Datasets:

TaylorAI

/

pubmed_author_manuscripts

like 3

Follow

Taylor 16

PMC007xxxxxx/PMC7614033.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101674869 Nat Microbiol Nat Microbiol Nature microbiology 2058-5276 35918425 7614033 10.1038/s41564-022-01178-w EMS159074 Article Viral biogeography of the mammalian gut and parenchymal organs Shkoporov Andrey N. †12 Stockdale Stephen R. 1 Lavelle Aonghus 1 Kondova Ivanela 3 Heuston Cara 1 Upadrasta Aditya 1 Khokhlova Ekaterina V. 1 van der Kamp Imme 1 Ouwerling Boudewijn 3 Draper Lorraine A. 1 Langermans Jan A.M. 34 Ross R Paul 12 Hill Colin †12 1 APC Microbiome Ireland, University College Cork, Cork, Ireland 2 School of Microbiology, University College Cork, Cork, Ireland 3 Biomedical Primate Research Centre, Rijswijk, The Netherlands 4 Department of Population Health Sciences, Veterinary Faculty, Utrecht University, Utrecht, The Netherlands † Corresponding authors: andrey.shkoporov@ucc.ie, c.hill@ucc.ie 01 8 2022 02 8 2022 27 12 2022 09 1 2023 7 8 13011311 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. The mammalian virome has been linked to health and disease but our understanding of how it is structured along the longitudinal axis of the mammalian gastrointestinal tract (GIT) and other organs is limited. Here we report a metagenomic analysis of the prokaryotic and eukaryotic virome occupying luminal and mucosa-associated habitats along the GIT, as well as parenchymal organs (liver, lung and spleen), in two representative mammalian species, the domestic pig and rhesus macaque (six animals per species). Luminal samples from the large intestine of both mammals harboured the highest loads and diversity of bacteriophages (class Caudoviricetes, family Microviridae and others). Mucosal samples contained much lower viral loads but a higher proportion of eukaryotic viruses (families Astroviridae, Caliciviridae, Parvoviridae). Parenchymal organs contained significant numbers of bacteriophages of gut origin, in addition to some eukaryotic viruses. Overall, GIT virome composition was specific to anatomical region and host species. Upper GIT and mucosa-specific viruses were greatly under-represented in distal colon samples (a proxy for faeces). Nonetheless, certain viral and phage species were found ubiquitously in all samples from the oral cavity to the distal colon. The dataset and its accompanying methodology may provide an important resource for future work investigating the biogeography of the mammalian gut virome. pmcIntroduction The gastrointestinal tracts (GIT) of humans and other mammals contain highly individualized microbiomes1–4, composed of bacteria5, archaea6, eukaryotic microorganisms7, and viruses8–10. The close association of microbes and their mammalian host as an ecological unit is increasingly recognised as important for health11,12. The gut virome, which is largely composed of bacterial viruses (bacteriophages or phages) remained a relatively unexplored area until recently, when a potential role for the virome in shaping bacterial communities was postulated13–16. A number of potential mechanisms by which such shaping could occur have been suggested, and include “kill-the-winner” dynamics in bacterial communities caused by phage predation (at least at strain and sub-strain level17); diversifying selection acting upon both adaptive mutations18,19 and phase variations20,21; as well as phage-mediated horizontal gene transfer (HGT) that could involve diverse mechanisms such as generalised, specialised and lateral transduction22–24. Our current understanding of the virome, and the phageome in particular, is limited and based mostly on sequencing-based studies of faecal samples, which represent static snapshots of the distal gut virome. Neither the temporal dynamics17, nor the variation and flux of viral populations along the longitudinal and transverse axes of GIT (the “viral biogeography” of the gut25,26), have received proper attention. Recent human cohort studies highlighted a tight association between the gut virome and gut bacteriome in terms of both α- and β-diversity17,27. Additionally, multiple lines of evidence suggest that many successful gut bacteriophages, such as the crAss-like phages, engage in long-term persistent relationships with their hosts28,29, in line with the “piggyback-the-winner” dynamics of temperate bacteriophages30. It is important to obtain a more detailed view of both the temporal and spatial dynamics of the virome in order to understand its interplay with the bacterial microbiome, its significance for human health and potential role in disease31,32. Complex macro- and micro-anatomy of the digestive tract, together with exocrine functions of GIT mucosa and accessory organs create a series of longitudinal and radial biochemical gradients, affecting the composition of local resident microbiota, including viruses25,33,34. Adaptation to such microhabitats is clearly evident amongst bacteria, such as body site-specific lactobacilli35 or various mucin-foraging bacteria36. Host-associated mutualistic and commensal bacteria have evolved persistence mechanisms such as adsorption and embedding into mucus layers, and potentially have access to anatomical sites protected from the luminal stream and the action of bacteriophages25,34. Similarly, the ability to bind to and accumulate in the mucous layer and potentially restrict bacterial invasion was also reported for certain bacteriophages, which prompted a discussion on the role of bacteriophages as a quasi-immune system of the digestive tract37–39. Pronounced physiological and anatomical differences between homologous GIT segments in different species of mammals, associated with digestive adaptations, adds another layer of complexity to this system40. In this study, we present a comprehensive biogeographical analysis of viruses in the GIT of two mammalian species, the domestic pig (Sus scrofa domesticus) and rhesus macaque (Macaca mulatta), chosen for their phylogenetic, physiological and anatomical relevance for humans. We focuse our attention on bacteriophage populations and attempt to answer two key questions. Firstly, what are the differences in virome composition between different digestive tract regions and how representative are distal gut samples of the virome in the upper GIT? Secondly, to what extent can the virome be shared between the digestive tract regions and with extra-GIT organs? Results Virome sequencing approach We applied shotgun metagenomic sequencing of VLP-enriched samples41–43, to characterize luminal/mucosal viral DNA and RNA content in different locations along the digestive tract. In order to adopt a broader taxonomic outlook and get insights into spatial virome organisation that go beyond the physiological and anatomical specifics of a particular mammalian species, we included six healthy domestic pigs (Sus scrofa domesticus) and six rhesus macaques (Macaca mulatta). Thirteen anatomical locations were sampled for each species, including skin, tongue, stomach, small intestine (SI; proximal [duodenum], medial [jejunum], and distal [ileum]), caecum, large intestine (LI; proximal, medial, and distal colon), as well as parenchymal organs (liver, lung and spleen). At relevant sites, both the mucosal and the surrounding luminal content were sampled (Fig. 1). Given the overwhelming prevalence of bacteriophages in mammalian faecal viromes17,41, their possible role in shaping the gut bacteriome16,19,44 and a lack of knowledge on their spatial distribution and populations dynamics in the GIT9,34, bacterial viruses were the primary focus of this study. Genomic DNA and cDNA of mixed viral populations were sequenced using the Illumina Novaseq platform to a median depth of 6.2±5.8M per sample (median±IQR; Supplementary Information). Unlike many previous studies, our viral metagenomics approach was designed to be relatively unbiased. A simple nucleic acid extraction procedure was adopted that deliberately avoided the use of micro-filtration, VLP precipitation using PEG/NaCl, chloroform extraction, or density gradient ultracentrifugation; all of which are known to introduce different biases in virome profiling studies42,43,45. By avoiding whole-genome amplification we also avoided artificial virome composition skewness, loss of viral diversity, and over-amplification of small circular ssDNA genomes17,46. Lastly, including lactococcal phage Q33 as an artificial internal viral standard in our extraction procedure allowed us to estimate abundance of viral genomes in the sample by comparing their mean sequence coverage with that of the internal standard17,43. Assembly of reads into contigs47, removal of redundancy across individual samples and animals17, and selection of viral sequences from a bacterial and mammalian host DNA background yielded of catalogue of 107,680 contigs, corresponding to putative complete and fragmented viral genomes (Extended Data Fig. 1). At least 24 families of prokaryotic and eukaryotic viruses48,49 were recognised across the viromes of the two mammalian species using an automated taxonomic assignment algorithm (Extended Data Fig. 2). Approximately half (58,573) of the contigs were broadly similar (≥50% sequence identity over 85% of contig length) to previously reported genomes of either cultured or uncultured viruses50–55, but the remaining half were only identifiable as viral using a de novo multi-classifier approach56. However, even within sequences homologous to previously reported viral genomes and genome fragments, the majority (31,032) constitute unclassified viral species by the recently proposed standard of metagenomic viral species delineation (≥95% sequence identity over ≥85% of its length)57. Absolute viral counts along the GIT proximal-distal axis Absolute quantitation of viral genomic contigs with ≥50% calculated completeness level (n = 2,442), grouped at viral family level, revealed pronounced differences in the virome between GIT locations, as well as the differences between the two animal species. The pig LI lumen is dominated by tailed bacteriophages (class Caudoviricetes, including crAss-like phages9,53,58) with total viral loads exceeding 109 genome copies g-1 contents. Similar total counts are evident in macaques, although small ssDNA Microviridae phages59 are the most numerous group of taxonomically classified viruses (Fig. 1). Total viral loads in large intestinal mucosa samples were two orders of magnitude lower than matched luminal samples, and eukaryotic viruses (families Circoviridae, Astroviridae, Calicivirdae and Parvoviridae) had higher relative weights in those locations. Stomach and SI lumen and mucosa were colonised by relatively even mixes of bacteriophages and eukaryotic viruses, with a characteristic prevalence of Parvoviridae in the pig small intestinal mucosa. Similar combinations of viral families were detectable in tongue mucosa and skin samples in both animal species. Interestingly, samples taken from lung, spleen and liver parenchyma in both species contained unexpectedly high viral loads, approaching and exceeding 106 genome copies g-1 of tissue. In macaques, these viral populations that are apparently associated with interior body milieu of healthy animals, were mainly represented by eukaryotic viruses of Circoviridae and Caliciviridae families. In both species, and especially in pigs, the viral consortia of interior milieu included bacteriophages, primarily from the Microviridae family (Fig. 1). We then used all 107,680 viral contigs, both high quality and highly fragmented57, to identify compositional virome differences between different body sites in both animal species (Fig. 2; Extended Data Fig. 1). While highly fragmented viral contigs are less useful for taxonomic classification and host identification purposes17,32, omitting them from diversity analyses would leave the majority of viral diversity untapped (>50% of all Illumina reads from most body sites) (Supplementary Information). To compensate for inter-individual virome differences and make the virome more comparable across animal cohorts we used gene sharing networks60 to group all non-singleton viral genomic contigs (n = 12,262) into 3,770 Viral Clusters (VCs). Virome composition along the GIT proximal-distal axis Multivariate virome comparison, based on fractional abundance of VCs at different sites, revealed a strong separation of large intestinal viromes from the small intestinal and gastric viromes in both animal species (Fig. 2A; Extended Data Fig. 3). When viewed across the two species, differences between organs were responsible for 11.1% of variance (Adonis with 1000 permutations, p ≤ 0.001). Surprisingly, inter-individual virome differences accounted for 9.6% of variance, higher than percent variance explained by animal species (4.9%; p ≤ 0.001). This is despite the fact that within each cohort animals were relatively inbred, lived in the same facility and were fed with a standardised diet. Moreover, between organ variance in interaction with the individual animal factor accounted for 30.0% of virome data variance (p ≤ 0.001), much higher than percent variance explained by similar interaction between organ and animal species factors (9.4%, p ≤ 0.001). Differences between mucosal and luminal virome explained only a relatively minor fraction of variance (1.0% for the main effect, 1.9% in interaction with organ factor; p ≤ 0.001). The major compositional separation between viromes of LI, SI and other organs seems to be closely aligned with overall diversity and total viral load (p ≤ 0.001 in PERMANOVA), with caecal and LI viromes being simultaneously the most taxonomically diverse and the most populous (Fig. 1; Fig. 2A-C). In a single macaque (M6) and pig (E6), all mucosal sites were sampled twice, with 1 cm separation between each pair of samples, to assess whether close proximity of mucosal sites in the gut correlates with increased similarity of the virome composition. In both small and large intestine there was a tendency for these paired samples to resemble each other more closely than more distant sites within these and other animals, but this did not reach the level of statistical significance (see Supplementary Information). We attempted to identify specific VCs driving the separation between organ-specific viromes (Fig. 2D), as well as VCs responsible for separation between luminal and mucosal viromes and the two animal species. Across the two species, a total of 217 VCs were differentially abundant between organ pairs in the following sequence: skin-tongue-stomach-SI-Caecum-LI (p < 0.05 in ANCOM test with Benjamini-Hochberg correction; see Supplemental results for more detail), with the largest fraction of these VCs (n = 119) being discriminatory between the SI and caecum/LI. Twenty VCs were found to be differentially abundant between luminal and mucosal sites in both species of animals, eleven of them being over-represented in mucosal sites compared to the luminal sites in the same organs. As described in the Supplemental results, many of the organ-discriminatory VCs were positively correlated with bacterial genera characteristic of a particular segment in the GIT (see Supplementary Information). Sharing of virome components between different regions in the GIT Having observed only this partial separation of GIT sites by virome composition, we reasoned that there should be extensive sharing of individual viral species/strains between multiple GIT sites in each of the animals. To investigate this, we returned to the level of individual viral contigs and visualised their sharing between organs in a particular sequence. Agreeing with the individualised nature of gut viromes demonstrated above, patterns of viral contig sharing between different organs were also unique, not only between pigs and macaques, but also between individual animals within each cohort (Extended Data Figs. 4-5). Despite that, common trends in viral sharing between organs could also be easily observed. As shown in aggregate maps of viral contig sharing, summarizing the data from all pigs (Fig. 3) and all macaques (Fig. 4), high diversity populations of LI bacteriophages (Fig. 2B-C) are also efficiently shared between all locations in caecum and colon (Fig. 3 and Fig. 4). Summing up data from all pigs, for instance, >2,000 crAss-like phage and 65-130 Microviridae genomic contigs are shared between sites from the caecum to the distal colon (in luminal and mucosal samples together). Similarly, in macaques, 65-109 crAss-like contigs and 33-98 Microviridae were found shared between same anatomical sites. The same pattern was also true for tailed bacteriophage genomic contigs in the families Siphoviridae, Myoviridae and Podoviridae (Extended Data Fig. 6). As a rule of thumb, >50% of viral contig diversity in pigs and >30% in macaques was shared between all locations in the LI. Extensive sharing of viral contigs was observed within SI of both animal species. By contrast, only 1-2% of pig distal SI viral diversity and <1% of the same in macaques was detectable in caecum samples (Fig. 3 and Fig. 4). This picture is however, complicated by the fact that some of the gastric and SI viral contigs that we failed to detect in the distal segment, were nevertheless found in caecum and/or in the lower segments of LI. This suggests that limitations in sequencing depth and/or strict criteria of contig detection might introduce some artificial gaps of contig detection across multiple anatomical sites in our data. Distal colonic samples (a proxy for faecal samples in our study), appeared to be good representatives of total viral diversity in the lower GI tract (>50% represented), and poorer representatives of the upper GI tract (~10% represented of gastric virome). Only a small fraction of tongue virome could be detected in the distal colon (Fig. 3 and Fig. 4). Nevertheless, our data contains numerous examples of prokaryotic and eukaryotic viruses (genomic contigs with ≥50% estimated completeness) shared across six or more different anatomical sites. In pigs, such examples include Astroviridae and Caliciviridae species in luminal, Parvoviridae in mucosal samples and parenchymal organs, as well as numerous bacteriophage types across anatomical locations. In macaques Circoviridae and and Caliciviridae species were ubiquitously found (Extended Data Fig. 7). Livers, lungs and spleens of both animal species, shared with the GIT sites not only the genomes of eukaryotic viruses (Circo-, Calici-, Parvoviridae) but also small genomes of Microviridae phages and other phage genomic contigs (Figs. 3-4; Extended Data Fig. 4-6). In the light of some recent publications, this can be interpreted as evidence for possible translocation of some digestive tract bacteriophages across healthy gut epithelia61, ending up in the internal organs (liver, lung, spleen), presumably via macropinocytosis, the portal vein (liver), lymphatic system, or perhaps via regurgitation of stomach contents (lung). Discussion Recent studies have observed correlations in gut bacteriome and phageome composition and claimed associations between altered virome composition and GIT diseases in humans31,32. It has been speculated that phages could play a decisive role in controlling bacterial population density and structure via “kill-the winner” or similar types of ecological dynamics62–64. Indeed, in simplified microbiota models exponential growth of phage under optimal conditions can lead to the rapid collapse of sensitive bacterial populations65, resulting in cascades of knock-on effects in non-susceptible bacterial populations via inter-bacterial interactions16. On the other hand, there is also convincing evidence that points toward a much less disruptive role of phages in microbiome composition, in that most numerically prevalent phage types are either temperate (existing in the form of prophages as well as free viral particles), or have evolved to support a long term, stable persistence in the microbiome with only limited effects on the density of bacterial host populations66. A number of potential persistence mechanisms have been proposed that includes phase variation of phage receptors in bacteria20,28 leading to herd immunity21, or physical segregation of mucus-embedded sensitive bacteria from luminal phages (“source-sink” model)34. It would be impossible to fully understand the dynamics of phage-host interaction and therefore the role of phages as either “drivers” or “passengers” in real-world complex microbiomes without having a detailed map of the virome in both temporal and spatial (biogeographic) dimensions. In this study we provide such a spatial map for two mammalian species, pigs and macaques. From a technical perspective the study was designed to minimize the biases typically associated with virome analysis42,47,67. We used unamplified nucleic acids and assembly-based cataloguing of unclassified viruses, coupled with quantitation by comparison against a spike-in viral standard. We also clustered individual sequences into VCs to allow us to robustly detect and quantify both known and unclassified viruses with DNA or RNA genomes. Unlike in many previous studies67, we revealed an abundance of RNA viruses, including unclassified phages belonging to Leviviricetes class, and mammalian viruses belonging to the Astroviridae and Caliciviridae (Supplementary Information). Small ssDNA Microviridae phages were found to be a dominant group in the macaque colon, a finding that previously would have been dismissed as a DNA amplification bias17,46. A limitation of this assembly-based approach was, however, that we almost certainly missed some of the low abundance viruses seen in a previous study of the macaque virome26. In the mammalian GIT, a number of factors may influence differences in the bacterial microbiota and virome between small and large intestines. Lower pH, higher oxygen tension, faster transit time and bile acid activity may limit bacterial growth in the SI, while a thicker mucus gel layer, slower transit time and shift to fermentation contribute to a large increase in microbial density in the LI68. As expected, the vast majority of phage biomass and diversity was concentrated in the colonic lumen, reflecting the dense community of bacterial hosts in that site. Upper GIT viromes were distinctly different and reflective of differences in bacteriome composition between different GIT regions (Supplementary Information). Direct correlations in density and composition between virome and bacteriome in the gut have been reported before17 and are consistent with the “piggyback-the-winner” ecological model30. Interestingly, distal gut luminal viromes appeared to be very homogenous, from caecum to distal colon and compositionally much more reflective of an individual animal, than of a particular location in the colon. This confirms that inter-individual variability remains a hallmark feature of the intestinal virome17, even within these highly controlled environments. Recently, stochastic assembly effects have been shown to drive inter-individual variability in the bacterial microbiota in mice69 and this phenomenon should apply similarly to the virome in pigs and macaques. Our results are in close agreement with research conducted on bacterial biogeography in the macaque gut, where Yasuda et al. observed predominantly inter-individual, and less location-specific, variation of luminal microbiota in jejunal, ileal, and colonic sites70. The same authors noted significant differences between luminal and mucosal microbiota in the same locations, with the latter being more influenced by biogeography than by an individual animal. In line with this, subsets of VCs appear to be specifically associated with both types of habitat (Supplementary Information). At the same time, mucosal samples show drastically reduced viral load and increased prevalence of viruses infecting mammalian cells. These results may support a recently proposed “source-sink” model34 arguing that exclusion of bacteriophages from mucous layer creates a refuge for bacterial cells, allowing the co-existence of virulent phage and sensitive bacterial cells in close proximity. This apparently disagrees with an earlier “bacteriophage adherence to mucus” model (BAM)37, which argued that an accumulation of bacteriophages and an increased virus-to-microbe ratio (VMR ~ 39:1) in the mucus creates a barrier limiting bacterial invasion and segregating bacterial population to the luminal space. In the absence of quantitative data on bacteria, our study cannot testify to the VMR ratios in the lumen and mucosa. The BAM model therefore, can still accommodate our results, with a caveat that certain bacteriophages possessing Ig-like protein domains required for binding to mucus37 are equally abundant in the mucus and in the lumen, while phages lacking this ability are excluded into the luminal space. One can envisage complex scenarios of phage-host interaction in the GIT, with some phage-host pairs following “source-sink” dynamics, while others showing behaviours more conforming with the BAM model. We observed extensive sharing of individual viral strains throughout the entire GIT. The most prominent examples were phages found continuously across multiple sites. For the majority of strains however, the continuous flow of phages from small to large intestine seems to be interrupted at the ileocaecal valve. This can be explained in part by drastic differences in composition (and presumably total biomass) of bacteriomes between SI and LI, which in turn support the growth of completely different phage populations. However, a complete extinction of small intestinal phages during passage from SI to LI seems unlikely, and therefore, the dilution effect, caused by vastly larger viral biomass supported by greater numbers of bacteria, combined with limitations imposed by sequencing depth, is a likely cause of the apparent disappearance of gastric and small intestinal phages in the caeca and LI. Despite our original expectations, we could not definitively confirm a tendency for mucosal samples taken at 1cm distance to be closer in virome composition to each other than to other sites in the same anatomical region (Supplementary Information), which again suggests a relative homogeneity of virome along the proximal-distal axis within each region of the digestive tract. This observation calls for future longitudinal studies to examine viral flow and local temporal differences in virome composition in the gut. Luminal samples from the distal colon, which can roughly be equated with faecal samples for the purpose of this study, are only representative of a fraction of the viral diversity present in different segments of the digestive tract. This is especially evident in the case of eukaryotic viruses, many of which are readily detectable in colonic mucosa (Astroviridae in pigs) or SI lumen (Caliciviridae in both pigs and macaques), and in parenchymal organs such as liver, lung and spleen, but not in the distal LI lumen. Interestingly, and agreeing with our earlier notion of virome individuality, each animal harboured a unique pattern of eukaryotic viruses, with regards to their taxonomic composition, strain variation and biogeographic distribution (Supplementary Information). The epidemiological and pathological significance of biogeographic distribution of these common viruses in porcine and murine GIT (in particular porcine Astroviridae71) is difficult to establish without further extensive population and longitudinal data collection. Our findings in this study were largely consistent between pigs and macaques, despite differences in species, environment, diet and age. Notably, the pigs at three months were weaned and in early adolescence, while macaques were adults (5-12 years old). We note that all animals were female, thus preventing any determination of possible sex-effects on intestinal biogeography. One of the interesting findings in this study was possible evidence of bacteriophage translocation from the gut into the systemic circulation and eventually parenchymal organs such as the liver, spleen and lungs. While animal dissection and sample collection for this study was conducted within a sanitary research environment, we could not achieve fully aseptic conditions in our pig facility. Therefore, it is possible some of the viral biomass in pig parenchymal organs that was orders of magnitude lower than was found in the gut could represent cross-contamination of solid organ samples. Nevertheless, we believe that this cannot fully explain our findings. Parenchymal organ viromes were dominated by eukaryotic viruses, and while phages present in these organs were specific strains shared with digestive tract viromes, they were not the most dominant strains. It has previously been demonstrated that at least specific phage types are able to adhere and translocate through the intestinal epithelial lining37,61. In our study, a tendency towards enrichment for smaller phages (family Microviridae) was observed in parenchymal organ viromes, which might indicate increased transepithelial diffusion of small viral particles. The exact fate of translocating phage and their systemic effects has so far remained unclear72,73, and our observations might be insightful for studying anti-phage immune responses74. This work highlights that focussing on distal LI sampling (or faecal sampling) dramatically under-represents GI viral communities (particularly eukaryotic viruses), and points to consistent drop-out of upper GI viral communities in colonic samples. In addition to these findings, we detected some overlap between viral communities in parenchymal organs and the GIT which was not related to their overall abundance, suggesting that there may be some degree of specificity to viral translocation. Finally, we propose that this dataset and its accompanying methodology may provide an important catalogue of gut viruses and resource for future investigators in the field. Methods Ethical approval and study design The study design was developed with consideration to the three Rs for ethical use of animals in science: replacement, reduction, and refinement. The proposed euthanasia only study was reviewed by the Animal Welfare Body (AWB) of University College Cork (Euthanasia Only Authorisation 17-005). With authorisation and under the remit of authorised and experienced personnel, the study was performed succinctly and with minimal distress to the animals involved. No statistical methods were used to pre-determine sample sizes but our sample sizes are similar to those reported in previous publications26,70. Data collection and analysis were not performed blind to the conditions of the experiments. No randomisation procedures were used and no data points were excluded from any of the analyses. Animal sampling procedures (i) Sus scrofa domesticus – pigs Six healthy female Landrace pigs (body mass approximately 30 kg, approximately 3 months of age) were sourced from a local farm in Cork, Ireland. All pigs were raised in a shared environment and on the same diet, although the relatedness of their parentage is unknown. Pigs were transported to the research facility on the morning they were to be euthanised, with two animals sampled back-to-back per day. Before euthanasia, work surfaces and necessary tools were disinfected using Virkon S disinfectant. Following anaesthetic overdose with Pentobarbital (150mg/kg) death was confirmed by an authorised person, and tissue samples were collected. All biopsies (min. 3 cm × 3 cm) were minimally handled on site. Therefore, samples were not washed or stored in a buffer but placed directly into 50 ml Falcon tubes and stored on dry ice and then at -80°C. Initially, external biopsies of the tongue and skin were collected. Skin biopsies were taken from around the shoulder. Once external biopsies were obtained, pigs were rolled onto their back and a midline incision was performed from below the neckline of the animals to immediately preceding the genitalia. The complete gastrointestinal tract was removed from the abdominal cavity, with the connective tissue severed where required. Surgical thread was used to seal sections of the gastrointestinal tract. Two knots, approximately 2 cm apart, were tied tightly without severing the gastrointestinal tract. Subsequently, sections of the GI tract were separated by cutting between the tied knots that prevented the intestinal contents from leaking. Both the small and large intestines of animals were sealed in three approximately equal length sections to represent the proximal, medial, and distal regions. All sections of the GI tract were treated similarly. Briefly, an opening into the sealed GI tract tube was created and the contents removed before large representative sections of the bowel were cut and stored. Finally, stomach mucosa was from fundic region, and parenchymal organs were removed from the abdominal cavity of animals with large biopsies sections stored for later analysis. The processing time per animal was approximately 3 hours. (ii) Macaca mulatta – rhesus macaques Six healthy Indian-origin, female adult rhesus macaques aged 5-12 years with bodyweight 5.3 to 10.6 kg were used. All animals were born and raised in naturalistic multi-generational breeding groups at the Biomedical Primate Research Centre (BPRC), Rijswijk, The Netherlands, in comparable environments. All enclosures contained environmental enrichment and bedding to stimulate their natural behaviour. They were daily fed monkey chow pellets (Ssniff, Soest, Germany) in the morning, complemented with fruit and vegetables. Over a period of 5 months animals were euthanised using pentobarbital (70 mg/kg) following sedation with ketamin (10 mg/kg). The necropsy and collection of samples were done immediately after euthanasia. For isolation and collection of macaque samples strict sterility protocol and safety procedures were used. The sterility of the necropsy table and the surgical instruments were assured using Virkon S, sterilization procedures and use of disposable scalpels. Macaque tissue samples were retrieved and stored similarly to the procedures outlined for pigs. For the collection of the parenchymal and intestinal samples disposable scalpels and autoclaved scissors and forceps were used. To avoid contamination, after opening the thoracic and abdominal cavity the first samples collected were from the parenchymal organs- liver, spleen, and lung following by the intestinal samples. After each animal, the table was thoroughly cleaned with hot water and detergent followed by disinfection by Virkon S, to prepare for the next animal. All samples were immediately placed on dry ice and stored at -80°C. Tissue samples were transported on dry ice to APC Microbiome Ireland for further processing Biopsy preparation procedure, VLP enrichment, and nucleic acid sequencing GI and parenchymal organ sections of pigs and macaques were processed identically, in the same research facility, by the same team members, but on different days. Tissue samples were thawed on ice until completely defrosted. Excess faecal material on caecal and colon tissue sections were washed with sterile SM buffer (50 mM Tris-HCl; 100 mM NaCl; 8.5 mM MgSO4; pH 7.5). Tissue sections were stretched and pinned to a Styrofoam board using sterile syringes. Defined volume pinch biopsies of mucosal surfaces were collected with an endoscopic biopsy forceps. A “double-bite” of tissue samples at the same site ensured the accurate and complete loading of the forceps’ jaws. Mucosal pinches were removed from the forceps directly into pre-labelled Eppendorf tubes, filled with 400 μL of sterile SM buffer for processing. To enable comparisons of viral load across biopsy samples, 10 μL of 107 plaque forming units per millilitre of lactococcal phage Q33 were added to all samples. Additionally, Q33 in SM buffer or SM buffer-only were processed as negative controls. Fresh 0.5 M dithiothreitol (DTT) was prepared in 1 mL of SM buffer. A volume of 16 μL of the DTT stock was added to samples to achieve a final concentration of 20 mM, and samples were incubated at 37°C for 30 minutes. DTT was used to gentle solubilize mucin with minimal disruption of phage virions, as this disulphide bond reducing agent was previously demonstrated to release large quantities of non-mucin proteins from small intestine porcine preparations75. Host cellular debris and bacterial cells were pelleted by gentle centrifugation at 4000 g for 30 minutes at room temperature. Subsequently, 400 μL of liquid was aspirated and treated with 40 μL of DNase/RNase buffer (50 mM CaCl2; 10 mM MgCl2), 12 μL of DNase (manufacturer), 4 μL of RNase, and incubated at 37°C for 1 hour with intermittent inversion approx. every 15 minutes. Enzymes were inactivated by incubating at 65°C for 10 minutes. Viral-enriched samples void of free nucleic acids were lysed using the QIAgen Blood and Tissue Kit following the manufacturers guidelines. However, samples were eluted in only 20 μL of AE elution buffer to increase the final concentration of nucleic acid obtained. Virome shotgun library preparation and sequencing Reverse transcription (RT) reaction was performed using SuperScript IV First Strand Synthesis System (Invitrogen/ThermoFisher Scientific) with 11 μL of purified VLP nucleic acids sample and random hexamer oligonucleotides according to manufacturer’s protocol. Concentration of DNA purified using DNeasy Blood & Tissue kit (QIAGEN) was determined using the Qubit dsDNA HS kit and the Qubit 3 fluorometer (Invitrogen/ThermoFisher Scientific). DNA/cDNA yields varied between 0.05 and 29 ng/μl, with some samples being below detection limit. Library preparation was carried out using Accel-NGS 1S Plus kit (Swift Biosciences) according to manufacturer’s instructions. Briefly, 20 μl of RT product (regardless of DNA concentration, as the kit is flexible with regards to the amount of input DNA) were taken for sonication after adjusting the volume to 52.5 μl with low-EDTA TE buffer. Shearing of unamplified DNA/cDNA mixture (variable amounts of DNA) was performed on M220 Focused-Ultrasonicator (Covaris) with the following settings: peak power of 50 W, duty factor of 20%, 200 cycles per burst, total duration of 35 s. All following steps were performed in accordance with the manufacturer’s protocol. A 0.8 DNA/AMPure beads v/v ratio was used across all purification steps in the Accel-NGS 1S Plus protocol. Post-preparation library QC (fragment length distribution and quantitation) was performed using Agilent Bioanalyzer 2100 with High Sensitivity DNA kit and Invitrogen Qubit. Dual-indexed pooled library was sequenced using 2×150 nt paired-end sequencing run on an Illumina NovaSeq platform at GENEWIZ (Leipzig, Germany). In order to control for contamination of samples with exogenous viruses and viral nucleic acids, including lab reagent-derived and environmental, we also performed extraction from 400 μL of sterile SM buffer alone. Two samples were processed simultaneously with pig and macaque samples using the same protocol. Both of them yielded DNA/cDNA below detection limit after extraction, and only trace (insufficient for sequencing) amount of DNA was visible. To compensate for low yield, third sample was subjected to whole-genome multiple displacement DNA amplification (MDA, Illustra GenomiPhi V2 DNA Amplification Kit) as described before17. Analysis of virome shotgun sequencing data Raw reads were processed using Cutadapt v2.4 to remove adaptor sequences. Trimmomatic v0.3676 was used for quality-based trimming and filtration of reads with the following parameters: ‘SLIDINGWINDOW:4:20 MINLEN:60 HEADCROP:10’. Reads aligning to mammalian genomes were identified using Kraken v1.1.1 in combination with Macaca mulatta (GCF_000772875.2_Mmul_8.0.1) and Sus scrofa (GCF_000003025.6_Sscrofa11.1) reference genome files. Following removal of mammalian reads, levels of contamination with bacterial genomic reads were assessed using ViromeQC tool77. Reads were then assembled into contigs on a per sample basis using SPAdes assembler v3.13.0 in metagenomic mode with standard parameters78. Additionally, in attempt to assemble low-abundance genomes, reads were pooled by animal and assembled using MEGAHIT v1.2.1-beta79. All contigs > 1 kb were then pooled together and an all-vs-all BLASTn search was performed with e-value cut-off of ≤ 1E-20. Contig redundancy was removed by identifying pairs sharing 90% identity over 90% of the length (of the shorter contig in each pair) retaining the longest contig in each case. To extract viral contigs from a background of bacterial contamination several selection criteria were used. Firstly, contigs aligning using BLASTn v2.10.0+80 against viral sequences in NCBI RefSeq database (release 208), Gut Virome Database51, JGI IMG/VR database (v3, release 12-10-2020)50, Gut Phageome Database54, and the recent human gut phage MGV database55, as well as our in-house database of crAss-like phage genomes (n=1,576), with at least 50% identity over 85% of contig length (e-value cut-off of individual hits ≤ 1E-10) were deemed as viral. Secondly, contigs that identified as viral using VirSorter2 pipeline56 with strict criteria (score ≥ 0.9 OR score ≥ 0.7 with at least 1 viral hallmark protein-coding gene present) were added. Completeness level of viral genomic contigs was determined using CheckV81 with default parameters. VirSorter2-identified viral contigs marked as prophages by CheckV (Provirus==Yes), were eliminated. Viral genomic contigs identified by these approaches constituted the final non-redundant viral sequence catalogue (n = 107,680). Protein coding genes on viral contigs were predicted using Prodigal82 (-meta mode). Translated protein sequences were searched against PHROGs database83 of virus-specific protein family profile HMMs, using hmmscan (HMMER v3.1b2; e-value cut-off of ≤ 1E-5); viral protein sequences from NCBI nr (as of 02-11-2021) and viral RefSeq (release 208) databases and crAss-like phage proteins from an in-house database (n=7,356) using BLASTp v2.10.0+. (e-value cut-off of ≤1E-10). Circular genomic contigs were identified using LASTZ. G+C content was calculated using EMBOSS geecee. Assignment of contigs to viral families was accomplished using Demovir script (https://github.com/feargalr/Demovir), as described before17. Clustering of viral genomic contigs (only for contigs with >3 kb in length) into viral clusters (VCs, approximately genus-level operational taxonomic groups) was done using vConTACT2 software60 with the following optional parameters: --rel-mode Diamond --db ProkaryoticViralRefSeq85-Merged --pcs-mode MCL --vcs-mode ClusterONE. Viral genomic contig catalogue was further manually curated to remove coliphage phiX174 genome (commonly used as a spike-in by sequencing facilities). Phage lifestyle (temperate vs. virulent) was predicted using BACPHLIP84 using 0.95 confidence threshold. Remaining (non-viral) non-redundant contigs were assigned to bacterial taxa by performing BLASTn search against bacterial RefSeq (release 99) and HMP Reference Genomes databases. Taxonomic assignments were made at genus level, for contigs having 90% identity over ≥85% of combined alignment(s) length against a reference bacterial genome. CRISPR arrays were predicted on bacterial contigs and spacer sequences were extracted using PILER-CR v1.0685. To predict the hosts of phage, data was aggregated from several sources. Firstly, previously predicted hosts for viral species included into IMG/VR database were assigned to viral contigs in our catalogue belonging to the same species (≥95% identity over ≥85% of viral genomic contig length, in accordance with MIUViG criteria for viral species demarcation in metagenomic sequence data57). Secondly, a search against an in-house CRISPR spacer database (derived from bacterial RefSeq [release 89] and HMP Reference Genomes) was performed as described before17 to assign hosts to viral contigs, missing close homologs in the IMG/VR database. In a similar fashion, matches were found with CRISPR spacers encoded by bacterial contigs (with taxonomy assigned as described above) in the present study dataset. Lastly, BLASTn similarity of viral contigs to closely related ≥90% identity over ≥85% of viral contig length) prophages in bacterial genomes (RefSeq database of bacterial genomes, release 99; HMP Reference genomes database86) was used to assign hosts where neither IMG/VR nor CRISPR approaches were successful. Lastly, tRNA gene hits against NCBI nt database (release 28-11-2020) and bacterial RefSeq database (release 99) were used to predict hosts for cases where all other methods failed. At the VC level, host was assigned using the majority vote rule, after aggregating host predictions from individual viral contigs – members of a particular VC. Quantitative analysis of viral metagenomic data was performed essentially as described before17. Quality filtered reads were aligned to the curated viral contig database on a per sample basis using Bowtie2 v2.3.4.1 in the ‘end-to-end’ mode. A count table of contigs versus samples was subsequently generated using SAMTools v1.7. Sequence coverage was calculated per nucleotide position per contig per sample using SAMTools ‘mpileup’ command. Read counts for contigs in samples showing less than a minimum of 1x coverage of 75% of a contig length, were set to zero17. Absolute viral counts were calculated for viral genomic contigs based on comparison of their relative abundance with that of the externally added standard (lactococcal phage Q33). Only viral contigs with estimated completeness of >50% were taken into account based on an assumption that additional genomic fragments, which together constitute the remaining <50% portion of the complete genome, will not be counted and therefore will not artificially inflate the calculated total viral loads. Bacterial 16S rRNA amplicon sequencing During the biopsy preparation procedure, the porcine and macaque biopsy samples were reduced by DTT followed by centrifugation to reduce host tissue and bacterial cells and enrich the viral-like particles. However, the bacterial-containing pellet was used as the starting material for complementary 16S rRNA analysis of bacterial communities associated with the same biopsy samples analysed with respect viromes. The preparation and sequencing of 16S rRNA gene V3-V4 segment libraries followed the procedure outlined previously43. Analysis of bacterial 16S rRNA amplicon sequencing data Bacterial 16S rRNA amplicon sequencing data was processed using a pipeline based on USEARCH v8.1 (64 bit). Forward and reverse reads of 16S rRNA V3-V4 segment were merged together allowing for an expected error rate of <0.5 per nucleotide position at overlap. Merged sequences were truncated to remove forward (first 17 nt) and reverse (last 21 nt) 16S rRNA primers. Reads were then de-replicated and singletons were removed, followed by clustering into OTUs at 97% sequence identity level. Chimeras were removed using -uchime_ref function with rdp_gold reference database. Individual reads were then assigned to OTUs generated above at 97% sequence identity cut-off and read count matrix was generated. Finally, taxonomic assignment of OTUs was performed using RDP Classifier v2.12. Statistical methods All statistical analysis of sequencing data was carried out in R environment v4.1.0. Descriptive statistical visualisations were created using ggplot2 v3.3.3. Network visualisations were created using igraph v1.2.6. Heat maps were produced using gplots v3.1.1. Sankey diagrams were made using networkD3 v0.4. Permutational multivariate analysis of variance was performed using the adonis() function in Vegan with Bray-Curtis distances. Virome β-diversity was visualised through canonical analysis of principal coordinates with Bray-Curtis distances [capscale() function in Vegan v2.5-7 with default parameters]. Comparison of Bray-Curtis distances between viromes within organs was done using Wilcoxon test with Benjamini-Hochberg corrections. VCs differentially abundant between organs, tissues and animal species were identified using ANCOM-II87,88 with Benjamini-Hochberg correction, α=0.05, and w0 threshold set at 0.7. For between-organ tests, individual animal was used as random effect variable and models were adjusted for tissue type (lumen vs. mucosa) as covariate. This was followed by post hoc ANCOM-II tests for specific pairs of organs. For between-tissue tests (lumen vs. mucosa) and between-species tests (macaques vs. pigs), models were adjusted for individual animal or organ type as covariate, respectively. Correlations between fractional abundances of individual viral genomic contigs (or VCs) and bacterial 16S rRNA OTUs (or genera) were calculated using Spearman rank correlation method with Bonferroni correction for multiple tests. Extended Data Extended Data Fig. 1 Catalogue of viral genomic contigs assembled from trimmed and filtered Illumina reads (n = 107,680). A, Average read coverage vs. contig length, categories of viral genomic contigs identified by CheckV (high-quality genomes vs genome fragments according to definitions given by the MIUViG standard); B, distribution of viral genomic contigs by completeness level as predicted by CheckV with high quality draft and complete genomes by MIUViG standard highlighted in blue; C, cumulative fractional abundance of genomic contigs with different levels of completeness. Extended Data Fig. 2 Taxonomic distribution, size, and completeness of viral genomic contigs. Different viral families are shown in separate panels. Assignments are based on Demovir script. Contig size is plotted on log10-scaled x-axis. Contig completeness is predicted using CheckV. Extended Data Fig. 3 Aggregated fractional abundance of viral families across all anatomical sites in pigs (n = 6) and macaques (n = 6) in the study. Rows represent viral families, columns – sites in individual animals; the top annotation bar represent tissue types (lumen vs mucosa). Data is log10-transformed and presented with hierarchical clustering based on relative abundance patterns. Extended Data Fig. 4 Sharing of viral genomic contigs between different anatomical sites in individual pigs (n = 6). Vertical grey rectangles height is proportional to viral richness (individual genomic contig counts) at each location, aggregated across luminal and mucosal samples; thickness of coloured connectors is proportional with the number of genomic contigs of each viral family shared between pairs of locations; SI, small intestine; LI, large intestine; Prox/Mid/Dist, proximal, medial and distal portions, respectively; unclassified genomic contigs were excluded; C, fraction of viral contig diversity from each organ represented in the distal LI. Extended Data Fig. 5 Sharing of viral genomic contigs between different anatomical sites in individual macaques (n = 6). Vertical grey rectangles height is proportional to viral richness (individual genomic contig counts) at each location, aggregated across luminal and mucosal samples; thickness of coloured connectors is proportional with the number of genomic contigs of each viral family shared between pairs of locations; SI, small intestine; LI, large intestine; Prox/Mid/Dist, proximal, medial and distal portions, respectively; unclassified genomic contigs were excluded; C, fraction of viral contig diversity from each organ represented in the distal LI. Extended Data Fig. 6 Numbers of viral genomic contigs shared between pairs of organs in pigs and macaques. Numbers of shared contigs are expressed as aggregate counts of unique contigs shared between sites across all animals for each of the two species; SI, small intestine; LI, large intestine; Prox/Mid/Dist, proximal, medial and distal portions, respectively. Extended Data Fig. 7 Absolute counts of some of the most ubiquitous viral genomic contigs present in pigs and macaques. Only contigs with >50% estimated completeness and shared between 6 or more sites in any of the animals are displayed. Each line corresponds to an individual genomic contig (potentially collapsing multiple viral strains). Colours are according to viral families. Each panel represent an individual animal. Supplementary Material Supplemental Materials Acknowledgements This research was conducted with the financial support of Science Foundation Ireland (SFI) under Grant Number SFI/12/RC/2273 (Colin Hill & Paul Ross), a Science Foundation Ireland’s Spokes Programme which is co-funded under the European Regional Development Fund under Grant Number SFI/14/SP APC/B3032, Wellcome Trust Research Career Development Fellowship [220646/Z/20/Z] (Andrey Shkoporov); and a research grant from Janssen Biotech, Inc (Colin Hill & Paul Ross). This research was funded in whole, or in part, by the Wellcome Trust [220646/Z/20/Z]. For the purpose of open access, the authors have applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. Authors wish to thank Mr. Tom Haaksma (BPRC) for his technical help with animal sampling, as well as Dr. Jamie Fitzgerald and Ms. Julia Eckenberger for the fruitful discussion of statistical methods used in the study. Schematics in Figures 1-4 were created with BioRender.com. Data availability All data needed to evaluate the conclusions in the paper are present in the paper, Supplementary Information file, and the additional dataset available at https://doi.org/10.6084/m9.figshare.15149247.v2. Raw sequencing data are available from NCBI databases under BioProject PRJNA753514. Code availability Source data and custom R code used in this study are available at https://doi.org/10.6084/m9.figshare.15149247.v2. Further information and requests for data, code and resources should be directed to and will be fulfilled by Prof. Andrey Shkoporov (andrey.shkoporov@ucc.ie) and Prof. Colin Hill (c.hill@ucc.ie).. Fig. 1 Abundance of different viral families along the GIT and in parenchymal organs of domestic pigs (A, n=6) and rhesus macaques (B, n=6). SI, small intestine; LI, large intestine; Prox/Mid/Dist, proximal, medial and distal portions, respectively. Absolute abundance of viral genomes was calculated by comparing coverage with that of the spike-in standard (phage Q33). Only genomes with >50% of estimated completeness were taken into account when calculating viral loads. Bar heights correspond to median values across six animals of each species, error bars denote interquartile ranges. Rows of plots in each of panel A and B are tissue types (top to bottom: lumen, mucosa, and skin/parenchyma). Columns of plots are anatomical sites. Middle portion of the figure shows schematic locations of sampled anatomical sites. Fig. 2 α- and β-Diversity of viromes in various anatomical sites in domestic pigs (n=6) and rhesus macaques (n=6). A, Canonical analysis of principal coordinates (CAPSCALE) of Bray-Curtis dissimilarities between virome samples, based on fractional VC counts; anatomical locations, Shannon diversity index and total viral load used as constraining explanatory variables (vectors are only shown for the latter two); ellipses represent 95% confidence regions (see colour legend below, common with panels B and C); B and C, Shannon diversity index calculated with read counts for individual viral genomic contigs (pigs and macaques, respectively); SI, small intestine; LI, large intestine; Prox/Mid/Dist, proximal, medial and distal portions, respectively; organ colours are matched with those in panel A, dashed boxplots represent mucosal sites; boxplots are standard Tukey type with interquartile range (box), median (bar) and Q1 – 1.5 × IQR/Q3 + 1.5 × IQR (whiskers); D, VCs differentially abundant between organs selected using ANCOM-II test (p < 0.05 after Benjamini-Hochberg correction); rows represent VCs, columns – sites in individual animals; a series of post-hoc tests identified VCs (annotated with black bricks) discriminatory between the following anatomic locations: Skin-Tongue, Tongue-Stomach, Stomach-SI, SI-Caecum, and SI-LI; the top and the right-hand side annotation bars represent tissue types (lumen vs mucosa) and viral families of VCs respectively; tree represents hierarchical clustering of VCs based on relative abundance patterns. An expanded version of this panel is provided as supplementary Fig. 5. Fig. 3 Sharing of viral genomic contigs between different anatomical sites in pigs (n=6). A, fraction of viral contig diversity shared between pairs of sites in both directions (white arrows/boxplots are forward direction, grey are reverse), in the order indicated in panel B; Boxplots are standard Tukey type with interquartile range (box), median (bar) and Q1 – 1.5 × IQR/Q3 + 1.5 × IQR (whiskers). B, aggregated map of viral contig sharing across six animals; vertical grey rectangles height is proportional to viral richness (individual genomic contig counts) at each location, aggregated across luminal and mucosal samples; thickness of coloured connectors is proportional with the number of genomic contigs of each viral family shared between pairs of locations; SI, small intestine; LI, large intestine; Prox/Mid/Dist, proximal, medial and distal portions, respectively; unclassified genomic contigs were excluded; C, fraction of viral contig diversity from each organ represented in the distal LI; Boxplots are standard Tukey type as above. Fig. 4 Sharing of viral genomic contigs between different anatomical sites in macaques (n=6). A, fraction of viral contig diversity shared between pairs of sites in both directions (white arrows/boxplots are forward direction, grey are reverse), in the order indicated in panel B; Boxplots are standard Tukey type with interquartile range (box), median (bar) and Q1 – 1.5 × IQR/Q3 + 1.5 × IQR (whiskers). B, aggregated map of viral contig sharing across six animals; vertical grey rectangles height is proportional to viral richness (individual genomic contig counts) at each location, aggregated across luminal and mucosal samples; thickness of coloured connectors is proportional with the number of genomic contigs of each viral family shared between pairs of locations; SI, small intestine; LI, large intestine; Prox/Mid/Dist, proximal, medial and distal portions, respectively; unclassified genomic contigs were excluded; C, fraction of viral contig diversity from each organ represented in the distal LI; Boxplots are standard Tukey type as above. Author contributions Conceptualization, A.N.S., S.R.S., R.P.R, and C.Hill; Design of work, A.N.S., S.R.S., A.L., I.K., L.A.D., C.Heuston and J.L.; Acquisition of data, A.N.S., S.R.S., A.L., I.K., C.Heuston, A.U., E.V.K., I. v. d. K., B.O.; Analysis & Interpretation, A.N.S., S.R.S., A.L., I.K., C.Heuston, J.L., L.A.D., R.P.R, and C.Hill; Software, A.N.S.; All authors contributed to drafting and revising of the manuscript. Competing interests The authors declare that they have no competing interests. 1 Pasolli E Extensive Unexplored Human Microbiome Diversity Revealed by Over 150,000 Genomes from Metagenomes Spanning Age, Geography, and Lifestyle Cell 2019 176 649 662 e20 30661755 2 Zhu A Sunagawa S Mende DR Bork P Inter-individual differences in the gene content of human gut bacterial species Genome Biology 2015 16 82 25896518 3 Flores GE Temporal variability is a personalized feature of the human microbiome Genome Biology 2014 15 531 25517225 4 Mehta RS Stability of the human faecal microbiome in a cohort of adult men Nature Microbiology 2018 3 347 355 5 Arumugam M Enterotypes of the human gut microbiome Nature 2011 473 174 180 21508958 6 Wampach L Colonization and Succession within the Human Gut Microbiome by Archaea, Bacteria, and Microeukaryotes during the First Year of Life Front Microbiol 2017 8 7 Laforest-Lapointe I Arrieta MC Microbial Eukaryotes: a Missing Link in Gut Microbiome Studies mSystems 2018 3 e00201-17 29556538 8 Liang G The stepwise assembly of the neonatal virome is modulated by breastfeeding Nature 2020 581 470 474 32461640 9 Shkoporov AN Hill C Bacteriophages of the Human Gut: The “Known Unknown” of the Microbiome Cell Host & Microbe 2019 25 195 209 30763534 10 Manrique P Healthy human gut phageome Proc Natl Acad Sci U S A 2016 113 10400 10405 27573828 11 Shaw LP Modelling microbiome recovery after antibiotics using a stability landscape framework The ISME Journal 2019 13 1845 1856 30877283 12 Lloyd-Price J Abu-Ali G Huttenhower C The healthy human microbiome Genome Medicine 2016 8 51 27122046 13 Breitbart M Viral diversity and dynamics in an infant gut Res Microbiol 2008 159 367 373 18541415 14 Roux S Hallam SJ Woyke T Sullivan MB Viral dark matter and virus–host interactions resolved from publicly available microbial genomes eLife Sciences 2015 4 e08490 15 Minot S Rapid evolution of the human gut virome Proc Natl Acad Sci U S A 2013 110 12450 12455 23836644 16 Hsu BB Dynamic Modulation of the Gut Microbiota and Metabolome by Bacteriophages in a Mouse Model Cell Host & Microbe 2019 25 803 814 e5 31175044 17 Shkoporov AN The Human Gut Virome Is Highly Diverse, Stable, and Individual Specific Cell Host & Microbe 2019 26 527 541 e5 31600503 18 De Sordi L Khanna V Debarbieux L The Gut Microbiota Facilitates Drifts in the Genetic Diversity and Infectivity of Bacterial Viruses Cell Host & Microbe 2017 22 801 808 e3 29174401 19 De Sordi L Lourenço M Debarbieux L The Battle Within: Interactions of Bacteriophages and Bacteria in the Gastrointestinal Tract Cell Host & Microbe 2019 25 210 218 30763535 20 Porter NT Phase-variable capsular polysaccharides and lipoproteins modify bacteriophage susceptibility in Bacteroides thetaiotaomicron Nature Microbiology 2020 5 1170 1181 21 Turkington CJR Morozov A Clokie MRJ Bayliss CD Phage-Resistant Phase-Variant Subpopulations Mediate Herd Immunity Against Bacteriophage Invasion of Bacterial Meta-Populations Front Microbiol 2019 10 22 Chiang YN Penadés JR Chen J Genetic transduction by phages and chromosomal islands: The new and noncanonical PLOS Pathogens 2019 15 e1007878 31393945 23 Chen J Genome hypermobility by lateral transduction Science 2018 362 207 212 30309949 24 Kleiner M Bushnell B Sanderson KE Hooper LV Duerkop BA Transductomics: sequencing-based detection and analysis of transduced DNA in pure cultures and microbial communities Microbiome 2020 8 158 33190645 25 Donaldson GP Lee SM Mazmanian SK Gut biogeography of the bacterial microbiota Nature Reviews Microbiology 2016 14 20 32 26499895 26 Zhao G Virome biogeography in the lower gastrointestinal tract of rhesus macaques with chronic diarrhea Virology 2019 527 77 88 30468938 27 Moreno-Gallego JL Virome Diversity Correlates with Intestinal Microbiome Diversity in Adult Monozygotic Twins Cell Host & Microbe 2019 25 261 272 e5 30763537 28 Shkoporov AN Long-term persistence of crAss-like phage crAss001 is associated with phase variation in Bacteroides intestinalis BMC Biology 2021 19 163 34407825 29 Siranosian BA Tamburini FB Sherlock G Bhatt AS Acquisition, transmission and strain diversity of human gut-colonizing crAss-like phages Nature Communications 2020 11 280 30 Silveira CB Rohwer FL Piggyback-the-Winner in host-associated microbial communities npj Biofilms and Microbiomes 2016 2 16010 28721247 31 Norman JM Disease-specific Alterations in the Enteric Virome in Inflammatory Bowel Disease Cell 2015 160 447 460 25619688 32 Clooney AG Whole-Virome Analysis Sheds Light on Viral Dark Matter in Inflammatory Bowel Disease Cell Host & Microbe 2019 26 764 778 e5 31757768 33 Lloyd-Price J Strains, functions and dynamics in the expanded Human Microbiome Project Nature 2017 550 61 28953883 34 Lourenço M The Spatial Heterogeneity of the Gut Limits Predation and Fosters Coexistence of Bacteria and Bacteriophages Cell Host & Microbe 2020 28 390 401 e5 32615090 35 Pan M Hidalgo-Cantabrana C Barrangou R Host and body site-specific adaptation of Lactobacillus crispatus genomes NAR Genomics and Bioinformatics 2020 2 36 Tailford LE Crost EH Kavanaugh D Juge N Mucin glycan foraging in the human gut microbiome Front Genet 2015 6 37 Barr JJ Bacteriophage adhering to mucus provide a non–host-derived immunity PNAS 2013 110 10771 10776 23690590 38 Barr JJ Subdiffusive motion of bacteriophage in mucosal surfaces increases the frequency of bacterial encounters PNAS 2015 112 13675 13680 26483471 39 Yutin N Discovery of an expansive bacteriophage family that includes the most abundant viruses from the human gut Nature Microbiology 2018 3 38 46 40 Karasov WH Douglas AE Comparative Digestive Physiology Comprehensive Physiology American Cancer Society 2013 741 783 10.1002/cphy.c110054 41 Hoyles L Characterization of virus-like particles associated with the human faecal and caecal microbiota Res Microbiol 2014 165 803 812 25463385 42 Kleiner M Hooper LV Duerkop BA Evaluation of methods to purify virus-like particles for metagenomic sequencing of intestinal viromes BMC Genomics 2015 16 7 25608871 43 Shkoporov AN Reproducible protocols for metagenomic analysis of human faecal phageomes Microbiome 2018 6 68 29631623 44 Breitbart M Bonnain C Malki K Sawaya NA Phage puppet masters of the marine microbial realm Nature Microbiology 2018 3 754 45 Conceição-Neto N Modular approach to customise sample preparation procedures for viral metagenomics: a reproducible protocol for virome analysis Scientific Reports 2015 5 srep16532 46 Roux S Towards quantitative viromics for both double-stranded and single-stranded DNA viruses PeerJ 2016 4 e2777 28003936 47 Sutton TDS Clooney AG Ryan FJ Ross RP Hill C Choice of assembly software has a critical impact on virome characterisation Microbiome 2019 7 12 30691529 48 Adriaenssens EM Taxonomy of prokaryotic viruses: 2018-2019 update from the ICTV Bacterial and Archaeal Viruses Subcommittee Arch Virol 2020 165 1253 1260 32162068 49 Koonin EV Global Organization and Proposed Megataxonomy of the Virus World Microbiol Mol Biol Rev 2020 84 50 Roux S IMG/VR v3: an integrated ecological and evolutionary framework for interrogating genomes of uncultivated viruses Nucleic Acids Research 2021 49 D764 D775 33137183 51 Gregory AC The Gut Virome Database Reveals Age-Dependent Patterns of Virome Diversity in the Human Gut Cell Host & Microbe 2020 28 724 740 e8 32841606 52 Guerin E Biology and Taxonomy of crAss-like Bacteriophages, the Most Abundant Virus in the Human Gut Cell Host & Microbe 2018 24 653 664 e6 30449316 53 Yutin N Analysis of metagenome-assembled viral genomes from the human gut reveals diverse putative CrAss-like phages with unique genomic features Nat Commun 2021 12 1044 33594055 54 Camarillo-Guerrero LF Almeida A Rangel-Pineros G Finn RD Lawley TD Massive expansion of human gut bacteriophage diversity Cell 2021 184 1098 1109 e9 33606979 55 Nayfach S Metagenomic compendium of 189,680 DNA viruses from the human gut microbiome Nat Microbiol 2021 6 960 970 34168315 56 Guo J VirSorter2: a multi-classifier, expert-guided approach to detect diverse DNA and RNA viruses Microbiome 2021 9 37 33522966 57 Roux S Minimum Information about an Uncultivated Virus Genome (MIUViG) Nature Biotechnology 2019 37 29 37 58 Koonin EV Yutin N The crAss-like Phage Group: How Metagenomics Reshaped the Human Virome Trends in Microbiology 2020 28 349 359 32298613 59 Roux S Krupovic M Poulet A Debroas D Enault F Evolution and Diversity of the Microviridae Viral Family through a Collection of 81 New Complete Genomes Assembled from Virome Reads PLOS ONE 2012 7 e40418 22808158 60 Bin Jang H Taxonomic assignment of uncultivated prokaryotic virus genomes is enabled by gene-sharing networks Nat Biotechnol 2019 37 632 639 31061483 61 Bichet MC Bacteriophage uptake by mammalian cell layers represents a potential sink that may impact phage therapy iScience 2021 24 102287 33855278 62 Mills S Movers and shakers: influence of bacteriophages in shaping the mammalian gut microbiota Gut Microbes 2013 4 4 16 23022738 63 Sutton TDS Hill C Gut Bacteriophage: Current Understanding and Challenges Front Endocrinol 2019 10 64 Mukhopadhya I Segal JP Carding SR Hart AL Hold GL The gut virome: the ‘missing link’ between gut bacteria and host immunity? Therap Adv Gastroenterol 2019 12 1756284819836620 65 Reyes A Wu M McNulty NP Rohwer FL Gordon JI Gnotobiotic mouse model of phage–bacterial host dynamics in the human gut Proc Natl Acad Sci U S A 2013 110 20236 20241 24259713 66 Guerin E Isolation and characterisation of ΦcrAss002, a crAss-like phage from the human gut that infects Bacteroides xylanisolvens Microbiome 2021 9 89 33845877 67 Callanan J Biases in Viral Metagenomics-Based Detection, Cataloguing and Quantification of Bacteriophage Genomes in Human Faeces, a Review Microorganisms 2021 9 524 33806607 68 Nandhra GK Normative values for region-specific colonic and gastrointestinal transit times in 111 healthy volunteers using the 3D-Transit electromagnet tracking system: Influence of age, gender, and body mass index Neurogastroenterology & Motility 2020 32 e13734 31565841 69 Martínez I Experimental evaluation of the importance of colonization history in early-life gut microbiota assembly eLife 2018 7 e36521 30226190 70 Yasuda K Biogeography of the Intestinal Mucosal and Lumenal Microbiome in the Rhesus Macaque Cell Host & Microbe 2015 17 385 391 25732063 71 Qin Y Molecular epidemiology and viremia of porcine astrovirus in pigs from Guangxi province of China BMC Veterinary Research 2019 15 471 31881886 72 Górski A Bacteriophage translocation FEMS Immunology & Medical Microbiology 2006 46 313 319 16553803 73 Sausset R Petit MA Gaboriau-Routhiau V De Paepe M New insights into intestinal phages Mucosal Immunol 2020 13 205 215 31907364 74 Majewska J Oral Application of T4 Phage Induces Weak Antibody Production in the Gut and in the Blood Viruses 2015 7 4783 4799 26308042 75 Meldrum OW Mucin gel assembly is controlled by a collective action of non-mucin proteins, disulfide bridges, Ca 2+ -mediated links, and hydrogen bonding Scientific Reports 2018 8 5802 29643478 76 Bolger AM Lohse M Usadel B Trimmomatic: a flexible trimmer for Illumina sequence data Bioinformatics 2014 30 2114 2120 24695404 77 Zolfo M Detecting contamination in viromes using ViromeQC Nat Biotechnol 2019 37 1408 1412 31748692 78 Nurk S Meleshko D Korobeynikov A Pevzner PA metaSPAdes: a new versatile metagenomic assembler Genome Res 2017 27 824 834 28298430 79 Li D Liu C-M Luo R Sadakane K Lam T-W MEGAHIT: an ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph Bioinformatics 2015 31 1674 1676 25609793 80 Altschul SF Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res 1997 25 3389 3402 9254694 81 Nayfach S CheckV assesses the quality and completeness of metagenome-assembled viral genomes Nat Biotechnol 2021 39 578 585 33349699 82 Hyatt D Prodigal: prokaryotic gene recognition and translation initiation site identification BMC Bioinformatics 2010 11 119 20211023 83 Terzian P PHROG: families of prokaryotic virus proteins clustered using remote homology NAR Genomics and Bioinformatics 2021 3 lqab067 34377978 84 Hockenberry AJ Wilke CO BACPHLIP: predicting bacteriophage lifestyle from conserved protein domains PeerJ 2021 9 e11396 33996289 85 Edgar RC PILER-CR: Fast and accurate identification of CRISPR repeats BMC Bioinformatics 2007 8 18 17239253 86 The Human Microbiome Jumpstart Reference Strains Consortium A Catalog of Reference Genomes from the Human Microbiome Science 2010 328 994 999 20489017 87 Mandal S Analysis of composition of microbiomes: a novel method for studying microbial composition Microb Ecol Health Dis 2015 26 27663 26028277 88 Nearing JT Microbiome differential abundance methods produce different results across 38 datasets Nat Commun 2022 13 342 35039521

PMC007xxxxxx/PMC7614366.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101688624 Sci Immunol Sci Immunol Science immunology 2470-9468 36706172 7614366 10.1126/sciimmunol.ade1413 EMS172514 Article T-independent responses to polysaccharides in humans mobilize marginal zone B cells prediversified against gut bacterial antigens Weller Sandra 1* Sterlin Delphine 23 Fadeev Tatiana 1 Coignard Eva 1 de los Aires Alba Verge 1 Goetz Clara 1 Fritzen Rémi 14 Bahuaud Mathilde 56 Batteux Frederic 56 Gorochov Guy 23 Weill Jean-Claude 1* Reynaud Claude-Agnès 1* 1 Université Paris Cité, INSERM U1151, CNRS UMR-8253, Institut Necker Enfants Malades (INEM), F-75015 Paris, France 2 Sorbonne Université, INSERM, CNRS, Centre d’Immunologie et des Maladies Infectieuses (CIMI-Paris), F-75013 Paris, France 3 Département d’Immunologie, Assistance Publique Hôpitaux de Paris (AP-HP), Hôpital Pitié-Salpêtrière, F-75013 Paris, France 4 School of Medicine, University of St Andrews, St Andrews, UK 5 Université Paris Cité, INSERM U1016, Institut Cochin, F-75014 Paris, France 6 Service d’Immunologie Biologique, Assistance Publique-Hôpitaux de Paris (AP-HP), Hôpital Universitaire Paris Centre (HUPC), Centre Hospitalier Universitaire (CHU) Cochin, F-75014 Paris, France * Corresponding author. sandra.weller@inserm.fr (S.W.); jean-claude.weill@inserm.fr (J.-C.W.); claude-agnes.reynaud@inserm.fr (C.-A.R.) 27 1 2023 27 1 2023 19 3 2023 27 7 2023 8 79 eade1413eade1413 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Marginal zone (MZ) B cells are one of the main actors of T-independent (TI) responses in mice. To identify the B cell subset(s) involved in such responses in humans, we vaccinated healthy individuals with Pneumovax, a model TI vaccine. By high-throughput repertoire sequencing of plasma cells (PCs) isolated 7 days after vaccination and of different B cell subpopulations before and after vaccination, we show that the PC response mobilizes large clones systematically, including an immunoglobulin M component, whose diversification and amplification predated the pneumococcal vaccination. These clones could be mainly traced back to MZ B cells, together with clonally related IgA+ and, to a lesser extent, IgG+CD27+ B cells. Recombinant monoclonal antibodies isolated from large PC clones recognized a wide array of bacterial species from the gut flora, indicating that TI responses in humans largely mobilize MZ and switched B cells that most likely prediversified during mucosal immune responses against bacterial antigens and acquired pneumococcal cross-reactivity through somatic hypermutation. pmcIntroduction B cell responses are classified as either T-dependent (TD) or T-independent (TI) based on the requirement for T cell help in antibody (Ab) production. TI antigens (Ags) are essentially nonproteins with repeated molecular structures [e.g., bacterial capsular polysaccharides (capPS) of Streptococcus pneumoniae] and generally induce strong B cell receptor (BCR) cross-linking responsible for rapid B cell activation, proliferation, and plasmablastic differentiation (1–4). While being independent of cognate T cell help, TI responses benefit from additional stimuli provided by noncognate T cells, Toll-like receptor agonists, cytokines, and/or signals from accessory cells, which are needed for class switch recombination (CSR) (4–9). In contrast to TD responses, TI responses are generally considered to be unable to generate fully developed germinal centers (GCs) or a memory response characterized by affinity maturation and a stronger, faster response of B cells to Ag rechallenge. Recent data, however, point to some forms of B cell memory in mice arising in response to different model TI Ags and the formation of rapidly collapsing GCs (10–12). In humans, Pneumovax is a vaccine composed of capPS from 23 pneumococcal serotypes that is considered a model TI Ag. Vaccination with Pneumovax generates serum responses lasting for up to 5 years in humans (13). However, secondary challenges with Pneumovax or other plain polysaccharidic vaccines do not generate an enhanced response and can even face a state of hyporesponsiveness (14–17). Children below 2 years of age and elderly people have poor or no Ab responses to TI Ags (18), and conjugated vaccines composed of polysaccharides linked to peptide moieties have been successfully developed to circumvent this unresponsiveness (19). Marginal zone (MZ) and B1b B cells are considered to be the main actors of TI responses in the mouse (20), where Abs against model TI Ags are mainly unmutated. In contrast, the anti-capPS Abs generated after vaccination with polysaccharidic vaccines are highly mutated in humans (21–23), being of immunoglobulin M (IgM), IgG, and IgA classes (mostly IgG2 and IgA2) (24, 25). The B cell subset involved in human TI responses is still a debated question, because there is no clear equivalent in humans of the mouse B1b subset (26), and splenic IgM+IgD+CD27+ B cells that have characteristics similar to mouse MZ B cells also display marked differences. They express mutated Ig genes and are able to recirculate in the blood (27, 28). Arguments in favor of their diversification outside classical TD responses came from studies in patients with genetic deficiencies impairing T-B collaboration and GC formation (29). In addition, IgM+IgD+CD27+ B cells are already diversified in infants without obvious marks of Ag-driven selection (30). Together, these findings led us to propose that blood IgM+IgD+CD27+ B cells are circulating splenic MZ B cells harboring a prediversified Ig repertoire (28). However, mutations in the Bcl6 gene as well as clonal relationships between blood IgM+IgD+CD27+ and switched B cells support the view that they are GC-derived memory cells that can participate in TD immune responses (31–33). A small subset of IgM-only B cells also exists, but IgD expression appears insufficient by itself to discriminate IgM memory B cells that might be involved in TD responses from those activated by TI Ags. Both subsets overlap in their expression of surface proteins commonly used to phenotype human B cells (27). In favor of a distinct human MZ lineage as in mice, we have previously identified a distinct precursor capable of differentiating into MZ-like B cells upon Notch2--triggering signals (34). In line with our result, recent work identified a bifurcation of human B cell maturation at the transitional 1 (T1) stage, giving rise to IgMlo and IgMhi T2 cells, the latter being selectively recruited into gut-associated lymphoid tissues (GALTs). The authors proposed a developmental continuum from the IgMhi T2 stage to MZBs, with a stage of repertoire diversification in the GCs of GALTs (35). Concerning the role of IgM+IgD+CD27+ in TI responses, evidence is largely indirect and mainly based on the correlation of their reduced presence in the blood of young children (<2 years), elderly, or immunodeficient individuals showing increased susceptibility to encapsulated bacteria and/or poor responses to plain polysaccharide vaccines (36–40). Along with these observations, clonal amplification after vaccination (28) and phenotypic analysis of pneumococcal polysaccharide-specific B cells (41) suggest a dominant role of IgM+CD27+ cells in the response to Pneumovax vaccination. However, the study of anti-capPS responses in a humanized severe combined immunodeficient (SCID)/SCID mouse model suggested that IgM+IgD+CD27+ B cells are not the sole contributors of TI immune responses (42). To further examine TI responses in humans, we took advantage of the vaccination of six healthy individuals with Pneumovax. Our aim was to identify the B cell subset(s) that is responsible for the production of anti-pneumococcal capPS IgM, IgG, and IgA in response to immunization. We observed that the pneumococcal vaccine generated a large plasma cell (PC) response mobilizing pre-diversified and preamplified clones systematically with an IgM component that could in most cases be traced back to MZ B cells, together with IgA and, to a lesser extent, IgG. Pneumococcal-specific clones showed recognition of a wide array of bacterial species from the gut flora, indicating that TI responses in humans largely mobilize MZ and switched B cells that prediversified during mucosal immune responses against commensals harboring cross-reactive glycans. Results A major but transient expansion of the plasmablast/plasma cell pool in blood 7 days after vaccination with Pneumovax To identify the B cell subset(s) from which anti-capPS–secreting plasmablasts (PBs)/PCs are derived, we isolated day 7–PBs/PCs (PB/PC-d7) along with different B cell subsets sorted before and at days (d) 7, 14, 28, and 56 after vaccination. Whereas high-throughput BCR repertoire sequencing (HTS) was done for PB/PC-d7 of all vaccinees, only three donors were further selected for HTS of their B cell subsets sampled at different time points (Fig. 1A and fig. S1, A and B). In the six vaccinees, Pneumovax induced a strong increase in the proportion of PBs/PCs that ranged from 3 to 19% of total peripheral blood B cells at d7 (average, 11.6%) (Fig. 1B). By d14, however, and for later time points, the proportions of PBs/PCs returned to values that were close to preimmunization levels for most donors. Pre- and post-immunization serum IgG and IgA Ab levels against seven capPS serotypes included in the vaccine were measured, with responses varying greatly among the six vaccinees (Fig. 1C and fig. S1C). Except for P02 (who had an already high preimmunization serum IgG level against serotype 14 and showed a lower overall response), all the vaccinees showed a significant increase in mean IgG and IgA levels after immunization, with at least a twofold increase of Ab levels for most or all the tested serotypes (Fig. 1C). Together, and as suggested by their large PB/PC expansion at d7, the six participants were all responders to the 23-valent vaccine. The PB/PC-d7 pool contains large clonal expansions PB/PC-d7 were bulk-sorted for each vaccinee, with the mean number of cells processed for RNA extraction being about 149,000 per sample (45,000 to 275,000) (data file S1). After HTS, the number of Ig sequences ranged from 112,000 to 218,000 (average of 153,000). The sequences were partitioned into clones with a clonal assignment performed on pooled sequences of the three isotypes (IgM, IgA, and IgG). The mean number of PB/PC-d7 clones was 5666 (range, 4947 to 6644). The D50 values (number of clones accounting for 50% of total sequences) ranged from 7 to 46 clones, with an average value of 28 clones, thus revealing large clonal expansions in the PB/PC-d7 repertoire (data file S1). The clones were ranked by size (according to their number of sequences), and the first 100 clones were defined as the “top 100,” which represented on average 70% of the total sequences (Fig. 2A and fig. S2, A and B). Among the top 100 PB/PC clones of each donor, the first 10 clones alone represented at least one-third of the total sequences, and for each vaccinee, the 100th clone had still more than a hundred sequences (mean, 192; range, 120 to 263). We thus assume that the majority of top 100 PB/PC-d7 had been mobilized in specific anti-capPS Ab responses. Most of the top 100–ranked PB/PC clones include IgM, IgG, and IgA isotypes with a high mutation load, but IgM sequences have a prominent representation We analyzed the isotype composition of the top 100 PB/PB-d7 clones of each vaccinee. Almost all the clones (from 97 to 100%, depending on the individual) contained IgM sequences, and a mean of 83% of the clones (range, 77 to 91%) were constituted of IgM, IgA, and IgG sequences present in various proportions (Fig. 2, B and C). Clones with a majority of IgM sequences were nevertheless the most numerous in the six vaccinees (Fig. 2D). The five most represented isotype combinations, found in 84% of the clones, all included IgM, either coexpressed with various sub-isotypes or alone (Fig. 2B). For each vaccinee, the relative proportion of sub-isotype sequences was calculated for the top 100 PB/PC-d7 clones (Fig. 2E). IgM accounted for half the sequences, whereas IgG2, IgA2, and IgA1 averaged 18 ± 4.4%, 17 ± 4.1%, and 10 ± 2.6%, respectively, with IgG1 and IgG3 being much less frequent. Together IgM was dominant among the sequences expressed by the top 100 PB/PC-d7 clones, but IgG and IgA were also coexpressed in a large fraction of them. For PB/PC-d7 clones, the mean mutation number per clone showed a wide distribution, with an average of 21 mutations per clone and very few unmutated ones (Fig. 2F). The top 100 PB/PC-d7 harbored a similar mutation load, but no unmutated clones could be observed. For those comprising IgM, IgA, and IgG sequences, the mean mutations per clone were calculated by isotype (fig. S2C). No significant differences were seen between the means of mutations per clone of IgA- and IgG-associated V-region sequences, both carrying on average ≍21 mutations, whereas, for four of the six donors, IgM sequences were slightly less mutated than IgG. We conclude that the high mutation load of the most expanded PB/PC-d7 clones implies that their direct B cell precursors were already mutated. Moreover, the prominent representation of mutated IgM sequences in the top 100 PB/PC-d7 clones suggests that such precursor B cell clones, whatever their expressed isotype, descended from a common prediversified IgM+ ancestor. Evidence of recent isotype switch events in the top 100–ranked PB/PC-d7 CSR occurs very rapidly after infection or immunization, during both TD and TI responses (43, 44). Because the majority of the top 100 PB/PC-d7 clones contained the three IgM, IgA, and IgG isotypes, we wondered if they may have undergone CSR recently, before their PB differentiation. We focused on the three vaccinees for which PB/PC-d7 had been also single cell–sorted. Among the unique sequences of the top 100 clones, we searched for strictly identical VDJ sequences associated with different constant regions, suggestive of recent switch events. Evidence of this was numerous, implying mainly switching from IgM to one or several downstream sub-isotypes, with the first three combinations, from IgM toward IgA1 and/or IgA2, accounting for 41% of the switch events (fig. S3A). For IgG sub-isotypes, CSR events leading to IgG2 were the most abundant (18.6%), well ahead of those leading to IgG1 (4.4%). Together, direct or sequential switching events to IgA1 or IgA2 were a majority, accounting for 20 and 55% of all events, respectively. Along with HTS, VDJ-μ, VDJ-α, and VDJ-γ transcripts of PB/PC-d7 sorted as single cells (in 96-well plates) were amplified and sequenced. We analyzed the proportion of wells in which identical VDJ transcripts were associated with two different isotypes (fig. S3B). No wells with a double IgM/IgG isotype were identified for any of the vaccinees. In contrast, a few percent of the single PB/PC-d7 had IgM/IgA, as well as IgG/IgA double transcripts, a substantial frequency considering the transient nature of simultaneous expression of two transcripts after isotype switching. For P03 and P06, HTS was done on bulk-sorted PB/PC-d14 cells. For each of the top 100 PB/PC-d7 clones, we identified (if present) the corresponding PB/PC-d14 clone belonging to the same clonal group. For d7-d14 clone pairs,the proportion of IgA sequences significantly increased at d14, whereas proportions of IgM and IgG sequences decreased in both donors (fig. S3C). This observation fits well with a preferential switch toward IgA sub-isotypes during the plasmablastic differentiation, with α-transcripts becoming predominant at d14. Another not exclusive possibility is that PBs/PCs originating from already switched IgA+ cells are more persistent in the blood at d14. Public clonotypes are shared between Pneumovax vaccinees and match previously described anti-pneumococcal capPS mAbs The repertoire of published anti-capPS monoclonal Abs (mAbs) shows a bias in VH gene usage for members of the VH3 family, the first five genes in the ranking being VH3-7, VH3-30/33 [frequently used by cell wall polysaccharide (CWPS)–binding Abs (45), VH3-23, VH3-48, and VH3-74 (fig. S4). We thus analyzed the repertoire of both PB/PC-d7 and top 100 PB/PC-d7 clones of donors P03, P05, and P06, for whom heavy chain (HC) sequencing of naive B cells at d0 was also performed. The frequency of VH gene usage, relative to total sequences, was calculated for each repertoire and showed a significant increase in VH3-7, VH3-23, and VH3-74 usage in both PC/PB groups, consistent with previous data on anti-pneumococcal cap PSmAbs (Fig. 3A). We also noticed the increased usage in the top 100 PB/PC-d7 clones of the VH4-59 and VH4-61 genes and especially of the VH6-1 gene. Beyond the observation of a repertoire bias, we searched for public clonotypes shared by two or more of the vaccinated donors (Fig. 3, B and C). Among the PB/PC-d7 clones of the six donors, we identified a total of 233 public clonotypes, most of them shared by two (82%) or three donors (14%). Seven clonotypes were shared by four individuals, and two clonotypes [whoseH-CDR3 sequences were close to a previously described anti-pneumococcal capPS mAb (23, 45, 46)] were shared by five vaccinees (Fig. 3D and data file S2). When the clonotypes were shared between 3, 4, and 5 donors, 45, 71, and 100% of them, respectively, were contributed by one or more of the top 100 PB/PC clones. We further broadened the comparison of anti-pneumococcal mAbs from the literature to all our PB/PC-d7 clones to identify additional shared clonotypes. We found 23 additional clonotypes that matched with known anti-capPS mAbs (Fig. 3D). Among the clonotypes that matched with anti-capPS mAbs both by their H-CDR3 size and amino acid sequences, seven of them expressed different VH genes, which suggests a prominent role of these particular H-CDR3s in defining certain anticapPS specificities (in these particular cases, against serotypes 2 and 4). Representative VH-VL pairs of largely expanded PB/PC-d7 clones show anti-capPS reactivity that is dependent on somatic hypermutation BCR-HTS of bulk-sorted PB/PC-d7 cells highlighted the presence of very large clones. For donors P03, P05,and P06, we amplified and sequenced VDJ-μ, VDJ-α, VDJ-γ, VJ-κ, and VJ-λ transcripts of PB/ PC-d7 cells isolated as single cells in 96-well plates. Among wells for which both productive VH and VL chains were obtained, we selected those whose sequence belonged to a top 100–ranked PB/PC-d7 clone. Corresponding VH and VL pairs (irrespective of the starting HC isotype) were expressed as IgG1 Abs. We produced a total of 28 mAbs (data file S3), of which originally 15 were IgA, 6 were IgG, and 7 were IgM. All were mutated, with on average 21.6 ± 9.5 mutations per VH (ranging from 7 to 40 mutations). Eighty-five percent of the mAbs were reactive against one of the 23 tested serotypes, with one mAb binding both serotypes 9V and 9N, and two capPS having very similar nonbranched structures. Together, a broad range of serotypes contained in the vaccine was recognized by mAbs from the three donors (Fig. 4A). These mAbs displayed relatively high affinity for their cognate Ag, with Kd (dissociation constant) values ranging from 0.8 × 10−9 to 176 × 10−9, with most of them having Kd in the nanomolar range (Fig. 4B). Four mAbs, however, did not recognize any of the tested polysaccharides, including CWPS. Noteworthy, three of these four mAbs without measurable reactivity against any tested capPS were originally IgM. It might be that their affinity for a given serotype could be too weak to be detected when expressed as a soluble IgG1 fraction as opposed to an IgM membrane form. Eight of the 28 mAbs were chosen to be reverted to their VH and VL germline (GL) configuration. After expression and purification, the GL versions tested at the same concentration (1 μg/ml) lost any measurable reactivity toward the capPS serotypes. Together, the majority of expressed mutated VH-VL pairs selected from the top 100 PB/PC-d7 clones bind with relatively high affinity to one of the Pneumovax serotypes, their anti-capPS reactivity relying on the presence of somatic mutations. Top 100 PB/PC-d7 clones are clonally related to large clones that exist before vaccination and include MZB, IgM-only, and switched IgG+ and IgA+ cells We sought to identify the subset from which top 100 PB/PC-d7 cells originated by examining their clonal relationships with different B cell subsets at d0. For P03, P05, and P06, BCR-HTS was done for eight different B cell fractions: naive, MZB, IgG+, IgA+, and IgM-only B cells from IgD−CD27+ subsets and IgG+, IgA+, or IgM+ from IgD−CD27− subsets [double-negative (DN) cells] (fig. S1B). We found no clonal relationships between the top 100 PB/PC-d7 clones and naive B cells in any of the vaccinees. No or very few relationships were found with DN cells either. Percentages of clonal relationships between PB/PC-d7 and single CD27+ B cell subsets were relatively low (from 2.3 to 5%), except for IgA+CD27+ cells (12% on average) (Fig. 5A). PB/PC-d7 clones were mainly clonally related to complex clonal entities encompassing MZBs and one or several other CD27+ members (IgM-only, IgA+CD27+, and IgG+-CD27+ cells) (fig. S5A). Less frequently, PB/PC-d7 clones were related to d0 clonal entities devoid of MZB cells, but including IgM-only and IgA+ and/or IgG+ cells. However, whereas IgM-only CD27+ is a minor population among sorted IgD−CD27+ cells (14% for P03 and P05, and 11% for P06) (fig. S1A), the Ig library preparation and indexing protocol followed for multiplex HTS resulted in a representation in sequence reads similar to other fractions (data file S1), which artificially increased the clonal relationships observed between PB/PC-d7 clones and the IgM-only subset. Because of the limited sampling (about 20 ml of blood were taken at each time point) and sequencing depth, it is important to note that not finding a clonal relationship does not imply that it does not exist. Conversely, finding relationships with clones present at d0 that did not undergo postvaccine blastic amplification implies that these clones were already large. In conclusion, the majority of PB/PC-d7 originated from preexisting clones encompassing MZB, IgM-only, and switched B cells, confirming that they necessarily descended from a common IgM+ Ag-experienced ancestor and that MZB and IgM-only B cell subsets harbor repertoires with largely overlapping clonal expansions. Moreover, the d0 B cell representation showed a strong TI signature involving 60 to 75% of IgG2 among the IgG isotypes (Fig. 5B). The large clones from which the top 100 PB/PC-d7 clones originate remain globally stable 2 months after Pneumovax vaccination Having shown that top 100– ranked PB/PC-d7 clones originate from large preexisting clones, including MZB, IgM-only, and switched B cells, we next followed these clones at 7, 14 (for P03 and P06), 28, and 56 days after vaccination (Fig. 5, C and D). We found numerous cases of clonal relationships between the subsampled PB/PC-d7 clones and all the other subsets (MZB, IgM-only, IgA+, and IgG+ cells) at all time points (Fig. 5C). Again, considering sampling limitations, this result emphasizes the large size of the preexisting clones mobilized by the Pneumovax vaccine and their persistence after vaccination. We evaluated the percentages of top 100 PB/PC-d7 clones related to MZB, IgM-only, and switched IgG+ or IgA+ clones at d0 and d56. For this calculation, an “All” category was defined for each B cell subset, a category including the clonal relationship with PB/PC-d7 of this subset either alone or in combination with others (Fig.5D). We assumed that, if the size of the pre existing clones that fueled the anti-capPS response did not markedly decrease 2 months after vaccination, the percentage of clonal relationships between these clones and their related PB/PC-d7 clones should remain roughly similar between d0 and d56. Although the results observed individually differ from patient to patient (patients P03 and P06 behaving globally the same and patient P05 showing opposite trends),on average,there was no consistent trend toward a decrease in the percentage of clonal relationships between top 100 PB/PC-d7 clones and the different subsets at d56 compared with d0. Accordingly, we propose that the observed variations (that are of the same order in all subsets) are too small to be compatible with the hypothesis of a strong depletion of the pool of the B clones consecutive to their mobilization in the anti-capPS PC response. This suggests a proliferative maintenance of the preexisting clones while they differentiate into PC. Somatic mutations do not increase after Pneumovax vaccination, and mutation trees of capPS-responding clones do not reveal a time-dependent pattern of mutation accumulation In a typical TD response, a time-dependent accumulation of somatic mutations can be seen, which results from selection, in GCs, for cells expressing higher-affinity Ab variants. Having found that top 100 PB/PC-d7 clones originate from large B cell clones that include already mutated MZB, IgM-only, and switched cells, we wondered if the load of somatic hypermutation increased in these B cell clones 2 months after vaccination. We thus selected clones encompassing top 100 PB/PC-d7 and for which we had sequences both at d0 and d56 for P03, P05, and P06. Because these clones were generally made of MZB, IgM-only, and switched cells, the mean of mutations (in VH sequences) per clone was calculated separately for each subset [sequences were pooled for switched (IgG + IgA) cells] sorted at d0 and d56. For the three donors before vaccination, the mean mutation levels of MZB cells were lower than those of IgM-only and switched B cells, with 18.8, 22.4, and 22.4 mutations/VH, respectively. Two months after vaccination, a slight increase in mutation level was observed for MZB cells of donor P03 and P06 and for switched (IgG + IgA) cells of donor P03, with differences together modest, whereas, in all the different subsets of P05 and remaining ones of P03 and P06, no significant increase of the mutation levels was observed at d56 (Fig. 6A). We further considered the B cell clones (made of MZB, IgM-only, and switched B cells) from which the top 100 PB/PC-d7 clones were derived, and we focused on those for which one PC-associated VH/VL pair had a validated anti-capPS reactivity. Starting from sets of clonally related Ig sequences, we constructed Ig lineage trees using the IgPhyML algorithm. Figure 6 (B and C) shows representative trees in which, with the exception of the VH sequence of the produced anti-capPS mAbs, PB/PC-d7 sequences are not shown. Starting from the “trunk” originating from an inferred ancestor, it appears that the sequences corresponding to cells from different subsets (MZB, switched IgA and IgG, and IgM-only) were distributed all over the trees, from the first forks and lowest branches to the highest branches and terminal leaves, independently of the time points at which they had been sampled (d0, d7, d14, d28, and d56). Cells sampled at d0 (see arrows on Fig. 6, B and C) often mapped on terminal leaves, where they could also be found in the close vicinity (i.e., attached to the same node) of cells sampled at the latest time points (d28 or d56). In line with this observation, the lineage trees revealed no clear accumulation of somatic mutations with time, with again many examples of d0 cells being more highly mutated than cells sampled at later time points, the reverse being true as well, with examples of d28 or d56 cells being less mutated than cells sampled before the vaccination. The depicted mutation trees represent clones that included one PC with validated anti-capPS specificity. It is clear that, depending on the impact of mutations, different members of the clone may or may not share this antigenic recognition. To address this question, we included PC sequences in the phylogenetic analysis and, assuming that they largely reflect the anti-capPS response, analyzed their positioning in the tree. Two different examples are shown in fig. S5 (B and C): In the tree depicted in fig. S5C, PCs are distributed all over the different branches, which could correspond to a widely shared anti-capPS specificity by most clone members, irrespective of their intraclonal diversification. In the tree depicted in fig. S5B, PCs (PCs present at d14) are much more clustered, which could correspond in contrast to a case where anti-capPS specificity was acquired or improved by somatic mutations. Together, these results support the conclusion that the process that generated this clonal diversification predated the pneumococcal vaccine response and that no major maturation occurred thereafter. Anti-capPS mAbs cross-react against gut bacterial strains belonging to different genera, with contrasted roles of somatic mutations We have shown that anti-capPS Abs are produced by PB/PC-d7 clones that differentiate from B cells belonging to large clones encompassing prediversified MZB, IgM-only, and switched B cells. Recently, a high-dimensional analysis has documented the possible diversification of MZ B cells in GCs of GALTs and their dissemination to widely distant sites via blood recirculation (47). We thus wondered if the anti-capPS mAbs that we obtained may also cross-react with bacterial structures present on the gut flora. To this end, we tested the binding of 25 mAb (expressed as IgG1) to three different fecal microbiota obtained from healthy individuals by a flow cytometry assay (Fig. 7, A and B) (48, 49). Two mAbs [06(2)-E6 and 06(1)-C6] had no detectable binding to any of the microbiota tested, and two others (05-F10 and 05-C6) bound only and very weakly to microbiota II, the percentage of labeled bacteria being nevertheless more than three times above the observed value for the isotype control. All the remaining mAbs (i.e., 84%) bound to a small but substantial fraction of one or more of the tested microbiota, fractions whose values ranged from 0.1 to 6% of the total bacteria. To gain more insights into the commensal bacteria bound by the mAbs, we selected eight of them (identified by an asterisk in Fig. 7 and data file S3 and available in their mutated or GL version), and we sorted by fluorescence-activated cell sorting (FACS) the respective mAb+ and mAb− microbiota fractions for 16S ribosomal RNA (rRNA) gene sequencing (Fig. 7B). The sequencing (successfully completed for seven mAbs) showed that they all recognized bacteria from the four major phyla, namely, Actinobacteria, Bacteroidetes, Firmicutes, and Proteobacteria, but each one exhibited a distinct binding profile (Fig. 7C). The mAbbinding pattern was further characterized by calculating a log-based enrichment index (EI) at the genus level. All mAbs were cross-reactive against a median of 12 genera among the 100 most abundant ones (range, 6 to 21) (Fig. 7D) and thus recognized a wide panel of bacteria. EIs varied among the genera targeted by the mAbs, indicating that they might bind distinct genera with different affinities. Because each of the mAbs may potentially target a unique combination of bacteria belonging to different genera, or may also recognize more species within a genus, we searched for representative bacterial strains that are both cultivable and available (see Materials and Methods). Six mAbs were subsequently evaluated for their binding to isolated strains (fig. S4C), and three of the mutated mAbs showed a low to high binding to specific bacterial strains (Fig. 7E), thus confirming the validity of some of the mAb targets identified by 16S rRNA. No binding was detected for the three remaining mutated mAbs, a result that may rely on the large number of different species present among each genus [e.g., with regard to the Lactobacillus genus only, at least 50 species have been repeatedly detected in the stool of humans (50)]. As mentioned above, the eight mAbs, when reverted to their VH-VL GL configurations, lost their anti-capPS reactivity (data file S3). We thus compared in the same FACS flow experiment the binding of the mutated or GL version of each of the eight mAbs to the same microbiota (I, II, or III) that was used for the 16S RNA sequencing experiment. Whereas the percentages of bacteria stained by GL compared with the mutated mAbs were on average lower (1 ± 0.5% versus 2 ± 1.3%, the difference being not significant), the results were nevertheless heterogeneous: One GL mAb lost almost all its reactivity toward the microbiota, four mAbs showed a marked decrease in their binding, one mAb was unaffected, and two GL mAbs stained a larger bacterial fraction than their mutated counterparts. Among the mAbs tested for their binding to isolated bacterial strains (Fig. 7E), the results were again complex because GL mAb 06(1)-B7 and 03(4)-E04 did not recognize the Staphylococcus and/or Lactobacillus species anymore, whereas GL mAb 06(2)-D11 gained some reactivity against Bifidobacterium species and one Peptococcus sp. Together, these results illustrate that several GL mAbs were already reactive against the gut microbiota, whereas all required further diversification to recognize capPS Ags. Anti-capPS mAbs are mostly non-polyreactive In a recent work, it was shown that the cross-species binding and high microbiota reactivity of human intestinal IgA did not correlate with Ab polyreactivity (51). We thus screened our 24 anti-capPS mAbs for binding to a panel of four unrelated Ags commonly used to test for polyreactivity (52, 53). As shown in fig. S7, only two of them showed moderate binding to more than three Ags and were thus classified as polyreactive. Together, these findings strongly suggest that the specific binding of our anti-capPS mAbs to the microbiota samples is mediated by gut bacterial glycans. Discussion Whereas MZ and B1 B cells are responsible for responses to TI Ags in the mouse, the identity of the B cell subset involved in similar responses in humans remains debated. The notion that IgM+IgD+CD27+ B cells may represent the functional and developmental equivalent of the mouse MZ B cell lineage is still controversial, despite their widely accepted name of MZB cells, and observations in favor of their contribution to TI responses are largely based on correlative evidence. We therefore vaccinated healthy volunteers with Pneumovax, a plain polysaccharide vaccine, and compared the repertoire of d7 PBs/PCs with the repertoire of various B cell compartments analyzed before and after vaccination (with a longitudinal follow-up of up to 2 months) to identify the subset(s) from which the PC amplification originated and the possible repertoire modifications induced in the B cell populations engaged in this response. The strong d7 PB/PC response in blood mobilized very large clones, whose size and Ig mutation level can only be accounted for by the engagement of a preamplified and prediversified B cell population, clones that frequently included a major IgM component together with clonally related IgA (IgA2 more frequently than IgA1) and IgG (mostly IgG2) PCs. Clonal relationships with B cell subsets present in blood before immunization revealed a high frequency of clones present in the MZB and IgA compartments, as well as, but to a lesser extent, IgG. A similar frequency of clonal relatedness was observed between the PB/PC-d7 and the blood CD27+ B cell subsets before and 2 months after vaccination, indicating, at least for this population of young adults (18 to 40 years of age), that the large PC response did not exhaust preexisting clones. Similarly, mutation frequencies of clones observed at these two time points were similar, again supporting the notion of an anti-capPS response taking place through extrafollicular amplification of Ag-experienced cells without contribution of affinity maturation. However, it cannot be excluded that a few additional mutations did occur during the switching from IgM to the other isotypes (9). Nevertheless, in all cases of capPS-specific mAbs analyzed, the mutations of their VH and VL genes appeared mandatory for their specificity. No changes in isotype distribution among CD27+ B cells were either observed with time, whereas, in contrast, the frequency of IgA+ cells among PCs increased at d14 in the two cases where PCs could be analyzed up to this stage, suggesting, together with the identification of single cell–sorted PCs at d7 harboring two isotypes, that isotype switch was occurring during clonal expansion, leading to PC differentiation. Cloning and reexpression of several Abs expressed by the major PB/PC clones and showing specific recognition of distinct pneumococcal capPS serotypes revealed in most cases a clear binding to components of the gut microbiota. Isolation and identification through 16S sequencing of the bacterial species recognized revealed the capacity of these anti-pneumococcal Abs to bind to diverse types of phyla and genera. Moreover, because the majority of the anti-capPS mAbs (more than 90%) did not harbor any polyreactivity, this binding most likely proceeded through recognition of cross-reactive bacterial glycans. In line with this, it was shown that a reduction of circulating MZ B cells was correlated with a drop in binding to bacterial glycans (39). Although GL-reverted anti-pneumococcal Abs lost their specificity, the impact on their recognition of the gut flora was much more heterogeneous, with binding enhanced, decreased, or maintained upon GL reversion. In a similar experiment, gut IgA Abs were shown to lose their bacterial specificity when reverted to their GL configuration (51). Affinity maturation during gut responses is thus responsible for acquisition of pneumococcal cross-reactivity. Public clonotypes and preferential VH gene usage have been linked to pneumococcal anti-capPS recognition, and we observed similar repertoire convergence among our six vaccinees, as well as with previously published repertoire data. Such public specificities suggest an extensive sharing among individuals of cross-reactive bacterial glycans that elicited and selected these clonotypes. Overall, work thus far supports that diversification against gut-associated Ags, notably glycans, establishes a peripheral IgM B cell compartment, together with clonally related IgA and IgG2 subsets characteristic of a TI signature, with IgG2 representing 60 to 75% of the IgG sub-isotypes. A similar TI polarization is observed upon vaccination during the differentiation of MZ B cells into IgA and IgG PCs. No clear distinction could be made between MZB and IgM-only CD27+ B cells, this latter subset being much less frequent, suggesting that their repertoire may largely overlap, at least among B cell fractions sorted on the basis of IgD and CD27 surface expression. Moreover, IgA+CD27− B cells have been shown to display some antibacterial binding properties (54), but very few clonal relationships were observed in our study between the PC pool mobilized in the pneumococcal response and the DN compartment. The stability over time of large IgM, IgA, and IgG2 clones circulating in blood has been described recently in a longitudinal follow-up of two healthy donors, showing distinct characteristics from smaller IgG1 clones specific for TD vaccines, and our data fully agree with this observation (55). The concept of a prediversified MZ B cell repertoire obviously raises the question of the origin and location of such a prediversification step. A recent study reported that MZ B cell precursors differentiated and diversified their BCRs in GCs from GALTs (35). Along with our results of acquisition of capPS recognition through somatic mutations, this suggests a model in which, after an initial commitment at the MZP/T2 stage, MZ B cells can mature their BCRs by somatic hypermutation in GALT, in a process driven by gut commensals. Although less likely in normal physiological conditions, some of this process could also occur in the periphery upon translocation of gut bacterial Ags (56). Moreover, because these responses take place in the specific environment of the gut Peyer’s patches and unexpectedly harbor a typical IgG2 and IgA TI isotypic profile, this would suggest a diversification step triggered by noncanonical T cell help (57). In addition, one cannot formally exclude that asymptomatic carriage of S. pneumoniae may have triggered these diversified MZ B cells (58), but we find this unlikely to have occurred in all six individuals analyzed in this study. Moreover, because immune responses against other types of encapsulated bacteria also mobilize B cells with mutated BCR (59), it would imply the improbable scenario that such silent carriage always occurs. There remain several questions on the physiology of human MZ B cells particularly concerning their homeostatic regulation and recall properties. TI vaccinations in humans induce the production of protective Abs but do not generate an enhanced Ab response upon a boost, evoking the absence of functional memory. These TI vaccines are inefficient in children below 2 years, and it is generally assumed that this is due to the immaturity of the splenic MZ constituents. Our results on the necessity of a fully diversified repertoire to recognize the pneumococcal serotypes would suggest, as an alternate explanation, an incomplete maturation of the Abs involved. At this young age, the pneumococcal conjugated vaccine gives rise to a classical TD immuneresponse including affinity maturation and B cell memory. Paradoxically, if given in older infants and adults, this conjugated formulation induces, rather than a memory response, a response somehow similar to that obtained with the plain TI vaccine, even if in some reports it seems more robust (60). We propose that the functional splenic MZ B cells could sequester some of these capPS whether they are conjugated or not to a carrier protein, thus preventing them from reaching naive follicular B cells. It would be therefore very instructive to study in a similar way a group of adult participants receiving the conjugated pneumococcal vaccine (Prevnar) to evaluate the B cell repertoire this vaccine mobilizes. Our study was obviously limited to a small number of volunteers and we were not able to perform analyses beyond 56 days, which would be informative regarding the stability and the persistence of the expanded clones observed. The general relevance of our observations should also be extended to other types of polysaccharide vaccines, for example, against Neisseria meningitidis. In conclusion, we propose that MZ B cells generate a prediversified B cell repertoire in GALT that is driven by commensal bacterial Ags, which thereafter enable the rapid production of opsonizing Abs against highly pathogenic encapsulated bacteria. Many pathogens such as HIV are covered with glycans, which are often the main epitopes recognized by specific neutralizing Abs (61, 62). It is tempting to speculate that a vaccine containing only these glycans may trigger a fast and efficient MZ B cell response in humans (63). Evolutionarily, these results suggest that humans have retained, within a subset of B cells, the strategy used by several species such as chicken, sheep, and rabbit to diversify their pre-immune B cell repertoire through hypermutation and/or gene conversion in GALT to subsequently respond to external challenges (64). Materials and Methods Study design To identify the B cell subset(s) from which anti-capPS–secreting PBs/PCs are derived, and given that the number of serotype-specific PBs/PCs was shown to peak around d7 upon immunization with Pneumovax (65, 66), our strategy was based on the sorting and BCR repertoire sequencing of PB/PC-d7 and of different B cell subsets isolated before and after vaccination. Whereas sequencing was done for PB/PC-d7 of all vaccinees, only three donors (for which higher number of cells could be FACS-sorted) were further selected for HTS of their B cell subsets sampled at different time points. We took advantage of the unique Ig HC VDJ signature expressed by each B cell clone to search for clonal filiations between anti-capPS PB/PC-d7 clones and isolated B cell subsets. Because a direct isolation of anti-capPS PBs/PCs was not possible, likely anticapPS VDJ signatures were first validated through the expression of the corresponding mAbs and their testing against the 23 capPS of the vaccine. Through ad hoc amplification of VDJ-μ, VDJ-α, and VDJ-γ transcripts of the sorted cell fractions and HTS sequencing, the clonal relationships between anti-capPS PBs/PCs and eight different B cell subsets (MZB cells, CD27+ or CD27– IgG+, IgA+, and IgM-only B cells, and last naive B cells) were investigated. Last, we determined the reactivity of pneumococcal-specific mAbs against bacterial species from the gut flora through bacterial flow cytometry assays and 16S rRNA gene sequencing of the mAb-bound microbiota. Human Pneumovax immunization and sample collection The study was approved by the Institutional Review Board #00001072 (CPP Ile-de-France II). All the participants gave written informed consent. For details on the Pneumovax-23 vaccine (named Pneumovax throughout this study) and participant inclusion criteria, see the Supplementary Materials. The final six participants are referred as P02, P03, P05, P06, P07, and P08. Serum was obtained before and after vaccination (d0 and d28), and heparinized blood samples (about 25 ml) were collected before vaccination (d0) and at d7, d14, d28, and d56 after vaccination. Multicolor flow cytometry analysis and cell sorting Peripheral blood mononuclear cells were isolated by centrifugation on Ficoll and surface-stained for 20 min at 4°C in phosphate-buffered saline (PBS) + 2% fetal calf serum, with Abs listed in Supplementary Materials and Methods. The detailed gating strategy is depicted in fig. S1. FACS-sorted B cell fractions were collected in Eppendorf tubes filled with PBS + 2% fetal calf serum. After centrifugation, the cell pellets were resuspended in 75 μl of reverse transcriptase lysis buffer and RNA extraction was done using the RNeasy Kit (Qiagen). PB/PC-d7 clones were also single cell–sorted into 96-well polymerase chain reaction (PCR) plates (Bio-Rad) containing ice-cold lysis buffer (4 μl per well) as described in (67). Library preparation and Illumina sequencing The library preparation for BCR-HTS, designed for paired-end Illumina sequencing, was done as described in (68). Briefly, the method allows the introduction, during the synthesis of double-strand (ds) IgHV-D-J-CH cDNA, of unique molecular identifiers (UMIs) used to reduce PCR amplification biases and sequencing error rate in downstream IgH sequencing analysis. The synthesized ds cDNA was then subjected to a first constant-region multiplex PCR using a mix of CH isotype-specific reverse primers. The resulting PCR product was then used in separate seminested PCRs, with respective inner CH-μ, CH-α,and CH-γ primers (including Illumina adaptor sequences). For more details on library preparation and sequencing, see the Supplementary Materials. Bioinformatics analysis of IgH sequencing data UMI Barcoded MiSeq 2×300 reads were processed as described in (68) and in the Supplementary Materials. Clonal assignment and GL reconstruction were performed with Change-O toolkit (69) on sequences having at least two representative reads for PB/PC-d7 and on all sequences for other populations. Clonal assignment was based on the following criteria: sequences that had the same V-gene, J-gene, and junction length, with maximal nucleotide hamming distance in CDR3 of 0.06 for PB/PC-d7 population, 0.15 for switched population, and 0.1 for the other subsets (the optimal threshold determined with the SHazaM R package) (69). For B cell populations (other than PB/PC-d7), clones were assigned by isotype fraction. All the details for the calculation of total sequence number by clone and ranking of PB/PC-d7 clones (Fig. 2 and fig. S2), isotype switch event quantification in top 100 PB/PC-d7 (fig. S3A), and calculation of the proportion of IgM, IgA, and IgG sequences relative to the total of sequences in each PB/PC-d7–d14 clone pair (fig. S3C) are given in the Supplementary Materials, as for the search for public clonotypes among PB/PC-d7 clones of the six vaccinees (Fig. 3). Regarding the public clonotypes shared between the vaccinees of this study, the nucleotide sequences encoding the corresponding H-CDR3 were verified (data file S2) to dispel any PCR contamination issue. For each vaccinee, we performed clonal relationship analysis between PB/PC-d7 clones having more than 100 sequences and all other populations. The clones having the same V-gene, J-gene, junction length, and minimal hamming nucleotide distance less than 0.15 were considered the same clone. Accordingly, for subsequent analysis, we added to every PB/PC-d7 clone the corresponding clones from MZB, switched IgA, and switched IgG. The anti-PC sequences obtained on a single-cell level were also added to corresponding clones. Lineage trees were built on the basis of the maximum likelihood phylogenetic method using the IgPhyML tool (70, 71). For visibility reason, we subsampled all large fractions to 10 sequences by fraction. When tracing MZB-d0 clones at other time points in Circos plots, we chose only clones related to the top 50–ranked PB/PC-d7 clones and showed for each such clone only one instance of relation, if found, to other populations. Recombinant Ab production and purification Starting from single PB/PC-d7 sorted into 96-well PCR plates, single-cell reverse transcription PCR (RT-PCR) was performed and followed by Ig-μ, Ig-γ, Ig-λ, and Ig-κ gene PCR amplifications as described (67). An additional 3'CH1-alpha primer (5′-GAGTGGCTCCTGGGGGAAGA-3′) was designed to amplify also Ig-α chains. Paired IgH and IgL chains from clones of interest were cloned in γ1 HC and κ or λ light chain (LC) expression vectors (67) and transfected into human embryonic kidney (HEK) 293T cells using a jetPRIME kit (Polyplus). Recombinant Abs were purified from supernatants using protein G HP SpinTrap (GE Health-Care) and subsequently dialyzed overnight into PBS [0.01 M sodium phosphate (pH 7.4) and 0.137 M sodium chloride]. To obtain the GL-reverted versions of eight mAbs (marked by anasterisk in data file S2), the GL genes corresponding to each VH-D-JH and Vλ-Jλ (or Vκ-Jκ) pair were synthetized with proper restriction sites (and a codon optimization) and cloned into the γ1 HC and κ or λ LC expression vectors by ProteoGenix (France), transfected, and purified as described above. Anti-polysaccharide ELISA and affinity measurement Pre-and post-immunization serum IgG concentrations for seven pneumococcal serotypes (4, 6B, 9V, 14, 18C, 19F, and 23F) were measured as described in (72). Purified recombinant mAbs were tested against each of the capPS of Pneumovax. Briefly, 96-well plates were coated with individual capPS. After blocking of wells, purified mAbs diluted at 1μg/ml in enzyme-linked immunosorbent assay (ELISA) buffer containing CWPS (5 μg/ml) were added to the plates. Revelation was done with incubation with horseradish peroxidase–conjugated goat anti-human IgG followed by addition of KPL SureBlue TMB peroxidase substrate, after washing steps. The reaction was stopped by H2SO4, and optical density was measured at 450 nm. Background values given by incubation of PBS alone in coated wells were subtracted. All mAbs were tested in duplicate in two independent experiments. Ab affinities (Kd) were determined as described in (45) by curve fitting analysis of individual ELISA curves plotted from a dilution series of Ab beginning at 10 μg/ml. Each mAb was tested in duplicate in two experiments whose results were then averaged to obtain the Kd reported in data file S2 (see the Supplementary Materials for details). Polyreactivity ELISA mAbs were tested at four consecutive 1:4 dilutions (starting from 1 μg/ml) against dsDNA, keyhole limpet hemocyanin, lipopolysaccharide, and insulin as previously described (52). Abs were considered polyreactive when they recognized at least three structurally different Ags in two distinct experiments. Threshold values for reactivity were determined by using control Abs mGO53 (negative) (53) and ED38 (high positive). Bacterial strains and fecal bacteria The different isolated bacterial strains (Fig. 7 and fig. S6) were obtained from the Saint-Antoine Hospital (Paris),the Necker-Enfants Malades Hospital (Paris), the INRA (Jouy en Josas), and the American Type Culture Collection strain collections. Their culture conditions are described in the Supplementary Materials. The fecal bacteria from three healthy individuals were purified by gradient purification under anaerobic conditions as described in (48, 73) and in the Supplementary Materials. Bacterial flow cytometry mAb binding to microbiota or bacterial strains was evaluated as previously described (49, 74). All buffers were passed through sterile 0.22-μm filters before use. Thawed microbiota or bacterial strains (106 per condition) were fixed in 500 μl of paraformaldehyde solution (4% in 1× PBS) and stained with Cell Proliferation Dye eFluor 450 (eBioscience) for 25 min at 4°C. After washing with 1× PBS (10 min, 4000g, 4°C), bacteria were suspended in 1× PBS, 2% bovine serum albumin (Sigma-Aldrich), and 0.02% sodium azide (Sigma-Aldrich). mAb or human monoclonal anti–tumor necrosis factor–α (TNF-α) IgG1 (REMICADE; MSD France) was added at a final concentration of 1 μg/ml and incubated for 30 min at 4°C. After washing, goat anti-human IgG–Alexa Fluor 647 or isotype control (both from Jackson ImmunoResearch Laboratories) was incubated for 20 min at 4°C. Then, bacteria were washed and resuspended in sterile PBS. Samples were run using BD FACSCanto II. Analysis was performed with FlowJo software (TreeStar). Medians of fluorescence were used to measure mAb-binding levels for pure strains. Sorting of IgG-bound microbiota Gut microbiota were incubated with purified mAbs (at 1 μg/ml) in a 96-well V-bottom plate (106 bacteria per well, 10 wells per mAb) for 30 min at 4°C. After washing in 1 × PBS, live microbiota were stained as described above. Then, sorting was performed using a microbiota-dedicated single laser S3 cell sorter (Bio-Rad Laboratories, USA). Sorted bacteria (9 × 105) were collected in 1× PBS at 4°C, centrifuged (8000g, 10 min, 4°C), and stored at −80°C until DNA extraction. Purity for both fractions was systematically verified after sorting. To check the absence of contaminants in flow cytometer fluid lines, we regularly incubated sheath fluid in Brain-Heart Infusion Broth (BioMérieux) at 37°C for 7 days. 16S rRNA gene sequencing and analysis DNA extraction, generation of amplicons of V3-V4 regions of 16S rRNA genes, and multiplexed sequencing (MiSeq, Illumina) were done as described in (49) and in the Supplementary Materials. De-multiplexed paired-end 250 nucleotide reads were processed using MG-RAST analysis pipeline. Sequencing artifacts, host DNA contamination, and sequences fewer than 200 base pairs in length were removed. Insufficient quality reads were discarded (<5% of total reads). Sequences were then clustered into operational taxonomic units (OTUs) with a 97% homology using Greengenes database. OTUs containing only a single sequence were discarded. Previously published contaminant sequences (75) were removed if present in only one sorted fraction and absent from paired fractions. OTUs detected at >0.1% relative abundance in at least two samples were finally conserved. The OTU table was rarefied to the minimum sample’s depth (24,785 reads). In calculating the EI, we scored a pseudo-relative abundance equal to 0.0001, which was the lower limit of detection, if a taxon was not detected in a given fraction. Statistical analyses Statistics were calculated using GraphPad Prism (v9) and R (version 4.0.4). When a statistical analysis was done on data presented in a given figure, the test used is specified in the corresponding figure legend. Supplementary Material Fig S1-S7 Acknowledgments We thank J. Mégret (Cell Sorting Facility of the SFR Necker) for cell sorting, D. Bagnara for sharing expertise on BCR HTS, P. Villarese and the McIntyre/Asnafi team for access to an Illumina MiSeq sequencer, H. Mouquet and C. Planchais for the gift of mAbs ED38 and MOG53 and helpful discussions, C. Parizot for help in bacterial flow cytometry, S. Storck and P. Chappert for helpful discussions and support, and H. Lecuyer for the gift of bacterial strains. Funding This work was supported by a Fondation Princesse Grace grant and by an ERC Advanced Grant (B-response) awarded to J.-C.W. and C.-A.R. Data and materials availability The immunoglobulin repertoire sequencing data are available under ArrayExpress accession code E-MTAB-12117. All other data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. Fig. 1 A strong but transient PB/PC response in blood at day 7 after Pneumovax vaccination. (A) Study design and sample collection (see fig. S1). (B) Percentage of PBs/PCs relative to CD19+ B cells before and at different time points after vaccination with Pneumovax of six individuals. (C) Fold changes in serum levels of serotype-specific anti-capPS IgG and IgA between d0 and d28 after Pneumovax vaccination. The dashed line indicates a fold change of 2 (see fig. S1C). Fig. 2 High-throughput sequencing of the Ig repertoire of PB/PC-d7 cells after Pneumovax vaccination shows large clonal expansions, a high load of somatic hypermutations, and a representation of μ, α, and γ isotypes in the majority of the top 100–ranked clones. (A) Percentages of sequences (relative to the total number of PB/PC-d7 sequences) represented by each of the top 100 PB/PC-d7 clones for each vaccinee (see fig. S2A). (B) Combinations of IgH sub-isotypes expressed by PB/PC-d7 clones. The analysis is done on pooled top 100 PB/PC-d7 clones of P03, P05, and P06. The number on top of each bar represents the number of clones, with the isotype combination indicated below. The total percentage of the five most frequent combinations is indicated above the corresponding histogram bars. (C) Ternary plots depicting the relative proportion of IgA, IgG, and IgM sequences in the top 100 PB/PC-d7 clones of each vaccinee. Each black dot corresponds to one clone. A small ternary plot below shows a theoretical example of a clone with its projected coordinates on the three axes, indicating the proportion of each isotype. (D) Number of PB/PC-d7 clones, among the top 100 clones, that have a majority of IgM, IgA, or IgG sequences. (E) Sub-isotype representation in the top 100 PB/PC-d7 clones of each vaccinee. The relative percentages of sub-isotype sequences were calculated per clone (to allow weighting to clone size) and averaged for the top 100 PB/PC clones. (F) Scatterplots representing the means of Ig-VH mutations per clone for the totality (left) or for the top 100 PB/PC-d7 clones (right) of each vaccinee. Black (or gray) lines and error bars represent means ± SD. Fig. 3 Occurrence of public clonotypes in the repertoire of PB/PC-d7 that are shared by two or more Pneumovax vaccinees, some of these clonotypes being expressed by previously described anti-capPS mAbs. (A) Frequency of VH gene usage (relative to the total sequences of each subset) for naive (at d0), PB/PCd7, and top 100 PB/PC-d7 (pooled sequences from P03, P05, and P06). Only informative VH genes for the comparison between naive and PB/PC-d7, or between naive and top 100 PB/PC-d7, were plotted, i.e., VH genes either absent or representing less than 0.005% in two of three of the comparison groups were filtered out. Significance was calculated with Fisher’s exact test with a false discovery rate correction. P values are indicated only for VH genes that are significantly more expressed in PB/PC-d7 and/or top 100 PB/PC-d7 versus naive B cells. The color of the lines on top of the histogram bars indicates the comparison groups: red, naive versus PB/PC-d7; orange, naive versus top 100 PB/PC-d7; black, PB/PC-d7 versus top 100 PB/PC-d7. *P < 0.05; **P < 0.005; ***P < 0.0005. (B) Percentages of public clonotypes among the total number of PB/PC-d7 clones. Numbers on top of each histogram bar represent the number of clonotypes. (C) Proportion of public clonotypes, among those identified in the six vaccinees, that were shared by two, three, four, or five vaccinees. (D) Characteristics of the public clonotypes that matched with published anti-capPS mAb. The left part illustrates the sharedness of the clonotypes between the six vaccinees, each column corresponding to one of them, as indicated. The color of the dots in the boxes indicates the rank of the PB/PC-d7 clone associated to the clonotype whose characteristics are indicated on the right (dark blue: rank ≤100; light blue: 101 < rank ≤ 200; gray: rank > 200). Characteristics of the published mAb are indicated in the far right part of the table. “Pool” refers to the cases where a pool of capPS was used to test the specificity of the mAbs/single-chain fragment variable in the cited articles (see fig. S4). *, clonotypes matching with anti-capPS mAbs by both their H-CDR3 size and amino acid sequences, but expressing different VH or JH genes. Fig. 4 Representative VH-VL pairs of largely expanded PB/PC-d7 clones show anti-capPS reactivity that is dependent on somatic hypermutation. (A) Pneumococcal serotypes recognized by the mAbs obtained from vaccinees P03, P05, and P06. Each mAb recognized only one serotype (except for 9N/9V). (B) Dissociation constants (Kd, expressed as moles per liter) of the 28 expressed mAbs against their specific serotypes (see data file S2). Red dots correspond to mAbs for which a GL version of the VH/VL pair was expressed. (The two half red circles correspond to the mAb that binds to serotype 9N and 9V with different Kd.) mAbs with no measurable affinity were considered nonbinders (Kd ≥ 10−5). Fig. 5 The top 100 PB/PC-d7 clones are clonally related to large preexisting clonal entities, including MZB, as well as IgM-only, IgG+, and IgA+ CD27+ B cells that remain stable over time. (A) Percentage of top 100 PB/PC-d7 clones having clonal relationships at d0 with only one subset or with several subsets. The categories “MZB & one or more subset”and “IgMonly & one or more switched subset”include six and three combinations, respectively (see tables on the right of the graph).“+”means that a PB/PC-d7 clone is clonally linked to a given B cell subpopulation. (B) Isotype representation in the switched IgG and IgA d0 clones that are clonally related to the top 100 PB/PC-d7 clones. The relative percentages of sub-isotype sequences were calculated per clone (to allow weighting to clone size) and averaged. (C) Clonal overlap of d0 MZB clones that are related to the largest PB/PC-d7 clones with all other subsets at different time points, for P03, P05, and P06. For clarity, only the top 50 PB/PC-d7 clones that have clonal relationships with MZB cells at d0 (MZB-d0) were taken into consideration in this analysis, and their relationships with the different subsets are illustrated through an “MZB-d0-centric view” (PB/PC-d7 subsets are not depicted on the plot; see fig. S5). Accordingly, the many intersubset clonal relationships (e.g., between switched IgA, IgG, and IgM-only cells sampled over time) are not shown. (D) Percentage of top 100 PB/PC-d7 clones having clonal relationships with MZBs (All), IgM-only (All), switched IgG (All), or switched IgA (All) B cells at d0 and d56. Each “All”category includes eight possible combinations (see tables on the right of the graph). Fig. 6 Somatic mutations in clones related to the top 100 PB/PC-d7 clones do not increase 2 months after Pneumovax vaccination, and Ig mutation trees do not reveal a time-dependent pattern of mutation accumulation. (A) Mean of mutation numbers (in VH sequences) per clone for clone pairs present at both d0 and d56 in the MZB, IgM-only, and switched (IgG + IgA) subsets, respectively. Dashed lines represent the mean mutation number in VH sequences of the naive cells of each donor. Significance was calculated with the Wilcoxon matched-pairs two-tailed signed-rank test (*P ≤ 0.05; ns, not significant). (B and C) Representative Ig lineage trees of two B cell clones with anti-capPS specificities. The trees were built, using the IgPhyML algorithm, with a subsampling of 10 sequences (or fewer if not available) by B cell subset and by time point. The black symbols correspond to the sequence of the anti-capPS mAb that was isolated from single cell–sorted PCs (see data file S3). The gray rectangle indicates the inferred common progenitor. Each lineage tree has been duplicated to indicate the number of mutations per VH of each sequence. (Arrowheads mark examples of cells sampled at d0 but mapped on terminal leaves.) Fig. 7 Anti-capPS mAbs cross-react against gut bacteria from various genera and show a largely preserved cross-reactivity when reverted to their VH-VL GL configuration. (A) Representative flow cytometry plots of microbiota reactivity of two mAbs toward human gut bacteria. Commensal bacteria were isolated from fecal samples of three healthy individuals (I to III). Data are representative of two independent experiments. (B) Microbiota reactivity of the different mAbs toward human gut microbiota. Frequency of mAb-coated bacteria is represented with a color scale. Data are representative of two independent experiments. Serotype specificity is indicated on the right part of the graph. Asterisks (*) indicate mAbs for which GL versions have been produced. Hash symbols (#) indicate microbiota that have been sorted and analyzed by 16S ribosomal DNA sequencing in (C) and (D). (C) Relative abundance of the four main phyla in mAb+ fractions. (D) Heatmap diagram of EI of the most frequent genera from the three healthy microbiota. Selected genera are recognized by at least one mAb (EI > 0.2) and belong to the 100 most frequent genera in microbiota I, II, and III. Genera are grouped according to their phylum. (E) Representative flow cytometry analysis of mAb (red line) or irrelevant IgG (gray, anti–TNF-α IgG1) staining of pure bacterial strains. (F) Pairwise comparison of microbiota reactivity of mAbs and their respective GL revertants. Author contributions: Conceptualization: S.W., J.-C.W., and C.-A.R. Methodology: S.W., J.-C.W., and C.-A.R. Software: T.F. Validation: S.W. and D.S. Investigation: S.W., D.S., E.C., A.V.d.l.A., C.G., R.F., and M.B. Formal analysis: S.W., D.S., T.F., E.C., A.V.d.l.A., C.G., R.F., and M.B. Visualization: S.W., D.S., and T.F. Supervision: S.W., J.-C.W., and C.-A.R. Funding acquisition: J.-C.W. and C.-A.R. Resources: F.B. and G.G. Project administration: S.W., J.-C.W., and C.-A.R. Writing—original draft: S.W., J.-C.W., and C.-A.R. Writing—review and editing: S.W., A.V.d.l.A., C.G., R.F., J.-C.W., and C.-A.R. Competing interests: J.-C.W. received consulting fees from the Fondation Mérieux outside of this work. The other authors declare that they have no competing interests. References and Notes 1 Mosier DE Mond JJ Goldings EA The ontogeny of thymic independent antibody responses in vitro in normal mice and mice with an X-linked B cell defect J Immunol 1977 119 1874 1878 334975 2 Mond JJ Lees A Snapper CM T cell-independent antigens type 2 Annu Rev Immunol 1995 13 655 692 7612238 3 Mond JJ Vos Q Lees A Snapper CM T cell independent antigens Curr Opin Immunol 1995 7 349 354 7546399 4 Vos Q Lees A Wu ZQ Snapper CM Mond JJ B-cell activation by T-cell-independent type 2 antigens as an integral part of the humoral immune response to pathogenic microorganisms Immunol Rev 2000 176 154 170 11043775 5 Balázs M Martin F Zhou T Kearney JF Blood dendritic cells interact with splenic marginal zone B cells to initiate T-independent immune responses Immunity 2002 17 341 352 12354386 6 García de Vinuesa C MacLennan ICM Holman M Klaus GGB Anti-CD40 antibody enhances responses to polysaccharide without mimicking T cell help Eur J Immunol 1999 29 3216 3224 10540333 7 Jeurissen A Ceuppens JL Bossuyt X T lymphocyte dependence of the antibody response to “T lymphocyte independent type 2” antigens Immunology 2004 111 1 7 14678191 8 Litinskiy MB Nardelli B Hilbert DM He B Schaffer A Casali P Cerutti A DCs induce CD40-independent immunoglobulin class switching through BLyS and APRIL Nat Immunol 2002 3 822 829 12154359 9 Puga I Cols M Barra CM He B Cassis L Gentile M Comerma L Chorny A Shan M Xu W Magri G B cell–helper neutrophils stimulate the diversification and production of immunoglobulin in the marginal zone of the spleen Nat Immunol 2012 13 170 180 10 Defrance T Taillardet M Genestier L T cell-independent B cell memory Curr Opin Immunol 2011 23 330 336 21482090 11 Liu X Zhao Y Qi H T-independent antigen induces humoral memory through germinal centers J Exp Med 2022 219 e20210527 35019947 12 Obukhanych TV Nussenzweig MC T-independent type II immune responses generate memory B cells J Exp Med 2006 203 305 310 16476769 13 Musher DM Manoff SB Liss C McFetridge RD Marchese RD Bushnell B Alvarez F Painter C Blum MD Silber JL Safety and antibody response, including antibody persistence for 5 years, after primary vaccination or revaccination with pneumococcal polysaccharide vaccine in middle-aged and older adults J Infect Dis 2010 201 516 524 20092407 14 Cripps AW Folaranmi T Johnson KD Musey L Niederman MS Buchwald UK Immunogenicity following revaccination or sequential vaccination with 23-valent pneumococcal polysaccharide vaccine (PPSV23) in older adults and those at increased risk of pneumococcal disease: A review of the literature Expert Rev Vaccines 2021 20 257 267 33567914 15 Granoff DM Pollard AJ Reconsideration of the use of meningococcal polysaccharide vaccine Pediatr Infect Dis J 2007 26 716 722 17848884 16 Granoff DM Gupta RK Belshe RB Anderson EL Induction of immunologic refractoriness in adults by meningococcal C polysaccharide vaccination J Infect Dis 1998 178 870 874 9728562 17 O’Brien KL Hochman M Goldblatt D Combined schedules of pneumococcal conjugate and polysaccharide vaccines: Is hyporesponsiveness an issue? Lancet Infect Dis 2007 7 597 606 17714673 18 Siegrist C-A Aspinall R B-cell responses to vaccination at the extremes of age Nat Rev Immunol 2009 9 185 194 19240757 19 Lesinski GB Westerink MAJ Novel vaccine strategies to T-independent antigens J Microbiol Methods 2001 47 135 149 11576678 20 Martin F Oliver AM Kearney JF Marginal zone and B1 B cells unite in the early response against T-independent blood-borne particulate antigens Immunity 2001 14 617 629 11371363 21 Kolibab K Smithson SL Shriner AK Khuder S Romero-Steiner S Carlone GM Westerink MJ Immune response to pneumococcal polysaccharides 4 and 14 in elderly and young adults. I Antibody concentrations, avidity and functional activity Immun Ageing 2005 2 10 15982420 22 Lucas AH Moulton KD Tang VR Reason DC Combinatorial library cloning of human antibodies to Streptococcus pneumoniae capsular polysaccharides: Variable region primary structures and evidence for somatic mutation of Fab fragments specific for capsular serotypes 6B, 14, and 23F Infect Immun 2001 69 853 864 11159978 23 Zhou J Lottenbach KR Barenkamp SJ Lucas AH Reason DC Recurrent variable region gene usage and somatic mutation in the human antibody response to the capsular polysaccharide of Streptococcus pneumoniae type 23F Infect Immun 2002 70 4083 4091 12117915 24 Barrett DJ Ayoub EM IgG2 subclass restriction of antibody to pneumococcal polysaccharides Clin Exp Immunol 1986 63 127 134 3955880 25 Tarkowski A Lue C Moldoveanu Z Kiyono H McGhee JR Mestecky J Immunization of humans with polysaccharide vaccines induces systemic, predominantly polymeric IgA2subclass antibody responses J Immunol 1990 144 3770 3778 2110213 26 Haas KM B-1 lymphocytes in mice and nonhuman primates Ann N Y Acad Sci 2015 1362 98 109 25930711 27 Bagnara D Squillario M Kipling D Mora T Walczak AM Da Silva L Weller S Dunn-Walters DK Weill J-C Reynaud C-A A Reassessment of IgM memory subsets in humans J Immunol 2015 195 3716 3724 26355154 28 Weller S Braun MC Tan BK Rosenwald A Cordier C Conley ME Plebani A Kumararatne DS Bonnet D Tournilhac O Tchernia G Human blood IgM “memory” B cells are circulating splenic marginal zone B cells harboring a prediversified immunoglobulin repertoire Blood 2004 104 3647 3654 15191950 29 Weller S Faili A Garcia C Braun MC Le Deist F de Saint Basile G Hermine O Fischer A Reynaud CA Weill JC CD40-CD40L independent Ig gene hypermutation suggests a second B cell diversification pathway in humans Proc Natl Acad Sci U S A 2001 98 1166 1170 11158612 30 Weller S Mamani-Matsuda M Picard C Cordier C Lecoeuche D Gauthier F Weill J-C Reynaud C-A Somatic diversification in the absence of antigen-driven responses is the hallmark of the IgM+IgD+CD27+ B cell repertoire in infants J Exp Med 2008 205 1331 1342 18519648 31 Tangye SG Good KL Human IgM+CD27+ B cells: Memory B cells or “memory” B cells? J Immunol 2007 179 13 19 17579014 32 Budeus B Schweigle de Reynoso S Przekopowitz M Hoffmann D Seifert M Küppers R Complexity of the human memory B-cell compartment is determined by the versatility of clonal diversification in germinal centers Proc Natl Acad Sci U S A 2015 112 E5281 E5289 26324941 33 Seifert M Küppers R Molecular footprints of a germinal center derivation of human IgM+(IgD+)CD27+ B cells and the dynamics of memory B cell generation J Exp Med 2009 206 2659 2669 19917772 34 Descatoire M Weller S Irtan S Sarnacki S Feuillard J Storck S Guiochon-Mantel A Bouligand J Morali A Cohen J Jacquemin E Identification of a human splenic marginal zone B cell precursor with NOTCH2-dependent differentiation properties J Exp Med 2014 211 987 1000 24733829 35 Tull TJ Pitcher MJ Guesdon W Siu JHY Lebrero-Fernández C Zhao Y Petrov N Heck S Ellis R Dhami P Kadolsky UD Human marginal zone B cell development from early T2 progenitors J Exp Med 2021 218 e20202001 33538776 36 Carsetti R Rosado MM Donnanno S Guazzi V Soresina A Meini A Plebani A Aiuti F Quinti I The loss of IgM memory B cells correlates with clinical disease in common variable immunodeficiency J Allergy Clin Immunol 2005 115 412 417 15696104 37 Hart M Steel A Clark SA Moyle G Nelson M Henderson DC Wilson R Gotch F Gazzard B Kelleher P Loss of discrete memory B cell subsets is associated with impaired immunization responses in HIV-1 infection and may be a risk factor for invasive pneumococcal disease J Immunol 2007 178 8212 8220 17548660 38 Kruetzmann S Rosado MM Weber H Germing U Tournilhac O Peter H-H Berner R Peters A Boehm T Plebani A Quinti I Human immunoglobulin M memory B cells controlling Streptococcus pneumoniae infections are generated in the spleen J Exp Med 2003 197 939 945 12682112 39 Maglione PJ Simchoni N Black S Radigan L Overbey JR Bagiella E Bussel JB Bossuyt X Casanova J-L Meyts I Cerutti A IRAK-4 and MyD88 deficiencies impair IgM responses against T-independent bacterial antigens Blood 2014 124 3561 3571 25320238 40 Morrow M Valentin A Little R Yarchoan R Pavlakis GN A splenic marginal zone-like peripheral blood CD27+B220− B cell population is preferentially depleted in HIV type 1-infected individuals AIDS Res Hum Retroviruses 2008 24 621 633 18426338 41 Khaskhely N Mosakowski J Thompson RS Khuder S Smithson SL Westerink MAJ Phenotypic analysis of pneumococcal polysaccharide-specific B cells J Immunol 2012 188 2455 2463 22271652 42 Moens L Wuyts M Meyts I De Boeck K Bossuyt X Human memory B lymphocyte subsets fulfill distinct roles in the anti-polysaccharide and anti-protein immune response J Immunol 2008 181 5306 5312 18832686 43 Roco JA Mesin L Binder SC Nefzger C Gonzalez-Figueroa P Canete PF Ellyard J Shen Q Robert PA Cappello J Vohra H Class-switch recombination occurs infrequently in germinal centers Immunity 2019 51 337 350 e7 31375460 44 Xu Z Zan H Pone EJ Mai T Casali P Immunoglobulin class-switch DNA recombination: Induction, targeting and beyond Nat Rev Immunol 2012 12 517 531 22728528 45 Smith K Muther JJ Duke AL McKee E Zheng N-Y Wilson PC James JA Fully human monoclonal antibodies from antibody secreting cells after vaccination with Pneumovax®23 are serotype specific and facilitate opsonophagocytosis Immunobiology 2013 218 745 754 23084371 46 Adler AS Mizrahi RA Spindler MJ Adams MS Asensio MA Edgar RC Leong J Leong R Roalfe L White R Goldblatt D Rare, high-affinity anti-pathogen antibodies from human repertoires, discovered using microfluidics and molecular genomics MAbs 2017 9 1282 1296 28846502 47 Zhao Y Uduman M Siu JHY Tull TJ Sanderson JD Wu Y-CB Zhou JQ Petrov N Ellis R Todd K Chavele K-M Spatiotemporal segregation of human marginal zone and memory B cell populations in lymphoid tissue Nat Commun 2018 9 3857 30242242 48 Fadlallah J Sterlin D Fieschi C Parizot C Dorgham K El Kafsi H Autaa G Ghillani-Dalbin P Juste C Lepage P Malphettes M Synergistic convergence of microbiota-specific systemic IgG and secretory IgA J Allerg Clin Immunol 2019 143 1575 1585 e4 49 Sterlin D Fadlallah J Adams O Fieschi C Parizot C Dorgham K Rajkumar A Autaa G El-Kafsi H Charuel J-L Juste C Human IgA binds a diverse array of commensal bacteria J Exp Med 2020 217 e20181635 31891367 50 Rossi M Martínez-Martínez D Amaretti A Ulrici A Raimondi S Moya A Mining meta-genomic whole genome sequences revealed subdominant but constant Lactobacillus population in the human gut microbiota Environ Microbiol Rep 2016 8 399 406 27043715 51 Kabbert J Benckert J Rollenske T Hitch TCA Clavel T Cerovic V Wardemann H Pabst O High microbiota reactivity of adult human intestinal IgA requires somatic mutations J Exp Med 2020 217 e20200275 32640466 52 Planchais C Kök A Kanyavuz A Lorin V Bruel T Guivel-Benhassine F Rollenske T Prigent J Hieu T Prazuck T Lefrou L HIV-1 envelope recognition by polyreactive and cross-reactive intestinal B cells Cell Rep 2019 27 572 585 e7 30970259 53 Wardemann H Yurasov S Schaefer A Young JW Meffre E Nussenzweig MC Pre-dominant autoantibody production by early human B cell precursors Science 2003 301 1374 1377 12920303 54 Berkowska MA Schickel J-N Grosserichter-Wagener C de Ridder D Ng YS van Dongen JJM Meffre E van Zelm MC Circulating human CD27-IgA+ memory B cells recognize bacteria with polyreactive Igs J Immunol 2015 195 1417 1426 26150533 55 Phad GE Pinto D Foglierini M Akhmedov M Rossi RL Malvicini E Cassotta A Fregni CS Bruno L Sallusto F Lanzavecchia A Clonal structure, stability and dynamics of human memory B cells and circulating plasmablasts Nat Immunol 2022 23 1 10 56 Fine RL Manfredo Vieira S Gilmore MS Kriegel MA Mechanisms and consequences of gut commensal translocation in chronic diseases Gut Microbes 2020 11 217 230 31306081 57 Biram A Shulman Z T cell help to B cells: Cognate and atypical interactions in peripheral and intestinal lymphoid tissues Immunol Rev 2020 296 36 47 32557712 58 Ramos-Sevillano E Ercoli G Brown JS Mechanisms of naturally acquired immunity to Streptococcus pneumoniae Front Immunol 2019 10 358 30881363 59 Galson JD Clutterbuck EA Trück J Ramasamy MN Münz M Fowler A Cerundolo V Pollard AJ Lunter G Kelly DF BCR repertoire sequencing: Different patterns of B-cell activation after two Meningococcal vaccines Immunol Cell Biol 2015 93 885 895 25976772 60 Weller S Reynaud C-A Weill J-C Vaccination against encapsulated bacteria in humans: Paradoxes Trends Immunol 2005 26 85 89 15668123 61 Lorin V Fernández I Masse-Ranson G Bouvin-Pley M Molinos-Albert LM Planchais C Hieu T Péhau-Arnaudet G Hrebík D Girelli-Zubani G Fiquet O Epitope convergence of broadly HIV-1 neutralizing IgA and IgG antibody lineages in a viremic controller J Exp Med 2022 219 e20212045 35230385 62 Shivatare VS Shivatare SS Lee C-CD Liang C-H Liao K-S Cheng Y-Y Saidachary G Wu C-Y Lin N-H Kwong PD Burton DR Unprecedented role of hybrid N-glycans as ligands for HIV-1 broadly neutralizing anti-bodies J Am Chem Soc 2018 140 5202 5210 29578688 63 Williams WB Meyerhoff RR Edwards RJ Li H Manne K Nicely NI Henderson R Zhou Y Janowska K Mansouri K Gobeil S Fab-dimerized glycan-reactive antibodies are a structural category of natural antibodies Cell 2021 184 2955 2972 e25 34019795 64 Weill J-C Weller S Reynaud C-A A bird’s eye view on human B cells Semin Immunol 2004 16 277 281 15522625 65 Heilmann C Pedersen FK Quantitation of blood lymphocytes secreting antibodies to pneumococcal polysaccharides after in vivo antigenic stimulation Scand J Immunol 1986 23 189 194 3513298 66 Kehrl JH Fauci AS Activation of human B lymphocytes after immunization with pneumococcal polysaccharides J Clin Invest 1983 71 1032 1040 6339558 67 Tiller T Meffre E Yurasov S Tsuiji M Nussenzweig MC Wardemann H Efficient generation of monoclonal antibodies from single human B cells by single cell RT-PCR and expression vector cloning J Immunol Methods 2008 329 112 124 17996249 68 Crickx E Chappert P Sokal A Weller S Azzaoui I Vandenberghe A Bonnard G Rossi G Fadeev T Storck S Fadlallah J Rituximab-resistant splenic memory B cells and newly engaged naive B cells fuel relapses in patients with immune thrombocytopenia Sci Transl Med 2021 13 eabc3961 33853929 69 Gupta NT Vander Heiden JA Uduman M Gadala-Maria D Yaari G Kleinstein SH Change-O: A toolkit for analyzing large-scale B cell immunoglobulin repertoire sequencing data Bioinformatics 2015 31 3356 3358 26069265 70 Hoehn KB Lunter G Pybus OG A phylogenetic codon substitution model for antibody lineages Genetics 2017 206 417 427 28315836 71 Hoehn KB Vander Heiden JA Zhou JQ Lunter G Pybus OG Kleinstein SH Rep-ertoire-wide phylogenetic models of B cell molecular evolution reveal evolutionary signatures of aging and vaccination Proc Natl Acad Sci U S A 2019 116 22664 22672 31636219 72 Gerard A-L Goulenok T Bahuaud M Francois C Aucouturier P Moins-Teisserenc H Batteux F Papo T Sacre K Serum IgG2 levels predict long-term protection following pneumococcal vaccination in systemic lupus erythematosus (SLE) Vaccine 2020 38 6859 6863 32896468 73 Juste C Kreil DP Beauvallet C Guillot A Vaca S Carapito C Mondot S Sykacek P Sokol H Blon F Lepercq P Bacterial protein signals are associated with Crohn’s disease Gut 2014 63 1566 1577 24436141 74 Moor K Fadlallah J Toska A Sterlin D Balmer ML Macpherson AJ Gorochov G Larsen M Slack E Analysis of bacterial-surface-specific antibodies in body fluids using bacterial flow cytometry Nat Protoc 2016 11 1531 1553 27466712 75 Salter SJ Cox MJ Turek EM Calus ST Cookson WO Moffatt MF Turner P Parkhill J Loman NJ Walker AW Reagent and laboratory contamination can critically impact sequence-based microbiome analyses BMC Biol 2014 12 87 25387460 76 Vander Heiden JA Yaari G Uduman M Stern JNH O’Connor KC Hafler DA Vigneault F Kleinstein SH pRESTO: a toolkit for processing high-throughput sequencing raw reads of lymphocyte receptor repertoires Bioinformatics 2014 30 1930 1932 24618469 77 Brochet X Lefranc M-P Giudicelli V IMGT/V-QUEST: The highly customized and integrated system for IG and TR standardized V-J and V-D-J sequence analysis Nucleic Acids Res 2008 36 W503 W508 18503082 78 Ademokun A Wu Y-C Martin V Mitra R Sack U Baxendale H Kipling D Dunn-Walters DK Vaccination-induced changes in human B-cell repertoire and pneumococcal IgM and IgA antibody at different ages Aging Cell 2011 10 922 930 21726404 79 Baxendale HE Davis Z White HN Spellerberg MB Stevenson FK Goldblatt D Immunogenetic analysis of the immune response to pneumococcal polysaccharide Eur J of Immunol 2000 30 1214 1223 10760811 80 Bryson S Thomson CA Risnes LF Dasgupta S Smith K Schrader JW Pai EF Structures of preferred human IgV genes–based protective antibodies identify how conserved residues contact diverse antigens and assign source of specificity to CDR3 loop variation J Immunol 2016 196 4723 27183571 81 Chen Z Cox KS Tang A Roman J Fink M Kaufhold RM Guan L Xie A Boddicker MA Mcguinness D Xiao X Human monoclonal antibodies isolated from a primary pneumococcal conjugate Vaccinee demonstrates the expansion of an antigen-driven Hypermutated memory B cell response BMC Infect Dis 2018 18 613 30509199 82 Shaw DR Kirkham P Schroeder HW Roben P Silverman GJ Structure-function studies of human monoclonal antibodies to pneumococcus type 3 polysaccharide Ann N Y Acad Sci 1995 764 370 373 7486548 83 Zhou J Lottenbach KR Barenkamp SJ Reason DC Somatic hypermutation and diverse immunoglobulin gene usage in the human antibody response to the capsular polysaccharide of Streptococcus pneumoniae Type 6B Infect Immun 2004 72 3505 3514 15155658

PMC007xxxxxx/PMC7614726.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101525337 ACS Chem Neurosci ACS Chem Neurosci ACS chemical neuroscience 1948-7193 31845794 7614726 10.1021/acschemneuro.9b00338 EMS177433 Article Novel HDAC6 inhibitors increase tubulin acetylation and rescue axonal transport of mitochondria in a model of Charcot-Marie-Tooth Type 2F Adalbert Robert †“¶ Kaieda Akira ‡ Antoniou Christina † Loreto Andrea † Yang Xiuna † Gilley Jonathan † Hoshino Takashi ‡ Uga Keiko ‡# Makhija Mahindra T. ‡ Coleman Michael P. †§ † John van Geest Centre for Brain Repair, Department of Clinical Neurosciences, University of Cambridge, Forvie Site, Robinson Way, Cambridge CB2 0PY, UK ‡ Takeda Pharmaceutical Company Limited, 26-1, Muraoka-higashi 2-chome, Fujisawa, Kanagawa 251-8555, Japan ‡ Takeda Development Centre Europe Ltd., 61 Aldwych London WC2B 4AE UK § Babraham Institute, Babraham, Cambridge, CB22 3AT, UK “ Department of Anatomy, Histology and Embryology, Faculty of Medicine, University of Szeged, Szeged, Hungary Corresponding Authors *(M.P.C.) Tel: +44 1223 362151, Fax: +44 1223 331174. mc469@cam.ac.uk *(M.M) Tel/Fax: +44 20 3116 8965. mahindra.makhija@takeda.com ¶ Present address: Comparative Neuromuscular Disease Laboratory, Royal Veterinary College, 4 Royal College Street, Camden, London NW1 0TU, UK # Present address: Axcelead Drug Discovery Partners, Inc., 26-1, Muraoka-higashi 2-chome, Fujisawa, Kanagawa 251-0012, Japan 05 2 2020 08 1 2020 29 6 2023 06 7 2023 11 3 258267 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Disruption of axonal transport causes a number of rare, inherited axonopathies and is heavily implicated in a wide range of more common neurodegenerative disorders, many of them age-related. Acetylation of α-tubulin is one important regulatory mechanism, influencing microtubule stability and motor protein attachment. Of several strategies so far used to enhance axonal transport, increasing microtubule acetylation through inhibition of the deacetylase enzyme HDAC6 has been one of the most effective. Several inhibitors have been developed and tested in animal and cellular models but better drug candidates are still needed. Here we report the development and characterisation of two highly potent HDAC6 inhibitors, which show low toxicity, promising pharmacokinetic properties, and enhance microtubule acetylation in the nanomolar range. We demonstrate their capacity to rescue axonal transport of mitochondria in a primary neuronal culture model of the inherited axonopathy Charcot-Marie-Tooth Type 2F, caused by a dominantly acting mutation in heat shock protein beta 1. HDAC6 CMT Axonal transport Mitochondria Axonopathy α-tubulin pmcIntroduction The bidirectional movement of macromolecules and organelles along axons is essential for axon survival and function, and requires a complex machinery involving motor proteins, adapters coupling to specific cargoes, microtubule tracks and regulators of all the above. Not surprisingly, such an essential process involving many components can malfunction in a number of ways and when it does the consequences can be profound. Mutation of genes encoding axonal transport machinery and regulators cause a number of axonopathies, and especially diseases of long axons. For example, mutations in KIF5A encoding a major anterograde motor protein are an established cause of hereditary spastic paraplegia SPG101, and have also been linked to Charcot-Marie-Tooth Disease type 2 (CMT2)2, amyotrophic lateral sclerosis (ALS)3, 4 and neonatal intractable myoclonus5. Mutant dynactin causes motor neuron disease and distal spinal and bulbar muscular atrophy 6, 7 and in mice the mutation of tubulin chaperone protein Tbce causes a severe, early onset loss of motor axons with a major deficiency of microtubules8. Charcot-Marie-Tooth Disease type 2 (CMT2) is an axonal, non-demyelinating peripheral neuropathy characterized by distal muscle weakness and atrophy, mild sensory loss, and normal or near-normal nerve conduction velocities9. The Charcot-Marie-Tooth disease subtype 2F (CMT2F) and distal hereditary motor neuropathy subtype 2B (dHMN2B) are caused by autosomal dominantly inherited mutations in the small heat shock protein B1 (HSPB1) gene10, 11. The gene codes for heat shock protein beta-1 (HSPB1, also known as HSP27), which is a member of the small heat shock protein family comprising a highly conserved α-crystalline domain. HSPB1 acts as a chaperone binding with partially denatured proteins to prevent aggregation12, 13. Up to now, 18 mutations in HSPB1 have been linked to CMT2F and 27 mutations to dHMN 214. The S135F and P182L mutations are among the best characterized mutations so far11, 15, 16. The S135F mutation is the only one that causes both CMT2 and dHMN2B. P182L mutation is associated only with dHMN2B15. The S135F mutation is located in the α-crystallin domain while P182L mutation lies in the short C-terminal tail of the protein15. Interestingly the localization of the mutation was shown to have different effects on the protein function. While the S135F mutation caused the protein to increase its chaperone activity accompanied with an increased in its monomeric state the chaperone activity of HSPB1 was not affected by the P182L mutation17. Four mutant HSPB1 transgenic mouse models of CMT2F and/or dHMN2B have been developed so far, which partially recapitulate the hallmarks of peripheral neuropathy11, 18–20. S135F and P182L transgenic mice generated by d’Ydewalle et al. demonstrated noticeable phenotypes, the latter presents more like dHMN2B than CMT2F with a lack of sensory symptoms11 which recapitulates all key features of CMT2F or distal HMN2B, dependent on the mutation. However, CMT2F mouse models generated by other groups had notable differences. S135F transgenic mice reported by Lee et al. had no sensory phenotype and presented only a strict motor loss, similar to the P182L, but not the S135F mice of d’Ydewalle et al18. In further contrast, the R136W mouse model did not demonstrate any functional or behavioral deficits20. When R127W and P182L mutant proteins were expressed at physiological levels to alleviate concerns of artifacts due to overexpression, no pathology and behavioural deficits were found in mice19. This could be due to insufficient expression of HSPB1 under the ROSA26 locus. In addition to rare disorders and animal models, axonal transport deficiency is heavily implicated in many more common neurodegenerative and axonal disorders. Several cancer chemotherapeutics that cause peripheral neuropathy as a dose-limiting complication target microtubules21 and disrupt axonal transport 22. In Alzheimer’s disease, aggregation of microtubule associated protein tau, whose normal functions include regulation of microtubule stability and motor protein attachment23, 24 plays a prominent role, exogenously applied Aβ1-42 is able to disrupt axonal transport in a tau-dependent manner25. The two may also interact26 and impairment of axonal transport exacerbates animal models27. There are many indications of a wider role also in ALS28, Huntington’s disease29, Parkinsonism and frontotemporal dementia30, 31, and normal ageing, the biggest risk factor in each of these32,is accompanied by a twofold decline in axonal transport33. Thus, rare but often better-understood inherited disorders involving an axonal transport mechanism are an important starting point to develop therapies that could have far wider application in neurodegenerative disease. Axonal microtubules exist in a state of dynamic instability34, 35, constantly both growing and severing to maintain them typically between 0.15-20 μm in length36. Acetylation of α-tubulin at Lys40 is reported to increase microtubule stability under mechanical stress37 and to influence severing by katanin38. It also enhances the binding of kinesin-1 and axonal transport39, 40. Beside SIRT241 HDAC6 is the other major deacetylase for α-tubulin and its inhibition increases axonal transport of some cargoes in models of Charcot-Marie-Tooth disease types 2F11 and 2D42, ALS43, and Vincristine neuropathy44, also alleviating some symptoms. Beneficial outcomes have also been reported in models of Alzheimer’s disease45 and stroke46. Early studies of HDAC6 inhibition used tubacin, whose high lipophilicity and short in vivo half-life limited its usefulness. This was largely superseded by the development of Tubastatin A47. However, further improvement of potency is possible48, 49 so it is important to develop new compounds targeting HDAC6 with greater potential for clinical application. The HDAC inhibitors share a well-recognized pharmacophore that consists of three parts: a zinc binding group (ZBG), a linker, and a cap moiety. Classical HDAC inhibitors typically have the hydroxamic acid moiety as ZBG but the hydroxamic acid causes poor pharmacokinetics, low selectivity profiles, and production of active metabolites50. These features of hydroxamic acid are red flags for drug discovery in chronic diseases that are not life threatening. Therefore, we focussed here on the discovery of non-hydroxamic acid derivatives. High throughput screening (HTS) with a Takeda internal library provided several non-hydroxamic acid derivatives as hit compounds against HDAC6. By our medicinal chemistry efforts, we developed two compounds T-3796106 and T-3793168 that are highly selective for HDAC6, show CNS penetration and low toxicity both in vivo and in vitro. We report their dose-response effects for α-tubulin acetylation in primary neuronal cultures and their influence on axonal transport of mitochondria in a primary culture model of CMT2F. Results Evaluation of inhibitory potencies (IC50) of T-3796106 and T-3793168 The inhibitory potencies (IC50) of T-3796106 and T-3793168, which were developed through medicinal chemistry campaign from HTS hit compounds, were evaluated in HDAC panel assay (Table 1). T-3796106 showed potent inhibitory activity against HDAC6 with the IC50 value of 12 nM. IC50 values for HDAC3, HDAC8, HDAC5, HDAC7, and HDAC9 were in the range of 1,000-3,000 nM. IC50 values for HDAC1 and HDAC4 were over 6,000 nM. T-3796106 did not show inhibitory activity against HDAC2, HDAC10, and HDAC11 up to 10,000 nM. IC50 values of T-3793168 were 86 nM for HDAC6 and over 2,000 nM against other HDACs. T-3796106 and T-3793168 do not cause neuronal toxicity even at high concentrations First, we tested whether T-3796106 or T-3793168 induces any cytotoxicity in murine neuronal explant cultures, using concentrations substantially higher than those we subsequently used for axonal transport studies to indicate a large therapeutic window. We used superior cervical ganglion (SCG) explants because this neuron type is well-suited for genetic manipulation by microinjection and for axonal transport studies 51, 52, and incubated with concentrations from 1 μM to 100 μM for 24 h. No toxicity was observed at any concentration. Neurites remained morphologically similar to vehicle-treated or untreated cultures, even in their distal terminals which are typically the most vulnerable site (Fig 1A, B). Thus, both compounds are safe for neurons up to 100 μM for at least 24 h. Increased acetylation levels of α-tubulin after T-3796106 and T-3793168 treatment in neurons We next confirmed that steady state α-tubulin acetylation increases with either T-3796106 or T-3793168 within the above concentration range (data not shown), before titrating down to determine the dose-response curve for α-tubulin acetylation at sub-saturating levels, thereby minimizing the risk of off-target effects. Both compounds showed a clear dose-response effect between 1 nM and 250 nM in a 24 h incubation (Fig 2A), reaching significance at 50 nM for T-3796106 and 250 nM for T-3793168 (Fig 2B). Based on this characterization we used concentrations of 100 nM and 250 nM respectively in our subsequent axonal transport experiments. At these concentrations, there was no effect on histone acetylation which indicates a high selectivity of these compounds towards HDAC6 (Supplementary Figure 1). Axonal transport of mitochondria in wild type SCG cultures is not altered by T-3796106 or T-3793168 In the absence of a pathogenic mutation, we found no significant change in either the number or the average and maximum velocity of axonally transported mitochondria in dissociated wild-type SCG neurons treated with T-3796106, T-3793168 or Tubastatin A (Fig 3B, C, D, E). Thus, there is no change in basal axonal transport parameters for this cargo. Mitochondrial transport impairment induced by S135F mutation is rescued by T-3793168 In the presence of the HSPB1S135F mutation, which causes CMT2 and distal HMN in patients15, both the numbers of anterogradely and retrogradely moving mitochondria were significantly decreased relative to wild type neurons 12 h after microinjection (Fig 4A, B), mirroring similar changes reported in sensory neurons11. The transport deficits in both directions were significantly rescued in neurons treated for 24 h with 250 nM T-3793168, while those treated with 100 nM compound T-3796106 showed a trend towards increased mitochondrial transport but the difference was not statistically significant (Fig 4A, B). As previously reported11, we also found a rescue of anterograde axonal transport with 1 μM Tubastatin A, but in the retrograde direction the trend towards a rescue with Tubastatin A was not significant. The average and maximum speed of mitochondria movement was not significantly altered in the neurons with the HSPB1S135F mutation and was unaffected by any of these treatments (Fig 4C, D). P182L mutation does not alter mitochondrial transport in SCG neurons Consistent with previous findings in sensory neurons11, SCG neurons expressing HSPB1P182L showed no significant changes in mitochondrial transport compared to wild type neurons (Fig 5). Treatment of these mutant-expressing neurons with T-3796106, T-3793168 or Tubastatin A also had no effect on mitochondrial movement (Fig 5). T-3796106 and T-3793168 increase α-tubulin acetylation in human whole blood We finally investigated the effects of T-3796106 and T-3793168 on acetylation of α-tubulin in human cells, using whole blood. For both compounds, clear dose-response effects were observed between 10 nM and 30 μM in a 4 h incubation (Fig 6). Over 100 μM of our compounds and 300 μM of hydroxamic acid-based HDAC6 inhibitor ACY-1215 showed some precipitate when the compounds were added in the culture medium. We also observed the effect of Tubastatin A on acetylated a-tubulin in the human whole blood assay. It showed a similar trend. In brief, the levels of acetylated a-tubulin were almost the same at 10 and 30 μM of Tubastatin A (data not shown) as well as that of ACY-1215. Discussion We report the development and characterization of two novel non-hydroxamic acid-based inhibitors with high potency and specificity for HDAC6, low toxicity in murine primary neuronal cultures and a dose-dependent effect on neuronal α-tubulin acetylation between 1250 nM. T-3793168 significantly increases both the anterograde and retrograde flux of mitochondria in axonal transport within 24 h of application to neurons expressing the CMT-2F HSPB1 mutation S135F, and T-3796106 shows a similar, albeit non-significant trend. Neither alters axonal transport in wild-type cells. For both compounds the changes in acetylated tubulin in whole blood was several orders of magnitude greater than those in mouse primary neuronal cultures. This suggests a tissue-, tubulin isoform-, or species-specific effect on the efficacy of these HDAC6 inhibitors indicating significant scope of lead compound optimization. Further studies to understand the basis of this specificity should help to optimize their efficacies to achieve substantial enhancement of axonal transport in human axonal disorders. A major advantage over hydroxamic acid-based inhibitors is greater selectivity over other HDAC family members and this is the case for these compounds. For example, hydroxamic acid-based HDAC6 inhibitor ACY-1215 has a high HDAC6 enzyme potency with IC50 value of 4.7 nM but much lower selectivity (12-fold selectivity for HDAC6 and HDAC1 at IC50)53. In contrast, non-hydroxamic acid-based HDAC6 inhibitors T-3796106 and T-3793168 showed excellent selectivity (>25-fold over other HDAC family members; >100-fold selectivity over HDAC1 at IC50). It will be important now to test the effect of these compounds on axonal transport in other disease models where transport is impaired and the effect on other axonal transport cargoes. For example, axonal transport defects underlie vincristine neuropathy and some forms of hereditary spastic paraplegia and ALS, and have also been implicated in glaucoma, Alzheimer’s disease and multiple sclerosis. Many axonal transport cargoes need to be continuously shuttled back and forth but among some of the most important ones are NMNAT2, whose absence limits the survival of transected axons51 and the retrograde transport of lysosomes to maintain efficient autophagy and mitochondrial quality control54, and neurotrophins. Finally, it will be important to test the efficacy of HDAC6 inhibition and rescue of axonal transport in vivo using methods for live imaging of transport cargoes in live nerves and CNS tissue33, 55, 56, to assess how HDAC6 inhibition compares to other methods of boosting axonal transport in experimental models57, 58 and to further develop these lead compounds. Methods Chemicals T-3796106 and T-3793168 are novel HDAC6 inhibitors developed by Takeda Pharmaceutical Company Limited (Patent WO2017014321)59.The purity of T-3796106 and T-3793168 was determined to be ≥ 95% by elemental analysis which was performed by Sumika Chemical Analysis Service, Ltd. experimentally determined hydrogen, carbon, and nitrogen composition by elemental analysis was within ±0.4% of the expected value, implying a purity of ≥95%. ACY-1215 was synthesized and determined to be ≥ 95% purity by elemental analysis by Takeda Pharmaceutical Company Limited. Enzyme assay HDAC panel assay was performed by Reaction Biology Corp. (Malvern, PA, USA) according to their validated protocol. To evaluate the potency and selectivity of T-3796106 and T- 3793168, HDAC panel assay was carried out by Reaction Biology Corp. Briefly, the deacetylation reaction was performed in buffer conditions of 50 mM Tris-HCl pH 8.0, 137 mM NaCl, 2.7 mM KCl, 1 mM MgCl2, and 1 mg/mL bovine serum albumin (BSA), and 1%DMSO. The fluorogenic peptide, RHK-K(Ac)-AMC, is used as substrate for Class1 and 2B HDACs, RHK(Ac)K(Ac)-AMC for HDAC8, and Boc-Lys(trifluoroacetyl)-AMC for Class2A HDACs. After the reaction, by Developer with Trichostatin A, a fluorescence signal (Ex. 360 nm/Em. 460 nm) developed. Animals C57BL/6JOlaHsd mice were obtained from Harlan UK (Bicester, UK). All animal work was carried out in accordance with the Animals (Scientific Procedures) Act, 1986, under Project License 70/7620. Cell culture Explant SCG cultures SCGs were dissected from 0 to 2 days old C57BL/6 (wild-type) mouse pups. Cleaned explants were placed in the centre of 3.5 cm tissue culture dishes pre-coated with poly-L-lysine (20 mg/mL for 1–2 h; Sigma) and laminin (20 mg/mL for 1–2 h; Sigma). Explants were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) with 4,500 mg/L glucose and 110 mg/L sodium pyruvate (Sigma), 2 mM glutamine (Invitrogen), 1% penicillin/streptomycin (Invitrogen), 100 ng/mL 7S NGF (Invitrogen), and 10% fetal bovine serum (Sigma). Four μM aphidicolin (Calbiochem) was used to reduce proliferation and viability of small numbers of non-neuronal cells. Cultures were used after 6 days. Dissociated SCG cultures SCG ganglia were dissociated by incubation in 0.025% trypsin (Sigma) in PBS (without CaCl2 and MgCl2) for 30 min followed by 0.2% collagenase type II (Gibco) in PBS for a further 20 min. Ganglia were then gently triturated using a pipette. After a 2 h pre-plating stage to remove non-neuronal cells, 5,000–10,000 dissociated neurons were plated in a 1 cm2 poly-L-lysine and laminin-coated area in the centre of 3.5 cm ibidi μ-dishes (Thistle Scientific, Glasgow, UK) for microinjection experiments or in the centre of 3.5 cm tissue culture dishes for analysis by western blotting. Dissociated cultures were maintained the same as explant cultures. HDAC6 inhibitors treatment The SCG explants and dissociated cultures were treated for 24 h at 37°C with compound dosages ranging from 1 μM to 100 μM for toxicity experiments and 1nM to 250 nM for testing the dose-response study of α-tubulin acetylation. For axonal transport rescue experiments, dissociated SCG cultures were treated with either 100 nM T-3796106, 250 nM T-3793168, 1 μM Tubastatin A or an equivalent amount of DMSO. Plasmid constructs The S135F and P182L mutations were introduced separately by QuikChange II site-directed mutagenesis (Stratagene) into the complete open reading frame of human HSPB1 isoform cloned into expression vector pCMV-Tag2 (Stratagene). The mito-EGFP construct was kindly provided by Dr Andrea Loreto. Microinjection Microinjection was performed using a Zeiss Axiovert 200 microscope with an Eppendorf 5171 transjector and 5246 micromanipulator system and Eppendorf Femtotips. Microinjection mixes of plasmid DNA were prepared in 0.5× PBS(–), passed through a Spin-X filter (Costar, Glasgow, UK) Eppendorf and injected directly into the nuclei of SCG neurons in dissociated cultures. Femtotips were loaded with the microinjection mix and injection was performed using an Eppendorf 5171 transjector and 5246 micromanipulator system on a Zeiss Axiovert 200 microscope. All injections were carried out directly into the nuclei of dissociated SCG neurons. A maximum total DNA concentration of 0.05 μg/μL in the injection mix was used. Forty cells were injected per dish and imaging was performed 12 hours after microinjection. Western Blotting Following treatment, ganglia and neurites were collected and washed in PBS with complete, ethylenediaminetetraacetic acid (EDTA)-free protease inhibitor cocktail tablets (Sigma-Aldrich), and lysed directly into 2x Laemmli sample buffer. A total of 10 μL of each sample were separated on a 12% SDS-PAGE and transferred to PVDF membrane (Millipore) using the Bio-Rad Mini-PROTEAN III wet transfer system. Blots were blocked and incubated with primary antibodies overnight (in 1xTBS pH. 8.3, with 0.05% Tween 20 and 5% milk powder or 5% BSA). The antibodies were directed against α-tubulin (1/5,000; ab 15246, Abcam) and acetylated α-tubulin (1/5,000; T7451, Sigma) and detected with mouse-700 (Life Technologies) and rabbit-800 (LI-COR) secondary antibodies. Blots were then scanned and quantified using the Odyssey imaging system (LI-COR Biosciences, Lincoln, North Carolina). Live imaging of mitochondrial transport and image analysis Mitochondria were labelled by microinjection of mito-EGFP and their movement along the neurites was recorded with an inverted spinning-disk confocal microscope Olympus IX70 using a 100x 1.49 NA oil immersion objective (Olympus), and controlled with MetaMorph 7.7 software (Molecular Devices). The environment was controlled with a stage top incubator (model INUBG2E-ZILCS; Tokai Hit), set at 37°C and CO2 set to 5%. Time lapse images of mitochondrial movements were acquired every 1 s for 2 min (120 frames in total). A total of 4-5 movies from different neurons were captured from each culture dish. Individual neurites were straightened using the Straighten plugin in ImageJ software version 1.44 (Rasband, W.S., ImageJ, U. S. National Institutes of Health, Bethesda, MD; http://imagej.nih.gov/ij/, 1997e2012). Transport parameters were determined for individual neurites using the Difference Tracker set of ImageJ plugins 060. The principal output of these plugins is the number of moving particles identified in each frame of the image, normalized to 1000 pixels (Figure 3A). Human whole blood assay Study design In human whole blood assay, whole human blood was collected from healthy volunteers after informed consent at Takeda Pharmaceutical Company Limited. 25 μL of the collected human whole blood was put into each well of a 96-well round-bottom plate. The whole blood was treated with 10 μL of diluted compounds in 10% FBS containing RPMI 1640 medium (Gibco). In the control group, 0.1% DMSO was added as a final concentration. After that, the treated whole blood was incubated at 37°C for 30 minutes at 5% CO2. Next, 65 μL of RPMI 1640 medium was added onto each well, and the samples were incubated at 37°C for 3.5 hours. T-3796106, T-3793168, and ACY-1215 were dissolved in 100% DMSO (to a stock concentration of 10-300 mM for our in vitro studies). Measurements For flow cytometry analyses, the compound-treated whole blood samples were transferred to an assay block (Costar). Diluted Lyse/Fix buffer (BD Biosciences) in dH2O was added to each sample with pipetting well. The samples were put at RT for 10 minutes and then centrifuged at 400xg for 5 minutes. After centrifugation, the supernatant was removed by aspiration. 250 μL of Perm/Wash buffer I (BD Biosciences) was added to each well and the samples were transferred into a 96-well V-bottom plate, and they were incubated on ice for 20 minutes. These samples were centrifuged at 400xg for 5 minutes at RT, and the supernatant was removed by aspiration. The cells were stained with Zenon conjugated AF647 acetylated α-tubulin (ab179484, Abcam) or matched isotype control (ab172730, Abcam) for 20-30 minutes on ice. Zenon Rabbit IgG Labeling Kit, AF647 (Molecular Probes) was used according to the provided protocol. The cells were centrifuged at 400xg at 5 minutes, and the supernatant was removed, and then washed with 200 μL of Perm/Wash buffer I. After re-centrifugation and removal of the supernatant, the samples were suspended with 200 μL of FACS stain buffer (1% FBS/PBS). The samples were analyzed using lymphocyte gate by BD Fortessa, and the results were analyzed with FlowJo software. Therefore, only lymphocytes were analyzed neither erythrocytes nor thrombocytes were included in the assay. Statistical analysis Statistical tests, as described in the figure legends, were performed using Prism software (GraphPad Software Inc, La Jolla, CA, USA). A p value of >0.05 was considered not significant (ns) and *p < 0.05 was significant. Supplementary Material Supporting information Acknowledgments The authors thank Daniel Curran and Tauhid Ali for their contribution to the discussions; Masashi Toyofuku, Kousuke Hidaka and Fumiaki Kikuchi for the synthesis of T-3796106, T- 3793168 and ACY-1215, Andrea Loreto for mito-EGFP construct and Myra Ng for the development of human whole blood assay.. Funding Funding for this work was provided by Takeda Development Centre Europe Ltd. M.P.C. is funded by the John and Lucille van Geest Foundation. Abbreviations HDAC histone deacetylase KIF5A kinesin heavy chain isoform 5A Aβ1-42 amyloid beta peptide 42 HMN hereditary motor neuropathy NMNAT2 nicotinamide mononucleotide adenylyltransferase 2 DMSO dimethyl sulfoxide FBS fetal bovine serum Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Table 1 HDAC panel assay. The selectivity of T-3796106 and T-3793168 was analyzed based on HDAC enzyme inhibition. [a] The compound activity against 11 HDACs represented with the IC50 value. The IC50 values shown are the mean values of duplicate measurements; the numbers in parentheses represent each data. [b] No inhibition or compound activity that could not be fit to an IC50 curve. Target Compounds IC50 (nM)a T-3796106 T-3793168 Tubastatin A HDAC1 6200 (5820-6660) b >10000 HDAC2 >10000 b >10000 HDAC3 4000 (3480-4470) b >10000 HDAC8 >1000 5600 (4170-7080) >1000 HDAC4 6200 (5970-6380) >10000 6200 (6030-6330) HDAC5 1700 (1610-1800) 2300 (1550-3060) 1900 (1580-2310) HDAC7 1100 (1050-1130) 5800 (4000-7660) 590 (530-641) HDAC9 2700 (2540-2840) 5000 (4950-4960) 1100 (913-1380) HDAC6 12 (12.3-12.4) 86 (67.4-104) 15 (14.3-15.2) HDAC10 >10000 b >10000 HDAC11 >10000 b >10000 Authors Contributions Research design: R.A., M.T. M., and M.P.C. Experimental work: R.A., A.K, A.L., C.A., X.Y., J.G., T.H., and K.U. Data analyses and interpretation: R.A., A.L., C.A., M. T. M, and M.P.C. Writing the manuscript: R.A., M.P.C. and M.T.M. Notes The authors declare no competing financial interests. Table of contents was created with BioRender.com. 1 Reid E Kloos M Ashley-Koch A Hughes L Bevan S Svenson IK Graham FL Gaskell PC Dearlove A Pericak-Vance MA Rubinsztein DC A kinesin heavy chain (KIF5A) mutation in hereditary spastic paraplegia (SPG10) Am J Hum Genet 2002 71 1189 1194 12355402 2 Crimella C Baschirotto C Arnoldi A Tonelli A Tenderini E Airoldi G Martinuzzi A Trabacca A Losito L Scarlato M Benedetti S Mutations in the motor and stalk domains of KIF5A in spastic paraplegia type 10 and in axonal Charcot-Marie-Tooth type 2 Clin Genet 2012 82 157 164 21623771 3 Brenner D Yilmaz R Muller K Grehl T Petri S Meyer T Grosskreutz J Weydt P Ruf W Neuwirth C Weber M German, A. L. S. n. M. N. D. N. E. T Hot-spot KIF5A mutations cause familial ALS Brain 2018 141 688 697 29342275 4 Nicolas A Kenna KP Renton AE Ticozzi N Faghri F Chia R Dominov JA Kenna BJ Nalls MA Keagle P Rivera AM Genome-wide Analyses Identify KIF5A as a Novel ALS Gene Neuron 2018 97 1268 1283 e1266 29566793 5 Duis J Dean S Applegate C Harper A Xiao R He W Dollar JD Sun LR Waberski MB Crawford TO Hamosh A KIF5A mutations cause an infantile onset phenotype including severe myoclonus with evidence of mitochondrial dysfunction Ann Neurol 2016 80 633 637 27463701 6 Puls I Jonnakuty C LaMonte BH Holzbaur EL Tokito M Mann E Floeter MK Bidus K Drayna D Oh SJ Brown RH Jr Mutant dynactin in motor neuron disease Nat Genet 2003 33 455 456 12627231 7 Puls I Oh SJ Sumner CJ Wallace KE Floeter MK Mann EA Kennedy WR Wendelschafer-Crabb G Vortmeyer A Powers R Finnegan K Distal spinal and bulbar muscular atrophy caused by dynactin mutation Ann Neurol 2005 57 687 694 15852399 8 Martin N Jaubert J Gounon P Salido E Haase G Szatanik M Guenet JL A missense mutation in Tbce causes progressive motor neuronopathy in mice Nat Genet 2002 32 443 447 12389029 9 Bird TD Charcot-Marie-Tooth Neuropathy Type 2 Adam MP Ardinger HH Pagon RA Wallace SE Bean LJH Stephens K Amemiya A GeneReviews((R)) 1993 Seattle (WA) 10 Ismailov SM Fedotov VP Dadali EL Polyakov AV Van Broeckhoven C Ivanov VI De Jonghe P Timmerman V Evgrafov OV A new locus for autosomal dominant Charcot-Marie-Tooth disease type 2 (CMT2F) maps to chromosome 7q11-q21 Eur J Hum Genet 2001 9 646 650 11528513 11 d'Ydewalle C Krishnan J Chiheb DM Van Damme P Irobi J Kozikowski AP Vanden Berghe P Timmerman V Robberecht W Van Den Bosch L HDAC6 inhibitors reverse axonal loss in a mouse model of mutant HSPB1-induced Charcot-Marie-Tooth disease Nat Med 2011 17 968 U986 21785432 12 Bakthisaran R Tangirala R Rao Ch M Small heat shock proteins: Role in cellular functions and pathology Biochim Biophys Acta 2015 1854 291 319 25556000 13 Haslbeck M Vierling E A first line of stress defense: small heat shock proteins and their function in protein homeostasis J Mol Biol 2015 427 1537 1548 25681016 14 Schwartz NU Charcot-Marie-Tooth 2F (Hsp27 mutations): A review Neurobiol Dis 2019 130 104505 31212070 15 Evgrafov OV Mersiyanova I Irobi J Van Den Bosch L Dierick I Leung CL Schagina O Verpoorten N Van Impe K Fedotov V Dadali E Mutant small heat-shock protein 27 causes axonal Charcot-Marie-Tooth disease and distal hereditary motor neuropathy Nat Genet 2004 36 602 606 15122254 16 Juneja M Burns J Saporta MA Timmerman V Challenges in modelling the Charcot-Marie-Tooth neuropathies for therapy development J Neurol Neurosurg Psychiatry 2019 90 58 67 30018047 17 Almeida-Souza L Goethals S de Winter V Dierick I Gallardo R Van Durme J Irobi J Gettemans J Rousseau F Schymkowitz J Timmerman V Increased monomerization of mutant HSPB1 leads to protein hyperactivity in Charcot-Marie-Tooth neuropathy J Biol Chem 2010 285 12778 12786 20178975 18 Lee J Jung SC Joo J Choi YR Moon HW Kwak G Yeo HK Lee JS Ahn HJ Jung N Hwang S Overexpression of mutant HSP27 causes axonal neuropathy in mice J Biomed Sci 2015 22 43 26141737 19 Bouhy D Geuens T De Winter V Almeida-Souza L Katona I Weis J Hochepied T Goossens S Haigh JJ Janssens S Timmerman V Characterization of New Transgenic Mouse Models for Two Charcot-Marie-Tooth-Causing HspB1 Mutations using the Rosa26 Locus J Neuromuscul Dis 2016 3 183 200 27854215 20 Srivastava AK Renusch SR Naiman NE Gu S Sneh A Arnold WD Sahenk Z Kolb SJ Mutant HSPB1 overexpression in neurons is sufficient to cause age-related motor neuronopathy in mice Neurobiol Dis 2012 47 163 173 22521462 21 Hoke A Ray M Rodent models of chemotherapy-induced peripheral neuropathy ILAR journal 2014 54 273 281 24615440 22 LaPointe NE Morfini G Brady ST Feinstein SC Wilson L Jordan MA Effects of eribulin, vincristine, paclitaxel and ixabepilone on fast axonal transport and kinesin-1 driven microtubule gliding: Implications for chemotherapy-induced peripheral neuropathy Neurotoxicology 2013 37 231 239 23711742 23 Trinczek B Ebneth A Mandelkow EM Mandelkow E Tau regulates the attachment/detachment but not the speed of motors in microtubule-dependent transport of single vesicles and organelles J Cell Sci 1999 112 Pt 14 2355 2367 10381391 24 Dixit R Ross JL Goldman YE Holzbaur EL Differential regulation of dynein and kinesin motor proteins by tau Science 2008 319 1086 1089 18202255 25 Vossel KA Zhang K Brodbeck J Daub AC Sharma P Finkbeiner S Cui B Mucke L Tau reduction prevents Abeta-induced defects in axonal transport Science 2010 330 198 20829454 26 Adalbert R Milde S Durrant C Ando K Stygelbout V Yilmaz Z Gould S Brion JP Coleman MP Interaction between a MAPT variant causing frontotemporal dementia and mutant APP affects axonal transport Neurobiol Aging 2018 68 68 75 29729423 27 Stokin GB Lillo C Falzone TL Brusch RG Rockenstein E Mount SL Raman R Davies P Masliah E Williams DS Goldstein LS Axonopathy and transport deficits early in the pathogenesis of Alzheimer’s disease Science 2005 307 1282 1288 15731448 28 De Vos KJ Hafezparast M Neurobiology of axonal transport defects in motor neuron diseases: Opportunities for translational research? Neurobiol Dis 2017 29 Smith AL Teener JW Callaghan BC Harrington J Uhlmann WR Amyotrophic lateral sclerosis in a patient with a family history of huntington disease: genetic counseling challenges Journal of genetic counseling 2014 23 725 733 24763861 30 Ittner LM Fath T Ke YD Bi M van Eersel J Li KM Gunning P Gotz J Parkinsonism and impaired axonal transport in a mouse model of frontotemporal dementia Proc Natl Acad Sci U S A 2008 105 15997 16002 18832465 31 Alami NH Smith RB Carrasco MA Williams LA Winborn CS Han SSW Kiskinis E Winborn B Freibaum BD Kanagaraj A Clare AJ Axonal transport of TDP-43 mRNA granules is impaired by ALS-causing mutations Neuron 2014 81 536 543 24507191 32 Adalbert R Coleman MP Review: Axon pathology in age-related neurodegenerative disorders Neuropathol Appl Neurobiol 2013 39 90 108 23046254 33 Milde S Adalbert R Elaman MH Coleman MP Axonal transport declines with age in two distinct phases separated by a period of relative stability Neurobiol Aging 2015 36 971 981 25443288 34 Stepanova T Slemmer J Hoogenraad CC Lansbergen G Dortland B De Zeeuw CI Grosveld F van Cappellen G Akhmanova A Galjart N Visualization of microtubule growth in cultured neurons via the use of EB3-GFP (end-binding protein 3-green fluorescent protein) J Neurosci 2003 23 2655 2664 12684451 35 Baas PW Qiang L Neuronal microtubules: when the MAP is the roadblock Trends Cell Biol 2005 15 183 187 15817373 36 Yu W Baas PW Changes in microtubule number and length during axon differentiation J Neurosci 1994 14 2818 2829 8182441 37 Portran D Schaedel L Xu Z Thery M Nachury MV Tubulin acetylation protects long-lived microtubules against mechanical ageing Nat Cell Biol 2017 19 391 398 28250419 38 Sudo H Baas PW Acetylation of microtubules influences their sensitivity to severing by katanin in neurons and fibroblasts J Neurosci 2010 30 7215 7226 20505088 39 Reed NA Cai D Blasius TL Jih GT Meyhofer E Gaertig J Verhey KJ Microtubule acetylation promotes kinesin-1 binding and transport Curr Biol 2006 16 2166 2172 17084703 40 Chen S Owens GC Makarenkova H Edelman DB HDAC6 regulates mitochondrial transport in hippocampal neurons PloS one 2010 5 e10848 20520769 41 North BJ Marshall BL Borra MT Denu JM Verdin E The human Sir2 ortholog, SIRT2, is an NAD+-dependent tubulin deacetylase Mol Cell 2003 11 437 444 12620231 42 Benoy V Van Helleputte L Prior R d’Ydewalle C Haeck W Geens N Scheveneels W Schevenels B Cader MZ Talbot K Kozikowski AP HDAC6 is a therapeutic target in mutant GARS-induced Charcot-Marie-Tooth disease Brain 2018 141 673 687 29415205 43 Guo WT Naujock M Fumagalli L Vandoorne T Baatsen P Boon R Ordovas L Patel A Welters M Vanwelden T Geens N HDAC6 inhibition reverses axonal transport defects in motor neurons derived from FUS-ALS patients Nature communications 2017 8 44 Van Helleputte L Kater M Cook DP Eykens C Rossaert E Haeck W Jaspers T Geens N Berghe PV Gysemans C Mathieu C Inhibition of histone deacetylase 6 (HDAC6) protects against vincristine-induced peripheral neuropathies and inhibits tumor growth Neurobiol Dis 2018 111 59 69 29197621 45 Zhang L Liu C Wu J Tao JJ Sui XL Yao ZG Xu YF Huang L Zhu H Sheng SL Qin C Tubastatin A/ACY-1215 Improves Cognition in Alzheimer’s Disease Transgenic Mice Journal of Alzheimers Disease 2014 41 1193 1205 46 Wang Z Leng Y Wang J Liao HM Bergman J Leeds P Kozikowski A Chuang DM Tubastatin A, an HDAC6 inhibitor, alleviates stroke-induced brain infarction and functional deficits: potential roles of alpha-tubulin acetylation and FGF-21 up-regulation Scientific reports 2016 6 19626 26790818 47 Butler KV Kalin J Brochier C Vistoli G Langley B Kozikowski AP Rational design and simple chemistry yield a superior, neuroprotective HDAC6 inhibitor, tubastatin A J Am Chem Soc 2010 132 10842 10846 20614936 48 Benoy V Vanden Berghe P Jarpe M Van Damme P Robberecht W Van Den Bosch L Development of Improved HDAC6 Inhibitors as Pharmacological Therapy for Axonal Charcot-Marie-Tooth Disease Neurotherapeutics : the journal of the American Society for Experimental NeuroTherapeutics 2017 14 417 428 27957719 49 He JC Yao W Wang JM Schemmer P Yang Y Liu Y Qian YW Qi WP Zhang J Shen Q Yang T TACC3 overexpression in cholangiocarcinoma correlates with poor prognosis and is a potential anti-cancer molecular drug target for HDAC inhibitors Oncotarget 2016 7 75441 75456 27705912 50 Flipo M Charton J Hocine A Dassonneville S Deprez B Deprez-Poulain R Hydroxamates: relationships between structure and plasma stability J Med Chem 2009 52 6790 6802 19821586 51 Gilley J Coleman MP Endogenous Nmnat2 is an essential survival factor for maintenance of healthy axons PLoS Biol 2010 8 e1000300 20126265 52 Milde S Gilley J Coleman MP Axonal trafficking of NMNAT2 and its roles in axon growth and survival in vivo Bioarchitecture 2013 3 133 140 24284888 53 Santo L Hideshima T Kung AL Tseng JC Tamang D Yang M Jarpe M van Duzer JH Mazitschek R Ogier WC Cirstea D Preclinical activity, pharmacodynamic, and pharmacokinetic properties of a selective HDAC6 inhibitor, ACY-1215, in combination with bortezomib in multiple myeloma Blood 2012 119 2579 2589 22262760 54 Maday S Wallace KE Holzbaur EL Autophagosomes initiate distally and mature during transport toward the cell soma in primary neurons J Cell Biol 2012 196 407 417 22331844 55 Misgeld T Kerschensteiner M Bareyre FM Burgess RW Lichtman JW Imaging axonal transport of mitochondria in vivo Nat Methods 2007 4 559 561 17558414 56 Gibbs KL Kalmar B Sleigh JN Greensmith L Schiavo G In vivo imaging of axonal transport in murine motor and sensory neurons J Neurosci Methods 2016 257 26 33 26424507 57 Zhang B Maiti A Shively S Lakhani F McDonald-Jones G Bruce J Lee EB Xie SX Joyce S Li C Toleikis PM Microtubule-binding drugs offset tau sequestration by stabilizing microtubules and reversing fast axonal transport deficits in a tauopathy model Proc Natl Acad Sci U S A 2005 102 227 231 15615853 58 Lin MY Cheng XT Tammineni P Xie Y Zhou B Cai Q Sheng ZH Releasing Syntaphilin Removes Stressed Mitochondria from Axons Independent of Mitophagy under Pathophysiological Conditions Neuron 2017 94 595 610 e596 28472658 59 Kaieda A T M Daini M Nara H Yoshikawa M Ishii N Hidaka K Oxadiazole derivatives useful as HDAC inhibitors Patent WO2017014321 2017 60 Andrews S Gilley J Coleman MP Difference Tracker: ImageJ plugins for fully automated analysis of multiple axonal transport parameters J Neurosci Methods 2010 193 281 287 20869987

PMC007xxxxxx/PMC7614733.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 20810170R Soc Probl Soc Probl Social problems 0037-7791 1533-8533 37426294 7614733 10.1093/socpro/spac015 EMS177419 Article Rights vs. Lived Realities: Women’s Views of Gender Equality in Relationships in Rural South Africa https://orcid.org/0000-0002-3760-6271 Sennott Christie * https://orcid.org/0000-0001-7570-0763 Kane Danielle Purdue University * Corresponding Author: Department of Sociology, Purdue University, 700 W. State St., West Lafayette, IN 47906, USA. csennott@purdue.edu 3 3 2022 26 6 2023 07 7 2023 2022 spac015This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. South Africa’s Constitution is among the world’s most ambitious in promoting gender equality, but the country continues to be marked by inequality and gender-based violence. Given this context, we analyze 43 interviews with Black women aged 18-55 in rural South Africa to explore how the constitutional ideal of gender equality—or “50/50”—has been interpreted and applied in women’s intimate relationships. Overall, we found that inequality and gender hierarchy were common in relationships. Women relied on two logics to explain the persistence of inequality in their relationships. First, women offered ideological support for gender norms supporting hierarchy by linking 50/50 to the abandonment of culture, tradition, and respect. Second, women viewed reaffirmation of gender inequality within relationships as a pragmatic way to avoid men’s violence and infidelity, thus protecting women from abandonment and HIV. Women’s views about equality in relationships were shaped by dominant gender norms, precarity in the local political economy, and the risks of violence and HIV/AIDS. Our findings expand theories of social change by highlighting how not only longstanding social norms, but also local political-economic and health conditions can influence views of equality and ultimately the local adoption or dismissal of international standards of rights and equality. gender inequality Africa rights HIV violence pmcGender equality and the empowerment of all women and girls is the fifth United Nations Sustainable Development Goal, which countries around the world are working toward meeting by 2030 (United Nations 2020). Despite progress over the past several decades, gender inequality in both the public and private spheres remains a significant social problem across the globe (World Bank 2020). Thus, scholars and policymakers have called for more attention to gender equality as a public good, for reducing health disparities, and as a key driver of economic development (see for example Baten and de Plejit 2019; Helman and Ratele 2016; Kabeer 2016). The current study examines women’s views of gender equality in intimate relationships in South Africa, a country in which progressive democratic ideals coexist with extreme inequality (World Bank 2019). The end of the apartheid system of government in 1994 witnessed the adoption of a new South African Constitution, stipulating not only racial equality but also a host of other progressive social policies, including gender equality (see Hassim 2006, 2018). While many new democracies have taken only nominal steps toward increasing gender equality, South Africa aimed to create “a truly inclusive polity…[that] involved a remarkable effort to confront gender inequalities” (Seidman 2001:220). Given the governmental support for equal rights, there was an assumption that “real change in a feminist direction was possible through the state” (Hassim 2003:505). Indeed, explicit governmental support for gender equality has continued, evidenced most recently by the 2020 passage of the Emergency Response Action Plan and the introduction of several new bills to combat gender-based violence and femicide.1 Despite the institutional support for gender equality, however, there remains a vast divide between “South Africa’s formal status as a ‘rights paradise’ and the grim realities of people’s daily lives” (Robins 2008:2). Following the transition to democracy and the widespread promotion of women’s rights, researchers documented a “crisis in masculinity” (Walker 2005:226) and backlash against women, contributing to high rates of intimate partner violence (IPV) and HIV (Jewkes et al. 2010; Seedat et al. 2009). Though significant progress has been made in HIV testing and treatment programs, South Africa continues to have the world’s largest HIV/AIDS epidemic (UNAIDS 2018) with a much higher prevalence among women (26 percent) than men (15 percent) (HSRC 2018). Additionally, estimates suggest nearly one-third (30 percent) of women in South Africa have experienced recent IPV (UNAIDS 2018). These social and health epidemics have prompted numerous interventions aimed at curbing gender-based violence and reducing women’s risk of HIV, in part by altering gender norms that position women as subordinate to men (Dworkin et al. 2012; Gibbs et al. 2015, 2020; Jewkes and Morrell 2010; Willan et al. 2020). In this study, we explore the disconnect between South Africa’s governmental commitment to equal rights and women’s views of gender equality in intimate relationships, referred to colloquially as “50/50”. To do so, we analyze 43 in-depth interviews with Black women in rural Mpumalanga Province, South Africa. Our research is grounded in work that addresses how international discourses of rights get (re)interpreted at the local level (e.g., Kurzman et al. 2019; Thomas 2007; Unnithan and Pigg 2014) by examining how women interpret and apply ideals of equal rights in their relationships. This approach also aligns with evidence from sub-Saharan Africa that the backlash to women’s rights often plays out within relationships (e.g., Jewkes et al. 2010; Pierotti, Lake, and Lewis 2018; Walker 2005; Wyrod 2016). Overall, we found that inequality and gender hierarchy were common in intimate relationships. Women relied on two logics to explain the persistence of inequality in their relationships. First, women offered ideological support for longstanding norms supporting gender hierarchy by linking 50/50 to the abandonment of culture, tradition, and respect. Second, women viewed the reaffirmation of gender inequality in relationships as a pragmatic way to avoid men’s violence and infidelity, thus protecting women from abandonment and HIV. Taken together, women’s views about equality in relationships were shaped by dominant gender norms, precarity in the local political economy, and the significant risks of violence and HIV. Our findings highlight the importance of social and cultural factors as well as the material conditions that structure life in rural areas in how ideals of equal rights are adopted or dismissed in relationships. This study is one of the first in rural sub-Saharan Africa to ask women directly about their views of gender equality in relationships. Although research from the United States has documented attitudes among men and women signaling a stalled gender revolution (e.g., England, Levine, and Mishel 2020), there are few comparable studies of women in sub-Saharan Africa, where equal rights were established more recently and societies continue to undergo rapid social change. Additionally, although there is a growing body of research examining men’s responses to women’s rights in sub-Saharan Africa (e.g., Dworkin et al. 2012; Dworkin, Fleming, and Colvin 2015; Pierotti et al. 2018; Wyrod 2008), few studies have analyzed the factors influencing women’s own views of gender equality in relationships (see Pettifor et al. 2012; Wyrod 2016 for exceptions). The current study contributes to this work by focusing on a distinct rural setting marked by structural disadvantage in which narratives about equal rights and a reassertion of tradition both compete and intertwine (Ainslie and Kepe 2016; Mathis 2011), allowing us to parse out the local dynamics that are often obscured in national-level survey data. Additionally, our in-depth interviews provide insight into the processes through which ideals of equality are interpreted and applied within the context of women’s relationships. This approach is valuable for better understanding the reasons why legislating gender equality does not always translate into changes in women’s lived realities. Background Explanations for the persistence of gender inequality often hinge on longstanding social and cultural norms that position women as subordinate to men (see for example Ridgeway 2011). More broadly, norms scholars argue that because social norms—or group-level expectations for behavior and the sanctions for breaking them—are fundamental to societies, they have a way of persisting and maintaining inequality even in the face of significant social change (Horne and Mollborn 2020). Individual beliefs and behaviors are influenced by norms, which become internalized (Horne and Mollborn 2020; Risman 2018). Once internalized, these expectations for behavior become a routine and taken for granted part of how individuals evaluate their own actions and the actions of others, similar to what Risman calls “the cultural component of the social structure” or “the interactional expectations that each of us meet in every social encounter” (2004:433; see also Giddens 1984; Horne and Mollborn 2020). Although structural changes promoting gender equality might be expected to alter gender hierarchy and interactions between men and women, Ridgeway argues individuals bring “trailing gender beliefs” into new social contexts, reinscribing the status assumptions tied to these beliefs onto new social arrangements and ultimately reinforcing existing inequalities (2011:28). This suggests that analyses centering on how individuals interpret and apply abstract ideals of gender equality into the fabric of their everyday lives may be particularly fruitful for understanding the persistence of inequality in the face of social change (Risman 2004), such as the documented disconnect between policy change in South Africa and resistance on the ground. This also allows us to see how women employ agency in making decisions to maximize their security and wellbeing within restrictive cultural and social norms (Kandiyoti 1988; Mathis 2011; Mphaphuli and Smuts 2021; Willan et al. 2020), rather than viewing women simply as victims of inequality. Ideological Support for Inequality Normative systems of hegemonic masculinity and emphasized femininity reinforce gender inequality by dictating complementary and hierarchical relations between men and women (Connell 1987, 1995). More specifically, although they are not monolithic, hegemonic constructions of masculinity across South Africa emphasize dominance, control, sexual prowess, and economic provision (Hunter 2005; Jewkes and Morrell 2012), whereas dominant constructions of emphasized or acquiescent femininity emphasize deference, modesty, respectability, and caretaking (Bhana 2016; Gqola 2007; Harrison 2008; Jewkes and Morrell 2012; Sennott and Mojola 2017). Because these norms reinforce gender hierarchy and inequality, women’s fulfilment of the social and cultural expectations surrounding femininity effectively subordinates them to men (Bhana 2016; Harrison 2008; Jewkes and Morrell 2012; Mathis 2011; Mphaphuli and Smuts 2021; Sennott and Mojola 2017). The persistence of dominant gender norms has contributed to high rates of IPV and HIV, prompting several interventions across South Africa. For example, the One Man Can Programme, implemented by the South African NGO Sonke Gender Justice, and the Stepping Stones, Creating Futures Programme, implemented by the South African Medical Research Council and Project Empower, focus on altering norms, beliefs, and behaviors related to masculinity (see van den Berg et al. 2013; Dworkin et al. 2012; Gibbs et al. 2015, 2020; Ratele 2015; Treves-Kagan et al. 2020; Willan et al. 2020). Nonetheless, research has documented resistance among men to ideals of equality due to the assumed tradeoff between improvements in women’s status and a loss of men’s power and respect (Dworkin et al. 2012; Posel 2004; Walker 2005). Thus, equality is viewed through a zero-sum approach in which women’s gains equal men’s losses. For example, research from South Africa found that men viewed women’s empowerment and increasing access to jobs and money as disruptive to gender relations and disempowering for men (Dworkin et al. 2012). Similar dynamics have been noted in other sub-Saharan African settings such as Uganda where men reported feeling victimized and disempowered due to efforts to promote women’s rights and equality (Wyrod 2008, 2016) and in the Democratic Republic of Congo where men who participated in gender sensitization workshops were somewhat more willing to share domestic tasks, but not power (Pierotti et al. 2018). Taken together, these studies highlight how men’s compliance with hegemonic masculinity can impede the acceptance of women’s rights and gender equality. Tensions in Women’s Support for Equality Research has shown that dominant gender norms can also influence women’s beliefs and actions regarding equality in relationships. Specifically, women may implicitly support hegemonic masculinity through their enactments of femininity (Jewkes and Morrell 2012; Schippers 2007). For example, women who idealize men who fulfill the provider role and therefore encourage men to enact this behavior buttress the power of norms that reinforce women’s dependence on men (Bhana and Patttman 2011; Pettifor et al. 2012; Talbot and Quayle 2010). Though “modern” or “resistant” femininities that reject gender hierarchy and women’s submission are becoming more prevalent across South Africa, women who push for equality are typically viewed as challenging dominant norms, thereby incurring rebuke from families, friends, and communities (Dworkin et al. 2015; Jewkes and Morrell 2012; Pettifor et al. 2012; Sennott and Angotti 2016). This may encourage women to temper their support for equality in relationships or dismiss changes that would threaten what is often perceived as the “natural” gender order. For example, although women in urban Uganda were critical of men’s abuses of power, they remained emphatic that men should have more power and authority because these traits were given to men “by God” (Wyrod 2008). Further, women may be compelled to abide by norms of acquiescent femininity if there are few other options or if doing so provides women with respect, status, and security (Graham 2016; Jewkes and Morrell 2012; Kandiyoti 1988; Sennott and Mojola 2017). These dynamics put pressure on women to adhere to dominant gender norms dictating their own submission (see also Campbell and Nair 2014; Ranganathan et al. 2021). We build on this work by focusing specifically on how women view and navigate ideals of equal rights within their relationships. In doing so, we aim to identify local factors that create tension in women’s support for gender equality. This is critical as scholars have argued that universal notions of rights—such as those adopted by the South African government—lack grounding in women’s lived experiences; therefore, accounting for local factors and structures is necessary to understand why legislated rights may not translate into equality on the ground (see Undie and Izugbara 2011). Further, research from South Africa has highlighted the importance of recognizing “the everyday complexities of women’s lives, including the patriarchal social structures that constrain them” (Ranganathan et al. 2021:3). Thus, despite the government’s commitment to gender equality across all realms of society—including the family—local cultural frameworks may still encourage the belief that men deserve more rights and power than women (Thomas 2007). In rural South Africa, high levels of economic precarity (Blalock 2014; Statistics South Africa 2020) and the ubiquity of HIV/AIDS (Gómez-Olivé et al. 2013) are important local conditions likely to affect how women weigh equality in relationships vis-à-vis their own security and wellbeing. Additionally, because of its prevalence, IPV constitutes a significant public health concern and is closely tied to men’s fulfilment of hegemonic masculinity norms and the risk of HIV (Dunkle et al. 2004; Jewkes et al. 2010; Treves-Kagan et al. 2020). Research from South Africa has shown that defying dominant norms can be detrimental for women’s security, respectability, health, and wellbeing in the context of economic precarity and uncertainty due to HIV/AIDS (Ratele 2015; Sennott, Madhavan, and Nam 2020). Building on this work, we examine women’s views about gender equality in relationships as a window into the local norms, political-economic structures, and health uncertainties that shape why women’s rights and equality are adopted or dismissed. Methods This study is nested in the Agincourt Health and Socio-Demographic Surveillance System (Agincourt) site in rural Mpumalanga Province, South Africa. Agincourt incorporates approximately 110,000 individuals in 21,000 households in 31 villages and has conducted an annual census of demographic and health information since 1992 (see www.agincourt.co.za). The majority of Agincourt residents are of the Shangaan ethnic group and one-third are of Mozambican origin. The area is a former apartheid-era “homeland” where Black South Africans were forced to resettle and endured the compounding hardships of inadequate healthcare, employment, education, and infrastructure (Worden 2007). The area still has limited resources: water is provided through neighborhood taps, there is no formal sanitation system, and electricity is unreliable and expensive (Kahn et al. 2012). Challenges in the local political economy have rendered most paid work temporary, informal, and underpaid, leading to high unemployment (Blalock 2014; Statistics South Africa 2020). These economic challenges push many men and increasingly women to out-migrate for work (Blalock 2014). HIV/AIDS is endemic in the area: prevalence is nearly 20 percent among adults, with a much higher rate among women (24 percent) than men (11 percent) (Gómez-Olivé et al. 2013). Our findings reflect how these local conditions influence rural women’s beliefs about gender equality. The larger study from which this analysis is drawn was designed to explore rural South African women’s views and experiences related to relationship formation, marriage processes, gender equality in relationships, and health and wellbeing (see also Sennott et al. 2020). In 2015, a research team comprised of the first author and three local research assistants conducted in-depth interviews in XiShangaan/XiTonga (local language) with 43 Black women aged 18-55 in Agincourt. All participants provided written informed consent. The interview protocol was semistructured, allowing for probing on issues unique to participants. Interviews lasted 1-2 hours, were audio-recorded, and translated and transcribed by the research team. The findings drawn from the interview data were supported by the first author’s extensive experiences working in this area of South Africa. The study received ethical approval from the Mpumalanga Province Department of Health, local leadership, and university institutional review boards in the United States and South Africa. We use pseudonyms to protect the identities of participants. Participants were recruited through a quota snowball sampling technique based on recommendations from a Community Advisory Group that advises on best practices in the site, local research assistants’ social networks, and participants’ referrals of the study to other women. Because marriage is an important marker of women’s status and wellbeing in Agincourt, the aim was to recruit roughly equal numbers of women based on relationship status and bridewealth (“lobola” in Shangaan). The analytic sample included 13 women (30 percent of the sample) who were married with lobola paid, 13 women (30 percent) who were cohabiting with a partner without lobola, and 17 women (40 percent) who were neither married nor cohabiting (labeled “single” below). Participants were 34 years old on average. Nearly 70 percent had finished secondary school and 26 percent were employed in either the formal or informal labor market. Additionally, one woman was volunteering at a preschool, one was in the process of opening a small business, and one was in school. One woman was pregnant and participants had 2.6 children on average (range 0-9), consistent with fertility patterns in Agincourt (Williams et al. 2013). We used a combination of deductive and inductive coding strategies to analyze the data (Charmaz 2001; Strauss and Corbin 1990). We first engaged in structured coding by interview question, focusing primarily on the question: “Do you think men and women should be equal in their relationships? Why/why not?” We also analyzed several follow-up questions addressing the good and bad things about 50/50; whether women’s own relationships were 50/50; and significant others’ views of 50/50. Thus, interviews focused on how ideals about gender equality were interpreted and applied in intimate relationships and did not engage in broader discussions about how rights and equality function across other societal spheres (see Hames 2006 for a discussion of the latter). Although the term “50/50” was introduced by the South African government and is often used publicly to refer to gender representation in government and employment, it has been appropriated at the local level to refer to equality in relationships and households, as several studies have reported (e.g., Dworkin et al. 2012; Hunter 2010; Pettifor et al. 2012). After structured coding, we engaged in inductive coding based on emergent themes and found an overarching pattern of women reaffirming gender hierarchy and inequality in relationships. Based on this theme, we returned to the data to reanalyze and recode the transcripts. Through this iterative process, we found no systematic differences in views of gender equality based on women’s socio-demographic characteristics. The quotes provided below illustrate the most prevalent themes across interviews. To aid in readability of the quotes, we made minor edits to grammar and include clarifying phrases in brackets. Findings We found that inequality and gender hierarchy were common in intimate relationships. Women relied on two logics to explain the persistence of gender inequality in their relationships. First, 86 percent (n=37) of the sample offered ideological support for longstanding norms supporting gender hierarchy by linking 50/50 to the abandonment of culture, tradition, and respect. This explanation drew on gender essentialist scripts defining differences between men and women and men as “natural” and unmalleable. Second, 77 percent (n=33) viewed the reaffirmation of gender hierarchy in relationships as a pragmatic way to avoid men’s violence and infidelity, thus protecting women from abandonment and HIV. This explanation highlighted local political-economic uncertainties and health risks and emphasized that 50/50 relationships were available only to women who were highly educated and employed, and thus not economically dependent on men. Overall, 67 percent (n=29) provided both ideological and pragmatic explanations for reaffirming gender inequality in relationships and 95 percent (n=41) relied on at least one of these logics. Four participants (9 percent) were ideologically supportive of gender equality in relationships but did not fully support it because of pragmatic barriers. Only two participants2 offered full support of gender equality in relationships. Taken together, women’s views about equality in relationships were shaped by dominant gender norms, precarity in the local political economy, and the significant risks of violence and HIV. Ideological Explanations for Gender Inequality Women often explained their reaffirmation of gender inequality in relationships by relying on scripts essentializing differences between men and women, reinforcing gender hierarchy, and linking women’s fulfilment of dominant gender norms to respect, culture, and tradition. When asked whether men and women should be equal in relationships, Alice (age 37, married) said: “No, they are not equal. A man is the head of the family. Even when a man and a woman are both working, a man will always be a man.” Similarly, Alina (age 44, married) supported gender hierarchy by emphasizing her partner’s position as household head and women’s duty to comply with the gender order: We are not equal, and he [her partner] also has more power than me because I am a woman. So, a woman must always humble herself to her man. A man will always be a man and it will never change. Even if a woman can have more money, it will never change. A man will always be head of the family. Alice’s and Alina’s comments emphasize ideological support for gender hierarchy over pragmatic concerns in arguing that even women who are employed—and may have some measure of economic independence—must “humble” themselves or be dutiful to their partners. This belief reinforces men’s “natural” position over women in relationships, as reflected in Alina’s claim that her husband has more “power” than her as a woman. Further, each woman’s quote followed a gender essentialist script by reinforcing the permanence of gender difference (“always”, “never”). As Tengisa (age 19, single) succinctly summed up: “It’s [50/50] bad because a man will always be a man, and [a] woman will always be a woman. They will never be equal.” The ideological support women offered for gender hierarchy and inequality in relationships was supported by dominant gender norms, which were held sacred among community members and buttressed by local institutions, including churches. For example, women frequently invoked the Bible or their pastor’s teachings in explaining their views of 50/50 in relationships. As Nyeleti (age 26, cohabiting) put it: Nyeleti: Our pastor is saying we must not live by 50/50. A man should always remain a man. He is the head of the family. He [the pastor] is saying in the beginning God created a man and the second one was a woman. Women are after men in everything. Interviewer: How do women feel about this preaching? Nyeleti: We do understand and accept that. We also pray to God to help us live according to the pastor’s preachings. By teaching that gender essentialism and hierarchy are ordained by God, some pastors participate in perpetuating “an inherited matrix of gender relations” (Burchardt 2010:77) that reinforces inequality. Women may be motivated to adhere to these teachings as a way of maintaining a valued identity as a pious, humble, and respectable woman. Dominant gender norms were also reinforced by the community and served as a form of social control. As Rhandzu (age 48, cohabiting) observed: “People don’t feel good seeing a man sweeping…seeing him helping me push a wheelbarrow.” The stigmatizing of a man helping a woman push a wheelbarrow underscores the gendered nature of domestic responsibilities such as farming and gathering water (both assigned to women) and highlights the community’s active role in sanctioning behaviors that challenge gendered expectations. Ellen (age 46, cohabiting) described men’s policing of each other regarding hierarchy in their relationships. When asked whether she and her partner practiced 50/50, Ellen replied: I don’t use 50/50 with my husband3. He doesn’t like it…He said he won’t live it in his house, as he was telling his friends. I listened to him talk while telling his friends about their wives who are on 50/50. He said those men are giving their wives permission to be on 50/50 just because they [the men] are weak. It means their houses are ruled by their wives. I supported his statement as he was right. There are few expectations for men’s involvement in household labor, and men who engage in tasks coded as feminine are regularly stigmatized and sanctioned (van den Berg et al. 2013; Dworkin et al. 2012). In describing men in 50/50 relationships as “weak” and “ruled by their wives”, Ellen’s husband upbraids his peers for not adhering to hegemonic masculinity norms and failing to uphold their power as men. This type of sanctioning can be an effective form of social control by bringing behavior back into alignment with dominant norms and serving as a cautionary tale (see also Horne and Mollborn 2020; Sennott and Angotti 2016). The combination of religious teachings on gender essentialism and community sanctioning reinforcing gendered expectations provided women a strong incentive to conform to dominant norms. Another ideological explanation for women’s reaffirmation of gender inequality in relationships hinged on culture, respect, and tradition. Hlekani (age 23, single) described how 50/50 contradicts the local ideal of respectful womanhood and the fulfillment of tradition: 50/50 is proof that there is no more respect. A woman will tell her husband, “Hey, can you come with that cup?” Meanwhile in the past it was not like that. I have seen from my mother…She was cooking and after that she will bring the food to my father putting it on the smaller handmade table. She will kneel down [a sign of respect] and say, “You will eat the food, father.” But nowadays things have changed. They [women] don’t kneel and they just stand with food on the plate. That’s why there is a difference with our cultures. If a man can marry a Shangaan woman like us, we are well known for respect…Women told themselves that they are free. They don’t want to be under pressure…Like I said men need women who respect, women like Shangaans who are humble. A woman’s dutifulness to her partner is an important tenet of acquiescent femininity in this context, and as illustrated by Hlekani’s comment that Shangaan women are “known for respect,” closely tied to women’s cultural identities. Further, by linking ideals of respectful feminine behavior to women “in the past,” Hlekani underscores the importance of fulfilling traditions to women attaining valued cultural identities, as opposed to women “nowadays” who are neither respectful nor cultural women. Thus, it was common for women to tie the “modern” practice of 50/50 to the erosion of culture, respect, and tradition. Finally, ideological explanations for gender inequality in relationships were closely tied to the difficulty the local context presented for cultural change, as described by Glory (age 52, married): In my community it’s rare to see people accepting 50/50…I think in our [rural] place like this one, it’s difficult to live the lifestyle of 50/50. 50/50 only applies to the location [urban areas] where most of the people are educated and they are working…I think the other thing that made us to be like this is the culture. Changing from the culture is still a challenge. By distinguishing educated women in urban areas from rural women who embody cultural values and traditions, Glory distances herself and her community from 50/50. Like Hlekani, Glory claims a cultural and moral identity for rural women who embody dominant gender norms rather than pursuing 50/50. That is, women like Hlekani and Glory are staking a claim to a valued identity, and this act of prioritizing culture and tradition can be read as a form of agency (Mphaphuli and Smuts 2021). However, this agency cannot be understood apart from the significant constraints of the local context, as described below. Pragmatic Explanations for Gender Inequality Pragmatic explanations for gender inequality in relationships focused on the strategies women employed to protect themselves from violence, HIV, and abandonment amidst local economic precarity. These explanations thus weighed women’s concerns about their health, wellbeing, and economic survival against concerns about gender imbalances in relationships. In doing so, women highlight the formidable structural constraints in the local political economy and health environment that serve as barriers to equality in relationships. This type of explanation is pragmatic in that women focused on the potential consequences of 50/50 rather than ideological motivations for dismissing 50/50. Research demonstrates that challenges to men’s masculine identities and essentialized rights to power and control—such as through seeking gender equality in relationships—are often associated with efforts to shore up masculinity in other ways (Dworkin et al. 2012; Pierotti et al. 2018) including violence (Jewkes and Morrell 2012; Mphaphuli and Smuts 2021). Indeed, participants frequently linked 50/50 directly to the risk of IPV. For example, Gavaza (age 43, married) said violence was common in 50/50 relationships because “Most women [who use 50/50] want to be a boss to their husbands, and they end up fighting because men are the leaders of the household.” Similarly, Basani (age 20, single) said that a man may beat his wife “because she forces him to live by 50/50, meanwhile he doesn’t understand it.” These comments show the direct relationship women perceived between 50/50 and men’s violence. In another example, Fanisa (age 31, married) explained that 50/50 was related to violence because women in these relationships “don’t have a good approach. They start to undermine their husbands and think as if they are better than them.” Thus, in Fanisa’s view, women who used 50/50 were disrupting the normative gender order and therefore challenging men’s power. She went on to describe how this situation could result in violence or even death: Fanisa: From there you will hear that they were fighting, and the woman has [moved] back to her parents’ home. Or you will hear that she was beaten to death by her husband. Interviewer: Is that common nowadays? Fanisa: This is very common…That’s where you will see that [violence] is not happening…in America only but even here locally things are happening. Women must be submissive to their husbands. The threat of violence against women who support 50/50 in relationships serves as a form of social control that motivates women to abide by dominant gender norms. Further, this explanation elucidates the common view that 50/50 leads to violence because of women’s actions. Rather than blaming men for being violent, Gavaza, Basani, and Fanisa attribute men’s violence to a woman’s approach to implementing 50/50 by saying she “forces him” or “wants to be the boss.” This perspective aligns with a zero-sum approach to equality in relationships and leaves little ground from which women might negotiate for greater power. Women also focused on how 50/50 could prompt men to have outside sexual affairs, which led to a heightened risk of HIV. For example, Alina (age 44, married) said: “…that [50/50] will lead a man to have an affair outside, where he will have peace of mind.” Alina’s comment about “peace of mind” highlights the belief that a man would leave a woman who was interested in 50/50 in favor of a woman who was more dutiful. Similarly, Priscilla (age 36, cohabiting) described how the push for 50/50 could encourage a man to seek a partner who would show him respect by attending to his needs: If I’m abusing my man here at home by saying he needs to help me by doing the household chores…he will go out and find a charming girl and they will fall in love. From there he will pay a visit to that girl and he will find that there is no 50/50. She is giving him water to bathe and [he is] drinking tea while in bed and that girl is taking care of everything. Do you think he can come back to you, the woman of 50/50? I don’t think so. Men don’t want to be abused…He knows that with you he is working until late, and he doesn’t have time to rest, but with that girl he is able to rest. As highlighted by Alina and Priscilla, men’s infidelity was characterized as an expected response to women’s attempts to have a more equal relationship. However, linking 50/50 to the “abuse” (Priscilla) of men, which was common across the sample, obfuscates the real risks of violence and HIV women face in seeking equality. Nearly half of individuals in Agincourt are HIV positive by the end of their reproductive years (Gómez-Olivé et al. 2013), therefore, remaining monogamous is an important way couples can avoid infection. This was reflected by Samara (age 38, married) who described a man’s reaction if a woman in a 50/50 relationship refuses sex: “If your husband wants to have sex with you and then you refuse, he will go out and start cheating, and it will hurt you and you will blame him. He will come back with infections [HIV], and he will infect you.” Comments like those from Alina, Priscilla, and Samara reinforce how the interrelated threats of infidelity and HIV incentivized women to adhere to norms dictating their submission. However, fulfilling the tenets of acquiescent femininity does not necessarily safeguard women from a partner’s infidelity or HIV transmission (see for example Jewkes and Morrell 2010; Sennott et al. 2020). Moreover, this approach ultimately relies on a gendered script that holds women responsible for how men treat them (see Gqola 2015; Mphaphuli and Smuts 2021). Lastly, women offered a pragmatic explanation for reaffirming gender inequality in relationships by arguing 50/50 is only possible for women with a measure of economic independence. For example, Ripfumelo (age 35, married) said 50/50 was only good in couples where women were employed: With us, as we are unemployed, it is not good to use it [50/50]. Even where the man is working and the woman is not, she should do everything in the household because she knows exactly what to do…Maybe I would use 50/50 if I was employed, but the fact is that at home I was taught about how I should behave when I am married so I won’t change myself. Underscoring the significant challenges in the local political economy, both Ripfumelo and her husband were unemployed, which she viewed as a barrier to 50/50. This was common among the sample. Although Ripfumelo indicated she might seek more equality in her relationship if she was working, she ultimately falls back on her gendered responsibilities to her husband (“how I should behave when I am married”). In doing so, she prioritizes fulfilling norms of acquiescent femininity, something she can control, over pursuing a more equal partnership, an outcome she cannot fully control. Moreover, the former is a reliable path to respectability, regardless of whether she is working, whereas the latter may increase risk, as discussed above. Twelani (age 28, single) illustrates these dynamics in explaining how educated and uneducated women differ in terms of respect and 50/50: Twelani: A woman who is not educated is full of respect, and if you want to see a man being happy just show him some respect, he will always be happy. And what I have realized is a woman who is not educated doesn’t live by 50/50 because she knows that she is not educated and not working. Once her husband can leave her, she will have nowhere to go, so that is why she will make sure that she respects her husband in all ways. Interviewer: What about women who are educated? Twelani: Mostly they don’t have respect to their men, do you know why? They have money. They are working, so there is nothing their man can say, because they have everything…Even when they can divorce, they will move on with their lives because they have money. Twelani’s comments highlight the common perception that education and employment provide women with the space to negotiate for 50/50 as well as the economic resources and independence to survive if the relationship dissolves. These options are not necessarily open to women without the same resources, who instead may be compelled to rely on deference to men and the fulfillment of tradition to ensure economic stability and avoid abandonment (see also Mathis 2011; Sennott and Mojola 2017). Twelani and Ripfumelo also emphasize that women who use 50/50 lack respect. Thus, although education and employment may provide women with independence and the leverage to have more equal partnerships, doing so challenges norms and traditions surrounding respect and thus the identities of women without those resources. Ultimately, the pragmatic explanations for women’s reaffirmation of gender inequality in relationships underscore how the risks of violence, HIV, and abandonment encourage women to strategize about how to ensure their own security, safety, and stability. Finally, of the 33 women who offered pragmatic explanations for reaffirming gender inequality, four did not ideologically support gender hierarchy and inequality in relationships4. Though their ideological views differed, their pragmatic explanations for reaffirming inequality echoed those above, namely the risks of abandonment (Maria, Olive), violence (Khongelani, Maria, Rhandzu), and infidelity and HIV (Khongelani). For instance, although Olive (age 41, single) felt that men and women “should be equal,” she said 50/50 was risky because: Men don’t want 50/50. You cannot say “cook” or “take care of the children while I go somewhere.” Men don’t want that. They want to be free to walk and drink with friends. So, I don’t want to take chances…You will hear men who fought with their wives saying “It’s all about 50/50. She is doing that because she got powers. Just leave her like that [divorce her], you will see where she is going [back to her parents’ home].” In saying she does not “want to take chances” by seeking a more equal relationship, Olive’s comments reflect the tension between ideological support for 50/50 and the risks of pursuing equality. Maria (47, cohabiting) offered a similar explanation for avoiding 50/50, even though she ideologically supported gender equality in relationships: “Men always want to be on top…If you tell him to come back home early, he will tell you that you can’t set his limits as he is the family head and it is his home…He will say, ‘This is my home. Go back where you are coming from.’“ Finally, despite her ideological support of 50/50 and her belief that men and women “need to share equally,” Khongelani (age 34, single) described why 50/50 was not practically feasible: It’s the men. They are cheating…We are not feeling well about this but men don’t change…We are not equal as men are beating women for nothing. It’s like with me, the way [her previous boyfriend] treated me was not 50/50…Men are cheating and forgot that nowadays there is illness. I mean the illness of HIV. They are going to infect us. For these women the risks of abandonment, infidelity, HIV, and violence, ultimately outweighed their ideological support for 50/50. These findings highlight the formidable barriers to embracing 50/50 given local conditions. Discussion and Conclusions This study examined women’s views about gender equality in relationships in South Africa, a country in which democratic ideals and a governmental commitment to equality exist alongside extreme levels of inequality, gender norms that essentialize difference, a resurgence of tradition, and high rates of IPV and HIV (Ainslie and Kepe 2016; Mathis 2011; UNAIDS 2018; World Bank 2019). Given these contradictions, a growing body of research has examined the social and cultural factors upholding gender inequality, often focusing on norms about hegemonic masculinity and the backlash among men to women’s rights (van den Berg et al. 2013; Dworkin et al., 2012; Gibbs et al. 2015, 2020; Ratele 2015; Walker 2005; Willan et al. 2020). This has left a dearth of evidence about how women themselves—the targets of numerous governmental policies and programmatic efforts—view gender equality in relationships. Our findings suggest that women in rural South Africa accounted for gender inequality in relationships on both ideological and pragmatic grounds. First, women based their ideological support of gender hierarchy and inequality on longstanding norms of gender essentialism that are widely supported by local institutions, community members, and women’s social networks. Thus, for many women, fulfilling the tenets of acquiescent femininity was closely tied to respect, culture, and tradition and thus their duty as women. These findings echo research from across sub-Saharan Africa showing that women in rural areas in particular are likely to defer to dominant norms due to their ideological commitments and the status these ideals provide, but also because of a lack of viable alternatives (Evans 2018; Pettifor et al. 2012; Wyrod 2008, 2016). For example, Evan’s (2018) research from Zambia showed that women in rural areas subscribed to dominant gender ideologies and restricted their aspirations to being wives and mothers largely because of limited exposure to alternative discourses and a lack of confidence in the likelihood of social change. Similarly, Campbell and Nair’s study in South Africa found that although women lamented the gender inequality they faced, they were unwilling to challenge men’s power because they viewed it as something to “work around” rather than something they could change (2014:1222). Our findings that women relied on gender essentialist scripts and the fulfillment of culture and tradition in their reaffirmation of inequality demonstrate that dominant norms about gender hierarchy in relationships can be particularly impervious to change. However, our findings that women take a pragmatic approach to 50/50 to protect themselves from men’s violence, infidelity, HIV, and abandonment also show the limits of cultural explanations for the reproduction of gender inequality. That is, women in our study often weighed their concerns about relationship inequality vis-à-vis the uncertainties of living in a rural setting marked by structural and health disadvantages and a lack of opportunities. At times this was acknowledged directly in women’s assessments that 50/50 was only possible for those who were educated or working and thus had some measure of economic independence. In essence, women’s pragmatic explanations for avoiding 50/50 reflected the idea that rights cannot be realized without resources (Campbell and Nair 2014; Englund 2006). Nonetheless, our findings emphasize that it is not just economic uncertainty that influences women’s views about equality in relationships, rather, it is the interrelationship between local norms supporting gender hierarchy, precarity in the political economy, and the substantial risks of violence and HIV. Extant research from South Africa underscores the challenge of reducing IPV where normative support for change is lacking. In work by Treves-Kagan and colleagues (2020), for example, culture and tradition were invoked as barriers to reducing violence (see also Dworkin et al. 2012, 2015; Hatcher et al. 2014). Thus, as we show, where economic uncertainty is rife, women may forgo gender equality in their relationships to avoid violence and a partner’s infidelity (Kandiyoti 1988). These findings emphasize the importance of embedding understandings of women’s beliefs about gender equality and their concomitant behaviors within local contexts by considering the political economic factors as well as health conditions—such as HIV/AIDS—that may complicate the adoption of equality in relationships (Kurzman et al. 2019; Thomas 2007; Unnithan and Pigg 2014). Although our study captures how women in rural South Africa view gender equality, we are unable to discern the interpersonal processes through which women may gain more power in relationships. Research is necessary to explore, for example, whether women who forgo 50/50 because of pragmatic concerns would be more likely to implement it given different local conditions. Future research should also examine the strategies women employ to negotiate for more equal relationships and compare women across settings to determine how local contexts might differentially influence women’s views of equality and the extent to which they are able to challenge unequal gender relations. Because our study focuses on the views of women in rural South Africa, future research among women in peri-urban or urban areas would be particularly instructive in this regard. As Evans (2018) suggests—and supported by our participants—urban areas may offer more opportunities for patriarchal norms to be disrupted and therefore for women to challenge the structures that subordinate them. This type of analysis could also introduce more variation among participants in race, ethnicity, and social class, allowing for more axes of comparison. Our findings highlight the view that 50/50 may be more attainable for women with education or economic resources, suggesting that social class may be especially important to examine. Our research also reinforces the importance of continuing gender transformative work with men, which has already begun across many areas of South Africa (e.g., van den Berg et al. 2013; Dworkin et al. 2012, 2015). Recent research suggests that though progress has been slow, these interventions can help shift community norms and gender relations in a more equitable direction (Gibbs et al. 2015, 2020; Ratele 2015; Treves-Kagan et al. 2020; Willan et al. 2020). Including more women and networks of community members in these endeavors may enhance the likelihood of sustaining change (Treves-Kagan et al. 2020). Finally, future research on gender equality that includes extended kin would be insightful for understanding what other types of status and hierarchy—such as age or seniority—within family systems can influence women’s views and experiences of 50/50 in relationships and more broadly (Moore 2015; Oyewùmí 1997). This study builds on and extends research from the Global North that has found that largescale change toward gender equality has been slowed by the persistence of longstanding gender norms (e.g., Ridgeway 2011). It is clear that broader structural changes, such as gender-progressive legislation, are necessary for more equitable societies; however, rights only go so far (Englund 2006; Hames 2006; Undie and Izugbara 2011). Despite the implementation of equal rights in South Africa, women in rural areas still face multiple structural barriers—alongside longstanding norms—that inhibit equality in relationships. We suspect that women across the Global South frequently contend with similar local material conditions and social structures that shape their views about equality and ultimately their health and wellbeing. In rural South Africa, for example, high rates of employment precarity and poverty, threats of gender-based violence, and the risk of HIV are omnipresent regardless of women’s ideological perspectives on gender equality. This suggests research focusing solely on cultural explanations for the persistence of gender inequality will be incomplete in overlooking pragmatic concerns. Therefore, our findings underscore the importance of incorporating both ideological and pragmatic factors in analyses of what inhibits or encourages greater equality in relationships in the Global South and beyond. We believe this approach has the potential to identify the complex interplay of local norms, material conditions, and risks to wellbeing that need to be addressed to effectively transform gender hierarchies and move closer to equality. Acknowledgements We thank the MRC/Wits Rural Public Health and Health Transitions Research Unit (Agincourt); the Lobola Project research team: Meriam Maritze, Nester Monareng, and Ellah Sihlangu; and Emily Dye and Abigail Nawrocki for research assistance. We thank Youngeun Nam and Daniel Winchester for their helpful feedback on earlier versions of the manuscript. Finally, we thank the women who made this project possible by sharing their views and experiences. Funding This work was supported by grants from the Purdue Research Foundation, the College of Liberal Arts, and the Department of Sociology at Purdue University. The MRC/Wits Rural Public Health and Health Transitions Research Unit and Agincourt Health and Socio-Demographic Surveillance System, a node of the South African Population Research Infrastructure Network (SAPRIN), is supported by the Department of Science and Innovation, the University of the Witwatersrand, and the Medical Research Council, South Africa, and previously the Wellcome Trust, UK (grants 058893/Z/99/A; 069683/Z/02/Z; 085477/Z/08/Z; 085477/B/08/Z). The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the funders. The authors declare that they have no conflicts of interest.A 1 These efforts are in response to the 2019 protests against the rising tide of violence against women in South Africa. See https://www.gov.za/blog/desk-president-36#. 2 Grace (age 23, single) described her views this way: “50/50 is good. I cannot do all the household activities meanwhile he [boyfriend] is sitting under the shadows. We must help each other. If I’m washing the dishes, he is sweeping the house, if I’m washing clothes, he is cooking.... We are helping each other. We are equal. It doesn’t mean…he doesn’t bathe [their son]. He has to.” Similarly, Gift (age 46, cohabiting) said: “I think it [50/50] helped us a lot as we were oppressed as women. Men were abusing us as we were not allowed to work, meanwhile they abandoned us. His parents also didn’t care about you as long as you woke up and did all the household activities. But now 50/50 allowed me to do [what I want] according to my will and without consulting him.” 3 Women often refer to their long-term or cohabiting partners as “husbands” even if they are not officially married. 4 All the women quoted above as offering pragmatic explanations also ideologically supported gender hierarchy in relationships. Ainslie Andrew Kepe Thembela Understanding the Resurgence of Traditional Authorities in Post-Apartheid South Africa Journal of Southern African Studies 2016 42 1 19 33 Baten Jorg de Plejit Alexandra Gender Equality Isn’t Just Moral – It’s Also Good Economics World Economic Forum 2019 Retrieved https://www.weforum.org/agenda/2019/06/history-shows-that-women-s-autonomy-leads-to-long-term-development van den Berg Wessel Hendricks Lynn Hatcher Abigail Peacock Dean Godana Patrick Dworkin Shari ‘One Man Can’: Shifts in Fatherhood Beliefs and Parenting Practices Following a Gendertransformative Programme in Eastern Cape, South Africa Gender & Development 2013 21 1 111 25 24000276 Bhana Deevia Virginity and Virtue: African Masculinities and Femininities in the Making of Teenage Sexual Cultures Sexualities 2016 10.1177/1363460715613298 Bhana Deevia Patttman Rob Girls Want Money, Boys Want Virgins: The Materiality of Love amongst South African Township Youth in the Context of HIV and AIDS Culture, Health & Sexuality 2011 13 8 961 72 Blalock Casey Labor Migration and Employment in Post-Apartheid Rural South Africa University of Colorado Boulder 2014 Burchardt Marian Ironies of Subordination: Ambivalences of Gender in Religious AIDS Interventions in South Africa Oxford Development Studies 2010 38 1 63 82 Campbell Catherine Nair Yugi From Rhetoric to Reality? Putting HIV and AIDS Rights Talk into Practice in a South African Rural Community Culture, Health & Sexuality 2014 16 10 1216 30 Charmaz Kathy Grounded Theory Emerson R Contemporary Field Research: Perspectives and Formulations 2001 335 52 Waveland Press Prospect Heights Connell RW Gender and Power: Society, the Person and Sexual Politics University of California Press Palo Alto, CA 1987 Connell Raewyn W Masculinities 1995 Cambridge Polity Dunkle Kristin L Jewkes Rachel K Brown Heather C Gray Glenda E McIntryre James A Harlow Siobán D Gender-Based Violence, Relationship Power, and Risk of HIV Infection in Women Attending Antenatal Clinics in South Africa Lancet 2004 363 1415 21 15121402 Dworkin SL Colvin C Hatcher A Peacock D Men’s Perceptions of Women’s Rights and Changing Gender Relations in South Africa: Lessons for Working With Men and Boys in HIV and Antiviolence Programs Gender & Society 2012 26 1 97 120 Dworkin Shari L Fleming Paul J Colvin Christopher J The Promises and Limitations of Gender Transformative Health Programming with Men: Critical Reflections from the Field Culture, Health & Sexuality 2015 17 sup2 128 43 England Paula Levine Andrew Mishel Emma Progress Toward Gender Equality in the United States Has Slowed or Stalled Proceedings of the National Academy of Sciences of the United States (PNAS) 2020 117 13 6990 97 Englund H Prisoners of Freedom: Human Rights and the African Poor U. Press Berkeley, CA 2006 Evans Alice Cities as Catalysts of Gendered Social Change? Reflections from Zambia Annals of the American Association of Geographers 2018 108 4 1096 1114 Gibbs Andrew Jewkes Rachel Sikweyiya Yandisa Willan Samantha Reconstructing Masculinity? A Qualitative Evaluation of the Stepping Stones and Creating Futures Interventions in Urban Informal Settlements in South Africa Culture, Health & Sexuality 2015 17 2 208 22 Gibbs Andrew Washington Laura Abdelatif Nada Chirwa Esnat Willan Samantha Shai Nwabisa Sikweyiya Yandisa Mkhwanazi Smanga Ntini Nolwazi Jewkes Rachel Stepping Stones and Creating Futures Intervention to Prevent Intimate Partner Violence Among Young People : Cluster Randomized Controlled Trial Journal of Adolescent Health 2020 66 3 323 35 Giddens Anthony The Constitution of Society: Outline of the theory of Structuration University of California Press Berkeley 1984 Gómez-Olivé Francesc Xavier Angotti Nicole Houle Brian Klipstein-Grobusch Kerstin Kabudula Chodziwadziwa Menken Jane Williams Jill Tollman Stephen Clark Samuel J Prevalence of HIV among Those 15 and Older in Rural South Africa AIDS Care 2013 January 25 1122 28 23311396 Gqola Pumla Dineo How the ‘Cult of Femininity’ and Violent Masculinities Support Endemic Gender Based Violence in Contemporary South Africa African Identities 2007 5 1 111 24 Gqola Pumla Dineo Books M Rape: A South African Nightmare Johannesburg, South Africa 2015 Graham Lauren ‘I Am a Mother’: Young Women’s Negotiation of Femininity and Risk in the Transition to Adulthood Culture Health & Sexuality 2016 18 3 265 79 Hames Mary Rights and Realities: Limits to Women’s Rights and Citizenship after 10 Years of Democracy in South Africa Third World Quarterly 2006 27 7 1313 27 Harrison Abigail Hidden Love: Sexual Ideologies and Relationship Ideals among Rural South African Adolescents in the Context of HIV/AIDS Culture, Health & Sexuality 2008 10 2 175 89 Hassim Shireen The Gender Pact and Democratic Consolidation: Institutionalizing Gender Equality in the South African State Feminist Studies 2003 29 3 504 28 Hassim Shireen Women’s Organizations and Democracy in South Africa: Contesting Authority The University of Wisconsin Press Madison 2006 Hassim Shireen Decolonising Equality: The Radical Roots of the Gender Equality Clause in the South African Constitution South African Journal on Human Rights 2018 34 3 342 58 Hatcher Abigail M Colvin Christopher J Ndlovu Nkuli Dworkin Shari L Intimate Partner Violence among Rural South African Men: Alcohol Use, Sexual Decision-Making, and Partner Communication Culture, Health & Sexuality 2014 16 9 1023 39 Helman Rebecca Ratele Kopano Everyday (In)Equality at Home: Complex Constructions of Gender in South African Families Global Health Action 2016 9 1 31122 27293123 Horne Christine Mollborn Stefanie Norms: An Integrated Framework Annual Review of Sociology 2020 46 467 87 Human Sciences Research Council (HSRC) The Fifth South African National HIV Prevalence, Incidence, Behaviour and Communication Survey, 2017: HIV Impact Assessment Summary Report Cape Town, South Africa 2018 Hunter Mark Cultural Politics and Masculinities: Multiple-partners in Historical Perspective in KwaZulu-Natal Culture, Health & Sexuality 2005 7 3 209 23 Hunter Mark Love in the Time of AIDS: Inequality, Gender, and Rights in South Africa Indiana University Press Bloomington, IN 2010 Jewkes Rachel K Dunkle Kristin Nduna Mzikazi Shai Nwabisa Intimate Partner Violence, Relationship Power Inequity, and Incidence of HIV Infection in Young Women in South Africa: A Cohort Study The Lancet 2010 376 9734 41 48 Jewkes Rachel Morrell Robert Gender and Sexuality: Emerging Perspectives from the Heterosexual Epidemic in South Africa and Implications for HIV Risk and Prevention Journal of the International AIDS Society 2010 13 6 1 11 20205768 Jewkes Rachel Morrell Robert Sexuality and the Limits of Agency among South African Teenage Women: Theorising Femininities and Their Connections to HIV Risk Practices Social Science & Medicine 2012 74 1729 37 21696874 Kabeer Naila Gender Equality, Economic Growth, and Women’s Agency: The ‘Endless Variety’ and ‘Monotonous Similarity’ of Patriarchal Constraints Feminist Economics 2016 22 1 295 321 Kahn Kathleen Collinson Mark A Xavier Gómez-Olivé F Mokoena Obed Twine Rhian Mee Paul Afolabi Sulaimon A Clark Benjamin D Kabudula Chodziwadziwa W Khosa Audrey Khoza Simon Profile: Agincourt Health and Socio-Demographic Surveillance System International Journal of Epidemiology 2012 41 988 1001 22933647 Kandiyoti Deniz Bargaining with Patriarchy Gender & Society 1988 2 3 274 90 Kurzman Charles Dong Willa Gorman Brandon Hwang Karam Ryberg Renee Zaidi Batool Women’s Assessments of Gender Equality Socius 2019 5 1 13 Mathis Sarah M Disobedient Daughters? Changing Women’s Roles in Rural Households in KwaZulu-Natal Journal of Southern African Studies2 2011 37 4 831 48 Moore Elena Forms of Femininity at the End of a Customary Marriage Gender & Society 2015 29 6 817 40 Mphaphuli Memory Smuts Letitia ‘Give It to Him’: Sexual Violence in the Intimate Relationships of Black Married Women in South Africa Signs 2021 46 2 443 64 Oyewùmí Oyèrónké The Invention of Women: Making an African Sense of Western Gender Discourses U. of M. Press London 1997 Pettifor Audrey MacPhail Catherine Anderson Althea D Maman Suzanne ‘If I Buy the Kellogg’s Then He Should [Buy] the Milk’: Young Women’s Perspectives on Relationship Dynamics, Gender Power and HIV Risk in Johannesburg, South Africa Culture, Health & Sexuality 2012 14 5 477 90 Pierotti Rachael S Lake Milli Lewis Chloe Equality on His Terms: Doing and Undoing Gender Through Men’s Discussion Groups Gender & Society 2018 32 4 540 62 Posel Deborah Afterword: Vigilantism and the Burden of Rights: Reflections on the Paradoxes of Freedom in Post-apartheid South Africa African Studies 2004 63 2 231 36 Ranganathan Meghna Stern Erin Knight Louise Muvhango Lufono Molebatsi Mpho Polzer-Ngwato Tara Lees Shelley Stockl Heidi Women’s Economic Status, Male Authority Patterns and Intimate Partner Violence: A Qualitative Study in Rural North West Province, South Africa Culture, Health & Sexuality 2021 Ratele Kopano Working through Resistance in Engaging Boys and Men Towards Gender Equality and Progressive Masculinities Culture, Health & Sexuality 2015 17 sup2 144 58 Ridgeway Cecilia L Framed by Gender: How Gender Inequality Persists in the Modern World Oxford University Press New York 2011 Risman Barbara Gender as a Social Structure: Theory Wrestling with Activism Gender & Society 2004 18 4 429 50 Risman Barbara Where the Millennials Will Take Us: A New Generation Wrestles with the Gender Structure O. U. Press New York 2018 Robins S From Revolution to Rights in South Africa: Social Movements, NGOs and Popular Politics after Apartheid J. Currey Oxford 2008 Schippers Mimi Recovering the Feminine Other: Masculinity, Femininity, and Gender Hegemony Theory and Society 2007 36 1 85 102 Seedat Mohamed Van Niekerk Ashley Jewkes Rachel Shahnaaz Suffla Ratele Kopano Violence and Injuries in South Africa: Prioritising an Agenda for Prevention Lancet 2009 374 1011 22 19709732 Seidman Gay W ‘Strategic’ Challenges to Gender Inequality: The South African Gender Commission Ethnography 2001 2 219 41 2, Special Issue: Global Ethnography (June 2001) Sennott Christie Angotti Nicole Reconsidering Gendered Sexualities in a Generalized AIDS Epidemic Gender & Society 2016 30 6 935 57 35571270 Sennott Christie Madhavan Sangeetha Nam Youngeun Modernizing Marriage: Balancing the Benefits and Liabilities of Bridewealth in Rural South Africa Qualitative Sociology 2021 44 55 75 34168388 Sennott Christie Mojola Sanyu ‘Behaving Well’: The Transition to Respectable Womanhood in Rural South Africa Culture, Health and Sexuality 2017 19 7 781 95 Statistics South Africa Quarterly Labour Force Survey Pretoria, South Africa 2020 Strauss Anselm L Corbin Juliet Basics of Qualitative Research: Grounded Theory Procedures and Techniques Sage Newbury Park, CA 1990 Talbot Kirsten Quayle Michael The Perils of Being a Nice Guy: Contextual Variation in Five Young Women’s Constructions of Acceptable Hegemonic and Alternative Masculinities Men and Masculinities 2010 13 2 255 78 Thomas Felicity Global Rights, Local Realities: Negotiating Gender Equality and Sexual Rights in the Caprivi Region, Namibia Culture, Health & Sexuality 2007 9 6 599 614 Treves-Kagan Sarah Maman Suzanne Khoza Nomhle MacPhail Catherine Peacock Dean Twine Rhian Kahn Kathleen Lippman Sheri A Pettifor Audrey Fostering Gender Equality and Alternatives to Violence: Perspectives on a Gender-Transformative Community Mobilisation Programme in Rural South Africa Culture, Health & Sexuality 2020 22 S1 127 44 UNAIDS Country Fact Sheet: South Africa UNAIDS 2018 Retrieved January 15, 2021 https://www.unaids.org/en/regionscountries/countries/southafrica Undie Chi-Chi Izugbara Chimaraoke O Unpacking Rights in Indigenous African Societies: Indigenous Culture and the Question of Sexual and Reproductive Rights in Africa BMC International Health and Human Rights 2011 11 Suppl 3 S2 United Nations Sustainable Development Goals Goal 5: Achieve Gender Equality and Empower All Women and Girls 2020 Retrieved https://www.un.org/sustainabledevelopment/gender-equality/ Unnithan Maya Pigg Stacy Leigh Sexual and Reproductive Health Rights and Justice - Tracking the Relationship Culture, Health & Sexuality 2014 16 10 1181 87 Walker Liz Men Behaving Differently: South African Men since 1994 Culture, Health & Sexuality 2005 7 3 225 38 Willan Samantha Gibbs Andrew Shai Nwabisa Ntini Nolwazi Petersen Inge Did Young Women in South African Informal Settlements Display Increased Agency after Participating in the Stepping Stones and Creating Futures Intervention? A Qualitative Evaluation Social Science & Medicine 2020 265 113302 32890814 Williams Jill Ibisomi Latifat Sartorius Benn Kahn Kathleen Collinson Mark Tollman Stephen Garenne Michel Convergence in Fertility of South Africans and Mozambicans in Rural South Africa, 1993-2009 Global Health Action 2013 6 19236 23364078 Worden Nigel The Making of Modern South Africa: Conquest, Apartheid, Democracy Fourth Wiley-Blackwell Oxford 2007 World Bank The World Bank in South Africa 2019 Retrieved https://www.worldbank.org/en/country/southafrica/overview World Bank The World Bank in Gender 2020 Retrieved https://www.worldbank.org/en/topic/gender/overview Wyrod Robert Between Women’s Rights and Men’s Authority: Masculinity and Shifting Discourses of Gender Difference in Urban Uganda Gender & Society 2008 22 6 799 823 19862350 Wyrod Robert AIDS and Masculinity in the African City: Privilege, Inequality, and Modern Manhood University of California Press Oakland 2016

PMC007xxxxxx/PMC7614799.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9702877 CNS Spectr CNS Spectr CNS spectrums 1092-8529 35593450 7614799 10.1017/S1092852922000815 EMS177445 Article Religiosity, Impulsivity, and Compulsivity in University Students Grant Jon E. 1 Lust Katherine 2 Chamberlain Samuel R. 3 1 Department of Psychiatry & Behavioral Neuroscience University of Chicago, Chicago, IL, USA 2 Boynton Health Service, University of Minnesota USA 3 Department of Psychiatry, Faculty of Medicine, University of Southampton, UK; and Southern Health NHS Foundation Trust, Southampton, UK Address correspondence to: Jon E. Grant, JD, MD, MPH, Professor, Department of Psychiatry & Behavioral Neuroscience, University of Chicago, Pritzker School of Medicine, 5841 S. Maryland Avenue, MC 3077, Chicago, IL 60637, Phone: 773-834-1325; Fax: 773-834-6761; jongrant@uchicago.edu 01 6 2023 20 5 2022 04 7 2023 25 7 2023 28 3 367373 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Background Prior research suggests that religiosity may be associated with higher levels of mental health in certain domains (e.g. self-esteem, rates of substance use problems). However, very little is known about religiosity and impulsive plus compulsive tendencies. This study examined associations between religiosity and impulsive and compulsive behaviors and traits among university students. Methods 9,449 students received a 156-item anonymous online survey which assessed religiosity, alcohol and drug use, mental health issues, and impulsive and compulsive traits. Two groups of interest were defined: those with high religiosity, and those with low religiosity, based on z-scores. The two groups were compared on the measures of interest. Results 3,572 university students (57.1% female) responded to the survey. Those with high levels of organizational religious activity (ORA), as well as those with high levels of intrinsic or subjective religiosity (IR) differed from their fellow students in having better self-esteem, being less likely to have alcohol or drug problems, and generally being less impulsive in terms of attention and planning. These associations were of small effect size except for the link between religiosity and lower impulsivity, which was of medium effect size. Conclusion This study shows a link between higher religiosity and lower impulsivity, as well as higher levels of mental health across several domains. Whether these associations are causal – and if so, the direction of such causality – requires rigorous longitudinal research. religiosity spirituality addiction impulsivity pmcIntroduction Religion has had an enduring impact on human society and has shaped how countless people perceive themselves and their world (1). Although the evolutionary basis of religion continues to be debated, some conceptual approaches view religion as either a byproduct of fundamental cognitive processes or as an adaptive social system designed to promote cooperation and other prosocial behaviors (2). Given that religion appears to have adaptive value, it is perhaps unsurprising that most studies support a positive association between religiosity and mental health (3,4). The behavioral mechanisms that may explain these findings across diverse cultures, however, are a matter of controversy (5). Religiosity has also shown some association with spirituality, but there are differences between these constructs. Whereas religion represents a socially-organized system of beliefs (13), spirituality is usually defined by the person and often refers to a person’s sense of meaning in life and a connection to a power greater than the self. Studies in the field of addictions have suggested that both religiosity and spirituality often increase self-control over unhealthy behaviors by giving people a feeling of purpose, reinforcing core values, and promoting cognitive changes (14–16). According to an emerging body of evidence, the religion/health relationship may be partly explained by the concept of self-control (6–8). Self-control is a construct linked to several distinct cognitive processes and personality traits, including conscientiousness (an index of one’s tendency to be organized, responsible, and hard-working) and the ability to delay immediate gratification. By these measures, the more religious a person is the greater the capacity for self-control, on average, compared to non-religious counterparts (7–10). Indeed, self-control is thought to be a crucial element of religious practice—consider that virtually all religions require their members to participate in effortful ritual practices or behaviors, such as public prayer or fasting, that require the exercise of self-control (7,8). In turn, self-control may promote greater subjective psychological well-being (11) or mediate the relationship between religiosity and health behaviors (e.g., substance use) (12). Notably, studies of diverse social groups have found that subjective well-being is better predicted by involvement in institutional religious practices (e.g., attendance at religious services) than by private religious practice or personal religious belief (8). Private or subjective forms of religiosity, however, may preferentially benefit some clinical populations (8). Relevant to the social construct of ‘self-control’ are the concepts, from the neurosciences, of impulsivity and compulsivity. Impulsivity refers to the tendency towards hasty or poorly thought out actions, leading to untoward actions (cite); whereas compulsivity is the tendency towards repetitive habitual actions that persist despite consequent functional impairment (cite). These two processes contribute at different stages in the progression from a potentially risky act (e.g. drinking alcohol or gambling) through to getting ‘stuck’ in these behaviors. Impulsivity and compulsivity can be fruitfully measured using convenient self-report questionnaires (cite). Young adulthood is a time where many individuals engage in and struggle with controlling unhealthy behaviors, but little is known about the influences of religiosity in this age cohort. Our study examined links between impulsivity, compulsivity, and multiple dimensions of religiosity in a large sample of university students using a voluntary, anonymous internet-based survey. We hypothesized that religiosity (specifically, the frequency of religious behaviors performed in a group or social setting) would be associated with lower trait impulsivity, lower trait compulsivity, and lower rates of non-substance or “behavioral” addictions (gambling disorder, compulsive sexual behavior, binge eating disorder) (17). Methods Survey Design The Department of Psychiatry and Behavioral Neuroscience at the University of Chicago and Boynton Health at the University of Minnesota jointly developed the Health and Addictive Behaviors Survey to assess mental health and well-being in a large sample of university students. The survey included basic demographics as well as questions from a number of validated screening tools examining mental health and psychological well-being. All study procedures were carried out in accordance with the Declaration of Helsinki and were approved by the Institutional Review Board of the University of Minnesota. Participants A sub-sample of 10,000 college and graduate students at a large, nondenominational and coeducational Midwestern university were chosen by random, computer-generated selection from a total pool of approximately 60,000 students at the university. The survey was distributed over a three-week period during the fall 2016 semester, with invitations sent by email and surveys completed online. Of the 10,000 email invitations, 9449 were successfully received by the recipients (i.e., did not bounce back). Recipients of the email were first required to view the IRB-approved online informed consent page, at which point students could choose to participate in the survey or opt out. The survey asserted that all information was both anonymous and confidential. Compensation was offered at the conclusion of the survey by randomly selecting respondents to receive tablet computers (n = 3) or gift certificates to an online retailer in the amounts of $250 (n = 4), $500 (n = 2), and $1000 (n = 1). Participants were required to review all survey questions to be eligible for prize drawings, but were not required to answer any question they did not wish to answer. Of the 9449 students who received the invitation to participate, 3659 (38.7%) completed the survey. Assessments The self-report survey consisted of 156 questions and took participants approximately 30 minutes to complete. Survey questions assessed demographic information (including religious affiliation), sexual behavior, self-reported academic achievement (i.e., grade point average [GPA]), and clinical characteristics of participants, including levels of mental health and substance use. In order to assess other aspects of mental health function and religiosity, participants were also asked to complete the following measures: Religiosity was assessed using the Duke University Religion Index (DUREL). The DUREL is a valid and reliable, 5-item measure of religious involvement across three domains: organizational religious activity (ORA), non-organizational religious activity (NORA), and intrinsic or subjective religiosity (IR) (18). The ORA domain assesses frequency of participation in religious services (1 = never to 6 = more than once/week). The NORA domain measures the extent of involvement in private religious activities, such as prayer or the study of religious texts (1 = rarely or never to 6 = more than once a day). The IR domain (3 items) assesses the degree to which the participant is motivated by or committed to his or her religion (1 = definitely not true to 5 = definitely true of me). Higher scores reflect greater religiosity. The DUREL demonstrated good internal consistency in our sample (Cronbach α = 0.924). Putative disorders of impulse control were screened for using the Minnesota Impulsive Disorders Interview (MIDI). In this study, we focused on gambling disorder and compulsive sexual behavior (19–20). Alcohol use behaviors and related problems were assessed using the Alcohol Use Disorders Identification Test (AUDIT). Each item is scored 0-4, with a maximum of 40 points possible. A score of 8 or greater indicates hazardous or harmful alcohol use (21). Problematic substance use was identified using the Drug Abuse Screening Test (DAST-10). A score of 3 is used to screen for a drug use disorder (22–23). Depressive symptoms were measured using the Patient Health Questionnaire (PHQ-9). The PHQ-9 is based directly on DSM-IV-TR criteria for major depressive disorder (24). Posttraumatic stress disorder (PTSD) was screened for using the Primary Care PTSD Screen (PC-PTSD). The PC-PTSD is based on DSM-IV PTSD criteria (25). A score of ≥3 indicates probable PTSD. Generalized anxiety disorder (GAD) was screened for using the Generalized Anxiety Disorder 7 (GAD-7). Total scores of 10 or greater indicate clinically significant anxiety (26). Attention-deficit/hyperactivity disorder (ADHD) was screened for using the Adult ADHD Self-Report Scale (ASRS-v1.1). The ASRS has demonstrated strong psychometric properties (27). Global feelings of self-worth or self-regard were measured using the Rosenberg Self-Esteem Scale (RSES). Scores below 15 suggest low self-esteem (28). Impulsivity was assessed using the Barratt Impulsiveness Scale, Version 11 (BIS-11). The BIS-11 is a 30-item measure designed to assess impulsivity across three dimensions: attentional (inability to concentrate), motor (acting without thinking), and non-planning (lack of future orientation) (29–30). Each of the 30 items is rated on a 4-point scale of 1 (rarely/never) to 4 (almost always), where 4 indicates greater impulsiveness. Compulsive traits were measured using the Cambridge–Chicago Compulsivity Trait Scale (CHI-T). The scale has shown excellent psychometric properties, with high internal consistency (Cronbach’s alpha = 0.8), excellent convergent validity against gold-standard assessments for a variety of compulsive disorders (each p < .001 for gambling disorder, obsessive-compulsive disorder, and substance use disorder symptoms), and excellent discriminant validity against other constructs such as depression (31). The validity of the CHI-T has also been confirmed in a large sample of the general population and in people with mental health and psychiatric diagnoses (32). Data Analysis Only respondents with complete data on at least one the DUREL subscales were included in the analyses (N = 3564; 99.8%). Total scores for each of the three subscales were transformed to standardized z scores and participants were categorized based on level of religiosity: low (z<−1) or high (z>+1). Participants not scoring in these ranges played no further role in the analysis. Distributions of the third score (NOR) did not permit this approach and so we focused on the two where we had adequately large samples of folks with low and high. Groups were compared on demographic and clinical measures using independent sample t tests for continuous variables (or equivalent nonparametric tests, as indicated in the text) and chi-square tests for categorical variables. Effect sizes were calculated for all significant differences, which were determined for Likelihood ratio test using Cramer’s V (V = 0.1 is considered a small effect size, 0.3 is medium, and 0.5 is large) (28). Continuous variables were tested for statistical difference using F-test and Cohen’s d for effect size, effect sizes of 0.02, 0.15, and 0.35 are termed small, medium, and large, respectively (32). SPSS was used for all statistical analyses (version 25; IBM). Statistical significance was defined as p ≤ 0.01 to account for multiple comparisons. Results The demographic characteristics of the 3,572 participants (57.1% female) are presented in Table 1. Overall, the mean ORA score was 2.52 (1.49), mean NORA score was 1.98 (1.51) and the mean IRA score was 7.66 (4.17). Participants with high ORA and high IR showed the same patterns in terms of demographics. That is, they were more likely to be female, married or engaged, and identify as Catholic, Muslim, protestant, or “other Christian” than those low on ORA or IR. Grade point average did not differ based on religiosity. In terms of mental health, the data are presented in Table 2. Participants who scored high on ORA and IR were significantly less likely to have alcohol or drug problems and less likely to have low self-esteem. In addition, those with high ORA were significantly less likely to screen positive for PTSD. Finally, although participants with higher levels of ORA and IR did not differ significantly on a measure of compulsivity from those with lower levels, they did significantly differ in two domains of impulsiveness, attentional and non-planning impulsiveness (Table 3). Discussion This study examined two aspects religiosity and their links with mental health with a particular focus on impulsive and compulsive tendencies. The two aspects of religiosity examined were organizational religiosity (propensity to attend and engage with formal religious services) and intrinsic religiosity (propensity to integrate religion into one’s life endeavors) (cite). We focused on a large sample of university students and the possible associations between religiosity and a range of demographic/clinical measures, and questionnaire-based measures of impulsivity. We found that students who scored high on either types of religiosity were less impulsive, had better self-esteem and were less likely to have alcohol or drug problems. These results seem generally in keeping with previous examinations of religiosity in young adults. In a previous study using the DRUEL in a small sample of 93 patients with mental illness who had attempted suicide and 61 healthy individuals, Caribe and colleagues (2015) found that the healthy individuals scored higher scores in the religiosity domains and this was associated with lower scores on the BIS impulsiveness scale. Similarly, a study of 448 students in Iran found that those who engaged more often in organized religious activities and had higher intrinsic religiosity were less likely to engage in risky behaviors such as sexual risk taking, careless driving, violence, smoking, as well as alcohol and drug abuse (Ameri et al., 2017). The links between religiosity and other measures in the current study were generally of small effect size, which would be in keeping with prior cross-sectional research in other areas of mental health (cites) including more recent longitudinal work (cite). However, the one finding in this study that demonstrated a moderate effect size was that higher religiosity was associated with less attentional impulsiveness. This BIS subscale reflects a tendency to have rapid shifts in attention, to have difficulties in task focus, and to become impatient with complexity. The fact that religiosity was not associated with compulsivity is a novel finding, contrary to our predictions, and is in contrast to the link found with impulsivity. These results may suggest that people with high religiosity are less likely to engage in impulsive acts on the spur of the moment (e.g. early stages of alcohol use or gambling), but are just as likely to develop habitual repetitive behaviors over time after initially engaging in these activities. It is interesting to consider how this may reflect the focus of several mainstream religions on often complete avoidance of certain addictive substances and behaviors (e.g. alcohol, gambling). Does this reflect our innate tendency to develop habits irresepective of religiosity, whereas avoiding early stages of potentially problematic behavior is something we are abler to do and this is aided by religious frameworks? In terms of mental health problems, we found that higher levels of religiosity were significantly associated with higher self-esteem and, in the case of organized religion, with lower levels of PTSD. Thus our findings add to growing evidence of the potential small effect size protective factors of religiosity in young people. A study of Veterans similarly found that PTSD was less likely in those with greater religiosity (Sharma et al., 2017). This finding could be explained by the sense of purpose and community that organized religion instills in some people, or it could be an indirect effect. That is, those with higher organizational religiosity also had better self-esteem, were less impulsive and less likely to have alcohol and drug problems. Given that PTSD has been associated with alcohol and drug problems (Panza et al., 2022), and that less impulsive people may be less likely to have traumatic situations (Santos et al., 2022), multiple interacting variables may explain the lower rates of PTSD in those who are more religious. This study of religiosity in young adults has the advantage of being relatively large. Nonetheless, there are several limitations that should be considered. The study was cross-sectional and hence the direction of causality of any effects cannot be established – this would require longitudinal research on the topic; however, we hope that such cross-sectional data will encourage such follow-up. Given that associations were generally of small effect size, we did not attempt to examine mediation between variables. There are limitations inherent in the study being conducted using an online interface via the Internet – diagnostic assessment may be less accurate via such an online survey compared to in-person assessment by a clinician; there may be responder biases; and there may be under-reporting (though this possibility is reduced by individuals’ responses not being lacked to personally identifiable information). Our splitting of the sample into those with high and low religiousity was a useful way of presenting the data since it is intuitive to the reader; however of course there are other ways of operationalizing high and low religiosity that could be used. In summary, we found that higher levels of religiosity in university students were associated with lower rates of impulsivity (medium effect size) as well as relatively higher levels of mental health (small effect size), but not with different levels of compulsivity. Whether religiosity leads to being less impulsive or vice versa, both, or the link can be accounted for by other variables, remains unclear. The link with impulsive traits may indicate less propensity of people with high religiousity to spontaneously undertake or engage with potentially harmful activities (e.g. alcohol or gambling) but that once initiated, there is a similar tendency to get stuck in a given habitual pattern as compared to people with low levels of religiosity. Table 1 Demographics of university students based on level of religiositya   Organizational religious activity Statistic Intrinsic religiosity Statistic Z Score <-1.00 N=1338 Z Score >1.00 N=450 Z Score <-1.00 N=958 Z Score >1.00 N=867 Gender Male 516(41.1) 170(40.3) LR=11.789 df=2 P=.003 V=.073 373(41.8) 261(32.2) LR=29.093 df=2 P<.001 V=.129 Female 704(56.0) 250(59.2) 494(55.4) 543(67.0) Transgender, genderqueer, or alternative descriptor 37(2.9) 2(0.5) 25(2.8) 7(0.9) Religious affiliation Agnostic 352(26.3) 1(0.2) LR=1129.31 df=11 P=.000 V=.770 240(25.1) 8(0.9) LR=1573.25 df=11 P<.001 V=.829 Atheist 369(27.6) 2(0.4) 358(37.4) 4(0.5) Buddhist 17(1.3) 2(0.4) 11(1.1) 11(1.3) Catholic 44(3.3) 100(22.2) 27(2.8) 172(19.8) Hindu 13(1.0) 2(0.4) 1(0.1) 16(1.8) Jewish 8(0.6) 3(0.7) 12(1.3) 9(1.0) Muslim 12(0.9) 28(6.2) 1(0.1) 51(5.9) Protestant 8(0.6) 123(27.3) 2(0.2) 193(22.3) Other Christian 61(4.6) 144(32.0) 13(1.4) 289(33.3) Other 119(8.9) 2(0.4) 56(5.8) 19(2.2) Chose more than one religion 225(16.8) 39(8.7) 172(18.0) 83(9.6) Prefer to not answer 110(8.2) 4(0.9) 65(6.8) 12(1.4) Student Status Undergraduate 885(66.1) 290(64.4) LR=1.00 df=2 P=.606 V=.024 640(66.8) 548(63.2) LR=4.976 df=2 P=.083 V=.052 Graduate/Professional 446(33.3) 156(34.7) 315(32.9) 311(35.9) Non-degree seeking 7(0.5) 4(0.9) 3(0.3) 8(0.9) Race/ethnicity, Caucasian 913(72.7) 282(66.7) LR=5.604 df=1 P=.018 V=.058 685(76.8) 599(73.8) LR=2.093 df=1 P=.148 V=.035 Relationship Status Single 578(43.2) 231(51.3) LR=62.905 df=3 P=.000 V=.186 394(41.1) 414(47.8) LR=59.713 df=3 P<0.001 V=.180 Dating 578(43.2) 110(24.4) 441(46.0) 265(30.6) Engaged/married 169(12.6) 106(23.6) 111(11.6) 182(21.0) Other 13(1.0) 3(0.7) 12(1.3) 6(0.7) College GPA Below 2.50 26(2.0) 5(1.1) LR=4.647 df=3 P=.200 V=.051 18(1.9) 11(1.3) LR=1.435 df=3 P=.697 V=..028 2.50-2.99 120(9.1) 36(8.2) 78(8.2) 69(8.0) 3.00-3.49 452(34.2) 135(30.6) 325(34.2) 286(33.3) 3.50-4.00 725(54.8) 265(60.1) 528(55.6) 492(57.3) a Measured by the Duke University Religion Index (DUREL) Data refer to N (percentage), LR=Likelihhod Ratio, V=Cramer’s V GPA = grade point average Table 2 Mental health problems of university students based on level of religiositya   Organizational religious activity Statistic Intrinsic religiosity Statistic Z Score <-1.00 N=1338 Z Score >1.00 N=450 Z Score <-1.00 N=958 Z Score >1.00 N=867 PHQ9-Major depression disorderb 69(5.4) 12(2.9) LR=5.078 df=1 P=.024 V=.052 54(6.0) 33(4.1) LR=3.179 df=1 P=.075 V=.043 PC-PTSDc 212(16.6) 44(10.3) LR=10.821 df=1 P=.001 V=.077 137(15.0) 106(12.9) LR=1.651 df=1 P=.199 V=.031 Generalized anxiety disorderd 232(18.5) 57(13.6) LR=5.407 df=1 P=.020 V=.056 168(18.6) 130(16.0) LR=2.005 df=1 P=.157 V=.034 Compulsive sexual behavior 46(3.7) 14(3.3) LR=0.118 df=1 P=.731 V=.008 32(3.6) 26(3.2) LR=0.164 df=1 P=.686 V=.010 Binge eating disorder 37(2.9) 8(1.9) LR=1.440 df=1 P=.230 V=.028 28(3.1) 11(1.3) LR=6.203 df=1 P=.013 V=.059 ADHD 241(19.1) 62(14.6) LR=4.602 df=1 P=.032 V=.051 174(19.4) 120(14.8) LR=6.371 df=1 P=.012 V=.061 Gambling disorder 5(0.4) 0(0.0) LR=2.895 df=1 P=.089 V=.031 2(0.2) 3(0.4) LR=0.317 df=1 P=.574 V=.014 Low self-esteeme 213(17.1) 45(10.8) LR=9.965 df=1 P=.002 V=.075 156(17.7) 83(10.5) LR=18.110 df=1 P<0.001 V=.103 AUDIT score >=8 335(25.3) 36(8.1) LR=69.653 df=1 P=.000 V=.184 238(25.1) 150(17.5) LR=15.649 df=1 P<0.001 V=.093 DAST-10 score >=3 142(10.8) 12(2.7) LR=33.540 df=1 P=.000 V=.124 95(10.1) 43(5.0) LR=16.558 df=1 P<0.001 V=.095 a Measured by the Duke University Religion Index (DUREL) Data refer to N (percentage), LR=Likelihhod Ratio, V=Cramer’s V ADHD = attention-deficit/hyperactivity disorder; GAD-7 = General Anxiety Disorder-7; PC-PTSD = Primary Care PTSD Screen; PHQ-9 = Patient Health Questionnaire a Measured by the Duke University Religion Index (DUREL) b PHQ-9 score ≥10 c PC-PTSD score ≥3 d GAD-7 score ≥10 e RSES score <15 Table 3 Impulsivity and compulsivity of university students based on level of religiositya   Organizational religious activity Statistic Intrinsic religiosity Statistic Z Score <-1.00 N=1338 Z Score >1.00 N=450 Z Score <-1.00 N=958 Z Score >1.00 N=867 Cambridge-Chicago Compulsivity Trait Scale 9.83(13.69) 8.63(13.22) F(1,1713)=2.506 P=.114 d=.09 9.79(13.56) 8.31(13.20) F(1,1742)=5.296 P=.021 d=0.11 Barratt Impulsiveness Scale (BIS-11) Attentional impulsiveness 16.70(4.08) 15.25(3.83) F(1,1639)=39.98 P=.000 d=0.36 16.76(4.25) 15.48(3.85) F(1,1666)=41.42 P=.000 d=0.32 Motor impulsiveness 20.46(4.01) 20.02(4.12) F(1,1642)=3.716 P=.054 d=0.11 20.22(4.01) 20.25(4.14) F(1,1668)=0.024 P=.876 d=0.008 Non-planning impulsiveness 23.29(4.88) 22.21(4.46) F(1,1634)=15.736 P=.000 d=0.23 23.01(4.88) 22.28(4.74) F(1,1665)=9.388 P=.002 d=0.15 a Measured by the Duke University Religion Index (DUREL) Data refer to Mean (standard deviation), d=Cohen’s d Cambridge–Chicago Compulsivity Trait Scale Barratt Impulsiveness Scale (BIS-11) Attentional impulsiveness Motor impulsiveness Non-planning impulsiveness Disclosures: Dr. Grant has received research grants from Otsuka and Biohaven Pharmaceuticals. Dr. Grant receives yearly compensation from Springer Publishing for acting as Editor-in-Chief of the Journal of Gambling Studies and has received royalties from Oxford University Press, American Psychiatric Publishing, Inc., Norton Press, and McGraw Hill. This research was funded in whole, or in part, by Wellcome [110049/Z/15/Z & 110049/Z/15/A]. For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. Dr. Chamberlain’s role in this study was funded by a Wellcome Trust Clinical Fellowship (110049/Z/15/Z & 110049/Z/15/A). Dr. Chamberlain receives honoraria from Elsevier for editorial work. 1 Bloom P Religion, Morality, Evolution Annu Rev Psychol 2012 63 1 179 99 21943167 2 Shaver JH Purzycki G Sosis R Evolutionary Theory The Oxford Handbook of the Study of Religion Oxford University Press New York 2016 cited 2017 Oct 17 [Internet]Available from: http://www.oxfordhandbooks.com/view/10.1093/oxfordhb/9780198729570.001.0001/oxfordhb-9780198729570-e-9 3 Moreira-Almeida A Lotufo Neto F Koenig HG Religiousness and mental health: a review Rev Bras Psiquiatr 2006 Sep 28 3 242 50 16924349 4 Bonelli RM Koenig HG Mental disorders, religion and spirituality 1990 to 2010: a systematic evidence-based review J Relig Health 2013 Jun 52 2 657 73 23420279 5 George LK Ellison CG Larson DB Target article: Explaining the relationships between religious involvement and health Psychol Inq 2002 Jul 1 13 3 190 200 6 Koole SL McCullough ME Kuhl J Roelofsma PHMP Why religion’s burdens are light: from religiosity to implicit self-regulation Personal Soc Psychol Rev Off J Soc Personal Soc Psychol Inc 2010 Feb 14 1 95 107 7 McCullough ME Willoughby BLB Religion, self-regulation, and self-control: Associations, explanations, and implications Psychol Bull 2009 Jan 135 1 69 93 19210054 8 Wood C Ritual well-being: toward a social signaling model of religion and mental health Relig Brain Behav 2017 Jul 3 7 3 223 43 9 Carter EC McCullough ME Kim-Spoon J Corrales C Blake A Religious people discount the future less Evol Hum Behav 2012 May 1 33 3 224 31 10 Paglieri F Borghi AM Colzato LS Hommel B Scorolli C Heaven can wait. How religion modulates temporal discounting Psychol Res 2013 Nov 77 6 738 47 23344901 11 Hofmann W Luhmann M Fisher RR Vohs KD Baumeister RF Yes, but are they happy? Effects of trait self-control on affective well-being and life satisfaction J Pers 2014 Aug 82 4 265 77 23750741 12 Walker C Ainette MG Wills TA Mendoza D Religiosity and substance use: test of an indirect-effect model in early and middle adolescence Psychol Addict Behav J Soc Psychol Addict Behav 2007 Mar 21 1 84 96 13 Longshore D Anglin MD Conner BT Are religiosity and spirituality useful constructs in drug treatment research? J Behav Health Serv Res 2008 36 177 188 19023659 14 Grodzicki J Galanter M Spirituality and addiction Subst Abus 2005 26 1 4 15 Galanter M Spirituality and recovery in 12-step programs: An empirical model J Subst Abuse Treat 2007 33 265 272 17889297 16 Jennings TL Lyng T Gleason N Finotelli I Coleman E Compulsive sexual behavior, religiosity, and spirituality: A systematic review J Behav Addict 2021 Dec 31 10 4 854 878 34971357 17 Grant JE Potenza MN Weinstein A Gorelick DA Introduction to Behavioral Addictions Am J Drug Alcohol Abuse 2010 Sep 36 5 233 41 20560821 18 Koenig HG Büssing A The Duke University Religion Index (DUREL): A Five-Item Measure for Use in Epidemological Studies Religions 2010 Dec 1 1 1 78 85 19 Grant JE Impulse Control Disorders: A Clinician’s Guide to Understanding and Treating Behavioral Addictions W W Norton Company New York 2008 224 20 Odlaug BL Grant JE Impulse-Control Disorders in a College Sample: Results From the Self-Administered Minnesota Impulse Disorders Interview (MIDI) Prim Care Companion CNS Disord 2010 Apr 22 12 2 0 21 Saunders JB Aasland OG Babor TF de la Fuente JR Grant M Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption--II Addict Abingdon Engl 1993 Jun 88 6 791 804 22 Skinner HA The drug abuse screening test Addict Behav 1982 7 4 363 71 7183189 23 Yudko E Lozhkina O Fouts A A comprehensive review of the psychometric properties of the Drug Abuse Screening Test J Subst Abuse Treat 2007 Mar 32 2 189 98 17306727 24 Kroenke K Spitzer RL Williams JB The PHQ-9: validity of a brief depression severity measure J Gen Intern Med 2001 Sep 16 9 606 13 11556941 25 Prins A Ouimette P Kimerling R Cameron RP Hugelshofer DS Shaw-Hegwer J The primary care PTSD screen (PC-PTSD): Development and operating characteristics Prim Care Psychiatry 2003 9 9 14 26 Spitzer RL Kroenke K Williams JBW Löwe B A brief measure for assessing generalized anxiety disorder: the GAD-7 Arch Intern Med 2006 May 22 166 10 1092 7 16717171 27 Kessler RC Adler L Ames M Demler O Faraone S Hiripi E The World Health Organization Adult ADHD Self-Report Scale (ASRS): a short screening scale for use in the general population Psychol Med 2005 Feb 35 2 245 56 15841682 28 Rosenberg M Society and the adolescent self-image Princeton University Press Princeton, NJ 1965 29 Patton JH Stanford MS Barratt ES Factor structure of the Barratt impulsiveness scale J Clin Psychol 1995 Nov 51 6 768 74 8778124 30 Stanford MS Mathias CW Dougherty DM Lake SL Anderson NE Patton JH Fifty years of the Barratt Impulsiveness Scale: An update and review Personal Individ Differ 2009 Oct 47 5 385 95 31 Chamberlain SR Grant JE Initial validation of a transdiagnostic compulsivity questionnaire: The Cambridge-Chicago Compulsivity Trait Scale CNS Spectr Press 2017 28 Cohen J Statistical Power Analysis for the Behavioral Sciences 2nd ed Erlbaum Hillsdale, NJ 1988 31 Tiego J Trender W Hellyer PJ Grant JE Hampshire A Chamberlain SR Measuring compulsivity as a self-reported multi-dimensional transdiagnostic construct: Large-scale (N=182,000) validation of the Cambridge-Chicago Compulsivity Trait Scale Submitted 2022 Caribé AC Rocha MF Junior DF Studart P Quarantini LC Guerreiro N Miranda-Scippa  Religiosity and Impulsivity in Mental Health: Is There a Relationship? J Nerv Ment Dis 2015 Jul 203 7 551 4 10.1097/NMD.0000000000000316 26020819 Ameri Z Mirzakhani F Nabipour AR Khanjani N Sullman MJM The Relationship Between Religion and Risky Behaviors Among Iranian University Students J Relig Health 2017 Dec 56 6 2010 2022 10.1007/s10943-016-0337-1 27990616 Panza KE Kline AC Na PJ Potenza MN Norman SB Pietrzak RH Epidemiology of DSM-5 alcohol use disorder in U.S. military veterans: Results from the National Health and Resilience in Veterans Study Drug Alcohol Depend 2022 Feb 1 231 109240 10.1016/j.drugalcdep.2021.109240 Epub 2021 Dec 27 34974271 Santos LL Netto LR Cavalcanti-Ribeiro P Pereira JL Souza-Marques B Argolo F Lira SB Fontes G Moreira EC Anthony JC Koenen KC Trauma, Anxiety Disorders Study Group-TADSG. Drugs age-of-onset as a signal of later post-traumatic stress disorder: Bayesian analysis of a census protocol Addict Behav 2022 Feb 125 107131 10.1016/j.addbeh.2021.107131 Epub 2021 Oct 13 34763301 Sharma V Marin DB Koenig HK Feder A Iacoviello BM Southwick SM Pietrzak RH Religion, spirituality, and mental health of U.S. military veterans: Results from the National Health and Resilience in Veterans Study J Affect Disord 2017 Aug 1 217 197 204 10.1016/j.jad.2017.03.071 Epub 2017 Apr 11 28415007

PMC007xxxxxx/PMC7614800.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0376331 J Psychiatr Res J Psychiatr Res Journal of psychiatric research 0022-3956 1879-1379 33189357 7614800 10.1016/j.jpsychires.2020.11.004 EMS178725 Article Eating disorders with over-exercise: A cross-sectional analysis of the mediational role of problematic usage of the internet in young people Ioannidis Konstantinos abc Hook Roxanne W b Grant Jon E d Czabanowska Katarzyna ce Roman-Urrestarazu Andres abc‡ Chamberlain Samuel R ab‡ a Department of Psychiatry, University of Cambridge, UK b Cambridge and Peterborough NHS Foundation Trust, Cambridge, UK c Care and Public Health Research Institute, Maastricht University, Maastricht, Netherlands d Department of Psychiatry and Behavioral Neuroscience, University of Chicago, Chicago, Illinois, USA e Institute of Public Health, Faculty of Health Sciences, Jagiellonian University, Krakow, Poland Corresponding authors: Konstantinos Ioannidis, Consultant Psychiatrist, Eating Disorder Service, Addenbrookes Hospital, Hills Road, Cambridge, CB2 0QQ, UK; ioannik@doctors.org.uk AND Samuel R Chamberlain, Wellcome Trust Clinical Fellow, Department of Psychiatry, University of Cambridge, Herchel Smith Building for Brain and Mind Sciences, Forvie Site, Robinson Way Cambridge Biomedical Campus, Cambridge, CB2 0SZ, UK; srchamb@gmail.com ‡ Joint last authorship 01 1 2021 04 11 2020 10 7 2023 25 7 2023 132 215222 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Background Eating disorders are widespread illnesses with significant impact. There is growing concern about how young people overuse online resources leading to mental health sequelae. Methods We gathered data from 639 individuals from a population cohort. Participants were all young adults at the point of contact and were grouped as having probable eating disorder with excessive exercise (n=37) or controls (n=602). We measured obsessionality, compulsivity, impulsivity, and problematic internet use. Group differences in these domains were evaluated; and structural equation modelling (SEM) was used to assess structural relationships between variables. Results Cases had higher scores of obsessional thoughts of threat (Cohen’s d=0.94, p <0.001), intolerance towards uncertainty (Cohen’s d=0.72; p <0.001), thoughts of importance and control (Cohen’s d=0.65, p <0.01), compulsivity (Cohen’s d=0.72; p <0.001), negative urgency (Cohen’s d=0.75, p<0.001), and higher problematic usage of the internet (Cohen’s d=0.73; p-corrected <0.001). Our SEM showed significant partial mediation of problematic internet use on both the effect of obsessionality latent factor on cases (z-value=2.52, p<0.05), as well as of sensation seeking latent factor on cases (z-value=2.09, p<0.05). Discussion Youth with eating disorder and heightened exercise levels have increased obsessive thoughts of threat, compulsivity traits and sensation seeking impulsivity. The association between obsessive thoughts and eating disorders, as well as sensation seeking and eating disorder symptoms were partially mediated by problematic internet use. Excessive use of online resources may be playing a role in the development or maintenance of eating disorder symptoms in the background of obsessional thoughts and sensation seeking impulsive traits. eating disorder anorexia nervosa bulimia nervosa internet addiction problematic internet use pmcIntroduction Eating disorders (EDs) have the highest morbidity and mortality of all mental illnesses (Arcelus, Mitchell, Wales, & Nielsen, 2011) and affect a significant proportion of the population. Depending on the cohort and definition, anorexia nervosa (AN) has a lifetime prevalence of between 1.2% and 4.3% (broad definition) in females (Smink, Van Hoeken, & Hoek, 2012) and 0.24% in men, whereas bulimia nervosa (BN) has a lifetime prevalence of 1.0-2.9% in females, and 0.5% in men (Smink et al., 2012). Incidence of AN has increased 50-fold since the 1930s and has remained relatively stable since the 1970s (Hoek, 2006); however, some studies suggest an ongoing increase of incidence in younger populations (Zipfel, Giel, Bulik, Hay, & Schmidt, 2015) and eating disorders still remain an important health burden for societies worldwide (Erskine, Whiteford, & Pike, 2016; Treasure et al., 2015). Problematic usage of the internet and eating disorders Over the last decade, there has been growing concern over the impact of online media on eating disorders (Ioannidis et al., 2020; Mingoia, Hutchinson, Wilson, & Gleaves, 2017; Tiggemann & Miller, 2010; Tiggemann & Slater, 2013, 2017). Problematic usage of the internet (PUI) is an umbrella term used to describe maladaptive behaviors manifesting on the online milieu (Fineberg et al., 2018) and PUI is an age-related multifaceted construct that encompasses a number of maladaptive online behaviors (Ioannidis et al., 2018) linked with heightened levels of psychiatric comorbidity (Ho et al., 2014). Cross-sectional correlations between PUI and eating disorder psychopathology, body dissatisfaction, restrained eating, drive for thinness have been shown in meta-analysis (Ioannidis et al., 2020), while studies of bulimia and PUI show similar correlations and group comparisons (Butkowski, Dixon, & Weeks, 2019; Melioli, Rodgers, Rodrigues, & Chabrol, 2015; Smith, Hames, & Joiner, 2013; Tao, 2013). A number of prospective studies support the notion that effects of PUI on eating disorders do exist and exposure to particular types of online content e.g. social networking site (SNS) use may have accumulating effects over time (de Vries, Peter, de Graaf, & Nikken, 2016; Ferguson, Muñoz, Garza, & Galindo, 2014; Hsieh, Hsiao, Yang, Liu, & Yen, 2018; Hummel & Smith, 2015; Smith et al., 2013; Tiggemann & Slater, 2017). Experimental studies in the field have demonstrated direct effects of SNS usage or consumption of pro-ED content (e.g. “fitspiration”) on body dissatisfaction, internalization of the thin ideal and weight and shape concerns (Fardouly, Diedrichs, Vartanian, & Halliwell, 2015; Mabe, Forney, & Keel, 2014; Prichard, Kavanagh, Mulgrew, Lim, & Tiggemann, 2020; Slater, Halliwell, Jarman, & Gaskin, 2017; Tiggemann & Zaccardo, 2015). Excessive exercise Excessive exercise is a particularly challenging eating disorder behavior that can lead to catastrophic consequences e.g. precipitous weight loss, coupled with exercise through injury, heart abnormalities (e.g. life threatening bradycardia), rhabdomyolysis, among other complications (Ghoch, Calugi, & Grave, 2016; Peñas-Lledó, Vaz Leal, & Waller, 2002). AN cohorts present with excessive exercise in up to ~80% (Rizk, Lalanne, Berthoz, Kern, & Godart, 2015)and in a 15-year-prospective study of AN showed compulsive excessive exercise at the time of discharge being one of the most significant predictors of chronic outcome and early time to relapse (HR = 2.2, 95% CI = 1.1–4.9) (Strober, Freeman, & Morrell, 1997). Excessive exercise has been linked with the consumption of “thinspiration” or “fitspiration” online content (Carrotte, Vella, & Lim, 2015; Quesnel, Cook, Murray, & Zamudio, 2018), as well as weight loss and fitness applications (Apps) in both males and females (Almenara, Machackova, & Smahel, 2019; Embacher Martin, McGloin, & Atkin, 2018; Levinson, Fewell, & Brosof, 2017; Linardon & Messer, 2019; Simpson & Mazzeo, 2017). Obsessionality, intolerance of uncertainty and compulsivity in eating disorders Obsessional ideas about own body image are core symptoms of AN (Collier & Treasure, 2004) and have causally been linked to starvation since the early exploration of consequences of starvation to healthy individuals (Minnesota study) (Keys, Brozek, Henschel, Mickelsen, & Taylor, 1950). In the 1980s’ EDs were considered as extreme manifestations of societal obsession with thinness (Collier & Treasure, 2004) and we now know that AN has a genetic linkage with obsessionality on chromosome 1 locus (Devlin, 2002). Furthermore, restricting AN was demonstrated to have reduced cognitive flexibility both during AN episodes and after recovery (Tchanturia et al., 2004), while individuals with AN experience obsessional thoughts linked to their compulsive exercise, eating and weight related obsessionality (Byrne et al., 2018; Godier & Park, 2015). Linked to obsessional traits, ‘intolerance of uncertainty’ (IU) is the tendency for a negative emotional, cognitive and behavioral reaction to uncertain situations and events. Compulsive eating disorder behaviors have been linked to IU (Boswell, Thompson-Hollands, Farchione, & Barlow, 2013; Parkes et al., 2019). IU has been quantitatively demonstrated as prevalent in AN (Frank et al., 2012; Sternheim, Startup, & Schmidt, 2011) and qualitatively explored to show that IU in AN manifests as fear of unduly evaluation from others, leading to social problem solving difficulties (Sternheim, Danner, van Elburg, & Harrison, 2020) and compulsive planning and action (Sternheim, Konstantellou, Startup, & Schmidt, 2011). Compulsivity has been defined as a trait in which actions are persistently repeated despite adverse consequences (Robbins, Gillan, Smith, de Wit, & Ersche, 2012). Behavioral traits of compulsivity covary with eating disorder psychopathology (Godier & Park, 2014). Extreme dietary restriction and over-exercise may reflect excessive habit formation leading to compulsive starvation or over-activity behavior (Everitt & Robbins, 2005; Fladung et al., 2010). Impulsivity and sensation seeking in eating disorders Impulsivity is a multi-faceted construct referring to acting without forethought or reflection or consideration of the consequences. PUI has been linked with increased levels of trait impulsivity and compulsivity (Ioannidis et al., 2016) and sensation seeking (Lin & Tsai, 2002). Impulsivity in eating disorders has been linked to poor long-term AN outcomes (Fichter, Quadflieg, & Hedlund, 2006), but also strongly related with bulimia with or without purging and binge eating disorder (Collier & Treasure, 2004; Fahy & Eisler, 1993). Heightened sensation seeking impulsivity has been particularly demonstrated in bulimia as compared to controls (Rossier, Bolognini, Plancherel, & Halfon, 2000) even after controlling for victimization and traumatic experiences (Brewerton, Cotton, & Kilpatrick, 2018). Impulsivity, compulsivity and obsessionality, when considered together they are found as prevalent behaviors in purging anorexia (Hoffman et al., 2012). Obsessional thinking is strongly positively correlated with compulsive behavior(Kim et al., 2016). Impulsivity and compulsivity exist cross-diagnostically in latent functionally impairing forms which are positively correlated (Chamberlain, Stochl, Redden, & Grant, 2018). Aims and hypotheses This current study had two aims: first we aimed to compare the behavioral characteristics of eating disorder traits with heightened levels of exercise in respect to their levels of (1) impulsivity, (2) compulsivity, (3) obsessionality, (4) sensation seeking, (5) intolerance to uncertainty and (6) problematic usage of the internet against controls. By doing so, we aim to quantify differences on group level in our dataset, as they have been demonstrated in previous research, and establish that our cohort does share the behavioral characteristics in line with current literature. Therefore, we hypothesized that participants with eating disorders and heightened exercise will present with increased levels of trait impulsivity, compulsivity, obsessionality, as well as heightened levels of intolerance for uncertainty and sensation seeking when compared to controls. Our second aim would be to statistically explore the structural relationship between the variables in our model and consecutively the potential mediating effect that problematic internet use may have on these neurobiological dimensions on their effect on eating disorders. To date there is no study exploring those mediating effects of PUI in eating disorders. Methods Study criteria and recruitment Participants were recruited from the Neuroscience in Psychiatry Network (NSPN) UK youth cohort:, which is a longitudinal cohort, exploring brain development trajectories and mental health outcomes (Kiddle et al., 2018). The sample was originally recruited on an age-sex stratified basis, in order to maximize representativeness of the normal population in the catchment areas covered (Cambridge and London). In this study, we contacted all individuals (adults, Mean [sd] age: 23.4 [3.2]) who were still enrolled in this cohort at the time of data collection (2017–2018) via email and invited them to take part in an online study being conducted via SurveyMonkey. Participants received £15 compensation in the form of a gift voucher. Further methodological details about the recruitment and instruments are presented in previous work (Chamberlain et al., 2019). The data that support the findings of this study are available on request from the corresponding author, subject to agreement of the Chief Investigator. The data are not publicly available due to privacy or ethical restrictions Ethical considerations The procedures of this study were carried out in accordance with the Declaration of Helsinki and the study was approved by the Cambridge East Research Ethics Committee (Study approval number 16/EE/0260). All subjects gave informed consent online. Assessments Participants were classified as cases according to whether or not they displayed high probability of having an eating disorder diagnosis and excessive exercise. For the assessment of their eating disorder diagnosis we used the SCOFF Eating Disorder Questionnaire (Luck et al., 2002). This is a validated screening tool for detection of eating disorders, specifically either anorexia nervosa or bulimia nervosa. The scale is sensitive to the presence of different aspects of eating disorder symptoms, such as purging, weight loss, distorted body image, loss of control over eating and food preoccupation. A score of 2 and above indicates high likelihood of an eating disorder reason why we used this cut-off to define our cases. For the assessment of excessive exercise, we used the exercise addiction inventory (EAI) (Griffiths, Szabo, & Terry, 2005). Due to prior absence of an established cut-off, we defined excessive exercise based on scores being >1 standard deviation from that of the cohort that was examined (19 and above; the mean cohort score was 13.4). Therefore, our cases were characterized as probable eating disorder (either anorexia nervosa or bulimia nervosa) plus excessive exercise symptoms. Behavioral Assessments Problematic usage of the internet Problematic usage of the internet was quantified using the Internet Addiction Test short-12 item version (IAT-12) (Pawlikowski, Altstötter-Gleich, & Brand, 2013). The IAT-12 consists of twelve items ascertaining the level of problematic internet use, and was developed from Kimberley Young’s Internet Addiction Test, based on rigorous psychometric refinement of the original scale. Compulsivity Cambridge–Chicago Compulsivity Trait Scale (CHI-T) (Chamberlain & Grant, 2018). This is a scale designed to capture the comprehensive aspects of compulsivity, viewed trans-diagnostically. The scale comprises 15 items, each scored on a Likert scale of 1–4, from “strongly disagree” to “strongly agree.” The total score is 60, with higher scores indicating higher compulsivity. The scale is sensitive to compulsivity across a range of pathologies, such as disordered gambling, substance use, and obsessive-compulsive symptoms (Albertella et al., 2019). Impulsivity The Barratt Impulsiveness Scale short version (BIS-8) to quantify impulsive personality (Patton, Stanford, & Barratt, 1995) is a self-report questionnaire used to determine levels of impulsiveness. The short Urgency, Premeditation (lack of), Perseverance (lack of), Sensation Seeking, Positive Urgency, Impulsive Behavior Scale (S-UPPS) is a measurement of impulsivity as a multi-faceted and multi-dimensional construct, comprising five impulsive personality traits (Whiteside & Lynam, 2001). The proposed model of impulsivity includes five specific (distinct) dispositions a) sensation seeking (i.e. pursuit of novel/exciting stimuli), b) lack of planning (i.e. action without advanced planning), c) lack of perseverance (i.e., limited capacity for focus maintenance), d) positive urgency (i.e., rash action in response to intense positive emotions), and e) negative urgency (i.e., rash action in response to intense negative emotions) (Whiteside & Lynam, 2001). The S-UPPS scale has five subscales (first order factors) and four items per sub-scale. Sensation seeking comprises a second order factor alone, whilst positive and negative urgency comprise ‘emotion based rushed action’ and lack of premeditation and perseverance comprise ‘deficits in conscientiousness’ (Cyders, Littlefield, Coffey, & Karyadi, 2014). Sensation Seeking The Brief Sensation Seeking Scale (BSSS-8) is a measure of sensation seeking as a psychobiological trait of need for novelty, complexity and intensity (Hoyle, Stephenson, Palmgreen, Lorch, & Donohew, 2002). It comprises of eight items of a five-point-Likert scale from “strongly disagree” to “strongly agree”. Intolerance of Uncertainty Intolerance of Uncertainty scale (Buhr & Dugas, 2002). This is a scale original developed in French, but validated for English speakers as well. The scale comprises 27 items, each scored on a Likert scale of 1–5, from “not at all characteristic of me” to “entirely characteristic of me”; with higher scores indicating higher degree of intolerance to uncertainty. The measure of interest was the total score. Statistical analysis Data processing and statistical analyses were conducted using statistical software R version 3.4.2 and “dplyr” (Wickham, François, Henry, Müller, & RStudio, 2020) and “lavaan” (Rosseel, 2012) R packages. We performed direct comparisons of our cases and controls using student t-test under the assumption of normal distribution of behavioral characteristics in our cohort. The NSPN is a representative cohort of the catchment area and behavioral characteristics are expected to have normal distributions. We used chi-square to compare non-parametric values e.g. gender. Finally, we also performed a structural equation modelling (SEM) to explore structural relationships between the variables at hand; this also enabled us to ascertain whether problematic internet use has any mediation influence on the effect of behavioral traits on eating disorder cases. Our SEM initial (hypothesized) model included four latent variables predicted by manifest variables as such: a) ‘Obsessionality’ latent factor predicted by the four subscales of OBQ (“Obsessional thoughts of threat”, “Obsessional intolerance towards uncertainty”, “Obsessional thoughts of importance and control”, “Obsessional thoughts of inflated responsibility”; b) “Compulsivity” latent factor predicted by the two factors of CHI-T “reward-seeking and need for perfection” and “anxiolytic/soothing compulsivity”; c) “Impulsivity” latent factor predicted by the four factors of S-UPPS “Negative urgency”, “Lack of perseverance”, “Lack of premeditation”, “Positive urgency”, also “BIS-8 total score”; d) “Sensation seeking” latent factor predicted by “BSSS total score” and “S-UPPS Sensation seeking”. We hypothesized that those variables were predictive of problematic usage of the internet “IAT-12 total score”, as well as “cases”, as defined above (SCOFF≥2, EAI≥19). We also used “Intolerance to Uncertainty” as (IUS total score) as a separate path, predicting both PUI and “cases”. We calculated regression coefficients for all predictors as well as covariances between all latent variables between themselves and with IUS. We calculated the indirect effect of problematic internet use for every latent and IUS variable on cases. Our initial (hypothesized) model was plotted and presented in Figure 1. We then followed a step wise change of our model by adding relationships based on their modification indices and subtracting relationships based on non-significant covariances, aiming to improve the model’s goodness of fit statistics. We added new relationships with high modification indices taking into account the theoretical implications of adding those relationships into the model. We calculated the degrees of freedom, goodness and badness of fit statistics (“AIC” = Akaike information criterion; “CFI” = comparative fit index; “TLI” = Tucker–Lewis index; “RMSEA” = root mean square error) for every model and compared each model with the previous one using chi-square comparisons. We finalized our SEM when reached non-significant improvement in our model via path change (Schermelleh-Engel, Moosbrugger, & Müller, 2003). Results Our final sample comprised 37 cases (i.e. individuals meeting criteria for probable eating disorder plus having excessive exercise), and 602 controls. Group comparison results are shown in Table 1. Case control comparison Cases had higher scores of obsessional thoughts of threat (t-test, Cohen’s d=0.94, p<0.001), obsessional intolerance towards uncertainty (OBQ) (Cohen’s d=0.72; p<0.001), obsessional thoughts of importance and control (Cohen’s d=0.65, p<0.01), transdiagnostic compulsivity traits (Cohen’s d=0.72; p<0.001), negative urgency (Cohen’s d=0.75, p<0.001), intolerance of uncertainty (IUS) (Cohen’s d=0.55; p <0.05), and higher problematic usage of the internet (Cohen’s d=0.73; p corrected <0.001). All other results were non-significant or became non-significant after the application of Bonferroni correction for multiple comparisons. We used standardized mean difference (Cohen’s d) under the assumption of normality and homogeneity of variances. Structural equation modelling Problematic internet use was associated with obsessionality (regression coef. z=7.61, p<0.001) and sensation seeking (z=4.65, p<0.001); eating disorder case was associated with obsessionality (z=4.53, p<0.001), sensation seeking (z=2.26, p<0.02) and PUI (z=2.32, p<0.02). The chosen model with the lowest RMSEA was model 8 (see Table 2) which had Comparative Fit Index (CFI) 0.975 and Tucker-Lewis Index (TLI) 0.956 indicating good fit. Root Mean Square Error of Approximation was less than 0.1 (mean=0.059; 1000-iterations-bootstrap 95%CI 0.059-0.060). The indirect (mediation) effect of PUI on obsessionality effect on cases was statistically significant (z=2.25, p=0.024) indicating partial mediation (Obsessionality ~ case standardized effect reduction from 0.24 to 0.21 [12.5% reduction]). The indirect (mediation) effect of PUI on sensation seeking effect on cases was statistically significant (z=2.08, p=0.037) indicating partial mediation (Sensation seeking ~ case standardized effect reduction from 0.11 to 0.09 [18% reduction]). Initial models did not have acceptable goodness of fit statistics and were rejected (see Table 2). Full mediation SEM results are presented in Table 3. Model 8 (chosen model) is graphically presented in Figure 2. Comparative statistics between hypothesized model and final model are presented in Table 2. Discussion This is the first study to examine the problematic online behaviors, coupled with behavioral characteristics of a putative eating disorders cohort with heightened excessive exercise behaviors. In our study, we identified, through group comparisons, that cases, as compared to controls, had heightened degree of obsessive thoughts of threat, obsessional intolerance towards uncertainty, obsessional thoughts of importance and control, high cross-diagnostic traits of compulsivity, negative urgency impulsivity and higher levels of problematic internet use. Those results are in line with previous research, that obsessional preoccupation with food and food predominance, as well as and intolerance to uncertainty manifesting with deficits in social decision making, planning and action, as well as fear of unduly evaluation from others (Davis & Kaptein, 2006; Sternheim, Konstantellou, et al., 2011). Increased compulsivity, manifesting as compulsive restriction of food intake and compulsive exercise is also in line with previous research (Davis & Kaptein, 2006; Godier & Park, 2014), as well as a higher level of negative urgency impulsive, particularly in cohorts of heightened impulsivity during negative emotional states (e.g. binge/purging AN or BN) (Westwater et al., 2019). The increased level of PUI is also in line with previous research. Furthermore, our study is the first to explore the mediation effect of problematic internet behaviors on the impact of obsessionality and sensation seeking, to eating disorder symptoms, via SEM. Our analysis showed partial mediation, for both effects of obsessionality (0.24 to 0.21) and sensation seeking (0.11 to 0.09) suggestive that obsessional thoughts and sensation seeking impulsivity traits, when present in in young populations, may be impacting on the development or perseverance of eating disorders, partially via the problematic usage of online resources. While a SEM analysis does not provide evidence for causal directional link, it highlights the importance for future studies that can potentially examine this interaction further. Previous longitudinal research has shown that the use of social media (Facebook) maintained weight and shape concerns as well as state anxiety (Mabe et al., 2014) as compared to alternate online activity. Also, social media use has been associated with perseverance of obsessional body image symptoms (Tiggemann & Slater, 2013) and found to causally associate with obsessive drive for thinness longitudinally (Tiggemann & Slater, 2017). We argue that our mediation model is grounded on robust theory of obsessional thoughts and sensation seeking behavior strongly associate with both with PIU and ED and the mediation pathway in proposition is both statistically demonstrable and theoretically plausible. We argue that enhancing our understanding of the behavioral underpinnings of this effect may be helpful in the developing appropriate interventions and therapeutic targets, including health recommendations about the use of novel technology, digital interventions and appropriate clinical interviews and screening of symptoms. Obsessional thoughts linked with compulsive usage of the internet (e.g. calorie counting via apps, fitness apps, obsessing over body image content consumption, step counting etc.) and sensation seeking online behaviors (e.g. consumption of ‘fitspiration’ or food related or ‘ mukbang’ content etc.) may be potential such targets. Finally, the current manuscript is prepared in the unusual times of the COVID-19 pandemic. The global social distancing measures have driven people to rely more that even on online resources for their work, leisure and social connectedness. It is unclear what effect this may have, but it is possible that we may see higher levels of problematic usage of the internet in the future; this may mean that it would be pertinent for future research to unravel the causal links between behavioral traits predisposing for both PUI and EDs, to enable us to think about how to target those in our diagnostics, therapies and prevention programs. Limitations We have several limitations to consider in this study deriving from our data collection process and instruments used. Given that this is an online survey, it has less quality control and less accuracy for measuring psychopathology constructs as compared to face-to-face clinical assessments. For example, we used the SCOFF questionnaire to ascertain putative AN or BN diagnosis. While the SCOFF is an efficient screening tool for AN and BN, and its specificity and positive predictive value are reasonable (Spec.: 89.6%, PPV: 24.4%) (Luck et al., 2002) for a screening tool, it does not have ‘gold-standard’ diagnostic validity that can be provided by a clinical or DSM-5 structured interview. Furthermore, due to the survey being delivered online, there is also a potential sampling bias, since returning participants of the NSPN cohort may be those who are more technologically adept or responsive to email requests. In respect to our SEM analysis, it is important to note that mediational models are presumptuously causal models in which the mediator is presumed to cause the outcome and not vice versa (Baron & Kenny, 1986). Here, we model on the basis that latent cross-diagnostic traits e.g. sensation seeking, compulsivity, obsessionality are factors predisposing to eating disorder behaviors, however, we cannot draw causal effects; this would require a different study design. Future research with appropriate (longitudinal, randomized, controlled) design can explore further whether those causal links exist and in which direction. Furthermore, for our SEM we used the CHI-T two-factor structure as reported in first publication (Chamberlain & Grant, 2018), however this factor analysis is considered preliminary; future research on the instrument in larger samples may replicate this finding or demonstrate a different factor structure for the instrument. It is important to note that BSSS total score and sensation seeking UPPS scores considered individually were not statistically significantly higher for cases in group comparisons; however, the sensation seeking latent factor was predictive of cases (p=0.02, see Table 3). This may imply that latent sensation seeking as modelled in our SEM brings together a wider range of sensation seeking parameters, rendering the latent construct predictive of eating disorder symptoms, an attribute that the instruments may not possess if considered individually. Conclusion We have shown that our case group of putative eating disorders with heightened levels of activity have increased levels of obsessionality, cross-diagnostic compulsivity, negative urgency impulsivity, intolerance of uncertainty and higher levels of problematic usage of the internet. We have demonstrated that obsessionality latent factor and sensation seeking latent factor predict cases, and that problematic usage internet resources mediates that relationship. This mediation provides us novel insight into the potential role of problematic use of online resources for the development and perseverance of eating disorder psychopathology with heightened exercise levels. Acknowledgement We are indebted to the volunteers who participated in the study. Figure 1 Structural equation model hypothesized model Legend: IUS = intolerance to uncertainty score; IU = intolerance to uncertainty, as directly measured by IUS; SS = sensation seeking latent factor; BSSS = Brief Sensation Seeking Scale (BSSS) total score; BIS-8 = The Barratt Impulsiveness Scale, short version total score; Imp = impulsivity latent factor; SUPPS = Short UrgencyPremeditation-Perseverance-Sensation Seeking-Positive Urgency Scale (SUPPS); SUPPS pu = SUPPS Positive urgency; SUPPS pr = SUPPS (Lack of) perseverance; SUPPS pm = SUPPS (Lack of) premeditation; SUPPS nu = SUPPS Negative urgency; CHI-T = Cambridge-Chicago Compulsivity Trait Scale (CHI-T); CHI-T rs = CHI-T reward-seeking and need for perfection; CHI-T as = CHI-T anxiolytic/soothing compulsivity; Comp = compulsivity latent factor; Obs = obsessionality latent factor; OBQ = Obsessive Beliefs Questionnaire, short version (OBQ-20); OBQ m = Obsessional thoughts of importance and control; OBQ n = Obsessional intolerance towards uncertainty; OBQ p = Obsessional thoughts of inflated responsibility; OBQ t = Obsessional thoughts of threat; Problematic Internet use = Internet Addiction Test, short version (IAT-12) total score; Eating disorder = case; Numeric scores are standardized regression coefficients in direct lines and standardized covariance coefficients in curves lines. Figure 2 Structural equation model chosen model Legend: BSSS = Brief Sensation Seeking Scale (BSSS) total score; BIS-8 = The Barratt Impulsiveness Scale, short version total score; Imp = impulsivity latent factor; OBQ = Obsessive Beliefs Questionnaire, short version (OBQ-20); OBQ m = Obsessional thoughts of importance and control; OBQ n = Obsessional intolerance towards uncertainty; OBQ p = Obsessional thoughts of inflated responsibility; OBQ t = Obsessional thoughts of threat; Problematic Internet use = Internet Addiction Test, short version (IAT-12) total score; Eating disorder = case; Numeric scores are standardized regression coefficients in direct lines and standardized covariance coefficients in curves lines. Table 1 Demographic and behavioral characteristics of study cohort TOTAL N = 639 Controls N = 502 ED Low exercise N= 100 ED High exercise N= 37 Group t-test comparison p-value† Signif. †† Cohen’s d Mean (sd) Mean (sd) Mean (sd) Mean (sd) Age 23.4(3.2) 23.4(3.1) 23.4(3.9) 23.4(3.1) - - - - Gender [%Female] 65% 64.7% 70% A < B Obsessional thoughts of threat (OBQ) 15.7 (6.1) 14.8(5.8) 18.1 (6.4) 21.0 (5.0) A < B < C Obsessional intolerance towards uncertainty (OBQ) 18.2 (6.9) 17.3 (6.6) 20.6 (7.5) 22.8 (6.2) A < B , C Obsessional thoughts of importance and control (OBQ) 14.1 (6.2) 13.4 (5.9) 16.4 (6.7) 17.8 (6.0) A < B , C Obsessional thoughts of inflated responsibility (OBQ) 20.9 (6.3) 20.7 (6.4) 21.4 (6.3) 22.6 (5.6) - Transdiagnostic compulsivity traits (CHI-T) 24.3 (6.0) 23.6 (5.9) 26.2 (6.0) 28.3 (4.8) A < B < C Impulsivity traits (BIS-8) 16.4 (3.9) 15.9 (3.7) 18.4 (4.3) 17.1 (3.6) A < B Negative urgency (SUPPS) 4.74 (2.49) 4.3 (2.3) 6.4 (2.4) 6.5 (2.1) A < B , C Lack of perseverance (SUPPS) 4.44 (1.85) 4.4 (1.8) 4.8 (2.1) 3.7 (1.6) A, B > C Lack of premeditation (SUPPS) 3.95 (1.85) 3.8 (1.8) 4.7 (2.0) 3.9 (2.2) A < B Sensation seeking (SUPPS) 5.99 (2.60) 6.1 (2.6) 5.4 (2.6) 6.6 (3.1) A, C > B Positive urgency (SUPPS) 3.14 (2.14) 3.0 (2.1) 3.8 (2.2) 4.0 (2.1) A < B, C Intolerance of Uncertainty (IUS) 58.3 (21.1) 55.3 (20.0) 69.2 (22.7) 69.0 (18.9) A < B, C BSSS 24.11(6.8) 24.2 (6.6) 23 (7.1) 25.5 (7.0) - Internet use (IAT-12) 13.1(8.0) 11.7 (7.2) 16.6 (9.9) 18.5 (7.7) A < B, C Quality of life 55.3 (17.8) 57.3 (17.6) 46.5 (15.4) 50.9 (17.7) A > B, C Group A = controls; Group B = ED with low exercise; Group C = ED with high exercise; † Two sample t-test p-values; †† Significance: ‘*’ <0.05; ‘**’ <0.01; ‘***’ <0.001; ††† Chi-square; ED = Eating disorders; Obsessive Beliefs Questionnaire, short version (OBQ-20); Internet Addiction Test, short version (IAT-12); Cambridge–Chicago Compulsivity Trait Scale (CHI-T); The Barratt Impulsiveness Scale, short version (BIS-8); Brief Sensation Seeking Scale (BSSS); Short Urgency-Premeditation-Perseverance-Sensation Seeking-Positive Urgency Scale (SUPPS). Table 2 Structural equation modelling Model DF χ2 diff Pr (>Chisq) AIC CFI TLI RMSEA 95%CI RMSEA Path FIRST MODEL 90 - - 44816.52 0.653 0.538 0.158 0.158 - 0.158 - 2 89 187.68 *** 44624.60 0.706 0.603 0.146 0.146 - 0.147 comp_rsfr ~~ supps_lackpersevrnce 3 88 23.463 *** 44611.65 0.712 0.607 0.146 0.145 - 0.146 comp_rsfr ~~ bis8_total 4 87 122.89 *** 44488.21 0.748 0.653 0.137 0.137 - 0.137 bis8_total ~~ supps_lackpremed 5 67 147.12 *** 39466.08 0.735 0.640 0.143 0.142 - 0.143 Compulsivity =~ comp_rsfr + comp_arss 6 66 63.713 *** 39395.04 0.759 0.667 0.137 0.137 - 0.137 OBQ_perfec_intoluncert ~~ supps_lackpersevrnce 7 22 342.93 *** 28183.77 0.792 0.659 0.164 0.164 - 0.164 Impulsivity =~ exogenous 8 16 284.52 *** 23472.67 0.971 0.949 0.062 0.062 - 0.063 ius_total ~ 9 8 11.157 0.19 17743.64 0.970 0.944 0.072 0.071 - 0.072 SS =~ bsss_total + supps_sensseek Legend: Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 ; DF = Degrees of freedom; χ2 diff = chi square difference; Pr(>Chisq) = p-value for the chi square test, tests compare consecutive models; AIC = Akaike information criterion; CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation; 95%CI RMSEA = 95% confidence intervals for RMSEA; Path = step change for each model. All results are averages and 95%CI intervals from 1000 iteration boot strap estimates of model fit. Table 3 Structural equation regressions, total and indirect effects of H8 model Regressions Estimate Standard Errors z-value P(>|z|) Std.all Case by ~ Obsessionality 0.03 0.005 5.15 <0.001 0.26 Sensation seeking 0.01 0.005 2.01 0.04 0.08 PUI 0.01 0.003 3.34 0.001 0.15 PUI by ~ Obsessionality 0.463 0.074 6.23 <0.001 0.30 Sensation seeking 0.275 0.069 3.97 <0.001 0.16 Total effects Obsessionality on case 0.032 0.005 6.23 <0.001 0.30 Sensation seeking on case 0.012 0.005 2.59 0.01 0.11 Indirect effects PUI~obsessionality 0.005 0.002 3.02 0.002 0.044 PUI~sens.seeking 0.003 0.001 2.57 0.01 0.024 Legend: SS = sensation seeking latent factor; Obs = obsessionality latent factor; OBQ = Obsessive Beliefs Questionnaire, short version (OBQ-20); OBQ m = Obsessional thoughts of importance and control; OBQ n = Obsessional intolerance towards uncertainty; OBQ p = Obsessional thoughts of inflated responsibility; OBQ t = Obsessional thoughts of threat; Problematic Usage of the Internet (PUI) = Internet Addiction Test, short version (IAT-12) total score; z-values = regression coefficients; Std.all = standardized coefficients; Case = SCOFF > 2, Exercise Addiction Inventory > 18. Disclosures Dr Chamberlain’s involvement in this research was funded by a Wellcome Trust Clinical Fellowship (110049/Z/15/Z). Dr Chamberlain consults for Promentis and Ieso; and receives stipends from Elsevier for journal editorial work. Dr Grant reports grants from the National Center for Responsible Gaming, Forest Pharmaceuticals, Takeda, Brainsway, and Roche, and others from Oxford Press, Norton, McGraw-Hill, and American Psychiatric Publishing outside of the submitted work. Authors received no funding for the preparation of this manuscript. The other authors report no financial relationships with commercial interest. Dr Roman-Urrestarazu work received funding from the Gillings Fellowship in Global Public Health Grant Award YOG054 and the Commonwealth Fund with a Harkness Fellowships in Health Care Policy and Practice 2020-2021. Author contributions KI designed the idea for the manuscript, analyzed the data, wrote the majority of the manuscript and coordinated the co-authors’ contributions. SRC, RH designed and coordinated the study and collected and managed the data. All authors read and approved the final manuscript and contributed to the drafting and revising of the paper as well as to interpreting the results. Albertella L Chamberlain SR Le Pelley ME Greenwood L-M Lee RSC Den Ouden L Yücel M Compulsivity is measurable across distinct psychiatric symptom domains and is associated with familial risk and reward-related attentional capture CNS Spectrums 2019 1 8 10.1017/s1092852919001330 Almenara CA Machackova H Smahel D Sociodemographic, Attitudinal, and Behavioral Correlates of Using Nutrition, Weight Loss, and Fitness Websites: An Online Survey Journal of Medical Internet Research 2019 21 4 e10189 10.2196/10189 30946018 Arcelus J Mitchell AJ Wales J Nielsen S Mortality Rates in Patients With Anorexia Nervosa and Other Eating Disorders Archives of General Psychiatry 2011 68 7 724 10.1001/archgenpsychiatry.2011.74 21727255 Baron R Kenny D The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations Journal of Personality and Social Psychology 1986 51 6 1173 1182 https://www2.psych.ubc.ca/~schaller/528Readings/BaronKenny1986.pdf 3806354 Boswell JF Thompson-Hollands J Farchione TJ Barlow DH Intolerance of Uncertainty: A Common Factor in the Treatment of Emotional Disorders Journal of Clinical Psychology 2013 69 6 630 645 10.1002/jclp.21965 23381685 Brewerton TD Cotton BD Kilpatrick DG Sensation seeking, binge-type eating disorders, victimization, and PTSD in the National Women’s Study Eating Behaviors 2018 30 120 124 10.1016/j.eatbeh.2018.07.001 30012278 Buhr K Dugas MJ The intolerance of uncertainty scale: Psychometric properties of the English version Behaviour Research and Therapy 2002 40 8 931 945 10.1016/S0005-7967(01)00092-4 12186356 Butkowski CP Dixon TL Weeks K Body Surveillance on Instagram: Examining the Role of Selfie Feedback Investment in Young Adult Women’s Body Image Concerns Sex Roles 2019 81 5-6 385 397 10.1007/s11199-018-0993-6 Byrne CE Wonderlich JA Curby T Fischer S Lock J Le Grange D Using bivariate latent basis growth curve analysis to better understand treatment outcome in youth with anorexia nervosa European Eating Disorders Review 2018 26 5 483 488 10.1002/erv.2596 29691947 Carrotte ER Vella AM Lim MSC Predictors of “Liking” Three Types of Health and Fitness-Related Content on Social Media: A Cross-Sectional Study Journal of Medical Internet Research 2015 17 8 e205 10.2196/jmir.4803 26297689 Chamberlain SR Grant JE Initial validation of a transdiagnostic compulsivity questionnaire: The Cambridge-Chicago Compulsivity Trait Scale CNS Spectrums 2018 23 5 340 346 10.1017/S1092852918000810 29730994 Chamberlain SR Stochl J Redden SA Grant JE Latent traits of impulsivity and compulsivity: Toward dimensional psychiatry Psychological Medicine 2018 48 5 810 821 10.1017/S0033291717002185 28805173 Chamberlain SR Tiego J Fontenelle LF Hook R Parkes L Segrave R Yücel M Fractionation of impulsive and compulsive trans-diagnostic phenotypes and their longitudinal associations Australian and New Zealand Journal of Psychiatry 2019 53 9 896 907 10.1177/0004867419844325 31001986 Collier DA Treasure JL The aetiology of eating disorders British Journal of Psychiatry 2004 November 185 363 365 10.1192/bjp.185.5.363 Cyders MA Littlefield AK Coffey S Karyadi KA Examination of a short English version of the UPPS-P Impulsive Behavior Scale Addictive Behaviors 2014 39 9 1372 1376 10.1016/j.addbeh.2014.02.013 24636739 Davis C Kaptein S Anorexia nervosa with excessive exercise: A phenotype with close links to obsessive-compulsive disorder Psychiatry Research 2006 142 2-3 209 217 10.1016/j.psychres.2005.11.006 16697469 de Vries DA Peter J de Graaf H Nikken P Adolescents’ Social Network Site Use, Peer Appearance-Related Feedback, and Body Dissatisfaction: Testing a Mediation Model Journal of Youth and Adolescence 2016 45 1 211 224 10.1007/s10964-015-0266-4 25788122 Devlin B Linkage analysis of anorexia nervosa incorporating behavioral covariates Human Molecular Genetics 2002 11 6 689 696 10.1093/hmg/11.6.689 11912184 Embacher Martin K McGloin R Atkin D Body dissatisfaction, neuroticism, and female sex as predictors of calorie-tracking app use amongst college students Journal of American College Health 2018 66 7 608 616 10.1080/07448481.2018.1431905 29405860 Erskine HE Whiteford HA Pike KM The global burden of eating disorders Current Opinion in Psychiatry 2016 October 1 29 346 353 10.1097/YCO.0000000000000276 27532942 Everitt BJ Robbins TW Neural systems of reinforcement for drug addiction: From actions to habits to compulsion Nature Neuroscience 2005 November 8 1481 1489 10.1038/nn1579 16251991 Fahy T Eisler I Impulsivity and eating disorders British Journal of Psychiatry 1993 FEB 162 193 197 10.1192/bjp.162.2.193 Fardouly J Diedrichs PC Vartanian LR Halliwell E Social comparisons on social media: The impact of Facebook on young women’s body image concerns and mood Body Image 2015 13 38 45 10.1016/j.bodyim.2014.12.002 25615425 Ferguson CJ Muñoz ME Garza A Galindo M Concurrent and Prospective Analyses of Peer, Television and Social Media Influences on Body Dissatisfaction, Eating Disorder Symptoms and Life Satisfaction in Adolescent Girls Journal of Youth and Adolescence 2014 43 1 1 14 10.1007/s10964-012-9898-9 23344652 Fichter MM Quadflieg N Hedlund S Twelve-year course and outcome predictors of anorexia nervosa International Journal of Eating Disorders 2006 39 2 87 100 10.1002/eat.20215 16231345 Fineberg N Demetrovics Z Stein D Ioannidis K Potenza M Grünblatt E Chamberlain S Manifesto for a European research network into Problematic Usage of the Internet European Neuropsychopharmacology 2018 10.1016/J.EURONEURO.2018.08.004 Fladung AK Grön G Grammer K Herrnberger B Schilly E Grasteit S Von Wietersheim J A neural signature of anorexia nervosa in the ventral striatal reward system American Journal of Psychiatry 2010 167 2 206 212 10.1176/appi.ajp.2009.09010071 19833790 Frank GKW Roblek T Shott ME Jappe LM Rollin MDH Hagman JO Pryor T Heightened fear of uncertainty in anorexia and bulimia nervosa International Journal of Eating Disorders 2012 45 2 227 232 10.1002/eat.20929 21495057 El Ghoch M Calugi S Grave RD Management of Severe Rhabdomyolysis and Exercise-Associated Hyponatremia in a Female with Anorexia Nervosa and Excessive Compulsive Exercising Case Reports in Medicine 2016 2016 10.1155/2016/8194160 Godier LR Park RJ Compulsivity in anorexia nervosa: A transdiagnostic concept Frontiers in Psychology 2014 5 778 10.3389/fpsyg.2014.00778 25101036 Godier LR Park RJ Does compulsive behavior in Anorexia Nervosa resemble an addiction A qualitative investigation Frontiers in Psychology 2015 OCT 6 10.3389/fpsyg.2015.01608 Griffiths MD Szabo A Terry A The exercise addiction inventory: a quick and easy screening tool for health practitioners British Journal of Sports Medicine 2005 39 6 e30 10.1136/bjsm.2004.017020 15911594 Ho RC Zhang MWB Tsang TY Toh AH Pan F Lu Y Mak KK The association between internet addiction and psychiatric co-morbidity: A meta-analysis BMC Psychiatry 2014 14 1 183 10.1186/1471-244X-14-183 24947851 Hoek HW Incidence, prevalence and mortality of anorexia nervosa and other eating disorders Current Opinion in Psychiatry 2006 July 19 389 394 10.1097/01.yco.0000228759.95237.78 16721169 Hoffman ER Gagne DA Thornton LM Klump KL Brandt H Crawford S Bulik CM Understanding the association of impulsivity, obsessions, and compulsions with binge eating and purging behaviours in anorexia nervosa European Eating Disorders Review 2012 May 20 e129 10.1002/erv.2161 22351620 Hoyle RH Stephenson MT Palmgreen P Lorch EP Donohew RL Reliability and validity of a brief measure of sensation seeking Personality and Individual Differences 2002 32 401 414 www.elsevier.com/locate/paid Hsieh KY Hsiao RC Yang YH Liu TL Yen CF Predictive effects of sex, age, depression, and problematic behaviors on the incidence and remission of internet addiction in college students: A prospective study International Journal of Environmental Research and Public Health 2018 15 12 10.3390/ijerph15122861 Hummel AC Smith AR Ask and you shall receive: Desire and receipt of feedback via Facebook predicts disordered eating concerns International Journal of Eating Disorders 2015 48 4 436 442 10.1002/eat.22336 25060558 Ioannidis K Chamberlain SR Treder MS Kiraly F Leppink EW Redden SA Grant JE Problematic internet use (PIU): Associations with the impulsive-compulsive spectrum. An application of machine learning in psychiatry Journal of Psychiatric Research 2016 83 94 102 10.1016/j.jpsychires.2016.08.010 27580487 Ioannidis K Taylor C Holt L Brown K Lochner C Fineberg NA Czabanowska K Problematic usage of the internet and eating disorders: a multifaceted, systematic review and metaanalysis MedRxiv 2020 10.1101/2020.08.20.20177535 Ioannidis K Treder MS Chamberlain SR Kiraly F Redden SA Stein DJ Grant JE Problematic internet use as an age-related multifaceted problem: Evidence from a two-site survey Addictive Behaviors 2018 81 157 166 10.1016/j.addbeh.2018.02.017 29459201 Keys A Brozek J Henschel A Mickelsen O Taylor HL The Biology of Human Starvation University of Minnesota Press Oxford 1950 Kiddle B Inkster B Prabhu G Moutoussis M Whitaker KJ Bullmore ET Jones PB Cohort Profile: The NSPN 2400 Cohort: a developmental sample supporting the Wellcome Trust NeuroScience in Psychiatry Network International Journal of Epidemiology 2018 47 1 18 19g 10.1093/ije/dyx117 29177462 Kim SK McKay D Taylor S Tolin D Olatunji B Timpano K Abramowitz J The structure of obsessive compulsive symptoms and beliefs: A correspondence and biplot analysis Journal of Anxiety Disorders 2016 38 79 87 10.1016/j.janxdis.2016.01.003 26851748 Levinson CA Fewell L Brosof LC My Fitness Pal calorie tracker usage in the eating disorders Eating Behaviors 2017 27 14 16 10.1016/j.eatbeh.2017.08.003 28843591 Lin SSJ Tsai C-C Sensation seeking and internet dependence of Taiwanese high school adolescents Computers in Human Behavior 2002 18 411 426 Retrieved from www.elsevier.com/locate/comphumbeh Linardon J Messer M My fitness pal usage in men: Associations with eating disorder symptoms and psychosocial impairment Eating Behaviors 2019 33 13 17 10.1016/j.eatbeh.2019.02.003 30772765 Luck AJ Morgan JF Reid F O’Brien A Brunton J Price C Lacey JH The SCOFF questionnaire and clinical interview for eating disorders in general practice: Comparative study British Medical Journal 2002 325 7367 755 756 10.1136/bmj.325.7367.755 12364305 Mabe AG Forney KJ Keel PK Do you “like” my photo Facebook use maintains eating disorder risk International Journal of Eating Disorders 2014 47 5 516 523 10.1002/eat.22254 25035882 Melioli T Rodgers RF Rodrigues M Chabrol H The role of body image in the relationship between internet use and bulimic symptoms: Three theoretical frameworks Cyberpsychology, Behavior, and Social Networking 2015 18 11 682 686 10.1089/cyber.2015.0154 26378881 Mingoia J Hutchinson AD Wilson C Gleaves DH The relationship between social networking site use and the internalization of a thin ideal in females: A meta-analytic review Frontiers in Psychology 2017 AUG 8 10.3389/fpsyg.2017.01351 Parkes L Tiego J Aquino K Braganza L Chamberlain SR Fontenelle L Yücel M Transdiagnostic variations in impulsivity and compulsivity in obsessive-compulsive disorder and gambling disorder correlate with effective connectivity in cortical-striatal-thalamic-cortical circuits NeuroImage 2019 202 116070 10.1016/J.NEUROIMAGE.2019.116070 31382045 Patton JH Stanford MS Barratt ES Factor structure of the Barratt impulsiveness scale Journal of Clinical Psychology 1995 51 6 768 774 8778124 Pawlikowski M Altstötter-Gleich C Brand M Validation and psychometric properties of a short version of Young’s Internet Addiction Test Computers in Human Behavior 2013 29 3 1212 1223 10.1016/j.chb.2012.10.014 Peñas-Lledó E Vaz Leal FJ Waller G Excessive exercise in anorexia nervosa and bulimia nervosa: Relation to eating characteristics and general psychopathology International Journal of Eating Disorders 2002 31 4 370 375 10.1002/eat.10042 11948642 Prichard I Kavanagh E Mulgrew KE Lim MSC Tiggemann M The effect of Instagram #fitspiration images on young women’s mood, body image, and exercise behaviour Body Image 2020 33 1 6 10.1016/j.bodyim.2020.02.002 32062021 Quesnel DA Cook B Murray K Zamudio J Inspiration or Thinspiration: the Association Among Problematic Internet Use, Exercise Dependence, and Eating Disorder Risk International Journal of Mental Health and Addiction 2018 16 5 1113 1124 10.1007/s11469-017-9834-z Rizk M Lalanne C Berthoz S Kern L Godart N Problematic Exercise in Anorexia Nervosa: Testing Potential Risk Factors against Different Definitions PLOS ONE 2015 10 11 e0143352 10.1371/journal.pone.0143352 26618359 Robbins TW Gillan CM Smith DG de Wit S Ersche KD Neurocognitive endophenotypes of impulsivity and compulsivity: Towards dimensional psychiatry Trends in Cognitive Sciences 2012 January 16 81 91 10.1016/j.tics.2011.11.009 22155014 Rosseel Y Lavaan: An R package for structural equation modeling Journal of Statistical Software 2012 48 1 1 36 10.18637/jss.v048.i02 Rossier V Bolognini M Plancherel B Halfon O Sensation seeking: a personality trait characteristic of adolescent girls and young women with eating disorders European Eating Disorders Review 2000 8 3 245 252 10.1002/(SICI)1099-0968(200005)8:3<245::AID-ERV308>3.0.CO;2-D Schermelleh-Engel K Moosbrugger H Müller H Evaluating the Fit of Structural Equation Models: Tests of Significance and Descriptive Goodness-of-Fit Measures Methods of Psychological Research Online 2003 8 Simpson CC Mazzeo SE Calorie counting and fitness tracking technology: Associations with eating disorder symptomatology Eating Behaviors 2017 26 89 92 10.1016/j.eatbeh.2017.02.002 28214452 Slater A Halliwell E Jarman H Gaskin E More than Just Child’s Play: An Experimental Investigation of the Impact of an Appearance-Focused Internet Game on Body Image and Career Aspirations of Young Girls Journal of Youth and Adolescence 2017 46 9 2047 2059 10.1007/s10964-017-0659-7 28316057 Smink FRE Van Hoeken D Hoek HW Epidemiology of eating disorders: Incidence, prevalence and mortality rates Current Psychiatry Reports 2012 14 4 406 414 10.1007/s11920-012-0282-y 22644309 Smith AR Hames JL Joiner TE Status Update: Maladaptive Facebook usage predicts increases in body dissatisfaction and bulimic symptoms Journal of Affective Disorders 2013 149 1-3 235 240 10.1016/j.jad.2013.01.032 23453676 Sternheim L Danner U van Elburg A Harrison A Do anxiety, depression, and intolerance of uncertainty contribute to social problem solving in adult women with anorexia nervosa? Brain and Behavior 2020 10.1002/brb3.1588 Sternheim L Konstantellou A Startup H Schmidt U What does uncertainty mean to women with anorexia nervosa An interpretative phenomenological analysis European Eating Disorders Review 2011 19 1 12 24 10.1002/erv.1029 20669152 Sternheim L Startup H Schmidt U An experimental exploration of behavioral and cognitive-emotional aspects of intolerance of uncertainty in eating disorder patients Journal of Anxiety Disorders 2011 25 6 806 812 10.1016/j.janxdis.2011.03.020 21555203 Strober M Freeman R Morrell W The long-term course of severe anorexia nervosa in adolescents: Survival analysis of recovery, relapse, and outcome predictors over 10-15 years in a prospective study International Journal of Eating Disorders 1997 22 4 339 360 10.1002/(SICI)1098-108X(199712)22:4<339::AID-EAT1>3.0.CO;2-N 9356884 Tao Z The relationship between Internet addiction and bulimia in a sample of Chinese college students: Depression as partial mediator between Internet addiction and bulimia Eating and Weight Disorders 2013 18 3 233 243 10.1007/s40519-013-0025-z 23760906 Tchanturia K Anderluh MB Morris RG Rabe-Hesketh S Collier DA Sanchez P Treasure JL Cognitive flexibility in anorexia nervosa and bulimia nervosa Journal of the International Neuropsychological Society 2004 10 4 513 520 10.1017/S1355617704104086 15327730 Tiggemann M Miller J The internet and adolescent girls’ weight satisfaction and drive for thinness Sex Roles 2010 63 1 79 90 10.1007/s11199-010-9789-z Tiggemann M Slater A NetGirls: The internet, facebook, and body image concern in adolescent girls International Journal of Eating Disorders 2013 46 6 630 633 10.1002/eat.22141 23712456 Tiggemann M Slater A Facebook and body image concern in adolescent girls: A prospective study International Journal of Eating Disorders 2017 50 1 80 83 10.1002/eat.22640 27753130 Tiggemann M Zaccardo M “Exercise to be fit, not skinny”: The effect of fitspiration imagery on women’s body image Body Image 2015 15 61 67 10.1016/j.bodyim.2015.06.003 26176993 Treasure J Zipfel S Micali N Wade T Stice E Claudino A Wentz E Anorexia nervosa Nature Reviews Disease Primers 2015 November 26 1 1 21 10.1038/nrdp.2015.74 Westwater ML Mancini F Gorka AX Shapleske J Serfontein J Grillon C Fletcher PC Prefrontal responses during proactive and reactive inhibition are differentially impacted by stress in anorexia and bulimia nervosa BioRxiv Preprint 2019 10.1101/2020.02.27.968719 Whiteside SP Lynam DR The five factor model and impulsivity: Using a structural model of personality to understand impulsivity Personality and Individual Differences 2001 30 4 669 689 10.1016/S0191-8869(00)00064-7 Wickham H François R Henry L Müller K RStudio A Grammar of Data Manipulation [R package dplyr version 085] 2020 Retrieved April 24, 2020 https://cran.r-project.org/package=dplyr Zipfel S Giel KE Bulik CM Hay P Schmidt U Anorexia nervosa: Aetiology, assessment, and treatment The Lancet Psychiatry 2015 2 1099 1111 10.1016/S2215-0366(15)00356-9 26514083

PMC007xxxxxx/PMC7614801.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7911385 Psychiatry Res Psychiatry Res Psychiatry research 0165-1781 1872-7123 37163882 7614801 10.1016/j.psychres.2023.115245 EMS177468 Article Suicidal Ideation and Attempts in Trichotillomania Grant Jon E. a* Collins Madison a Chesivoir Eve a Chamberlain Samuel R. b a Department of Psychiatry & Behavioral Neuroscience, University of Chicago, Pritzker School of Medicine, Chicago, IL, USA b Department of Psychiatry, Faculty of Medicine, University of Southampton, UK; and Southern Health NHS Foundation Trust, Southampton, UK * Address correspondence to: Jon E. Grant, JD, MD, MPH, Professor, Department of Psychiatry & Behavioral Neuroscience, University of Chicago, 5841 S. Maryland Avenue, MC 3077, Chicago, IL 60637, Phone: 773-834-1325; Fax: 773-834-6761; jongrant@uchicago.edu 01 7 2023 06 5 2023 05 7 2023 25 7 2023 325 115245115245 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Trichotillomania is characterized by chronic pulling out of one’s hair. Given the negative sequelae of trichotillomania, we examined rates of suicidal ideation and suicide attempts. Of the 219 adults (mean age = 29.5 years; 88% female) recruited, 40 (18.3%) reported lifetime suicidal ideation, and 5 (2.3%) reported a lifetime suicide attempt. Those with histories of suicidal ideation were significantly more likely to have major depressive disorder. Our findings suggest that suicidal ideation and attempts are common in trichotillomania and support the idea that comorbid depression should be considered a risk factor for suicidality. trichotillomania suicidality comorbidity pmc1 Introduction Trichotillomania is characterized by the failure to resist impulses to pull out one’s hair often resulting in noticeable hair loss (American Psychiatric Association, APA, 2013). Trichotillomania is classified as an obsessive-compulsive and related disorder in the Diagnostic and Statistical Manual Version 5 (DSM-5) (APA, 2013). The condition is frequently associated with poor self-esteem, depression, psychosocial dysfunction, and overall lower quality of life (Woods et al., 2006). More specifically, approximately 45% of people with trichotillomania struggle with comorbid major depressive disorder (Grant et al., 2020). Given the negative consequences associated with trichotillomania, and the elevated rates of co-occurring major depressive disorder, we would expect that rates of suicidal ideation may also be elevated in this population, and yet we have found no data on rates of suicidal ideation or suicide attempts among people with trichotillomania. To address this unmet need, the present study assessed suicidal ideation and suicide attempts in a large sample of adults with trichotillomania using a secondary analysis of pooled research studies. We hypothesized that both suicidal ideation and suicide attempts would be elevated in individuals with trichotillomania compared to normative data from the US general population. 2 Methods 2.1 Participants Adults with trichotillomania who had participated in a range of research studies were included. Inclusion criteria for all studies were: current DSM-5 trichotillomania, written informed consent, and the ability to understand the study and the consent form. Exclusion criteria were: bipolar I disorder, schizophrenia, or an alcohol/substance use disorder in the preceding three months. Data from baseline visits were used for the current study. The sample was recruited from metropolitan areas in the USA. After receiving a complete description of the study, participants provided written informed consent. All procedures involving human subjects were approved by the Institutional Review Board at the University of Chicago. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. 2.2 Assessments Participants were assessed for age, biological sex at birth, self-reported gender, and racial-ethnic identity. Suicidality was assessed using the Columbia Suicide Severity Rating Scale (C-SSRS) (Posner et al., 2011), a valid and reliable semi-structured clinical interview consisting of subscales assessing current and lifetime suicidal ideation and suicidal behavior. Other assessments included: Massachusetts General Hospital Hairpulling Scale (MGH-HPS), a 7-item self-report scale (total scores range from 0 to 28) assessing hair pulling over the past seven days (Keuthen et al., 1995); Clinical Global Impression-Severity scale (CGI) (Guy, 1976) to assess overall reported symptoms as well as observations of hair loss; Mini International Neuropsychiatric Interview (MINI; Sheehan et al., 1998) or the Structured Clinical Interview for DSM-IV (First et al., 1995) to examine co-occurring psychiatric disorder; Quality of Life Inventory (QOLI) (Frisch et al., 2005), a self-report scale assessing satisfaction in sixteen domains of life; and the Barratt Impulsiveness Scale 11 (BIS-11) (Patton et al., 1995). 2.3 Statistics We present descriptive characteristics of the sample including percentage of those with lifetime suicidal ideation and lifetime suicide attempts. Furthermore, the sample was divided into two subgroups according to the lifetime presence or absence of suicidal ideation: trichotillomania with lifetime suicidal ideation (score ≥ 1 on the suicidal severity subscale) compared to trichotillomania without any lifetime suicidal ideation (score = 0). To analyze the sociodemographic and clinical features associated with lifetime suicidal ideation, groups were compared using independent sample t-tests for continuous variables and Pearson's chi-square tests for categorical variables. Statistical significance was defined as P < .05, Bonferroni corrected for the number of measures at the level of type of measurement. 3 Results The study comprised 219 adults with trichotillomania (mean age = 29.5 [SD=7.63) years [range 18 to 64 yrs]; 88.0%) female). Of 219 participants, number and percentages of people in different racial-ethnic categories were: Caucasian 174 (79.5%), Black 23 (10.5%), Asian 13 (5.9%), Latino/Hispanic 5 (2.3%),), and Other/Mixed Race 4 (1.8%). Of the 219 participants, 40 (18.3%) reported lifetime suicidal ideation, and 5 (2.3%) reported a lifetime suicide attempt. When we compared those with lifetime suicidal ideation to those without lifetime suicidal ideation, we found no significant differences with respect to age (t(211) = 0.699, p = 0.49), gender (X2(2) = 0.60, p = 0.74), or race/ethnicity (X2(6) = 12.12, p = 0.06). In terms of comorbidity, those with lifetime suicidal ideation were significantly more likely to have co-occurring major depressive disorder (62.5% compared to 34.7%; χ2(1) = 10.48; p<.001) (see Table 1). The groups did not significantly differ with respect to other psychiatric comorbidities (all p>0.10). The participants with lifetime suicidal ideation did not significantly differ from those without suicidal ideation with respect to trichotillomania symptom severity (MGH-HPS scores of 18.3 [SD = 3.52] compared to 18.4 [SD = 3.53], respectively; t(209) = .14, p=.891), overall global mental health severity (CGI scores of 4.4 [SD = 0.59] compared to 4.4 [SD = 0.69] respectively; t(209) = .49, p=.627), quality of life (QOLI t-scores of 31.5 [SD = 25.1] compared to 42.91 [SD = 14.0] respectively; t(81) = 1.8 p=.075), or impulsivity (BIS total scores of 62.7 [SD = 12.3] compared to 61.4 [SD = 12.1] respectively; t(181) = -.64, p=.526). 4 Discussion This study is the first that we are aware of that examined suicidal ideation and attempts in adults with trichotillomania. We found that the rate of lifetime suicidal ideation was almost one in five (18.3%) and that approximately 2% had lived through one or more suicide attempt(s) in their lifetimes. To put these data into a larger context, a study exploring suicidal behaviors across 17 countries using data from the World Health Organization World Mental Health Survey Initiative, found a cross-national lifetime prevalence of 9.2% for suicidal ideation and 2.7% for suicide attempts (Nock et al., 2008). In a National Comorbidity Survey (1990-1992) in the USA, using a nationally representative sample, 13.5% of individuals reported lifetime suicidal ideation and 4.6% reported a lifetime attempt (Kessler et al., 1999). These data for trichotillomania are interesting when also compared to data regarding suicidal ideation and attempts in other obsessive-compulsive spectrum disorders. A meta-analysis of suicidal ideation in obsessive-compulsive disorder (OCD) found that at least one person out of ten with OCD experiences one or more suicide attempt(s) during their lifetime (pooled prevalence rate 13%), while nearly half of individuals with OCD have experienced suicidal ideation (pooled prevalence rate of lifetime suicidal ideation 47%) (Pelligrini et al 2020). An earlier meta-analysis found that comorbid disorders, increased severity of obsessions, feelings of hopelessness and past history of suicide attempts were associated with worsening levels of suicidality in OCD (Angelakis et al., 2015). Another study of suicide attempts in adults with OCD found that 14.6% of the sample (n=425) reported at least one suicide attempt during their lifetime (Dell’Osso et al., 2018). Finally, a recent meta-analysis of OCD and related disorders found that for body dysmorphic disorder, the pooled prevalence of lifetime suicide attempts and lifetime suicidal ideation were, respectively, 35.2% and 66.1%, whereas for grooming disorders (these studies only included skin picking), the pooled prevalence of lifetime suicide attempts and current suicidal ideation were 13.3% and 40.4%, respectively (no data was available for lifetime suicidal ideation) (Pelligrini et al., 2021). Based on our data, trichotillomania appears to have a lower rate of both lifetime suicidal ideation and suicide attempts compared to certain other obsessive-compulsive spectrum disorders. One possible explanation for these differences could be that the symptoms of trichotillomania, while impairing, are less likely to result in suicidal ideation or attempts as compared to the symptoms of OCD or BDD. Given the reported low quality of life and depression, however, it will be important to determine what protective factors against suicidality people with trichotillomania possess that others in the OCD spectrum may not have. Suicidal ideation in trichotillomania was significantly associated with having comorbid major depressive disorder. Having said that, it is also important to note that approximately 40% of those with suicidal ideation did not have comorbid major depressive disorder. It is likely that those with greater general psychopathology who have both depression and trichotillomania are also more likely to have suicidal ideation. It is not possible to unentangle the complex relationship between suicidality, depression and hair pulling based on our limited data. Some of these participants might have been depressed and suicidal due to their pulling, others may have been depressed and suicidal independent of their pulling, and others may have had suicidal ideation independent of pulling and of depressive symptoms. Another way of interpreting these results, however, is not that depression is a risk factor but instead suicide is an independent problem and thus a shared outcome of a subset of trichotillomania patients and a subset of some depressed patients. Regardless, these data do suggest that everyone with trichotillomania should be screened for major depressive disorder and suicidal ideation. There are important limitations to this study. First, the sample for the present study was recruited from a diverse group of research studies and the recruitment method may have varied slightly. Second, the findings of this study may not be generalizable to people with trichotillomania in the general community, or to specific clinical practice situations, since all participants were part of various research studies. Third, we only used the Columbia Suicide Rating Scale to examine suicidality, and did not include a measure of disability, and therefore future research may consider adding other measures of interest. Fourth, we did not ask what if any relationship hair pulling had to suicidal ideation. Finally, the study enrolled 80% Caucasians and therefore minority racial-ethnic groups were underrepresented, relative to the general population, which may affect the generalizability of the results. Prior work indicates that suicidality in general maybe higher in particular minority racial-ethnic groups, notably in relation to indigenous populations (Troya et al., 2022), highlighting the need to study this issue in future trichotillomania research. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Table 1 Psychiatric Co-Morbidities by Lifetime Suicidal Ideation a Lifetime Suicidal Ideation (N = 40) No Lifetime Suicidal Ideation (N = 173) MDD 25 (62.5%) *** 60 (34.7%) *** OCD 1 (2.5%) 11 (6.4%) ADHD 2 (5%) 12 (6.9%) Any Anxiety Disorder 7 (17.5%) 45 (26.0%) Any Eating Disorder 2 (5%) 3 (1.7%) Any SUD 0 (0%) 9 (5.2%) Note. A Bonferroni correction was applied to account for multiple comparisons, leaving the critical p-value at 0.007 (0.05/7). a Data are presented in N (%) form unless otherwise indicated. *** p < .001 Competing Interests: Dr. Grant has received research grants from Janssen, Boehringer Ingelheim, and Biohaven Pharmaceuticals. He receives yearly compensation from Springer Publishing for acting as Editor-in-Chief of the Journal of Gambling Studies and has received royalties from Oxford University Press, American Psychiatric Publishing, Inc., Norton Press, and McGraw Hill. Dr. Chamberlain receives a stipend from Elsevier for editorial work. Ms. Chesivoir and Ms. Collins report no conflicts. American Psychiatric Association Diagnostic and statistical manual of mental disorders 5th ed Washington, DC 2013 Angelakis I Gooding P Tarrier N Panagioti M Suicidality in obsessive compulsive disorder (OCD): a systematic review and meta-analysis Clin Psychol Rev 2015 39 1 15 10.1016/j.cpr.2015.03.002 Epub 2015 Mar 25 25875222 Dell’Osso B Benatti B Arici C Palazzo C Altamura AC Hollander E Fineberg N Stein DJ Nicolini H Lanzagorta N Marazziti D Prevalence of suicide attempt and clinical characteristics of suicide attempters with obsessive-compulsive disorder: a report from the International College of Obsessive-Compulsive Spectrum Disorders (ICOCS) CNS Spectr 2018 23 1 59 66 10.1017/S1092852917000177 Epub 2017 Mar 16 28300008 First MB Spitzer RL Gibbon M Williams JBW Structured Clinical Interview for DSM-IV-Patient Edition (SCID-I/P, Version 20) Biometrics Research Department, New York State Psychiatric Institute New York 1995 Frisch MB Clark MP Rouse SV Rudd MD Paweleck JK Greenstone A Kopplin DA Predictive and treatment validity of life satisfaction and the quality of life inventory Assessment 2005 12 1 66 78 10.1177/1073191104268006 15695744 Grant JE Dougherty DD Chamberlain SR Prevalence, gender correlates, and comorbidity of trichotillomania Psychiatry Res 2020 288 112948 10.1016/j.psychres.2020.112948 Epub 2020 Apr 18 32334275 Kessler RC Borges G Walters EE Prevalence of and risk factors for lifetime suicide attempts in the National Comorbidity Survey Arch Gen Psychiatry 1999 56 7 617 26 10.1001/archpsyc.56.7.617 10401507 Keuthen NJ O’Sullivan RL Ricciardi JN Shera D Savage CR Borgmann AS Jenike MA Baer L The Massachusetts General Hospital (MGH) Hairpulling Scale: 1. development and factor analyses Psychother Psychosom 1995 64 3–4 141 5 10.1159/000289003 8657844 Nock MK Borges G Bromet EJ Alonso J Angermeyer M Beautrais A Bruffaerts R Chiu WT de Girolamo G Gluzman S de Graaf R Cross-national prevalence and risk factors for suicidal ideation, plans and attempts Br J Psychiatry 2008 192 2 98 105 10.1192/bjp.bp.107.040113 18245022 Patton JH Stanford MS Barratt ES Factor structure of the Barratt Impulsiveness Scale J Clin Psychol 1995 51 6 768 774 10.1002/1097-4679(199511)51:6<768::aid-jclp2270510607>3.0.co;2-1 8778124 Pellegrini L Maietti E Rucci P Casadei G Maina G Fineberg NA Albert U Suicide attempts and suicidal ideation in patients with obsessive-compulsive disorder: A systematic review and meta-analysis J Affect Disord 2020 276 1001 1021 10.1016/j.jad.2020.07.115 Epub 2020 Jul 22 32750613 Pellegrini L Maietti E Rucci P Burato S Menchetti M Berardi D Maina G Fineberg NA Albert U Suicidality in patients with obsessive-compulsive and related disorders (OCRDs): A meta-analysis Compr Psychiatry 2021 108 152246 10.1016/j.comppsych.2021.152246 Epub 2021 May 19 34062378 Posner K Brown GK Stanley B Brent DA Yershova KV Oquendo MA Currier GW Melvin GA Greenhill L Shen S Mann JJ The Columbia-Suicide Severity Rating Scale: initial validity and internal consistency findings from three multisite studies with adolescents and adults Am J Psychiatry 2011 168 12 1266 77 10.1176/appi.ajp.2011.10111704 22193671 Sheehan DV Lecrubier Y Sheehan KH Amorim P Janavs J Weiller E Hergueta T Baker R Dunbar GC The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10 J Clin Psychiatry 1998 59 Suppl 20 22 33 9881538 Troya MI Spittal MJ Pendrous R Crowley G Gorton HC Russell K Byrne S Musgrove R Hannah-Swain S Kapur N Knipe D Suicide rates amongst individuals from ethnic minority backgrounds: A systematic review and meta-analysis EClinicalMedicine 2022 Apr 28 47 101399 10.1016/j.eclinm.2022.101399 35518122 Woods DW Flessner CA Franklin ME Keuthen NJ Goodwin RD Stein DJ Walther MR Trichotillomania Learning Center-Scientific Advisory Board The Trichotillomania Impact Project (TIP): exploring phenomenology, functional impairment, and treatment utilization J Clin Psychiatry 2006 67 12 1877 88 10.4088/jcp.v67n1207 17194265

PMC007xxxxxx/PMC7614802.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0111052 Aust N Z J Psychiatry Aust N Z J Psychiatry The Australian and New Zealand journal of psychiatry 0004-8674 1440-1614 34903086 7614802 10.1177/00048674211066004 EMS177462 Article Natural Recovery in Trichotillomania Grant Jon E. a Chamberlain Samuel R. b a Department of Psychiatry & Behavioral Neuroscience, University of Chicago, Chicago, IL, USA b Department of Psychiatry, University of Southampton, UK * Address correspondence to: Jon E. Grant, JD, MD, MPH, Professor, Department of Psychiatry & Behavioral Neuroscience, University of Chicago, Pritzker School of Medicine, 5841 S. Maryland Avenue, MC 3077, Chicago, IL 60637, Phone: 773-834-1325; Fax: 773-834-6761; jongrant@uchicago.edu 01 10 2022 14 12 2021 05 7 2023 25 7 2023 56 10 13571362 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Objectives Trichotillomania is characterized by repetitive pulling out of one’s hair, leading to distress and/or functional impairment. Long considered a chronic condition if left untreated (albeit with fluctuating intensity), there have been intimations that the disorder may be of limited duration in some people. Methods A sample of 10,169 adults, aged 18-69 years, representative of the general US population, were recruited and screened for current and lifetime trichotillomania. Potential differences in demographic and clinical variables and lifetime comorbidities, between those with natural recovery from trichotillomania, and those with current trichotillomania, were identified using analysis of variance or Likelihood-Ratio chi-square tests as appropriate. Additional analyses using binary logistic regression were used to control for potential confounding differences between the groups initially identified. Results 24.9% of the entire sample of people with lifetime trichotillomania reported that they no longer had symptoms of trichotillomania and had never received therapy or medication treatment for it (i.e. they experienced natural recovery). Those who experienced natural recovery did not differ from those with current trichotillomania in terms of demographic or clinical characteristics, except that they were currently older. Natural recovery was associated with significantly lower rates of related comorbidities: obsessive-compulsive disorder (OCD), attention-deficit hyperactivity disorder (ADHD), panic disorder, skin picking disorder, and tic disorder. Discussion These findings from the first epidemiology study examining natural recovery in trichotillomania highlight the importance of screening for and treating such comorbidities in patients with trichotillomania, in order to maximize chance of clinical recovery. trichotillomania natural recovery outcomes comorbidity pmcIntroduction Trichotillomania (hair-pulling disorder) is characterized by recurrent pulling out of one’s own hair, leading to hair loss and oftentimes functional impairment (American Psychiatric Association, 2013). Although examined in the medical literature for decades (Grant and Chamberlain, 2016), the course of trichotillomania has been the subject of much debate. Long considered a chronic condition if left untreated (albeit with fluctuating intensity) (Christenson et al., 1991a, 1991b), there have been intimations that the disorder may be of limited duration in some people. For example, in the case of childhood-onset trichotillomania (i.e. those whose hair pulling begins before the age of 5 years), many simply stop pulling over time (Swedo and Leonard, 1992). Other research suggests that trichotillomania may become chronic in people who have pulled for more than 6 months (Chang et al., 1992). Research on the clinical characteristics, neurobiology, and treatment of trichotillomania has expanded over the last thirty years (Grant and Chamberlain, 2016), but research on the natural course of the disorder and any putative predictors of subsequent recovery is scant. Such research is needed in light of findings indicating that trichotillomania results in a heavy psychosocial burden of suffering (Grant et al., 2016; Franklin et al., 2008; Houghton et al., 2016; Tung et al., 2015). One area of trichotillomania research in need of further elaboration is that of natural recovery (i.e. having met diagnostic criteria for the disorder in the past but not meeting any diagnostic criteria for the disorder for the past 12 months and attaining this achievement without any formal psychological or pharmacological interventions). Our definition is consistent with the one proposed by Slutske (2006), although others have defined the term as achieving remission for the short term (i.e. 2-3 months) from a disorder but still possibly having some mild symptoms (Mekonen et al., 2021). When we look to other mental health disorders, we find some limited understanding of this phenomenon. In the case of major depressive disorder, a recent metaanalysis found a pooled estimate of 12.5% of people with untreated depression achieved remission within 12 weeks (Mekonen et al., 2021). In the area of obsessive compulsive disorder, a disorder with some phenomenological similarities to trichotillomania, we find that rates of remission before there were evidence-based treatment (i.e. a type of natural recovery) range from 20-24% (Pollitt, 1957; Skoog and Skoog, 1999) (keeping in mind that the diagnosis of “obsessional neurosis from many years ago may not track completely with the current DSM diagnosis of obsessive compulsive disorder). In the case of substance use disorders and gambling disorder, both of which may have some relationship to trichotillomania if it is conceptualized as a behavioral addiction, research suggests that many of those who recover are able to do so without any formal treatment. The rate of natural recovery among individuals with alcohol use disorders has ranged from 24.4% to 78% (Bischof et al., 2005; Dawson, 1996; Dawson et al., 2005; Sobell et al., 1996) (the range may be due to remission being defined by either abstinence or moderate drinking without meeting criteria for an alcohol use disorder) (Mellor et al., 2019). In the related area of gambling addiction, two large national U.S. surveys found that 33%–36% of people with gambling disorder experienced natural recovery (Slutske, 2006). Another study found that around 75% of young adults with subsyndromal disordered gambling at baseline no longer had such symptoms at 1-year follow up (Grant et al., 2014). Some predictors of natural recovery in the alcohol field include being female, older age, and married; however, severity of alcoholism was negatively correlated with chance of natural recovery (Dawson et al., 2005). In the area of gambling disorder, being female appears to be associated with a greater likelihood of natural recovery (Slutske et al., 2009). Natural recovery from sub-syndromal disordered gambling has been associated with lower amounts spent gambling at baseline, and with older age, albeit within the context of young adults followed up for 1-year (Grant et al., 2014). There is some indication in the obsessive compulsive disorder literature that older age of disorder onset was associated with greater likelihood of natural recovery (Skoog and Skoog, 1999). Based on this background, the aims of the present study were to document the rates of natural recovery among individuals with trichotillomania and to determine clinical and demographic variables associated with natural recovery. Material And Methods Participants and Procedures Data were collected in the context of market research for a client exploring a potential new treatment for trichotillomania. These market research data were then made available in anonymized form to the current researchers, without restriction. The current paper therefore comprises secondary analyses of de-identified data and was thus exempted from Institutional Review Board (IRB) procedures under current US guidelines. All participants had provided informed consent and had agreed that their data could be shared in anonymized form with external researchers. A convenience sample of approximately 10,000 individuals representative of the general US population, 18-69 years of age, were screened for trichotillomania. Survey respondents were recruited from the Schlesinger Group, an ESOMAR (European Society for Opinion and Marketing Research) member that adheres to a globally recognized code of conduct, the jointly developed ICC/ESOMAR Code, for the purposes of such marketing research. Survey respondents were recruited using the “General Population” panel, a well-known provider of panels for online surveys. Quotas were used to obtain a sample that was age and gender matched to the general US Population. Quality control procedures included: double opt-in; confirmation using photo ID validation (manual) at time of registration for panel; relevant ID and a programming (CAPTCHA) at registration to deter bots; a Red Herring survey to catch people outside of US, hidden questions in registration to catch bots, database checks to identify batches of similar email structure entering panel in short time period, profile checks to identify unlikely combinations of or too many combinations of ailments, and profile checks to identify selection of aberrant choices at different questions at registration and over time on the panel. Participants received 300 points for participation, which had a monetary value of $3.00. The total duration of the survey was approximately 15 minutes. Assessments Each individual completed a self-administered survey via the Internet, which comprised two segments. Part 1: Screening for prevalence: demographics and diagnosis of trichotillomania and comorbidities; and Part 2: Survey of people with current trichotillomania: survey of diagnosis, severity, and life impact. Part 1 of the survey asked about multiple psychiatric disorders with one question (“Please indicate whether you currently have or have ever had any of the following medical conditions”). Trichotillomania (hair-pulling disorder) was one of the listed conditions. All participants indicated either: “Never”, “In the past, but not currently”, or “Currently”. The general question was then followed by specific questions regarding who diagnosed the condition, age of onset, and treatment history. In addition, Part 2 of the survey asked about each of the following diagnostic criteria for trichotillomania: “Repeated pulling of my hair causing hair loss; repeated attempts to stop or decrease the hair pulling; the hair pulling is/was causing me personal distress or causing difficulty in areas of my life; realizing that the hair pulling, or hair loss was not related to some other medical problem or a skin condition; and the hair pulling was not done to try to improve my appearance or what I saw as a flaw.” It also asked each person if they had ever received psychotherapy or prescription medication for trichotillomania; and how severe their symptoms were ‘at their worst’ (mild, moderate, or severe). Only if the person answered affirmatively to hair-pulling/trichotillomania in the list of medical conditions were they then prompted to answer Part 2. “Natural recovery” was defined as an individual reporting they had experienced trichotillomania in the past but not currently; and that they had not received therapy or prescription medication for trichotillomania. The survey data were collected in January, 2019. The advertisements about the survey were designed to be neutrally worded; i.e. did not mention the purpose of the survey, no mention of “health”, nor of any diagnosis, in order to reduce participation bias. Data Analysis Potential differences in (1) demographic and clinical variables and (2) lifetime comorbidities, between those with natural recovery from trichotillomania, and those with current trichotillomania, were identified using analysis of variance (ANOVA) or Likelihood-Ratio (LR) chi-square tests as appropriate. Additional analyses using binary logistic regression were used to control for potential confounding differences between the groups initially identified. Statistical analyses were conducted using JMP Pro Software. Significance was defined as p<0.05. Results The total sample size was 10,169 adults, and this sample mirrored closely key demographic characteristics of the US population (Grant et al., 2020). In total, of 253 participants with lifetime trichotillomania, 78 participants (30.8%) reported that trichotillomania was a past but not current problem. Of the 78 participants who reported lifetime but not current trichotillomania, 63 said they stopped pulling their hair without treatment and 15 reported no longer pulling due to treatment. Therefore, the rate of natural recovery in the entire sample of people with lifetime trichotillomania was 24.9% (63/253). Demographic and clinical characteristics for those with natural recovery from trichotillomania (n=63), versus those with current trichotillomania (n=175), are shown in Table 1. Those who reported natural recovery were significantly more likely to be older. The two groups did not differ significantly on other demographic variables or on severity of trichotillomania symptoms. For those participants who reported natural recovery, their hair pulling stopped on average after 10.0 years (median 6 years). The rates of co-occurring lifetime disorders for the two groups are presented in Table 2. After controlling for age, those who reported natural recovery were significantly less likely to report histories of ADHD, OCD, panic disorder, skin picking disorder, and tic disorder. These findings were also significant without controlling for age. Lower rates of certain other disorders (anxiety disorder, bipolar disorder, eating disorder) were observed in the initial analyses, in those with natural recovery versus current trichotillomania, but these results were no longer significant once age was controlled for. Discussion This study, the first to examine natural recovery in TTM using an overall sample that was epidemiologically representative of the US general population, found that among adults with a history of the disorder, 24.9% reported that their trichotillomania remitted without any formal psychological or pharmacological treatment interventions. Interestingly, this rate is not dissimilar to natural recovery for obsessive-compulsive disorder, or substance or gambling addiction, as reported in much of that literature (Pollitt, 1957; Skoog and Skoog, 1999; Bischof et al., 2005; Sobell et al., 1996; Dawson, 1996; Dawson et al., 2005; Slutske, 2006). The rate found here is however higher than reported in depression (Mekonen et al., 2021). The finding that roughly one-fourth of adults with a history of trichotillomania recover from their problems suggests that trichotillomania does not always follow a chronic or persisting course, and that different individuals experience a very different course. In addition, the findings strongly suggest that the lack of comorbidity may have the strongest influence in natural recovery as those who experienced it generally had lower rates of several disorders. The strongest predictor of natural recovery, of the variables examined, was the lack of OCD comorbidity. What actually explains the influence of comorbidity in natural recovery? There are several possible, and non-mutually exclusive, explanations. First, it is possible that trichotillomania with OCD, for example, is different neurobiologically from pure trichotillomania and has a different course of illness. Second, the effects of related comorbidities, such as OCD, may simply make it more difficult for trichotillomania symptoms to improve spontaneously as OCD symptoms (or other traits associated with OCD, such as perfectionism or cognitive rigidity) may reinforce the trichotillomania symptoms (e.g., symmetry obsessions drive pulling hair to even it out). The finding of a fairly high rate of natural recovery needs to be understood in the context of fairly low rates of seeking treatment for trichotillomania. In fact, one study found that less than 20% of individuals with trichotillomania received psychotherapy for their pulling, despite its status as the first-line treatment approach (Woods et al., 2006). Another study found that only 40% of people with trichotillomania received any mental health care specific to trichotillomania (Cohen et al., 1995). This fairly low rate of treatment-seeking is likely due to personal factors (e.g. embarrassment) as well as external barriers (e.g., lack of knowledgeable clinicians) to obtaining help (Woods et al., 2006). The low rate of seeking treatment for trichotillomania is likely due to multiple reasons but one possible explanation, derived from the current study, is that perhaps some people are able to employ strategies themselves that are effective. Thus, evidence from natural recovery could potentially inform formal approaches for treating trichotillomania. For example, one obvious question is whether those who have overcome their trichotillomania on their own did so by completely abstaining from any pulling or whether they were able to continue to pull to some degree without problems. Most of what is known regarding trichotillomania has been garnered from observations in clinical settings when patients were receiving treatment. The results of this study suggest that clinically recruited samples are probably not ideal for some research purposes because findings may not generalize to people in the wider community with the condition. Studies involving people who have recovered from trichotillomania may inform and expand our understanding of the disorder. It is important to note that the demographic characteristics of the current sample of people with trichotillomania are likely to differ in important ways those of prior research studies. Much prior research into the potential prevalence of trichotillomania used college student convenience samples, which by definition would be younger and with a narrower age range than the current dataset, and would be less representative of the general population than those in the current study. Furthermore, clinical trial studies have tended to include relatively high proportions of women compared to men, and would have restricted participation based on extensive inclusion/exclusion criteria; whereas the current study had roughly equal proportions of male and female individuals with trichotillomania, and did not have such restrictions on study participation. Thus, it is important to consider that natural remission may be different to the rate reported here in particular subgroups of patients, such as those taking part in clinical trials. This study has several positive features, notably that it is the first large study of trichotillomania examining rates and features of natural recovery. Several limitations, however, should be considered. First, the study was a survey and as such no direct in person interviews were performed. The gold standard for diagnosis is of course clinical interview by a healthcare professional. Due to the bespoke convenience nature of the survey, it did not use gold-standard rating tools but rather pragmatic questions about whether trichotillomania had been diagnosed; and also about whether the different diagnostic criteria were met, from the individual’s perspective. Second, the survey used a non-probability sample. Although the sample demographics paralleled national demographics, it still raises the possibility of selection biases based on personality factors, etc. Third, data on comorbidities are also per participant report and as such may have over- or under- reported specific conditions. Fourth, natural recovery has no agreed upon definition in the field of trichotillomania. Is it enough not to meet DSM diagnostic criteria of should complete abstinence of pulling be the standard? Is a person recovered after one year? Or is more time needed? In summary, this study examined the rate of natural recovery in trichotillomania in a large representative sample of adults in the USA. Overall, natural recovery in trichotillomania was relatively common (24.9%), and was associated with lower rates of comorbidities of several related disorders, especially (but not exclusively) OCD. These results provide new information for people with trichotillomania in terms of natural recovery rates, and also have clinical implications, serving to highlight the importance of screening for these often over-looked related comorbidities in people with trichotillomania. We hope future studies may examine this important issue and include variables not examined here such as depression and anxiety symptoms, self-esteem, and psychosocial functioning to mention only a few that would deepen our understanding of who may experience natural recovery from trichotillomania. Funding and Disclosures of competing interests The researchers’ time for this study was funded by internal funds. The survey data were collected by Promentis Pharmaceuticals, Inc., and were made available for unrestricted use by the study authors. The authors received no funding from Promentis Pharmaceuticals, Inc., for this study. Promentis Pharmaceuticals, Inc., has had no influence on the analyses of data or the writing of this manuscript. Dr. Grant has received research grants from TLC Foundation, and Otsuka Pharmaceuticals. Dr. Grant receives yearly compensation from Springer Publishing for acting as Editor-in-Chief of the Journal of Gambling Studies and has received royalties from Oxford University Press, American Psychiatric Publishing, Inc., Norton Press, and McGraw Hill. Dr. Chamberlain’s time on this study was supported in part by a Wellcome Trust Clinical Fellowship (110049/Z/15/Z). Dr. Chamberlain receives a stipend for his work as Associate Editor at Neuroscience and Biobehavioral Reviews; and at Comprehensive Psychiatry. Table 1 Comparison of key demographic and cinical features of the two trichotillomania groups. Current Trichotillomania (N=175) Naturally Remitted Trichotillomania (N=63) F p Age, years 36.0 (12.8) 42.4 (16.3) 10.0223 0.002 Gender, female, N [%] 83 [47.4%] 31 [49.2%] 5.159 LR 0.3067 Racial-ethnic group, White Caucasian, N [%] 127 [72.6%] 47 [74.6%] 0.098 LR 0.7542 Age at onset, years 17.7 (10.8) 15.9 (10.5) 1.0409 0.3089 Reported severity of trichotillomania (at worst), N [%] 2.481 LR 0.2892    Mild 57 [32.6%] 27 [42.9%]    Moderate 106 [60.6%] 31 [49.2%]    Severe 12 [6.9%] 5 [7.9%] Household income, N [%] 0.892 LR 0.9708    <$25,000 32 [18.3%] 12 [19.1%]    $25,000 - $50,000 48 [27.4%] 14 [22.2%]    $50,001 - $75,000 36 [20.6%] 14 [22.2%[    $75,001 - $125,000 30 [17.1%] 13 [20.6%]    >$125,000 23 [13.1%] 8 [12.7%]    Prefer not to answer 6 [3.4%] 2 [3.2%] Current relationship status, N [%] 8.065 LR 0.1527    Single, not in relationship 60 [34.3%] 26 [41.3%]    Single, in a relationship 26 [14.9%] 11 [17.5%]    Cohabiting with partner 13 [7.4%] 4 [6.4%]    Domestic partnership 3 [1.7%] 2 [3.2%]    Married 72 [41.1%] 17 [27.0%]    Separated 1 [0.6%] 3 [4.8%] Education level, N [%] 5.560 LR 0.1351    High school or less 37 [21.1%] 8 [12.7%]    At least some college education or above 138 [78.9%] 55 [87.3%] LR = Likelihood Ratio chi-square test. Table 2 Lifetime history of other mental disorders in the trichotillomania study groups. Current Trichotillomania (N=175) Naturally Remitted Trichotillomania (N=63) LR chi-square p p value for regression model, controlling for age ADHD 60 (34.29%) 9 (14.29%) 9.894 0.0017 0.0116 Alcohol or drug abuse 67 (38.29%) 17 (26.98%) 2.666 0.1025 Anxiety Disorder 134 (76.57%) 40 (63.49%) 3.88 0.0489 0.1258 Bipolar Disorder 46 (26.29%) 9 (14.29%) 4.039 0.0445 0.1309 Depression 127 (72.57%) 43 (68.25%) 0.418 0.5181 Eating Disorder 61 (34.86%) 13 (20.63%) 4.604 0.0319 0.1267 OCD 76 (43.43%) 12 (19.05%) 12.678 <0.001 0.0023 Panic Disorder 71 (40.57%) 14 (22.22%) 7.152 0.0075 0.0268 PTSD 64 (36.57%) 16 (25.4%) 2.676 0.1019 Skin picking disorder 60 (34.29%) 11 (17.46%) 6.719 0.001 0.0315 Tic disorder 28 (16%) 2 (3.17%) 8.693 0.0032 0.0362 LR = Likelihood Ratio chi-square test; ADHD = attention deficit hyperactivity disorder; OCD = obsessive-compulsive disorder; PTSD = post-traumatic stress disorder American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders 5th ed American Psychiatric Association Washington, DC 2013 Bischof G Rumpf HJ Meyer C Hapke U John U Influence of psychiatric comorbidity in alcoholdependent subjects in a representative population survey on treatment utilization and natural recovery Addiction 2005 100 3 405 413 15733254 Chang CH Lee MB Chiang YC Lu LC Trichotillomania:A clinical study of 36 patients Journal of the Formosan Medical Assocication 1992 90 176 180 Christenson GA Pyle RL Mitchell JE Estimated lifetime prevalence of trichotillomania in college students Journal of Clinical Psychiatry 1991a 52 415 417 1938977 Christenson GA Mackenzie TB Mitchell JE Characteristics of 60 adult chronic hair pullers American Journal of Psychiatry 1991b 148 365 70 1992841 Cohen LJ Stein DJ Simeon D Spadaccini E Rosen J Aronowitz B Hollander E Clinical profile, comorbidity, and treatment history in 123 hair pullers: a survey study Journal of Clinical Psychiatry 1995 56 7 319 326 7615485 Dawson DA Correlates of past-year status among treated and untreated persons with former alcohol dependence: United States, 1992 Alcohol Clinical Experimental Research 1996 20 4 771 779 Dawson DA Grant BF Stinson FS Chou PS Huang B Ruan WJ Recovery from DSM-IV alcohol dependence: United States, 2001-2002 Addiction 2005 100 3 281 292 15733237 Franklin ME Flessner CA Woods DW Keuthen NJ Piacentini JC Moore P Stein DJ Cohen SB Wilson MA Trichotillomania Learning left-Scientific Advisory Board The child and adolescent trichotillomania impact project: descriptive psychopathology, comorbidity, functional impairment, and treatment utilization Journal of Developmental and Behavioral Pediatric 2008 29 6 493 500 Grant JE Chamberlain SR Trichotillomania American Journal of Psychiatry 2016 173 868 874 27581696 Grant JE Redden SA Leppink EW Odlaug BL Chamberlain SR Psychosocial dysfunction associated with skin picking disorder and trichotillomania Psychiatry Research 2016 239 68 71 27137963 Grant JE Derbyshire K Leppink E Chamberlain SR One-year follow-up of subsyndromal gambling disorder in non-treatment-seeking young adults Annals of Clinical Psychiatry 2014 26 3 199 205 25166482 Grant JE Dougherty DD Chamberlain SR Prevalence, gender correlates, and co-morbidity of trichotillomania Psychiatry Research 2020 288 112948 32334275 Houghton DC Maas J Twohig MP Saunders SM Compton SN Neal-Barnett AM Franklin ME Woods DW Comorbidity and quality of life in adults with hair-pulling disorder Psychiatry Research 2016 239 12 19 27137957 Lochner C Keuthen NJ Curley EE Tung ES Redden SA Ricketts EJ Bauer CC Woods DW Grant JE Stein DJ Comorbidity in trichotillomania (hair-pulling disorder): A cluster analytical approach Brain Behavior 2019 9 12 e01456 31692297 Mekonen T Ford S Chan GCK Hides L Connor JP Leung J What is the short-term remission rate for people with untreated depression? A systematic review and meta-analysis Journal of Affective Disorders 2021 296 17 25 34583099 Mellor R Lancaster K Ritter A Systematic review of untreated remission from alcohol problems: Estimation lies in the eye of the beholder Journal of Substance Abuse Treatment 2019 102 60 72 31202290 Pollitt J Natural history of obsessional states; a study of 150 cases British Medical Journal 1957 1 5012 194 8 13383228 Skoog G Skoog I A 40-year follow-up of patients with obsessive-compulsive disorder [see commetns] Archives of General Psychiatry 1999 56 2 121 127 10025435 Slutske WS Natural recovery and treatment-seeking in pathological gambling: results of two U.S. national surveys American Journal of Psychiatry 2006 163 2 297 302 16449485 Slutske WS Blaszczynski A Martin NG Sex differences in the rates of recovery, treatment-seeking, and natural recovery in pathological gambling: results from an Australian community-based twin survey Twin Research and Human Genetics 2009 12 5 425 32 19803770 Sobell LC Cunningham JA Sobell MB Recovery from alcohol problems with and without treatment: prevalence in two population surveys American Journal of Public Health 1996 86 7 966 972 8669520 Swedo SE Leonard HL Trichotillomania. An obsessive compulsive spectrum disorder? Psychiatric Clinics of North America 1992 15 4 777 790 1461795 Tung ES Flessner CA Grant JE Keuthen NJ Predictors of life disability in trichotillomania Comprehensive Psychiatry 2015 56 239 244 25281991 Woods DW Flessner CA Franklin ME Keuthen NJ Goodwin RD Stein DJ Walther MR Trichotillomania Learning left-Scientific Advisory Board The Trichotillomania Impact Project (TIP): exploring phenomenology, functional impairment, and treatment utilization Journal of Clinical Psychiatry 2006 67 12 1877 1888 17194265

PMC007xxxxxx/PMC7614803.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101300405 Brain Imaging Behav Brain Imaging Behav Brain imaging and behavior 1931-7557 1931-7565 34410609 7614803 10.1007/s11682-021-00533-5 EMS178724 Article Reward Processing in Trichotillomania and Skin Picking Disorder Grant Jon E. M.D., J.D., M.P.H a* Peris Tara S. Ph.D b Ricketts Emily J. Ph.D b Bethlehem Richard A.I. Ph.D c Chamberlain Samuel R. Ph.D d O’Neill Joseph PhD b Scharf Jeremiah M. M.D., Ph.D e Dougherty Darin D. M.D e Deckersbach Thilo e Woods Douglas W. Ph.D f Piacentini John Ph.D b** Keuthen Nancy J. Ph.D e** a Department of Psychiatry & Behavioral Neuroscience University of Chicago, Chicago, IL, USA b Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles, CA, USA c Department of Psychiatry, University of Cambridge, UK d Department of Psychiatry, Faculty of Medicine, University of Southampton, UK; and Southern Health NHS Foundation Trust, UK e Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, USA f Department of Psychology, Marquette University, Milwaukee, WI, USA * Address correspondence to: Jon E. Grant, JD, MD, MPH Professor, Department of Psychiatry & Behavioral Neuroscience University of Chicago, 5841 S. Maryland Avenue, MC 3077, Chicago, IL 60637 Phone: 773-834-1325; Fax: 773-834-6761; jongrant@uchicago.edu ** Co-Senior Authors 01 4 2022 19 8 2021 10 7 2023 25 7 2023 16 2 547556 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Background Trichotillomania (hair pulling disorder) and skin picking disorder are common and often debilitating mental health conditions, grouped under the umbrella term of body focused repetitive behaviors (BFRBs). Although the pathophysiology of BFRBs is incompletely understood, reward processing dysfunction has been implicated in the etiology and sustention of these disorders. The purpose of this study was to probe reward processing in BFRBs. Methods 159 adults (125 with a BFRB [83.2% (n=104) female] and 34 healthy controls [73.5% (n=25) female]) were recruited from the community for a multi-center between-group comparison using a functional imaging (fMRI) monetary reward task. Differences in brain activation during reward anticipation and punishment anticipation were compared between BFRB patients and controls, with stringent correction for multiple comparisons. All group level analyses controlled for age, sex and scanning site. Results Compared to controls, BFRB participants showed marked hyperactivation of the bilateral inferior frontal gyrus (pars opercularis and pars triangularis) compared to controls. In addition, BFRB participants exhibited increased activation in multiple areas during the anticipation of loss (right fusiform gyrus, parahippocampal gyrus, cerebellum, right inferior parietal lobule; left inferior frontal gyrus). There were no significant differences in the win-lose contrast between the two groups. Conclusions These data indicate the existence of dysregulated reward circuitry in BFRBs. The identified pathophysiology of reward dysfunction may be useful to tailor future treatments. trichotillomania skin picking disorder imaging reward fMRI neurobiology pmcIntroduction Trichotillomania (TTM; also known as Hair Pulling Disorder) and Skin Picking Disorder (SPD), are characterized by repeated pulling out of hair resulting in hair loss or picking at skin resulting in tissue damage, respectively. These disorders have been conceptualized under the larger umbrella of body focused repetitive behavior disorders (BFRBs) and often result in significant psychosocial impairment (Tucker et al., 2011). Although both psychosocial and psychopharmacological treatments have demonstrated some degree of efficacy for BFRBs, many people fail to respond or exhibit only partial responses to these interventions (Sani et al., 2019). One issue that has hampered treatment development to date, and the formulation of brain-based disease models, is that the understanding of the neurobiology of BFRBs remains limited. BFRB neuroimaging research comparing patients to matched controls has identified, in small, single-site samples (Ns ranging from 10-76) of participants, evidence of abnormalities in several regions of the brain which have included areas involved in habit formation (e.g., dorsal striatal areas), emotional regulation (e.g., amygdala and hippocampal areas), memory processing (e.g., temporal lobe), self-monitoring and awareness (e.g., precuneus), reward processing (e.g., ventral striatum, frontal hemisphere, bilateral cuneus), visual processing of disgust (e.g., insula and putamen), and generation and suppression of motor responses (e.g., inferior frontal gyrus) (Swedo et al., 1991; Grachev 1997; O’Sullivan et al., 1997; Stein et al., 1997; Rauch et al., 2007; Keuthen et al., 2007; Chamberlain et al., 2008, 2010; 2018; Lee et al., 2010; grant et al., 2013; Roos et al., 2015; Odlaug et al., 2016; Chienle et al., 2018; Isobe et al., 2018; Wabnegger and Schienle 2019). With such small samples and conflicting findings, if treatments are to target pathophysiology of BFRBs there remains a substantive need for further work aimed at understanding more precisely the neurobiological underpinnings of BFRBs. One potential promising area for exploration is reward processing, based on several convergent lines of thinking. The pulling and picking of BFRBs is often described as pleasurable (Arzeno et al., 2006) and people will report urges to engage in the behavior that mirror those described by people with substance use problems; and that undertaking the behavior leads to transient reduction or relief of the urge, though this is short-lived (Grant et al., 2007). Additionally, early data suggested that people with BFRBs were more likely than controls to have first-degree relatives with substance addictions, disorders typically characterized by reward processing abnormalities (Schlosser et al., 1994). Although limited in number, available double-blind placebo-controlled treatment trials that have shown some benefit for BFRBs have used pharmacological agents modulating glutamate and dopamine (Deepmala et al., 2015), both of which seem integral to reward processing. In a previous study examining reward circuitry activation, White and colleagues (2013) examined 13 adults with TTM (unmedicated and compared to 12 controls) using fMRI with a monetary incentive delay (MID) task. In a region-of-interest analysis, TTM patients exhibited decreased nucleus accumbens (NAcc) activation for gain anticipation and increased NAcc activation for gain and loss outcomes, versus controls. However, these findings were not significant at the whole brain level. At the level of the whole brain, loss anticipation showed less activation of the left putamen and insula in TTM than controls. Given this background, there is some suggestion that TTM and SPD may represent problems of disordered reward processing, but further testing of this hypothesis via neuroimaging is needed. Therefore, a greater understanding of reward processing should allow for improved prevention and treatment strategies. Thus, the objective of this study was to examine reward circuitry activation in a large multi-site sample of patients with BFRBs compared to controls. Methods Participants included 159 adults recruited from the community as having either a BFRB (trichotillomania, skin picking disorder, or both as their primary psychiatric problem) or being a healthy control. Three sites were involved in recruitment: University of Chicago, University of California, Los Angeles, and Massachusetts General Hospital/Harvard Medical School. Inclusion criteria for the clinical sample was: a) DSM-5 diagnosis of trichotillomania and/or skin picking disorder as the primary psychiatric conditions; b) aged 18 to 65; c) fluency in English; and d) capable of providing informed consent/assent. Inclusion criteria for the healthy controls were the same except they could have no current or lifetime history of any DSM-5 psychiatric disorder. BFRB participants could have comorbid psychiatric disorders (based on the Mini International Neuropsychiatric Interview 7.0; (Sheehan et al., 1998) as long as TTM or SPD were the primary psychiatric condition. Exclusion criteria for the clinical sample and healthy controls were: (a) current or lifetime diagnosis of any serious medical or psychiatric illness (including substance use disorder) that would preclude successful study participation; (b) neurological conditions that would preclude completion of neurocognitive tasks; (c) use of psychotropic medications unless the dose had been stable for at least the past 3 months; (d) body metal other than dental fillings (assessed using a neuroimaging screening form); (e) positive pregnancy test; and (f) medical condition or other factor contraindicating neuroimaging. Procedures Potential participants were screened by the study site coordinator. Prior to obtaining written informed consent, the investigators provided a complete description of the study, discussed potential risks, and answered questions regarding the study. After that, participants provided written informed consent. Participants received up to $200 for participation as reimbursement for their time. All participants underwent a diagnostic interview and were asked to complete an “MR Screening Form” to rule out any conditions that preclude MR scanning. MRI Neuroimaging We used a multi-site neuroimaging design involving equal numbers of participants across three sites: (1) in the Massachusetts General Hospital Martinos Center for Biomedical Imaging, (2) the Staglin Center for Cognitive Neuroscience at the UCLA Semel Institute for Neuroscience and Human Behavior, and (3) the Department of Psychiatry and Behavioral Neuroscience at the University of Chicago. As described above, participants were screened for scanner compatibility at the outset, and we scanned eligible participants sequentially. Imaging was performed on a 3-Tesla MRI scanner at all three sites with all scanners synchronized. Each MRI scanning session lasted no more than 75 minutes. Task order was pseudo-randomized and counterbalanced across subjects. All tasks were presented using Eprime software. Participants were instructed in the behavioral tasks that they would engage in while in the fMRI scanner. The participants were given headphones, ear protectors, and a head restraining device was used to reduce excess motion. We first acquired high resolution, anatomical images, typically about 15 minutes, before undertaking the fMRI sequence. fMRI Task: Monetary Reward Task The monetary reward task was used to examine reward processing (Knutson et al., 2001). The version of the task deployed enabled separate analysis of neural activation for anticipated monetary reward, and for anticipated monetary loss, on the task. The task is comprised of 72 trials, and each trial lasts approximately 6 seconds (range 3–10 sec). A cue was presented for 500 milliseconds and signaled a potentially rewarding (a circle) or non-rewarding (a square) trial. Participants were instructed that the game’s goal is to see how much money they can win. If they saw a circle with one line through it, it meant they could win $1. If they saw a square with one line through it, it meant they could lose $1 if they did not press the button at the correct time. If they saw a circle with two lines through it, they could win $5. If they saw a square with two lines through it, they could lose $5. Participants were given feedback (reward or no reward). Each participant underwent 10 practice trials, and based on these practice trials, each individual had a time limit to respond to the target during the task (based on their shortest reaction time during the practice sessions). By increasing and decreasing the time limit, participants were rewarded in 50% of the reward trials. The design of the task allows for the modeling of response to anticipation of reward or loss to be independent from that of actual reward or loss. Image acquisition Imaging was acquired across three scanning sites (University of Chicago, University of California, Los Angeles, and Massachusetts General Hospital/Harvard Medical School) using a unified acquisition protocol. Structural scans were acquired on a 2 Siemens Magnetom Prismafit 3T scanners (UCLA and MGH/Harvard) and one Philips Achieva 3T MRI scanner with dStream (Chicago) all with 32-channel head coils, using a MPRAGE acquisition sequence with the following parameters: slab orientation = sagittal, FOV 256x256x176, voxel size 1x1x1 mm3, inversion delay time TI = 900 ms, TR = 2310 ms, TE = 2.9 ms flip angle = 9 degree. Functional task MRI was acquired with a single-shot gradient echo planar imaging sequence. Thirty-nine interleaved axial slices parallel to the AC–PC line covering the whole brain were acquired: TR = 2000 m s, TE = 28 ms, flip angle = 90°, field of view = 210 × 210 mm, matrix = 205 × 205, voxels size 3.2x3.2x3.1mm and a 20% distance factor, GRAPPA acceleration factor = 2 (for the 2 Siemens scanners) and SENSE acceleration factor 2 (for the Philips scanner). Image processing First level fMRI data processing was performed using FEAT (FMRI Expert Analysis Tool) Version 6.00 (FMRIB’s Software Library, www.fmrib.ox.ac.uk/fsl). Registration to high resolution structural and/or standard space images was done using FLIRT (Jenkinson et al., 2002). Registration from high resolution structural to standard space was refined using FNIRT nonlinear registration (Andersson et al., 2007). The following pre-statistics processing was applied; motion correction using MCFLIRT (Jenkinson et al., 2002); slice-timing correction using Fourier-space time-series phase-shifting; non-brain removal using BET (Smith 2002); spatial smoothing using a Gaussian kernel of FWHM 5mm; grand-mean intensity normalization of the entire 4D dataset by a single multiplicative factor; highpass temporal filtering (Gaussian-weighted least-squares straight line fitting, with sigma=50.0s). Time-series statistical analysis was done using FILM with local autocorrelation correction fitting contrasts for anticipation in win and lose trials as well as the contrast between them. All first level outputs were manually inspected for potential registration errors. Contrasts from first level analysis were entered into higher level analyses carried out using FEAT (FMRI Expert Analysis Tool) Version 6.00 (FMRIB’s Software Library, www.fmrib.ox.ac.uk/fsl). Z (Gaussianised T/F) statistic images were thresholded non-parametrically using clusters determined by Z>2.3 corresponding to a p-value of 0.01 and a (corrected) cluster significance threshold of P=0.05. All group level analyses included age, sex and scanning site as potential confounding factors. Rendered thresholded z-maps were subsequently visualized using the glass-brain plotting tool and cope estimates using ggstatsplot. Following analysis of main effects, we extracted cope estimates for post-hoc analysis within condition and picker and puller sub-groups from within main effect thresholded regions of interest using Featquery to extract cope estimates for peak coordinates. All post-hoc analysis of estimates were conducted using non-parametric Kruskal Wallis one-way analysis of variance for main effect and a Dwass-Steel-Crichtlow-Fligner test for pairwise comparisons. All pairwise comparisons were corrected for multiple comparisons using Benjamini and Hochberg corrections (Benjamini and Hochberg 1995). For visualization we also included individual cope estimates of main effects as assessed using the Bayesian analog of a students t-test. Results Sample Characteristics The sample included 159 participants (125 adults with a BFRB and 34 adults as healthy controls), of which 49 had TTM, 51 had SPD, and 25 had both TTM and SPD. Of the 125 adults with BFRBs, 83.2% (n=104) were female, the mean age was 29.1 (9.2), and the mean age of BFRB onset was 13.1 (6.4) years. Of the 34 healthy controls, 73.5% (n=25) were female and mean age was 26.8 (7.6) years. Groups did not differ in terms of gender or age as assessed using a Wilcoxon signed rank test. Behavior BFRB participants did not significantly differ from controls in reaction times [Win anticipation; 391 vs 377 msec T(68.81)=1.40, p =.17. Loss anticipation; 389 vs 375 msec T(71.50)=1.57, p =.27]. Reaction times did not differ between TTM, SPD, and comorbid participants [Win anticipation; F(3,78.11)=1.64, p=.19. Loss anticipation; F(3,76.83)=1.04, p=.38]. Neuroimaging Compared to controls, BFRB participants showed an increased activation in two significant clusters during the anticipation of reward (see Table 1 and Figure 1). Cluster 1 was maximal at left inferior frontal gyrus (BA 44 & 45). Cluster 2 was maximal at right inferior frontal gyrus (BA 44 & 45). BFRB participants showed an increased activation in four significant clusters during the anticipation of loss (see Table 2 and Figure 2). Cluster 1 was maximal at right fusiform gyrus (BA 37), extending into inferior temporal gyrus, parahippocampal gyrus, and cerebellum. Cluster 2 was maximal at right inferior parietal lobule (BA 39), extending into angular gyrus and middle occipital gyrus. Cluster 3 was maximal at left inferior frontal gyrus (BA 44 & 45), extending into precentral gyrus and middle frontal gyrus. Cluster 4 was maximal at right inferior frontal gyrus (BA 44 & 45). The BFRB and control groups did not differ significantly in the win-lose contrast. Additionally, activation at the peak cluster coordinates did not differ significantly between patients who did versus did not have other mental health comorbidities (all p>0.15 uncorrected, Wilcoxon tests). After recognizing that BFRB participants differed from controls, we examined the groups within the BFRBs (TTM, SPD, and co-occurring TTM + SPD). Post-hoc analyses within condition are presented in Figure 3. Although the 3 clinical groups did not differ from each other on activation during anticipation of wins, all three clinical groups differed significantly from controls in Cluster 1 during anticipation of wins. In terms of loss anticipation, all three clinical groups exhibited significantly greater activation in Clusters 3 and 4 compared to controls. In addition, participants with SPD exhibited significantly greater activation in Cluster 1 compared to controls. Finally, participants with SPD and those with SPD+TTM showed significantly greater activation in Cluster 2 compared to controls. Discussion To our knowledge, this is the largest neuroimaging study of BFRBs to explore reward-related task activation. Our findings present evidence for dysregulated reward circuitry in BFRBs. Reward seeking and loss/harm avoidance play important roles in human behavior, and when there is dysfunction in reward processing, maladaptive behaviors may result. The DSM-5 diagnostic criteria for both TTM and SPD highlight failed attempts to reduce the behavior – which is suggestive of reward dysfunction. Here, the main finding was of marked hyperactivation of the bilateral inferior frontal gyrus (pars opercularis and pars triangularis) in people with BFRBs compared to controls, when anticipating reward or punishment. These results were significant with stringent statistical correction, including when controlling for potential confounders. The various versions of the monetary incentive task have been useful in understanding the neuropathology underpinning reward processing. Though earlier work with this task focused on the nucleus accumbens (NAcc) using a region-of-interest approach, is now widely established that reward and loss anticipation involve activation of distributed neural circuitry. One neuroimaging meta-analysis (using 20 studies) showed that healthy volunteers activate the nucleus accumbens, thalamus, insula, and medial frontal gyrus during reward processing (Knutson and Greer 2008). A larger meta-analysis of 142 studies found that healthy volunteers activated the nucleus accumbens, insula, inferior frontal gyrus, anterior cingulate cortex, and medial orbito-frontal cortex during reward processing (Liu et al., 2011). In terms of anticipation of rewards, we found strong evidence of hyperactivation in the inferior frontal gyrus (IFG) (as well as right fusiform gyrus, inferior temporal gyrus, parahippocampal gyrus, right inferior parietal lobule, and middle frontal gyrus) in all three BFRB groups compared to controls and these findings suggest that people with BFRBs are biologically hypersensitive to potential rewards. The salient role of the IFG is of particular interest given that above-mentioned meta-analysis found IFG activation (and the other regions) is common in reward processing tasks including during reward anticipation (Liu et al., 2011). Additionally, a small study of cognitive flexibility in TTM (n=12) found right frontal hyperactivation using fMRI (Grant et al., 2018) and excess IFG thickness was found in a recent meta-analysis in TTM (n=76 patients versus n=41 controls) (Wagnegger and Schienle 2019), which also seems to extend to first-degree relatives in a small sample (Odlaug et al., 2014). These findings collectively point to abnormalities of IFG as a potential core feature of BFRB, and now extends those previous TTM findings to SPD as well. In addition to its role in reward and punishment anticipation noted above, the IFG is involved in the detection of environmentally salient cues (Hampshire et al., 2010), and in the suppression of habitual response patterns (Aron et al., 2014). Response inhibition deficits have been found in individuals with right IFG damage, and the severity of the response inhibition deficit has been positively associated with the degree of right IFG damage (Aron et al., 2003). Additionally, research has demonstrated that when the right IFG is disrupted by transcranial magnetic stimulation, response inhibition is also impaired (Chambers et al., 2007). The finding of abnormal IFG function in BFRBs may help to account for neuropsychological findings with regards to these disorders. In particular, significant inhibitory control deficits have been reported in most but not all studies of BFRBs versus healthy controls. One interpretation of the current data alongside the neurocognitive literature is that the IFG may be over-activated by anticipation of reward or punishment in patients with BFRBs, in turn impeding the ability of this region to successfully undertake other cognitive functions such as top-down behavioral inhibition of automated behavior. This hypothesis could be tested in future work by also examining the monetary incentive task and stop-signal fMRI task performance in relation not only to brain activation but also connectivity metrics. If this hypothesis is correct, treatment of BFRBs using medications or psychotherapies that dampen the sub-cortical reward pathways may then in turn enable the IFG to exert more top-down control by freeing up this region’s processing capacity. Our results confirm an association between the anticipation of monetary reward and frontal hyperactivation. Having said that, the lack of a difference in striatal activation between BFRBs and controls merits further consideration. Unlike a recent fMRI study that found some evidence for abnormal accumbens activation in TTM patients during a reward task (White et al., 2013), our study found that activation of the accumbens did not differ between BFRB participants and controls. The task did activate this region across all subjects, indicating this null result was not simply due to the task failing to activate this region. The study by White and colleagues included only 13 subjects, all of whom had TTM, and thus differs from the current study in multiple ways. First, the accumbens finding in the previous study was not significant in the whole-brain analysis and so may reflect a false positive with respect to the region-of-interest analysis. Second, the role of the striatum in BFRBs is not well understood, with some previous structural studies finding that neither TTM nor SPD subjects differed significantly from controls in terms of dorsal and ventral striatum volumes (Odlaug et al., 2014) and others finding significant differences (Roos et al., 2015). A recent meta-analysis of all available literature did not find sub-cortical structural differences in TTM versus controls, including specifically in the NAcc (Chamberlain et al., 2018). Third, tasks that recruit both executive and reward networks may simply exhibit greater dyfunction in top-control elements of reward compared to bottom-up drive. Findings of no differences in NAcc activation during anticipated monetary reward seem to differ from those found in OCD (reduced NAcc activation to anticipated monetary reward in OCD (Figee et al., 2011) or those found in substance addictions (relatively increased NAcc response to rewarding outcome in cocaine addicted adults (Jia et al., 2011) and alcoholics (Bjork et al., 2008) and might suggest that BFRBs have distinct neurobiological substrates from these other conditions. Lastly, it should be noted that due to its small size, the NAcc can be considered difficult to visualize and measure in terms of activation, which could hinder ability to detect subtle differences in activation. Several limitations should be considered in relation to the current study. The study was neither designed nor powered to evaluate possible effects of previous treatment on brain activation, nor the contribution of specific types of comorbidities. Overall though, brain findings did not differ as a function of whether patients did or did not have mental health comorbidities. Also, although the total size of the study was fairly large, the number of participants with individual BFRBs may have been too small to detect differences between those with TTM, SPD or the comorbid condition. Finally, the current research was undertaken in a cohort that was largely female and of white racial-ethnic type and thus may not be representative of the larger population of people with BFRBs. In summary, this large multi-center fMRI study suggests that TTM and SPD are associated with disordered reward processing, linked to the inferior frontal gyrus. Ideally, understanding reward processing dysfunction should allow for improved treatment strategies via neuromodulation, pharmacotherapy or psychosocial interventions. Funding This study was funded by the Body-Focused Precision Medicine Initiative granted by The TLC Foundation for Body-Focused Repetitive Behaviors to University of Chicago (Dr. Grant), Massachusetts General Hospital (MGH)/Harvard (Dr. Keuthen), and University of California, Los Angeles (UCLA) (Dr. Piacentini). Dr. Chamberlain’s involvement in this study was funded by a Wellcome Trust Clinical Fellowship (refs. 110049/Z/15/Z & 110049/Z/15/A). The TLC Foundation for Body-Focused Repetitive Behaviors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. This study was completed with support from the REDCap project at the University of Chicago, which is hosted and managed by the Center for Research Informatics and funded by the Biological Sciences Division and by the Institute for Translational Medicine, CTSA grant number UL1 TR000430 from the National Institutes of Health. Availability of data and material Data available upon request Code availability NA Figure 1 Anticipation of reward during win trials Figure 2 Anticipation of reward during loss trials Figure 3 Win and loss anticipation across groups Table 1 Cluster sizes and p values, along with peak coordinates for anticipation of reward hyperactivation in the BFRB group Cluster Index Voxels P-corrected Z-MAX Z-MAX X (mm) Z-MAX Y (mm) Z-MAX Z (mm) 2 879 0.00102 4.14 44 10 30 1 481 0.048 4.19 -38 10 22 Table 2 Cluster sizes and p values, along with peak coordinates for anticipation of loss hyperactivation in the BFRB group. Cluster Index Voxels P-corrected Z-MAX Z-MAX X (mm) Z-MAX Y (mm) Z-MAX Z (mm) 4 816 0.00162 4.22 46 12 28 3 729 0.00365 4.08 -36 12 24 2 537 0.0248 3.98 38 -60 50 1 506 0.0343 4.16 30 -60 -10 Author Contributions: Jon E. Grant, Tara Peris, Emily Rickets, Samuel R. Chamberlain, Jeremiah Scharf, Darin Dougherty, Doug Woods, John Piacentini, Nancy J. Keuthen all made substantial contributions to the conception or design of the work as well as the acquisition, analysis, or interpretation of data; they all aided in drafting the work, gave final approval of the version to be published; and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Compliance with Ethical Standards: The Institutional Review Boards for the University of Chicago, University of California, Los Angeles, and Massachusetts General Hospital/Harvard Medical School approved the study and the informed consent. Data sharing agreements were arranged across all sites and neuroimaging equipment was synced across the sites. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Conflict of Interests: Dr. Grant has received research grants from Biohaven, Promentis, and Otsuka Pharmaceuticals. Dr. Grant receives yearly compensation from Springer Publishing for acting as Editor-in-Chief of the Journal of Gambling Studies and has received royalties from Oxford University Press, American Psychiatric Publishing, Inc., Norton Press, and McGraw Hill. Dr. Chamberlain consults for Promentis; and receives stipends from Elsevier for editorial journal work. The other authors report no conflicts. Dr. Dougherty receives research support and honoraria from Medtronic, Inc. Dr. Woods has received royalties from Oxford University Press and Springer Press. Dr. Piacentini has received research grants from NIMH, the Tourette Association of America, and Pfizer. He receives travel support and honoraria from the Tourette Association of America and the International OCD Foundation and book royalties from Guilford Publications and Oxford University Press. Dr Keuthen has received prior research grants from The TLC Foundation for Body-Focused Repetitive Behaviors and royalties from New Harbinger, Inc. The remaining authors have nothing to disclose. Consent to participate: Prior to obtaining written informed consent, the investigators provided a complete description of the study, discussed potential risks, and answered questions regarding the study. After that, participants provided written informed consent Consent for publication NA Andersson JLR Jenkinson M Smith S Non-linear optimisation; FMRIB Technial Report TR07JA1 2007 Aron AR Fletcher PC Bullmore ET Sahakian BJ Robbins TW Stop-signal inhibition disrupted by damage to right inferior frontal gyrus in humans Nat Neurosci 2003 6 2 115 6 Erratumin:NatNeurosci 2003 Dec 6 12 1329 10.1038/nn1003 12536210 Aron AR Robbins TW Poldrack RA Inhibition and the right inferior frontal cortex: one decade on Trends Cogn Sci 2014 18 4 177 85 10.1016/j.tics.2013.12.003 24440116 Arzeno Ferrao Y Almeida VP Bedin NR Rosa R D’Arrigo Busnello E Impulsivity and compulsivity in patients with trichotillomania or skin picking compared with patients with obsessive-compulsive disorder Compr Psychiatry 2006 47 4 282 8 10.1016/j.comppsych.2005.11.005 16769303 Benjamini Y Hochberg Y Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing J R Stat Soc B Stat Methodol 1995 57 289 300 Bjork JM Momenan R Smith AR Hommer DW Reduced posterior mesofrontal cortex activation by risky rewards in substance-dependent patients Drug Alcohol Depend 2008 95 1–2 1158 10.1016/j.drugalcdep.2007.12.014 Chamberlain SR Menzies LA Fineberg NA Del Campo N Suckling J Craig K Grey matter abnormalities in trichotillomania: morphometric magnetic resonance imaging study Br J Psychiatry 2008 193 3 216 221 10.1192/bjp.bp.107.048314 18757980 Chamberlain SR Hampshire A Menzies LA Garyfallidis E Grant JE Odlaug BL Reduced brain white matter integrity in trichotillomania: a diffusion tensor imaging study Arch Gen Psychiatry 2010 67 9 965 971 10.1001/archgenpsychiatry.2010.109 20819990 Chamberlain SR Harries M Redden SA Keuthen NJ Stein DJ Lochner C Cortical thickness abnormalities in trichotillomania: international multi-site analysis Brain Imaging Behav 2018 12 3 823 828 10.1007/s11682-017-9746-3 28664230 Chambers CD Bellgrove MA Gould IC English T Garavan H McNaught E Dissociable mechanisms of cognitive control in prefrontal and premotor cortex J Neurophysiol 2007 98 6 3638 47 10.1152/jn.00685.2007 17942624 Deepmala Slattery J Kumar N Delhey L Berk M Dean O Clinical trials of N-acetylcysteine in psychiatry and neurology: A systematic review Neurosci Biobehav Rev 2015 55 294 321 10.1016/j.neubiorev.2015.04.015 25957927 Figee M Vink M de Geus F Vulink N Veltman DJ Westenberg H Dysfunctional reward circuitry in obsessive-compulsive disorder Biol Psychiatry 2011 69 9 867 74 10.1016/j.biopsych.2010.12.003 21272861 Grachev ID MRI-based morphometric topographic parcellation of human neocortex in trichotillomania Psychiatry Clin Neurosci 1997 51 5 315 321 10.1111/j.1440-1819.1997.tb03205.x 9413880 Grant JE Odlaug BL Potenza MN Addicted to hair pulling How an alternate model of trichotillomania may improve treatment outcome Harv Rev Psychiatry 2007 15 2 80 5 10.1080/10673220701298407 17454177 Grant JE Odlaug BL Hampshire A Schreiber LR Chamberlain SR White matter abnormalities in skin picking disorder: a diffusion tensor imaging study Neuropsychopharmacology 2013 38 5 763 769 10.1038/npp.2012.241 23303052 Grant JE Daws R Hampshire A Chamberlain SR An fMRI Pilot Study of Cognitive Flexibility in Trichotillomania J Neuropsychiatry Clin Neurosci 2018 30 4 318 324 10.1176/appi.neuropsych.18030038 30141727 Hampshire A Chamberlain SR Monti MM Duncan J Owen AM The role of the right inferior frontal gyrus: inhibition and attentional control Neuroimage 2010 50 3 1313 9 10.1016/j.neuroimage.2009.12.109 20056157 Isobe M Redden SA Keuthen NJ Stein DJ Lochner C Grant JE Striatal abnormalities in trichotillomania: a multi-site MRI analysis Neuroimage Clin 2018 17 893 898 10.1016/j.nicl.2017.12.031 29515968 Jenkinson M Bannister P Brady M Smith S Improved optimization for the robust and accurate linear registration and motion correction of brain images Neuroimage 2002 17 825 841 10.1016/s1053-8119(02)91132-8 12377157 Jia Z Worhunsky PD Carroll KM Rounsaville BJ Stevens MC Pearlson GD An initial study of neural responses to monetary incentives as related to treatment outcome in cocaine dependence Biol Psychiatry 2011 70 6 553 60 10.1016/j.biopsych.2011.05.008 21704307 Keuthen NJ Makris N Schlerf JE Martis B Savage CR McMullin K Evidence for reduced cerebellar volumes in trichotillomania Biol Psychiatry 2007 61 3 374 381 10.1016/j.biopsych.2006.06.013 16945351 Knutson B Fong GW Adams CM Varner JL Hommer D Dissociation of reward anticipation and outcome with event-related fMRI Neuroreport 2001 12 17 3683 7 10.1097/00001756-200112040-00016 11726774 Knutson B Greer SM Anticipatory affect: neural correlates and consequences for choice Philos Trans R Soc Lond B Biol Sci 2008 363 1511 3771 86 10.1098/rstb.2008.0155 18829428 Lee JA Kim CK Jahng GH Hwang LK Cho YW Kim YJ A pilot study of brain activation in children with trichotillomania during a visual-tactile symptom provocation task: a functional magnetic resonance imaging study Prog Neuropsychopharmacol Biol Psychiatry 2010 34 7 1250 1258 10.1016/j.pnpbp.2010.06.031 20637819 Liu X Hairston J Schrier M Fan J Common and distinct networks underlying reward valence and processing stages: a meta-analysis of functional neuroimaging studies Neurosci Biobehav Rev 2011 35 5 1219 36 10.1016/j.neubiorev.2010.12.012 21185861 Odlaug BL Chamberlain SR Derbyshire KL Leppink EW Grant JG Impaired response inhibition and excess cortical thickness as candidate endophenotypes for trichotillomania J Psychiatr Res 2014 59 167 73 10.1016/j.jpsychires.2014.08.010 25223951 Odlaug BL Hampshire A Chamberlain SR Grant JE Abnormal brain activation in excoriation (skin-picking) disorder: evidence from an executive planning fMRI study Br J Psychiatry 2016 208 2 168 174 10.1192/bjp.bp.114.155192 26159604 O’Sullivan RL Rauch SL Breiter HC Grachev ID Baer L Kennedy DN Reduced basal ganglia volumes in trichotillomania measured via morphometric magnetic resonance imaging Biol Psychiatry 1997 42 1 39 45 10.1016/S0006-3223(96)00297-1 9193740 Rauch SL Wright CI Savage CR Martis B McMullin KG Wedig MM Brain activation during implicit sequence learning in individuals with trichotillomania Psychiatry Res 2007 154 3 233 240 10.1016/j.pscychresns.2006.09.002 17321724 Roos A Grant JE Fouche JP Stein DJ Lochner C A comparison of brain volume and cortical thickness in excoriation (skin picking) disorder and trichotillomania (hair pulling disorder) in women Behav Brain Res 2015 279 255 258 10.1016/j.bbr.2014.11.029 25435313 Sani G Gualtieri I Paolini M Bonanni L Spinazzola E Maggiora M Drug Treatment of Trichotillomania (Hair-Pulling Disorder), Excoriation (Skinpicking) Disorder, and Nail-biting (Onychophagia) Curr Neuropharmacol 2019 17 8 775 786 10.2174/1570159X17666190320164223 30892151 Schienle A Übel S Wabnegger A Visual symptom provocation in skin picking disorder: an fMRI study Brain Imaging Behav 2018 12 5 1504 1512 10.1007/s11682-017-9792-x 29305750 Schlosser S Black DW Blum N Goldstein RB The demography, phenomenology, and family history of 22 persons with compulsive hair-pulling Ann Clin Psychiatry 1994 6 147 52 7881494 Sheehan DV Lecrubier Y Sheehan KH Amorim P Janavs J Weiller E The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10 J Clin Psychiatry 1998 59 Suppl 20 22 57 Smith SM Fast robust automated brain extraction Hum Brain Mapp 2002 17 143 155 12391568 Stein DJ Coetzer R Lee M Davids B Bouwer C Magnetic resonance brain imaging in women with obsessive-compulsive disorder and trichotillomania Psychiatry Res 1997 74 3 177 182 10.1016/s0925-4927(97)00010-3 9255863 Swedo SE Rapoport JL Leonard HL Schapiro MB Rapoport SI Grady CL Regional cerebral glucose metabolism of women with trichotillomania Arch Gen Psychiatry 1991 48 9 828 833 10.1001/archpsyc.1991.01810330052008 1929773 Tucker BT Woods DW Flessner CA Franklin SA Franklin ME The Skin Picking Impact Project: phenomenology, interference, and treatment utilization of pathological skin picking in a population-based sample J Anxiety Disord 2011 25 1 88 95 10.1016/j.janxdis.2010.08.007 20810239 Wabnegger A Schienle A The Role of the Cerebellum in Skin-Picking Disorder Cerebellum 2019 18 1 91 98 10.1007/s12311-018-0957-y 29934941 White MP Shirer WR Molfino MJ Tenison C Damoiseaux JS Greicius MD Disordered reward processing and functional connectivity in trichotillomania:a pilot study J Psychiatr Res 2013 47 9 1264 72 10.1016/j.jpsychires.2013.05.014 23777938

PMC007xxxxxx/PMC7614804.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9716531 J Med Chem J Med Chem Journal of medicinal chemistry 0022-2623 1520-4804 35080396 7614804 EMS181407 10.1021/acs.jmedchem.1c01684 Article Discovery of Novel Inhibitors of Uridine Diphosphate-N-Acetylenolpyruvylglucosamine Reductase (MurB) from Pseudomonas aeruginosa, an Opportunistic Infectious Agent Causing Death in Cystic Fibrosis Patients https://orcid.org/0000-0002-6035-7525 Acebrón-García-de-Eulate Marta Department of Biochemistry, University of Cambridge, Cambridge CB2 1GA, U.K Mayol-Llinàs Joan Holland Matthew T. O. Yusuf Hamied Department of Chemistry, University of Cambridge, Cambridge CB2 1EW, U.K Kim So Yeon Department of Biochemistry, University of Cambridge, Cambridge CB2 1GA, U.K Brown Karen P. Molecular Immunity Unit, Department of Medicine, MRC Laboratory of Molecular Biology, University of Cambridge, Cambridge CB2 0QH, U.K.; Cambridge Centre for Lung Infection, Royal Papworth Hospital, Cambridge CB23 3RE, UK Marchetti Chiara Yusuf Hamied Department of Chemistry, University of Cambridge, Cambridge CB2 1EW, U.K https://orcid.org/0000-0001-5916-0728 Hess Jeannine Yusuf Hamied Department of Chemistry, University of Cambridge, Cambridge CB2 1EW, U.K https://orcid.org/0000-0002-2734-2444 Di Pietro Ornella Yusuf Hamied Department of Chemistry, University of Cambridge, Cambridge CB2 1EW, U.K https://orcid.org/0000-0002-2734-2444 Mendes Vitor Department of Biochemistry, University of Cambridge, Cambridge CB2 1GA, U.K Abell Chris Yusuf Hamied Department of Chemistry, University of Cambridge, Cambridge CB2 1EW, U.K Andres Floto R. Molecular Immunity Unit, Department of Medicine, MRC Laboratory of Molecular Biology, University of Cambridge, Cambridge CB2 0QH, U.K.; Cambridge Centre for Lung Infection, Royal Papworth Hospital, Cambridge CB23 3RE, UK https://orcid.org/0000-0003-0205-5630 Coyne Anthony G. Blundell Tom L. 1 Corresponding Authors Joan Mayol-Llinas – Yusuf Hamied Department of Chemistry, University of Cambridge, Cambridge CB2 1EW, U.K.; jm2293@cam.ac.uk; Anthony G. Coyne – Yusuf Hamied Department of Chemistry, University of Cambridge, Cambridge CB2 1EW, U.K.; agc40@cam.ac.uk; Tom L. Blundell – Department of Biochemistry, University of Cambridge, Cambridge CB2 1GA, U.K.; tom@cryst.bioc.cam.ac.uk 10 2 2022 26 1 2022 19 7 2023 25 7 2023 65 3 21492173 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Pseudomonas aeruginosa is of major concern for cystic fibrosis patients where this infection can be fatal. With the emergence of drug-resistant strains, there is an urgent need to develop novel antibiotics against P. aeruginosa. MurB is a promising target for novel antibiotic development as it is involved in the cell wall biosynthesis. MurB has been shown to be essential in P. aeruginosa, and importantly, no MurB homologue exists in eukaryotic cells. A fragment-based drug discovery approach was used to target Pa MurB. This led to the identification of a number of fragments, which were shown to bind to MurB. One fragment, a phenylpyrazole scaffold, was shown by ITC to bind with an affinity of Kd = 2.88 mM (LE 0.23). Using a structure guided approach, different substitutions were synthesized and the initial fragment was optimized to obtain a small molecule with Kd = 3.57 μM (LE 0.35). Graphic abstract pmc■ Introduction Pseudomonas aeruginosa, a rod-shaped Gram-negative bacterium, is a frequent opportunistic agent of hospital-acquired infections.1 In cystic fibrosis (CF), P. aeruginosa is responsible for 80% of the lung infections of CF patients by the age of 18.2 Moreover, acquisition of P. aeruginosa by CF patients leads to 2.6 times higher risk of death, making chronic infection by this pathogen the major cause of death in this type of patient.3,4 This bacterium has become resistant to current antibiotics such as β-lactams due to its low membrane permeability, abundant efflux pumps, and various antibiotic-degradative enzymes.5 Currently, the antibiotics used against P. aeruginosa in the clinic are limited and resistant strains are increasing in hospitals worldwide.6 Consequently, P. aeruginosa has been classified as one of the six pathogens in the world that most require new antibacterial drugs.6–9 Therefore, it is necessary to design new antibiotics that act on novel targets of P. aeruginosa. In the cell wall of most bacteria, one of the main components is peptidoglycan, a peptide cross-linked polymer of alternating N-acetyl-glucosamine and N-acetyl-muramic acid (UNAM) units.10–12 The bacterial cell wall offers osmotic stability, and as a result, the enzymes involved in peptidoglycan biosynthesis are essential for bacterium survival. Therefore, the inhibition of any of these enzymes results in loss of bacterial cell wall followed by cell lysis.13 In recent decades, attention has been paid to the family of Mur enzymes,14 which synthesize the cell wall from the nucleotide sugar uridine diphosphate N-acetylglucosamine. Antibiotics that block one of these enzymes are of interest, and some have already been developed. One example is fosfomycin, which inhibits MurA (UDP-N-acetylglucosamine-enolpyruvyl-transferase), the Mur enzyme that catalyzes the first step of the cell wall biosynthesis.15 Unfortunately, P. aeruginosa has also developed resistance to this antibiotic by enzymatic deactivation or by decreasing its cellular uptake.16 The second enzyme in the pathway, UDP-N-acetylenolpyruvoylglucosamine reductase (MurB), has also proved to be of interest as a novel target in P. aeruginosa because no MurB homologue exists in eukaryotic cells. MurB catalyzes the reduction of UDP-N-acetylglucosamine enolpyruvate (UN-AGEP), the product of MurA, to N-acetyl-muramic acid (UNAM) using FADH2 (Figure 1).17,18 P. aeruginosa MurB (Pa MurB) is a monomeric enzyme that comprised three different domains (PDB code 4JB1) (Figure 2). There are domain I (amino acids 1–75 and 336–339), domain II (amino acids 76-191) that binds the FAD cofactor, and domain III (amino acids 192–335), which has binding sites for the NADPH cofactor and substrate (UNAGEP).12 The X-ray crystal structure of the complex of Pa MurB with FAD and NADP+ (PDB code: 4JB1) and the crystal structure of Escherichia coli MurB in complex with the substrate UNAGEP (Ec MurB, PDB code: 2MBR)19 have very similar threedimensional structures with identical residues in the active site12 and FAD bound in the same manner. Both enzymes are type I UNAGEP reductases, similar to the Staphylococcus aureus MurB (SaMurB, type IIa) and Thermus caldophilus MurB (type IIb) structures20,21 but lacking an α-helix and a protruding βαββ fold in the domain III (Figure 2). Extensive biochemical characterization of Ec MurB has led to the description of the reaction as a ping-pong bi-bi mechanism,22 where first the NADPH transfers an hydrogen to FAD followed by NADP+ dissociation from the enzyme. Successively, UNAGEP binds and the hydride is transferred from FADH2 to the vinyl ether of UNAGEP, becoming UNAM. Several MurB inhibitors have been designed using structure-based approaches23 based on the cocrystal structures of Ec MurB and Sa MurB.24–26 However, currently there are no inhibitors described against Pa MurB. In the present study, an in-house fragment library was screened against Pa MurB and this led to the identification ofsmall molecules that bind to Pa MurB. X-ray crystallography of one of the fragment hits was shown that it binds at the interface of FAD and the substrate binding pockets, which offers a novel strategy for the development of Pa MurB inhibitors. ■ Results Chemical Scaffold Identification A library of 960 rule-of-three compliant fragments was screened at a concentration of 5.0 mM against MurB using differential scanning fluorimetry (DSF) (Table 1). Fourteen fragments were shown as positive hits, and they were verified in triplicate at a 1.0 mM concentration. As a result, nine hits still had a positive thermal shift compared to the negative control. In the screening, the MurB–FAD complex was used due to the high affinity of FAD for MurB.12 The fragment hits identified were validated by X-ray crystallography. However, only the pyrazole derivative 4 was successfully crystallized in complex with Pa MurB (Figure 3). X-ray crystallography showed that this fragment binds in the catalytic pocket in close proximity to FAD.17 The binding affinity of fragment 4 was then determined using isothermal titration calorimetry (ITC) where the affinity was measured to be Kd = 2.88 mM. All the analogues were screened using two different biophysical techniques, DSF (at 5 mM and/or 1 mM) and surface plasmon resonance (SPR) at 1 mM. Fragments with significant increases in ΔTM or R higher than fragment 4 (R > Rf4) (see Figure 4) were validated by ITC. The goal was to identify potential binding analogues with higher binding affinity than fragment 4. Modification of the Methyl Group The methyl group on the pyrazole ring of fragment 4 points into a small pocket that the NADP+ does not fill (see Figure 3b). This pocket contains mainly hydrophilic residues such as Glu335, Asn243, Ser239, and Arg166. In addition, there is also a water molecule (W1, Figure 3b) in this small pocket, which is tightly bound to the hydrophilic residues by several hydrogen bonds. In order to further explore the structure–activity relationships (SAR) with this pocket, the methyl substituent was replaced by different groups that could possibly interact with the hydrophilic residues or with the water molecule (fragments 15–18) (Table 2). The substitution at the 5-position (R in Table 2) in the pyrazole ring with a trifluromethyl group (fragment 18) was shown to give an increase in affinity where a Kd of 0.25 ± 0.04 mM was measured using ITC. This fragment was successfully crystallized in complex with Pa MurB, and the X-ray crystal structure shows a similar binding mode to the original fragment hit 4 (Figure 5a). However, the trifluromethyl group is shown to interact with the water W1 in the small pocket and another water (W3) next to it, and these interactions displace the rest of the fragment 18 closer to the α-helix 6 (see Figure 5a). Modification of the Pyrazole Ring The CH on the 3-position of the pyrazole ring of the fragment 4 was changed for a N to observe if a possible hydrogen bond interaction could be formed with residue Tyr132 (fragment 19). This change did not give an increase in the melting temperature (Table 3). However, the X-ray crystal structure of Pa MurB in complex with fragment 19 shows a 180° flip of the five-membered ring due to the formation of an interaction of the N at the 3-position of the triazole with Arg166 (see Figure 6). In addition, this methyl group points to a small pocket that mainly contains hydrophilic residues such as Lys227, Tyr132, and Arg196 and a highly bound water molecule. These observations could suggest that an analogue ring consisting ofa pyrrole or imidazole containing two substituents at the 5- and 2-positions could also be of interest. Therefore, the pyrrole (fragment 20) and imidazole (fragment 21) analogues were synthesized, screened, and compared to fragment 4 using two different biophysical techniques, DSF and SPR (Table 3). Fragments with significant increase in ΔTM or R higher than fragment 4 (R > Rf4) at 1.0 mM were validated again by ITC. The pyrrole analogue 20 showed a greater thermal shift. Interestingly, one pyrrole derivative (fragment 3) was also identified in the initial library of hits. Subsequently, the methyl group at the 5-position was replaced with a CF3 (fragment 22) in order to compare it with fragment 18. However, this modification showed no change in activity. Finally, the methyl at the 2-position was changed to a hydrophilic group to allow interaction with the hydrophilic residues or the water molecule of the small pocket (fragment 23). Unfortunately, a lower affinity was observed (Kd = 208 ± 15 μM). As a result, two series of compounds were taken for further optimization, the pyrazole fragment 18 and the pyrrole fragments 20 and 22. Exploration of Substituents on the Phenyl Ring Initially, the replacement of the phenyl group was studied (Table 4) and substitution for a benzyl or a thiophenemethyl group did not increase the affinity (fragments 24 and 25). The change of the phenyl ring for a thiophenyl or pyridinyl showed lower thermal shifts (fragments 26–28). Consequently, the phenyl ring was retained and the introduction of different substituents on the ring was explored (Table 4). An increase in the melting temperature was observed by adding large apolar groups such as halogens or methyl groups at the 2-substiuted-positions. It was observed that the larger the group, the higher the affinity (fragments 29–33). Introduction of polar and apolar groups in other positions on the phenyl ring did not improve affinity (fragments 34–40). The introduction of two apolar 3-substituted groups was shown to slightly increase the affinity (fragments 41 and 42). It was observed that some of the initial fragment hits contained phenyl groups with 3,4-dichlorophenyl substituents (fragments 5, 6, and 9). Therefore, this modification was incorporated into the developed compounds and this change showed an increase in affinity, Kd = 26.1 ± 2.7 μM (compound 43). The introduction of the 2-methylphenyl group could orientate the pyrazole ring and the phenyl ring at around 90°, which is the conformation observed in the crystal structure of fragment 18 (Figure 5). Molecular simulations were performed to examine this (Figure 8).28 In the Pa MurB crystal structure, fragment 18 shows a dihedral angle of 86.5° between the pyrazole and the phenyl ring, whereas the minimum energy structure of fragment 18 has a dihedral angle of 120°. If an ortho substituent is added in the phenyl ring, then the dihedral angle of the minimum energy structure becomes similar to the one in the crystal structure. Substituents that give closer angles to the crystal structure showed better affinity (59.8° for 2-fluorophenyl (29), 70.0° for 2-chlorophenyl (30), 69.4° for 2-bromophenyl (31), 79.6° for 2-methylphenyl (32), and 90.5° for 2,6-dimethylphenyl (41)) (Table 4). The 3,4-dichlorophenyl group could improve the binding by hydrophobic interactions with Leu300, Leu228, and Val301 (see Figure 5b). Unfortunately, attempts to obtain X-ray crystal structures of these substituted compounds with Pa MurB were not successful. As a result, the 2-methylphenyl substitution was merged with the 3,4-dichlorophenyl substitution (fragments 44 and 45) (Table 5). This allowed the identification of the best substitution pattern, which was a phenyl ring substituted with a methyl group in the 2-position and two chlorines in the 4- and 5-position on the phenyl ring (fragment 44). Subsequently, some of these modifications were successfully translated into the pyrrole fragments 20 and 22 to yield fragments 46–52 and fragments 53–55, respectively. Additionally, it was observed that both chlorine atoms are important for binding as there is a decrease in thermal shift if one of the chlorines is removed (fragments 49 and 50). However, the fragment containing the pyrazole ring gave the lowest Kd of 3.57 ± 0.76 μM (fragment 44) (see Figure 7a). In the pyrrole series, the two substituents at the 2- and 5-position of the pyrrole ring can also affect the dihedral angle. Consequently, molecular simulations were also performed. fragments 20 and 22, which have angles of 100 and 70.1°, respectively, have an angle more similar to the crystal structure (86.5°) than Fragment 18, which has an angle of 120°. Fragments 20 and 22 showed greater affinity (Table 3) (see Figure 7b). However, these two fragments both had similar binding affinities. This suggests that the CF3 group is not improving the affinity as in the pyrazoles. However, if a 2-methyl group is added in the phenyl ring of these fragments, then both resulting fragments (fragments 46 and 53) have an angle more similar to the one of the crystal structure (90.0 and 91.0°, respectively). Due to the fact that fragment 53 has a CF3 group in the 2-position instead of the CH3, a higher affinity is now observed. Consequently, once the dihedral angle is close to what is observed in the X-ray crystal structure, better affinity can be observed by the addition of the CF3 group. Addition of a second ortho-methyl group (fragment 47) did not change the dihedral angle (90.0°) as it did in the pyrazole series, and no change in affinity was observed. Modifications at the Carboxylic Acid Upon exploration of the SAR on the two rings, a further approach was explored to examine whether the compounds can be grown from the carboxylic acid moiety. NADP+ was shown by X-ray crystallography to form a “sandwich” π-stacking interaction with its adenine ring to the Tyr196 and Tyr264 at the entrance of the binding pocket (see Figure 3a). In the absence of NADP+, these two tyrosine residues are not stabilized and the α-helix and βαββ fold in domain III (see Figure 2a) are flexible. The carboxylic acid of fragment 4 forms a polar interaction with the Tyr132 and it points to the same direction as the adenine ring of NADP+ (see Figure 3b). Consequently, this is a good vector for developing these compounds. Therefore, the carboxylic acids of some of the previously developed fragments were grown with different functional groups (Table 6). Amide and ester derivatives did not show any activity (fragments 56–58). Only the sulfonamide derivatives were detected to be active by both DSF and ITC. If the methanesulfonyl group (fragment 59) was changed for a benzenesulfonyl group (fragment 60), then a higher ΔTM was observed (from +0.5 to +1.0 °C, respectively). However, fragment 60 (Kd = 0.32 ± 0.06 mM) showed a similar binding affinity than the acid analogue 18 (Kd = 0.25 ± 0.04 mM). The activity was lost when the phenyl group was changed for a benzyl group (fragment 61). Consequently, the effect in the binding affinity by the substitution of the acid moiety for a benzenesulfonyl group was investigated with the most optimized fragments such as 41 and 44. After growing fragment 41,a ΔTM = +3.0 °C and a Kd = 25.6 ± 2.9 μΜ were observed (fragment 62). The addition of a bromine substituent on this N-benzenesulfonyl group decreased the ΔTM to +1.3 °C (fragment 63). In the case of fragment 44, after the addition of the N-benzenesulfonyl group, a ΔTM of +2.7 °C and a Kd = 12.0 ± 3.7μM were observed (fragment 64). Attempts to obtain X-ray crystal structures of these fragments proved unsuccessful. Synthetic Chemistry Different synthetic routes were employed to prepare the different types of fragments (Scheme 1). Pyrazoles (fragments 17 and 24–45) were prepared using two key steps. The first step involved the reaction of α,β-unsaturated keto esters with different hydrazines in the presence of Et3N and EtOH at 80 °C to yield the desired 4,5-substituted pyrazoles.29 The 3,4-substituted regioisomers were only observed when aliphatic hydrazines were employed as starting materials (synthesis of fragments 24 and 25). The second step involved the hydrolysis of the resulting esters with 2 M NaOH in EtOH at 80 °C. The triazole 19 was prepared from ethyl acetoacetate and phenylboronic acid in the presence of sodium azide, copper acetate, and catalytic amounts of piperidine.30 No other major regioisomer was observed. The resulting product was hydrolyzed as previously. Pyrrole derivatives (20, 22, 23, and 46-55) were also prepared using two key steps. The first step involved the reaction of different α,β-keto esters with different amines using the Paal–Knorr reaction.31,32 This reaction used acetic acid as a solvent at 120 °C when R2 = CH3. However, when R2 = CF3, para-toluenesulfonic acid in toluene at 110 °C was used instead due to the degradation ofthe α,β-keto esters in the acetic acid. Finally, the resulting ester was hydrolyzed as before. Imidazole 21 was also prepared in two main steps. In the first step, an α,β-keto ester analogue reacted with aniline in the presence of trifluoroacetic acid and butyronitrile at 120 °C.33 The product was hydrolyzed by 6 M HCl at 100 °C. The basic conditions used previously were not used here due to the degradation of the product. Finally, the amide and sulfonamide derivatives (58 and 59–64, respectively) were prepared from the corresponding acids in a one-pot step. For the amide, the carboxylic acid was first converted into an acyl chloride by reaction with thionyl chloride at 80 °C. This intermediate was then reacted with benzyl amine in the presence of pyridine in 1,4-dioxane at room temperature to yield the corresponding amide. For the sulfonamides, the carboxylic acid reacted with the desired sulfonamide through a coupling reaction involving the use of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide and catalytic amounts of 4-dimethylaminopyridine in dichloromethane at room temperature. ■ Discussion And Conclusions The development of inhibitors of peptidoglycan targeting MurB enzymes has proved exceptionally challenging and currently, there are no approved antibiotics that target any of the nine subsequent steps after MurA. This study illustrates the successful application of a fragment-based approach to obtain, for the first time, a potent candidate that binds to the catalytic pocket of Pa MurB. Screening of 960 fragment libraries by DSF and validation using X-ray crystallography allowed the identification of a pyrazole derived fragment 4, which showed a Kd of 2.8 mM by ITC. This fragment was synthetically modified to improve its binding affinity. Different substitutions were tested in comparison with the initial fragment 4 using two different biophysical techniques, DSF and SPR at 1.0 mM. Binding parameters were calculated using ITC for those fragments with significant binding increase than the initial fragment. The substitution of the 5-methyl group on fragment 4 for a 5-trifluoromethyl group on the pyrazole ring or the substitution of the 5-methylpyrazole for a 2,5-dimethylpyrrole or 2-methyl-5-trifluoromethylpyrrole ring increased the binding affinity to values of Kd around 0.1–0.3 mM (fragments 18, 20, and 22, respectively). Subsequently, the introduction of an ortho-methyl group or a 3,4-dichloro group on the phenyl ring decreased the Kd to around 50–20 μM while maintaining the ligand efficiency. The initial library of hits contained fragments with a 2,5-dimethylpyrrole ring or 3,4-dichlolorophenyl groups, suggesting that it was important to look at these fragments for the optimization of fragment 4. The fact that the substitutions could be translated from pyrazoles to pyrroles suggests that the binding mode could be similar; however, pyrazoles showed to bind tighter. Exploration of the SAR on the carboxylic showed that a phenylsulfonamide group was tolerated, but no gain in affinity was observed. The best fragments were obtained by merging the ortho-methyl group with the meta,para-dichloro groups and showed an LE = 0.35 (fragment 44) and an LE = 0.32 (fragment 54). As a result, fragments with higher potency to that of the cofactors of Pa MurB have been designed. Consequently, these fragments can grow into Pa MurB inhibitors that would disrupt cell wall biosynthesis. This fragment is a good candidate because the binding is mediated by π–π interactions between FAD and Tyr132; thus, there is no possibility of P. aeruginosa becoming resistant by mutations in MurB. Although Tyr132 is quite conserved in bacteria (in MurB from E. coli and S. aureus), some bacteria such as T. caldophilus have R132. However, if Tyr132 were to mutate to another amino acid, then the π-π interaction would likely be replaced with a π-polar interaction. A future work can elaborate this fragment in order to increase the interactions with MurB in the catalytic pocket. ■ Experimental Section Cloning, Protein Expression, and Purification MurB gen of P. aeruginosa was designed based on the sequence available in the NCBI database and synthesized using GeneArt (Invitrogen). The gene was cloned between the BamHI and HindIII sites in pET28a vector (Novagen) and modified with an N-terminal 6xHis-SUMO tag. The resulting plasmid was confirmed by DNA sequencing and transformed into the E. coli BL21(DE3) strain (Invitrogen). Transformed cells were grown to OD610 = 0.6 in LB media (Invitrogen) containing 30 mg L-1 kanamycin at 37 °C. At this OD, protein expression was induced using 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) overnight at 18 °C. Cells were harvested by centrifugation and resuspended in 50 mM Tris (pH 8), 0.5 M NaCl, 5 mM MgCl2, and 20 mM imidazole with protease inhibitor tablets (Roche) and DNAse I. Cells were lysed, sonicated, and centrifuged at 30,000g for 45 min to collect the supernatant. Pa MurB was purified with a HiTrap IMAC Sepharose FF column (GE-Healthcare), equilibrated with 50 mM Tris (pH 7.5), 0.5 MNaCl, and 20 mM imidazole, and the elution was performed in the same buffer with 500 mM imidazole. Imidazole was removed with overnight dialysis in 50 mM Tris (pH 8) and NaCl (250 mM). Meanwhile, the SUMO tag was cleaved by adding Ulp1 Protease at a 1:100 ratio. The SUMO tag, Ulp1 protease, and Pa MurB were separated using a Superdex 200 column equilibrated with 50 mM Tris (pH 8) and NaCl (250 mM). Fraction purity was determined by SDS-PAGE, and the purest fractions were pooled and concentrated to 25.5 mg mL-1 in the same buffer, flash frozen in liquid nitrogen, and stored at −80 °C. Differential Scanning Fluorimetry Differential scanning fluorimetry was performed using a Bio-Rad CFX96 Touch PCR system from 25 to 95 °C in 0.5 °C increments of 30 s duration. Samples were run in 96-well clear-bottomed plates. For these experiments, each well contained a final volume of 25 μL, consisting of 25 mM Tris–HCl (pH 8.0), 150 mM NaCl, 5X SyproOrange, 10 μM Pa MurB, and either 5% DMSO-d6 or ligand at 5.0 or 1.0 mMin DMSO-d6 as specified. Controls were used for all experiments, with DMSO-d6 (reference) and NADP+ (positive control) used instead of the fragment. The resulting data (fluorescence intensity vs temperature) was fitted to obtain the denaturing temperature TM (point of sigmoidal inflection) as the maximum of each curve’s derivative. This analysis was performed using Microsoft Office Excel. The reference unfolding temperature of Pa MurB in 5% DMSO-d6 was subtracted from the values in the presence of the fragment to obtain the thermal shift. The thermal shifts at 5.0 mM were recorded once (n = 1), and the thermal shifts at 1.0 mM were recorded three times (n = 3). Crystallization and Data Collection Crystallization of the complexes was carried out by seeding using Pa MurB:NADPH microcrystals as a nucleation starting point. Pa MurB:NADPH crystals were set up, manually mixing 1 μL of Pa MurB at 25.5 mg μL-1 and 2 mM NADPH and 1 μL of crystallization condition mix (160 mM (NH4)SO4, 80 mM sodium acetate at pH 4.6, 20% PEG 4000, and 20% glycerol) using the hanging-drop vapor-diffusion method in 24-well VDX greased plates (Hampton Research, Aliso Viejo, California, USA). Crystallization of the fragment complexes or Apo crystal was prepared using the sitting-drop vapor-diffusion method at 25 °C, and the plates were mounted in the Mosquito Crystal robot (TPP Labtech, Hertfordshire, UK). In each crystallization drop, 0.4 μL of reservoir solution and 0.05 μL of microseeds were added to 0.2 μL of protein solution. The protein droplets were equilibrated over 70 μL of reservoir solution mix (30% glycerol and 22% PEG 1500). Suitable crystals for X-ray diffraction grew in 1 week. Diffraction data were processed and reduced using XDS30 and Aimless from the CCP4 suite.31 All the structures crystallized in the P61 space group with one protomer per asymmetric unit. Initial phases were determined using the previously published structure of Pa MurB (PDB code: 4JB1)12 as a model with the PHASER32 program from the PHENIX software package.33 Model building and structure validation were refined using PHENIX and Coot.34 All data sets were collected at stations I03 and I04-1 at the Diamond Light Source (Oxford, UK). Data collection and refinement statistics are summarized in Table S1 General Chemistry Commercially available starting materials and fragments 4, 15, 16, 18, and 56 were obtained from Sigma–Aldrich, Acros, Fluorochem, and Alfa Aesar. All non-aqueous reactions were performed under a nitrogen atmosphere unless otherwise stated. Watersensitive reactions were performed in anhydrous solvents in oven-dried glassware cooled under nitrogen before use. Petrol refers to petroleum spirit (b.p.: 40–60 °C), THF refers to tetrahydrofuran, and DCM refers to dichloromethane. A rotary evaporator was used to remove the solvents in vacuo. Thin-layer chromatography was performed using Merck glass-backed silica (Kieselgel 60 F254 0.25 mm plates). An ultraviolet lamp (λ-max = 254 nm) and KMnO4 were used for visualization. Flash column chromatographywas performed using automated Isolera Spektra One/Four purification systems and an appropriately sized Biotage SNAP column containing KP-silica gel (50 μM). A Perkin-Elmer One FT-IR spectrometer was used to analyze the infrared spectra. Absorptions are reported in wavenumbers (cm−1). An SQD2 mass spectrometer detector (Waters) utilizing electrospray ionization (ESI) was used for low-resolution mass spectrometry (MS). High-resolution mass spectrometry (HRMS) was recorded using a Waters LCT Premier Time of Flight (TOF) mass spectrometer or a Micromass Quadrapole-Time of Flight (Q-TOF) spectrometer. The purity of tested fragments was determined by high-performance liquid chromatography (HPLC). All final fragments had purity greater than 95% unless otherwise stated. HPLC was carried out using an Ultra Performance Liquid Chromatographic system (UPLC) Waters Acquity H-class. Samples were detected using a Waters Acquity TUV detector at two wavelengths (254 and 280 nm). Samples were run using an Acquity UPLC HSS column and a flow rate of 0.8 mL/min. The eluent consisted of 0.1% formic acid in water (A) and acetonitrile (B) (gradient, from 95% A to 5% A over a period of 4 min). Proton (1H), carbon (13C), and fluorine (19F) NMR data were collected on a Bruker 400 MHz spectrometer. Data were collected at 300 K. Chemical shifts (δ) are given in parts per million (ppm), and they are referenced to the residual solvent peak. 19F NMR spectra were references to TFA. Coupling constants (J) are reported in Hertz (Hz), and splitting patterns are reported in an abbreviated manner: app. (apparent), s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet), and br (broad). General Procedure A A dispersion of 20% Pd(OH)2 on carbon (0.5 equiv) was added to a solution of the benzyl derivative (1.0 eq) in ethanol (0.1 M). The mixture was stirred under hydrogen at room temperature for 24 h. Subsequently, it was filtered through celite and concentrated in vacuo to give a crude product. General Procedure B An aqueous solution of 2 M NaOH (6.0 eq) was added dropwise to a solution of the ester derivative (1.0 equiv) in ethanol (0.2 M), and the resulting mixture was stirred at 80 °C. After consumption of the starting material, the solvent was removed in vacuo, water (5 mL for each 1.00 mmol of the ester derivative) was added, and the mixture was washed with ethyl acetate (5 mL for each 1.00 mmol of the ester derivative). Successively, the aqueous layer was acidified to pH 4 with an aqueous solution of 1 M HCl and extracted with ethyl acetate (3 X 5 mL for each 1.00 mmol of the ester derivative). The organic phases were combined, washed with brine (5 mL for each 1.00 mmol of the ester derivative), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to yield a crude product or the corresponding acid derivative. General Procedure C.36 The amine derivative (1.0 equiv) was added to a solution of the carbonyl derivative (1.0 equiv) in acetic acid (0.27 M), and the mixture was stirred at 120 °C until consumption of the starting material. After cooling to room temperature, water (4 mL for each 1.00 mmol of amine derivative) was added and the mixture was extracted with EtOAc (3 × 8 mL for each 1.00 mmol of amine derivative). The organic phases were combined, dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. General Procedure D.37 TsOH.H2O (0.5 equiv) was added to a solution of the amine derivative (1.0 equiv) and the dicarbonyl derivative S9 (1.0 equiv) in toluene (concentration of the amine derivative: 0.27 M), and the mixture was stirred at 110 °C. After consumption of the starting materials, the mixture was allowed to cool to room temperature and water (3 mL for each 1.00 mmol of the amine derivative) and EtOAc (3 mL for each 1.00 mmol of the amine derivative) were added. The phases were separated, and the aqueous phase was extracted with EtOAc (3 × 3 mL for each 1.00 mmol of the amine derivative). The organic phases were combined, dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. General Procedure E.29 Triethylamine (1.2 or 2.4 equiv if the hydrazine dihydrochloride salt derivative was used) was added dropwise to a stirred solution of ethyl-3-ethoxy-2-(2,2,2-trifluoroacetyl)acrylate (1.0 equiv) and the hydrazine hydrochloride salt derivative (1.0 equiv) in ethanol (concentration of acrylate derivative: 0.4 M), and the resulting mixture was stirred at 80 °C. After consumption ofthe starting material, the mixture was allowed to cool to room temperature and the solvent evaporated under reduced pressure. EtOAc (2 mL for each 1.00 mmol of the hydrazine derivative) and water (2 mL for each 1.00 mmol of the hydrazine derivative) were added, the phases were separated, and the aqueous phase was extracted with EtOAc (3 × 2 mL for each 1.00 mmol of the hydrazine derivative). The organic phases were combined, washed with brine (2 mL for each 1.00 mmol of the hydrazine derivative), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. General Procedure F A solution of NaNO2 (1.2 equiv) in H2O (1.8 M) was added dropwise to a solution of the amine derivative (1.0 equiv) in concentrated HCl (0.3 M) at 0 °C. The reaction was stirred for 30 min at 0 °C, and the insolubilities were removed. Successively, a solution of SnCl2η(H2O)2 (3.0 equiv) in 1:1 concentrated HCl-H2O (1.1 M) was added. After stirring the reaction for 2.5 h at 0 °C, the precipitate was filtered, washed with a cold aqueous solution of 6.0 M HCl, washed with hexane, and dried in vacuo to give the hydrazine derivative. General Procedure G 1-Ethyl-3-(3-dimethylaminopropyl)-carbodiimide hydrochloride (1.1 equiv), the sulfonamide derivative (1.1 equiv), and 4-dimethylaminopyridine (0.1 equiv) were added to a solution of the carboxylic acid derivative (1.0 equiv) in DCM (0.1 M). After stirring at room temperature for 18 h, water (8 mL for each 1.00 mmol of carboxylic acid derivative) was added, the phases were separated, and the aqueous phase was extracted with DCM (3 × 8 mL for each 1.00 mmol of carboxylic acid derivative). The organic phases were combined, dried under anhydrous sodium sulfate, filtered, and concentrated in vacuo to give a crude product or the sulfonamide derivative. Ethyl 4-(Benzyloxy)-3-oxobutanoate (S1) Benzyl alcohol (2.00 mL, 19.3 mmol) was added to a suspension of 60% NaH in mineral oil (1.16 g, 29.0 mmol) in THF (24 mL) at 0 °C. After stirring at room temperature for 2 h, ethyl 4-chloroacetoacetate (2.1 mL, 15.4 mmol) was added dropwise over 30 min and the reaction mixture was stirred at room temperature for 18 h. Successively, the mixture was acidified to pH 2 with an aqueous solution of 6.0 M HCl. Water (10 mL) and EtOAc (10 mL) were added, the phases were separated, and the aqueous phase was extracted with EtOAc (3 × 10 mL). The organic phases were combined, dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to yield a crude product. The crude product was purified by flash column chromatography (0–5% EtOAc in petrol) to give the benzyl derivative S139 as yellow oil (2.50 g, 10.6 mmol, 55%). Rf 0.41 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.40-7.27 (m, 5H), 4.58 (s, 2H), 4.16 (q, J = 7.1 Hz, 2H), 4.14 (s, 2H), 3.53 (s, 2H), 1.24 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 201.7, 167.0, 137.0, 128.6, 128.1, 127.9, 74.8, 73.5, 61.4,46.1, 14.1 ppm. MS (ESI): [M + H]+237.1. IR (neat): νmax 3064–2871 (w, C–H), 1721 (s, C⩵O), 1656 (w, C⩵O), 1454 (m), 1393 (m), 1367 (m), 1317 (m), 1229 (m), 1098 (m), 1030 (m) cm-1. Ethyl 5-[(Benzyloxy)methyl]-1-phenyl-1H-pyrazole-4-carboxylate(S2) The benzyl derivative S1 (0.50 g, 2.11 mmol) and N,N-dimethylformamide dimethyl acetal (0.35 mL, 2.63 mmol) were stirred at room temperature for 18 h and concentrated in vacuo to yield the crude product. This was dissolved in EtOH (4 mL), phenylhydrazine hydrochloride (0.30 g, 2.11 mmol) and Et3N (0.35 mL, 2.53 mmol) were added, and the reaction mixture was stirred at 80 °C for 18 h. After allowing it to cool to room temperature, the solvent was evaporated under reduced pressure and the residue was taken up in ethyl acetate (4 mL) and water (4 mL). The phases were separated, and the aqueous phase was extracted with EtOAc (3 × 4 mL). The organic phases were combined, washed with brine (4 mL), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. The crude product was purified using flash column chromatography (7% EtOAc in petrol) to give the benzyl derivative S2 as colorless oil (0.38 g, 1.13 mmol, 54%). Rf: 0.50 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.09 (s, 1H), 7.65 (d, J = 7.7 Hz, 2H), 7.50-7.40 (m, 3H), 7.36-7.25 (m, 5H), 4.81 (s, 2H), 4.63 (s, 2H), 4.35 (q, J = 7.1 Hz, 2H), 1.38 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 163.3, 142.0, 138.9, 137.6, 129.2, 128.7, 128.5.128.0. 127.9.125.1.114.9, 73.1, 60.4,60.3, 14.5 ppm. MS (ESI): [M + H]+ 337.3. IR (neat): νmax 3113–2868 (w, C–H), 1708 (s, C = O), 1549 (m), 1501 (m), 1379 (m), 1265 (s), 1241 (s), 1189 (m), 1093 (s), 1063 (s), 1022 (s) cm-1. Ethyl 5-(Hydroxymethyl)-1-phenyl-1H-pyrazole-4-carboxylate (S3) According to General Procedure A, the benzyl derivative S2 (0.32 g, 0.95 mmol) gave a crude product. The crude product was purified using flash column chromatography (5% EtOAc in petrol) to give the alcohol derivative S3 as colorless oil (0.22 g, 0.89 mmol, 94%). Rf. 0.50 (50:50 EtOAc.petrol). 1H NMR (400 MHz, CDCl3): δ 8.02 (s, 1H), 7.53-7.42 (m, 5H), 4.76 (s, 2H), 4.37 (q, J =7.1 Hz, 2H), 1.39 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDQ3): δ 164.9, 146.6, 141.8, 138.3, 129.4, 129.1, 125.4, 114.0, 61.0, 55.1, 14.4 ppm. MS (ESI): [M +H]+247.1. IR (neat): νmax3272 (s, O–H), 3057-2851 (w, C–H), 1698 (s, C⩵O), 1558 (m), 1497 (m), 1460 (m), 1410 (m), 1381(m), 1244 (s), 1207 (s), 1090 (s), 1026 (s) cm-1. 5-(Hydroxymethyl)-1-phenyl-1H-pyrazole-4-carboxylic Acid (17) According to General Procedure B, the alcohol derivative S3 (56 mg, 0.23 mmol) was stirred for 4 h to give the acid derivative 17 as a white amorphous solid (47 mg, 0.22 mmol, 94%). Rf: 0.16 (20:80 MeOH:DCM). 1H NMR (400 MHz, CD3OD): δ 8.05 (s, 1H), 7.65-7.47 (m, 5 H), 4.80 (s, 2H) ppm. 13C NMR (100 MHz, CD3OD): δ 166.5, 146.9, 143.0, 139.8, 130.3, 130.2, 126.5, 115.2, 53.8 ppm. MS (ESI): [M + H]+ 219.0. HPLC: retention time: 1.31 min (>99%). IR (neat): νmax 3295 (s, O–H), 3063-2548 (s, C-H), 1673 (s, C⩵O), 1560 (s), 1458 (m), 1416 (m), 1282 (s), 1262 (s), 1226 (m), 1096 (w), 1019 (s), 936 (s) cm−1. HRMS: calculated for C11H10N2O3 [M + H]+ = 219.0770, observed = 219.0765. Ethyl 5-Methyl-1-phenyl-1H-1,2,3-triazole-4-carboxylate (S4) According to a procedure,35 sodium azide (0.26 g, 4.10 mmol) and copper(II) acetate (37.2 mg, 0.20 mmol) were added to a solution of phenylboronic acid (0.25 g, 2.05 mmol) in DMSO (10 mL) and water (1 mL). After stirring for 4 h at room temperature, ethyl acetoacetate (0.28 mL, 2.25 mmol) and piperidine (41 μL, 0.41 mmol) were added and the reaction mixture was stirred at 80 °C for 18 h. Successively, the reaction was allowed to cool to room temperature and an aqueous solution of 5% ammonium hydroxide (20 mL) and EtOAc (20 mL) were added. The phases were separated, and the aqueous phase was extracted with EtOAc (3 × 10 mL). The organic phases were combined, washed with brine (3 × 30 mL), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product, which was purified by flash column chromatography (10-20% EtOAc in petrol) to give the ester derivative S440 as yellow oil (0.32 g, 1.38 mmol, 67%). Rf: 0.43 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.58-7.48 (m,3H), 7.45-7.38 (m, 2H), 4.43 (q, J = 7.1 Hz, 2H), 2.55 (s, 3H), 1.41 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 161.8, 138.9, 136.7, 135.5, 130.1, 129.7, 125.4, 61.1, 14.4, 10.0 ppm. MS (ESI): [M + H]+ 232.1. IR (neat): νmax 3061-2871 (w, C–H), 1712 (s, C⩵O), 1597 (w), 1566 (w), 1504 (m), 1423 (m), 1374 (m), 1350 (w), 1278 (w), 1244 (s), 1228 (s), 1207 (s), 1107 (s), 1010 (w), 980 (w) cm-1. 5-Methyl-1-phenyl-1H-1,2,3-triazole-4-carboxylic Add (19) According to General Procedure B, the ester derivative S4 (83 mg, 0.36 mmol) was stirred for 3 h to give the acid derivative 1941 as a white amorphous solid (70 mg, 0.34 mmol, 94%). Rf: 0.29 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 10.83 (br. s, 1H), 7.68-7.36 (m, 5H), 2.62 (s, 3H) ppm. 13CNMR (100 MHz, CDCl3): δ 165.5.140.0. 136.1.135.3.130.4.129.8.125.4.10.2ppm.MS (ESI): [M + H]+ 204.1. HPLC: retention time: 1.42 min (>99%). IR (neat): vmax 3067 (s, O–H), 2928-2580 (w, C–H), 1681 (s, C⩵O), 1566 (m), 1494 (m), 1452 (m), 1270 (m), 1241 (m), 1229 (m), 1117 (m), 1090 (m) cm-1. Ethyl 2-Acetyl-4-oxopentanoate (S5) According to a procedure,42 chloroacetone (1.26 mL, 15.8 mmol) was added to a solution of ethyl acetoacetate (2.00 mL, 15.8 mmol) in triethylamine (15 mL) and the reaction mixture was stirred at 90 °C for 18 h. After allowing the mixture to cool to room temperature, the mixture was concentrated in vacuo. Water (10 mL) and DCM (10 mL) were added, the phases were separated, and the aqueous phase was extracted with DCM (3 × 10 mL). The organic phases were combined, washed with brine (10 mL), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product, which was purified by flash column chromatography (0–10% EtOAc in petrol) to give the carbonyl derivative S537 as yellow liquid (1.20 g, 6.44 mmol, 41%). Rf: 0.51 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 4.16 (q, J =7.1 Hz, 2H), 3.98 (dd, J =8.2, 5.7 Hz, 1H), 3.11 (dd, J = 18.5, 8.2 Hz, 1H), 2.92 (dd, J = 18.5, 5.7 Hz, 1H), 2.32 (s, 3H), 2.16 (s, 3H), 1.25 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 205.7, 202.2, 168.8, 61.8, 53.8, 41.6, 30.1, 29.7, 14.1 ppm. MS (ESI): [M + H]+ 187.1. IR (neat): νmax 2984–2927 (w, C–H), 1739 (m, C⩵O), 1711 (s, C⩵O), 1359 (m), 1259 (m), 1229 (m), 1157 (m) cm−1. Ethyl 2,5-Dimethyl-1-phenyl-1H-pyrrole-3-carboxylate (S6) According to General Procedure C, aniline (0.12 mL, 1.34 mmol) and the carbonyl derivative S5 were stirred for 18 h to give a crude product, which was purified by flash column chromatography (5% EtOAc in petrol) to give the ester derivative S636 as yellow oil (0.27 g, 1.11 mmol, 83%). Rf 0.70 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.52-7.41 (m, 3H), 7.17 (d, J = 7.6 Hz, 2H), 6.37 (s, 1H), 4.28 (q, J = 7.1 Hz, 2H), 2.29 (s, 3H), 1.97 (s, 3H), 1.35 (t, J = 7.1, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 165.8, 137.8, 136.3, 129.4, 128.8, 128.6, 128.3, 111.5, 107.6, 59.3, 14.7, 12.7, 12.5 ppm. MS (ESI): [M + H]+ 244.2. IR (neat): νmax 2981–2898 (w, C–H), 1738 (w), 1693 (s, C= O), 1540 (w), 1500 (w), 1408 (m), 1369 (w), 1226 (s), 1216 (s), 1079 (s), 1014 (m) cm−1. 2,5-Dimethyl-1-phenyl-1H-pyrrole-3-carboxylic Acid (20) According to General Procedure B, the ester derivative S6 (0.10 g, 0.41 mmol) was stirred for 2 days to give the acid derivative 2041 as a white amorphous solid (70 mg, 0.32 mmol, 78%). Rf: 0.54 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 10.86 (br. s, 1H), 7.55-7.42 (m, 3H), 7.20 (d, J = 7.4 Hz, 2H), 6.44 (s, 1H), 2.32 (s, 3H), 1.99 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 171.4, 137.8, 137.7, 129.5, 129.2, 128.7, 128.3, 110.8, 108.2, 12.8, 12.7 ppm. MS (ESI): [M + H]+ 216.1. HPLC: retention time: 1.72 min (>99%). IR (neat): νmax 3056 (s, O–H), 2917-2584 (s, C–H), 1651 (s, C⩵O), 1579 (m), 1531 (m), 1494 (m), 1453 (m), 1400 (m), 1261 (s), 1084 (m) cm−1. Methyl 2-Acetamido-3-oxobutanoate (S7) According to a procedure,43 a solution of sodium nitrite (3.30 g, 48.0 mmol) in water (4 mL) was added dropwise to a solution of methyl acetoacetate (4.00 mL, 36.8 mmol) in acetic acid (10 mL). After stirring the solution at room temperature for 2 h, water (25 mL) was added and stirred for further 30 min. Successively, the solution was extracted with diethyl ether (3 × 50 mL). The organic phases were combined, washed with a saturated sodium bicarbonate aqueous solution (50 mL), dried under anhydrous sodium sulfate, filtered, and concentrated in vacuo. Acetic acid (31.5 mL) and acetic anhydride (9.20 mL) were added, and the mixture was cooled to 0 °C. Subsequently, zinc powder (12.0 g, 184 mmol) was slowly added and the reaction was stirred at room temperature for 18 h. After filtering the mixture through celite, water (30 mL) and DCM (50 mL) were added. The phases were separated, and the aqueous phase was extracted with DCM (3 × 50 mL). Successively, the organic phases were combined, washed with a saturated sodium bicarbonate aqueous solution (50 mL), filtered, and concentrated in vacuo. The crude product was purified by flash column chromatography (33% EtOAc in petrol) and triturated with Et2O to give the carbonyl derivative S743 as a white solid (2.55 g, 14.7 mmol, 40%). Rf 0.36 (90:10 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 6.76 (app. br. s, 1H), 5.25 (d, J = 6.6 Hz, 1H), 3.78 (s, 3H), 2.35 (s, 3H), 2.04 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 198.6, 169.9, 166.7, 63.0, 53.3, 28.1, 22.7 ppm. MS (ESI): [M + H]+ 174.1. IR (neat): νmax 3231 (s, N–H), 3028–2956 (w, C–H), 1741 (s, C⩵O), 1724 (s, C⩵O), 1632 (s, C⩵O), 1524 (s), 1433 (m), 1375 (m), 1290 (m), 1223 (s), 1157 (s), 1140 (s) cm−1. Methyl 2,5-Dimethyl-1-phenyl-1H-imidazole-4-carboxylate (S8) According to a modified procedure,38 trifluoroacetic acid (0.12 mL, 1.48 mmol) was added to a solution of the dicarbonyl derivative S7 (0.20 g, 1.14 mmol) and aniline (0.14 mL, 1.48 mmol) in butyronitrile (4.40 mL). After stirring at 120 °C for 4 h, DCM (10 mL) was added and the solution was washed with a saturated sodium carbonate aqueous solution (10 mL), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo. The crude product was purified by flash column chromatography (20% petrol in EtOAc) to give the ester derivative S844 as a white solid (0.11 g, 0.47 mmol, 41%). Rf 0.23 (80:20 EtOAc.petrol). 1H NMR (400 MHz, CDCl3): δ 7.50-7.40 (m, 3H), 7.11 (d, J = 7.1 Hz, 2H), 3.81 (s, 3H), 2.23 (s,3H),2.13 (s, 3H) ppm. 13C NMR(100MHz, CDCl3): δ 164.2,144.6,136.9, 135.5,129.8, 129.4, 127.3, 51.2, 13.7, 10.7 ppm. MS (ESI): [M + H] + 231.1. IR (neat): νmax 3049–2925 (w, C–H), 1709 (s, C⩵O), 1543 (m), 1494 (m), 1434 (m), 1403 (m), 1372 (m), 1360 (s), 1186 (s), 1176 (s), 1092 (s) cm−1. 2,5-Dimethyl-1-phenyl-1H-imidazole-4-carboxylic Acid (21) A 6 M HCl aqueous solution (0.50 mL) was added to the ester derivative S8 (10 mg, 43.4 μMol), and the mixture was stirred at 100 °C for 18 h. The solvent was removed in vacuo to yield the acid derivative 2144 as a hydrochloride salt (9 mg, 43.4 μMol, >99%). Rf 0.11 (20:80 MeOH.DCM). 1H NMR (400 MHz, CD3OD): δ 7.76-7.68 (m, 3H), 7.57-7.50 (m, 2H), 2.46 (s, 3H), 2.38 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.7, 147.8, 138.7, 133.5, 132.6, 131.7, 128.5, 121.4, 11.6, 10.4 ppm. MS (ESI): [M + H]+ 217.1. HPLC: retention time: 0.97 min (91%). IR (neat): νmax 2761 (s, O–H, C–H), 1721 (s, C⩵O), 1444 (m), 1331 (m), 1220 (m), 1170 (s), 1117 (s) cm−1. Ethyl 4-Oxo-2-(2,2,2-trifluoroacetyl)pentanoate (S9) Ethyl 4,4,4-trifluoroacetoacetate (2.31 mL, 15.8 mmol) was added dropwise to a suspension of 60% NaH in mineral oil (0.63 g, 15.8 mmol) in 1,2-dimethoxyethane (8.00 mL) at 0 °C. After stirring for 30 min at 0 °C, chloroacetone (1.45 mL, 18.2 mmol) in 1,2-dimethoxyethane (2 mL) and KI (32 mg, 0.19 mmol) were added and the reaction was stirred at 85 °C for 18 h. The mixture was diluted with water (20 mL) and extracted with Et2O (3 × 20 mL). The organic phases were combined, washed with brine (20 mL), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. The crude product was purified by vacuum distillation to yield the carbonyl derivative S945 as orange oil (2.31 g, 9.62 mmol, 61%). Rf 0.44 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 4.34 (dd, J = 9.6, 4.7 Hz, 1H), 4.20 (app. qd, J = 7.2, 3.1 Hz, 2H), 3.28 (dd, J = 18.5, 9.6 Hz, 1H), 3.12 (dd, J = 18.5, 4.7 Hz, 1H), 2.20 (s, 3H), 1.25 (t, J = 7.2 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 204.1, 187.2 (q, J = 36.8 Hz), 166.7, 115.3 (q, J =291.3 Hz), 62.7,47.7,42.1, 29.2, 13.9ppm. 19F NMR (376 MHz, CDCl3): δ -78.0 ppm. MS (ESI): [M - H]- 239.0. IR (neat): νmax 2988–2924 (w, C–H), 1718 (s, C⩵O), 13.70 (w), 1267 (m), 1157 (s), 1096 (m), 1041 (m) cm−1. Ethyl 5-Methyl-1-phenyl-2-(trifluoromethyl)-1H-pyrrole-3-car-boxylate (S10) According to General Procedure D, aniline (37.3 μL, 0.41 mmol) was stirred for 2.5 h to give a crude product, which was purified by flash column chromatography (0-5% EtOAc in petrol) to give the ester derivative S10 as yellow oil (18.0 mg, 60.5 μMol, 15%). Rf: 0.59 (30:70EtOAc:petrol). 1H NMR(400MHz, CDCl3): δ7.51-7.44 (m, 3H), 7.25-7.20 (m, 2H), 6.47 (s, 1H), 4.33 (q, J = 7.1 Hz, 2H), 1.94 (s, 3H), 1.35 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 163.6,137.6,133.7, 129.3,129.3,127.8 (q, J = 0.7Hz), 122.2 (q, J = 38.3 Hz), 120.5 (q, J = 269.4 Hz), 118.4 (q, J = 2.1 Hz), 110.0, 60.7, 14.2, 12.5 ppm. 19F NMR (376 MHz, CDCl3): δ -54.4 ppm. MS (ESI): [M + H] +298.1. IR(neat): νmax 2981–2853 (w, C–H), 1721 (s, C⩵O), 1511 (m), 1495 (m), 1417 (m), 1276 (m), 1219 (s), 1174 (s), 1114 (s), 1039 (s), 996 (s) cm−1. 5-Methyl-1-phenyl-2-(trifluoromethyl)-1H-pyrrole-3-carboxylic acid (22) According to General Procedure B, the ester derivative S10 (8 mg, 26.9 μMol) was stirred for 4 h to give a crude product. The crude product was purified by flash column chromatography (30% EtOAc in petrol) to give the acid derivative 22 as a colorless amorphous solid (6 mg, 20.4 μMol, 76%). Rf. 0.34 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.55-7.49 (m, 3H), 7.31-7.24 (m, 2H), 6.60 (s, 1H), 1.99 (s,3H) ppm. 13C NMR (100MHz, CDCl3): δ 168.6, 137.6, 133.8 (q, J = 1.3 Hz), 129.4, 129.3, 127.7 (q, J = 0.7 Hz), 123.4 (q, J = 38.5 Hz), 120.3 (q, J = 269.6 Hz), 116.9 (q, J = 2.0 Hz), 110.9, 12.6 ppm. 19F NMR (376 MHz, CDCl3): δ -53.5 ppm. MS (ESI): [M + H]+ 270.0. HPLC: retention time: 1.88 min (>99%). IR (neat): νmax 3100–2600 (s, O–H, C–H), 1665 (s, C⩵O), 1517 (m), 1497 (m), 1456 (m), 1418 (m), 1255 (s), 1138 (s), 1000 (m) cm−1. HRMS: calculated for C13H10F3NO2 [M + H]+ = 270.0742, observed = 270.0735. 4-(Benzyloxy)butan-2-one (S11) Benzyl bromide (4.12 mL, 34.7 mmol) was added to a mixture of 4-hydroxy-2-butanone (2.00 mL, 23.15 mmol) and N,N-diisopropylethylamine (6.4 mL, 37.0 mmol) at room temperature. The reaction was stirred at 150 °C for 2 h and allowed to cool to room temperature. EtOAc (10 mL) and an aqueous solution of 10% sodium bisulfate (10 mL) were added, the phases were separated, and the aqueous phase was extracted with EtOAc (3 × 2 mL). The organic phases were combined, dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to yield a crude product. The crude product was purified by flash column chromatography (10% EtOAc in petrol) to give the benzyl derivative S1146 as yellow liquid (2.60 g, 14.6 mmol, 63%). Rf·. 0.67 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.42-7.20 (m, 5H), 4.51 (s, 2H), 3.74 (t, J =6.3 Hz, 2H), 2.71 (t, J = 6.3 Hz, 2H), 2.18 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 207.2,138.1,128.5,127.8,127.7,73.3,65.3,43.8, 30.5 ppm. MS (ESI): [M + H]+ 179.0. IR (neat): νmax 2863 (m, C–H), 1714 (s, C⩵O), 1454 (w), 1363 (m), 1170 (w), 1104 (s), 1085 (s) cm−1. 4-(Benzyloxy)-1-bromobutan-2-one (S12) Bromine (0.51 mL, 10.0 mmol) was added dropwise to a solution of the benzyl derivative S11 (1.78 g, 10.0 mmol) in methanol (18 mL) at0 °C, and the reaction was stirred at room temperature for 18 h. An aqueous solution of 1.0 M K2CO3 (20 mL) and Et2O (20 mL) were added. The phases were separated, and the aqueous phase was extracted with Et2O (2 × 20 mL). The organic phases were combined, dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. The crude product was dissolved in THF (72 mL), and an aqueous solution of 1.0 M H2SO4 (36.0 mL) was added. After stirring for 2 h at 65 °C and concentrating the mixture in vacuo, Et2O (20 mL) and water (20 mL) were added and the phases were separated. The organic phase was washed with an aqueous solution of 2.0 M KHCO3 (10 mL), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. The crude product was purified by flash column chromatography (6% EtOAc in petrol) to give the bromine derivative S1247 as pale yellow oil (1.30 g, 5.05 mmol, 51%). Rf: 0.44 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.40-7.25 (m, 5H), 4.51 (s, 2H), 3.94 (s, 2H), 3.77 (t, J = 6.1 Hz, 2H), 2.92 (t, J = 6.1 Hz, 2H) ppm. 13C NMR (100 MHz, CDCl3): δ 200.4, 137.9, 128.5, 127.9, 127.8, 73.4, 65.4, 40.3, 35.0 ppm. MS (ESI): [M + Na]+ 279.2. IR (neat): νmax 3087–2865 (m, C–H), 1715 (s, C⩵O), 1495 (w), 1054 (w), 1390 (m), 1365 (m), 1326 (w), 1255 (w), 1205 (w), 1178 (w), 1095 (s), 1075 (s), 1026 (m) cm−1. Ethyl 2-Acetyl-6-(benzyloxy)-4-oxohexanoate (S13) Ethyl acetoacetate (0.12 mL, 0.97 mmol) was added dropwise to a suspension of 60% NaH in mineral oil (39 mg, 0.97 mmol) in 1,2-dimethoxyethane (1 mL) at 0 °C. After stirring for 10 min at 0 °C, the halogen derivative S12 (0.25 g, 0.97 mmol) was added and the reaction was stirred at room temperature for 18 h. Successively, an aqueous solution of 1.0 M HCl (1 mL) and EtOAc (1 mL) were added, the phases were separated and the aqueous phase was extracted with EtOAc (3 × 1 mL). The organic phases were combined, dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. The crude product was purified by flash column chromatography (10-20% EtOAc in petrol) to give the benzyl derivative S1348 as colorless liquid (0.28 g, 0.91 mmol, 94%). Rf 0.50 (40:60 EtOAcpetrol). 1H NMR (400 MHz, CDCl3): δ 7.40-7.21 (m, 5H), 4.49 (s, 2H), 4.18 (q, J = 7.1 Hz, 2H), 4.02 (app. t, J =6.9 Hz, 1H), 3.79-3.66 (m, 2H), 3.15 (dd, J = 18.5, 8.2 Hz, 1H), 2.96 (dd, J = 18.5, 5.7 Hz, 1H), 2.74 (t, J = 6.2 Hz, 2H), 2.34 (s, 3H), 1.26 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 206.3, 202.2, 168.8, 138.1, 128.5, 127.8, 127.7, 73.3, 65.1, 61.8, 53.7, 42.9, 41.3, 30.1, 14.1 ppm. MS (ESI): [M + Na]+ 329.2. IR (neat): νmax 2981–2869 (m, C–H), 1739 (m, C⩵O), 1712 (s, C⩵O), 1454 (w), 1399 (w), 1360 (m), 1256 (m), 1205 (m), 1147 (m), 1097 (m), 1022 (m) cm−1. Ethyl 5-[2-(Benzyloxy)ethyl]-2-methyl-1-phenyl-1H-pyrrole-3-carboxylate (S14) According to General Procedure C, aniline (63 μL, 0.68 mmol) and the carbonyl derivative S13 were stirred for 18 h to give a crude product, which was purified by flash column chromatography (5% EtOAc in petrol) to give the ester derivative S14 as a white solid (0.16 g, 0.44 mmol, 65%). Rf 0.48 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.50-7.40 (m, 3H), 7.36-7.22 (m, 5H), 7.17-7.10 (m, 2H), 6.45 (s, 1H), 4.44 (s, 2H), 4.29 (q, J = 7.1 Hz, 2H), 3.54 (t, J = 7.3 Hz, 2H), 2.64 (t, J = 7.3 Hz, 2H), 2.27 (s, 3H), 1.36 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3). δ 165.7, 138.3, 137.5, 136.5, 130.0, 129.5, 128.8, 128.5, 128.4, 127.8, 127.7, 111.8, 107.6, 73.0, 69.1, 59.4, 27.3, 14.7, 12.3 ppm. MS (ESI): [M + H] + 364.3. IR(neat): νmax 2981-2854 (m, C–H), 1685 (s, C⩵O), 1526 (w), 1494 (m), 1430 (m), 1379 (m), 1365 (m), 1221 (s), 1119 (m), 1084 (s), 1072 (s), 1024 (m), 1000 (m) cm−1. Ethyl 5-(2-Hydroxyethyl)-2-methyl-1-phenyl-1H-pyrrole-3-carboxylate (S15) According to General Procedure A, the benzyl derivative S14 (0.13 g, 0.36 mmol) gave a crude product. The crude product was purified using flash column chromatography (30% EtOAc in petrol) to give the alcohol derivative S15 as colorless oil (86.0 mg, 0.31 mmol, 87%). Rf: 0.31 (50:50 EtOAc.petrol). 1H NMR (400 MHz, CDCl3). δ 7.52-7.39 (m, 3H), 7.16 (d, J = 7.2 Hz, 2H), 6.46 (s, 1H), 4.26 (q, J = 7.1 Hz, 2H), 3.60 (t, J = 6.7 Hz, 2H), 2.56 (t, J = 6.7 Hz, 2H), 2.25 (s, 3H), 2.04 (br. s, 1H), 1.33 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 165.7, 137.4, 136.9, 129.6, 129.5, 128.8, 128.4, 111.8, 107.8, 61.2, 59.4, 30.0, 14.6, 12.3 ppm. MS (ESl): [M + H]+ 274.2. IR (neat): νmax 3447 (s, O–H), 2978-2875 (m, C–H), 1693 (s, C⩵O), 1674 (s), 1597 (w), 1572 (w), 1529 (m), 1498 (m), 1418 (m), 1378 (m), 1352 (w), 1219 (s), 1079 (s), 1047 (m), 1012 (m) cm−1. 5-(2-Hydroxyethyl)-2-methyl-1-phenyl-1H-pyrrole-3-carboxylic acid (23) According to General Procedure B, the ester derivative S15 (56 mg, 0.20 mmol) was stirred for 5 h to give a crude product. The crude product was purified by flash column chromatography (50% EtOAc in petrol) to give the acid derivative 23 as a white amorphous solid (36 mg, 0.15 mmol, 75%). Rf. 0.50 (EtOAc). 1H NMR (400MHz, CD3OD): δ 7.60-7.46 (m, 3H), 7.25 (d, J =7.7 Hz, 2H), 6.42 (s, 1H), 3.55 (t, J = 7.2 Hz, 2H), 2.54 (t, J = 7.2 Hz, 2H), 2.23 (s, 3H) ppm. 13C NMR (100 MHz, CD3OD): δ 169.4, 138.8, 137.8, 131.3, 130.6, 130.0, 129.6, 112.6, 109.2, 61.9, 31.0, 12.5 ppm. MS (ESI): [M + H]+ 246.1. HPLC: retention time: 1.50 min (>99%). IR (neat): νmax 3326 (m, OH), 2958–2850 (m, C–H), 1673 (s, C⩵O), 1568 (m), 1529 (m), 1492 (m), 1430 (m), 1375 (m), 1357 (m), 1217 (s), 1054 (s), 1025 (s) cm−1. HRMS: calculated for C14H15NO3 [M + H]+ = 246.1130, observed = 246.1124. Ethyl 1-Benzyl-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S16) According to General Procedure E, benzylhydrazine dihydrochloride (0.50 g, 2.56 mmol) was stirred for 3.5 h to give a crude product. The crude product was purified by flash column chromatography (5% EtOAc in petrol) to give the ester derivative S16 as colorless oil (0.48 g, 1.60 mmol, 63%). Rf: 0.56 (20:80 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.01 (s, 1H), 7.40–7.10 (m, 5H), 5.54 (s, 2H), 4.32 (q, J = 7.1 Hz, 2H), 1.35 (t, J = 7.1 Hz, 3H) ppm. 13C NMR(100 MHz, CDCl3): δ 161.0, 146.9, 142.0, 135.3, 131.9 (q, J = 40.1 Hz), 128.9, 128.4, 127.2, 119.6 (q, J = 271.0 Hz), 61.2, 56.7 (q, J = 3.2 Hz), 14.2 ppm. 19F NMR (376 MHz, CDCl3): δ -57.4 ppm. MS (ESI): [M + H]+ 299.2. IR (neat): νmax 2982 (w, C–H), 1735 (s, C⩵O), 1559 (m), 1477 (m), 1294 (s), 1222 (s), 1187 (s), 1152 (s), 1042 (s) cm−1. 1-Benzyl-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (24) According to General Procedure B, the ester derivative S16 (0.20 g, 0.67 mmol) was stirred for 5.5 h to give the acid derivative 24 as a white amorphous solid (0.16 g, 0.60 mmol, 90%). Rf: 0.18 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.08 (s, 1H), 7.407.10 (m, 5H), 5.55 (s, 2H) ppm. 13C NMR (100 MHz, CDCl3): δ 166.5, 146.0, 142.8, 135.1, 132.7 (q, J = 40.0 Hz), 128.9, 128.5, 127.2, 119.4 (q, J = 271.7 Hz), 56.8 (q, J = 3.2 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ - 57.4 ppm. MS (ESI): [M - H]- 269.0. HPLC: retention time: 1.79 min (>99%). IR (neat): νmax 3040 (m, O–H), 2928-2525 (w, C–H), 1700 (m, C⩵O), 1562 (m), 1480 (m), 1410 (m), 1300 (s), 1231 (s), 1174 (s), 1136 (s), 1037 (s), 1014 (s) cm−1. HRMS: calculated for C12H9F3N2O2 [M + H]+ = 271.0694, observed = 271.0682. [(Thiophen-2-yl)methyl]hydrazine Dihydrochloride (S17) tert-Butyl carbazate (1.50 g, 11.3 mmol) was added to a solution of 2-thiophenecarboxaldehyde (1.00 mL, 10.8 mmol) in MeOH (25 mL) at room temperature. After stirring the mixture at 65 °C for 1 h, the solvent was removed in vacuo, the crude product was dissolved in THF (45 mL), and sodium cyanoborohydrate (1.00 g, 16.2 mmol) was added. Subsequently, AcOH (17.0 mL) was added dropwise and the mixture was stirred at room temperature for 24 h. Successively, a saturated aqueous solution of NaHCO3 (20 mL) was slowly added and the mixture was extracted with EtOAc (2 × 10 mL). The organic phases were combined, washed with brine (10 mL), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. EtOH (50 mL) and an aqueous solution of concentrated HCl (5 mL) were added. After stirring the mixture at 80 °C for 18 h, the solvent was removed in vacuo to yield the hydrazine derivative S1749 as a white amorphous solid (1.90 g, 9.44 mmol, 87%). Rf: 0.65 (20:80 MeOH:DCM). 1H NMR (400 MHz, CD3OD) δ 7.49 (dd, J = 5.1, 1.1 Hz, 1H), 7.20 (dd, J = 3.5, 1.1 Hz, 1H), 7.06 (dd, J = 5.1, 3.5 Hz), 4.37 (s, 2H) ppm. 13C NMR (100 MHz, CD3OD): δ 135.5, 130.4, 128.4, 128.3, 49.9 ppm. MS (ESI): [M + H] + 128.9. IR (neat): νmax 3205 (m, N-H), 2990-2862 (s, N-H, C–H), 1582 (m), 1501 (m), 1375 (m), 1243 (w), 1047 (w), 1022 (w) cm−1. Ethyl 1-[(Thiophen-2-yl)methyl]-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S18) According to General Procedure E, the hydrazine derivative S17 (0.25 g, 1.24 mmol) was stirred for 18 h to give a crude product. The crude product was purified by flash column chromatography (5-30% EtOAc in petrol) to give the ester derivative S18 as pale brown oil (0.20 g, 0.66 mmol, 53%). Rf: 0.56 (20:80 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.98 (s, 1H), 7.27 (dd, J = 5.1, 1.2 Hz, 1H), 7.04 (d, J = 3.5 Hz, 1H), 6.95 (dd, J = 5.1, 3.5 Hz, 1H), 5.67 (s, 2H), 4.31 (q, J = 7.1 Hz, 2H), 1.34 (t, J = 7.1 Hz, 3H) ppm. 13CNMR(100 MHz, CDCl3): δ 160.9, 142.2, 136.8, 131.4 (q, J = 40.3 Hz), 127.6, 127.0, 126.8, 119.6 (q, J = 271.1 Hz), 116.4 (q, J = 1.6 Hz), 61.2, 51.4 (q, J = 3.5 Hz), 14.2 ppm. 19F NMR (376 MHz, CDCl3): δ -57.1 ppm. MS (ESI): [M + H]+ 305.1. IR (neat): νmax 2983 (w, C–H), 1732 (s, C⩵O), 1558 (w), 1476 (w), 1408 (w), 1373 (w), 1292 (s), 1218 (s), 1188 (s), 1148 (s), 1039 (s), 1021 (s) cm−1. 1-[(Thiophen-2-yl)methyl]-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (25) According to General Procedure B, the ester derivative S18 (0.10 g, 0.33 mmol) was stirred for 6.5 h to give the acid derivative 25 as a white amorphous solid (72.0 mg, 0.26 mmol, 79%). Rf: 0.29 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 11.02 (br. s, 1H), 8.09 (s, 1H), 7.29 (dd, J = 5.2, 1.2 Hz, 1H), 7.07 (d, J = 3.5 Hz, 1H), 6.97 (dd, J = 5.2, 3.5 Hz, 1H), 5.71 (s, 2H) ppm. 13C NMR (100 MHz, CDCl3): δ 166.2, 143.1, 136.4, 132.4 (q, J = 41.1 Hz), 127.8, 127.1, 127.0, 119.4 (q, J = 271.7 Hz), 115.1 (q, J = 1.4Hz), 51.6 (q, J = 3.6 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ -57.2 ppm. MS (ESI): [M - H] - 275.0. HPLC: retention time: 1.65 2864-2557 (m, O–H, C–H), 1687 (s, C⩵O), 1565 (m), 1479 (m), 1422 (m), 1334 (m), 1302 (s), 1239 (s), 1129 (w), 1029 (s), 1009 (s) cm−1. HRMS: calculated for C10H7F3N2O2S [M - H]- = 275.0102, observed = 275.0104. N-(Thiophen-2-yl)(tert-butoxy)carbohydrazide (S19) tert-Butyl carbazate (3.50 g, 26.8 mmol), Cs2CO3 (6.90 g, 21.2 mmol), CuI (0.25 g, 1.40 mmol), and trans-4-hydroxy-L-proline (0.35 g, 2.67 mmol) were added to a solution of 2-bromothiophene (1.00 mL, 10.3 mmol) in DMSO (50 mL), and the reaction was stirred at 80 °C for 18 h. After allowing the mixture to cool to room temperature, water (40 mL) and EtOAc (10 mL) were added. The phases were separated, and the aqueous phase was extracted with EtOAc (2 × 10 mL). The organic phases were combined, washed with brine (10 mL), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo. The crude product was purified by flash column chromatography (10% EtOAc in petrol) to give the carbamate derivative S1950 as pale brown oil (0.52 g, 2.42 mmol, 24%). Rf: 0.60 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 6.88 (br. s, 1H), 6.84-6.78 (m, 2H), 4.56 (s, 2H), 1.56 (s, 9H) ppm. 13C NMR (100 MHz, CDCl3): δ 153.2, 146.8, 125.3, 117.3, 112.6, 83.2, 28.3 ppm. MS (ESI): [M + H] + 158.9 (tBu lost during MS). IR (neat): vmax 3324 (m, N–H), 3273 (w, N–H), 3201 (w, N–H), 2977–2930 (m, C–H), 1692 (s, C⩵O), 1621 (w), 1534 (m), 1473 (w), 1447 (m), 1368 (s), 1322 (s), 1280 (m), 1251 (m), 1224 (m), 1150 (s), 1082 (m), 1050 (m), 998 (s) cm–1. Ethyl 1-(Thiophen-2-yl)-5-(trifluoromethyl)-1 H-pyrazole-4-carboxylate (S20) A solution of 4.0 M HCl in 1,4-dioxane (1.00 mL) was added to a solution of the derived carbamate S19 (0.10 g, 0.47 mmol) in DCM (1 mL). The reaction was stirred at room temperature for 3 days, and the solvent was removed in vacuo to yield a crude product. The crude product was dissolved in EtOH (1.00 mL), and ethyl-3-ethoxy-2-(2,2,2-trifluoroacetyl)acrylate (0.10 mL, 0.53 mmol) and Et3N (79 μL, 0.57 mmol) were added. The mixture was stirred at 80 °C for 5 h. After allowing the reaction to cool to room temperature, the solvent was evaporated in vacuo and EtOAc (1 mL) and water (1 mL) were added. The phases were separated, and the aqueous phase was extracted with EtOAc (3 × (1 mL)). The organic phases were combined, washed with brine (1 mL), dried under anhydrous magnesium sulfate, filtered, and concentrated in vacuo to give a crude product. The crude product was purified by flash column chromatography (5% EtOAc in petrol) to give the ester derivative S20 as yellow oil (75 mg, 0.26 mmol, 55%). Rf: 0.62 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.10 (s, 1H), 7.37 (dd, J = 5.6, 1.4 Hz, 1H), 7.17 (dd, J = 3.8,1.4 Hz, 1H), 7.01 (dd, J = 5.6, 3.8 Hz, 1H), 4.37 (q, J = 7.1 Hz, 2H), 1.37 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.7, 142.8, 139.0, 133.9 (q, J = 40.1 Hz), 126.0, 125.6, 125.5 (q, J = 1.5 Hz), 118.9 (q, J = 271.7 Hz), 117.1 (q, J =1.3 Hz),61.5, 14.2ppm. 19F NMR (376 MHz, CDCl3): δ –57.3 ppm. MS (ESI): [M + H] + 291.1. IR (neat): νmax 3109–2907 (w, C–H), 1734 (s, C⩵O), 1566 (w), 1554 (w), 1466 (w), 1397 (w), 1377 (w), 1291 (s), 1230 (s),1185 (s), 1140 (s), 1035 (s) cm−1. 1-(Thiophen-2-yl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (26) According to General Procedure B, the ester derivative S20 (60 mg, 0.21 mmol) was stirred for 5.5 h to give the acid derivative 26 as a white amorphous solid (50 mg, 0.19 mmol, 90%). Rf: 0.33 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 11.34 (br. s, 1H), 8.21 (s, 1H), 7.40 (dd, J = 5.6, 1.4 Hz, 1H), 7.21 (dd, J = 3.8, 1.4 Hz, 1H), 7.04 (dd, J = 5.6, 3.8 Hz) ppm. 13C NMR (100 MHz, CDCl3): δ 166.1, 143.6, 138.7, 134.9 (q, J = 40.7Hz), 126.3, 125.8 (q, J = 1.5 Hz), 125.7, 118.7 (q, J = 272.1 Hz), 115.8 (q, J = 1.3 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ –56.3 ppm. MS (ESI): [M – H]– 261.0. HPLC: retention time: 1.64 min (>99%). IR (neat): νmax 2856–2583 (m, O–H, C–H), 1702 (m, C⩵O), 1568 (w), 1545 (w), 1418 (w), 1299 (m), 1254 (m), 1235 (m), 1187 (m), 1134 (s), 1026 (m) cm–1. HRMS: calculated for C9H5F3N2O2S [M + H]+ = 263.0102, observed = 263.0110. Ethyl 1-(Pyridin-3-yl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxy-late (S21) According to General Procedure E, 3-hydrazinopyridine dihydrochloride (0.30 g, 1.64 mmol) was stirred for 18 h to give a crude product. The crude product was purified by flash column chromatography (20–40% EtOAc in petrol) to give the ester derivative S2151 as yellow oil (86 mg, 0.30 mmol, 18%). Rf: 0.40 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.75 (app. d, J = 4.9 Hz, 1H), 8.71 (s, 1H), 8.14 (s, 1H), 7.77 (app. dt, J =8.2, 1.9 Hz, 1H), 7.46 (dd, J = 8.2, 4.9 Hz, 1H), 4.36 (q, J = 7.1 Hz, 2H), 1.36 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.7, 150.9, 146.8 (q, J =1.1 Hz), 143.2, 136.2, 133.3 (q, J = 0.9 Hz), 133.1 (q, J = 40.1 Hz), 123.7, 119.0 (q, J =271.5 Hz), 117.5 (q, J = 1.5 Hz), 61.5, 14.1 ppm. 19F NMR (376 MHz, CDCl3): δ –56.1 ppm. MS (ESI): [M + H]+ 286.2. IR (neat): νmax 2988 (w, C–H), 1710 (m, C⩵O), 1554 (w), 1491 (w), 1465 (w), 1411 (w), 1384 (w), 1295 (w), 1245 (m), 1223 (m), 1190 (s), 1148 (m), 1082 (s), 1027 (m) cm–1. 1-(Pyridin-3-yl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (27) According to General Procedure B, the ester derivative S21 (66 mg, 0.23 mmol) was stirred for 18 h to give the acid derivative 2T47 as a white amorphous solid (57 mg, 0.22 mmol, 96%). Rf: 0.16 (20:80 MeOH:DCM). 1H NMR (400 MHz, CD3OD): δ 8.75 (app. d, J = 4.9 Hz, 1H), 8.72 (d, J =2.5 Hz, 1H), 8.21 (s, 1H), 8.02 (app. dt, J = 8.3,2.0 Hz, 1H), 7.65 (dd, J 8.3, 4.9 Hz, 1H) ppm. 13C NMR (100 MHz, CDCl3): δ 163.3,151.7,147.6 (q, J = 1.1 Hz), 144.5, 138.1, 135.7 (q, J = 1.0 Hz), 134.0, 125.5, 120.5 (q, J =270.6 Hz), 119.1 (q, J = 1.4 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ – 57.5 ppm. MS (ESI): [M + H]+ 258.1. HPLC: retention time: 1.36 min (>99%). IR (neat): νmax 3118–2413 (m, O–H, C–H), 1878 (w), 1718 (m, C⩵O), 1562 (m), 1489 (w), 1433 (m), 1378 (w), 1366 (w), 1299 (m), 1256 (m), 1225 (m), 1184 (s), 1144 (s), 1082 (m), 1045 (s), 1027 (s) cm–1. Ethyl l-(Pyridin-4-yl)-5-(trifluoromethyl)-lH-pyrazole-4-carboxylate (S22) According to General Procedure E, 4-hydrazinopyridine hydrochloride (20 mg, 0.14 mmol) was stirred overnight to give a crude product. The crude product was purified by flash column chromatography (30% EtOAc in petrol) to give the ester derivative S2251 as colorless oil (18 mg, 63.1 μMol, 45%). Rf: 0.38 (60:40 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.80 (d, J = 5.3 Hz, 2H), 8.15 (s, 1H), 7.43 (d, J = 5.3 Hz, 2H), 4.38 (q, J = 7.1 Hz, 2H), 1.38 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.6, 151.1, 146.4, 143.4, 132.6 (q, J = 40.7 Hz), 120.0, 119.0 (q, J = 271.6 Hz), 118.2 (q, J = 1.4 Hz), 61.7, 14.2 ppm. 19F NMR (376 MHz, CDCl3): δ – 55.9 ppm. MS (ESI): [M + H]+286.1. IR (neat): νmax 1737 (s, C⩵O), 1590 (s), 1501 (w), 1299 (m), 1245 (s), 1148 (s), 1040 (m) cm–1. HRMS: calculated for C12H10F3N3O2 [M + H]+ = 286.0803, observed = 286.0800. l-(Pyridin-4-yl)-5-(trifluoromethyl)-lH-pyrazole-4-carboxylic Acid (28) According to General Procedure B, the ester derivative S22 (18 mg, 63.1 μMol) was stirred for 2 h to give the acid derivative 2851 as a white amorphous solid (15 mg, 59.3 μMol, 94%). Rf 0.13 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 13.49 (br. s, 1H), 8.82 (d, J = 6.2 Hz, 2H), 8.32 (s, 1H), 7.64 (d, J = 6.2 Hz, 2H) ppm. 13C NMR (100 MHz, DMSO-d6): δ 162.1, 151.6, 146.4, 143.7, 132.0, 131.7, 121.0, 118.4 (q, J = 1.4 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ –54.4 ppm. MS (ESI): [M + H]+ 258.1. HPLC: retention time: 1.34 min (>99%). IR (neat): νmax 2447 (w, O–H), 1690 (s, C⩵O), 1601 (m), 1555 (m), 1400 (m), 1239 (m), 1209 (m), 1187 (m), 1126 (s), 1029 (m), 970 (m). HRMS: calculated for C10H6F3N3O2 [M + H]+ = 258.0490, observed = 258.0486. Ethyl 1-(2-Fluorophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S23) According to General Procedure E, 2-fluorophenylhy-drazine hydrochloride (0.34 g, 2.10 mmol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (30% EtOAc in petrol) to give the ester derivative S2353 as a white amorphous solid (0.46 g, 1.53 mmol, 73%). Rf: 0.56 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.20 (s, 1H), 7.56–7.51 (m, 1H), 7.48 (td, J = 7.6, 1.7 Hz, 1H), 7.31 (tt, J = 7.6, 1.1 Hz, 1H), 7.28 (dt, J =8.4, 1.1 Hz, 1H), 4.40 (q, J = 7.2 Hz, 2H), 1.40 (t, J = 7.2 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 159.9, 155.9 (d, J = 253.4 Hz), 133.3 (q, J = 40.4Hz), 131.1 (d, J = 7.9 Hz), 127.6, 126.5 (d, J =12.7 Hz), 123.8 (d, J = 4.0 Hz), 118.0 (q, J = 271.7 Hz), 115.7, 115.5 ppm. 19F NMR (376 MHz, DMSO-d6): δ – 54.2, –60.4 ppm. MS (ESI): [M + H]+ 303.1. IR (neat): νmax 1733 (s, C⩵O), 1599 (w), 1567 (m), 1512 (s), 1302 (s), 1246 (s), 1147 (s), 1039 (s) cm–1. HRMS: calcd for C13H10F4N2O2 [M + H]+ = 303.0756, observed = 303.0751. 1-(2-Fluorophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (29) According to General Procedure B, the ester derivative S23 (0.43 g, 1.41 mmol) was stirred for 2 h to give the acid derivative 2953 as a white amorphous solid (0.36 g, 1.30 mmol, 92%). Rf: 0.32 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 8.25 (s, 1H), 7.58–7.50 (m, 1H), 7.47 (dd, J = 7.8, 1.7 Hz, 1H), 7.30 (t, J = 7.8 Hz, 1H), 7.25 (t, J = 5.3 Hz, 1H) ppm. 13C NMR (100 MHz, CDCl3): δ 165.6, 156.8 (d, J = 253.8 Hz), 143.9,135.1 (q, J = 40.4Hz), 132.1 (d, J = 7.8 Hz), 128.4,127.3 (d, J = 12.6 Hz), 124.7 (d, J = 4.1 Hz), 120.1 (q, J = 271.8 Hz), 116.6, 115.4 (d, J =1.3 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ –57.7, –62.2 ppm. MS (ESI): [M – H]- 273.1. HPLC: retention time: 1.76 min (>99%). IR (neat): νmax 2844 (w, O–H), 1704 (s, C⩵O), 1600 (w), 1572 (m), 1509 (m), 1419 (w), 1299 (s), 1261 (s), 1189 (m), 1135 (s), 1032 (s), 970 (m) cm−1. Ethyl 1-(2-Chlorophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (57) According to General Procedure E, 2-chlorophenylhy-razine hydrochloride (0.11 g, 0.63 mmol) was stirred for 2 h to give a crude product. The crude product was purified using flash column chromatography (5% EtOAc in petrol) to give the ester derivative 5753 9 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.18 (s, 1H), 7.55 (d, J = 8.0 Hz, 1H), 7.51-7.45 (m, 1H), 7.44-7.39 (m, 2H), 4.38 (q, J = 7.1 Hz, 2H), 1.38 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.8, 143.1, 137.3, 134.1 (q, J = 40.1 Hz), 132.2, 131.6, 130.4, 128.8 (q, J = 0.6 Hz), 127.6, 119.0 (q, J = 271.6 Hz), 116.3 (q, J = 1.4 Hz), 61.4.14.2 ppm. 19F NMR (376 MHz, CDCl3): δ -58.2 ppm. MS (ESI): [M + H]+ 319.1. HPLC: retention time: 2.20 min (>99%). IR (neat): νmax 3071-2850 (w, C–H), 1718 (s, C⩵O), 1562 (m), 1497 (m), 1444 (w), 1386 (w), 1299 (m), 1245 (s), 1221 (m), 1147 (s), 1097 (m), 1049 (m), 1017 (m) cm−1. 1-(2-Chlorophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (30) According to General Procedure B, the ester derivative 57 (45 mg, 0.14 mmol) was stirred for 1 h to give the acid derivative 3053 as a yellow amorphous solid (41 mg, 0.14 mmol, >99%). Rf. 0.32 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 11.62 (br. s, 1H), 8.28 (s, 1H), 7.57 (d, J = 8.0 Hz, 1H), 7.51 (td, J = 8.0, 7.0, 2.8 Hz, 1H),7.48-7.40 (m, 2H) ppm. 13C NMR (125 MHz, CD3OD): δ 163.4, 144.2, 138.5, 135.1 (q, J = 40.0 Hz), 133.1, 133.0, 131.2, 130.2, 129.0, 120.4 (q, J = 270.8 Hz), 117.9 (q, J = 1.5 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ -58.3 ppm. MS (ESI): [M + H]+ 291.1. HPLC: retention time: 1.91 min (>99%). IR (neat): νmax 3000-2500 (w, O–H, C–H), 1702 (m, C⩵O), 1571 (m), 1495 (m), 1461 (w), 1424 (w), 1303 (m), 1268 (m), 1182 (s), 1143 (s), 1101 (m), 1050 (m), 1034 (s) cm−1. Ethyl 1-(2-Bromophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-car-boxylate (S24) According to General Procedure E, 2-bromophenylhydrazine hydrochloride (0.37 g, 1.66 mmol) was stirred for 2 h to give a crude product. The crude product was purified using flash column chromatography (0-10% EtOAc in petrol) to give the ester derivative S24 as orange oil (65 mg, 0.18 mmol, 11%). Rf 0.28 (25:75 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.18 (s, 1H), 7.73 (dd, J =8.0, 1.6 Hz, 1H), 7.50-7.38 (m, 3H), 4.38 (q, J =7.1 Hz, 2H), 1.39 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.9, 143.1, 138.9, 134.0 (q, J = 40.0 Hz), 133.5, 131.8, 128.9 (q, J = 0.7 Hz), 128.2, 121.9 (q, J = 0.8 Hz), 119.0 (q, J = 271.7 Hz), 116.3 (q, J = 1.5 Hz), 61.4.12.2 ppm. 19FNMR(376MHz, CDCl3): δ -58.0 ppm. MS (ESi): [M + H]+ 365.1. IR (neat): νmax 3070-2908 (w, C–H), 1716 (s, C⩵O), 1561 (m), 1494 (m), 1385 (w), 1298 (m), 1244 (s), 1221 (s), 1146 (s), 1092 (s), 1040 (s), 970 (s) cm−1. 1-(2-Bromophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (31) According to General Procedure B, the ester derivative S24 (44 mg, 0.12 mmol) was stirred for 1 h to give the acid derivative 3154 as a yellow amorphous solid (37 mg, 0.11 mmol, 92%). Rf. 0.30 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 8.29 (s, 1H), 7.77 (m, 1H), 7.55-7.40 (m, 3H) ppm. 13C NMR (100 MHz, DMSO-d6): δ 162.1, 143.4, 138.8, 133.6, 132.9 (q, J = 39.5 Hz), 132.9, 129.9, 129.3, 121.3, 119.3 (q, J = 270.9 Hz), 117.3 (q, J = 2.9 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ -57.2 ppm. MS (ESI): [M - H]- 334.9. HPLC: retention time: 1.80 min (95%). IR (neat): νmax 2852 (w, O–H), 1702 (s, C⩵O), 1571 (m), 1493 (m), 1464 (m), 1447 (m), 1427 (m), 1144 (s) cm−1. HRMS: calculated for CuH6BRf3N2O2 [M - H]- = 334.9643, observed = 334.9635. Ethyl 1-(o-Tolyl)-5-(trifluoromethyl)-1 H-pyrazole-4-carboxylate (S25) According to General Procedure E, o-tolylhydrazine hydrochloride (0.10 g, 0.63 mmol) was stirred for 3 h to give a crude product. The crude product was purified using flash column chromatography (0-10% EtOAc in petrol) to give the ester derivative S2553 as yellow oil (0.11 g, 0.37 mmol, 59%). Rf. 0.60 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.15 (s, 1H), 7.42 (td, J = 7.5, 1.5 Hz, 1H), 7.33 (d, J = 7.1 Hz, 1H), 7.30 (t, J = 7.4 Hz, 1H), 7.24 (d, J = 7.5 Hz, 1H), 4.38 (q, J = 7.1 Hz, 2H), 2.04 (s, 3H), 1.38 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 161.1, 142.6, 138.6, 135.4, 133.4 (q, J = 39.7 Hz), 131.0, 130.4, 127.1 (q, J = 0.9 Hz), 126.6, 119.1 (q, J = 271.5 Hz), 115.9 (q, J = 1.3 Hz), 61.3, 16.9, 14.2 ppm. 19F NMR (376 MHz, CDCl3): δ -57.8. MS (ESI): [M + H]+ 299.2. IR (neat): νmax 2984 (w, C–H), 1735 (s, C⩵O), 1561 (m), 1501 (m), 1467 (m), 1383 (m), 1298 (s), 1228 (s), 1147 (s), 1072 (m), 1040 (s) cm−1. 1-(o-Tolyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic acid (32) According to General Procedure B, the ester derivative S25 (0.19 g, 0.63 mmol) was stirred for 1 h to give the acid derivative 3253 as yellow oil (0.16 g, 0.58 mmol, 92%). Rf: 0.20 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 8.25 (s, 1H), 7.45 (app. td, J = 7.5, 1.4 Hz, 1H), 7.36 (d, J = 7.0 Hz, 1H), 7.33 (app. t, J = 7.5 Hz, 1H), 7.26 (d, J = 7.2 Hz, 1H), 2.07 (s, 3H) ppm. 13C NMR (125 MHz, CDCl3): δ 163.6, 143.8, 139.9, 136.6, 134.3, 132.0, 131.6, 128.2 (q, J = 0.9 Hz), 127.7, 120.5 (q, J = 270.8 Hz), 117.6 (q, J = 1.3 Hz), 16.8 ppm. 19F NMR (376 MHz, CDCl3): δ -57.9 ppm. MS (ESI): [M + H]+ 271.2. HPLC: retention time: 1.92 min (>99%). IR (neat): νmax 2927-2588 (m, O–H, C–H), 1701 (s, C⩵O), 1566 (m), 1500 (w), 1466 (w), 1421 (w), 1301 (m), 1255 (m), 1223 (m), 1185 (s), 1128 (s), 1068 (w), 1033 (s) cm−1. Ethyl 5-(Trifluoromethyl)-1-(2-(trifluoromethyl)phenyl)-1H-pyra-zole-4-carboxylate (S26) According to General Procedure E, 2-(trifluoromethyl)phenylhydrazine hydrochloride (0.35 g, 1.66 mmol) was stirred for 18 h to give a crude product. The crude product was purified by flash column chromatography (10-20% EtOAc in petrol) to give the ester derivative S2652 as yellow oil (49 mg, 0.14 mmol, 8%). Rf: 0.45 (30:70EtOAc:petrol). 1HNMR(400MHz, CDCl3): δ 8.19 (s, 1H), 7.91-7.82 (m, 1H), 7.79-7.67(m, 2H), 7.49-7.39 (m, 1H),4.40 (q, J = 7.1 Hz, 2H), 1.41 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (150 MHz, CDCl3): δ 160.7,142.7,136.8,134.5 (q, J = 40Hz), 132.7,130.8,129.4, 127.9 (q, J = 32Hz), 127.4(q, J =5.0Hz), 122.5 (q, J = 274Hz), 118.9 (q, J = 272 Hz), 116.4, 61.3, 14.1 ppm. 19F NMR (376 MHz, CDCl3): δ -56.5, -60.6 ppm. MS (ESI): [M + H]+ 353.2. IR (neat): νmax 1724 (s, C⩵O), 1563 (m), 1507 (m), 1319 (s), 1249 (s), 1161 (s), 968 (s) cm−1. HRMS: calculated for C14H10F6N2O2 [M + H]+ = 353.0724, observed = 353.0721. 5-(Trifluoromethyl)-1-(2-(trifluoromethyl)phenyl)-1H-pyrazole-4-carboxylic Acid (33) According to General Procedure B, the ester derivative S26 (48.0 mg, 0.14 mmol) was stirred for 2 h to give the acid derivative 3 352 as a yellow amorphous solid (29.1 mg, 90.0 μMol, 64%). Rf: 0.30 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 10.05 (br. s, 1H), 8.26 (s, 1H), 7.87 (dd, J = 6.9, 2.5 Hz, 1H), 7.77-7.70 (m, 2H), 7.45 (dd, J = 6.9,2.5 Hz, 1H) ppm. 13C NMR (150 MHz, CDCl3): δ 177.7, 165.9, 143.4, 136.5, 135.4 (q, J = 41 Hz), 132.8, 131.0, 129.3, 127.5 (q, J = 32 Hz), 127.5 (q, J =5Hz), 122.4(q, J = 274Hz),118.6(q, J = 272 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ -56.6, -60.6 ppm. MS (ESI): [M - H]- 323.0. HPLC: retention time: 1.80 min (>99%). IR (neat): νmax 2924 (w, O–H), 1703 (s, C⩵O), 1573 (m), 1320 (s), 1258 (m), 1137 (s), 1034 (m) cm−1. HRMS: calculated for C12H6F6N2O2 [M + H]+ = 325.0411, observed = 325.0400. Ethyl 5-(Trifìuoromethyl)-1-[4-(trifìuoromethyl)phenyl]-1H-pyra-zole-4-carboxylate (S27) According to General Procedure E, 4-(trifluoromethyl)phenylhydrazine hydrochloride (0.25 g, 1.17 mmol) was stirred for 4 h to give a crude product. The crude product was purified by flash column chromatography (5% EtOAc in petrol) to give the ester derivative S2752 as yellow oil (0.27 g, 0.77 mmol, 66%). Rf 0.68 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.13 (s, 1H), 7.77 (d, J = 8.3 Hz, 2H), 7.57 (d, J = 8.3 Hz, 2H), 4.37 (q, J = 7.1 Hz, 2H), 1.37 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.8, 143.0,142.2, 132.8 (q, J = 40.4 Hz), 132.1 (q, J = 33.1 Hz), 126.5 (q, J = 3.7 Hz), 126.4 (q, J = 1.1 Hz), 123.5 (q, J = 272.5 Hz), 119.1 (q, J = 271.5 Hz), 117.5 (q, J = 1.3 Hz), 61.5, 14.1 ppm. 19F NMR (376 MHz, CDCl3): δ -56.0, -63.7 ppm. MS (ESI): [M + H]+ 353.2. IR (neat): νmax 2991-2909 (w, C–H), 1737 (s, C⩵O), 1562 (w), 1466 (w), 1323 (m), 1300 (m), 1223 (s), 1175 (s), 1123 (s), 1062 (s), 1035 (s) cm−1. 5-(Trifluoromethyl)-1-[4-(trifluoromethyl)phenyl]-1H-pyrazole-4-carboxylic Acid (34) According to General Procedure B, the ester derivative S27 (0.10 g, 0.28 mmol) was stirred for 4 h to give the acid derivative 3452 as an amorphous white solid (88 mg, 0.27 mmol, 96%). Rf: 0.26 (20:80MeOH:DCM). 1HNMR(400 MHz, CDCl3): δ 8.25 (s, 1H), 7.81 (d, J = 8.3 Hz, 2H), 7.60 (d, J = 8.3 Hz, 2H) ppm. 13C NMR (100 MHz, CDCl3): δ 166.0, 143.8, 141.9, 133.8 (q, J = 40.7 Hz), 132.4 (q, J = 33.1 Hz), 126.6 (q, J = 3.7 Hz), 126.5, (q, J = 1.1 Hz), 123.5 (q, J = 272.7 Hz), 118.9 (q, J = 271.9 Hz), 116.1 (q, J = 1.4 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ -56.1, -63.7 ppm. MS (ESI): [M - H]- 323.1. HPLC: retention time: 1.99 min (>99%). IR (neat): νmax 3100–2587 (m, O–H, C–H), 1701 (m, C⩵O), 1570 (w), 1467 (w), 1422 (w), 1407 (w), 1327 (m), 1295 (m), 1259 (m), 1224 (m), 1179 (m), 1141 (s), 1112 (s), 1063 (m), 1031 (m) cm−1. Ethyl 1-(4-Methoxyphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S28) According to General Procedure E, 4-methox-yphenyl hydrazine hydrochloride (0.22 g, 1.24 mmol) was stirred for 2 h to give a crude product. The crude product was purified by flash column chromatography (10% EtOAc in petrol) to give the ester derivative S2855 as an amorphous orange solid (0.19 g, 0.60 mmol, 48%). Rf: 0.53 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CD3OD): δ 8.08 (s, 1H), 7.33 (d, J = 9.0 Hz, 2H), 6.98 (d, J = 9.0 Hz, 2H), 4.37 (q, J = 7.1 Hz, 2H), 3.87 (s, 3H), 1.38 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CD3OD): δ 161.2, 160.6, 142.3, 132.7 (q, J = 39.8 Hz), 132.4, 127.2, 119.2 (q, J = 271.4 Hz), 116.5 (q, J = 1.0 Hz), 114.3,61.3, 55.7, 14.2 ppm. 19F NMR (376 MHz, CD3OD): δ -56.5 ppm. MS (ESI): [M + H]+ 315.1. IR (neat): νmax 3117–2843 (w, C–H), 1735 (m, C⩵O), 1565 (w), 1516 (s), 1465 (w), 1446 (w), 1372 (w), 1302 (s), 1223 (s), 1185 (m), 1129 (s), 1080 (m), 1043 (s), 1018 (s), 971 (m) cm−1. 1-(4-Methoxyphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carbox-ylic Acid (35) According to General Procedure B, the ester derivative S28 (62 mg, 0.20 mmol) was stirred for 1 h to give the acid derivative 3 555 as an amorphous white solid (50 mg, 0.18 mmol, 90%). Rf 0.30 (20:80 MeOH.DCM). 1H NMR (400 MHz, CDCl3): δ 11.43 (br. s, 1H), 8.19 (s, 1H), 7.35 (d, J = 8.5 Hz, 2H), 7.00 (d, J = 8.5 Hz, 2H), 3.87 (s, 3H) ppm. 13C NMR (400 MHz, CDCl3): δ 166.4, 160.7, 143.1, 133.6 (q, J =40.2 Hz), 132.1, 127.2, 119.0 (q, J = 271.6 Hz), 115.2 (q, J = 1.2 Hz), 114.4, 55.7 ppm. 19F NMR (376 MHz, CDCl3): δ -56.5 ppm. MS (ESI): [M - H]- 285.0. HPLC: retention time: 1.80 min (>99%). IR (neat): νmax 2968-2591 (m, O–H, C–H), 1700 (s, C⩵O), 1679 (m), 1563 (w), 1517 (s), 1466 (w), 1423 (w), 1302 (s), 1249 (s), 1175 (m), 1140 (s), 1075 (m), 1028 (s) cm−1. (2-Methyl-4-nitrophenyl)hydrazine (S29) According to a proce-dure,56 hydrazine monohydrate (2.00 mL, 40.6 mmol) was added to a solution of 1-fluoro-2-methyl-4-nitrobenzene (3.00 g, 19.3 mmol) in isopropyl alcohol (30.0 mL). After stirring the mixture at 90 °C for 2 h, additional hydrazine monohydrate (2.00 mL, 40.6 mmol) was added and the mixture was kept at 90 °C for a further 2 h. The mixture was allowed to cool to room temperature, and diethyl ether (25 mL) was added. The precipitate was collected by filtration, washed with water (12 mL) and diethyl ether (12 mL), and dried in vacuo. Subsequently, a 6 M HCl aqueous solution (18 mL) was added. After stirring the mixture at room temperature for 1 h, the precipitate was collected by filtration, washed with a 6 M HCl aqueous solution, and dried in vacuo to yield the hydrochloride salt of the hydrazine derivative S2956 as a white-yellow amorphous solid (2.00 g, 9.86 mmol, 51%). Rf: 0.75 (20:80 MeOH:DCM). 1H NMR (400MHz, DMSO-d6): δ 10.56 (br. s, 2H), 8.88 (s, 1H), 8.09 (dd, J = 9.0, 2.7 Hz, 1H), 8.02 (d, J = 2.7 Hz, 1H), 7.04 (d, J = 9.0 Hz, 1H), 2.27 (s, 3H) ppm. 13C NMR (100 MHz, DMSO-d6): δ 149.4, 140.2, 125.4, 124.4, 123.0, 111.2, 17.2 ppm. MS (ESI): [M + H] + 168.0. IR (neat): νmax 3280 (s, N-H), 3074 (m, NH), 2817-2607 (m, N-H, C–H), 1593 (m), 1584 (m), 1495 (s), 1335 (s), 1298 (m), 1253 (m), 1216 (m), 1102 (m) cm−1. Ethyl 1-(2-Methyl-4-nitrophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S30) According to General Procedure E, the hydrazine derivative hydrochloride S29 (0.25 g, 1.22 mmol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (0-10% EtOAc in petrol) to give the ester derivative S30 as yellow oil (0.24 g, 0.70 mmol, 57%). Rf 0.40 (20:80 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.22 (d, J = 2.5 Hz, 1H), 8.18 (s, 1H), 8.17 (dd, J = 8.6,2.5 Hz, 1H), 7.44 (d, J =8.6 Hz, 1H),4.37(q, J = 7.1Hz, 2H),2.16(S, 3H),1.37(t, J = 7.1Hz,3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.6, 148.6, 143.4, 138.0, 133.6 (q, J = 40.0 Hz), 128.6,128.5,126.1,121.9,118.9 (q, J = 271.6 Hz), 116.8 (q, J = 1.4 Hz), 61.6, 17.3, 14.1 ppm. 19F NMR (396 MHz, CDCl3): δ - 57.5 ppm.MS(ESI):[M + H]+ 344.2. IR (neat): νmax 2984-2873 (w, C–H), 1726 (s, C⩵O), 1566 (w), 1534 (m), 1499 (w), 1354 (m), 1300 (m), 1229 (s), 1180 (s), 1139 (s), 1101 (s), 1039 (s) cm−1. 1-(2-Methyl-4-nitrophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (36) According to General Procedure B, the ester derivative S30 (0.10 g, 0.29 mmol) was stirred for 6 h to give the acid derivative 36 as a white amorphous solid (76.0 mg, 0.24 mmol, 83%). Rf: 0.33 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 8.32 (d, J = 2.5 Hz, 1H), 8.24 (s, 1H), 8.23 (dd, J =8.6, 2.5 Hz, 1H), 7.62 (d, J = 8.6 Hz, 1H), 2.16 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 163.3, 150.1, 144.6, 139.3, 134.6 (q, J = 40.1 Hz), 129.9, 126.9, 122.9, 120.4 (q, J = 270.9 Hz), 118.4 (d, J = 1.0 Hz), 17.1 ppm. 19F NMR (396 MHz, CDCl3): δ -58.8 ppm. MS (ESI): [M - H]- 314.0. HPLC: retention time: 1.84 min (>99%). IR (neat): νmax 3100-2580 (w, O–H, C–H), 1705 (m, C⩵O), 1686 (m), 1572 (m), 1527 (m), 1496 (m), 1347 (m), 1293 (m), 1263 (m), 1228 (m), 1179 (m), 1140 (s), 1098 (m), 1034 (m) cm−1. HRMS: calculated for C12H8F3N3O4 [M + H]+ = 316.0545, observed = 316.0542. 3-Hydrazinylbenzamide Hydrochloride (S31) According to General Procedure F, 3-aminobenzamide (1.00 g, 7.34 mmol) gave the hydrazine derivative S3157 as a pale brown amorphous solid (1.36 g, 7.25 mmol, 99%). Rf: 0.20 (20:80 MeOH:DCM). 1H NMR (400 MHz, DMSO-d6): δ 10.3 (br. s, 3H), 7.93 (br. s, 1H), 7.53 (s, 1H), 7.44 (app. d, J = 7.7 Hz, 1H), 7.34 (app. t, J = 8.0,1H), 7.13 (dd, J =8.0 and 2.4 Hz, 1H), 4.83 (br. s, 2H) ppm. 13C NMR (100 MHz, DMSO-d6): δ 167.7, 145.6, 135.1, 128.8, 120.1, 117.1, 113.7 ppm. MS (ESI): [M + H] + 152.0. IR (neat): νmax 3458 (w, N-H), 3358 (w, N-H), 3271 (m, NH), 3176-2700 (s, N-H, C–H), 1646 (s, C⩵O), 1561 (s), 1540 (s), 1456 (s) cm−1. Ethyl 1-(3-Carbamoylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S32) According to General Procedure E, the hydrazine derivative S31 (0.25 g, 1.33 mmol) was stirred for 6 h to give a crude product. The crude product was purified using flash column chromatography (40-60% EtOAc in petrol) to give the ester derivative S32 as a white amorphous solid (0.20 g, 0.61 mmol, 46%). Rf 0.53 (90:10 EtOAc:petrol). 1H NMR (400 MHz, DMSO-d6): δ 8.33 (s, 1H), 8.16 (br. s, 1H), 8.11 (app. dt, J =7.5, 1.6 Hz, 1H), 8.02 (s, 1H), 7.75-7.65 (m, 2H), 7.61 (br. s, 1H), 4.32 (q, J = 7.1 Hz, 2H), 1.30 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, DMSO-d6): δ 166.2, 160.2, 142.3, 138.9, 135.3, 131.4, 129.4, 129.11, 128.8, 125.1, 118.9 (q, J = 271.1 Hz), 116.3 (q, J = 1.5 Hz), 61.1, 13.9 ppm. 19F NMR (376 MHz, DMSO-d6): δ -55.5 ppm. MS (ESI): [M + H]+ 328.1. IR (neat): νmax 3442 (m, N-H), 3199 (m, N-H), 2984 (w, C–H), 1729 (s, C⩵O), 1697 (s, C⩵O), 1623 (w), 1566 (w), 1449 (w), 1374 (m), 1299 (s), 1248 (s), 1224 (s), 1190 (s), 1133 (s), 1079 (m), 1038 (s), 986 (m) cm−1. 1-(3-Carbamoylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-car-boxylic Acid (37) According to General Procedure B, the ester derivative S32 (85 mg, 0.26 mmol) was stirred for 6 h to give the acid derivative 37 as a white amorphous solid (70 mg, 0.23 mmol, 88%). Rf 0.10 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 8.22 (app. dt, J = 7.4, 1.6 Hz, 1H), 8.18 (s, 1H), 8.08 (s, 1H), 7.75-7.64 (m, 2H), 5.05 (br. s, 2H) ppm. 13C NMR (100 MHz, CDCl3): δ 167.9, 163.5, 144.0, 140.9, 133.4, 132.1, 131.4, 130.6, 128.2 (q, J = 0.8 Hz), 124.6, 120.5 (q, J = 270.7 Hz), 118.6 (q, J = 1.4 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ -56.6 ppm. MS (ESI): [M - H]- 298.0. HPLC: retention time: 1.53 min (94%). IR (neat): νmax 3100-2555 (s, N-H, O–H, C–H), 1689 (s, C⩵O), 1571 (m), 1472 (w), 1448 (w), 1418 (m), 1398 (w), 1298 (s), 1265 (s), 1220 (m), 1196 (m), 1117 (s), 1135 (s), 1125 (s), 1069 (m), 1032 (s) cm−1. HRMS: calculated for C12H8F3N3O3 [M + H]+ = 300.0596, observed = 300.0592. Ethyl 1-(2,3-Dimethylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S33) According to General Procedure E, 2,3-dimethylphenylhydrazine hydrochloride (0.25 g, 1.44 mmol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (5% EtOAc in petrol) to give the ester derivative S33 as yellow oil (0.30 g, 0.96 mmol, 67%). Rf: 0.66 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.14 (s, 1H), 7.29 (d, J = 7.7 Hz, 1H), 7.18 (t, J = 7.7 Hz, 1H), 7.08 (d, J = 7.7, 1H), 4.37 (q, J = 7.2, 2H), 2.33 (s, 3H), 1.87 (s, 3H), 1.38 (t, J = 7.2, 3H) ppm. 13C NMR(100MHz, CDCl3): δ 161.1,142.5,138.6,138.4,134.0, 133.5 (q, J = 39.9 Hz), 131.7, 125.9, 124.7, 119.1 (q, J = 271.5 Hz), 115.8 (d, J = 1.3 Hz), 61.2,20.2, 14.1, 13.8 ppm. 19F NMR (376 MHz, CDCl3): δ - 57.8 ppm. MS (ESI): [M + H]+ 313.1. IR (neat): νmax 2995-2924 (w, C–H), 1713 (s, C⩵O), 1554 (m), 1481 (m), 1384 (m), 1293 (m), 1244 (s), 1230 (s), 1176 (s), 1144 (s), 1038 (s), 1017 (m) cm−1. 1-(2,3-Dimethylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (38) According to General Procedure B, the ester derivative S33 (0.12 g, 0.38 mmol) was stirred for 4.5 h to give the acid derivative 38 as a yellow amorphous solid (95 mg, 0.33 mmol, 88%). Rf 0.22 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 11.44 (br. s, 1H), 8.25 (s, 1H), 7.33 (d, J = 7.7 Hz, 1H), 7.22 (t, J =7.7 Hz, 1H), 7.12 (d, J = 7.7 Hz, 1H), 2.36 (s, 3H), 1.91 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3). δ 166.5, 143.4, 138.6, 138.5, 134.5 (q, J = 40.2 Hz), 134.0, 131.9, 126.0, 124.7, 118.9 (q, J =271.8 Hz), 114.6 (d, J = 1.1 Hz), 20.3, 13.8 ppm. 19F NMR (376 MHz, CDCl3): δ -56.9 ppm. MS (ESI): [M + H]+ 285.1. HPLC: retention time: 1.88 min (>99%). IR (neat): 2926-2580 (m, O–H, C–H), 1698 (s, C⩵O), 1681 (s), 1564 (m), 1482 (w), 1456 (w), 1417 (w), 1299 (s), 1258 (s), 1225 (s), 1186 (s), 1138 (s), 1033 (s), 968 (m) cm−1. HRMS: calculated for C13H11F3N2O2 [M + H]+ = 285.0851, observed = 285.0851. Ethyl 1-(2,4-Dimethylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S34) According to General Procedure E, 2,4-dimethyphenylhydrazine hydrochloride (0.21 g, 1.22 mmol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (15% EtOAc in petrol) to give the ester derivative S34 as a pale red amorphous solid (0.25 g, 0.81 mmol, 66%). Rf: 0.44 (15:85 EtOAc.petrol). 1H NMR (600 MHz, CDCl3): δ 8.14 (s, 1H), 7.18-7.06 (m, 3H), 4.38 (q, J = 7.2 Hz, 2H), 2.39 (s, 3H), 2.00 (s, 3H), 1.39 (t, J = 7.2 Hz, 3H) ppm. 13C NMR (150 MHz, CDCl3): δ 161.1, 142.4, 140.4, 136.0, 134.9, 133.4 (q, J = 39.7 Hz), 131.5, 127.1, 126.7, 119.1 (q, J = 271.4Hz), 115.7, 61.2, 21.2, 16.7, 14.1 ppm. 19F NMR (376 MHz, CDCl3): δ -56.9 ppm. MS (ESI): [M + H]+ 313.2. IR (neat): 1715 (s, C⩵O), 1553 (m), 1550 (w), 1389 (m), 1255 (m), 1241 (s), 1149 (s), 1144 (s), 1041 (w), 1038 (m), 972 (m) cm−1. HRMS: calculated for CˆH15F3N2O2 [M + H]+ = 313.1163, observed = 313.1153. 1-(2,4-Dimethylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-car-boxylic Acid (39) According to General Procedure B, the ester derivative S34 (0.22 g, 0.70 mmol) was stirred for 2 h to give the acid derivative 39 as a red amorphous solid (0.19 g, 0.67 mmol, 96%). Rf. 0.22 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 11.74 (s, 1H), 8.25 (s, 1H), 7.21-7.11 (m, 3H), 2.40 (s, 3H), 2.02 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 166.4, 143.3, 140.7, 135.8, 134.8, 134.3 (q, J = 40.2 Hz), 131.6, 127.2, 126.7 (q, J = 0.7 Hz), 118.9 (q, J = 271.8 Hz), 114.6 (d, J = 1.2 Hz), 21.2, 16.7 ppm. 19F NMR (396 MHz, CDCl3): δ -57.0 ppm. MS (ESI): [M + H]+ 285.1. HPLC: retention time: 1.95 min (>99%). IR (neat): 2927 (w, O–H), 1720 (s, C⩵O), 1558 (m), 1506 (m), 1305 (s), 1233 (s), 1163 (s), 1068 (m), 1031 (s) cm−1. HRMS: calculated for C13H11F3N2O2 [M + H]+ = 285.0851, observed = 285.0841. Ethyl 1-(2,5-Dimethylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S35) According to General Procedure E, 2,5-dimethylphenylhydrazine hydrochloride (0.25 g, 1.44 mmol) was stirred for 16 h to give a crude product. The crude product was purified using flash column chromatography (5% EtOAc in petrol) to give the ester derivative S35 as a yellow amorphous solid (0.28 g, 0.90 mmol, 63%). Rf. 0.70 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.14 (s, 1H), 7.24-7.16 (m, 2H), 7.05 (s, 1H), 4.36 (q, J = 7.1 Hz, 2H), 2.34 (s, 3H), 1.98 (s, 3H), 1.37 (t, J =7.1 Hz, 3H) ppm. 13CNMR(100 MHz, CDCl3): δ 161.1, 142.5, 138.4, 136.6, 133.4 (q, J = 39.8 Hz), 132.1, 131.1, 130.7, 127.5, 119.1 (q, J = 271.4Hz), 115.8 (d, J = 1.4Hz), 61.2, 20.7, 13.4,14.2 ppm. 19F NMR (376 MHz, CDCl3): δ -57.9 ppm. MS (ESI): [M + H]+ 313.1. IR (neat): νmax 3122-2926 (w, C–H), 1717 (s, C⩵O), 1553 (m), 1468 (m), 1384 (w), 1298 (m), 1234 (s), 1174 (s), 1135 (s), 1070 (w), 1036 (m) cm−1. 1-(2,5-Dimethylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-car-boxylic Acid (40) According to General Procedure B, the ester derivative S35 (0.11 g, 0.35 mmol) was stirred for 4.5 h to give the acid derivative 40 as yellow oil (83.0 mg, 0.29 mmol, 83%). Rf: 0.37 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 8.24 (s, 1H), 7.26-7.20 (m, 2H), 7.08 (s, 1H), 2.37 (s, 3H), 2.01 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 166.5, 143.4, 138.3, 136.7, 134.4 (q, J = 40.1 Hz), 132.1, 131.4, 130.8, 127.5, 118.9 (q, J = 271.8 Hz), 114.6, 20.8, 16.5 ppm. 19F NMR (376 MHz, CDCl3). δ -57.0 ppm. MS (ESI): [M - H]- 283.0. HPLC: retention time: 1.90 2929-2600 (m, O–H, C–H), 1708 (m C⩵O), 1679 (m), 1559 (m), 1510 (w), 1458 (m), 1421 (w), 1297 (m), 1259 (s), 1232 (m), 1182 (m), 1143 (s), 1033 (m) cm 1. HRMS: calculated for C13H11F3N2O2 [M + H]+ = 285.0851, observed = 285.0855. Ethyl 1-(2,6-dimethylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-arboxylate (S36) According to General Procedure E, 2,6-dimethyl-phenylhydrazine hydrochloride (0.32 g, 1.87 mmol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (30% EtOAc in petrol) to give the ester derivative S36 as a white amorphous solid (93.7 mg, 0.30 mmol, 16%). Rf. 0.59 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.24 (s, 1H), 7.31 (t, J = 7.7 Hz, 1H), 7.17 (d, J = 7.7 Hz, 2H), 4.40 (q, J = 7.1 Hz, 2H), 1.99 (s, 6H), 1.41 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (150 MHz, CDCl3): δ 161.0, 143.1, 138.0, 135.5, 133.5 (q, J = 40.0 Hz), 130.0, 128.3, 119.1 (q, J =271.0 Hz), 115.6 (q, J = 1.5 Hz), 61.2, 16.9, 14.1 ppm. 19F NMR (376 MHz, CDCl3): δ -58.4ppm. MS (ESI): [M + H]+ 313.2. IR (neat): 1736 (s, C⩵O), 1560 (s), 1484 (s), 1297 (s), 1222 (s), 1145 (s), 1039 (s), 966 (s) cm−1. HRMS: calculated for C15H15F3N2O2 [M + H]+ = 313.1163, observed = 313.1155. 1-(2,6-Dimethylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-car-boxylic Acid (41) According to General Procedure B, the ester derivative S36 (92.0 mg, 0.29 mmol) was stirred for 2 h to give the acid derivative 41 as a white amorphous solid (59.0 mg, 0.21 mmol, 70%). Rf. 0.38 (20:80MeOH:DCM). 1HNMR(400 MHz, CDCl3): δ 8.31 (s, 1H), 7.32 (t, J = 7.6 Hz, 1H), 7.17 (d, J = 7.6 Hz, 2H), 1.99 (s, 6H) ppm. 13C NMR (150 MHz, CDCl3). δ 166.3, 144.0, 138.0, 135.5, 134.5 (q, J = 40.2 Hz), 130.2, 128.4, 118.9(q, J = 271.8 Hz), 114.5 (q, J = 1.2 Hz), 17.0 ppm. 19F NMR (376 MHz, CDCl3). δ -58.5 ppm. MS (ESI). [M + H]+ 285.2. HPLC. retention time. 1.90 min (>99%). IR (neat). 2924 (w, O–H), 1704 (s, C⩵O), 1571 (m), 1485 (m), 1303 (m), 1266 (m), 1249 (m), 1149 (s), 1035 (s), 967 (s) cm−1. HRMS. calculated for C13H11F3N2O2 [M + H]+ = 285.0851, observed = 285.0847. Ethyl 1-(2,6-Dichlorophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-arboxylate (S37) According to General Procedure E, 2,6-dichlor-ophenylhydrazine hydrochloride (0.13 g, 0.63 mmol) was stirred for 2 h to give a crude product. The crude product was purified using flash column chromatography (10% EtOAc in petrol) to give the ester derivative S37 as a pale yellow amorphous solid (0.14 g, 0.40 mmol, 63%). Rf. 0.47 (30.70 EtOAc.petrol). 1H NMR (400 MHz, CDCl3). δ 8.26 (s, 1H), 7.50-7.45 (m, 2H), 7.42 (dd, J = 9.5,6.4 Hz, 1H), 4.39 (q, J = 7.1 Hz, 2H), 1.39 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3). δ 160.7, 143.9, 135.5, 134.5 (q, J = 40.4 Hz), 134.4 (q, J = 0.7 Hz), 131.8, 128.7, 118.9 (q, J = 271.6 Hz), 116.4 (q, J = 1.6 Hz), 61.4, 14.2 ppm. 19F NMR (376 MHz, CDCl3). δ -59.6 ppm. MS (ESI). [M + H]+ 353.1. IR (neat). 3090–2942 (w, C–H), 1729 (s, C⩵O), 1571 (m), 1496 (m), 1442 (m), 1294 (m), 1240 (s), 1221 (s), 1149 (s), 1084 (m), 1040 (s) cm−1. 1-(2,6-Dichlorophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-car-boxylic Acid (42) According to General Procedure B, the ester derivative S37 (88.3 mg, 0.25 mmol) was stirred for 1 h to give the acid derivative 42 as a yellow amorphous solid (78.0 mg, 0.24 mmol, 98%). Rf. 0.25 (EtOAc). 1H NMR (400 MHz, CDCl3). δ 11.01 (br. s, 1H), 8.35 (s, 1H), 7.53-7.41 (m,3H) ppm. 13C NMR (100 MHz, CDCl3). δ 166.0, 144.6, 135.5 (q, J = 41.2 Hz), 134.3, 134.3 (q, J = 0.6 Hz), 132.0, 128.7, 118.6 (q, J = 272.0 Hz), 115.2 (q, J = 1.5 Hz) ppm. 19F NMR (376 MHz, CDCl3). δ -59.7 ppm. MS (ESI). [M - H]- 323.0. HPLC. retention time. 1.90 min (94%). IR (neat). 2873 (m, O–H), 2599–2565 (w, C–H), 1699 (s, C⩵O), 1572 (m), 1494 (m), 1444 (m), 1424 (m), 1302 (m), 1259 (m), 1223 (m), 1148 (s), 1033 (m) cm−1. HRMS. calculated for C11H5Cl2F3N2O2 [M + H]+ = 324.9758, observed = 324.9758. Ethyl 1-(3,4-Dichlorophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S38) According to General Procedure E, 3,4-dichlor-ophenylhydrazine hydrochloride (0.25 g, 1.17 mmol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (5% EtOAc in petrol) to give the ester derivative S3852 as a yellow-white amorphous solid (0.21 g, 0.59 mmol, 51%). Rf. 0.38 (10.90 EtOAc.petrol). 1H NMR (400 MHz, CDCl3). δ 8.11 (s, 1H), 7.62-7.56 (m, 2H), 7.29 (d, J = 8.7 Hz, 1H), 4.38 (q, J = 7.1 Hz, 2H), 1.38 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3). δ 160.8, 143.0, 138.4, 134.7, 133.5 (q, J = 40.3 Hz), 130.9, 128.1 (q, J = 1.1 Hz), 125.2 (q, J = 1.1 Hz), 122.2, 119.1 (q, J = 271.6 Hz), 117.5, 61.6, 14.2 ppm. 19F NMR (376 MHz, CDCl3): δ -55.2 ppm. MS (ESI): [M + H] + 353.1. IR (neat): νmax 3068-2906 (w, CH), 1702 (s, C⩵O), 1595 (w), 1557 (m), 1486 (m), 1417 (w), 1403 (w), 1385 (m), 1356 (w), 1295 (m), 1254 (s), 1221 (m), 1174 (m), 1149 (s), 1135 (s), 1084 (m), 1039 (m), 1013 (m), 982 (m) cm−1. 1-(3,4-Dichlorophenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylic Acid (43) According to General Procedure B, the ester derivative S38 (0.10 g, 0.28 mmol) was stirred for 4 h to give the acid derivative 4352 as a light brown amorphous solid (40 mg, 0.12 mmol, 43%). Rf: 0.30 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 9.57 (br. s, 1H), 8.22 (s, 1H), 7.65-7.55 (m, 2H), 7.31 (dd, J = 8.6, 2.5 Hz, 1H) ppm. 13C NMR (100 MHz, CDCl3): δ 166.0, 143.7, 138.2, 135.0, 133.8 (q, J = 40.6 Hz), 133.6, 131.0, 128.1 (q, J = 1.2 Hz), 125.2 (q, J = 1.2 Hz), 118.8 (q, J = 272.0 Hz), 116.1 (q, J = 1.2 Hz) ppm. 19F NMR (376 MHz, CDCl3): δ -55.2 ppm. MS (ESI): [M - H]- 323.0. HPLC: retention time: 2.06 min (>99%). IR (neat): νmax 3102-2872 (m, O–H, C–H), 1703 (s, C⩵O), 1562 (m), 1484 (m), 1424 (w), 1297 (m), 1265 (m), 1249 (m), 1216 (m), 1185 (s), 1152 (s), 1132 (s), 1076 (m), 1031 (s) cm−1. (4,5-Dichloro-2-methylphenyl)hydrazine Hydrochloride (S39) According to General Procedure F, 4,5-dichloro-2-methylaniline (0.69 g, 3.91 mmol) gave the hydrazine derivative S39 as a pale brown amorphous solid (0.49 g, 2.15 mmol, 55%). Rf 0.47 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CD3OD): δ 7.34 (s, 1H), 7.05 (s, 1H), 2.22 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 143.7, 133,1, 131.0, 127.5, 126.2, 115.2, 16.5 ppm. MS (ESI): [M + H]+ 228.0. IR (neat): νmax 3273 (m, N-H), 2872 (s, N-H), 2693 (s, N-H, C–H), 1586 (w), 1536 (s), 1493 (s), 1422 (m), 1380 (m), 1219 (w), 1174 (m), 1149 (m), 1118 (m), 941 (m) cm−1. Ethyl 1-(4,5-Dichloro-2-methylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S40) According to General Procedure E, the hydrazine derivative S39 (0.49 g, 2.15 mmol) was stirred for 4 h to give a crude product. The crude product was purified using flash column chromatography (0-5% EtOAc in petrol) to give the ester derivative S40 as a yellow amorphous solid (0.43 g, 1.17 mmol, 54%). Rf: 0.63 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.15 (s, 1H), 7.45 (s, 1H), 7.38 (s, 1H), 4.38 (q, J =7.1 Hz, 2H), 2.00 (s, 3H), 1.38 (t, J =7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.7, 143.1, 137.6, 135.8, 134.7, 133.6 (q, J = 40.1 Hz), 132.4, 130.4, 129.0, 119.0 (q, J = 271.6 Hz), 116.5 (q, J = 1.5 Hz), 61.5, 16.5, 14.2 ppm. 19F NMR (376 MHz, CDCl3): δ -57.6 ppm. MS (ESI): [M + H]+ 367.1. IR (neat): νmax 3122-2869 (w, C–H), 1716 (m, C⩵O), 1555 (m), 1488 (m), 1419 (w), 1400 (w), 1383 (m), 1352 (w), 1294 (m), 1245 (m), 1149 (s), 1132 (s), 1077 (m), 1035 (m), 1008 (m), 984 (m) cm−1. 1-(4,5-Dichloro-2-methylphenyl)-5-(trifluoromethyl)-1H-pyra-zole-4-carboxylic Acid (44) According to General Procedure B, the ester derivative S40 (0.43 g, 1.17 mmol) was stirred for 5 h to give a crude product. The crude product was triturated with hexane to give the acid derivative 44 as a white amorphous solid (0.32 g, 0.94 mmol, 80%). Rf: 0.38 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 11.12 (br. s, 1H), 8.26 (s, 1H), 7.47 (s, 1H), 7.40 (s, 1H), 2.03 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 166.0, 143.9, 137.4, 135.6, 135.0, 134.6 (q, J = 40.4 Hz), 132.5, 130.6, 129.0, 118.7 (q, J = 272.0 Hz), 115.2, 16.5 ppm. 19F NMR (376 MHz, CDCl3): δ - 57.7 ppm. MS (ESI): [M - H]- 337.0. HPLC: retention time: 2.15 min (>99%). IR (neat): νmax 2924-2600 (m, O–H, C–H), 1701 (m, C⩵O), 1681 (m), 1561 (m), 1490 (m), 1452 (m), 1418 (w), 1300 (m), 1260 (m), 1223 (m), 1178 (m), 1149 (s), 1074 (w), 1032 (m) cm−1. HRMS: calculated for C12H7Cl2F3N2O2 [M + H]+ = 338.9915, observed = 338.9915. (3,4-Dichloro-2-methylphenyl)hydrazine Hydrochloride (S41) According to General Procedure F, 3,4-dichloro-2-methylaniline (0.20 g, 1.13 mmol) gave the hydrazine derivative S41 as a pale brown amorphous solid (0.23 g, 0.63 mmol, 56%). Rf: 0.54 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CD3OD): δ 7.41 (d, J = 8.7 Hz, 1H), 6.90 (d, J = 8.7 Hz, 1H), 2.37 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 143.6, 134.2, 128.8, 127.7, 127.5, 113.5, 14.9 ppm. MS (ESI): [M + H]+228.1. IR (neat): νmax 3181-3156 (m, N-H), 2990–2653 (s, N-H, C–H), 1556 (s), 1517 (m), 1483 (w), 1456 (s), 1401 (w), 1190 (m), 1170 (s), 1062 (m) cm−1. Ethyl 1-(3,4-Dichloro-2-methylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxylate (S42) According to General Procedure E, the hydrazine derivative S41 (0.12 g, 0.53 mmol) was stirred for 5.5 h to give a crude product. The crude product was purified using flash column chromatography (5% EtOAc in petrol) and triturated with hexane to give the ester derivative S42 as a colorless amorphous solid (0.11 g, 0.30 mmol, 57%). Rf: 0.58 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.16 (s, 1H), 7.45 (d, J =8.5 Hz, 1H), 7.15 (d, J = 8.5 Hz, 1H), 4.38 (q, J = 7.1, 2H), 2.07 (s, 3H), 1.39 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 160.8, 143.1, 137.7, 136.8, 135.5, 134.1, 133.7 (q, J = 40.0 Hz), 127.9, 126.1, 119.0 (q, J = 271.6 Hz), 116.5 (q, J = 1.4 Hz), 61.5, 16.1, 14.2 ppm. 19F NMR (376 MHz, CDCl3): δ -57.6 ppm. MS (ESI): [M + H] + 367.1. IR (neat): νmax 3120 (w, C–H), 2980 (w, C–H), 1715 (s, C⩵O), 1555 (m), 1476 (m), 1385 (m), 1295 (m), 1248 (s), 1150 (s), 1086 (m), 1039 (m), 1013 (m), 974 (m) cm−1. 1-(3,4-Dichloro-2-methylphenyl)-5-(trifluoromethyl)-1H-pyra-zole-4-carboxylic Acid (45) According to General Procedure B, the ester derivative S42 (0.11 g, 0.30 mmol) was stirred for 5 h to give a crude product. The crude product was triturated with hexane to give the acid derivative 45 as a white amorphous solid (90 mg, 0.27 mmol, 90%). Rf: 0.25 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 10.43 (br. s, 1H), 8.26 (s, 1H), 7.47 (d, J = 8.5 Hz, 1H), 7.17 (d, J = 8.5 Hz, 1H), 2.10 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 166.0, 143.8, 137.5, 136.7, 135.8, 134.7 (q, J = 40.4 Hz), 134.2, 127.9, 126.1, 118.8 (q, J = 272.0 Hz), 115.2, 16.1 ppm. 19F NMR (376 MHz, CDCl3): δ -57.7 ppm. MS (ESI): [M - H]- 337.0. HPLC: retention time: 2.08 min (>99%). IR (neat): νmax 3127-2592 (m, O–H, C–H), 1678 (m, C⩵O), 1561 (m), 1459 (m), 1293 (m), 1254 (m), 1228 (m), 1183 (m), 1147 (s), 1088 (m), 1035 (m) cm−1. HRMS: calculated for C12H7Cl2F3N2O2 [M + H]+ = 338.9915, observed = 338.9910. Ethyl 2,5-Dimethyl-1-(2-methylphenyl)-1H-pyrrole-3-carboxylate (S43) According to General Procedure C, 2-methylaniline (57 μL, 0.54 mmol) and the carbonyl derivative S5 were stirred for 7 h to give a crude product. The crude product was purified by flash column chromatography (0-5% EtOAc in petrol) to give the ester derivative S4358 as a colorless oil (0.12 g, 0.47 mmol, 87%). Rf: 0.53 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.42-7.27 (m, 3H), 7.10 (d, J = 7.6 Hz, 1H), 6.39 (s, 1H), 4.28 (q, J = 7.1 Hz, 2H), 2.19 (s, 3H), 1.93 (s, 3H), 1.87 (s,3H), 1.35 (t, J = 7.1, 3H) ppm. 13C NMR (100MHz, CDCl3): δ 165.9, 136.9, 136.7, 135.8, 131.0, 129.1, 128.6, 128.2, 127.0, 111.4, 107.4, 59.3, 17.1, 14.7, 12.3, 12.0 ppm. MS (ESI): [M + H]+ 258.1. IR (neat): vmiD1 2979-2921 (w, C–H), 1695 (s, C⩵O), 1579 (w), 1534 (m), 1495 (m), 1411 (m), 1335 (w), 1241 (m), 1215 (s), 1198 (m), 1121 (w), 1073 (s), 1002 (w) cm−1. 2,5-Dimethyl-1-(2-methylphenyl)-1H-pyrrole-3-carboxylic Acid (46) According to General Procedure B, the ester derivative S43 (30 mg, 0.11 mmol) was stirred for 2 days to give a crude product. The crude product was purified using flash column chromatography (30% EtOAc in petrol) to give the acid derivative 4658 as a white amorphous solid (21 mg, 92 μMol, 83%). Rf: 0.21 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.43-7.28 (m, 3H), 7.13 (d, J = 7.6 Hz, 1H), 6.46 (s, 1H), 2.23 (s, 3H), 1.95 (s, 3H), 1.89 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 171.4, 137.3, 136.8, 136.7, 131.1, 129.4, 128.7, 128.6, 127.1, 110.8, 108.1, 17.1, 12.4, 12.3 ppm. MS (ESI): [M + H] + 230.1. HPLC: retention time: 1.90 min (>99%). IR (neat): νmax 3033–2850 (s, O–H, C–H), 2746-2509 (m, C–H), 1652 (s, C⩵O), 1575 (w), 1530 (m), 1494 (m), 1425 (m), 1401 (w), 1257 (s), 1197 (w), 1122 (w), 1078 (m), 1026 (w), 1004 (w), 931 (m) cm−1. Ethyl 1-(2,6-Dimethylphenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylate (S44) According to General Procedure C, 2,6-dimethylaniline (65 μL, 0.53 mmol) and the carbonyl derivative S5 were stirred for 18h to give a crude product. The crude product was purified by flash column chromatography (5% EtOAc in petrol) to give the ester derivative S44 as colorless oil (0.13 g, 0.48 mmol, 91%). Rf: 0.71 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.31-7.23 (m, 1H), 7.19 (d, J = 7.5 Hz, 2H), 6.45 (s, 1H), 4.30 (q, J = 7.1 Hz, 2H), 2.18 (s, 3H), 1.93 (s, 6H), 1.86 (s, 3H), 1.38 (t, J = 7.1, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 166.0, 136.7, 135.9, 135.0, 128.8, 128.4, 127.2, 111.6, 107.7, 59.3, 17.5, 14.7, 12.0, 11.7 ppm. MS (ESI): [M + H] + 272.5, IR (neat): νmax 2978–2920 (w, C–H), 1696 (s, C⩵O), 1533 (w), 1478 (w), 1410 (m), 1380 (w), 1214 (s), 1098 (m), 1073 (s), 1001 (w) cm−1. 1-(2,6-Dimethylphenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylic Acid (47) According to General Procedure B, the ester derivative S44 (61 mg, 0.22 mmol) was stirred for 2 days to give the acid derivative 47 as a white amorphous solid (40 mg, 0.16 mmol, 73%). Rf 0.70 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 7.31-7.26 (m, 1H), 7.20 (d, J = 7.6 Hz, 2H), 6.52 (s, 1H), 2.21 (s, 3H), 1.96 (s, 6H), 1.87 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 171.5,136.7,136.5, 135.9,128.9, 128.5, 127.7, 111.0, 108.4, 17.5, 12.1, 11.9 ppm. MS (ESI): [M - H]- 242.8. HPLC: retention time: 2.01 min (>99%). IR (neat): νmax 3033–2573 (O–H, C–H), 1649 (s, C⩵O), 1574 (w), 1530 (m), 1475 (w), 1425 (m), 1376 (w), 1330 (w), 1265 (s), 1250 (s), 1101 (w), 1078 (m), 1025 (w), 1001 (w), 972 (w), 935 (m) cm−1. HRMS: calculated for C15H17NO2 [M + H]+ = 244.1338, observed = 244.1335. Ethyl 1-(3,4-Dichlorophenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylate (S45) According to General Procedure C, 3,4-dichloroaniline (0.11 g, 0.67 mmol) and the carbonyl derivative S5 were stirred for 18 h to give a crude product. The crude product was purified by flash column chromatography (0-5% EtOAc in petrol) to give the ester derivative S45 as colorless oil (0.15 g, 0.48 mmol, 72%). Rf 0.57 (20:80 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.58 (d, J = 8.4 Hz, 1H), 7.32 (s, 1H), 7.05 (d, J = 8.4 Hz, 1H), 6.36 (s, 1H), 4.27 (q, J = 7.1 Hz, 2H), 2.29 (s, 3H), 1.98 (s, 3H), 1.34 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 165.5, 137.2, 136.0, 133.5, 133.2, 131.2, 130.3, 128.6, 127.7, 112.3, 108.2, 59.5, 14.6, 12.7, 12.4ppm. MS (ESI): [M + H]+ 312.1. IR (neat): νmax 2976-2858 (w, C–H), 1690 (s, C⩵O), 1585 (w), 1533 (m), 1473 (m), 1454 (m), 1410 (m), 1352 (m), 1324 (w), 1253 (w), 1219 (s), 1129 (m), 1084 (s), 1028 (s) cm−1. 1-(3,4-Dichlorophenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylic Acid (48) According to General Procedure B, the ester derivative S45 (28.0 mg, 89.7 μMol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (30% EtOAc in petrol) to give the acid derivative 4837 as a white amorphous solid (15 mg, 52.8 μMol, 59%). Rf 0.22 (30:70 EtOAc:petrol). 1H NMR (400 MHz, DMSO-d6): δ 11.7 (br. s, 1H), 7.81 (d, J = 8.5 Hz, 1H), 7.77 (s, 1H), 7.37 (dd, J = 8.5, 2.4 Hz, 1H), 6.24 (s, 1H), 2.22 (s, 3H), 1.95 (s, 3H) ppm. 13C NMR (100 MHz, DMSO-d6): δ 166.0, 136.9,135.0,131.8,131.5,131.2,130.2,128.7,128.0,111.8,108.0, 12.3, 12.0 ppm. MS (ESI): [M + H]+284.0. HPLC: retention time: 1.99 min (>99%). IR (neat): νmax 3069-2574 (s, O–H, C–H), 1644 (s, C⩵O), 1583 (w), 1530 (m), 1465 (s), 1403 (m), 1383 (m), 1327 (w), 1255 (s), 1241 (s), 1129 (m), 1091 (m), 1030 (m), 927 (m) cm−1. Ethyl 1-(3-chlorophenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylate (S46) According to General Procedure C, 3-chloroaniline (57 μL, 0.54 mmol) and the carbonyl derivative S5 were stirred for 18 h to give a crude product. The crude product was purified by flash column chromatography (0-5% EtOAc in petrol) to give the ester derivative S4637 as colorless oil (0.12 g, 0.43 mmol, 80%). Rf 0.54 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.47-7.40 (m, 2H), 7.21 (s, 1H), 7.12-7.05 (m, 1H), 6.37 (s, 1H), 4.28 (q, J = 7.1 Hz, 2H), 2.29 (s,3H), 1.98 (s, 3H), 1.34 (t, J =7.1 Hz, 3H) ppm. 1C NMR (100 MHz, CDCl3): δ 165.6, 139.0, 136.1, 135.1, 130.4, 129.0, 128.6, 126.6, 125.1, 112.0, 108.0, 59.4, 14.6, 12.7, 12.4 ppm. MS (ESI): [M + H] + 278.2. IR (neat): νmax 2978-2850 (m, C–H), 1695 (s, C⩵O), 1592 (m), 1581 (m), 1535 (m), 1479 (m), 1409 (m), 1373 (m), 1354 (w), 1216 (s), 1098 (m), 1077 (s), 1026 (w) cm−1. 1-(3-Chlorophenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylic Acid (49) According to General Procedure B, the ester derivative S46 (30 mg, 0.11 mmol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (30% EtOAc in petrol) to give the acid derivative 4937 as a colorless amorphous solid (19 mg, 76.1 μMol, 69%). Rf: 0.22 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.50-7.41 (m,2H), 7.23 (s, 1H), 7.15-7.08 (m, 1H), 6.43 (s, 1H), 2.32 (s, 3H), 1.99 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 171.2, 138.9, 137.6, 135.2, 130.5, 129.2, 129.1, 128.6, 126.6, 111.2, 108.5, 12.8, 12.6 ppm. MS (ESI): [M + H] +250.1. HPLC: retention time: 1.95 min (>99%). IR (neat): νmax 3086–2586 (s, O–H, C–H), 1646 (s, C⩵O), 1584 (m), 1534 (m), 1466 (m), 1430 (m), 1399 (w), 1378 (w), 1330 (w), 1271 (m), 1250 (s), 1122 (w), 1079 (w), 1034 (w), 945 (w) cm−1. Ethyl 1-(4-Chlorophenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylate (S47) According to General Procedure C, 4-chloroaniline (0.10 g, 0.54 mmol) and the carbonyl derivative S5 were stirred for 18 h to give a crude product. The crude product was purified by flash column chromatography (0-5% EtOAc in petrol) to give the ester derivative S4737 as a colorless amorphous solid (0.11 g, 0.40 mmol, 74%). Rf: 0.57 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.47 (d, J = 8.2 Hz, 2H), 7.12 (d, J = 8.2 Hz, 2H), 6.37 (s, 1H),4.27 (q, J =7.1 Hz, 2H), 2.28 (s, 3H), 1.97 (s, 3H), 1.34 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 165.7, 136.3, 136.1, 134.6, 129.7, 129.6, 128.7, 111.9, 107.9, 59.4, 14.6, 12.7, 12.4ppm. MS (ESI): [M + H]+278.2. IR (neat): νmax 2922 (s, C–H), 2852 (m, C–H), 1694 (s, C⩵O), 1535 (w), 1492 (m), 1415 (m), 1371 (w), 1221 (s), 1081 (s), 1000 (m) cm−1. 1-(4-Chlorophenyl)-2,5-dimethyl-1 H-pyrrole-3-carboxylic Acid (50) According to General Procedure B, the ester derivative S47 (30 mg, 0.11 mmol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (30% EtOAc in petrol) to give the acid derivative 5032 as a white amorphous solid (18 mg, 72.1 μMol, 66%). Rf: 0.22 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.48 (d, J = 8.4 Hz, 2H), 7.14 (d, J = 8.4 Hz, 2H), 6.43 (s, 1H), 2.30 (s, 3H), 1.98 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 170.8, 137.6, 136.2, 134.8, 129.8, 129.6, 129.1, 111.1, 108.5, 12.8,12.6 ppm. MS (ESI): [M +H]+250.1. HPLC: retention time: 1.95 min (>99%). IR (neat): νmax 3089-2508 (s, O–H, C–H), 1646 (s, C⩵O), 1580 (w), 1538 (m), 1495 (m), 1468 (w), 1428 (m), 1401 (m), 1366 (w), 1329 (w), 1258 (s), 1083 (m), 1018 (w), 1002 (w), 942 (m) cm−1. Ethyl 1-(4,5-Dichloro-2-methylphenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylate (S48) According to General Procedure C, 4,5-dichloro-2-methylaniline (95 mg, 0.54 mmol) and the carbonyl derivative S5 were stirred for 5 h to give a crude product. The crude product was purified by flash column chromatography (5% EtOAc in petrol) to give the ester derivative S48 as pale yellow oil (0.15 g, 0.46 mmol, 85%). Rf: 0.48 (20:80 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.45 (s, 1H), 7.25 (s, 1H), 6.39 (s, 1H), 4.27 (q, J = 7.1 Hz, 2H), 2.20 (s, 3H), 1.90 (s, 3H), 1.89 (s, 3H), 1.34 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 165.6, 137.1, 136.3, 135.5, 133.3, 132.4, 130.6, 130.4, 128.0, 112.2, 108.1, 59.5, 16.7, 14.6, 12.3, 12.0ppm. MS (ESI): [M + H]+ 326.1. IR (neat): νmax 2979-2922 (w, C–H), 1695 (s, C⩵O), 1581 (w), 1535 (w), 1479 (s), 1414 (m), 1331 (w), 1216 (s), 1187 (s), 1132 (m), 1077 (s), 1028 (m) cm−1. 1-(4,5-Dichloro-2-methylphenyl)-2,5-dimethyl-1H-pyrrole-3-car-boxylic Acid (51) According to General Procedure B, the ester derivative S48 (0.12 g, 0.36 mmol) was stirred for 1 day to give a crude product. The crude product was purified using flash column chromatography (0-30% EtOAc in petrol) to give the acid derivative 51 as an amorphous white solid (81 mg, 0.27 mmol, 75%). Rf: 0.20 (30:70 EtOAc:petrol). 1H NMR (400 MHz, DMSO-d6): δ 7.75 (s, 1H), 7.62 (s, 1H), 6.25 (s, 1H), 2.10 (s, 3H), 1.83 (s, 6H) ppm. 13C NMR (100 MHz, DMSO-d6): δ 166.0, 137.6, 136.2, 134.6, 132.3, 131.7, 130.4, 129.0, 127.4, 111.7, 107.9, 15.9, 11.8, 11.6 ppm. MS (ESI): [M + H]+ 298.0. HPLC: retention time: 2.18 min (>99%). IR (neat): νmax 2918-2579 (m, O–H, C–H), 1646 (s, C⩵O), 1581 (w), 1536 (w), 1477 (m), 1429 (w), 1406 (w), 1383 (w), 1327 (w), 1260 (s), 1233 (m), 1187 (w), 1134 (m), 1083 (w), 1028 (m), 951 (w) cm−1. HRMS: calculated for C14H13Cl2NO2 [M + H]+ = 298.0402, observed = 298.0399. Ethyl 1-(3,4-Dichloro-2-methylphenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylate (S49) According to General Procedure C, 3,4-dichloro-2-methylaniline (0.10 g, 0.54 mmol) and the carbonyl derivative S5 were stirred for 3 h to give a crude product. The crude product was purified by flash column chromatography (5% EtOAc in petrol) to give the ester derivative S49 as an amorphous white solid (0.15 g, 0.46 mmol, 85%). Rf. 0.68 (30:70 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 7.44 (d, J =8.4Hz, 1H), 7.02 (d, J = 8.4 Hz, 1H), 6.39 (s, 1H), 4.28 (q, J = 7.1 Hz, 2H), 2.19 (s,3H), 1.99 (s,3H), 1.87 (s, 3H), 1.35 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 165.6, 137.9, 136.3, 135.8, 134.2, 134.0, 128.3, 128.2, 127.6, 112.1, 108.0, 59.5, 16.0, 14.6, 12.3, 12.0 ppm. MS (ESI): [M + H]+ 328.0. IR (neat): νmax 3075–2917 (w, C–H), 1696 (s, C⩵O), 1577 (w), 1535 (m), 1467 (m), 1411 (m), 1395 (m), 1372 (m), 1351 (w), 1333 (w), 1249 (m), 1219 (s), 1194 (s), 1127 (w), 1080 (s), 1046 (m), 1001 (m) cm−1. 1-(3,4-Dichloro-2-methylphenyl)-2,5-dimethyl-1H-pyrrole-3-carboxylic Acid (52) According to General Procedure B, the ester derivative S49 (0.12 g, 0.36 mmol) was stirred for 1 day to give a crude product. The crude product was purified using flash column chromatography (0-30% EtOAc in petrol) to give the acid derivative 52 as an amorphous white solid (90 mg, 0.30 mmol, 83%). Rf 0.15 (30:70 EtOAc.petrol). 1H NMR (400 MHz, DMSO-d6). δ 7.68 (d, J = 8.4 Hz, 1H), 7.31 (d, J = 8.4 Hz, 1H), 6.27 (s, 1H), 2.09 (s, 3H), 1.92 (s, 3H), 1.82 (s, 3H) ppm. 13C NMR (100 MHz, DMSO-d6). δ 166.0, 137.4, 136.0, 134.8, 132.6, 132.5, 128.6, 128.4, 127.7, 111.8, 107.9, 15.6, 11.8, 11.6 ppm. MS (ESI). [M + H]+ 298.1. HPLC. retention time. 2.20 min (>99%). IR (neat). νmax 3067-2521 (m, O–H, C–H), 1655 (s, C⩵O), 1577 (w), 1534 (m), 1469 (m), 1427 (m), 1392 (m), 1379 (m), 1359 (w), 1331 (w), 1261 (s), 1239 (m), 1194 (m), 1126 (w), 1087 (m), 1046 (w), 999 (m) cm−1. HRMS. calculated for C14H13Cl2NO2 [M + H]+ = 298.0402, observed = 298.0394. Ethyl 5-Methyl-1-(2-methylphenyl)-2-(trifluoromethyl)-1H-pyr-role-3-carboxylate (S50) According to General Procedure D, 2-methylaniline (0.10 g, 0.41 mmol) was stirred for 4.5 h to give a crude product. The crude product was purified by flash column chromatography (0-5% EtOAc in petrol) to give the ester derivative S50 as yellow oil (27 mg, 86.7 μMol, 21%). Rf. 0.57 (30.70 EtOAc.petrol). 1H NMR (400 MHz, CDCl3). δ 7.44-7.24 (m, 3H), 7.14 (d, J = 7.8 Hz, 1H), 6.53 (s, 1H), 4.33 (q, J = 7.1 Hz, 2H), 1.97 (s, 3H), 1.86 (s, 3H), 1.36 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3). δ 163.5, 136.8, 136.1, 133.1 (q, J = 1.4 Hz), 130.9, 129.6, 128.1 (q, J = 0.75 Hz), 126.8, 121.8 (q, J = 38.1 Hz), 120.5 (q, J = 269.3 Hz), 118.2 (q, J = 2.1 Hz), 110.3, 60.7, 16.9, 14.2, 12.1 ppm. 19F NMR (376 MHz, CDCl3). δ - 55.8 ppm. MS (ESI). [M + H]+ 312.1. IR (neat). νmax 2925 (m, C–H), 2853 (w, C–H), 1727 (m, C⩵O), 1511 (m), 1495 (m), 1460 (w), 1422 (m), 1378 (w), 1274(m), 1227 (s), 1199 (m), 1159 (s),1116 (s), 1035 (m) cm−1. 5-Methyl-l-(2-methylphenyl)-2-(trifluoromethyl)-lH-pyrrole-3-carboxylic Acid (53) According to General Procedure B, the ester derivative S50 (27.0 mg, 86.7 μMol) was stirred for 4 h to give a crude product. The crude product was purified using flash column chromatography (30% EtOAc in petrol) to give the acid derivative 53 as an amorphous white solid (16.0 mg, 56.5 μMol, 65%). Rf. 0.67 (20.80 MeOH.DCM). 1H NMR (400 MHz, CD3OD). δ 7.43-7.28 (m, 3H), 7.17 (d, J = 7.7 Hz, 1H), 6.63 (s, 1H), 1.99 (s, 3H), 1.88 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3). δ 168.5, 136.8, 136.0, 133.2 (q, J = 1.4Hz), 131.0, 129.7, 128.0, 126.9, 123.0 (q, J = 38.5 Hz), 120.3 (q, J = 269.9 Hz), 116.7 (q, J = 1.5 Hz), 111.2, 16.9, 12.2 ppm. 19FNMR(376 MHz, CDCl3). δ -54.9 ppm. MS (ESI). [M + H]+ 284.1. HPLC. retention time: 1.88 min (>99%). IR (neat): νmax 2920-2850 (s, O–H, C–H), 1690 (m, C⩵O), 1666 (m), 1519 (m), 1496 (w), 1460 (w), 1413 (m), 1336 (w), 1284 (m), 1252 (m), 1206 (w), 1172 (m), 1117 (s), 1046 (w), 1018 (m), 1002 (m) cm−1. HRMS. calculated for C14H10F3NO2 [M + H]+ = 284.0898, observed = 284.0896. Ethyl 1-(4,5-Dichloro-2-methylphenyl)-5-methyl-2-(trifluoromethyl)-1H-pyrrole-3-carboxylate (S51) According to General Procedure D, 4,5-dichloro-2-methylaniline (0.22 g, 1.25 mmol) was stirred for 24 h to give a crude product. The crude product was purified by flash column chromatography (0-5% EtOAc in petrol) to give the ester derivative S51 as yellow oil (0.13 g, 0.34 mmol, 27%). Rf. 0.68 (30.70 EtOAc.petrol). 1H NMR (400 MHz, CDCl3). δ 7.44 (s, 1H), 7.29 (s, 1H), 6.53 (s, 1H), 4.32 (q, J = 7.1 Hz, 2H), 1.93 (s, 3H), 1.89 (s, 3H), 1.35 (t, J = 7.1 Hz, 3H) ppm. 13C NMR (100 MHz, CDCl3). δ 163.2, 136.5, 136.1, 133.9, 132.9 (q, J = 1.5 Hz), 132.3, 130.5, 129.8 (q, J =1.1 Hz), 121.8 (q, J = 38.3 Hz), 120.3 (q, J = 269.6 Hz), 118.9 (q, J = 2.3 Hz), 110.8, 60.9, 16.5, 14.2, 12.2 ppm. 19F NMR (376 MHz, CDCl3). δ -55.6 ppm. MS (ESI). [M + H]+ 380.2. IR (neat). νmax 2983-2929 (w, C–H), 1726 (m, C⩵O), 1513 (m), 1477 (m), 1423 (m), 1276 (m), 1234 (s), 1194 (s), 1164 (s), 1121 (s), 1039 (m), 1027 (m), 1001 (m) cm−1. 1-(4,5-Dichloro-2-methylphenyl)-5-methyl-2-(trifluoromethyl)-1H-pyrrole-3-carboxylic Acid (54) According to General Procedure B, the ester derivative S51 (0.10 g, 0.26 mmol) was stirred for 18 h to give a crude product. The crude product was purified using flash column chromatography (0-30% EtOAc in petrol) and triturated with petrol to give the acid derivative 54 as an amorphous white solid (71 mg, 0.22 mmol, 85%). Rf. 0.58 (EtOAc). 1H NMR (400 MHz, CDCl3). δ 7.63 (s, 1H), 7.51 (s, 1H), 6.55 (s, 1H), 1.95 (s, 3H), 1.92 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3). δ 166.2, 138.3, 137.6, 134.7, 133.4, 133.4, 131.2, 131.1, 122.6 (q, J = 38.1 Hz), 121.7 (q, J = 268.7 Hz), 120.3 (q, J = 2.1 Hz), 111.9, 16.3, 12.0 ppm. 19F NMR (376 MHz, CDCl3). δ -56.8 ppm. MS (ESI): [M - H]- 350.0. HPLC. retention time. 2.23 min (>99%). IR (neat). νmax 2958-2559 (w, O–H, C–H), 1673 (m, C⩵O), 1517 (m), 1475 (m), 1420 (m), 1272 (m), 1205 (m), 1159 (m), 1130 (s), 1035 (m), 1006 (m) cm−1. HRMS. calculated for C14H10Cl2F3NO2 [M + H]+ = 352.0119, observed = 352.0114. Ethyl 1-(3,4-Dichloro-2-methylphenyl)-5-methyl-2-(trifluoro-methyl)-1H-pyrrole-3-carboxylate (S52) According to General Procedure D, 4,5-dichloro-2-methylaniline (0.22 g, 1.25 mmol) was stirred for 1 day to give a crude product. The crude product was purified by flash column chromatography (0-5% EtOAc in petrol) to give the ester derivative S52 as yellow oil (80 mg, 0.21 mmol, 17%). Rf. 0.67 (30.70 EtOAc.petrol). 1H NMR (400 MHz, CDCl3). δ 7.43 (d, J = 8.5 Hz, 1H), 7.07 (d, J = 8.5 Hz, 1H), 6.53 (s, 1H),4.32 (q, J =7.2 Hz, 2H), 2.01 (s,3H), 1.87 (s, 3H), 1.35 (t, J = 7.1 Hz, 3H) ppm. 13CNMR(100 MHz, CDCl3). δ 163.2,137.3, 136.0, 134.6, 134.1, 133.2 (q, J = 1.4 Hz), 128.0, 127.1 (q, J = 0.7 Hz), 121.9 (q, J = 38.2 Hz), 120.3 (q, J = 269.5 Hz), 118.9 (q, J = 2.1 Hz), 110.7, 60.9, 16.0, 14.2, 12.2 ppm. 19F NMR (376 MHz, CDCl3). δ -55.6 ppm. MS (ESI). [M + H]+ 380.2. IR (neat). νmax 3121-2851 (w, C–H), 1721 (m, C⩵O), 1512 (w), 1460 (w), 1420 (w), 1274 (m), 1234 (m), 1187 (m), 1124 (s), 1073 (m), 1053 (m), 1030 (m), 999 (m) cm−1. 1-(3,4-Dichloro-2-methylphenyl)-5-methyl-2-(trifluoromethyl)-1H-pyrrole-3-carboxylic Acid (55) According to General Procedure B, the ester derivative S52(80.0 mg, 0.21 mmol) was stirred for 24 h to give a crude product. The crude product was purified using flash column chromatography (0-30% EtOAc in petrol) and triturated with petrol to give the acid derivative 55 as an amorphous white solid (52.0 mg, 0.15 mmol, 71%). Rf. 0.27 (30.70 EtOAc.petrol). 1H NMR (400 MHz, CDCl3): δ 7.58 (d, J = 8.5 Hz, 1H), 7.25 (d, J = 8.5 Hz, 1H), 6.56 (s, 1H), 2.02 (s, 3H), 1.90 (s, 3H) ppm. 13C NMR (100 MHz, CD3OD): δ 166.3, 138.5, 137.5, 135.4, 134.9 (q, J = 1.4 Hz), 134.6, 129.3, 128.8 (q, J = 1.0 Hz), 122.7 (q, J = 38.1 Hz), 121.7 (q, J = 268.7 Hz), 120.3 (q, J = 2.1 Hz), 111.9, 16.0, 12.0ppm. 19FNMR(376MHz, CDCl3): δ -56.7 ppm. MS (ESI). [M - H]- 350.0. HPLC. retention time. 2.26 min (>99%). IR (neat). νmax 3084-2592 (m, O–H, C–H), 1689 (s), 1579 (w), 1519 (m), 1467 (m), 1424 (m), 1391 (w), 1334 (w), 1282 (m), 1257 (s), 1178 (m), 1158 (m), 1124 (s), 1052 (m), 1012 (m), 1000 (m) cm−1. HRMS. calculated for C14H10Cl2F3NO2 [M + H]+ = 352.0119, observed = 352.0112. N-Benzyl-5-methyl-1-phenyl-1 H-pyrazole-4-carboxamide (58) Thionyl chloride (1.40 mL) was added to the carboxylic acid derivative 4 (0.13 g, 0.63 mmol), and the solution was stirred at 80 °C for 3 h. After concentrating the solution in vacuo, 1,4-dioxane (3.1 mL), benzylamine (0.10 mL, 0.95 mmol), and pyridine (77 μL, 0.95 mmol) were added and the resulting mixture was stirred at room temperature for 18h. Successively, the mixture was concentrated in vacuo and EtOAc (6 mL) was added. The organic phase was washed with a 0.1 M HCl aqueous solution (3 × 10 mL) and brine (5 mL), dried under anhydrous magnesium sulfate, and concentrated in vacuo to yield a crude product, which was purified by flash column chromatography (EtOAc) to give the amide derivative 5859 as a pale brown amorphous solid (76 mg, 0.26 mmol, 42%). Rf 0.49 (50.50 EtOAc.petrol). 1H NMR (400 MHz, CDCl3). δ 7.78 (s, 1H), 7.53-7.48 (m, 2H), 7.47 7.40 (m, 3H), 7.39-7.35 (m, 4H), 7.33-7.28 (m, 1H), 6.07 (br. s, 1H), 4.63 (d, J = 5.7 Hz, 2H), 2.61 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 163.4, 142.3, 138.8, 138.3, 137.9, 129.1, 128.7, 128.5, 127.8, 127.5, 125.4, 115.3, 43.3, 11.8 ppm.MS (ESI). [M + H]+292.2. HPLC: retention time: 1.84 min (>99%). IR (neat): νmax 3315 (m, NH), 1629 (s, C⩵O), 1593 (m), 1567 (s), 1536 (m), 1503 (s), 1453 (m), 1394 (s), 1353 (w), 1287 (s), 1262 (w), 1138 (w), 938 (s) cm−1. N-Methanesulfonyl-1-phenyl-5-(trifluoromethyl)-1H-pyrazole-4-carboxamide (59) According to General Procedure G, the carboxylic acid derivative 18 (44 mg, 0.17 mmol) and methanesulfonamide gave the sulfonamide derivative 59 as a white amorphous solid (40 mg, 0.12 mmol, 70%). Rf 0.29 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 9.08 (s, 1H), 8.09 (s, 1H), 7.58-7.48 (m, 3H), 7.45-7.38 (m, 2H), 3.45 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 159.0, 140.1, 138.8, 132.8 (q, J = 40.4 Hz), 130.4, 129.4, 125.8 (q, J = 1.0 Hz), 118.9 (q, J = 271.6 Hz), 117.3 (q, J =1.2 Hz),41.9ppm. 19FNMR(376 MHz, CDCl3): δ -56.4 ppm. MS (ESI): [M + H]+ 334.1. HPLC: retention time: 1.80 min (>99%). IR (neat): νmax 3276 (m, N-H), 3137-2940 (w, C–H), 1703 (m, C⩵O), 1559 (w), 1502 (w), 1433 (m), 1405 (m), 1326 (m), 1294 (m), 1227 (m), 1158 (s), 1132 (s), 1082 (m), 1073 (m), 1022 (m), 973 (m) cm−1. HRMS: calculated for C12H10F3N3O3S [M + H]+ = 334.0473, observed = 334.0475. N-(Benzenesulfonyl)-1-phenyl-5-(trifluoromethyl)-1H-pyrazole-4-carboxamide (60) According to General Procedure G, the carboxylic acid derivative 18 (44 mg, 0.17 mmol) and benzenesulfo-namide gave the sulfonamide derivative 60 as a white amorphous solid (38 mg, 95.2 μMol, 56%). R 0.40 (20:80 MeOH:DCM). 1H NMR (400 MHz, CDCl3): δ 8.79 (s, 1H), 8.16 (d, J = 7.6 Hz, 2H), 8.00 (s, 1H), 7.69 (t, J = 7.4 Hz, 1H), 7.59 (t, J = 7.6 Hz, 2H), 7.55–7.46 (m, 3H), 7.41-7.35 (m, 2 H) ppm. 13C NMR (100 MHz, CDCl3): δ 157.7, 140.1, 138.1, 134.5, 133.0, 130.4, 129.4, 129.2, 128.7, 126.5, 125.8 (q, J = 0.8 Hz), 118.9 (q, J = 271.7 Hz), 117.7 (q, J = 1.0 Hz) ppm. 19FNMR (376 MHz, CDCl3): δ -56.2 ppm. MS (ESI): [M + H]+ 396.2. HPLC: retention time: 1.96 min (>99%). IR (neat): νmax 3247 (m, N-H), 2921-2850 (w, C–H), 1717 (m, C⩵O), 1563 (w), 1501 (w), 1448 (w), 1425 (m), 1408 (m), 1335 (m), 1294 (m), 1219 (m), 1169 (s), 1147 (s), 1135 (s), 1084 (s), 1020 (m) cm−1. HRMS: calculated for C17H12F3N3O3S [M + H]+ = 396.0630, observed = 396.0636. 1-Phenyl-N-phenylmethanesulfonyl-5-(trifluoromethyl)-1H-pyrazole-4-carboxamide (61) According to General Procedure G, the carboxylic acid derivative 4 (0.11 g, 0.55 mmol) and benzylsulfonamide gave a crude product. The crude product was purified by flash column chromatography (30% EtOAc in petrol) to give the ester derivative 61 as a pale yellow amorphous solid (11.7 mg, 33.0 μMol, 6%). Rf 0.31 (50:50 EtOAc:petrol). 1H NMR (400 MHz, CDCl3): δ 8.00 (s, 1H), 7.65-7.32 (m, 10H), 4.79 (s, 2H), 2.58 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 164.6, 146.4, 141.0, 139.6, 132.0, 130.6, 130.5, 130.4, 129.8, 129.7, 129.5, 126.9, 59.6, 12.1 ppm. MS (ESI): [M + H]+ 356.2. HPLC: retention time: 1.83 min (>99%). IR (neat): νmax 3330 (w, NH), 1682 (s, C⩵O), 1598 (w), 1549 (m), 1503 (m), 1455 (m), 1403 (m), 1337 (s), 1231 (m), 1154 (s), 1135 (m), 1056 (m) cm−1. HRMS: calculated for C18H17N3O3S [M + H]+ = 356.1068, observed = 356.1064. N-(Benzenesulfonyl)-1-(2,6-dimethylphenyl)-5-(trifluoromethyl)-1H-pyrazole-4-carboxamide (62) According to General Procedure G, the carboxylic acid derivative 41 (0.11 g, 0.39 mmol) and benzenesulfonamide gave a crude product. The crude product was purified by flash column chromatography (30-100% EtOAc in petrol) to give the sulfonamide derivative 62 as a white amorphous solid (38 mg, 89.7 μMol, 23%). Rf: 0.34 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 8.14 (s, 1H), 8.05 (d, J =7.6 Hz, 2H), 7.66-7.50 (m, 3H), 7.33 (t, J = 7.6 Hz, 1H), 7.20 (d, J = 7.6 Hz, 2H), 1.93 (s, 6H) ppm. 13C NMR (100 MHz, CDCl3): δ 165.6, 143.4, 142.4, 139.0, 137.2, 133.3, 133.2 (q, J = 39.8 Hz), 131.3, 129.5, 129.3, 128.6, 123.0, 120.6 (q, J = 270.5 Hz), 17.0 ppm. 19F NMR (376 MHz, CDCl3): δ -60.6 ppm. MS (ESI): [M + H] + 424.2. HPLC: retention time: 2.10 min (96%). IR (neat): νmax 3443 (w, N-H), 1701 (m, C⩵O), 1605 (m), 1560 (m), 1485 (m), 1448 (m), 1375 (m), 1304 (s), 1136 (s), 1087 (m), 1048 (m), 969 (m) cm−1. HRMS: calculated for C19H16F3N3O3S [M + H]+ = 424.0942, observed = 424.0947. N-(3-Bromobenzenesulfonyl)-1-(2,6-dimethylphenyl)-5-(trifluor-omethyl)-1H-pyrazole-4-carboxamide (63) According to General Procedure G, the carboxylic acid derivative 41 (0.15 g, 0.54 mmol) and 3-bromobenzene-1-sulfonamide gave a crude product. The crude product was purified by flash column chromatography (30% petrol in EtOAc) to give the sulfonamide derivative 63 as a yellow amorphous solid (21.7 mg, 43.2ˆmol, 8%). Rf: 0.12 (EtOAc). 1H NMR(400 MHz, acetone-d6): δ 8.22 (s, 1H), 8.18 (s, 1H), 8.05 (d, J = 7.9 Hz, 1H), 7.74 (d, J = 7.6 Hz, 1H), 7.47 (t, J = 7.9 Hz, 1H), 7.33 (t, J = 7.6 Hz, 1H), 7.21 (d, J = 7.6 Hz, 2H), 1.91 (s, 6H) ppm. 13C NMR (125 MHz, acetone-d6): δ 147.4, 143.6, 139.8, 136.9, 135.8, 134.9 (q, J =39.2 Hz), 131.7,131.0,130.9,129.3,129.0,128.9,127.1,122.8,120.8 (q, J = 270.2 Hz), 17.3 ppm. 19F NMR (376 MHz, acetone-R): δ -59.0 ppm. MS (ESI): [M - H]- 502.1. HPLC: retention time: 2.30 min (>99%). IR (neat): νmax 3457 (w, N-H), 1706 (s, C⩵O), 1559 (m), 1364 (s), 1294 (s), 1255 (m), 1222 (s), 1175 (s), 1139 (vs), 1098 (m), 968 (m) cm−1. HRMS: calculated for C19H15BRF3N3O3S [M + H]+ = 500.9969, observed = 500.9985. N-(Benzenesulfonyl)-1-(4,5-dichloro-2-methylphenyl)-5-(trifluor-omethyl)-1H-pyrazole-4-carboxamide (64) According to General Procedure G, the carboxylic acid derivative 44 (40 mg, 0.12 mmol) and benzenesulfonamide gave a crude product. The crude product was purified by flash column chromatography (0-50% EtOAc in petrol) to give the sulfonamide derivative 64 as a white amorphous solid (18 mg, 37.6 μMol, 31%). Rf: 0.40 (EtOAc). 1H NMR (400 MHz, CDCl3): δ 8.11 (d, J = 7.8 Hz, 2H), 8.07 (s, 1H), 7.63 (t, J = 7.5 Hz, 1H), 7.52 (app. t, J = 7.8 Hz, 2H), 7.42 (s, 1H), 7.29 (s, 1H), 1.93 (s, 3H) ppm. 13C NMR (100 MHz, CDCl3): δ 159.5, 141.3, 139.0, 137.8, 136.9, 135.8, 135.0, 134.0 (q, J = 3.1 Hz), 133.0 (q, J = 39.8 Hz), 132.4, 130.4, 129.1, 129.0, 128.2, 118.8 (q, J = 271.7 Hz), 16.5 ppm. 19FNMR(376 MHz, CDCl3): δ -57.1 ppm. MS (ESI): [M - H]- 476.1. HPLC: retention time: 2.12 min (>99%). IR (neat): νmax 3267 (m, N-H), 2925 (w, C–H), 1704 (m, C⩵O), 1559 (w), 1492 (w), 1432 (m), 1411 (m), 1335 (w), 1293 (m), 1236 (m), 1165 (s), 1153 (s), 1134 (s), 1080 (m), 1022 (m), 987 (m) cm−1. HRMS: calculated for C18H12Cl2F3N3O3S [M + H]+ = 478.0007, observed = 478.0001. Surface Plasmon Resonance Experiments were performed using as a running buffer that consisted of 10 mM NaPO4 (pH 7) 150 mM NaCl, and 2% DMSO at 25 °C in a Biacore T200 (GE Healthcare). For data analyses, bulk effects were corrected using solvent correction and were performed through the Biacore T200 evaluation software 2.0 (GE Healthcare). Pa MurB was covalently coupled to a CM5 chip (GE Healthcare) by standard amine coupling protocol. For the single concentration experiment, all fragments were diluted to 1 mM in running buffer injected for 30 s at 30 μL s−1 and the dissociation was for 320 s. All fragments were tested two times in reverse orders. Sensograms were visually inspected, and fragments with significant signal increase comparing with the original fragment were selected for affinity study by ITC. Isothermal Titration Calorimetry Isothermal titration calorimetry experiments to quantify ligand binding to Pa MurB were performed using a Malvern MicroCal Auto-iTC200 system at 25 °C. Titrations consisted of an initial injection of 0.4 μL, which was discarded during data processing, followed by 28 further injections of 1.5 μL separated by a 120 s interval. The Pa MurB protein was dialyzed overnight at 4 °C in 25 mM Tris-HCl (pH 8.0) and 150 mM NaCl. Sample cell and syringe solutions were prepared using the same buffer, with a final DMSO-R concentration of 5%. Pa MurB concentrations of 200-50 pM were used, with ligand concentrations of3.0-0.5 mM. The protein well had a volume of 400 μL, the ligand well a volume of 200 μL, and the blank well a volume of 400 μL. Titrations were fitted with Origin software (OriginLab, Northampton, MA, USA) using a one site binding model. All ITC titration curves are shown in the SI. Dihedral Angle Calculations The global ground state conformations and dihedral angle calculations were performed using Schrodinger Maestro 11.60 The scanning of the dihedral angles was performed using a MacroModel coordinate scan (force field: OPLS-2005, solvent: water, default settings). Supplementary Material Molecular formula strings Supplementary Material ■ Acknowledgments The authors would like to thank the Diamond Light Source for beam-time (proposals mx18548) and the staff of beamlines I03 and I04-1 for assistance with data collection. Funding M.A.-G.-D.-E. was supported by American leprosy Missions Grant (G88726). J.M.-L., S.Y.K., and O.D.P. were funded by the Cystic Fibrosis Trust and Fondation Botnar (grant no. 6063). C.M. was funded by the Bill and Melinda Gates Foundation, Hit-TB (OPP1024021). J.H. was funded by the Swiss National Science Foundation (SNSF Early PostDoc. Mobility Fellowship P2ZHP2_164947) and the Marie Curie Research Grant Scheme, EU H2020 Framework Programme (H2020-MSCA-IF-2017, ID: 789607). V.M. was funded by the Bill and Melinda Gates Foundation, Hit-TB (OPP1024021), and SHORTEN-TB (OPP1158806). K.P.B. and R.A.F. were funded by NIHR Cambridge Biomedical Research Centre. C.A., R.A.F., and T.L.B. were funded by the UK Cystic Fibrosis Trust (SRC010). R.A.F. was funded by the Wellcome Trust 107032AIA, and R.A.F. and T.L.B. were founded by the UK Cystic Fibrosis Trust (Innovation Hub grant 001). ■ Abbreviations CF cystic fibrosis DSF differential scanning fluorimetry EDC 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide Ec Escherichia coli FAD flavin adenine dinucleotide ITC isothermal titration calorimetry LE ligand efficiency MIC minimum inhibitory concentration MurB UDP-N-acetylenolpyruvoyl-glucosamine reductase Mw molecular weight NADP+ nicotinamide adenine dinucleotide phosphate Pa Pseudomonas aeruginosa PBPs penicillin-binding proteins RU response units Sa Staphylococcus aureus SAR structure–activity relationship SPR surf;ace plasmon resonance UDP uridine diphosphate UNAGEP UDP-N-acetylglucosamine enolpyruvate UNAM N-acetyl-muramic acid Figure 1 MurB enzymatic reaction. Figure 2 Catalytic pocket and domain distribution of MurB, illustrating domain I in red, domain II in blue, and domain III in pink. (a) Pa MurB in complex with NADP+ (PDB code: 4JB1).12 (b) Ec MurB in complex with UNAGEP (PDB code: 2MBR).19 FAD is shown in yellow, NADP+ in cyan, and UNAGEP in green. Figure 3 X-ray structure of fragment 4 (dark blue) (a, b, PDB code: 7OR2) and NADP+ (cyan) (a, PDB code: 4JB1)12 bound to the active site of Pa MurB. (a) Superimposition of Pa MurB in complex with fragment 4 (dark blue) and Pa MurB in complex with NADPH (cyan). (b) X-ray structure of fragment 4 bound to the active site of Pa MurB. FAD is depicted in yellow. Arpeggio27 was used to analyze the interactions. Hydrogen bonds are shown as red dashed lines, π–π interactions are indicated in green dashed lines, and water molecules are shown as red dots. The final 2F0-Fc map around the ligand is shown in blue at 1σ. Figure 4 Hit plot displaying results of analogues in SPR assay. Hits were characterized as higher RU (response units) normalized than fragment 4. Analogues in yellow showed curves that they did not reach equilibrium indicating aggregation and were ruled out as hits. Mw, molecular weight. Figure 5 X-ray structures of fragment 4 and fragment 18 bound to the active site of Pa MurB. (a) Superimposition of Pa MurB in complex with fragment 4 (dark blue) and with fragment 18 (white). (b) X-ray structure of fragment 18 bound to the active site of Pa MurB. FAD is depicted in yellow. Arpeggio27 was used to analyze the interactions. Hydrogen bonds are shown as red dashed lines, π–π interactions are indicated in green dashed lines, and water molecules are shown as red dots. The final 2F0–Fc map around the ligand is shown in blue at 1σ. Figure 6 X-ray structure of fragment 4 and fragment 19 bound to the active site of Pa MurB. (a) Superimposition of Pa MurB in complex with fragment 4 (dark blue) and with fragment 19 (orange). (b) X-ray structure of fragment 19 bound to the active site of Pa MurB. FAD is depicted in yellow. Arpeggio27 was used to analyze the interactions. Hydrogen bonds are shown as red dashed lines, π–π interactions are indicated in green dashed lines, and water molecules are as red dots. The final 2F0–Fc map around the ligand is shown in blue at 1σ. Figure 7 ITC titration curves at key stages of optimization for (a) pyrazole and (b) pyrrole series. Titrations performed at 50 μM Pa MurB with 3.0 mM fragments, except for fragments 44 and M 54 where 1.0 mM was used. See the SI for titration curves of other fragments and NADP+. Figure 8 (a) Comparison of the dihedral angles between the conformation of fragment 18 in complex with Pa MurB and (b) minimum energy conformations of fragments 18, 41, and 53. See the SI for minimum energy conformations of other fragments. Scheme 1 General Synthetic Routes for the Synthesis of the Tested Fragments Table 1 Hits from Differential Scanning Fluorimetry at Ligand Concentrations of 5.0 and 1.0 mMa Fragment ΔTM(°C) (±SEM) 5 mM1 mM Fragment ΔTM(°C) (±SEM) 5 mM 1 mM Fragment ΔTM(°C) (±SEM) 5 mM 1 mM 1 +3.5 +1.0(0) 6 +1.5 +1.0(0) 11 +1.0 +0.2(0.2) 2 +4.0 +1.0(0) 7 +1.0 +0.0(0) 12 +1.0 +0.0(0.0) 3 +2.0 +1.5(0) 8 +1.0 +0.0(0) 13 +1.0 +0.0(0.0) 4 +1.5 +0.5(0) 9 +1.0 +0.5(0) 14 +1.0 +0.5(0) 5 +1.5 +1.0(0) 10 +1.0 +0.0(0) a Fragment hits show shifts in protein melting temperatures (ΔTM) from DMSO-d6 at 10 μM Pa MurB concentration to the two different ligand concentrations. Each ligand is screened at concentrations of 5.0 mM (n = 1) and 1.0 mM (n = 3). Table 2 Structure and Biophysical Data for Fragments 4 and 15–18a Fragment R ΔTM(°C) (±SEM) 1.0 mM SPRRU>RUf4 1.0 mM Kd (mM) ITC LE 4 Me +0.5 (0) – 2.88 ± 0.25 0.23 15 NH2 0.0 (0) no – – 16 OH +0.7 (0.2) – No binding – 17 CH2OH +0.2 (0.2) – – – 18 CF3 +1.0 (0) no 0.25 ± 0.04 0.27 a Shift in protein melting temperatures (ΔTM) from DMSO-d6 at 10 μM Pa MurB concentration and at 1.0 mM fragment concentrations (n = 3). SPR RU of each fragment at 1 mM in comparison with RU of fragment 4 (RU > RUf4, RU = response units) (n = 2). Kd calculated using ITC (50 μM Pa MurB, 3.0 mM fragment). Ligand efficiencies were calculated as LE = −(RTlnKd)/(number of heavy atoms) and are reported in kcal/mol per heavy atom. Dash entries in the table mean not measured. Table 3 Biophysical Data for Fragments 19–23a Fragment R1 R2 X ΔTM (°C) (±SEM) 1.0 mM SPR RU>RUf4 1.0 mM Kd(μM) ITC LE 19 - +0.5 (0) no 877 ±154 0.28 20 Me Me CH +2.0 (0.2) yes 110+14 0.34 21 Me Me N +0.7 (0.2) no – – 22 CFs Me CH +2.0 (0) yes 112+4.4 0.28 23 Me CH2CH2OH CH +2.0 (0) – 280+15 0.28 a Shift in protein melting temperatures (ΔTM) from DMSO-d6 at a ligand concentration of 1.0 mM and 10 μM MurB concentration (n = 3). SPR RU of each fragment at 1 mM in comparison with RU of fragment 4 (RU > RUf4, RU = response units) (n = 2). Kd calculated using ITC (50 μM MurB, 3.0 mM fragment). Ligand efficiencies were calculated as LE = –(RTlnKd)/(number of heavy atoms) and are reported in kcal/mol per heavy atom. Dash entries in the table mean not measured. Table 4 Biophysical Data for Fragments 24-43a Fragment R ΔTM (°C) (±SEM) 1.0 mM SPR RU>RUf4 1.0 mM Kd(μM) ITC LE 24 benzyl +1.0(0) – – – 25 thiophen-2-yl)methyl +1.0(0) – – – 26 thiophen-2-yl +0.3 (0.2) – – – 27 3-pyridinyl +0.3 (0.2) – – – 28 4-pyridinyl 0.0(0) – – – 29 2-fluorophenyl +1.0(0) – – – 30 2-chlorophenyl +1.5(0) yes 148+15 0.27 31 2-bromophenyl +2.0(0) yes 85,.5 ± 7.7 0.29 32 2-methylphenyl +2.0(0) yes 44.6 + 3.1 0.31 33 2-trifluoromethylphenyl +1.0(0) – – – 34 4-trifluoromethylphenyl +1.0(0) – – – 35 4-methoxyphenyl +2.0(0) no – – 36 2-methyl-4-nitrophenyl +1.7(0.7) yes 208 ± 28 0.23 37 3-carbamoylphenyl +0.5(0) – – – 38 2,3-dimethylphenyl +2.0(0) yes 99.0 ± 7.2 0.27 39 2,4-dimethylphenyl +2.3(0.2) yes 85.5 ± 8.0 0.28 40 2,5-dimethylphenyl +2.0(0) yes 138 + 16 0.26 41 2,6-dimethylphenyl +2.5(0) – 25.8 + 5.5 0.31 42 2,6-dichlorophenyl 2.5(0) yes 60.2 ± 4.4 0.29 43 3,4-dichlorophenyl +2.3(0.2) yes 26.1+2.7 0.31 a Shift in protein melting temperatures (ΔTM) from DMSO at 1.0 mM fragment concentration and 10 μM MurB concentration (n = 3). SPR RU of each fragment at 1 mM in comparison with RU of fragment 4 (RU > RUf4, RU = response units) (n = 2). Kd calculated using ITC (50 μM MurB, 3.0 mM fragment; except for fragment 41, where 200 μM MurB was used). Ligand efficiencies were calculated as LE = −(RTlnKd)/(number of heavy atoms) and are reported in kcal/mol per heavy atom. Dash entries in the table mean not measured. Table 5 Biophysical Data for Fragments 44-55a Fragment R ΔTM (°C) (±SEM) 1.0 mM SPR RU>RUf4 1.0 mM Kd(μM) ITC LE 44 2-methyl-4,5-dichlorophenyl +5.0 (0) yes 3.57 ± 0.76 0.35 45 2-methyl-3,4-dichlorophenyl +3.7 (0.2) yes 11.3 + 2.5 0.32 46 2-methylphenyl +2.0 (0) yes 64.5 + 6.1 0.34 47 2,6-dimethylphenyl +3.0 (0) no 62.5 ± 3.5 0.32 48 3,4-dichlorophenyl +2.3 (0.2) no 47.8 ± 3.8 0.33 49 3-chlorophenyl +1.5 (0) yes – – 50 4-chlorophenyl +1.5 (0) no – – 51 2-methyl-4,5-dichlorophenyl +3.5 (0) – 24.1+4.0 0.33 52 2-methyl-3,4-dichlorophenyl +4.0 (0.2) – 24.3 ± 7.4 0.33 53 2-methylphenyl +2.5 (0) yes 40.2+1.4 0.30 54 2-methyl-4,5-dichlorophenyl +4.7 (0.2) – 8.00+1.1 0.32 55 2-methyl-3,4-dichlorophenyl +2.7 (0.2) yes 11.4 + 3.9 0.31 NADP+ – +3.5 (0) – 23.6 ± 2.4 – NADPH – +3.0 (0) – – – a Shift in protein melting temperatures (ΔTM) from DMSO-d6 at 1.0 mM fragment concentration and 10 μM Pa MurB concentration (n = 3). SPR RU of each compound at 1.0 mM in comparison with RU of fragment 4 (RU > RUf4, RU = response units) (n = 2). Kd calculated using ITC (50 μM Pa MurB, 3.0 mM fragment; except for fragments 44, 51, and 54, which were tested at 1.0 mM, and fragments 45 and 52, which were tested at 0.5 mM). Ligand efficiencies were calculated as LE = −(RTlnKd)/(number of heavy atoms) and are reported in kcal/mol per heavy atom. Dash entries in the table mean not measured. Table 6 Biophysical Data for Fragments 56–64a Fragment R X Y ΔTM (°C) (±SEM) 1.0 mM SPR RU>RUf4 1.0 mM Kd(μM) ITC LE 56 NH2 CF3 H 0.0 (0) – – – 57 OEt CF3 CL 0.0 (0) – – – 58 N-benzylamino Me H 0.0 (0) – – – 59 N-(methanesulfonyl)amino CF3 H +0.5 (0) no – – 60 N-(benzenesulfonyl)amino CF3 H +1.0 (0) yes 324 ± 59 0.18 61 N-(benzylsulfonyl)amino Me H 0.0 (0) – – – R1 R2 62 phenyl 2,6-dimethyl phenyl +3.0 (0) yes 25.6 + 2.9 0.22 63 3-bromophenyl +1.3 (0.2) – – – 64 phenyl 2-methyl-4,5-dichlorophenyl +2.7 (0.3) – 12.0 + 3.7 0.22 a Shift in protein melting temperatures (ΔTM) from DMSO-d6 at 1.0 mM fragment concentration and 10 μM Pa MurB concentration (n = 3). SPR RU of each fragment at 1.0 mM in comparison with RU of fragment 4 (RU > RUf4, RU = response units) (n = 2). Kd calculated using ITC (50 μM Pa MurB, 3.0 mM fragment). Ligand efficiencies were calculated as LE = −(RTlnKd)/(number of heavy atoms) and are reported in kcal/mol per heavy atom. Dash entries in the table mean not measured. Author Contributions ±M.A.-G.-D.-E. and J.M.-L. contributed equally. M.A.-G.-D.-E. carried out the structural biology and biochemical studies. J.M.-L., M.H., and C.M. synthesized the compounds used in this study. S.Y.K. cloned Pa MurB and designed the purification protocol. S.Y.K. and O.D.P. carried out the fragment screening of the 960-fragment library by thermal shift. J.M.-L. carried out thermal shift and ITC, and M.A.-G.-D.-E. carried out SPR. J.H. carried out the computational studies. M.A.-G.-D.-E. and J.M.-L. wrote the manuscript with contributions from A.G.C. and T.L.B. C.A., A.G.C., T.L.B., V.M., and R.A.F. supervised the project. All authors have given approval to the final version of the manuscript. Notes The authors declare no competing financial interest. Accession Codes Atomic coordinates for the X-ray structures of fragments 4 (PDB code: 7OR2), 18 (PDB code: 7ORZ), and 19 (PDB code: 7OSQ) are available from the RCPB Protein Data Bank (http://www.rcpb.org). Authors will release the atomic coordinates and experimental data upon article publication. (1) Aloush V Navon-Venezia S Seigman-Igra Y Cabili S Carmeli Y Multidrug-Resistant Pseudomonas aeruginosa: Risk Factors and Clinical Impact Antimicrob Agents Chemother 2006 50 43 48 16377665 (2) Rajan S Pulmonary Infections in Patients with Cystic Fibrosis Semin Respir Infect 2002 17 47 56 11891518 (3) Coutinho HDM Falcão-Silva VS Gonçalves FG Pulmonary Bacterial Pathogens in Cystic Fibrosis Patients and Antibiotic Therapy: A Tool for the Health Workers Int Arch Med 2008 1 24 18992146 (4) Lyczak JB Cannon CL Pier GB Lung Infections Associated with Cystic Fibrosis Clin Microbiol Rev 2002 15 194 222 11932230 (5) Gellatly SL Hancock REW Pseudomonas aeruginosa: New Insights into Pathogenesis and Host Defenses Pathog Dis 2013 67 159 173 23620179 (6) WHO Antimicrobrial Resistance Global Report on Surveillance 2014 (7) Helen WB George HT John SB John EE Gilbert D Rice LB Scheld M Spellberg B Bugs JBB No Drugs: No ESKAPE! An Update from the Infectious Diseases Society of America Clin Infect Dis 2009 48 1 12 19035777 (8) Antibiotic Resistance Threats in the United States 2013 (9) Döring G Flume P Heijerman H Elborn JS Treatment of Lung Infection in Patients with Cystic Fibrosis: Current and Future Strategies J Cyst Fibros 2012 11 461 479 23137712 (10) Anderson JS Matsuhashi M Haskin MA Strominger JL Biosythesis of the Peptidoglycan of Bacterial Cell Walls. II. Phospholipid Carriers in the Reaction Sequence J Biol Chem 1967 242 3180 3190 6027793 (11) Bugg TDH Walsh CT Intracellular Steps of Bacterial Cell Wall Peptidoglycan Biosynthesis: Enzymology, Antibiotics, and Antibiotic Resistance Nat Prod Rep 1992 9 199 215 1436736 (12) Chen MW Lohkamp B Schnell R Lescar J Schneider G Substrate Channel Flexibility in Pseudomonas aeruginosa MurB Accommodates Two Distinct Substrates PLoS One 2013 8 e66936 23805286 (13) El Zoeiby A Beaumont M Dubuc E Sanschagrin F Voyer N Levesque RC Combinatorial Enzymatic Assay for the Screening of a New Class of Bacterial Cell Wall Inhibitors Bioorg Med Chem 2003 11 1583 1592 12628682 (14) El Zoeiby A Sanschagrin F Levesque RC Structure and Function of the Mur Enzymes: Development of Novel Inhibitors Mol Microbiol 2003 47 1 12 12492849 (15) Gautam A Vyas R Tewari R Peptidoglycan Biosynthesis Machinery: A Rich Source of Drug Targets Crit Rev Biotechnol 2011 31 295 336 21091161 (16) De Groote VN Fauvart M Kint CI Verstraeten N Jans A Cornelis P Michiels J Pseudomonas aeruginosa Fosfomycin Resistance Mechanisms Affect Non-Inherited Fluoroquinolone Tolerance J Med Microbiol 2011 60 329 336 21212150 (17) Benson TE Marquardt JL Marquardt AC Etzkorn FA Walsh CT Overexpression, Purification, and Mechanistic Study of UDP-N-Acetylenolpyruvylglucosamine Reductase Biochemistry 1993 32 2024 2030 8448160 (18) Lovering AL Safadi SS Strynadka NCJ Structural Perspective of Peptidoglycan Biosynthesis and Assembly Annu Rev Biochem 2012 81 451 478 22663080 (19) Benson TE Walsh CT Hogle JM X-Ray Crystal Structures of the S229A Mutant and Wild-Type MurB in the Presence of the Substrate Enolpyruvyl-UDP-N-Acetylglucosamine at 1.8-Å Resolution Biochemistry 1997 36 806 811 9020778 (20) Benson TE Harris MS Choi GH Cialdella JI Herberg JT Martin JP Baldwin ET A Structural Variation for MurB: X-Ray Crystal Structure of Staphylococcus aureus UDP-N-Acetylenolpyruvylglucosamine Reductase (MurB) Biochemistry 2001 40 2340 2350 11327854 (21) Kim M-K Cho MK Song H-E Kim D Park B-H Lee JH Kang GB Kim SH Im YJ Lee D-S Eom SH Crystal Structure of UDP-N-Acetylenolpyruvylglucosamine Reductase (MurB) from Thermus caldophilus Proteins 2007 66 751 754 17120230 (22) Dhalla AM Yanchunas J Jr Ho H-T Falk P Villafranca J Robertson JG Steady-State Kinetic Mechanism of Escherichia coli UDP-N-Acetylenolpyruvylglucosamine Reductase Biochemistry 1995 34 5390 5402 7727397 (23) Katz A Caufield C Structure-Based Design Approaches to Cell Wall Biosynthesis Inhibitors Curr Pharm Des 2003 9 857 866 12678870 (24) Andres CJ Bronson JJ D’Andrea SV Deshpande MS Falk PJ Grant-Young KA Harte WE Ho H-T Misco PF Robertson JG Stock D 4-Thiazolidinones: Novel Inhibitors of the Bacterial Enzyme MurB Bioorg Med Chem Lett 2000 10 715 717 10782671 (25) Yang Y Severin A Chopra R Krishnamurthy G Singh G Hu W Keeney D Svenson K Petersen PJ Labthavikul P Shlaes DM 3,5-Dioxopyrazolidines, Novel Inhibitors of UDP-N-Acetylenolpyruvylglucosamine Reductase (MurB) Bacteria, with Activity against Gram-Positive Am Soc Microbiol 2006 50 556 564 (26) Sarver RW Rogers JM Epps DE Determination of Ligand-MurB Interactions by Isothermal Denaturation: Application as a Secondary Assay to Complement High Throughput Screening J Biomol Screen 2002 7 21 28 11897052 (27) Jubb HC Higueruelo AP Ochoa-Montaño B Pitt WR Ascher DB Blundell TL Arpeggio: A Web Server for Calculating and Visualising Interatomic Interactions in Protein Structures J Mol Biol 2017 429 365 371 27964945 (28) Hudson SA Surade S Coyne AG Mclean KJ Leys D Munro AW Abell C Overcoming the Limitations of Fragment Merging: Rescuing a Strained Merged Fragment Series Targeting Mycobacterium tuberculosisCYP121 ChemMedChem 2013 8 1451 1456 23788280 (29) Menozzi G Mosti L Schenone P Reaction of 2-Dimethylaminomethylene-1,3-Diones with Dinucleophiles. VI. Synthesis of Ethyl or Methyl 1,5-Disubstituted 1H-Pyrazole-4-Carboxylates J Heterocycl Chem 1987 24 1669 1675 (30) Kabsch W XDS Acta Crystallogr D Biol Crystallogr 2010 66 125 132 20124692 (31) Winn MD Ballard CC Cowtan KD Dodson EJ Emsley P Evans PR Keegan RM Krissinel EB Leslie AGW McCoy A McNicholas SJ Overview of the CCP4 Suite and Current Developments Acta Crystallogr D Biol Crystallogr 2011 67 235 242 21460441 (32) McCoy AJ Grosse-Kunstleve RW Adams PD Winn MD Storoni LC Read RJ Phaser Crystallographic Software J Appl Crystallogr 2007 40 658 674 19461840 (33) Liebschner D Afonine PV Baker ML Bunkóczi G Chen VB Croll TI Hintze B Hung L-W Jain S McCoy AJ Moriarty NW Macromolecular Structure Determination Using X-Rays, Neutrons and Electrons: Recent Developments in Phenix Acta Crystallogr Sect D Struct Biol 2019 75 861 877 31588918 (34) Emsley P Lohkamp B Scott WG Cowtan K Features and Development of Coot Acta Crystallogr, D Biol Crystallogr 2010 66 486 501 20383002 (35) Raeppel SL Raeppel F Therrien E Design and Synthesis of Close Analogs of LCRF-0004, a Potent and Selective RON Receptor Tyrosine Kinase Inhibitor Bioorg Med Chem Lett 2015 25 2527 2531 25953155 (36) Osman NA Ligresti A Klein CD Allarà M Rabbito A Di Marzo V Abouzid KA Abadi AH Discovery of Novel Tetrahydrobenzo[b]Thiophene and Pyrrole Based Scaffolds as Potent and Selective CB2 Receptor Ligands: The Structural Elements Controlling Binding Affinity, Selectivity and Functionality Eur J Med Chem 2016 122 619 634 27448919 (37) Mahy W Patel M Steadman D Woodward HL Atkinson BN Svensson F Willis NJ Flint A Papatheodorou D Zhao Y Vecchia L Screening of a Custom-Designed Acid Fragment Library Identifies 1-Phenylpyrroles and 1-Phenylpyrrolidines as Inhibitors of Notum Carboxylesterase Activity J Med Chem 2020 63 9464 9483 32787107 (38) Seo HJ Park E-J Kim MJ Kang SY Lee SH Kim HJ Lee KN Jung ME Lee M Kim M-S Son E-J Design and Synthesis of Novel Arylpiperazine Derivatives Containing the Imidazole Core Targeting 5-HT2A Receptor and 5-HT Transporter J Med Chem 2011 54 6305 6318 21823597 (39) Kong D Xue T Guo B Cheng J Liu S Wei J Lu Z Liu H Gong G Lan T Hu W Optimization of P2Y12 Antagonist Ethyl 6-(4-((Benzylsulfonyl)Carbamoyl)Piperidin-1-Yl)-5-Cyano-2-Methylnicotinate (AZD1283) Led to the Discovery of an Oral Antiplatelet Agent with Improved Druglike Properties J Med Chem 2019 62 3088 3106 30843696 (40) Guo N Liu X Xu H Zhou X Zhao H A Simple Route towards the Synthesis of 1,4,5-Trisubstituted 1,2,3-Triazoles from Primary Amines and 1,3-Dicarbonyl Compounds under Metal-Free Conditions Org Biomol Chem 2019 17 6148 6152 31187848 (41) He W Liu R Liao Y Ding G Li J Liu W Wu L Feng H Shi Z He M A New 1,2,3-Triazole and Its Rhodamine B Derivatives as a Fluorescence Probe for Mercury Ions Anal Biochem 2020 598 113690 32259511 (42) Peng Y Luo J Feng Q Tang Q Understanding the Scope of Feist−Bénary Furan Synthesis: Chemoselectivity and Diastereoselectivity of the Reaction Between Alpha-Halo Ketones and Beta-Dicarbonyl Compounds Eur J Org Chem 2016 5169 5179 (43) Lu B Wu X Li C Ding G Li W Xie X Zhang Z Highly Diastereo- and Enantioselective Access to Syn-α-Amido Β-Hydroxy Esters via Ruthenium-Catalyzed Dynamic Kinetic Resolution-Asymmetric Hydrogenation J Org Chem 2019 84 3201 3213 30776889 (44) Lange JHM van der Neut MAW Wals HC Kuil GD Borst AJM Mulder A den Hartog AP Zilaout H Goutier W van Stuivenberg HH van Vliet BJ Synthesis and SAR of Novel Imidazoles as Potent and Selective Cannabinoid CB2 Receptor Antagonists with High Binding Efficiencies Bioorg Med Chem Lett 2010 20 1084 1089 20031412 (45) Smith JO Mandal BK Filler R Beery JW Reaction of Ethyl 4,4,4-Trifluoroacetoacetate Enolate with 3-Bromo-1,1,1-Trifluoroacetone: Synthesis of 2,4-Bis(Trifluoromethyl)Furan J Fluorine Chem 1997 81 123 128 (46) Gathirwa JW Maki T Benzylation of Hydroxy Groups with Tertiary Amine as a Base Tetrahedron 2012 68 370 375 (47) Jakob B Voss G Gerlach H Synthesis of (S)- and (R)-3-Hydroxyhexadecanoic Acid Tetrahedron: Asymmetry 1996 7 3255 3262 (48) Nuss J Williams M Mohan R Martin R Wang T-L Tsuruoka H Aoki K Honzumi M Asoh Y Saito K Homma T 1-Phenylpyrrole Compounds 2010 (49) Wang Z Wang M Yao X Li Y Tan J Wang L Qiao W Geng Y Liu Y Wang Q Design, Synthesis and Antiviral Activity of Novel Pyridazines Eur J Med Chem 2012 54 33 41 22608761 (50) Charpe VP Hande AA Sagadevan A Hwang KC Visible-Light Induced Copper(I)-Catalysed Denitrogenative Oxidative Coupling of Hydrazinylpyridines with Terminal Alkynes Green Chem 2018 20 4859 4864 (51) Lu T Connolly PJ Cummings MD Barbay JK Kreutter KD Wu T Diels GSM Thuring JW Philippar U Edwards JP Berthelot DJ-C Pyrazole Derivatives as Malti Inhibitors. U.S. Patent Application 16/440 2019 302 (52) Gilmore JL Sheppeck JE Sphingosine-1-Phosphate Receptor Agonists 2011 (53) Wu Z Park HY Xie D Yang J Hou S Shahzad N Kim CK Yang S Synthesis, Biological Evaluation, and 3D-QSAR Studies of N-(Substituted Pyridine-4-Yl)-1-(Substituted Phenyl)-5-Trifluoromethyl-1 H-Pyrazole-4-Carboxamide Derivatives as Potential Succinate Dehydrogenase Inhibitors J Agric Food Chem 2021 69 1214 1223 33480684 (54) Schlosser M Volle JN Leroux F Schenk K Switchable Reactivity: The Site-Selective Functionalization of Trifluoromethyl-Substituted Pyrazoles Eur J Org Chem 2002 2913 2920 (55) Beck JR Wright FL Synthesis of 1-Aryl-5-(Trifluoromethyl)-1H-Pyrazole-4-Carboxylic Acids and Esters J Heterocycl Chem 1987 24 739 740 (56) Miles DH Guasch J Toste FD A Nucleophilic Strategy for Enantioselective Intermolecular Α-Amination: Access to Enantioenriched Α-Arylamino Ketones J Am Chem Soc 2015 137 7632 7635 26066512 (57) Pinto DJP Han W Hu Z Qiao J Glycinamides as Factor XA Inhibitors 2003 (58) Bock BL Adams R Stereochemistry of Phenyl Pyrrols XIX J Am Chem Soc 1931 53 3519 3522 (59) Giacomelli G Porcheddu A Salaris M Taddei M Microwave-Assisted Solution-Phase Synthesis of 1,4,5-Trisubstituted Pyrazoles Eur J Org Chem 2003 537 541 (60) Schrödinger Release 2021−1 Maestro, Schrödinger LLC New York, NY 2021

PMC007xxxxxx/PMC7614805.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0314024 Int J Parasitol Int J Parasitol International journal for parasitology 0020-7519 1879-0135 31153899 7614805 10.1016/j.ijpara.2019.04.001 EMS181173 Article Evolutionary sex allocation theory explains sex ratios in natural Plasmodium falciparum infections Schneider Petra petra.schneider@ed.ac.uk a* Babiker Hamza A. hbabiker@squ.edu.om b Gadalla Amal A. H. GadallaA1@cardiff.ac.uk bc Reece Sarah E. sarah.reece@ed.ac.uk a a Institute of Evolutionary Biology & Institute of Immunology and Infection Research, School of Biological Sciences, University of Edinburgh, Charlotte Auerbach Road, Edinburgh EH9 3FL, United Kingdom b College of Medicine and Health Sciences, Sultan Qaboos University, PO Box 35, Alkhoud, 123 Sultanate of Oman c Division of Population Medicine. School of Medicine, Cardiff University, Cardiff CF14 4XN, United Kingdom * Corresponding author: Petra Schneider, Institute of Evolutionary Biology & Institute of Immunology and Infection Research, Ashworth room 412, School of Biological Sciences, University of Edinburgh, Charlotte Auerbach Road, Edinburgh EH9 3FL, United Kingdom. Tel: +441316519219; Fax +441316506564; petra.schneider@ed.ac.uk 01 7 2019 31 5 2019 18 7 2023 25 7 2023 49 8 601604 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Malaria transmission is achieved by sexual stages, called gametocytes, and the proportion of gametocytes that are male versus female (sex ratio) influences transmission success. In malaria model systems, variation in gametocyte sex ratios can be explained by the predictions of evolutionary sex allocation theory. We test these predictions using natural Plasmodium falciparum infections. The predicted negative correlation between sex ratio and gametocyte density holds: sex ratio increases when gametocyte densities decrease, and this is most apparent in single genotype infections and in the dry season. We do not observe higher gametocyte sex ratios in mixed compared to single genotype infections. Fertility insurance local mate competition sex allocation gametocyte density seasonal malaria transmission competition pmcMalaria parasites (Plasmodium ssp.) replicate asexually in their vertebrate hosts, but require a single round of sexual reproduction to achieve between-host transmission via mosquito vectors. Sexual stages, called gametocytes, are formed in the vertebrate host and both the density and sex ratio (defined as the proportion that are male) of gametocytes shape parasite fitness/transmission (Reece et al., 2008; Mitri et al., 2009; Bradley et al., 2018). Gametocyte sex ratios of malaria parasites vary considerably, both during infections and in different hosts. Studies using a variety of model systems (rodent and lizard parasites, and the human parasite Plasmodium falciparum in vitro) have demonstrated that variation in gametocyte sex ratios is influenced by parasite genotype, genetic diversity of infections (multiplicity of infection, MOI), transmission-blocking immunity, and the density of red blood cells, reticulocytes and gametocytes (Paul et al., 2000; West et al., 2001; Paul et al., 2002; Reece et al., 2005; Reece et al., 2008; Ramiro et al., 2011). Evolutionary sex allocation theory can explain why parasite fitness is maximised by adjusting sex ratio in response to the above factors (West et al., 2001; Paul et al., 2002; Gardner et al., 2003; Reece et al., 2008; Neal and Schall, 2014). Evolutionary sex allocation theory predicts that each clone within an infection should adjust its gametocyte sex ratio in response to number of co-infecting clones within the host, gametocyte density and the number of gametes produced by each male gametocyte (Nee et al., 2002; Gardner et al., 2003). Specifically, in single genotype infections, the number of offspring is maximised (and thus, competition for females is minimised; “local mate competition theory”) when just enough males are produced to fertilise all females. For malaria parasites this means a female-biased gametocyte sex ratio (up to 8 females per male) because female and male gametocytes each transform into one female gamete or up to eight male gametes, respectively. However, malaria infections often contain multiple parasite genotypes and males can fertilise females from their own as well as unrelated genotypes. Whereas fitness for parasites in single genotype infections is determined by the total number of offspring, the fittest genotype for outbreeding parasites in mixed genotype infections is that with the highest genetic representation in the next generation (i.e. that with the highest proportion of matings). Thus, higher sex ratios are favoured in mixed genotype infections because each genotype can take advantage of the females belonging to co-infecting genotypes by producing male gametocytes, the more fecund sex (Figure S1A). However, at low gametocyte densities (e.g. due to low gametocyte conversion or during anaemia) or when male gametocyte survival and/or fecundity is low (e.g. due to immunity or drugs, which both can disproportionally affect males), the probability that a blood meal contains enough males to fertilise all females is reduced (Paul et al., 2002; Ramiro et al., 2011; Delves et al., 2013). Therefore, parasites should compensate by increasing their investment in males (West et al., 2001; Gardner et al., 2003) (Figure S1B). Experiments with P. chabaudi in vivo and P. falciparum in vitro are consistent with predictions that higher sex ratios are produced in mixed genotype infections, and when fertility insurance is required to avoid male limitation (Paul et al., 2002; Reece et al., 2008; Mitri et al., 2009). Data from natural P. falciparum infections are scarce, but higher sex ratios are observed in mixed genotype infections and in anaemic hosts (Read et al., 1992; Robert et al., 2003; Sowunmi et al., 2009a; Sowunmi et al., 2009b). Other studies - generally related to Plasmodium species infecting lizards - do not support fertility insurance theory but these produce high gametocyte densities and therefore may not require fertility insurance (Neal and Schall, 2014). However, whether sex ratios are adjusted according to variation in the genetic diversity of infections and male limitation have rarely been tested simultaneously in natural infections with parasite species expected to adopt both sex allocation strategies (Neal and Schall, 2014). Here, we examine whether the rules of sex allocation theory apply to the gametocyte sex ratios of P. falciparum in natural infections. Until the recent development of sex-specific RT-qPCR assays (Schneider et al., 2015; Santolamazza et al., 2017; Stone et al., 2017), methods to determine the sex ratio of P. falciparum gametocytes were laborious and suffered from low accuracy, particularly at low densities. We take advantage of sex-specific RT-qPCR assays to test whether gametocyte sex ratios in natural P. falciparum infections in Asar village, Eastern Sudan, follow the predictions of theories for local mate competition and fertility insurance. In this region, transmission is distinctly seasonal (September–December), chronic infections may persist asymptomatically throughout the year and mixed genotype infections are common (Gadalla et al., 2016). Ethical clearance was granted by the Ethical Committee of the Sudanese Ministry of Health. A detailed description of the study and methods used has been previously published (Gadalla et al., 2016). Briefly, individuals with symptomatic P. falciparum infections were enrolled after informed consent was obtained. Monthly blood samples were collected between November 2001 and October 2002, and another sample in December 2002. Symptomatic infections were treated with antimalarial drugs after samples were taken, according to national treatment guidelines at the time of the study. Therefore, we investigate sex ratios independently of any impact of drugs. Thirty one individuals had at least 3 parasite positive samples during the study and produced gametocytes, resulting in a total of 98 gametocyte-positive samples. Gametocytes, sex ratio and MOI (minimum number of genotypes present) were quantified by sex-specific RT-qPCRs and Pfg377 genotyping respectively (Abdel-Wahab et al., 2002; Gadalla et al., 2016) (see Supplement). Using these samples, we tested the predictions (Figure S1) that gametocyte sex ratios are negatively correlated with gametocyte densities (fertility insurance) and are higher in mixed vs. single genotype infections (local mate competition). Plasmodium falciparum gametocyte sex ratios vary considerably between samples, ranging from all male to all female. Overall, as expected, the majority of samples (94%) have a female-biased gametocyte sex ratio (mean 0.07 ± s.e.m. 0.02). The relationship between gametocyte density and sex ratio varies with MOI (interaction between MOI and gametocyte density: χ12=76.06, p<0.001) (Figure 1A, model A in Table 1). Despite low MOI in this area (mean minimum number of genotypes present 1.55 ± s.e.m. 0.07), we observe a difference in sex ratio between single and mixed genotype infections (Figure 1A). This is however, in the opposite pattern predicted by local mate competition theory: sex ratios are higher in single than in mixed genotype infections. Further, in single infections, sex ratio decreases with increasing gametocyte densities, following the logistic pattern predicted by fertility insurance (Figure S1B) (effect size -2.08 ± s.e.m. 0.13; z=-15.60, p<0.001), but not in mixed genotype infections (effect size 0.33 ± s.e.m. 0.39; z=0.84, p=0.40) (Figure 1A). Our data originate from an area with distinct seasonal transmission and parasites are suggested to adopt different transmission strategies in different seasons (Cornet et al., 2014; Gadalla et al., 2016; Pigeault et al., 2018). Accounting for the impact of season on fertility insurance (interaction between season and gametocyte density, χ12=76.06, p=0.008) and local mate competition (interaction between season and MOI, χ12=18.43, p<0.001) significantly improves our model explanatory power (ΔAIC=62.97; model B in Table 1). When season is taken into account, fertility insurance is detectable in both single and mixed genotype infections (Figures 1B and 1C). Specifically, fertility insurance is more apparent in single compared to mixed genotype infections (interaction between season and MOI, effect size -0.93 ± s.e.m. 0.38; z=-2.46, p=0.014), and in the dry season compared to the wet transmission season (interaction between season and gametocyte density, effect size -0.81 ± s.e.m. 0.31; z=-2.59, p=0.010). It is possible that the observed impact of MOI may vary with season, but we cannot formally test for this because the model including the interaction between season, gametocyte density and MOI does not converge. Adjusting the ratio of male vs. female gametocytes should enable malaria parasites to optimise gametocyte sex ratios in response to changing in-host conditions in the manners that maximise fitness. Our data suggest that the complex patterns of gametocyte sex ratios for P. falciparum parasites in natural infections are consistent with predictions of fertility insurance theory, concerning male limitation. In addition, our data suggest that fertility insurance is more apparent in single compared to mixed genotype infections, and in the dry compared to the wet transmission season. We find no evidence for local mate competition, concerning MOI. Our marker to assess the extent of multiplicity (Pfg377) has at least five alleles in this parasite population (Abdel-Wahab et al., 2002) and it is possible that multiple genotypes within an infection share the same allele. If so, then our apporach is conservative: the difference between single and mixed infections is likely greater. However, whilst we find a difference in sex ratios between single and mixed genotype infections, the pattern is opposite of that predicted: sex ratios are more female biased in mixed genotype infections. There are several potential explanations for this result. In theory, male gametocytes may be detected less efficiently in mixed genotype infections. However, sex-specific gametocyte assays (Schneider et al., 2015; Santolamazza et al., 2017; Stone et al., 2017) target conserved areas of the genome and there is no evidence that these assays function differently between single and mixed genotype infections. Our study reports a mean sex ratio lower than most field studies (Tadesse et al., 2019). This could be associated with males being detected less efficiently than females (Schneider et al., 2015; Stone et al., 2017). If samples with low gametocyte densities had particularly high sex ratios but were not detected, and therefore not included in our study, the true population mean may be closer to the theoretical minimum of approximately 0.1 (Tadesse et al, 2019). Moreover, any problems introduced by low gametocyte densities cannot explain why local mate competition is not apparent at higher gametocyte densities. Additionally, whereas mixed genotype infections in this seasonal transmission area are common, the genotype ratios within infections may be heavily skewed. If so, rare genotypes may be hard to detect by PCR. However, by definition rare genotypes will not influence the genetic diversity of gametocyte mating groups much and so, sex ratios best in single infections remain optimal (or very close to) in this case. Further research is required to investigate how proportional representation of genotypes within mixed genotype infections impacts upon sex ratios. Finally, we measured sex ratios of mature gametocytes and thus some, potentially sex-specific, gametocyte death may have occurred. Male gametocytes are more vulnerable to drug treatment and to immune attack (e.g. (Ramiro et al., 2011; Delves et al., 2013; Dicko et al., 2018)). If such immune responses are stronger in mixed genotype infections, male mortality rates may be elevated substantially beyond that which fertility insurance can compensate for. High male mortality could simultaneously dampen fertility insurance (especially in mixed genotype infections) and local mate competition. Elevated male mortality may also explain observations that: (i) P. chabaudi matches sex ratio to the predictions of local mate competition and fertility insurance before infections peak and immune responses develop (Reece et al., 2008); (ii) why parasites in natural, chronic P. mexicanum infections do not match local mate competition - compared to experimental infections of naïve lizards (Neal and Schall, 2014); and (iii) because antimalarial immune responses peak during the wet season, compared to dry season in Eastern Sudan (e.g. Cavanagh 1998), parasites match fertility insurance better in the dry season. However, explaining why parasites cannot sufficiently increase male production in mixed genotype infections to compensate for high levels of male-specific mortality is a challenge. Non-mutually exclusive hypotheses include that parasites may saturate in their ability to detect male killing (or its proxies); mechanistic constraints on development may limit the production of males; or increasing male production even more may be counterproductive if subsequent killing of these males disproportionately stimulates anti-male immune responses. Despite the unexpected results for sex ratios in mixed genotype infections, in a seasonal malaria setting in Eastern Sudan, we do observe fertility insurance in natural human malaria infections. Facultative adjustment of gametocyte sex ratios in natural infections has implications for disease control. For example, parasites that are exposed to interventions with sex-specific effects (e.g. drug treatment disproportionally affecting male gametocytes or transmission-reducing vaccines targeting females only) can adjust both their conversion rates and gametocyte sex ratio to partially compensate, thus undermining the efficacy of such interventions (Reece et al., 2008; Wu et al., 2008; Sowunmi et al., 2009b; Ramiro et al., 2011; Delves et al., 2013; Dicko et al., 2018). Furthermore, if the cues that parasites use to determine the appropriate level of fertility insurance can be identified, parasites could be tricked into producing suboptimal gametocyte sex ratios, thus reducing transmission (Carter et al., 2014). Finally, we recommend future studies extend our approach to different epidemiological settings, categorise data by season, and gather data on immunity to explore the extent that male-biased mortality affects the ability of P. falciparum to follow the rules of sex allocation theory. Supplementary Material Supplementary Material Acknowledgements We thank Aidan O’Donnell for assistance during the experiments. This work was supported by the National Environment Research Council [NE/I015329/1; NE/K006029/1]; the Wellcome Trust [WT082234MA; 202769/Z/16/Z]; and the Royal Society [UF110155]. The funders played no role in the study design; the collection, analysis and interpretations of the data; in the writing of the report and in the decision to submit the article for publication. Figure 1 Fertility insurance in natural P. falciparum infections. Fertility insurance is predicted in single, but – at first sight – not mixed genotype infections (A). When season is controlled for, fertility insurance is detected in all infections (B: dry season in red; C: wet transmission season in blue). Data shown are mean sex ratios ± s.e.m for single (grey circles) and mixed (black triangles) genotype infections. Predicted sex ratios are shown for single (coloured circles, solid line) and mixed genotype infections (coloured triangles, dashed line). Predictions from mixed effect logistic regression models (A: model A in Table 1; B and C: model B in Table 1) are at population level, i.e. after controlling inter- individual patient variation to reveal overall patterns. Shaded areas represent 95% prediction intervals, which were estimated by data simulation (n=1000; arm package, R v3.2.4, see Supplementary Methods). Predictions from mixed effect logistic regression models (model B in Table 1) before inverse logit back transformation are shown for the two season groups in Figure S2A-B. Table 1 Results from mixed effect logistic regression modelsa to analyse gametocyte sex ratiosb. Model A: SR˜GCT*MOI Chi square p-value AIC r2 marg/cond GCT*MOI(mix) 76.06 <0.001 479.24 0.52/0.75 Model B: SR˜GCT*MOI+SEAS*GCT+SEAS*MOI MOI*SEAS 18.43 <0.001 GCT:SEAS 7.06 0.008 416.27 0.54/0.77 GCT: 10log gametocyte density/μL blood; MOI: multiplicity of infection (single or mixed); SEAS: season (wet transmission season 1 August – 25 December or dry season 26 December – 31 July); r2 marg: marginal, r2 cond: conditional (i.e. taking into account inter-individual differences between patients). a Generalised linear mixed effect models (lme4, R v3.2.4; The R-Foundation, Vienna, Austria, see Supplementary Methods) with patient number included as a random effect to account for repeated sampling, and a binomial error structure. b Gametocyte sex ratios were analysed as a two-vector response variable (males, females) to enable weighted regression that takes into account sample sizes from which gametocyte sex ratios were estimated. Abdel-Wahab A Abdel-Muhsin AA Ali E Suleiman S Ahmed S Walliker D Babiker HA Dynamics of gametocytes among Plasmodium falciparum clones in natural infections in an area of highly seasonal transmission J Infect Dis 2002 185 1838 1842 12085337 Bradley J Stone W Da DF Morlais I Dicko A Cohuet A Guelbeogo WM Mahamar A Nsango S Soumare HM Diawara H Predicting the likelihood and intensity of mosquito infection from sex specific Plasmodium falciparum gametocyte density Elife 2018 7 e34463 29848446 Carter LM Schneider P Reece SE Information use and plasticity in the reproductive decisions of malaria parasites Malar J 2014 13 e115 Cornet S Nicot A Rivero A Gandon S Evolution of plastic transmission strategies in avian malaria PLoS Pathog 2014 10 e1004308 25210974 Delves MJ Ruecker A Straschil U Leliévre J Marques S López-Barragán MJ Herreros E Sinden RE Male and female P. falciparum mature gametocytes show different responses to antimalarial drugs Antimicrob Agents Chemother 2013 57 3268 3274 23629698 Dicko A Roh ME Diawara H Mahamar A Soumare HM Lanke K Bradley J Sanogo K Kone DT Diarra K Keita S Efficacy and safety of primaquine and methylene blue for prevention of Plasmodium falciparum transmission in Mali: a phase 2, single-blind, randomised controlled trial Lancet Infect Dis 2018 18 627 639 29422384 Gadalla AAH Schneider P Churcher TS Nassir E Abdel-Muhsin AA Ranford-Cartwright LC Reece SE Babiker HA Associations between season and gametocyte dynamics in chronic Plasmodium falciparum infections PLoS ONE 2016 11 e0166699 27870874 Gardner A Reece SE West SA Even more extreme fertility insurance and the sex ratios of protozoan blood parasites J Theor Biol 2003 223 515 521 12875828 Mitri C Thiery I Bourgouin C Paul RE Density-dependent impact of the human malaria parasite Plasmodium falciparum gametocyte sex ratio on mosquito infection rates Proc R Soc Lond Series B 2009 276 3721 3726 Neal AT Schall JJ Testing sex ratio theory with the malaria parasites Plasmodium mexicanum in natural and experimental infections Evolution 2014 68 1071 1081 24350982 Nee S West SA Read AF Inbreeding and parasite sex ratios Proc R Soc Lond Series B 2002 269 755 760 Paul REL Breya PT Robert V Plasmodium sex determination and transmission to mosquitoes Trends Parasitol 2002 18 32 38 11850012 Paul REL Coulson TN Raibaud A Brey PT Sex determination in malaria parasites Science 2000 287 128 131 10615046 Pigeault R Caudron Q Nicot A Rivero A Gandon S Timing malaria transmission with mosquito fluctuations Evol Lett 2018 2 378 389 30283689 Ramiro RS Alpedrinha J Carter L Gardner A Reece SE Sex and death: The effects of innate immune factors on the sexual reproduction of malaria parasites PLoS Pathog 2011 7 e1001309 21408620 Read AF Narara A Nee S Keymer AE Day KP Gametocyte sex ratios as indirect measures of outcrossing rates in malaria Parasitology 1992 104 387 395 1641238 Reece SE Drew DR Gardner A Sex ratio adjustment and kin discrimination in malaria parasites Nature 2008 453 609 614 18509435 Reece SE Duncan AB West SA Read AF Host cell preference and variable transmission strategies in malaria parasites Proc R Soc Lond Series B 2005 272 511 517 Robert V Sokhna CS Rogier C Ariey F Trape JF Sex ratio of Plasmodium falciparum gametocytes in inhabitants of Dielmo, Senegal Parasitology 2003 127 1 8 12885183 Santolamazza F Avellino P Siciliano G Yao FA Lombardo F Ouédraogo JB Modiano D Alano P Mangano VD Detection of Plasmodium falciparum male and female gametocytes and determination of parasite sex ratio in human endemic populations by novel, cheap and robust RTqPCR assays Malar J 2017 16 468 29149898 Schneider P Reece SE Schaijk BCLv Bousema T Lanke KHJ Meaden CSJ Gadalla A Ranford-Cartwright LC Babiker HA Quantification of female and male Plasmodium falciparum gametocytes by reverse transcriptase quantitative PCR Mol Biochem Parasitol 2015 199 29 33 25827756 Sowunmi A Balogun ST Gbotosho GO Happi CT Plasmodium falciparum gametocyte sex ratios in symptomatic children treated with antimalarial drugs Acta Trop 2009a 109 108 117 19027703 Sowunmi A Gbotosho GO Happi CT Folarin OA Balogun ST Population structure of Plasmodium falciparum gametocyte sex ratios in malarious children in an endemic area Parasitol Intl 2009b 58 438 443 Stone W Sawa P Lanke K Rijpma S Oriango R Nyaurah M Osodo P Osoti V Mahamar A Diawara H Woestenenk R A molecular assay to quantify male and female Plasmodium falciparum gametocytes: Results From 2 randomized controlled trials using primaquine for gametocyte clearance J Infect Dis 2017 216 457 467 28931236 Tadesse FG Meerstein-Kessel L Gonçalves BP Drakeley C Ranford-Cartwright L Bousema T Gametocyte sex ratio: the key to understanding Plasmodium falciparum transmission? Trends Parasitol 2019 35 226 238 30594415 West SA Reece SE Read AF Evolution of gametocyte sex ratios in malaria and related apicomplexan (protozoan) parasites Trends Parasitol 2001 17 525 531 11872397 Wu Y Ellis RD Shaffer D Fontes E Malkin EM Mahanty S Fay MP Narum D Rausch K Miles AP Aebig J Phase 1 Trial of Malaria Transmission Blocking Vaccine Candidates Pfs25 and Pvs25 Formulated with Montanide ISA 51 PLoS ONE 2008 3 e2636 18612426

PMC007xxxxxx/PMC7614806.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101698577 Nat Ecol Evol Nat Ecol Evol Nature ecology & evolution 2397-334X 30886375 7614806 10.1038/s41559-019-0831-4 EMS181172 Article The evolutionary ecology of circadian rhythms in infection Westwood Mary L 1* O’Donnell Aidan J 1 de Bekker Charissa 2 Lively Curtis M 3 Zuk Marlene 4 Reece Sarah E 1 1 Institute of Evolutionary Biology and Institute of Immunology and Infection Research, School of Biological Sciences, University of Edinburgh, Charlotte Auerbach Road, Edinburgh EH9 3FL, United Kingdom 2 Department of Biology, University of Central Florida, 4111 Libra Drive, Orlando, Florida 32816, United States of America 3 Department of Biology, Indiana University, Bloomington, 1001 East Third Street, Indiana 47405, United States of America 4 Department of Ecology, Evolution and Behavior, University of Minnesota, 1479 Gortner Ave, St. Paul, Minnesota 55108, United States of America * Corresponding author: Mary Westwood (mary.westwood@ed.ac.uk) 01 4 2019 18 3 2019 18 7 2023 25 7 2023 3 4 552560 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. pmcBiological rhythms coordinate organisms’ activities with daily rhythms in the environment. For parasites, this includes rhythms in both the external abiotic environment and the within-host biotic environment. Hosts exhibit rhythms in behaviours and physiologies, including immune responses, and parasites exhibit rhythms in traits underpinning virulence and transmission. Yet, the evolutionary and ecological drivers of rhythms in traits underpinning host defence and parasite offence are largely unknown. Here, we explore how hosts use rhythms to defend against infection, why parasites have rhythms, and whether parasites can manipulate host clocks to their own ends. Harnessing host rhythms or disrupting parasite rhythms could be exploited for clinical benefit; we propose an interdisciplinary effort to drive this emerging field forward. Circadian rhythms have long been taken for granted by science. Indeed, the first observation of a clock-controlled behaviour (leaf opening and closing in Mimosa pudica) was not recorded until the 18th century1. Following the fundamental observation that organisms can adaptively anticipate daily rhythms in their environment, the field of “chronobiology” took off in the mid-20th century with a focus on evolutionary and ecological questions. However, the advent of genetic tools a few decades later shifted the remit to determining the molecular and genetic workings of circadian clocks. Yet, despite their assumed major impact on fitness, circadian rhythms remain overlooked in evolutionary ecology2–4. Here, we propose that the integration of chronobiology and evolutionary ecology return to its roots to tackle a topic of growing and applied interest; the role of rhythms in host-parasite interactions. Note that we use the term “parasite” to collectively refer to all agents of infection (e.g. single-celled and multicellular eukaryotes, bacteria, viruses). One of the most fundamental ecological interactions is that between hosts and parasites. Research from diverse taxa (plants, mammals, and insects) reveals that host clocks drive daily rhythms in immune defences, disease severity and spread5,6. Parasites display daily rhythms in traits underpinning within-host survival and between-host transmission7,8. Rhythms in parasite activities and in host responses to infection could provide an advantage to parasites, hosts, both, or neither. To what extent parasites and hosts are in control of their own and/or each other’s rhythms is also poorly understood. Understanding the evolution (and possibly, coevolution) of rhythms may enable vaccines and drugs to take advantage of rhythmic vulnerabilities in parasites or harness host rhythms to improve efficacy and reduce drug toxicity. For such interventions to be robust to parasite evolution, understanding how host-parasite interactions shape rhythms in hosts and parasites is necessary7. Key questions include how rhythms in diverse host traits contribute to defence, how parasites cope with exposure to their host’s rhythms, and whether hosts and parasites can manipulate each other’s rhythms for their own benefit. We discuss these three scenarios, identify systems to explore them, and offer ways in which this knowledge can be exploited to improve health. An evolutionary ecologist’s introduction to chronobiology is provided in Boxes 1 and 2. Rhythms in host defence The most patent defence against infection is the immune response, and a wealth of evidence reveals that circadian clocks play a role in orchestrating immune defences5. Circadian clock genes are expressed in many types of immune cell, and the immune and circadian systems are connected in multiple ways9,10. For instance, the clock gene Bmal1 mediates the balance between pro- and anti-inflammatory responses11. Rhythmic production of the pro-inflammatory cytokines TNF-α and IL-6 by macrophages is clock controlled12, and mobilization of inflammatory monocytes is also regulated by the clock10. This phenomenon, termed “anticipatory inflammation”, appears uncoupled to metabolic rhythms and may defend against incoming parasites13. Similarly, in humans, proinflammatory cytokines peak in circulation during the day (active phase)14, whereas hematopoietic stem and progenitor cells, and most mature leukocytes, peak at night14,15. In nocturnal mammals, an inverse rhythm is often observed, with innate defences peaking at night (active phase) and repair mechanisms peaking during the day (resting phase)9. Observations of immune rhythms have given rise to the notion that organisms invest in defence during the active phase when parasite encounter is assumed most likely, and repair during the resting phase16. Temporal segregation of immune responses may thus solve problems caused by having immune defences continually tuned to maximal (e.g. collateral damage via immunopathology17). Also, energetic demands imposed by activity and metabolism may trade-off against immune defence18. Intuitively, “defence only during the active phase” suggests the host is achieving the most “bang for the buck” by ensuring activities that are energetically costly, or likely to cause collateral damage, are only performed when most useful. However, this intuition may be naïve. First, it ignores the potential for constraints imposed by the need to temporally couple (or de-couple) certain immune rhythms with other internal rhythms7. This includes separating the timing of metabolism from defensive actions within immune cells themselves5,16. Second, it assumes that a parasite encounter is rhythmic and predictably occurs in the active phase. This is clearly the case for food-borne parasites, but ingestion is not the only route into a host. Rather, the immune system functions within a broad set of energetic demands in which parasite defence is just one of many requirements. For example, rhythmic stomatal opening for gas exchange during the day is a well-used route into plants by bacterial pathogens19. Consequently, Arabidopsis is better able to detect and defend against parasites in the morning than evening20,21. Given the wealth and diversity of data (illustrated in Table 1), meta-analyses are needed to test whether the timing (phase) of rhythms in immune effectors relates to nocturnal vs diurnal lifestyles and whether they function in front-line or secondary defences, or healing. Infection in the active vs resting phase for diverse hosts (flies, plants, mammals) dramatically affects disease severity and mortality rates (Table 1), suggesting that the phase of immune rhythms upon infection matters. Most studies performed in plants (Table 1) point towards infection during the active phase resulting in greater resistance to infection and less damage to the plant. But the degree to which immune rhythms result in time-of-day differences in parasite control can be counter-intuitive. For example, mice mount higher clock-controlled proinflammatory responses against Salmonella enterica Typhimurium when challenged in their rest phase, but bacterial load is also higher and hosts have worse symptoms22. Furthermore, Leishmania parasites infect host neutrophils and macrophages, and the clock-controlled secretion of chemoattractants by these immune cells facilitates their infection, making parasite invasion more successful at night when immune activity is highest23. Thus, whether immune rhythms are sufficient to entirely explain divergent outcomes of time-of-day of infection is unclear (Table 1). Studies that separate the effects of immune rhythms on preventing infection from their role in dealing with ongoing infection will reveal the extent to which immune rhythms are beneficial and when they should be overruled to deal with a major threat. Additionally, most time-of-day immune challenges have used either bacteria or chemicals, raising the question of whether a more diverse array of challenges are needed to establish general patterns. That host circadian clocks impact on infection via traits other than immune responses has been largely overlooked. Rhythmicity in host activity may determine when hosts provide the best resources to their parasites and offer the most opportunities for onwards transmission24–26. For example, a recent study of the intestinal helminth Trichuris muris demonstrates the role of host rhythms in foraging. Mice infected in the morning (resting phase) expel worms sooner and have a stronger T-helper 2 response than dusk-infected (active phase) mice, and this effect is reversed when mice are fed only in the day, in an immune-independent manner27. Host feeding rhythms are relevant to gut microbiota, and a two-way feedback between host and microbe rhythms has been proposed28. Daily rhythms in host reproductive behaviours may make hosts vulnerable to infection. For example, the crepuscular and nocturnal singing activity of the cricket Teleogryllus oceanicus allows the acoustically-orienting parasitoid fly Ormia ochracea to locate hosts, but the flies are best able to hunt when darkness is incomplete29. A rhythmically expressed reproductive behaviour (singing) got the host into this mess, and it appears that natural selection has found two solutions (see Box 3). In addition to immune responses, infected hosts often exhibit adaptive sickness behaviours consisting of endocrine, autonomic, and behavioural changes that perturb circadian rhythms30,31. For example, wild red colobus monkeys (Procolobus rufomitratus tephrosceles) decrease energetically costly activities, and rest frequently, while shedding whipworm eggs32. Fever, another common sickness behaviour, is sufficiently advantageous to offset the 10-12.5% increase in metabolic rate required for each 1°C increase in temperature33 and has been conserved throughout more than 600 million years of vertebrate evolution34. Fever enhances an organisms chance of survival by creating a hostile environment for parasites and a more active immune response34–37. Under normal circumstances, the so-called central (SCN) clock controls body temperature rhythms, but how the SCN and inflammation interact to control temperature is unknown. Though many behaviours altered during infection are clock-controlled during health, the extent to which organisms become too sick to maintain normal behaviour or adaptively disrupt their rhythms is unclear. Additionally, clock-control could facilitate recovery of rhythms during the return to health. Viewing the host as a collection of traits connected by the circadian system has the potential to uncover novel strategies to resist infection and reveal the circumstance in which immune rhythms reflect constraints or adaptations. Indeed, rhythmic metabolism of xenobiotic substances (e.g. drugs and vaccines) influences efficacy and toxicity in a time-of-day dependent manner38. For example, halothane (a commonly used anaesthetic) administered to mice in the daytime results in low mortality (5%), but mortality increases (76%) if administered at night39 and half of the best-selling drugs in the USA for humans target the products of genes that are rhythmically expressed (in mice)40. A better understanding of host rhythms could be harnessed to make drugs and vaccines more effective, as well as mitigating the negative effects of modern lifestyles that involve shift work and jet lag. However, for such interventions to be sustainable in the face of parasite evolution, understanding the ecology of rhythms from the perspective of parasites is also required. Rhythms in parasite offence Scheduling activities to take advantage of daily rhythms in transmission opportunities could be a general explanation for rhythms in parasites. The most well-known example concerns the transmission forms (microfilariae) of different species of filarial worms. They move from the host’s organs to the capillaries during the day or night, depending on whether they are transmitted by day- or night-biting insect vectors41. In addition to the activity patterns of vectors, rhythmic interactions with hosts also matter. For example, the larval stage of the blood fluke Schistosoma japonicum emerge from their invertebrate host to seek a mammalian host at different times of day. Flukes emerge in the afternoon when the preferred host is nocturnal or in the morning if seeking a diurnal host42. Parasites that have free-living stages are also subject to rhythms in the abiotic environments. The coccidian parasite Isospora sheds from its host in the late afternoon to minimise UV exposure and desiccation risk whilst undergoing a developmental transition necessary to infect new hosts43. However, key questions remain about the adaptive nature of these rhythms. For example, why aren’t microfilariae located in the peripheral capillaries all day long? Is a cost associated with this location, which is only worth paying at times of day when vectors are active? In contrast to the role of parasite rhythms in transmission, their role in within-host survival has received less attention. Many host rhythms (in addition to immune rhythms) present opportunities and constraints for parasites. Trypanosoma brucei (which cause sleeping sickness) display circadian clock-driven rhythms in the expression of metabolic genes8. These rhythms correlate with time-of-day sensitivity to oxidative damage, thereby suggesting the need to cope with redox challenges caused by rhythmic digestion of food by hosts. In contrast, rhythms in the development of asexually replicating malaria parasites capitalise on daily variation in the nutritional content of blood caused by host immune responses and feeding patterns44,45. Whether malaria parasites cannot complete their developmental cycle until the host makes nutrients available, and/or use nutrients rhythms as a time-of-day cue to set the pace of their development, is unknown46 (see Box 3). Clocks in parasites or hosts could have fitness consequences for one or both parties, or neither. Fitness consequences for both hosts and parasites suggests that clocks could coevolve. Clock coevolution is suspected for the plant-pollinator system Petunia axillaris and Manduca sexta47, in which nocturnal scent emission by P. axillaris coincides with foraging activity in the hawkmoth M. sexta. Both traits are clock-controlled, and appear so well synchronized that, even in the absence of floral scent emission, M. sexta exhibits a burst in foraging activity at the same time that floral scent emission is expected to be greatest. However, foraging behaviour also remains sensitive to the environment, as evidenced by absence of activity when the moth is subjected to light at night. If rhythms in different organisms do coevolve, then they should use the same Zeitgeber, but how robust should their timing systems be to fluctuations in the environment? If the rhythm of one party is more readily disrupted (masked) by environmental change, or faster at tracking seasonal changes in photoperiod, then the relationship may be disrupted to the gain of hosts or parasites. Exploring the degree and consequences of plasticity in rhythms is pertinent because climate change is interfering with the ability of interacting species to synchronise48. The situation is further complicated when interactions between both host and parasite clocks shape disease trajectories. For example, in a plant-fungus system (Arabidopsis thaliana and Botrytis cinerea, respectively), when both parties are in the same photoperiod schedule, primary plant defences peak in the morning, and the fungus produces the biggest lesions when inoculated at dusk49. The authors were able to separate the contributions to pathogenicity by host and parasite clocks using reverse lighting schedules for fungus and plants: fungus at dusk produced more severe infections than fungus at dawn, regardless of time-of-day for recipient plants49. Furthermore, this suggests B. cinerea anticipates and exploits weaknesses in plant defence at dusk rather than attempting to overwhelm dawn defences (see section “Rhythms in host defence”). Separately assigning the contributions of rhythms in hosts/vectors and parasites to virulence and transmission is necessary to understand whose genes control which rhythms, and hence how they can be shaped by selection. If parasite rhythms are adaptive, then disrupting them could reduce disease severity as well as transmission. However, understanding the timing mechanisms of parasite rhythms is necessary to disrupt them7. Unravelling how parasite rhythms are controlled is a considerable challenge. Parasites might allow the host to inadvertently schedule their activities for them, in which case the genes encoding parasite timing mechanisms belong to hosts. Alternatively, parasites might keep time using a circadian clock (with the properties described in Box 1), as demonstrated for T. brucei and B. cinerea. Given the diversity in clock genes across taxa, searching genomes for known clock genes often yields “absence of evidence” not “evidence of absence.” Instead, round-the-clock transcriptomics or proteomics, paired with bioinformatics approaches to mine for known core clock-related functional domains and sequence patterns may find candidates. However, simpler time-keeping strategies exist, though they do not necessarily have the advantages of temperature compensation or anticipation. For example, cell division cycles are often controlled by hourglass mechanisms that rely upon threshold concentrations of substances, independently of periodic phenomena50. Alternatively, organisms can react directly (via “tracking”) to temporal changes in the environment. Note, this differs from masking, a chronobiological phenomenon in which the expression of a clock-controlled rhythm is suppressed by a change in the environment without having a direct effect on the period or phase of the underlying rhythm51. A response that directly tracks time-of-day cues may suit parasites with multi-host lifecycles if each host type provides a different time-cue. Given that rhythms in T. brucei metabolism and plasticity in development during the asexual cycle of Plasmodium spp. enables these parasites to tolerate drugs, there is an urgent need for proximate and ultimate explanations of their rhythms. The T. brucei clock is entrained by temperature cycles, but if other parasites use Zeitgebers to set their clocks, or respond directly to time-of-day cues, that are readily perturbed, it should be possible to reduce parasite fitness by interfering with their rhythms. Further, reports of changes to the biting time of mosquito populations that transmit malaria suggests that insecticide-treated bed nets are imposing selection on vector rhythms8,52,53. Given that rhythms of parasites and mosquitoes each affect malaria transmission in lab experiments54,55, what are the likely epidemiological consequences? Recent work suggests that mosquitoes are more susceptible to infection when they feed in the daytime and parasites are more infectious at night54. Thus, day-biting could increase the prevalence, but not burden, of malaria in mosquitoes. However, in the longer term, if parasites evolve to invert their rhythm but mosquitoes do not, both prevalence and burden may increase. Parasite manipulation of host rhythms Rhythms in host processes offer opportunities that parasites could exploit. Could parasite fitness be increased by coercing hosts into altering their rhythms? Although many striking examples of parasite manipulation of host phenotypes (i.e. changes to host traits that benefit parasites) are known56, the notion of “parasite manipulation of host clocks” is largely unexplored57. A pre-requisite for parasite manipulation is that a phenotypically plastic host trait is targeted; and circadian clocks are flexible. Because clocks control much of the host’s behaviour and physiology58 and clocks throughout a given host involve the same players in the canonical clock (the TTFL), manipulation of the host’s time-keeping may be an efficient way to simultaneously alter many aspects of the within-host environment. Alternatively, parasites interests may be served by bolstering circadian rhythms of their hosts during sickness to ensure they forage and interact with conspecifics, as usual. As outlined in the section “Rhythms in host defence,” separating the effects of being sick per se from host defence and parasite manipulation is challenging. Recently, a combination of culture and comparison of infection models has revealed that T. brucei alters expression rhythms of clock genes in host mice59. Specifically, infected hosts are more active in the resting phase (phase-advanced) because the clock runs faster (shorter period). Effects at organismal, cellular, and molecular levels suggests the behaviour is not just a result of sickness59. However, it is not clear how T. brucei achieves this, and whether the parasite benefits from altering host rhythms. One target of circadian disruption by viral parasites is the gene Bmal1, a core clock gene. Herpes and influenza A virus replication and dissemination within the host is enhanced in infections where Bmal1 is knocked out60. However, it remains unclear if virus replication is maximised by simply disturbing rhythmicity in host cell cycles or if this is a case of immune manipulation since Bmal1 appears involved in innate host defence60. Having observed changes to host clocks, the proceeding step is to decipher the ecological context behind these effects. The above examples lend proof-of-principle to the idea that parasites can manipulate host clocks and could be a general explanation for examples of host manipulation. Hairworms (Nematomorpha) are a well-known case of temporally linked behavioural manipulation. They infect various arthropods, notably crickets, and cause the host to wander in an erratic manner until a body of water is encountered. The host commits suicide by jumping in water, and the adult hairworm emerges. Infected hosts are found wandering only in the early part of the night61, and uninfected hosts are rarely motivated to jump into water. Infected crickets differentially express an array of proteins, some of which are linked to visual processes and circadian clocks62. Culturing isolated host cells with parasite products and quantifying the expression of clock genes (following Rijo-Ferreira 2018) could illuminate this case of parasite manipulation. For systems without relevant insect cells lines, or cases where manipulation is likely to be tissue/cell type specific, a transcriptomics approach may be useful63. Round the clock expression data can be mined for putative core clock genes and their phase, amplitude and period assessed in control and manipulated hosts. This however, is likely to be extremely challenging for host species whose timekeeping does not rely on a canonical circadian clock. Another putative case for clock manipulation concerns the New Zealand freshwater snail (Potamopyrgus antipodarum) infected with Microphallus trematodes64 (Trematoda: Microphallidae). Uninfected adult snails forage primarily at night on the upper surfaces of rocks in the shallow-water margins of lakes. These snails retreat to under rocks at sunrise, which likely reduces their risk of predation by waterfowl, which are the definitive host for Microphallus. Infected snails, however, show delayed retreating, potentially making them more likely to be consumed25. Crucially, the apparent manipulation only occurs when the parasite is mature. Snails infected with immature (non-transmissible) stages exhibit the same risk-averse retreating behaviour as uninfected snails25. In addition, snails infected with other species of sterilizing trematodes, which are not trophically transmitted, do not exhibit the same risky behaviour as those infected with Microphallus65, thereby eliminating the possibility that the Microphallus-induced behavioural change is a simple artefact of parasitic castration. Finally, Microphallus-infected snails spend more time foraging on the top of rocks, even when food was removed whereas uninfected snails retreated to shelter65. Taken together, the data suggest that Microphallus induce a change in snail behaviour that increases trophic transmission, potentially via manipulation of clock-controlled activity rhythms. There are many ways that parasites could interfere with clock-controlled host behaviours. A blunt instrument would be to alter perception/detection of the Zeitgeber that sets the time of the host’s clock, which is usually light. For example, Microphallus could interfere with photoreception to reduce the sensitivity of snails to dawn, causing their clocks to phase delay and forage at higher light intensities than un-manipulated snails. Alternatively, parasites could induce the host to ignore its clock (mask) or alter clock regulation of hormones that relay time-of-day information around the host. For example, baculoviruses appear to perturb the circadian rhythms of their caterpillar hosts by disrupting hormones that control climbing behaviour. In the baculovirus (Lymantria dispar nucleopolyhedrovirus), a single gene inactivates 20-hydroxyecdysone66 (a host hormone regulated by a circadian oscillator), motivating the caterpillar to climb high atop their host plants. Here, they liquefy and disseminate the virus to caterpillars below, as well as infecting birds who consume the corpses67. Similar to the manipulation of caterpillar hosts, many species of parasitic fungi (Ophiocordyceps spp. and Pandora spp.) alter the daily behavioural rhythm of a variety of ant species68,69 (See Box 3). Parsing out whether temporal disruption is a host response or clock manipulation is nearly, if not entirely, impossible without uncovering the mechanism of manipulation. The lack of insight into the mechanisms parasites use to interfere with their hosts has stalled progress in the field of “host manipulation by parasites”70. This gap could be filled by harnessing the tools and conceptual framework developed in chronobiology. Many of the examples above have employed an ecological approach, yet a chronobiological approach can help elucidate both proximate and ultimate explanations. Conclusion Over the past few decades, the focus of chronobiology has been to elucidate the mechanistic underpinnings of biological rhythms. We propose that now is the time to integrate this knowledge into parasitology, evolutionary ecology, and immunology (see Box 2). Indeed, the role of biological rhythms in infectious disease is a growing topic that holds promise for improving human and animal health. History clearly illustrates that attempts to control parasites are usually met with counter-evolution (in the form of drug resistance, vaccine escape, and host shifts). A comprehensive understanding of how rhythms affect parasite invasion and exploitation of a host (or vector) offers novel ways to disrupt the chain of transmission and treat disease. Further, clock coevolution may occur in host-parasite-vector interactions, resulting in complex arms races best understood through the lens of chronobiology coupled with evolutionary ecology. Chronobiology supplies a myriad of tools to help elucidate rhythmic phenotypes and reveal to what extent host and parasite genes are responsible for rhythms in disease phenotypes. Adding an evolutionary ecology framework will ensure this information is generalisable and used to make interventions as evolution-proof as possible. Acknowledgements We thank the Darwin Trust of Edinburgh (MLW), the National Science Foundation (MZ), NERC and BBSRC (NE/K006029/1; SER), the Royal Society (UF110155; SER), and the Wellcome Trust (202769/Z/16/Z; SER) for supporting this work. Box 1 What are circadian rhythms? Biological rhythms are deemed to be controlled by circadian clocks if they meet several criteria71. First, their duration (period) must be approximately 24 hours. Second, they must persist (free-run) in conditions without time-of-day cues, which is usually assessed by observation in constant light or dark. Third, the phase of the oscillator or outputs are set (entrained) by a time-of-day cue (Zeitgeber) which is usually light. Fourth, unlike the rate of many chemical reactions, the speed of a circadian clock varies little over a biologically realistic range of environmental temperatures (temperature compensation). Together, these criteria allow organisms to fulfil a key feature of circadian rhythms: anticipatory, rather than reactionary, behaviour. For instance, plants ready photosynthetic machinery in anticipation of sunlight72,73 and animals exhibit food-anticipatory activity (e.g. increases in core temperature, activity, serum corticosterone, and duodenal disaccharides) prior to foraging74. The workings of circadian clocks are sufficiently flexible to allow organisms to cope with gradual changes in photoperiod across seasons, but not flexible enough to instantly cope with changes in time zones (which is why travellers experience jet lag). The mammalian circadian system is composed of the “central” clock in the brain (suprachiasmatic nucleus; SCN) and “peripheral clocks” in other organs and tissues (A). Clocks in nucleated cells are run by transcription-translation feedback loops (TTFL). For example, in animals the proteins CLOCK and BMAL1 act as activators and members of the PER and CRY families are repressors75 (B). Retinal photoreceptors receive light cues which are carried through the hypothalamic optic tract and transmitted to the SCN, resulting in its synchronization/entrainment (C). Clocks in organs and tissues (peripheral clocks) can be entrained by feeding rhythms, and in taxa other than mammals, exercise, social cues, and abiotic rhythms in temperature and humidity may entrain clocks (D). Rhythms are often characterised by their period, amplitude, and markers for phase (E; grey bars illustrate night time for a rhythmic trait measured over 48 hours). They are described in relation to the time since the Zeitgeber (ZT) occurred (e.g. ZT6 refers to 6 hours after dawn) which usually differs from the actual time-of-day (Circadian Time; CT). Box 2 Why have circadian rhythms evolved? Circadian clocks appear so advantageous that nearly all eukaryotes have a circadian system in most cells76. Circadian clocks may confer two kinds of fitness benefit: coordinating behaviours with rhythms in the external environment (extrinsic adaptive value), and temporally compartmentalising incompatible processes (intrinsic adaptive value)2. For instance, intrinsic benefits are conferred when cell division in yeast is temporally constrained to the reductive phase of metabolism, minimising rates of genetic mutation77. However, most studies of the fitness consequences of circadian rhythms have focussed on the benefits of synchronizing activities with rhythms in the abiotic environment: matching the period of day-night rhythms enables cyanobacteria to outcompete strains whose clocks run faster or slower78 and enhances the survival of Arabidopsis73. Rhythms in the biotic environment2 matter too. For example, the sea urchin Centrostephanus coronatus avoids predatory sheephead wrasse (Pimelometopon pulchrum) by foraging at night and retreating to shelter prior to the onset of wrasse activity79. Despite the diversity of extrinsic rhythms that could select for the scheduling of diverse processes, there are surprisingly few demonstrations that circadian clocks actually affect fitness. For example, fitness is greater in wild-type mice than mutant mice with shortened periods80, flies with clock mutations die more rapidly than wild types after infection with bacteria81,82, and circadian knockout plants flower later and are less viable than wild-type plants3. However, depending on ecological context, rigidly scheduling activities according to day and night is not always the best strategy. For example, nocturnal mice boost energy efficiency by switching to diurnality when challenged with cold and hunger83. Nursing honeybees, that remain in the hive are arrhythmic, because round-the-clock care is necessary for larvae; and, if needed, diurnal foraging bees can revert to arrhythmic nursing behaviour84. Shorebirds also display considerable plasticity in activity rhythms during breeding, likely explained by predator avoidance strategies85. The above examples illustrate the gains to be made from integrating chronobiology with evolutionary ecology in general4. We propose that such an approach offers a novel advance to the study of host-parasite interactions and coevolution. Coupling the well-developed conceptual frameworks for unravelling how circadian oscillators operate, and probing the costs and benefits of phenotypically plastic traits that are relevant to infection, will explain why rhythms in immune defences and parasite traits occur. Box 3 Case studies illustrating the role of circadian rhythms in parasite offence, host defence, and host manipulation Host-parasite system Teleogryllus oceanicus (Pacific field cricket) & Ormia ochracea (parasitoid fly) What we know O. ochracea deposit larvae which burrow into the host and emerge 7-10 days later, resulting in host death. A flatwing morph that is physically incapable of calling has evolved to evade the risk of parasitism by acting as a silent, satellite male24. A more nuanced form of parasite evasion? In addition to the flatwing morph, natural selection may have found another solution. Some males condense singing activity to the darkest part of the night29 which may hamper the fly’s ability to use visual cues to home in on hosts. Parasite evasion (via a flatwing phenotype or phase-shifted calling) trades off against attracting females, potentially constraining selection on these strategies. Moreover, multiple activities need to be coordinated for successful reproduction (e.g. locomotion, foraging, spermatophore production). Given that many of these traits are clock-controlled, could altering the timing outputs of the clock be a streamlined way of phase-shifting all related activities and minimizing the costs of parasite evasion? [associated image = cricket_fly.png] Photo credit: Norman Lee Host-parasite system Carpenter ants & Ophiocordyceps spp. and Pandora spp. (fungi) What we know O. unilateralis s.l. induces workers of its carpenter ant host, ordinarily active during the night-time, to wander out of the ant nest during the day-time. Hosts then summit vegetation and adopt a mandibular death-grip in elevated positions. This manipulated behaviour is highly time-of-day and species-specific and occurs within a 3-hour window at dawn or in the mid-late morning, depending on the species68,86. Clinging to vegetation, the ant dies whilst the fungus completes its life cycle by growing a spore-producing stalk out of the dorsal region of the ant’s thorax86. A case for coevolution and ecosystem specificity? The jigsaw puzzle of how the fungus controls the ant is still being pieced together. Clocks may play a central role because infection alters the expression of host clock homologues period and cycle68. Host manipulation also appears to involve altering host chemosensory abilities, potentially via rhythmic secretion of enterotoxins87, all achieved from the fungus’s primary location in muscle tissues88. [associated image = ant_fungi.png] Photo credit: Miles Zhang Host-parasite system Mammals & Plasmodium spp. (malaria parasites) What we know Malaria parasites synchronously burst from the host’s blood cells every 24, 48, or 72 hours depending on the parasite species89. When out of synch with the host’s circadian rhythms, parasites incur an approximately 50 percent reduction in the densities of both asexual stages (necessary for in-host survival), and sexual stages (responsible for transmission)90 before they become rescheduled to be in synch with host feeding rhythms44,45. Three worlds collide: a complex system of interactions? Why aligning the phase of parasite rhythms with the host’s rhythms is important remains mysterious, but recent work suggests that parasites are also selected to coordinate with the time-of-day their mosquito vectors are active54,55 (see Rund et al. 2011 for information on Anopheles circadian rhythms). If differently phased rhythms for asexual replication are required to provide the best matches to host and vector rhythms, parasites face a trade-off between maximizing in-host survival and between-host transmission. Such a tension could be exploited by novel drug treatments to coerce parasites into a loss of fitness. Further, mosquito nets have induced a shift in Anopheles gambiae biting activity, ultimately resulting in a change in host-parasite timing8,52,53. The epidemiological consequences of this are unknown. [associated image = mosquito_malaria.png] Photo credit: Sinclair Stammers Table 1 Impact of immune challenge during the rest and active phases of hosts. A selection of studies identified as time-of-day immune challenges from PubMed searches for ““time of day” plus “immune and infection” and ““circadian rhythm” plus “immune and infection”. Articles were included if the study involved a time-of-day immune challenge; those without a time-of-day immune challenge were not included in the table. Time-of-day (ToD) is given as hours since lights on (ZT) for organisms in entrainment conditions, and as subjective day/night for those in constant light or dark conditions (i.e. corresponding to the light or dark portion of the cycle before experiencing constant conditions). Unless otherwise stated, entrainment conditions are 12 hour light:dark. Outcomes of challenge in the rest phase (daytime for nocturnal organisms, nighttime for diurnal organisms) are compared to challenge in the active phase in terms of virulence metrics and immune effectors measured. Host spp. Challenge ToD Outcome in rest versus active phase Ref Mus musculus – house mouse (nocturnal) Salmonella typhmurium ZT4/16 Greater inflammation and bacterial load when infected in the rest phase 22 Leishmania major Subjective day/night Lower parasite burden and lower severity when infected in the rest phase 23 Lipopolysaccharide (LPS) endotoxin Subjective day/night Lower concentrations of cytokines when infected in the rest phase 91 ZT11/19 Higher mortality when challenged in the rest phase 92 Subjective day/night Greater inflammatory responses and lower bacterial burden when challenged/infected in the rest phase 93 Streptococcus pneumoniae ZT0/12 Murid Herpesvirus 4 ZT0/10 Greater viral replication when infected in the rest phase 60 Helicobacter pylori ZT1/7/13 Lower lymphocyte numbers when infected in the rest phase 94 Vesicular stomatitis virus ZT0/12 Higher mortality when infected in the rest phase 95 Drosophila melanogaster – fruit fly (diurnal) Pseudomonas aeruginosa ZT1/5/9/13/17/21/1 Lowest mortality when infected in the rest phase (especially ZT21) 82 Subjective day/night Lowest bacterial burden when infected in the rest phase Streptococcus pneumoniae ZT7/19 Slowest rate of mortality when infected in the rest phase 81 Escherichia coli ZT0/6/12/18 Infection at all ZT induces sleep the morning after infection and sleep was more prolonged after infection in the rest phase 96 Anopheles stephensi - Asian malaria mosquito (nocturnal) Escherichia coli Morning/evening Lower bacterial growth and lower mortality when infected in the rest phase 97 Arabidopsis thaliana – thale cress (diurnal) Pseudomonas syringae ZT0/4/10/16 Immune defences are highest when inoculation occurs early in the active phase Note photoperiod is 9 hours light: 15 hours dark 98 Botrytis cinerea Dawn/dusk Larger lesions when inoculated in the rest phase 49 ZT0/3/6/9/12/15/18/21/24 Greater susceptibility when inoculated in the rest phase 21 Pseudomonas syringae Subjective day/night Lower infiltration of bacteria when infected in the rest phase 99 Subjective morning/evening Greater suppression of bacterial growth at the start of the rest phase when spray-inoculated, and greater suppression of bacterial growth at the start of the active phase when syringe-infiltrated 20 Hyaloperonospora arabidopsidis Dawn/dusk Highest percentage of leaves with sporangiophores when infected in the start of the rest phase 100 Danio rerio zebrafish (diurnal) Salmonella typhimurium ZT4/16 Lower survival when infected in the rest phase 101 Oreochromis niloticus – Nile tilapia (mostly diurnal) LPS ZT3/15 Greater humoral immune response when infected in the rest phase 102 Phodopus sungorus – Siberian hamster (nocturnal) LPS ZT1/16 Shorter febrile response and more persistent locomotor activity when infected in the rest phase. Note, photoperiod is 16 hours light: 8 hours dark 103 Contributions SER conceived the study, MLW and SER drafted the manuscript, and all authors provided substantial input into ideas and the writing of subsequent drafts. Competing interests The authors declare no competing interests. 1 de Mairan J Observation botanique Hist l’Academie R des Sci Paris 1729 2 Sharma VK Adaptive Significance of Circadian Clocks Chronobiol Int 2003 20 901 919 14680135 3 Green RM Tingay S Wang Z-Y Tobin EM Circadian Rhythms Confer a Higher Level of Fitness to Arabidopsis Plants Plant Physiol 2002 129 576 LP 584 12068102 4 Helm B Two sides of a coin: ecological and chronobiological perspectives of timing in the wild Philos Trans R Soc B Biol Sci 2017 372 5 Scheiermann C Gibbs J Ince L Loudon A Clocking in to immunity Nat Rev Immunol 2018 18 423 437 29662121 6 Martinez-Bakker M Helm B The influence of biological rhythms on host–parasite interactions Trends Ecol Evol 2015 30 314 326 25907430 7 Reece SE Prior KF Mideo N The Life and Times of Parasites: Rhythms in Strategies for Within-host Survival and Between-host Transmission J Biol Rhythms 2017 32 516 533 28845736 8 Rijo-Ferreira F Pinto-Neves D Barbosa-Morais NL Takahashi JS Figueiredo LM Trypanosoma brucei metabolism is under circadian control Nat Microbiol 2017 2 17032 28288095 9 Scheiermann C Kunisaki Y Frenette PS Circadian control of the immune system Nat Rev Immunol 2013 13 190 23391992 10 Curtis AM Bellet MM Sassone-Corsi P O’Neill LAJ Circadian Clock Proteins and Immunity Immunity 2014 40 178 186 24560196 11 Zasłona Z The circadian protein BMAL1 in myeloid cells is a negative regulator of allergic asthma Am J Physiol Cell Mol Physiol 2017 312 L855 L860 12 Keller M A circadian clock in macrophages controls inflammatory immune responses Proc Natl Acad Sci 2009 106 21407 LP 21412 19955445 13 Nguyen KD Circadian Gene <em>Bmal1</em> Regulates Diurnal Oscillations of Ly6C<sup>hi</sup> Inflammatory Monocytes Science (80-) 2013 341 1483 LP 1488 14 Haus E Smolensky MH Biologic Rhythms in the Immune System Chronobiol Int 1999 16 581 622 10513884 15 Haus E Lakatua DJ Swoyer J Sackett-Lundeen L Chronobiology in hematology and immunology Am J Anat 2018 168 467 517 16 Labrecque N Cermakian N Circadian Clocks in the Immune System J Biol Rhythms 2015 30 277 290 25900041 17 Graham AL Allen JE Read AF Evolutionary Causes and Consequences of Immunopathology Annu Rev Ecol Evol Syst 2005 36 373 397 18 Kerr AM Gershman SN Sakaluk SK Experimentally induced spermatophore production and immune responses reveal a trade-off in crickets Behav Ecol 2010 21 647 654 19 Roden LC Ingle RA Lights, Rhythms, Infection: The Role of Light and the Circadian Clock in Determining the Outcome of Plant–Pathogen Interactions Plant Cell 2009 21 2546 LP 2552 19789275 20 Bhardwaj V Meier S Petersen LN Ingle RA Roden LC Defence Responses of Arabidopsis thaliana to Infection by Pseudomonas syringae Are Regulated by the Circadian Clock PLoS One 2011 6 e26968 22066021 21 Ingle RA Jasmonate signalling drives time-of-day differences in susceptibility of Arabidopsis to the fungal pathogen Botrytis cinerea Plant J 2015 84 937 948 26466558 22 Bellet MM Circadian clock regulates the host response to Salmonella Proc Natl Acad Sci 2013 110 9897 LP 9902 23716692 23 Kiessling S The circadian clock in immune cells controls the magnitude of Leishmania parasite infection Sci Rep 2017 7 10892 28883509 24 Zuk M Rotenberry JT Tinghitella RM Silent night: adaptive disappearance of a sexual signal in a parasitized population of field crickets Biol Lett 2006 2 521 LP 524 17148278 25 Levri EP Lively CM The effects of size, reproductive condition, and parasitism on foraging behaviour in a freshwater snail, Potamopyrgus antipodarum Anim Behav 1996 51 891 901 26 Ponton F Water-seeking behavior in worm-infected crickets and reversibility of parasitic manipulation Behav Ecol 2011 22 392 400 22476265 27 Hopwood TW The circadian regulator BMAL1 programmes responses to parasitic worm infection via a dendritic cell clock Sci Rep 2018 8 3782 29491349 28 Johnson CH Zhao C Xu Y Mori T Timing the day: what makes bacterial clocks tick? Nat Rev Microbiol 2017 15 232 28216658 29 Zuk M Simmons L Cupp L Calling characteristics of parasitized and unparasitized populations of the field cricket Teleogryllus oceanicus Behav Ecol Sociobiol 1993 33 339 343 30 Clark IA Budd AC Alleva LM Sickness behaviour pushed too far–the basis of the syndrome seen in severe protozoal, bacterial and viral diseases and post-trauma Malar J 2008 7 208 18854046 31 Dantzer R O’Connor JC Freund GG Johnson RW Kelley KW From inflammation to sickness and depression: when the immune system subjugates the brain Nat Rev Neurosci 2008 9 46 18073775 32 Ghai RR Fugere V Chapman CA Goldberg TL Davies TJ Sickness behaviour associated with non-lethal infections in wild primates Proc R Soc B 2015 282 20151436 33 Kluger MJ Phylogeny of fever Federation Proceedings 1979 38 30 34 759235 34 Evans SS Repasky EA Fisher DT Fever and the thermal regulation of immunity: the immune system feels the heat Nat Rev Immunol 2015 15 335 25976513 35 Kluger MJ Ringler DH Anver MR Fever and survival Science (80-) 1975 188 166 168 36 Schulman CI The effect of antipyretic therapy upon outcomes in critically ill patients: a randomized, prospective study Surg Infect (Larchmt) 2005 6 369 375 16433601 37 Earn DJD Andrews PW Bolker BM Population-level effects of suppressing fever Proc R Soc B 2014 281 20132570 38 Levi F Schibler U Circadian rhythms: mechanisms and therapeutic implications Annu Rev Pharmacol Toxicol 2007 47 593 628 17209800 39 Matthews JH Marte E Halberg F A Circadian susceptibility-resistance cycle to fluothane in male B 1 mice Can Anaesth Soc J 1964 11 280 14161442 40 Zhang R Lahens NF Ballance HI Hughes ME Hogenesch JB A circadian gene expression atlas in mammals: implications for biology and medicine Proc Natl Acad Sci 2014 111 16219 16224 25349387 41 Hawking F The 24-hour periodicity of microfilariae: biological mechanisms responsible for its production and control Proc R Soc Lond B 1967 169 59 76 42 Mouahid G A new chronotype of Schistosoma mansoni: adaptive significance Trop Med Int Heal 2012 17 727 732 43 Martinaud G Billaudelle M Moreau J Circadian variation in shedding of the oocysts of Isospora turdi (Apicomplexa) in blackbirds (Turdus merula): An adaptative trait against desiccation and ultraviolet radiation Int J Parasitol 2009 39 735 739 19100268 44 Prior KF Timing of host feeding drives rhythms in parasite replication PLoS Pathog 2018 14 e1006900 29481559 45 Hirako IC Daily Rhythms of TNFα Expression and Food Intake Regulate Synchrony of Plasmodium Stages with the Host Circadian Cycle Cell Host Microbe 2018 46 Reece SE Prior KF Malaria Makes the Most of Mealtimes Cell Host Microbe 2018 23 695 697 29902432 47 Fenske MP Nguyen LP Horn EK Riffell JA Imaizumi T Circadian clocks of both plants and pollinators influence flower seeking behavior of the pollinator hawkmoth Manduca sexta Sci Rep 2018 8 2842 29434312 48 Kharouba HM Global shifts in the phenological synchrony of species interactions over recent decades Proc Natl Acad Sci 2018 115 5211 5216 29666247 49 Hevia MA Canessa P Müller-Esparza H Larrondo LF A circadian oscillator in the fungus Botrytis cinerea regulates virulence when infecting Arabidopsis thaliana Proc Natl Acad Sci 2015 112 8744 8749 26124115 50 Rensing L Meyer-Grahle U Ruoff P Biological timing and the clock metaphor: oscillatory and hourglass mechanisms Chronobiol Int 2001 18 329 369 11475408 51 Mrosovsky N Masking: history, definitions, and measurement Chronobiol Int 1999 16 415 429 10442236 52 Sougoufara S Biting by Anopheles funestus in broad daylight after use of long-lasting insecticidal nets: a new challenge to malaria elimination Malar J 2014 13 125 24678587 53 Rund SSC O’Donnell AJ Gentile JE Reece SE Daily rhythms in mosquitoes and their consequences for malaria transmission Insects 2016 7 14 27089370 54 Schneider P Adaptive periodicity in the infectivity of malaria gametocytes to mosquitoes bioRxiv 2018 294942 55 Pigeault R Caudron Q Nicot A Rivero A Gandon S Timing malaria transmission with mosquito fluctuations Evol Lett 2018 56 Thomas F Rigaud T Brodeur J Evolution of parasite-induced behavioral alterations 2010 57 De Bekker C Merrow M Hughes DP From behavior to mechanisms: an integrative approach to the manipulation by a parasitic fungus (Ophiocordyceps unilateralis sl) of its host ants (Camponotus spp.) Integr Comp Biol 2014 54 166 176 24907198 58 Ko CH Takahashi JS Molecular components of the mammalian circadian clock Hum Mol Genet 2006 15 R271 R277 16987893 59 Rijo-Ferreira F Sleeping sickness is a circadian disorder Nat Commun 2018 9 62 29302035 60 Edgar RS Cell autonomous regulation of herpes and influenza virus infection by the circadian clock Proc Natl Acad Sci 2016 113 10085 10090 27528682 61 Thomas F Do hairworms (Nematomorpha) manipulate the water seeking behaviour of their terrestrial hosts? J Evol Biol 2002 15 356 361 62 Biron DG ‘Suicide’of crickets harbouring hairworms: a proteomics investigation Insect Mol Biol 2006 15 731 742 17201766 63 Hughes ME Guidelines for genome-scale analysis of biological rhythms J Biol Rhythms 2017 32 380 393 29098954 64 Lively CM Evidence from a New Zealand snail for the maintenance of sex by parasitism Nature 1987 328 519 65 Levri EP Parasite-induced change in host behavior of a freshwater snail: parasitic manipulation or byproduct of infection? Behav Ecol 1999 10 234 241 66 Hoover K A gene for an extended phenotype Science (80-) 2011 333 1401 67 Goulson D Wipfelkrankheit: modification of host behaviour during baculoviral infection Oecologia 1997 109 219 228 28307172 68 de Bekker C Gene expression during zombie ant biting behavior reflects the complexity underlying fungal parasitic behavioral manipulation BMC Genomics 2015 16 620 26285697 69 de Bekker C Will I Das B Adams RMM The ants (Hymenoptera: Formicidae) and their parasites: effects of parasitic manipulations and host responses on ant behavioral ecology Myrmecological News 2018 28 1 24 70 Herbison R Lagrue C Poulin R The missing link in parasite manipulation of host behaviour Parasit Vectors 2018 11 222 29615121 71 Johnson CH Elliott J Foster R Honma K Kronauer R Chronobiology: Biological Timekeeping 2004 72 Michael TP Enhanced Fitness Conferred by Naturally Occurring Variation in the Circadian Clock Science (80-) 2003 302 1049 LP 1053 73 Dodd AN Plant Circadian Clocks Increase Photosynthesis, Growth, Survival, and Competitive Advantage Science (80-) 2005 309 630 LP 633 74 Stephan FK The “other” circadian system: food as a Zeitgeber J Biol Rhythms 2002 17 284 292 12164245 75 Young MW Kay SA Time zones: a comparative genetics of circadian clocks Nat Rev Genet 2001 2 702 11533719 76 Dunlap JC Molecular bases for circadian clocks Cell 1999 96 271 290 9988221 77 Chen Z Odstrcil EA Tu BP McKnight SL Restriction of DNA Replication to the Reductive Phase of the Metabolic Cycle Protects Genome Integrity Science (80-) 2007 316 1916 LP 1919 78 Ouyang Y Andersson CR Kondo T Golden SS Johnson CH Resonating circadian clocks enhance fitness in cyanobacteria Proc Natl Acad Sci 1998 95 8660 8664 9671734 79 Nelson BV Vance RR Diel foraging patterns of the sea urchin Centrostephanus coronatus as a predator avoidance strategy Mar Biol 1979 51 251 258 80 Spoelstra K Wikelski M Daan S Loudon ASI Hau M Natural selection against a circadian clock gene mutation in mice Proc Natl Acad Sci 2016 113 686 691 26715747 81 Stone EF The circadian clock protein timeless regulates phagocytosis of bacteria in Drosophila PLoS Pathog 2012 8 e1002445 22253593 82 Lee J-E Edery I Circadian regulation in the ability of Drosophila to combat pathogenic infections Curr Biol 2008 18 195 199 18261909 83 van der Vinne V Cold and hunger induce diurnality in a nocturnal mammal Proc Natl Acad Sci 2014 111 15256 15260 25288753 84 Bloch G Robinson GE Chronobiology: reversal of honeybee behavioural rhythms Nature 2001 410 1048 11323660 85 Bulla M Unexpected diversity in socially synchronized rhythms of shorebirds Nature 2016 540 109 27880762 86 Hughes DP Behavioral mechanisms and morphological symptoms of zombie ants dying from fungal infection BMC Ecol 2011 11 13 21554670 87 De Bekker C Ohm RA Evans HC Brachmann A Hughes DP Ant-infecting Ophiocordyceps genomes reveal a high diversity of potential behavioral manipulation genes and a possible major role for enterotoxins Sci Rep 2017 7 12508 28970504 88 Fredericksen MA Three-dimensional visualization and a deep-learning model reveal complex fungal parasite networks in behaviorally manipulated ants Proc Natl Acad Sci 2017 201711673 89 Garcia CRS Markus RP Madeira L Tertian and quartan fevers: temporal regulation in malarial infection J Biol Rhythms 2001 16 436 443 11669417 90 Donnell AJO Schneider P Mcwatters HG Reece SE Fitness costs of disrupting circadian rhythms in malaria parasites Fitness costs of disrupting circadian rhythms in malaria parasites Proc R Soc B Biol Sci 2011 2429 2436 10.1098/rspb.2010.2457 91 Gibbs JE The nuclear receptor REV-ERBα mediates circadian regulation of innate immunity through selective regulation of inflammatory cytokines Proc Natl Acad Sci 2012 109 582 587 22184247 92 Marpegan L Diurnal variation in endotoxin-induced mortality in mice: correlation with proinflammatory factors Chronobiol Int 2009 26 1430 1442 19916840 93 Gibbs J An epithelial circadian clock controls pulmonary inflammation and glucocorticoid action Nat Med 2014 20 919 25064128 94 Druzd D Lymphocyte circadian clocks control lymph node trafficking and adaptive immune responses Immunity 2017 46 120 132 28087238 95 Gagnidze K Nuclear receptor REV-ERBα mediates circadian sensitivity to mortality in murine vesicular stomatitis virus-induced encephalitis Proc Natl Acad Sci 2016 201520489 96 Kuo T-H Pike DH Beizaeipour Z Williams JA Sleep triggered by an immune response in Drosophila is regulated by the circadian clock and requires the NFκB Relish BMC Neurosci 2010 11 17 20144235 97 Murdock CC Moller-Jacobs LL Thomas MB Complex environmental drivers of immunity and resistance in malaria mosquitoes Proc R Soc B 2013 280 20132030 98 Griebel T Zeier J Light regulation and daytime dependency of inducible plant defenses in Arabidopsis: phytochrome signaling controls systemic acquired resistance rather than local defense Plant Physiol 2008 147 790 801 18434604 99 Korneli C Danisman S Staiger D Differential control of pre-invasive and post-invasive antibacterial defense by the Arabidopsis circadian clock Plant Cell Physiol 2014 55 1613 1622 24974385 100 Wang W Timing of plant immune responses by a central circadian regulator Nature 2011 470 110 21293378 101 Du LY The innate immune cell response to bacterial infection in larval zebrafish is light-regulated Sci Rep 2017 7 12657 28978916 102 Lazado CC Skov PV Pedersen PB Innate immune defenses exhibit circadian rhythmicity and differential temporal sensitivity to a bacterial endotoxin in Nile tilapia (Oreochromis niloticus) Fish Shellfish Immunol 2016 55 613 622 27346154 103 Prendergast BJ Circadian disruption alters the effects of lipopolysaccharide treatment on circadian and ultradian locomotor activity and body temperature rhythms of female Siberian hamsters J Biol Rhythms 2015 30 543 556 26566981

PMC007xxxxxx/PMC7614807.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7806090 Neurosci Biobehav Rev Neurosci Biobehav Rev Neuroscience and biobehavioral reviews 0149-7634 1873-7528 34563562 7614807 10.1016/j.neubiorev.2021.09.034 EMS179561 Article The road towards understanding embodied decisions Gordon Jeremy 1 Maselli Antonella 2 Lancia Gian Luca 2 Thiery Thomas 3 Cisek Paul 4 Pezzulo Giovanni 2 1 University of California, Berkeley, Berkeley, CA 94704, US 2 Institute of Cognitive Sciences and Technologies, National Research Council, Rome, Italy 3 Department of Psychology, University of Montréal, Montréal, Québec, Canada 4 Department of Neuroscience, University of Montréal, Montréal, Québec, Canada Contact information: Giovanni Pezzulo, Institute of Cognitive Sciences and Technologies, National Research Council, Via S. Martino della Battaglia 44, 00185 Rome, Italy, giovanni.pezzulo@istc.cnr.it 01 12 2021 23 9 2021 17 7 2023 25 7 2023 131 722736 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Most current decision-making research focuses on classical economic scenarios, where choice offers are prespecified and where action dynamics play no role in the decision. However, our brains evolved to deal with different choice situations: “embodied decisions”. As examples of embodied decisions, consider a lion that has to decide which gazelle to chase in the savannah or a person who has to select the next stone to jump on when crossing a river. Embodied decision settings raise novel questions, such as how people select from time-varying choice options and how they track the most relevant choice attributes; but they have long remained challenging to study empirically. Here, we summarize recent progress in the study of embodied decisions in sports analytics and experimental psychology. Furthermore, we introduce a formal methodology to identify the relevant dimensions of embodied choices (present and future affordances) and to map them into the attributes of classical economic decisions (probabilities and utilities), hence aligning them. Studying embodied decisions will greatly expand our understanding of what decision-making is. embodied decisions affordances action-perception loop planning pmcIntroduction The ability to make effective value-based decisions is crucial for the survival of living organisms. Despite its popularity in psychology and neuroscience (Gold and Shadlen, 2007), value-based decision-making is often studied in restricted laboratory settings, using simple tasks that are inspired by economic theory, such as binary choices between lotteries (e.g., do you prefer 50 dollars with 20% probability or 20 dollars with 50% probability) or intertemporal offers (e.g., do you prefer 20 dollars now or 50 dollars in a month?). In this “classical” setting, there are a limited number of choices to be selected (often two) that are prespecified by the experimenter and presented simultaneously. Furthermore, the relevant choice dimensions (e.g., rewards and probabilities) are usually fixed throughout the experiment and unambiguous. In human studies, choice dimensions are usually labeled with symbols and easy to identify. Similarly, in animal studies, choice dimensions become stereotyped and unambiguous as an effect of long learning periods. Finally, the action component is often trivialized, not just because the action itself is simple (e.g., a button press) but also because there is no effect of the action on subsequent perception - i.e., there is no action-perception loop. The experiments conducted in this classical setting have crystallized a serial view of decision-making, which identifies three distinct sequential stages (Fodor, 1983; Pylyshyn, 1984): the perception of the attributes of the predefined alternatives, the decision between the alternatives, and the reporting of the decision by action (i.e., decide-then-act). While using the classical setting permits methodological rigor, this comes at the expense of an excessive focus on just one kind of choice - economic choices that can be most easily studied in the lab - while disregarding the fact that there are other kinds of choices that are equally (or even more) frequent in our lives and important from an evolutionary perspective. The classical setting is relatively well suited to certain types of choices, such as choosing between a fixed number of dishes from a restaurant menu. However, these are just a small subset of the choices that we make in our everyday lives. For example, on the way to the restaurant one made countless choices about actions, such as whether to stop or try to pass through an intersection, how to get around other people walking on the sidewalk, which chair to sit in, etc. In many of these situations, the choices themselves as well as their costs and benefits are always in flux, and they are difficult to model using classical concepts from economic choice settings. Furthermore, classical economic choices are not the kinds of scenarios that drove the evolution of the brain’s neural mechanisms. Instead, throughout its long history the brain adapted to deal with very different choice situations, such as deciding whether to approach or avoid an object, how to navigate to feeding sites, and how to move among obstacles (some of which might themselves be in motion). These kinds of “embodied decisions” dominated animal behavior long before humans existed, and are accomplished by neural mechanisms that have been conserved for hundreds of millions of years (Rodríguez et al., 2002; Saitoh et al., 2007; Striedter and Northcutt, 2019). Furthermore, the innovations of neural circuitry introduced in mammals and more recently, primates, were all made within the constraints of that ancestral context (Cisek, 2019; Passingham and Wise, 2012). Even now, in our daily lives, humans continue to perform many similar embodied decisions every day, such as when we play a sport, drive on a busy road, prepare a meal or play hide-and-seek with children, and it has often been suggested that our cognitive abilities are constructed upon a scaffolding provided by the sensorimotor strategies of such embodied behavior (Hendriks-Jansen, 1996; Piaget, 1952). Thus, one could argue that understanding embodied decisions is of primary importance for understanding many aspects of human cognition and behavior. Embodied decisions − and how they differ from standard decision settings Embodied decision settings are very different from classical settings for a number of reasons (Cisek and Pastor-Bernier, 2014) (Figure 1). First, the number of offers that enter the deliberation is not predefined and can vary on a moment-by-moment basis. For example, it is likely that a lioness in front of hundreds of gazelles does not consider each as a separate offer, but instead clusters them into “patches” of potential food sources. Second, choice offers and their dimensions are rarely discretized or labeled with symbols and units (e.g., “option 1”, “option 2”, “dollars”, “percentages”). Rather, the decision-maker has to identify the options perceptually (under uncertainty) and to select the relevant choice dimensions; these choice dimensions often include geometric dimensions and affordances, such as the relative lion-gazelle distance, which are disregarded in classical settings. Finally, and most importantly, perception, decision and action dynamics are intertwined. The action component is not simply a way to report a choice but an essential component to secure the reward (i.e., the lion has to actually chase the gazelle). Furthermore, action dynamics change the perceptual landscape (e.g., one or more gazelles can go out of sight when the animal starts moving) and the decision landscape (e.g., some gazelles can go out of reach and hence cease to be valid offers, or conversely become more available and extend the offer menu). Action and decision dynamics cannot be fully separated in time and are instead “continuous” (Yoo et al., 2021). The decision-maker can start acting before completing the decision, to buy time (Barca and Pezzulo, 2012; Pezzulo and Ognibene, 2011) or to exploit an option that would otherwise disappear; for example, a lion can start running toward a group of gazelles before deciding which one to chase, otherwise the gazelles will simply run away (i.e., act while deciding). Furthermore, the decision-maker can change their mind while acting, due to reconsidering a previous decision (Resulaj et al., 2009), gathering novel evidence, or the appearance of novel opportunities, e.g., previously ignored gazelles may come within reach (i.e., decide while acting). These examples illustrate that there are crucial differences between the classical laboratory settings in which decisions are usually studied and the embodied settings in which real life decisions are deployed. While it is intuitively clear that studying decisions in restricted laboratory settings is methodologically simpler than studying them “in the wild”, it worth asking whether certain aspects of lab settings, such as the focus on stable, predefined alternatives and the trivialization of action, have contributed to a distorted (or at least incomplete) view of how we solve decision tasks. For example, one may ask to what extent the serial view of decision-making (decide-then-act) applies beyond simple binary decisions in the lab, or whether capturing the richness of real life decisions (act while deciding and decide while acting) requires a different class of (embodied) models, which acknowledge the fact that decision and action dynamics deploy in parallel and influence each other bidirectionally (Lepora and Pezzulo, 2015). For example, a classic model of decision-making posits that one deliberates by accumulating sensory evidence until it reaches a threshold, and initiates movements at that time (Gold and Shadlen, 2007; Ratcliff, 1978). But how does that model generalize to situations, common during real-time behavior, in which one must already be acting (and thus “above threshold”) while still deliberating? Must the very concept of a threshold be abandoned when considering embodied settings, and if so, then where does that leave current models? Recent studies are starting to examine how humans make decisions during ongoing actions (Grießbach et al., 2021; Michalski et al., 2020) but models of the underlying mechanisms will need to go beyond traditional ideas of initiation thresholds, possibly toward ideas of task-specific subspaces in high-dimensional neural populations (Kaufman et al., 2014). We argue that to broaden our understanding of decision-making, it is necessary to go beyond restricted laboratory settings and design novel experimental paradigms bringing embodied dynamics tasks under controlled conditions. To this end, it is necessary to remove some conceptual and methodological barriers that make embodied decisions challenging to address. Towards a deeper understanding embodied decisions This paper has three goals. The first goal is clarifying what are the key characteristics of embodied decisions and the novel questions they raise. The second goal is identifying a novel methodology for the study of embodied decisions, by distilling key insights from recent studies in sports analytics, experimental psychology and other fields. The third goal is discussing to what extent the study of embodied decisions requires novel theories − or will change our understanding of what decisions are. Specifically, the remainder of this paper addresses the following three points: (1) Novel questions. What are the key differences between classical and embodied decision settings? What are the novel experimental questions that only arise when studying embodied settings and are instead ignored in classical settings? What can we learn from embodied settings that we cannot learn in the “classical” way? (2) Novel methodologies. How can we experimentally address the above empirical questions? Given that studying embodied decisions poses additional challenges compared to restricted laboratory settings, is it possible to develop a methodology that does not sacrifice rigor? Can we identify success cases? Can we borrow methodologies from other fields that address similar problems? (3) Novel theories. Would a widespread use of embodied settings change the way we understand decision-making? Is it possible that by studying decision-making in the classical way we have mischaracterized its mechanisms? Will the study of embodied decisions require novel conceptual frameworks and how would they differ from standard decision models? What will be the impact of novel studies of embodied decision making for cognitive science, neuroscience, robotics and other fields? What are the novel questions that we can ask by studying embodied decisions? In the introduction, we argued that embodied decision settings include a number of dimensions that are missing from classical settings. Here, we reconsider these unique dimensions of embodied settings and highlight that they prompt novel research questions that are impossible or difficult to study in classical settings. Below we focus on three novel questions that, despite being not exhaustive, exemplify well the heuristic potential of embodied settings. To illustrate the novel questions, we will use the example of a soccer player who decides where (or to which teammate) to pass the ball. Question 1. How are offers and their attributes identified? In classical settings, the offers are clearly identifiable by the decision-maker and their number is fixed and known in advance. The situation is different for the soccer player, because (despite an awareness that she has 10 teammates) she might not know exactly where they are and would hardly consider all of them as good choices. This situation prompts the novel question of how many (and which) choice alternatives she considers in the first place. A closely related question is what are the dimensions that the soccer player considers during the decision. The choice can clearly benefit from considering multiple possible dimensions, such as the distance from the player, whether the positioning of the receiving player is appropriate or advantageous to score a goal, etc. We will argue in the next Section that it is advantageous to group these dimensions into the two classes of probabilities and utilities, because this permits establishing a mapping with the two usual dimensions of expected value in economics. Crucially, however, the soccer player is not provided a priori with a list of the dimensions to consider and of the values of these dimensions, but has to figure them out as part of the decision-making task. This prompts a number of additional questions: which dimensions are considered? To what extent are these dimensions context-sensitive and time-varying? What are the contextual factors (e.g., situated aspects of the choice) and individual differences (e.g. personality or cognitive factors) that influence the selection of relevant dimensions? Another set of questions regards the way the values of these dimensions are estimated under uncertainty and how this influences dimension selection (e.g., it would be ineffective to select a choice dimension for which one has no access to reliable values). In classical studies of value-based decision-making, the perceptual phase is often trivial. Under the rubric of perceptual decision-making or attention selection, perceptual dynamics are studied in isolation from value-based computations. In contrast, in the domain of embodied decisions, the different (e.g., perceptual, attentional, value-based) aspects of the decision process are tightly intertwined. While studying each in isolation may be fruitful as a first approximation, embodied settings may offer a unique window into how they interact. Finally, the classical setup suggests that the identification of offers and evaluation of their attributes are largely sequential processes. Conversely, embodied decision settings allow us to study their interactions. For example, as you identify the offers, you may discover what are the attributes that differentiate them, which then helps to eliminate some offers from consideration (e.g. some players may be just too far to be considered, or too close to an opponent), leaving a new set of offers with a new set of attributes to consider (e.g. some players may be moving too fast). These examples suggest that, first, the consideration of some attributes comes before and influences the identification of offers; and second, the identification of offers could be considered as the first stage of a decision process, which filters out irrelevant choice alternatives. These are all problems that hardly arise in classical settings. Question 2. How is the deliberation between choice offers performed? Classical settings start from the premise that choice offers are predefined and presented in parallel; and the offer-attribute mapping is fixed throughout the decision. These assumptions are directly incorporated into decision-making models, like drift-diffusion models (Ratcliff et al., 2016), accumulator models (Usher and McClelland, 2001), or attractor-based models (Wang, 2008). For example, the drift diffusion model requires fixing two thresholds (one for each offer) and the decision variable (that reflects the value of the first offer minus the value of the second offer) for all the computation. The attractor-based model requires fixing the synaptic connections between populations that encode attribute values and populations that encode offer values. Without these fixed mappings, the mathematical guarantees of these models break down. However, as we have discussed above, in embodied decisions neither offers nor their mapping with attributes are predefined or fixed, nor can we assume they are presented simultaneously. It is unclear whether the most popular decision-making models can deal with these complexities in embodied situations, both mathematically and conceptually (e.g., embodied choices would require a rapid reconfiguration of the neural architecture supporting attractor-based decisions, but it is unclear to what extent this is plausible). One may therefore argue that we need a different conceptualization of the deliberative process that leads to embodied decisions. At minimum, we should consider that the mathematical guarantees of models that deal with fixed data streams are not appropriate for embodied decisions, where the choice conditions change over time and hence the most recent data streams are more relevant (Cisek et al., 2009). A more drastic change of perspective comes from studies of foraging (Charnov, 1976) (which have strong decide-while-act components), where the standard assumption is that the deliberation is between “select the current offer” (exploit) vs. “search for other offers” (explore), not “decide between two fixed choices” as in classical decision settings. A series of studies suggest that birds (and other animals) use by default strategies for dealing with sequential choices and these lead them to make irrational decisions when faced with simultaneous choices (Kacelnik et al., 2011). However, it remains an open question whether this or other perspectives offer a broader and more appropriate conceptualization of situated decision-making in real life situations (we will return to this point later). Question 3. How do perception, decision and action processes influence each other? In embodied settings, perception, decision and action processes are deeply intertwined. Leaving aside perceptual processes (which we briefly discussed above), the interplay of decision and action - and the “continuity” of the decision (Yoo et al., 2021) - is evidenced by the fact that decision-makers can start acting before completing a decision (act-whiledeciding) and can change their mind along the way, perhaps to consider new offers that were not initially present (decide-while-acting). These situations create novel opportunities that are rarely (if ever) addressed in classical settings. As an example of a novel question, it is unclear at what point of the decision a commitment emerges for one of the alternatives and under which conditions a person can change their mind (Cos et al., 2021). Additionally, what is the range of novel strategies that embodied settings afford? For example, a common observation in studies that ask people to answer by clicking one of two response buttons with a computer mouse is that participants sometimes move rapidly in between the two buttons, perhaps as a strategy to “buy time” and avoid committing until necessary (Pezzulo and Ognibene, 2011). These strategies are not normally available in classical settings but are very important in real life, where decision commitment dynamics are less constrained and postponing is often an option. Finally, while decisions are normally treated as independent from the context and from one another, sequential dynamics are ever-present in embodied settings. As each decision and action influences subsequent decisions, actions and perception dynamics, the “dynamical affordance landscape” of decisions (Pezzulo and Cisek, 2016) cannot be easily disregarded. For example, if the soccer player has the ball and goes to the left, he will open some affordances (a left attack) but also close others (an attack to the right). What novel methodologies do we need to address these new questions? Embodied decision settings prompt several novel research questions. Some studies are already meeting these challenges. For example, studies that track continuous (eye, hand or mouse) movement dynamics during the choice are revealing how decision and action processes influence each other; for example, how decision uncertainty provokes changes of movement trajectories during a continuous decision; and how motor costs influence the initial decision and the subsequent changes of mind (Barca and Pezzulo, 2015; Burk et al., 2014, 2014; Cos et al., 2021; Marcos et al., 2015; Resulaj et al., 2009; Spivey, 2007; Spivey and Dale, 2006)(Spivey et al., 2005). Other studies are addressing how we make decisions while we are already engaged in performing an action are revealing the importance of geometric and situated aspects, such as the relative distances between currently pursued and another available target, influence the choice of changing target (Michalski et al., 2020). Studies of embodied decision-making sometimes target interactive situations too, such as joint decisions (Pezzulo et al., 2021b) and joint actions, either collaborative, as in grasping something together, or competitive, as in martial arts (Candidi et al., 2015; Pezzulo et al., 2013; Pezzulo and Dindo, 2011; Sebanz et al., 2006; Yamamoto et al., 2013). These studies reveal how the decision and action processes of two (or more) co-actors are interdependent and how they can become aligned over time, especially when the two can engage in a reciprocal sensorimotor (nonverbal) communication (Pezzulo et al., 2018). Another example is a promising trend in systems neuroscience toward studies of more naturalistic settings. This includes neural recordings in animals that are freely moving (Chestek et al., 2009; Schwarz et al., 2014; Sodagar et al., 2007) or navigating through controlled environments (Etienne et al., 2014; Krumin et al., 2018), as well as studies in humans using portable magnetoencephalography or encephalography equipment (Boto et al., 2018; Djebbara et al., 2021) (Djebbara et al., 2019) (Topalovic et al., 2020). Towards a methodology that connects classical and embodied domains of decision Despite these promising developments, most aspects of embodied decisions remain unaddressed − in part because embodied decisions are more challenging to study compared to classical settings. A key question is whether it is possible to develop a sound methodology for the study of embodied decisions in their full complexity, without sacrificing rigor. One aspect that makes the study of embodied decisions challenging is that most of the choice dimensions reflect spatial and geometric aspects of the situation, such as “passability” affordances for a soccer player. Unlike choice dimensions usually considered in classical settings, such as dollars and seconds, affordances are more difficult to formalize − but they are exactly the dimensions that our situated brains evolved for. The methodology we propose here involves mapping the choice dimensions of embodied problems into the two factors − probabilities and utilities − that are often used in classical problems to define expected value. Establishing a formal correspondence between classical and embodied settings will permit addressing classical questions about (for example) utility maximization, risk-avoidance and sunk costs in embodied settings. Example 1: The case of soccer Fortunately, there are some success cases that exemplify this methodology, especially in sports analytics, as in the case of statistical studies of basketball (Cervone et al., 2016) and soccer (Fernández et al., 2019). For example, Figure 2 shows an example of a soccer player (yellow circle) who has to decide to whom he wants to pass the ball (small green circle) − or more precisely, where to pass it, since the passage can be to any location of the field, not just to his teammates’ current positions. The figure shows the “expected pass value (EPV)” surface for the soccer player (zones in red and blue have high and low EPVs, respectively). The notion of EPV is exactly the same notion of expected value used in economic studies and it is calculated by combining the usual two factors of probabilities and utilities. In this setting, the probability that a pass will succeed is itself modeled as a function of several subfactors, such as the distance from teammates and whether the passage zone is “under control” by a teammate or opponent. The utility of a pass is modeled as spanning both positive and negative dimensions, in consideration of the fact that passing to a certain zone can change the chances of scoring a goal (positive dimension) or receiving a counterattack (negative dimension). The positive and negative dimensions of utility are thus calculated separately by considering several subfactors, such as where the target of the pass is located within a precomputed “zone value surface” that spans the whole playing field (and is higher in the opponent’s field), and how many times a successful pass to the target zone resulted in a goal or a counterattack in the model’s training data. The values of each of the subfactors that contribute to probabilities or utilities are in part derived analytically (e.g., by calculating the zones that each player “controls”) and in part learned from data (e.g., how many times a goal was scored from each zone of the playing field, in a large database of recent matches). While the ways probabilities and utilities are calculated is context dependent (e.g., depends on specific game models that differ for soccer, basketball and other games), the methodology is general and it permits establishing a strong formal connection between embodied and classical settings. In other words, regardless of the difficulties of establishing how probabilities and utilities should be modeled for each embodied setting, and how to estimate their values, once these two factors have been estimated they can be combined to form an expected value surface, in the same way they are combined in classical settings and economics. This is evident if one considers that in Figure 2, it is possible to identify the choices (passes) that provide the highest rewards, defined here as moving the ball into an area from which it is possible to shoot at the goal (red arrows), those that provide the best expected value (green arrow) and those that provide the lowest risk of a counterattack (cyan arrow). This makes it possible to ask (for example) whether the soccer player selected the pass that maximizes utility or whether he is risk-seeking or risk-averse. Interestingly, despite the EPV being continuous, in most cases it is possible to enumerate a discrete (and small) number of choice offers, by focusing on the “peaks” of EPV on the field. This exemplifies a possible approach to map continuous aspects of embodied decisions (e.g., physical space) to discrete choice dimensions, which are most commonly used in standard decision-making models. However, while standard decision-making models specify a-priori a discrete set of options, it is possible that these emerge from continuous surfaces. This phenomenon can be formalized using various computational models that show decision surfaces developing peaks and troughs (Amari, 1977; Cisek, 2006; Furman and Wang, 2008; Sandamirskaya et al., 2013; Schneegans and Schöner, 2008) and transitioning from representing multiple options to choosing a single winner (Grossberg, 1973; Standage et al., 2011). Example 2: The case of crossing the river We illustrate how to use the same methodology used in the soccer study to derive “expected value surfaces” to deal with other embodied decision settings; for example, the case of a person who has to cross a river by jumping between stones having different sizes and placed at different distances, see Figure 3. As in the soccer example, this situation can be addressed by deriving various “surfaces” that consider spatial and geometric elements of the task, such as child-stone distance and stone sizes (as examples of probability surfaces) and distance between stones and the end of the river (as an example of a utility surface). Before analyzing the task of crossing the river, it is useful to mention explicitly the guiding principle that we use to distinguish probability and utility factors. In our analysis, the former (probability) factor relates to “present affordances” and the latter (utility) factor relates to “future affordances”, respectively (Gibson, 1977; Pezzulo and Cisek, 2016). Probabilities (and present affordances) only depend on the current situation (e.g., current position of the jumper, distance from the stones and their size), not the intended destination. As a result, computing the probability (or “present affordance”) surface amounts to asking: “which stones are most ‘jumpable’ given my current situation”? This question can be asked irrespective of the intended destination. On the contrary, the notion of utility depends on the future, up until the goal destination (in our example, the end of the river). Hence, computing the utility (or “future affordance”) surface amounts to asking: “Will this jump increase the chances of reaching the goal destination (and by how much)? What affordances will I create (or destroy) by doing a certain jump?” A concrete example may help. Let’s consider the jumper shown in Figure 3. He has the choice between three options: the small stone to the right, the small stone to the left and the big stone to the center. The right stone is much closer and hence has the greatest present affordance, but it has lesser utility, as it does not allow progress towards the goal. The center stone has the greatest utility as it is closer to the destination, but (despite its size) it may have a lower present affordance, given it is farther from the jumper. The jumper actually selects the left stone, which seems a good compromise despite having neither the greatest present affordance nor the greatest utility. Yet, its utility is noteworthy as it “opens up” novel affordances: it permits reaching the next, bigger stone with small effort - and eventually the goal destination. Figure 3b-f shows a schematic view of an affordance landscape summarizing the child’s decision (Figure 3a), with black circles representing the stones, and the white triangle representing the child’s present location and orientation. In this example, we formalize the “probability” (or “immediate affordance”) component in terms of two subcomponents. The first subcomponent of probability, Ad, takes into consideration the child-stone distance and prioritizes shorter jumps. It is calculated as: Ad(x,y)=−(x−xchild)2+(y−ychild)2 Note that the units in which x and y can be expressed relative to the subject’s size, to account for the fact that distance depends on bodily parameters (Warren, 1984). The second subcomponent of probability takes into consideration “landability” and assigns positive values to locations with stones and zero value to locations without stones (as it is impossible to land on water). For illustrative purposes, we show two landability affordances,which cover the whole stone or only its center, respectively; see Figure 3c. The former may be perceived by very accurate (or risk-seeking) jumpers, whereas the latter may be perceived by less accurate (or risk-averse) jumper, who may not perceive the outer edge of the stone being landable, knowing that their jumps have some variance and hence aiming at the outer edge poses a risk of falling (Trommershäuser et al., 2008). Furthermore, we formalize the “utility” component as the forward progress towards the end of the river, see Figure 3d. This is approximated as Au,0(x,y)=y−ychild This naive method does not correspond to the optimal utility surface (see below) but it is sufficient in this simple example to provide some directionality. By combining the two (probability and utility) components it is possible to calculate an “expected value (EV) surface”, see Figure 3e. This can be calculated simply as the product of the scaled probability and utility surfaces discussed above: EV0=(Ad⋅Al)⋅Au,0 We use min-max scaling, where Xscaled=X−XminXmax−Xmin Note that the EV surface is shown in two versions, each integrating one of the two landability surfaces shown in Figure 3c. It can be informative, as well, to visualize the EV surface with subsets of its subcomponents. For example, Figure 3f shows only the distance subcomponent of probability and utility (Figures 3b & 3d), which expresses the preferred location of stones irrespective of their true position, which may provide a useful guide during visual search - or if the task is to place stones optimally. While useful, the EV value surface is myopic as it only considers the immediate utility of each jump derived from a naive notion of “direction to goal”, not long-term utility. Given that we are dealing with a sequential decision problem that involves multiple jumps, long-term utilities should consider not just the utility of the next jump but also of the successive ones that may become available or unavailable as a function of the next jump; for example, using a look-ahead planning algorithm such as REINFORCE (Sutton and Barto, 1998). Look-ahead planning permits addressing cases in which the myopic strategy is maladaptive and leads to dead ends. One such case is illustrated in Figure 4. This is a variant of the situation shown in Figure 3, in which we moved the final stone from the left to the right. A myopic strategy would still assign the highest EV to the left stone, even if this is a dead end (Figure 4a). However, using a one-step look-ahead planning algorithm such as REINFORCE (Sutton and Barto, 1998), will permit predicting how the value surface will change after jumping to the left or right, respectively. In turn, considering this future utility instead of the immediate utility would change the EV surface at the start position − and assign the highest utility to the right stone (Figure 4b). This simple example illustrates the benefits of planning in situations where naive notions of utility (such as direction to goal) are misleading. From an empirical perspective, people may be quite effective in avoiding dead ends while crossing the river (in person or in a videogame-like setting), but at the expense of investing some (cognitive) cost − which may reflect the process of planning ahead, while overriding current affordances. Interestingly, such cognitive costs can be modeled as the computational costs required to update the expected value surface, from the “immediate” utility of Figure 4a to the “future” utility of Figure 4b-c. By using REINFORCE or a similar Monte Carlo algorithm, we can generate predictions for this cost as the number of rollout steps required to disambiguate between competing immediate affordances. In general, the greater the differences between the “immediate” utility surface of Figure 4a and the “future” utility surface of Figure 4b, the greater the computational cost to compute the latter (Ortega and Braun, 2013; Todorov, 2009; Zénon et al., 2019). If these computational costs correspond to mental effort, they could plausibly become apparent when comparing the RT distributions of peoples’ choices before a jump, when the “immediate” and “future” utility surfaces are more or less similar. Note that calculating the most accurate expected value surface may have a high computational cost. If people optimize a trade-off between the accuracy of the expected value surface and the computational costs of obtaining it, they could select adaptively the amount of resources to invest (e.g., number of rollouts; compare the two updated expected utility surfaces of Figure 4b-c which are obtained using 2 rollouts (Figure 4b) or 10 rollouts (Figure 4c) of the REINFORCE algorithm). In tasks that involve multiple choices, people might also select the strategy to plan ahead (and invest cognitive resources) only during the most critical decisions; for example, when they need to choose between two sequences of stones in opposite directions, but not in simpler situations, such as the one shown in Figure 3, where the benefits of planning might not be worth its cost. By looking at subjects’ choices during the task, it would be possible to infer the expected value surface that they actually computed, which could provide an indication their (cognitive) effort investment – and whether and how often they look ahead during embodied choices. Example 3: The case ofclimbing We consider one final application of “expected value surfaces”: the case of a rock climber who has to plan the best way to climb a wall − and more specifically, the best sequence of holds to traverse with their hands and feet, until reaching the “top” hold that marks the end of a climbing problem. Figure 5a shows a climber solving an example problem on a bouldering wall (called MoonBoard). The MoonBoard wall implies a standard configuration of all possible holds, and a problem is defined as a subset of these holds that are allowed. The goal is to start with each hand touching two predefined holds, then executing a sequence of movements until reaching a predefined “top” hold with both hands. Figure 5a shows a climber on an example bouldering problem on the MoonBoard, whereas Figures 5b-g show various probability and utility surfaces for the same problem (the holds that belong to the problem are marked with circles Figures 5b-g; and the current positions of the climber’s hands and foot are marked with short arcs around the holds; the green arc around marks the position of the climber’s left hand). As in the case of crossing the river, the value surfaces combine the two factors of “probabilities” (or current affordance, e.g., which holds are reachable given the current situation?) and “utilities” (or future affordance, e.g., which holds are more useful to open up future affordances that permit progress towards the top of the wall?). However, there is an important difference in the way we calculated “probabilities” in the two cases of crossing the river and climbing. In the case of crossing the river, we calculated them analytically, by only considering geometric aspects of the situation: agent-stone distance (for the distance subcomponent) and stone sizes (for the landability subcomponent). However, this analytical approach is much more challenging in the case of climbing, as a climber’s current affordances plausibly depend on a much larger set of factors (e.g., climbing hold types, posture, distance) that are challenging to model. For this, to model probability surfaces, we first identified a small set of subcomponents and then used a data-driven approach to learn their associated probabilities from a large dataset of climbing data, much like the soccer example described above (Fernández et al., 2019). This implies that the probability surfaces considered here reflect the average climbing patterns of multiple climbers. Figure 5b-e show the subcomponents of the probability surface that we consider in this example (which are provided as examples, but may not fully capture the full complexity of climbing). Figure 5b shows the first subcomponent, called “absolute landscape”, which depends only on the specific climbing hold (i.e., its type, shape, size) in any given position, not on the climber’s current position, and only depends on the kind and shape of the climbing hold. Here, the holds in the positions marked in red are “good” (or “easy to grasp and hold”) while those in blue are “bad” (or “difficult to grasp and hold”). The absolute landscape subcomponent is calculated empirically, using data freely available from the MoonBoard app (https://www.moonboard.com/moonboard-app), under the assumption that “good” (bad) climbing holds occur more often in easy (difficult) climbing routes. Figure 5c shows a second subcomponent of probability: the relative landscape. This is the likelihood of moving each limb to a particular region of the board and hence capturing aspects of limb-hold distance. Note that this subcomponent is specific for the current position of the climber; in the figure, the positions of the hands and foot are indicated by the small curved lines (with green indicating the limb whose subcomponent is shown, i.e., the left hand). Furthermore, the relative landscape subcomponent is limb-specific, which means that there is a separate subcomponent for each limb: the two top panels of Figure 5c are for the two arms, while the two bottom panels are for the two feet. The relative landscapes are calculated empirically from sequences chosen by climbers (not based on geometric considerations as in the case of crossing the river), conditioned on the starting pose of each limb. Figure 5d shows the same landscape, but “filtered” to show only holds that exist in the given problem. Figure 5e shows the aggregate probability surface for the climber’s left hand. This is the combination of the two probability subcomponents introduced earlier (the absolute landscape of Figure 1b and the relative landscape of Figure 1c, here limited to the left arm, i.e., the top-left panel) plus a third subcomponent: the pose. The pose subcomponent favors more common spatial configurations of the hands and feet, and discourages those that are less common. For example, low probability is assigned to moves resulting in configurations that break simple biomechanical constraints such as hands positioned farther apart than a climber’s arm-span would allow (e.g. in our example above, the pose constraint limits movement of the left hand away from the other limbs, which can be seen in the second plot in 5e). Like the relative subcomponent, the pose subcomponent is computed empirically by fitting a set of bivariate Gaussian distributions over the distributions of the Euclidean distance of limb’s pairs as observed in the solutions dataset. Figure 5f shows a naïve utility surface for this climbing problem. This is a simple gradient that originates from the last climbing hold of the problem, similar to the utility surface for the case of cross the river. Figure 5g shows that the probability surfaces (Figure 1e) and the naïve utility surface (Figure 5f) can be combined, to derive a final expected value (EV) surface for the left hand of the climber (Figure 5g). Obviously, the same can be done for the other three limbs. As in the cases of soccer and crossing the river, deriving expected value surfaces permits modeling the embodied choices of a climber who faces the same problem as shown in Figure 5a. Figure 6 illustrates a simulated climber that uses the expected value surfaces described so far to solve the problem of Figure 5a. Note that at each step during the simulated climb, we actually consider four value surfaces, one for the movements of each arm and foot - which means that at each moment in time, the climber has a choice not just between where to move a limb but also which limb to move. Figure 6a-b illustrates the resulting climbing behavior of a “greedy” planner that always selects the climbing hold whose expected value is the highest across the four surfaces (the alternation of hands and feet is an emergent effect of this strategy). This figure shows that some climbing holds (those at the top of the wall) have very low expected value at the beginning but acquire value as the climber progresses to the top of the wall. This exemplifies the concept of an evolving “landscape” of affordances as a function of the changing position and pose of the climber; and the fact that by moving the climber creates some affordances (and destroys others) (Pezzulo and Cisek, 2016). Figure 6c-d shows an empirical solution of the same problem, i.e., the actual sequence of movements selected by an experienced climber, with an expected utility landscape (generated from the same model) superimposed above. In this particular example, some of the expert’s movements were well-predicted by the greedy planning strategy, but not all. This is not surprising, given that the climbing scenario poses significant modeling challenges. Furthermore, it is worth reminding that the simulated climber of Figure 6a-b uses a naive utility surface, without look-ahead, which can potentially lead to dead ends. A better utility surface can be calculated with model-based planning, e.g., by doing rollouts from the current position up until the goal to be reached (i.e., the top hold); see our previous discussion of planning while crossing the river. Interestingly, while we considered the four value surfaces as independent so far, they become interdependent during planning; for example, by moving the left arm I can change the value surface of the other limbs. This is very interesting from a planning perspective, because it makes it possible to model the fact that in climbing, foot movements are often instrumental to create future affordances for the hands (e.g., render a distal hold reachable). Given that the goal is reaching the top with the hands, feet have ancillary roles: they are fundamental to change the hand value surfaces. As our discussion exemplified, modeling climbing is significantly more challenging than modeling the case of crossing the river, as there are many more dimensions that contribute to the notion of “affordance” in this setup; and these are difficult to treat analytically. For this, we used a data-driven methodology, analogous to what done in other sports like basketball (Cervone et al., 2016) and soccer (Fernández et al., 2019). This methodology can be readily adapted to model other domains of embodied choice that resist an analytic treatment. However, it is worth noting that the data driven methodology we used here blurs the distinction between probabilities and utilities that we assumed so far, by endowing probability surfaces with some element of utility. This is evident from the fact that the maxima of the relative landscape are above the current positions of the limbs (this is different from crossing the river, where the distance subcomponent was independent of direction). The fact that the relative probability surface is imbued with some utility results from the fact that we compute the relative landscape from empirical data, from climbers that always go to the top of the wall. Borrowing methodologies from other fields Apart for sports analytics, many other disciplines have addressed embodied decision problems and developed methodologies to deal with them. Robotics One field where embodied decisions have been object of interest is robotics. For example, when a robot has to make a decision between which objects to grasp, it has to consider situated and embodied aspects of the task, such as biomechanical constraints and the presence of obstacles. Recent work illustrates that deriving good movement policies benefits from learning internal (latent) codes that represent body-scaled “distance” metrics, as illustrated for example in Figure 7A, taken from (Srinivas et al., 2018). In this figure, lighter colors indicate larger latent distance and darker colors indicate smaller latent distance learned by the robot. Crucially, latent distance is body-scaled and sensitive to the robot arm biomechanics and to the presence of obstacles, rather than reflecting Euclidean distance. This line of research shows that it is possible to derive body-scaled perceptions akin to affordance (reachability) surfaces that we considered above for the tasks of crossing the river or climbing, using methods from deep learning and robotics. See also (Roberts et al., 2020; Zech et al., 2017) for recent reviews of other models of affordances in robotics. Reinforcement Learning The Reinforcement Learning (RL) paradigm, which typically presupposes a situated agent interacting with a changing environment, is sufficiently general to take on some of the embodied dynamics of interest. Model-free techniques, such as Q-learning, center on the learning of a value function to score state-action candidates. This value function, however, is typically global and amounts to a reflexive stimulus-action mapping that is not easily extended to the kind of forward planning required by humans in complex environments. One attempt to bridge model-free and more flexible model-based approaches in RL, which exhibits some similarity with the relative landscape used in our climbing example, is the successor representation (SR) (Dayan, 1993; Gershman, 2018; Momennejad et al., 2017). The SR uses a simple store of discounted expected occupancy mapping each origin state to a distribution over destination states. This technique avails a temporally extended local representation from which expected value can be efficiently computed without the use of Monte Carlo sampling. The SR can be combined with online Dyna-like replay to accommodate changes in environment and reward structure that pose challenges for continual learning (Momennejad et al., 2017). Other work in RL has begun to integrate the concept of affordances more explicitly. In one recent effort affordances are defined as action intents in a Markov Decision Process setting in order to reduce the space of state-action pairs evaluated (Khetarpal et al., 2020), though empirical results are so far limited to simple grid-world settings. In another example, a neural network is trained to model contextual affordances in order to predict the consequences of an action using a particular object (identified a priori) and thereby avoid arrival in a failure condition from which the goal cannot be reached (Cruz et al., 2018). Ecological Psychology In ecological psychology, Gibson’s work on affordances (Gibson, 1977) has been extended into an approach known as ecological dynamics, which applies methods from dynamical systems theory to understand behavioral interactions between organisms and their environment. Consistent with the conception of decision-making presented in this work, the ecological dynamics literature acknowledges that decisions are temporally extended and not easily separated from their behavioral expression (Beer, 2003; Nolfi, 2009; Nolfi and Floreano, 2001). Araújo et al. further suggest that because decisions are ultimately expressed as actions, an ecological analysis of human movement offers a “grounded” method to understand decision-making (Araújo et al., 2006). In one demonstration of a such an analysis, a study of human locomotion and obstacle avoidance analyzed path data as participants chose either an “inside”, or less direct “outside”, path to a goal location (Warren and Fajen, 2004). The authors modeled behavior as a differential equation where acceleration was a function of both goal distance and the angle between the goal and an intervening obstacle. Their analysis finds “bifurcation points” in the space of initial conditions (e.g. when the obstacle-goal angle exceeds 4 degrees) from which participants consistently shift to the inside path. According to this model, the selection of a path can be seen as an emergent behavior that arises from the dynamics of steering interacting with a particular environmental structure (Araújo et al., 2006). Furthermore, paths could be explained by these on-line dynamics without reference to explicit planning within an internal world model. Another insight that comes from ecological theory of development is Eleanor Gibson’s notion of prospectivity (Gibson, 1997). Prospectivity is one of four fundamental aspects of human behavior and it describes anticipatory, future-oriented behaviors such as an infant who reaches out for an object in anticipation that it will move within catching distance. This is an example of what we might call a future affordance, as the movement of the hand creates a new affordance (catching) that did not exist in the original configuration. This concept, that planning can be seen as the search for useful transformations to the available affordance landscape (Pezzulo et al., 2010; Pezzulo and Cisek, 2016; Rietveld and Kiverstein, 2014), is a central theme for the “affordance calculus” we propose. Motor control While the two domains of decision-making and motor control have long remained separated, a more recent trend sees motor control as a form of decision-making - or the problem of maximizing the utility of movement outcomes under various sources of uncertainty (Shadmehr et al., 2016; Wolpert and Landy, 2012). This analogy licenses the cross-fertilization of ideas and formal tools (e.g., statistical theory) across the two domains; and the development of novel computational models that combine objective functions used in economics, such as the maximization of expected utility, and in motor control, such as the minimization of motor costs (Ganesh and Burdet, 2013; Lepora and Pezzulo, 2015; Wispinski et al., 2018) These formulations become relevant when considering complex redundant tasks for which the same outcome can be achieved with different actions, as in a throwing task for which it is possible to hit a target (a fixed locations) with virtually infinite combinations of the release velocity of the projectile (Figure 7B). This redundancy poses the issue of selecting one among a set of actions, or sequences of actions, that guarantee the same level of performance. Interesting insights on how humans select actions in redundant tasks have been provided by computational (Cohen and Sternad, 2009; Müller and Sternad, 2004) and analytical approaches (Cusumano and Dingwell, 2013; Scholz and Schöner, 1999; Tommasino et al., 2021) based on the characterization of the action-to-performance mapping geometry. Such mappings can be seen as landscapes describing performance in a continuous space of actions, and the way in which executed actions are distributed in relation to the landscape geometry (e.g. gradient and Hessians) can reflect idiosyncratic differences in task execution strategies or learning patterns (Sternad, 2018; Tommasino et al., 2021). A similar approach could be beneficial for modeling embodied decision-making. For example, in the case of soccer, action selection could be further informed by considering the EPV (expected pass value) surface gradients, rather than just the peaks, so to take into account the risk associated with the intrinsic variability of motor execution (i.e. motor noise). Also, when the overall goal requires the execution of subtasks, while for a given subtask the solution manifold may allow for distant actions, only a part of the manifold could be compatible with concurrent or subsequent tasks. In this case, planning could be formalized as reciprocal constraints on subtasks manifolds. It has been shown empirically that in sequential actions, each action can be executed differently as a function of the preceding and the subsequent actions in the sequence. This coarticulation effect has been observed across various domains, such as speech, fingerspelling, and reaching and grasping actions (Jerde et al., 2003; Rosenbaum et al., 2012). For example, people grasp (and place their fingers on) a bottle differently depending on the next intended movement (Sartori et al., 2011). This and other similar results can be accounted for by a computational model where the solution manifolds of subsequent actions reciprocally constrain each other (Donnarumma et al., 2017). In general, the close integration between decision-making and motor control is supported by a large body of neurophysiological data showing the neural correlates of decisions throughout cortical and subcortical regions that are clearly implicated in sensorimotor control (for review, see (Cisek and Kalaska, 2010; Gold and Shadlen, 2007). For example, decisions about where to move the eyes take place in oculomotor circuits (Basso and Wurtz, 1998; Ding and Gold, 2010; Platt and Glimcher, 1999; Shadlen and Newsome, 2001), while decisions about reaching or grasping occur in arm or hand areas (Baumann et al., 2009; Cisek and Kalaska, 2005). In particular, when decisions are made about different targets for reaching, neural activity in dorsal premotor cortex reflects the changing probability (Thura and Cisek, 2014) and relative utility of the choices (Pastor-Bernier and Cisek, 2011), the competition between targets is modulated by the geometry of their placement (Pastor-Bernier and Cisek, 2011), and the very same cells continue to reflect online changes-of-mind (Pastor-Bernier et al., 2012). In short, embodied decisions appear to unfold as a continuous competition between neural correlates of potential actions (affordances) biased by all factors relevant to the choice (Cisek, 2007; Shadlen et al., 2008) and modulated by an urgency signal that helps to control the trade-off between the speed and accuracy of decisions (Cisek et al., 2009); see also (Standage et al., 2011; Thura and Cisek, 2017). Will the study of embodied decisions require the development of novel theories of decision-making and how would they differ from classical theories? In the introduction, we reviewed the fundamental differences between classical and embodied decision settings. There is not just one but several types of decisions − for example, choosing which class to take is not the same as choosing where to sit in the classroom − and these may require different mechanisms, serial or parallel, dynamic or static, etc. While the study of classical settings has advanced our knowledge in many ways, it is important to ask whether the knowledge we gathered by studying classical choices, and the models we developed to explain them, generalize to other situations, or whether they produced a distorted (or at least incomplete) view of how we solve decision tasks. The literature on value-based decision-making is currently fragmented into two main classes of models. The first (classical) class of models directly stems from economic theory and target economic-like laboratory tasks − and indeed, the study of the neural underpinnings of decision-making is often called “neuroeconomics” (Glimcher and Rustichini, 2004). In this approach, decision-making is described as the selection between a menu of prespecified offers whose values are putatively coded by the orbitofrontal cortex (Padoa-Schioppa, 2011). In this class of model, decision is therefore a centralized (prefrontal) process that only assigns action and perception systems ancillary roles. The second (action-based) class of models instead assumes that decision-making is a more distributed process within which the motor system plays an important role. When choices map directly to specific actions, as is the case in many of the choices for which our brains plausibly evolved, the selection can directly involve motor or premotor cortices − hence consisting in a competition between action affordances (Cisek, 2007). The action-based view does not assume that the prefrontal structures highlighted by the “classical” view are irrelevant; but rather that the decision emerges from a “distributed consensus” between several brain areas, each contributing different inputs to the decision (e.g., subjective values versus motor costs) (Cisek, 2012). Indeed, mapping at least some aspects of the choice into the action system is more advantageous than a centralized model for animals that have to decide and act in real time. Since it is implausible that we have developed a completely different decision architecture to deal with economic decision in the lab, it is possible that the ancestral action-based architecture is also in play when choice-action mappings are more arbitrary. We risk, perhaps, only observing parts of this system when we contrive lab-based choice settings, which often remove the situated aspects that motivated its evolution. Elaborations of action-based models, called “embodied models”, highlight that the distributed consensus architecture does not complete the decision before action initiation but rather continues deliberating afterwards, as the action unfolds in time. This is important for at least two reasons. First, as the deliberation continues, it is possible to change one’s mind or revise an initial plan along the way if novel opportunities arise. Second, action dynamics change the dynamical landscape of choice (e.g., the motor costs required to pursuing the selected plans). To the extent that these changing factors can be incorporated in the deliberative process, action deployment feeds back on the decision process, hence breaking the serial assumption of classical models (Lepora and Pezzulo, 2015). While these models seem more appropriate than classical models to deal with embodied decisions, current research has just scratched the surface of the possible ways the brain may implement decisions “in the wild” (and perhaps also in the lab). In the same way we are in need of novel methodologies to study embodied decisions, we also need a new theoretical framework to study them. It is important to acknowledge that future models of decisionmaking may radically change some of the core assumptions of current models. For example, embodied choice models are at odds with the classical serial view of decision-making. Decision models used in foraging theory (Charnov, 1976; Hayden and Moreno-Bote, 2018) are at odds with the standard definition of a decision as a competition between two (or more)offers and propose instead that the competition is between “exploit the current offer” and “explore alternatives”. While these decision models were originally developed to deal with offers that are not presented simultaneously (as is common in foraging studies), they can deal equally well with offers presented simultaneously (as common in lab studies) if one assumes that the offers are attended to serially (Hayden and Moreno-Bote, 2018). Another fundamental limitation of classical decision setups (and corresponding models) is that they keep perceptual aspects of the task extremely simplified (to avoid confounds). However, the situation is completely different in embodied settings, which often require making decisions based on perceived affordances (Gibson, 1979) and body-scaled perceptions (Proffitt, 2006). Similarly, the field of decision-making often assumes that we have ready-to-use representations of utilities (Padoa-Schioppa and Assad, 2006), but instead we showed that utility surfaces for embodied decisions need to be computed in context dependent ways − for example, to reflect future affordances − and in real time. We showed examples of how to quantify present affordances (or probabilities) and future affordances (or utilities) to study embodied decisions in ways that are commensurate to the study of classical neuroeconomic tasks. However, it remains to be assessed whether these formalizations really reflect the ways our brains process affordances or are only convenient ways to study them (Pezzulo and Cisek, 2016). In sum, the field of embodied decisions offers opportunities not just for conducting novel experiments but also for developing novel theories of decision-making and other cognitive processes. Future models of decision-making may not even answer the questions currently being pursued but instead lead to new types of questions; for example, if one sees the brain as a control system, the most important question is not “how does the brain represent knowledge about the world and subjective values”, but instead becomes “how does the brain learn to control increasingly complex interaction with the world”. In turn, the development of these theories could lead to a novel fundamental understanding of the brain as an organon that evolved for interaction, not for contemplation. An additional challenge of future theories of embodied decisions is offering a more comprehensive description of “embodied decision-makers,” which considers for example the fact that they feel emotions, remember past experiences, and monitor their ongoing performance. In our literature review, we largely omitted the important influences of arousal, motivation and emotion on decision-making processes and the roles of our neurocognitive capacities for executive functioning, attention and working memory in the solution of embodied decision tasks. While a comprehensive analysis of the above processes is beyond the scope of this article, advancing the field of embodied decision-making will require their systematic analysis and integration. In turn, a systematic study of embodied decision settings might potentially help refine our taxonomies of cognitive, emotional, motivational and executive processes, which we often take for granted, but which might not correspond to separate processes or neuronal substrates (Barrett and Finlay, 2018; Buzsaki, 2019; Cisek, 2019; Passingham and Wise, 2012). One example that we discussed multiple times already is the deep involvement of sensorimotor brain areas in decision-making processes, which is at odds with classical taxonomies (Cisek and Kalaska, 2010; Gold and Shadlen, 2007; Song and Nakayama, 2009); but there are other putative segregations (e.g., between planning and attention processes) that might be less compelling if one considers embodied choice situations. Potential impact of novel studies and theories of embodied decision-making The novel insights gathered from embodied decision experiments could have a significant impact on several fields, such as cognitive science, neuroeconomics, cognitive robotics and other disciplines interested in decision-making and sensorimotor control − as well as their deficits. As we discussed, most decision-making studies in cognitive science and neuroeconomics focus on simple choices between fixed menus. Embodied decision studies can shed light on other kinds of decisions − and possibly decision-making circuits in the brain − that are more deeply integrated with sensorimotor processes than traditionally considered. As noted in the introduction, brain evolution has for millions of years been driven by the challenges of embodied decisions during closed-loop interaction with a dynamic environment (Ashby, 1952; Cisek, 1999; Gibson, 1979; Maturana and Varela, 1980; Pezzulo and Cisek, 2016; Powers, 1973), and the neural systems meeting those challenges have been highly conserved (Cisek, 2019; Striedter and Northcutt, 2019). It is likely that even the abstract abilities of modern humans are built atop that sensorimotor architecture (Hendriks-Jansen, 1996; Pezzulo et al., 2021a; Pezzulo and Cisek, 2016; Piaget, 1952) and would be difficult to understand outside of its context (Buzsaki, 2019; Cisek, 2019). Furthermore, clarifying the neuronal and computational processes that living organisms use to make embodied choices would be extremely important for sports psychology and analytics as well as for cognitive robotics. All these fields deal natively with decisions deployed in embodied settings and have developed useful methodologies to study them (see our discussion above of analysis methodologies) that have a great potential for cross-fertilization. From a technological perspective, the realization of effective robots able to operate in unconstrained environments (e.g., homes or hospitals) critically depends on their ability to address embodied decisions, such as the choice between available affordances (including social affordances). Unlike current AI systems that are largely non-embodied, these robots will need to more deeply integrate decision and sensorimotor processes, and this is why the novel paradigms of embodied decision studies (e.g., decide-while-acting as opposed to decide-then-act) could be most beneficial. Finally, the novel insights gathered via embodied decision studies can help shed light on deficits that involve a combination of motor and decision processes. For example, it has been suggested (Mink, 1996) that the inability of Parkinsonian patients to take a step (apparently, a purely motor deficit) might be due to an inability to resolve the competition between actions (hence, a decision deficit) or to endow them with sufficient vigor or urgency. The development of novel theories of how we make embodied decisions in the first place could help us better understand how these processes can break down, as apparent in Parkinson’s or other diseases. Conclusions Every day we make countless embodied decisions, when we drive in a busy road, prepare a meal or play hide-and-seek with children. In this paper, we discussed how embodied decisions differ from classical economic decisions and which novel questions they raise that are poorly investigated. We then reviewed recent studies in fields like sports analytics and experimental psychology that start addressing these questions and distilled from them a general methodology to study embodied decisions more rigorously. We provided examples of how to cast key dimensions of embodied choices (namely, present and future affordances) into the attributes of classical economic decisions (namely, probabilities and utilities) - hence helping to align the two decision settings. Our hope is that this contribution will raise interest towards the emerging field of embodied decisions and pave the way for the realization of novel experiments, novel formal tools and novel theories that capture more appropriately its most distinctive features. Acknowledgements This research received funding from the European Union’s Horizon 2020 Framework Programme for Research and Innovation under the Specific Grant Agreement 945539 (Human Brain Project SGA3) to GP; the Office of Naval Research Global (ONRG, Award N62909-19-1-2017) to GP; and the European Research Council under the Grant Agreement No. 820213 (ThinkAhead) to GP. The GEFORCE Titan and Quadro RTX 6000s GPU cards used for this research were donated by the NVIDIA Corp. Figure 1 Examples of classical decision settings (left) versus embodied decision settings (right). Figure 2 Expected pass value (EPV) surface in soccer, from (Fernández et al., 2019), permission pending. Warm (red) colors represent higher EPV whereas cool (blue) colors represent lower EPV. Yellow circles represent the location of the player holding the ball (green circle) and his teammates. Blue circles represent the locations of the opponents. The colored arrows identify potential passes having the highest EPV on the surface (green arrow), the highest utility (red arrow) or the lower turnover expected value (cyan arrow), which (roughly) corresponds to the lowest probability of a counterattack in this setup. This visualization illustrates that a continuous decision process can be described in terms of classical concepts and analyzed with established techniques. For example, this formalism permits assessing whether soccer players chose “risky” (i.e., higher utility, lower probability) or “safe” (i.e., higher probability, lower utility) passes. Figure 3 Cross the river setup. This figure illustrates the two components of the affordance landscape for the situation shown in the photo in a), where a child must decide between three candidate stones for their next jump. In (b-f) we show a schematic view of an affordance landscape summarizing the child’s decision, with black circles representing the stones, and the white triangle at the child’s present location. The figure shows both probability surfaces and utility surfaces and how they are integrated to form an expected value surface for jumping. See the main text for explanation. Figure 4 An example of planning while crossing the river. (a) Expected value surface without planning, and below, the expected result of each available action (left, right). (b-c) Expected value surfaces updated based on 1-step planning. The cost of computing these updated expected value surface is a function of various parameters specific to both the planner (e.g. learning rate, planning depth) and problem context (e.g. number and differentiation of offers available, branching factor, etc). Here, we show two updated utility surfaces, calculated using 2 rollouts (b) or 10 rollouts (c) of the REINFORCE algorithm with the same parameterization (Sutton and Barto, 1998). Please note that these are just examples of updated utility surfaces, as the updating depends on the specific parameterization of the algorithm (e.g., its learning rate). Figure 5 The present affordance landscape of a bouldering wall (the MoonBoard). (a) A climber attempts an example problem on the MoonBoard. (b-d) Example subcomponents of the probability surface: absolute landscape, relative landscape (unfiltered) and relative landscape (filtered to show only the available holds for this problem, with a sketch of a climber superimposed to show the posture). The black circles show the positions of the climbing holds in the example problem. The curved lines show the current positions of the hands and feet of the climber in Figure 1a, with the green curved line highlighting the limb whose landscape is shown. (e) The three subcomponents that are combined to form the probability surface of the climber’s left hand, which is the next to move; these are the absolute landscape (which is the same as figure 1b), the pose subcomponent, and the relative landscape of the left hand (which is the same as the top-left panel of figure 1c). (f) This is a naïve utility component: a distance gradient starting from the final hold of this MoonBoard problem. (g) The final Expected Value for the left hand, which is obtained by combining probability subcomponents of Figure 1e and the utility subcomponent of Figure 1f. See the main text for explanation. Figure 6 Partial action sequences for a selected problem showing the complete underlying affordance landscapes (a & c), and the value of each hold (b & d). The first two plots show a greedy policy in which the maximum hold value (across all four limbs) is selected and the corresponding expected value map shown. The second two plots show a partial sequence from the true solution as annotated by an expert climber. Figure 7 Examples from other fields. (A) A learned distance metric in a robotic setting, from (Srinivas et al., 2018). Lighter (darker) colors indicate smaller (larger) latent body-scaled distances. Please note that the learned distance metric takes obstacles and arm biomechanics into account. (B) Example solution manifold in (Sternad, 2018). The upper panel depicts a two-dimensional throwing task in which the throwing position is fixed and one can modulate the velocity vector on a vertical plane. The gray shaded landscapes in the lower panel show performance (i.e. quadratic distance from the target) as a function of the vertical and horizontal velocities, while the yellow line represent the solution manifold, i.e. the set of actions with perfect performance (zero-error). The super-imposed dots represent the action distributions for two iso-performing throwers, showing how it is possible to trade-off accuracy (compact action distributions) with higher motor cost (larger velocities). Amari S Dynamics of pattern formation in lateral inhibition type neural fields Biological Cybernetics 1977 27 77 87 911931 Araújo D Davids K Hristovski R The ecological dynamics of decision making in sport Psychology of Sport and Exercise, Judgement and Decision Making in Sport and Exercise 2006 7 653 676 10.1016/j.psychsport.2006.07.002 Ashby WR Design for a brain Wiley Oxford, England 1952 Barca L Pezzulo G Tracking Second Thoughts: Continuous and Discrete Revision Processes during Visual Lexical Decision PLOS ONE 2015 10 e0116193 10.1371/journal.pone.0116193 25699992 Barca L Pezzulo G Unfolding Visual Lexical Decision in Time PLoS ONE 2012 Barrett LF Finlay BL Concepts, Goals and the Control of Survival-Related Behaviors Curr Opin Behav Sci 2018 24 172 179 10.1016/j.cobeha.2018.10.001 31157289 Basso MA Wurtz RH Modulation of Neuronal Activity in Superior Colliculus by Changes in Target Probability J Neurosci 1998 18 7519 7534 9736670 Baumann MA Fluet M-C Scherberger H Context-specific grasp movement representation in the macaque anterior intraparietal area J Neurosci 2009 29 6436 6448 10.1523/JNEUROSCI.5479-08.2009 19458215 Beer RD The Dynamics of Active Categorical Perception in an Evolved Model Agent Adaptive Behavior 2003 11 209 243 10.1177/1059712303114001 Boto E Holmes N Leggett J Roberts G Shah V Meyer SS Muñoz LD Mullinger KJ Tierney TM Bestmann S Barnes GR Moving magnetoencephalography towards real-world applications with a wearable system Nature 2018 555 657 661 10.1038/nature26147 29562238 Burk D Ingram JN Franklin DW Shadlen MN Wolpert DM Motor Effort Alters Changes of Mind in Sensorimotor Decision Making PLOS ONE 2014 9 e92681 10.1371/journal.pone.0092681 24651615 Buzsaki G The brain from inside out Oxford University Press USA 2019 Candidi M Curioni A Donnarumma F Sacheli LM Pezzulo G Interactional leader-follower sensorimotor communication strategies during repetitive joint actions Journal of The Royal Society Interface 2015 12 20150644 10.1098/rsif.2015.0644 26333815 Cervone D D’Amour A Bornn L Goldsberry K A Multiresolution Stochastic Process Model for Predicting Basketball Possession Outcomes Journal of the American Statistical Association 2016 111 585 599 10.1080/01621459.2016.1141685 Charnov EL Optimal foraging, the marginal value theorem Theoretical Population Biology 1976 9 129 136 10.1016/0040-5809(76)90040-X 1273796 Chestek CA Gilja V Nuyujukian P Kier RJ Solzbacher F Ryu SI Harrison RR Shenoy KV HermesC: low-power wireless neural recording system for freely moving primates IEEE Trans Neural Syst Rehabil Eng 2009 17 330 338 10.1109/TNSRE.2009.2023293 19497829 Cisek P Resynthesizing behavior through phylogenetic refinement Atten Percept Psychophys 2019 10.3758/s13414-019-01760-1 Cisek P Making decisions through a distributed consensus Current Opinion in Neurobiology 2012 22 927 936 10.1016/j.conb.2012.05.007 22683275 Cisek P Cortical mechanisms of action selection: the affordance competition hypothesis Phil Trans R Soc B 2007 362 1585 1599 17428779 Cisek P Integrated neural processes for defining potential actions and deciding between them: a computational model J Neurosci 2006 26 9761 9770 10.1523/JNEUROSCI.5605-05.2006 16988047 Cisek P Beyond the computer metaphor: Behavior as interaction Journal of Consciousness Studies 1999 6 125 142 Cisek P Kalaska JF Neural mechanisms for interacting with a world full of action choices Annu Rev Neurosci 2010 33 269 298 10.1146/annurev.neuro.051508.135409 20345247 Cisek P Kalaska JF Neural correlates of reaching decisions in dorsal premotor cortex: specification of multiple direction choices and final selection of action Neuron 2005 45 801 814 15748854 Cisek P Pastor-Bernier A On the challenges and mechanisms of embodied decisions Phil Trans Roy Soc B 2014 Cisek P Puskas GA El-Murr S Decisions in changing conditions: the urgency-gating model J Neurosci 2009 29 11560 11571 10.1523/JNEUR0SCI.1844-09.2009 19759303 Cohen RG Sternad D Variability in motor learning: relocating, channeling and reducing noise Exp Brain Res 2009 193 69 83 10.1007/s00221-008-1596-1 18953531 Cos I Pezzulo G Cisek P Changes of mind after movement onset depend on the state of the motor system bioRxiv 2021 2021.02.15.431196 10.1101/2021.02.15.431196 Cruz F Parisi GI Wermter S Multi-modal Feedback for Affordance-driven Interactive Reinforcement Learning 2018 International Joint Conference on Neural Networks (IJCNN) Presented at the 2018 International Joint Conference on Neural Networks (IJCNN) 2018 1 8 10.1109/IJCNN.2018.8489237 Cusumano JP Dingwell JB Movement variability near goal equivalent manifolds: Fluctuations, control, and model-based analysis Human Movement Science 2013 32 899 923 10.1016/j.humov.2013.07.019 24210574 Dayan P Improving Generalization for Temporal Difference Learning: The Successor Representation Neural Computation 1993 5 613 624 10.1162/neco.1993.5.4.613 Ding L Gold JI Caudate Encodes Multiple Computations for Perceptual Decisions J Neurosci 2010 30 15747 15759 21106814 Djebbara Z Fich LB Gramann K The brain dynamics of architectural affordances during transition Sci Rep 2021 11 2796 10.1038/s41598-021-82504-w 33531612 Djebbara Z Fich LB Petrini L Gramann K Sensorimotor brain dynamics reflect architectural affordances PNAS 2019 116 14769 14778 10.1073/pnas.1900648116 31189596 Donnarumma F Dindo H Pezzulo G Sensorimotor coarticulation in the execution and recognition of intentional actions Frontiers in psychology 2017 8 Etienne S Guthrie M Goillandeau M Nguyen TH Orignac H Gross C Boraud T Easy Rider: Monkeys Learn to Drive a Wheelchair to Navigate through a Complex Maze PLOS ONE 2014 9 e96275 10.1371/journal.pone.0096275 24831130 Fernández J Bornn L Cervone D Decomposing the Immeasurable Sport: A deep learning expected possession value framework for soccer 13th MIT Sloan Sports Analytics Conference 2019 Fodor JA Modularity of Mind: An Essay on Faculty Psychology MIT Press Cambridge, Mass 1983 Furman M Wang X-J Similarity Effect and Optimal Control of Multiple-Choice Decision Making Neuron 2008 60 1153 1168 10.1016/j.neuron.2008.12.003 19109918 Ganesh G Burdet E Motor planning explains human behaviour in tasks with multiple solutions Robotics and Autonomous Systems, Models and Technologies for Multi-modal Skill Training 2013 61 362 368 10.1016/j.robot.2012.09.024 Gershman SJ The Successor Representation: Its Computational Logic and Neural Substrates J Neurosci 2018 38 7193 7200 10.1523/JNEUROSCI.0151-18.2018 30006364 Gibson EJ An ecological psychologist’s prolegomena for perceptual development: Afunctional approach Evolving Explanations of Development: Ecological Approaches to Organism-Environment Systems American Psychological Association Washington, DC, US 1997 23 45 10.1037/10265-001 Gibson JJ The ecological approach to visual perception Lawrence Erlbaum Associates, Inc Mahwah, NJ 1979 Gibson JJ Perceiving, acting, and knowing: Toward an ecological psychology The Theory of Affordances 1977 67 82 Glimcher PW Rustichini A Neuroeconomics: the consilience of brain and decision Science 2004 306 447 452 10.1126/science.1102566 15486291 Gold JI Shadlen MN The neural basis of decision making Annu Rev Neurosci 2007 30 535 574 10.1146/annurev.neuro.29.051605.113038 17600525 Grießbach E Incagli F Herbort O Cañal-Bruland R Body dynamics of gait affect value-based decisions Sci Rep 2021 11 11894 10.1038/s41598-021-91285-1 34088941 Grossberg S Contour Enhancement, Short Term Memory, and Constancies in Reverberating Neural Networks Studies in Applied Mathematics 1973 52 213 257 10.1002/sapm1973523213 Hayden BY Moreno-Bote R A neuronal theory of sequential economic choice Brain and Neuroscience Advances 2018 2 2398212818766675 10.1177/2398212818766675 32166137 Hendriks-Jansen H Catching ourselves in the act: Situated activity, interactive emergence, evolution, and human thought MIT Press 1996 Jerde TE Soechting JF Flanders M Coarticulation in fluent fingerspelling J Neurosci 2003 23 2383 2393 12657698 Kacelnik A Vasconcelos M Monteiro T Aw J Darwin’s “tug-of-war” vs. starlings’“horse-racing”: how adaptations for sequential encounters drive simultaneous choice Behavioral Ecology and Sociobiology 2011 65 547 558 Kaufman MT Churchland MM Ryu SI Shenoy KV Cortical activity in the null space: permitting preparation without movement Nat Neurosci 2014 17 440 448 10.1038/nn.3643 24487233 Khetarpal K Ahmed Z Comanici G Abel D Precup D What can I do here? A Theory of Affordances in Reinforcement Learning International Conference on Machine Learning PMLR 2020 5243 5253 Krumin M Lee JJ Harris KD Carandini M Decision and navigation in mouse parietal cortex eLife 2018 7 e42583 10.7554/eLife.42583 30468146 Lepora NF Pezzulo G Embodied Choice: How Action Influences Perceptual Decision Making PLoS Comput Biol 2015 11 e1004110 10.1371/journal.pcbi.1004110 25849349 Marcos E Cos I Girard B Verschure PFMJ Motor Cost Influences Perceptual Decisions PLoS ONE 2015 10 e0144841 10.1371/journal.pone.0144841 26673222 Maturana HR Varela FJ Autopoiesis and Cognition: the Realization of Living D. Reidel Pub Dordrecht, Holland 1980 Michalski J Green AM Cisek P Reaching decisions during ongoing movements Journal of Neurophysiology 2020 123 1090 1102 10.1152/jn.00613.2019 32049585 Mink JW The basal ganglia: focused selection and inhibition of competing motor programs Prog Neurobiol 1996 50 381 425 9004351 Momennejad I Russek EM Cheong JH Botvinick MM Daw ND Gershman SJ The successor representation in human reinforcement learning Nature Human Behaviour 2017 1 680 692 10.1038/s41562-017-0180-8 Müller H Sternad D Decomposition of variability in the execution of goal-oriented tasks: three components of skill improvement J Exp Psychol Hum Percept Perform 2004 30 212 233 10.1037/0096-1523.30.1.212 14769078 Nolfi S Behavior and cognition as a complex adaptive system: Insights from robotic experiments Philosophy of Complex Systems, Handbook on Foundational/Philosophical Issues for Complex Systems in Science Elsevier 2009 Nolfi S Floreano D Evolutionary robotics The biology, intelligence, and technology of self-organizing machines MIT press 2001 Ortega PA Braun DA Thermodynamics as a theory of decision-making with information-processing costs Proc R Soc A 2013 469 10.1098/rspa.2012.0683 Padoa-Schioppa C Neurobiology Of Economic Choice: A Good-Based Model Annu Rev Neurosci 2011 34 333 359 10.1146/annurev-neuro-061010-113648 21456961 Padoa-Schioppa C Assad JA Neurons in the orbitofrontal cortex encode economic value Nature 2006 441 223 226 10.1038/nature04676 16633341 Passingham RE Wise SP The Neurobiology of the Prefrontal Cortex: Anatomy, Evolution, and the Origin of Insight 1 Oxford University Press USA 2012 Pastor-Bernier A Cisek P Neural Correlates of Biased Competition in Premotor Cortex J Neurosci 2011 31 7083 7088 21562270 Pastor-Bernier A Tremblay E Cisek P Dorsal premotor cortex is involved in switching motor plans Front Neuroeng 2012 5 5 10.3389/fneng.2012.00005 22493577 Pezzulo G Barca L Bocconi AL Borghi AM When Affordances Climb into your Mind: Advantages of Motor Simulation in a Memory Task Performed by Novice and Expert Rock Climbers Brain and Cognition 2010 73 68 73 20381226 Pezzulo G Cisek P Navigating the Affordance Landscape: Feedback Control as a Process Model of Behavior and Cognition Trends in Cognitive Sciences 2016 10.1016/j.tics.2016.03.013 Pezzulo G Dindo H What should I do next? Using shared representations to solve interaction problems Experimental Brain Research 2011 211 613 630 21559745 Pezzulo G Donnarumma F Dindo H Human Sensorimotor Communication: A Theory of Signaling in Online Social Interactions PLoS ONE 2013 8 e79876 10.1371/journal.pone.0079876 24278201 Pezzulo G Donnarumma F Dindo H D’Ausilio A Konvalinka I Castelfranchi C The body talks: Sensorimotor communication and its brain and kinematic signatures Physics of Life Reviews 2018 10.1016/j.plrev.2018.06.014 Pezzulo G Ognibene D Proactive Action Preparation: Seeing Action Preparation as a Continuous and Proactive Process Motor Control 2011 Pezzulo G Parr T Friston KJ The evolution of brain architectures for predictive coding and Active Inference Philosophical Transactions of the Royal Society B: Biological Sciences 2021a Pezzulo G Roche L Saint-Bauzel L Haptic communication optimises joint decisions and affords implicit confidence sharing Sci Rep 2021b 11 1051 10.1038/s41598-020-80041-6 33441715 Piaget J The Origins of Intelligence in Children International Universities Press New York 1952 Platt ML Glimcher PW Neural correlates of decision variables in parietal cortex Nature 1999 400 233 238 10.1038/22268 10421364 Powers WT Behavior: The Control of Perception Aldine Hawthorne, NY 1973 Proffitt DR Embodied Perception and the Economy of Action Perspectives on Psychological Science 2006 1 110 122 10.1111/j.1745-6916.2006.00008.x 26151466 Pylyshyn ZW Computation and Cognition: Toward a Foundation for Cognitive Science MIT Press 1984 Ratcliff R A theory of memory retrieval Psychological Review 1978 85 59 108 Ratcliff R Smith PL Brown SD McKoon G Diffusion Decision Model: Current Issues and History Trends in Cognitive Sciences 2016 20 260 281 10.1016/j.tics.2016.01.007 26952739 Resulaj A Kiani R Wolpert DM Shadlen MN Changes of mind in decision-making Nature 2009 461 263 266 10.1038/nature08275 19693010 Rietveld E Kiverstein J A Rich Landscape of Affordances Ecological Psychology 2014 26 325 352 10.1080/10407413.2014.958035 Roberts SF Koditschek DE Miracchi LJ Examples of Gibsonian Affordances in Legged Robotics Research Using an Empirical, Generative Framework Front Neurorobot 2020 14 10.3389/fnbot.2020.00012 Rodríguez F López JC Vargas JP Gómez Y Broglio C Salas C Conservation of Spatial Memory Function in the Pallial Forebrain of Reptiles and Ray-Finned Fishes J Neurosci 2002 22 2894 2903 11923454 Rosenbaum DA Chapman KM Weigelt M Weiss DJ van der Wel R Cognition, Action, And Object Manipulation Psychol Bull 2012 138 924 946 10.1037/a0027839 22448912 Saitoh K Ménard A Grillner S Tectal control of locomotion, steering, and eye movements in lamprey J Neurophysiol 2007 97 3093 3108 10.1152/jn.00639.2006 17303814 Sandamirskaya Y Zibner SK Schneegans S Schöner G Using dynamic field theory to extend the embodiment stance toward higher cognition New Ideas in Psychology 2013 31 322 339 Sartori L Straulino E Castiello U How Objects Are Grasped: The Interplay between Affordances and End-Goals PLOS ONE 2011 6 e25203 10.1371/journal.pone.0025203 21980396 Schneegans S Schöner G 13-Dynamic Field Theory as a Framework for Understanding Embodied Cognition Calvo P Gomila A Handbook of Cognitive Science, Perspectives on Cognitive Science Elsevier San Diego 2008 241 271 10.1016/B978-0-08-046616-3.00013-X Scholz JP Schöner G The uncontrolled manifold concept: identifying control variables for a functional task Exp Brain Res 1999 126 289 306 10.1007/s002210050738 10382616 Schwarz DA Lebedev MA Hanson TL Dimitrov DF Lehew G Meloy J Rajangam S Subramanian V Ifft PJ Li Z Ramakrishnan A Chronic, wireless recordings of large-scale brain activity in freely moving rhesus monkeys Nat Methods 2014 11 670 676 10.1038/nmeth.2936 24776634 Sebanz N Bekkering H Knoblich G Joint action: bodies and minds moving together Trends Cogn Sci 2006 10 70 76 10.1016/j.tics.2005.12.009 16406326 Shadlen MN Kiani R Hanks TD Churchland AK Neurobiology of Decision Making: An Intentional Framework Engel C Singer W Better than Conscious?: Decision Making, the Human Mind, and Implications for Institutions The MIT Press 2008 Shadlen MN Newsome WT Neural basis of a perceptual decision in the parietal cortex (area LIP) of the rhesus monkey J Neurophysiol 2001 86 1916 1936 11600651 Shadmehr R Ahmed A Vigor: Neuroeconomics of movement control MIT Press 2020 Shadmehr R Huang HJ Ahmed AA A Representation of Effort in Decision-Making and Motor Control Curr Biol 2016 26 1929 1934 10.1016/j.cub.2016.05.065 27374338 Sodagar AM Wise KD Najafi K A Fully Integrated Mixed-Signal Neural Processor for Implantable Multichannel Cortical Recording IEEE Transactions on Biomedical Engineering 2007 54 1075 1088 10.1109/TBME.2007.894986 17554826 Song J-H Nakayama K Hidden cognitive states revealed in choice reaching tasks Trends Cogn Sci 2009 13 360 366 10.1016/j.tics.2009.04.009 19647475 Spivey M The continuity of mind Oxford University Press USA 2007 Spivey M Grosjean M Knoblich G Continuous attraction toward phonological competitors Proceedings of the National Academy of Sciences of the USA 2005 102 10393 10398 15985550 Spivey MJ Dale R Continuous Dynamics in Real-Time Cognition Curr Dir Psychol Sci 2006 15 207 211 10.1111/j.1467-8721.2006.00437.x Srinivas A Jabri A Abbeel P Levine S Finn C Universal Planning Networks 2018 arXiv:1804.00645 [cs, stat] Standage D You H Wang D-H Dorris MC Gain modulation by an urgency signal controls the speed-accuracy trade-off in a network model of a cortical decision circuit Front Comput Neurosci 2011 5 7 10.3389/fncom.2011.00007 21415911 Sternad D It’s Not (Only) the Mean that Matters: Variability, Noise and Exploration in Skill Learning Curr Opin Behav Sci 2018 20 183 195 10.1016/j.cobeha.2018.01.004 30035207 Striedter GF Northcutt RG Brains through time: a natural history of vertebrates Oxford University Press 2019 Sutton RS Barto AG Reinforcement Learning: An Introduction MIT Press Cambridge MA 1998 Thura D Cisek P The Basal Ganglia Do Not Select Reach Targets but Control the Urgency of Commitment Neuron 2017 95 1160 1170 e5 10.1016/j.neuron.2017.07.039 28823728 Thura D Cisek P Deliberation and commitment in the premotor and primary motor cortex during dynamic decision making Neuron 2014 81 1401 1416 10.1016/j.neuron.2014.01.031 24656257 Todorov E Efficient computation of optimal actions Proc Natl Acad Sci U S A 2009 106 11478 11483 10.1073/pnas.0710743106 19574462 Tommasino P Maselli A Campolo D Lacquaniti F d’Avella A A Hessian-based decomposition characterizes how performance in complex motor skills depends on individual strategy and variability bioRxiv 2021 645317 10.1101/645317 Topalovic U Aghajan ZM Villaroman D Hiller S Christov-Moore L Wishard TJ Stangl M Hasulak NR Inman CS Fields TA Rao VR Wireless Programmable Recording and Stimulation of Deep Brain Activity in Freely Moving Humans Neuron 2020 108 322 334 e9 10.1016/j.neuron.2020.08.021 32946744 Trommershäuser J Maloney LT Landy MS Decision making, movement planning and statistical decision theory Trends Cogn Sci 2008 12 291 297 10.1016/j.tics.2008.04.010 18614390 Usher M McClelland JL On the time course of perceptual choice: The leaky, competing accumulator model Psychological Review 2001 108 3 550 592 11488378 Wang XJ Decision making in recurrent neuronal circuits Neuron 2008 60 215 34 10.1016/j.neuron.2008.09.034 18957215 Warren WH Perceiving affordances: Visual guidance of stair climbing Journal of Experimental Psychology: Human Perception and Performance 1984 10 683 703 10.1037/0096-1523.10.5.683 6238127 Warren WH Fajen BR Behavioral Dynamics of Human Locomotion Ecological Psychology 2004 16 61 66 10.1207/s15326969eco1601_8 Wispinski NJ Gallivan JP Chapman CS Models, movements, and minds: bridging the gap between decision making and action Annals of the New York Academy of Sciences 2018 0 10.1111/nyas.13973 Wolpert DM Landy MS Motor control is decision-making Curr Opin Neurobiol 2012 22 996 1003 10.1016/j.conb.2012.05.003 22647641 Yamamoto Y Yokoyama K Okumura M Kijima A Kadota K Gohara K Joint Action Syntax in Japanese Martial Arts PLoS ONE 2013 8 e72436 10.1371/journal.pone.0072436 24023740 Yoo SBM Hayden BY Pearson JM Continuous decisions Philosophical Transactions of the Royal Society B: Biological Sciences 2021 376 20190664 10.1098/rstb.2019.0664 Zech P Haller S Lakani SR Ridge B Ugur E Piater J Computational models of affordance in robotics: a taxonomy and systematic classification Adaptive Behavior 2017 25 235 271 10.1177/1059712317726357 Zénon A Solopchuk O Pezzulo G An information-theoretic perspective on the costs of cognition Neuropsychologia 2019 123 5 18 10.1016/j.neuropsychologia.2018.09.013 30268880

PMC007xxxxxx/PMC7614808.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0103374 Caries Res Caries Res Caries research 0008-6568 1421-976X 36689939 7614808 10.1159/000529160 EMS174141 Article OXIS contacts and approximal caries in preschool children- A prospective cohort study Kirthiga M a Muthu M S a Kayalvizhi G b Mathur Vijay Prakash c Jayakumar Naveenkumar d Praveen R e a Centre for Early Childhood Caries Research (CECCRe), Department of Pediatric and Preventive Dentistry, Sri Ramachandra Faculty of Dental Sciences, Sri Ramachandra Institute of Higher Education & Research (SRIHER), Chennai, Tamilnadu, India a Centre for Early Childhood Caries Research (CECCRe), Department of Pediatric and Preventive Dentistry, Sri Ramachandra Faculty of Dental Sciences, SRIHER, Chennai, Tamilnadu, India b Department of Pedodontics and Preventive Dentistry, Syamala Reddy Dental College, Bangalore, India c Division of Pedodontics and Preventive Dentistry, Centre for Dental Education and Research, All India Institute of Medical Sciences, New Delhi, India d Department of Oral and Maxillofacial Surgery, Sri Ramachandra Faculty of Dental Sciences, SRIHER, Chennai, Tamilnadu, India e Department of Conservative Dentistry and Endodontics, Indira Gandhi Institute of Dental Sciences, Puducherry, India Corresponding Author M.S. Muthu MDS., Ph.D., MFDS RCPS (Glasg), Head, Centre for Early Childhood Caries and Research, Department of Pediatric and Preventive Dentistry, Sri Ramachandra Faculty of Dental Sciences, Sri Ramachandra Institute of Higher Education and Research (SRIHER), Sri Ramachandra Nagar, Porur, Chennai, Tamil Nadu 600116 (India), Tel: +91 9444045094, muthumurugan@gmail.com 23 1 2023 23 1 2023 18 4 2023 25 7 2023 57 2 133140 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. The present prospective cohort study was conducted to evaluate the susceptibility of OXIS contact areas namely O (open type), X (point type), I (straight type) and S (curved type) in the development of approximal caries. We conducted this study among 953 school children with 3812 contacts in Puducherry, India. At baseline, the contacts were assessed in accordance with OXIS criteria. At the end of 12 months, two calibrated dentists measured dental caries following the International Caries Detection and Assessment (ICDAS) criteria. Information about feeding practices, diet, and oral hygiene was collected by means of a structured questionnaire from each child’s parent. Data were analyzed by unadjusted and adjusted Poisson regression analysis with a multilevel approach. The two levels of analysis were tooth and child. Of 3,812 contacts observed during the follow-up, 127 (3.3%) were observed as carious. Poisson regression analysis revealed a significant association between type of contact and caries prevalence (p < 0.05). The risk ratios for the development of approximal caries in X contacts were 2.4 (0.3-17.2), p value 0.38; in I contacts - 4.9 (1.2-19.9), p value 0.027; and in S contacts 8.2 (1.9-34.2), p value 0.004, when compared with the O contacts. Among the child variables, male gender (RR=2.1; 95%CI-1.3,3.5), parental supervision while toothbrushing (RR=1.6; 95%CI-1.1,2.4) and the use of toothpaste (RR=1.9; 95%CI-1.3,3.1) were found to be associated with approximal caries after adjustment for the other variables. Among the OXIS contacts, the S type was most susceptible to approximal caries due to its complex morphology, followed by I, X, and O. Approximal Caries OXIS Contact areas Pre school children Primary molars pmcIntroduction In the primary dentition, the occlusal lesions are more common than the approximal lesions. However, as soon as the proximal contacts form, the prevalence of the approximal lesions increases which occurs at three years. This pattern continues in the mixed and permanent dentitions. Therefore understanding the variations of contacts is crucial in understanding the progression of the disease [Kennedy., 1986]. Proximal caries detection in primary teeth is of great importance because of the rapid rate of caries progression and the difficulty in determining the presence or absence of a lesion. This could be explained by some of the characteristics of primary teeth, such as the thinner enamel and dentin layers, lower degree of mineralization, wider dentinal tubules in comparison with permanent teeth, and broad area of contact allowing greater biofilm accumulation [Virajsilp V., 2005]. The OXIS classification of interproximal contact areas of primary molars was first described in 2018. It consists of four types of contact areas, namely, the open type, denoted as O; the point contact type, denoted as X; the straight contact type, denoted as I; and the curved contact type, denoted as S [Kirthiga et al., 2018]. Subsequently, studies have been performed to evaluate the prevalence of OXIS contacts in various geographic locations, namely, Puducherry [Muthu et al., 2020], Seoul (South Korea) [Kirthiga et al., 2021], and Ajman (UAE) [Walia et al., 2021]. Previous studies in this area evaluated the association of two types of contacts (open and closed) with approximal caries and concluded that the risk for approximal caries in the posterior primary dentition is increased if contact points are closed rather than open [Allison and Schwartz, 2003; Warren et al., 2003; Subramaniam et al., 2012]. A prospective study concluded that the risk for approximal caries development was higher when the approximal surfaces were concave when compared with other combinations, namely, concave-concave, concave-convex, convex-concave, and convex-convex [Cortes et al., 2018]. A recent investigation was performed to assess the influence of OXIS contact areas on approximal caries in a retrospective cohort study design. This study used existing Cone Beam Computed Tomography (CBCT images) and clinical photographs and concluded that a significant association exists between OXIS contacts and approximal caries. Another significant conclusion from this study was that the S contact had the maximum risk for approximal caries development in primary molars, followed by I, X, and O contacts [Muthu et al., 2021]. Although evidence exists regarding OXIS contacts as a risk factor for approximal caries, the role of the confounding factors in the prevalence of approximal caries has not been measured. A well-designed prospective cohort study is necessary to understand the association of OXIS contacts with approximal caries. The aim of the present study was to evaluate, prospectively, the susceptibility of OXIS contact areas in the development of approximal caries, in a group of pre-schoolers aged 3-4 years. The hypothesis tested in the present study was that the broad contact areas (S & I) would be more susceptible to approximal caries development in children when compared with the narrow or open contact areas (X & O). Materials and Methods Ethical Considerations and Permissions The study protocol was reviewed and approved by the Institutional Review Board (IECNI/16/AUG/55/54). Prior to commencement of the study, permissions were obtained from the Chief Educational Officer, Puducherry, and Principals of the respective schools. After the purpose of the study was explained, a written informed consent was obtained from parents of the included children, allowing for their participation and examination for the presence of dental caries. Study Design and Participants This study was the longitudinal portion of a previous published study [Muthu et al., 2020] performed in preschool children aged 3-5 years who had an ancestral nativity to Puducherry, a Union Territory in India. A two-stage simple random sampling methodology was used to select schools and children, respectively. The total number of schools included for the present study was 34. Finally, a cohort of 1,119 children (with 4,476 contacts) aged 3–4 years was recruited at baseline. The calculation of the sample is based on a previously conducted study [Allison and Schwartz, 2003] to estimate the prevalence of open contacts of primary molars. Thus, the sample size was calculated with an expected prevalence of 30% percent and a z-value of 1.96. A minimum sample size of 933 children was determined. Furthermore, the sample size was increased to 1119 to compensate for 20% of additional losses. The baseline examinations were performed between November 2018 and January 2019, and the 12-month follow-up examinations were performed between November 2019 and January 2020. The selection criteria and the sample selection were described in a previously published study [Muthu et al., 2020]. Assessment of Baseline Data (November 2018 - January 2019) A single paediatric dentist (KM) was extensively trained and calibrated under the supervision of an expert to clinically evaluate the contact areas over two months duration. The calibration process consisted of a 10 hour session conduced in two stages. The first stage included power point slide presentations with clinical photographs of the OXIS classification of contact areas to be observed clinically. In the second stage, a clinical exercise using 100 study models of 25 children was performed to provide a learning environment of previously acquired theoretical information. For the calculation of intra examiner reproducibility, 25 caries free children (with 100 contacts) were examined and reexamined clinically after a period of two weeks. The intra-examiner Cohen’s Kappa coefficient was 0.96. Clinical examinations were performed among 4,476 contacts of the 1,119 caries-free children (ICDAS=0) for the type of contact between the distal surface of the first primary molar and the mesial surface of the second primary molar according to OXIS criteria [Kirthiga et al., 2018]. Sectional maxillary and mandibular impressions were made on the day of clinical examination, for future record purposes. All the recruited children were provided with oral hygiene instructions regarding frequency and method of brushing. Calibration of the Examiners Prior to the commencement of the study, two examiners (BR & SP), pediatric dentists, were trained to diagnose dental caries clinically and radiographically using ICDAS. The method of training was carried out according to the recommendations for examiner training of the 2014 ICDAS criteria manual. For the calculation of inter-examiner reproducibility, 25 children were analyzed by both the examiners. The inter examiner kappa coefficients was 0.94. The intra examiner kappa coefficients of the examiners (BR, SP) was observed to be 0.92 and 0.89 respectively. Assessment of the Outcome by Clinical and Radiographic Examination at 12 Months (November 2019 - January 2020) The dependent outcome was the clinical or radiographic presence of caries in the approximal surfaces (the distal surface of the first primary molar and/or the mesial surface of the second primary molar) at the 12-month follow-up examination. The outcome assessment was performed independently by two trained and calibrated examiners who were blinded to the baseline data. Clinical Examinations Clinical examinations at the 12-month follow-up were performed in a suitable classroom by means of a mouth mirror, under natural light (Type III examination)[Peter., 2017]. Cotton rolls were used to clean the teeth of food debris and to dry them. The selected children were examined for dental caries according to the International Caries Detection and Assessment System (ICDAS II) criteria [Ismail et al., 2007] by examiners who were blinded to the baseline data. Teeth were initially assessed wet and then air-dried by means of a portable device providing compressed air. The examiners assessed the buccal, lingual, mesial, distal, and occlusal surfaces of each tooth and recorded the findings on a custom-made assessment form. However, only the presence of caries in the approximal surfaces (the distal surface of the first primary molar or the medial surface of the second primary molar) was considered for the assessment. A CPITN probe was used for the assessment of enamel breakdown, if required. Radiographic Examination In addition to clinical examination, bitewing radiographs were taken when visual inspection of approximal surfaces was impossible (in accordance with the AAPD guidelines) or when the ICDAS score was 4 (underlying dark shadow from dentin). The radiographs were taken by means of a portable dental X-ray unit (Vatech EZRay Air Plus Portable X-Ray Machine, Vatech, New Delhi, India) and a digital scanner (SOREDEX™ DIGORA™ Optime, Brea, CA, USA) which were brought to the schools where the examinations took place. In addition, a ring holder was used to ensure standardization. All radiographs were read in a view box by both investigators on different occasions. The mesial surfaces of the second primary molars and the distal surfaces of the first primary molars were examined for dental caries. According to radiographic appearance, a code, as per ICDAS scoring, was assigned to the designated surface as follows: 0, No radiolucency; RA 1, Lesion in the external half of the enamel; RA 2, Lesion in the internal half of the enamel; RA 3, Lesion in the external third of the dentin; RB 4, Lesion in the middle third of the dentin; RC 5, Lesion in the internal third of the dentin, clinically cavitated; and RC 6, Radiolucency into the pulp, clinically cavitated. Missing (m) or filled (f) surfaces, if any, were coded separately. Although the ICDAS scores were used for the clinical and radiographic assessment of dental caries, the final outcome was determined as the absence or presence of approximal caries lesions (cavitated or non-cavitated). Data were recorded on a custom-made data sheet. Questionnaire Data and Diet Chart Assessment To ascertain the effects of independent variables related to feeding, oral hygiene, diet, and caries prevention, each child was given a validated questionnaire [Folayan MO et al., 2015] to take home on the day of the clinical examination. The parents’ socio-economic status was categorized by the modified Kuppuswamy scale as upper, upper middle, lower middle, upper lower, or lower class. The questionnaire was given to the classroom teachers, who communicated with the parents regarding its completion and return by the child within 24-48 hours. The parents were advised to contact the teachers (who were sensitized by the primary investigator) in case of any queries/doubts which may arise during questionnaire completion. The completed questionnaires were collected from the teachers of the recruited children after the stipulated time period. Due to the peak COVID situation and the travel restrictions during this period, the collection of questionnaires from all schools was not possible. Hence, in these situations, questionnaire data collection was performed via telephone interview by a trained investigator (KM). Statistical Analysis Demographic data were analyzed with descriptive statistics expressed as frequencies and percentages. The number of approximal caries was considered as count data. Data were entered in Excel and analyzed by STATA 16 software. The data had a hierarchy level which was tooth (level 1) and child (level 2). The data were analyzed by unadjusted and adjusted Poisson regression analysis with a multilevel approach (teeth [level 1, on which outcome was measured] and child [level 2]), since both dental variables and child characteristics could exert an influence on the outcome. The incidence of approximal caries in primary teeth was calculated as relative risk (RR) and respective 95% confidence intervals (CI). A p value of less than 0.05 was considered statistically significant. The stepwise method was used to select variables for the final model, maintaining only those dental variables that remained significant (p < 0.05) after the adjustments. For the child variables, those that remained significant (p < 0.05) after being controlled for the other variables were maintained in the final model. The goodness-of-fit of the model was analyzed using deviance (–2 log likelihood). Results Characteristics of the Participants Of the 1,119 children (with 4,476 contacts) who were recruited at baseline, 166 children (with 664 contacts) were lost to follow-up at 1 year, corresponding to a response rate of 85.2%. The drop-outs were primarily because they migrated from their schools or were not present on the day of the follow-up examination or did not complete the questionnaire. Finally, 953 (85.2%) subjects with 3,812 contacts remained in the study after completion of both clinical examinations and questionnaires. There were 759 questionnaires collected from schools. For the remaining 194 parents, questionnaire data collection was performed by means of a telephone interview. The gender distribution of the cohort present at one-year follow-up consisted of 478 (50.2%) female and 475 (49.8%) male children. At the final examination, the children were between 4 and 5 years of age (501 were 4 years old, 452 were 5 years old). Evaluation of the Questionnaires Supplementary file 1 denotes all the characteristics of the included participants based on the demographics and information collected from the questionnaire. Of the included children, 67.8% brushed once daily, 57.6% of the children had never visited a dentist before, and only 4% had received professional topical fluoride application. According to their parents, 78.2% of the children ate snacks at a frequency of 1-2 times per day. With respect to the knowledge-related questions on caries prevention, 52.7% of the parents/caregivers strongly disagreed that toothpaste containing fluoride has a role in caries prevention. With respect to frequency of snacking as a risk factor of dental caries, 61% of the parents agreed. According to the parents, 65.5% agreed that mouthrinsing after snacking is beneficial in the prevention of dental caries. Prevalence and Frequency of Contacts at Baseline Of the 4,476 contacts, the observed prevalence of O, X, I, and S types was 261 (5.8%), 148 (3.3%), 3,381 (75.5%), and 686 (15.3%), respectively. The most common contact observed was I, followed by S, O, and X. Assessment of the Outcome Among the 3,812 contacts, 127 contacts (3.3%) were found to be carious at the approximal surfaces (the distal surface of the first primary molar and/or the mesial surface of the second primary molar) during the follow-up visit. The remaining 3,685 caries lesions were sound. There were no restored or missing teeth identified. Among the 127 caries-affected contacts, 102 were cavitated lesions and 25 were non-cavitated lesions. Of the caries-affected contacts, 25 contacts on both approximal surfaces were caries-affected (the mesial surface of the primary second molar and the distal surface of the primary first molar). In the other 77 caries-affected contacts, 26 were mesial (primary second molar) and 51 were distal surfaces (primary first molar) showing the presence of approximal caries. Seventy-six lesions required a bitewing radiograph for confirmation of approximal caries. Among them, 32 lesions were found to be cavitated and 44 were sound. All the parents of the caries-affected children were contacted and informed regarding the child’s dental health status. They were offered free dental treatment at a private dental center. Of the 121 parents, 83 gave consent for the treatment. Therefore, 64 restorations with glass-ionomer cement and 19 stainless steel crowns were provided for the children with caries-affected contacts. Univariate and Multivariate Poisson Regression Analysis Supplementary file 2 displays the results of the univariate analysis by Poisson regression with a multilevel approach for the incidence of approximal caries and the independent variables. Among the tooth variables, the type of arch and type of contact were significantly associated with the incidence of dental caries. Among the contact types the S contact has an increased risk of development of approximal caries (RR-7.3;CI,1.9-32.0). With respect to type of arch the left upper (RR-0.5), left lower (RR-0.4) and right upper (RR-0.3) quadrants showed decreased risk of approximal caries development when compared to right lower quadrant respectively. The incidence of approximal caries among S type and I type contacts was 31.5 % and 65.4%, respectively. Considering the child variables, male gender (RR-3.4), upper middle (RR-2.8) and lower middle category (RR-4.5) of socio-economic status were statistically significant. Further, toothbrushing by parent alone (RR-1.8), the absence of use of toothpaste (RR-1.8), and the frequency of toothbrushing once (RR-2.5) and twice daily (RR-2.5) were also found to be statistically significant. Table 1 displays the multilevel adjusted Poisson regression models with a multilevel approach after adjustment for the cluster effect at tooth and child levels. Model 1 represents the naive model, model 2 represents multilevel Poisson regression with dental variables (level 1), and model 3 represents multilevel Poisson regression with dental (level 1) and child variables (level 2). Model 2 indicates that the ”S” type of contact showed a more than 7.7 times incidence of approximal caries as compared with the “O” type of contact (RR=7.7, 95% CI=1.9-32.0; p value - 0.005). Further, the contacts in the maxillary right quadrant showed a more than 3.5 times incidence of approximal caries as compared with contacts in the right lower quadrant (p value - < 0.001). Model 3 suggests that the type of contact and type of quadrant had a significant association with approximal caries. The risk ratio for the development of approximal caries was significant for S contact (RR=8.2;CI:1.9-34.2) and I contact (RR=4.9;CI:1.2-19.9) when compared to O contact. Further, right upper quadrant (RR=3.3;CI:1.9-5.7) was observed to have a significant association with approximal caries when compared to the right lower quadrant. Among the child variables, male gender (RR-2.1), toothbrushing by sometimes child, sometimes parent (RR-1.6), absence of use of a toothpaste (RR-1.9) and upper lower category of socioeconomic status (RR-2.2) were found to have an increased risk with approximal caries after adjustment for the other variables. The schematic representation of OXIS types of contacts at baseline and after 12 months, respectively, is shown in Figure 1. Discussion The aim of the present prospective longitudinal study was to determine whether OXIS contacts are a risk factor for the development of approximal caries in the primary teeth. The focus was on the types of contacts categorized as OXIS, and there was a possibility that more than one type of contact could exist in the same child. The research question of interest was in knowing whether a tooth with a broad contact (I or S) was at greater risk of developing dental caries compared with a tooth with an O or X type of contact, while also considering variables related to the child. Therefore, a multilevel analysis was adopted The findings of the present study confirm the hypothesis that the presence of a broad contact (I or S) poses a greater risk towards approximal caries development when compared to a narrow or open contact. To the best of our knowledge, this is the first longitudinal study to investigate the association between individual OXIS contacts and approximal caries with the tooth and child as the units of analysis in a multilevel approach. At the end of 12 months, 14.8% of the entire sample had dropped out, which was well within the acceptable range. Of the total number of carious contacts, the contributions of I and S contacts were 65.4% and 31.5%, respectively, which equals 96.9% [Muthu et al., 2021]. This result was in correlation with that of another study where the contributions of I and S contacts were 68% and 30.9%, respectively, which equals 98.9%. From the above findings, it is clear that the maximum number of ap caries lesions in the primary molars occur where a broad contact area (I or S) is present. Of the 127 carious contacts, 25 contacts were carious on both ap surfaces (mesial surfaces of primary second molars and distal surfaces of primary first molars). In the other 77 carious contacts, there were 51 distal surfaces (primary first molars) and 26 mesial surfaces (primary second molars) showing the presence of approximal caries. This result was in agreement with those of other studies where the distal surfaces of primary first molars were most commonly affected compared with other surfaces [Cortes et al., 2018; Fan et al., 2019]. A possible explanation for this finding could be the early eruption of primary first molar compared to the second molar. Further the distal contact of primary first molar is established before the mesial contact of the primary second molar. The final model (model 3) revealed that type of contact, type of quadrant, gender gender, parental supervision during toothbrushing and absence of use of toothpaste were found to be associated with the development of approximal caries. Among the type of contacts, the S type of contacts (RR = 8.2; 95% CI, 1.9-34.2) had an eightfold greater risk and I contact had a nearly fivefold greater risk of exhibiting approximal caries when compared to teeth with O type of contacts. Therefore, the S type (RR = 8.2; 95% CI, 1.9-34.2) was most susceptible to approximal caries among the OXIS contacts, due to its complex concave-convex morphology. This was followed by the I contact (RR = 4.9; 95% CI, 1.2–19.9) when compared with the O type of contact. This result was in agreement with that of a previous retrospective cohort study conducted in this area. Also, the broad contact areas (I and S) were observed to be more susceptible to approximal caries than the narrow contact areas (O and X) [Muthu et al., 2021]. This result was in agreement with that of another study where the morphology of the approximal surfaces in the primary molar teeth, if both surfaces were concave, significantly influenced the risk of the individual developing approximal caries. The concave-concave surface can be considered synonymous with the S type among OXIS contacts [Cortes et al., 2018]. A logical explanation would be that this type of contact would lead to maximum plaque retention between the primary molars, since maintenance of oral hygiene by routine mechanical cleansing methods would be difficult in these areas, given their complex ‘concave followed by convex’ design. Another variable that was found to be significant in the adjusted model was the type of quadrant. The right upper quadrant was found to be most prone to approximal caries. Previous studies have observed that the right side is more prone to develop dental caries, since it is the dominant side of mastication when compared with the left side of the oral cavity [Demirci M., 2020]. In the present study, gender (male), parental supervision during toothbrushing (sometimes child, sometimes parent), and absence of the use of toothpaste were found to be associated with approximal caries. The girls exhibited fewer teeth with dental caries when compared with boys. This finding suggests that girls may have followed oral hygiene measures to a greater degree than did boys. Although numerous variables with respect to the child relative to feeding habits, oral hygiene, and diet were assessed, none of the other factors was found to have an association with approximal caries. Being a prospective cohort study including children who were 3 and 4 years old at baseline, the findings provide concrete evidence on whether OXIS contacts are a risk factor for approximal caries. Further, the use of the ICDAS system for diagnosis of non-cavitated lesions and the bitewing radiographs for hidden lesions could have prevented the underestimation of dental caries. One limitation of the study was that clinical examination of the children under natural light and in the upright position may have made it difficult to inspect and diagnose dental caries in the posterior teeth. Further, the COVID situation during the data collection period made it impossible to complete the questionnaire collection from the schools. The association of OXIS contacts with approximal caries by means of a prospective cohort design and a standardized methodology should be studied in different ethnic populations. This could help us understand whether OXIS contacts are a potential risk factor for approximal caries after consideration of confounding variables. Conclusion The present study confirms that the variations in OXIS contacts is a potential risk factor for approximal caries. Among the four contact types, S contact type was the most susceptible followed by I type. Supplementary Material Supp Table 1 Supp Table 2 Acknowledgements The authors thank Dr. Sujitha Ponraj, BDS and Dr. Bhavyaa R., BDS, Post Graduate students, Department of Pediatrics and Preventive Dentistry, Faculty of Dental Sciences, Sri Ramachandra Institute of Higher Education and Research, Chennai, Tamilnadu, India, for their assistance in the process of screening the schoolchildren for dental caries. The authors also thank Mrs. Jothi, MSc Statistics, SCARF Research Foundation, Chennai for her assistance with statistical analysis. Funding sources This study was funded by the Wellcome Trust DBT India Alliance (IA/CPHE/17/1/503352). Data Availability Statement All data generated or analysed during this study are included in this article or available as supplementary files. Further enquiries may be directed to the corresponding author. Figure 1 Schematic representation of OXIS types of contacts at baseline and after 12 months, respectively. (a,b) O types of contacts. (c,d) X types of contacts. (e,f) I types of contacts. (g,h) S types of contacts Table 1 Multilevel adjusted Poisson regression analysis of incidence of proximal caries on primary teeth considering child and tooth variables Variable Model1: Naive model Model 2 RR (95%CI) p value  Model 3 RR (95%CI) P value Intercept 0.3(0.2,0.4) 0.004 (0.001,0.018) 0.0006 (0.0001,0.0034) Type of contact O 1 1 I 3.8(0.9,15.5) 0.062 4.9(1.2,19.9) 0.027 S 7.7(1.9,32.0) 0.005 8.2(1.9,34.2) 0.004 X 1.9(0.3,13.7) 0.513 2.4(0.3,17.2) 0.38 Type of quadrant Right lower 1 1 Right upper 3.5(1.9,6.1) <0.001 3.3(1.9,5.7) <0.001 Left upper 1.8(0.9,3.4) 0.05 1.7(0.9,3.2) 0.078 Left lower 1.4(0.8,2.7) 0.269 1.4(0.7,2.6) 0.336 Gender Female 1 1 Male 2.1(1.3,3.5) 0.003 Parental supervision while tooth toothbrushing Child alone 1 1 Parent alone 1.0(0.5,1.9) 0.989 Sometimes child, sometimes parent 1.6(1.1,2.4) 0.029 Use of toothpaste Yes 1 1 No 1.9(1.3,3.1) 0.002 Don’t know 1.1(0.4,3.0) 0.907 Frequency of tooth brushing Every meal 1 Twice daily 1.3(0.5,2.9) 0.598 Once 1.5(0.8,3.2) 0.234 Occasionally 2.3(0.7,7.2) 0.15 Never 2.3(0.5,11.3) 0.321 Don’t know 1.18e-08(0,) 0.999 SES Upper middle 1 1 Lower middle 2.0(0.9,4.5) 0.1 Upper lower 2.2(0.9,4.9) 0.058 log likelihood(ll) -558.919 -533.2311 -500.9509 Deviance (-2 log likelihood) 1117.837 1066.4622 1001.9018 CI: Confidence Interval; RR: Relative Risk Statement of Ethics This study protocol was reviewed and approved by the Institutional Ethics Committee, Sri Ramachandra University of Higher Education & Research, IEC-NI/16/AUG/55/54. A written informed consent was obtained from parents of the included children for participation and examination of the child for the presence of dental caries. Conflict of Interest Statement The authors have no conflicts of interest to declare. Author Contributions M Kirthiga contributed to conception, design, data acquisition, analysis and interpretation, and drafted and critically revised the manuscript. M S Muthu contributed to conception, design, data acquisition, analysis and interpretation, and drafted and critically revised the manuscript. G Kayalvizhi contributed to conception, design, analysis and interpretation, and critically revised the manuscript. Vijay Prakash Mathur contributed to conception, design, analysis and interpretation, and critically revised the manuscript. Naveenkumar Jayakumar contributed to conception, design, analysis and interpretation, and critically revised the manuscript. R Praveen contributed to data acquisition, analysis and interpretation, and critically revised the manuscript. All authors gave their final approval and agree to be accountable for all aspects of the work. 1 Allison PJ Schwartz S Interproximal contact points and proximal caries in posterior primary teeth Pediatr Dent 2003 25 5 334 40 13678098 2 Cortes A Martignon S Qvist V Ekstrand KR Approximal morphology as predictor of approximal caries in primary molar teeth Clin Oral Invest 2018 22 2 951 9 3 Demirci M Tuncer S Yuceokur AA Prevalence of caries on individual tooth surfaces and its distribution by age and gender in university clinic patients Eur J Dent 2010 Jul 4 3 270 9 20613915 4 Fan CC Wang WH Xu T Zheng SG Risk factors of early childhood caries (ECC) among children in Beijing - a prospective cohort study BMC Oral Health 2019 19 1 34 30777062 5 Folayan MO Kolawole KA Oziegbe EO Oyedele T Oshomoji OV Chukwumah NM Prevalence, and early childhood caries risk indicators in preschool children in suburban Nigeria BMC Oral Health 2015 15 72 26123713 6 Ismail AI Sohn W Tellez M Amaya A Sen A Hasson H The International Caries Detection and Assessment System (ICDAS): an integrated system for measuring dental caries Community Dent Oral Epidemiol 2007 35 3 170 8 17518963 7 Kennedy DB Anatomy of primary and permanent teeth, Pediatric Operative Dentistry 1986 Kennedy DB 3rd edn Bristol IOP Publishing Limited 37 8 Kirthiga M Muthu MS Kayalvizhi G Krithika C Proposed classification for inter proximal contacts of primary molars using CBCT: a pilot study Wellcome Open Res 2018 3 98 30345384 9 Kirthiga M Muthu MS Lee JJC Kayalvizhi G Mathur VP Song JS Prevalence and correlation of OXIS contacts using Cone Beam Tomography images (CBCT) and photographs Int J Paediatr Dent 2021 31 4 520 7 32621346 10 Muthu MS Kirthiga M Kayalvizhi G Mathur VP OXIS classification of inter proximal contacts of primary molars and its prevalence in 3-4 year old children Pediatr Dent 2020 42 197 202 32522322 11 Muthu MS Kirthiga M Lee JC Kayalvizhi G Mathur VP Kandaswamy D OXIS contactsas a risk factor for approximal caries: a retrospective cohort study Pediatr Dent 2021 43 4 296 300 34467847 12 Peter S Essentials of Public Health Dentistry 2017 6th edn New Delhi Arya Medi 13 Subramaniam P Babu GKl Nagarathna J Interdental spacing and dental caries in the primary dentition of 4-6 year old children J Dent (Tehran) 2012 9 3 207 14 23119129 14 Virajsilp V Thearmontree A Aryatawong S Paiboonwarachat D Comparison of proximal caries detection in primary teeth between laser fluorescence and bitewing radiography Pediatr Dent 2005 27 6 493 9 16532891 15 Walia T Kirthiga M Brigi C Muthu MS Odeh R Pakash Mathur V Interproximal contact areas of primary molars based on OXIS classification - a two centre cross sectional study Wellcome Open Res 2021 5 285 33537460 16 Warren JJ Slayton RL Yonezu T Kanellis MJ Levy SM Interdental spacing and caries in the primary dentition Pediatr Dent 2003 25 2 109 13 12723834

PMC007xxxxxx/PMC7614809.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0376333 J Psychosom Res J Psychosom Res Journal of psychosomatic research 0022-3999 1879-1360 32911441 7614809 10.1016/j.jpsychores.2020.110218 EMS181444 Article Different independent associations of depression and anxiety with survival in patients with cancer Walker Jane PhD 1 Magill Nicholas PhD 2 Mulick Amy MSc 3 Symeonides Stefan PhD 4 Gourley Charlie PhD 4 Toynbee Mark MBBS 1 van Niekerk Maike BScN 1 Burke Katy MBBS 1 Quartagno Matteo 3 Frost Chris MA 2 Sharpe Michael MD 1 1 Psychological Medicine Research, University of Oxford Department of Psychiatry, Warneford Hospital, Oxford, UK 2 Department of Medical Statistics, London School of Hygiene and Tropical Medicine, London, UK 3 Department of Non-Communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, UK 4 Cancer Research UK Edinburgh Centre, University of Edinburgh, Edinburgh, UK Correspondence: Jane Walker, Psychological Medicine Research, University of Oxford Department of Psychiatry, Warneford Hospital, Oxford, OX3 7JX, UK, jane.walker@psych.ox.ac.uk / t: +44 (0)1865 618229 01 11 2020 18 8 2020 19 7 2023 26 7 2023 138 110218110218 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Objective Depression and anxiety have both been reported to predict the worse subsequent survival of people with cancer. However, depression and anxiety are mutually associated and we lack understanding of their independent associations with survival. We therefore aimed to investigate these in a large sample of patients with common cancers. Methods We analysed data on 19,966 patients with common cancers (breast, colorectal, gynaecological, lung and prostate) who had attended specialist NHS outpatient clinics in Scotland, UK. Hospital Anxiety and Depression Scale (HADS) data were linked with demographic, cancer and mortality data. We estimated the independent associations of depression (HADS depression score) and anxiety (HADS anxiety score) with survival by fitting (separately for each cancer) Cox proportional hazards models which incorporated cubic splines to allow for non-linear associations. We also adjusted for potential confounders. Results The median time from HADS completion to death or censoring was 1.9 years. Greater depression was found to be strongly associated with worse survival from all cancers. When adjusted for anxiety, this association remained in males and increased in females. Greater anxiety was also associated with worse survival in nearly all cancers. However, when adjusted for depression, the association of anxiety with worse survival was lost. In females the association reversed direction so that greater anxiety was associated with better survival. Conclusion Although often considered together as aspects of ‘emotional distress’, depression and anxiety have different independent associations with survival in patients with cancer and should therefore be considered separately. depression anxiety cancer survival mortality neoplasms pmcIntroduction There is considerable interest in the relationship between psychological factors and the survival of people with cancer. A growing body of literature suggests that depression and anxiety, in particular, are associated with worse subsequent survival [1–3]. A hitherto neglected aspect of this literature is whether depression and anxiety, which commonly co-occur, have similar associations with survival when they are considered separately. This question arises from an increasing understanding that depression and anxiety are not just aspects of ‘emotional distress,’ but have distinct psychological and biological mechanisms [4]. We are unaware of any studies published to date that have examined this question. We therefore sought to answer it by conducting an analysis of prospectively collected data from a large cohort of patients with common cancers (breast, colorectal, gynaecological, lung and prostate cancers) who had completed depression and anxiety questionnaires as part of their routine cancer care and for whom we had survival data. The aims of our analysis were to examine the independent associations of depression and anxiety with subsequent survival in patients with common cancers by determining: (a) the association of depression with subsequent survival in patients with each cancer, with and without adjustment for anxiety and (b) the association of anxiety with subsequent survival in patients with each cancer, with and without adjustment for depression. Methods Study design and sample We analysed data from patients who had attended outpatient clinics of the Edinburgh, Glasgow and Dundee National Health Service (NHS) cancer centres in Scotland, UK. Each of these cancer centres provides a full range of diagnostic and treatment services in a large urban teaching hospital with outreach clinics in the smaller hospitals of surrounding towns. Together the three centres serve a geographically defined area of approximately four million people and provide specialist care for the vast majority of patients who have been diagnosed with cancer in this region. Patients attending these clinics were asked to complete a depression and anxiety questionnaire as part of their routine cancer care. Most patients (80%) completed this questionnaire (the main reason that patients did not complete the questionnaire was that their oncology appointment had begun before they could do so). We included a patient’s data in this analysis if: (a) they had attended an outpatient oncology consultation in a central or outreach cancer clinic between May 12, 2008 and Aug 24, 2011; (b) they had completed the depression and anxiety questionnaire (the Hospital Anxiety and Depression Scale, HADS) that was used routinely in the cancer clinics [5]; (c) the patient had no missing items on the HADS; (d) we could obtain their matched demographic and clinical data from the Scottish National Cancer Registry; (e) they had given consent for their relevant clinical data to be used for research; and (f) they had a primary breast, colorectal, gynaecological, lung or prostate cancer. We chose these cancers because they are the most common, they often form the basis for multidisciplinary cancer care (therefore the associations between depression and anxiety and survival in each group is clinically useful) and the number of patients within each grouping was sufficient to estimate these associations with acceptable accuracy. Measures Depression and anxiety The HADS was routinely given to everyone who attended the cancer clinics in order to assess how much depression and anxiety they had experienced over the preceding week [5]. The HADS has a total of 14 items; seven items make up the HADS depression subscale and seven make up the HADS anxiety subscale. The individual items are each scored from zero to three, resulting in maximum depression and anxiety subscale scores of 21, with higher scores indicating greater severity. Demographic and cancer data We obtained data on patients’ demographic and cancer characteristics from the NHS Scotland Cancer Registry. The Registry systematically collects information from hospitals throughout Scotland for all recorded cases of cancer. The data included sex, date of cancer diagnosis, age at cancer diagnosis, social deprivation score (calculated using the Scottish Index of Multiple Deprivation, based on area of residence at the time of cancer diagnosis; see Appendix A for details), primary cancer (see Appendix B for details) and initial cancer treatment objective (curative or palliative) which we used as a proxy for cancer severity that could be applied across all the cancers studied. Mortality data We obtained data on deaths up to April 30, 2012 (that is, 47 months from the first HADS completion on May 12, 2008 and eight months from the last HADS completion on Aug 24, 2011). These data were obtained from the National Records of Scotland (NRS) database and included the date and recorded cause of death of each patient. Data linkage To ensure data security and confidentiality the dataset of the patients’ HADS (depression and anxiety) scores was sent to the Information Services Division of NHS Scotland for linkage using unique patient identification numbers (Community Health Index numbers) and dates of birth. All identifying data were then removed in a one-way linkage to produce the anonymised dataset that was used for analysis. The study was approved by the South East Scotland Research Ethics Committee, the NHS Scotland Caldicott Guardian Forum, and the NHS Scotland Privacy Advisory Committee. Statistical analyses For each patient, we calculated the time to their death from the date they completed the HADS. We included deaths from any cause in our analysis because most of the deaths were recorded as being due to cancer (see results). If a patient had attended the cancer clinic and completed the HADS more than once during the study period, we used the data relating to the earliest of these clinic attendances. We censored patients who had left Scotland (at their date of emigration) and patients who were not known to have died or to have emigrated at the latest date on which data were available (April 30, 2012). Patients whose mortality status was unknown were followed to their last known appointment date (within the study period) or were excluded from the analysis if this was unavailable. We separately analysed the data from patients with each primary cancer (see Appendix B for details). Some of the cancers studied are sex-specific (prostate, breast and gynaecological). For the other, non-sex specific, cancers (colorectal cancer and lung cancer) we conducted separate analyses for males and females because inspection of the data suggested sex differences in the associations between anxiety and survival. Our main analysis consequently comprised seven sets of models with patients grouped as follows: prostate cancer, colorectal cancer – males, lung cancer – males, breast cancer, gynaecological cancer, colorectal cancer – females, lung cancer – females. For patients who had multiple primary cancers, we used the cancer diagnosis that most closely preceded their completion of the HADS to assign them to a group, except where two or more diagnoses were made on the same day (nine patients who were given two different cancer diagnoses on the same day were included in the analyses of both cancers). We used Cox proportional hazards models to estimate the associations of depression (HADS depression score) and anxiety (HADS anxiety score) with subsequent survival. As expected, depression and anxiety scores were associated (Pearson correlation = 0.60, see Appendix C). In order to determine their independent associations with survival, we therefore fitted models that included both as predictor variables (i.e. we calculated the association of depression with survival when adjusted for anxiety and vice versa). Because the associations appeared non-linear, we used restricted cubic splines with four knots (positioned at 5th, 35th, 65th and 95th percentiles) to model the associations of depression and anxiety with survival (we also performed an analysis using cubic splines with five knots, but choose to present results for four knots as five knots sometimes produced implausibly steep increases and decreases in the fitted relationships). We also extended these mutually adjusted models to incorporate interactions between depression and anxiety scores. In these models we first included all products of the linear term for depression and cubic spline terms for anxiety and vice-versa, as is recommended [6]. If these interaction terms were jointly statistically significant we additionally compared the fit of this model with a simpler one that included only the product of the linear terms for depression and anxiety. Having conducted separate analyses for males and females with each of the non-sex specific cancers (colorectal cancer and lung cancer), we performed secondary analyses in which we fitted models to all patients with lung cancer and (separately) all patients with colorectal cancer that included interactions between sex and the cubic spline terms for depression scores and anxiety scores. In all the models, we adjusted for the following covariates: age at cancer diagnosis, time between cancer diagnosis and completion of the HADS, social deprivation score, and initial treatment objective recorded at the time of cancer diagnosis. Depression, anxiety and all adjustment variables were either inherently or treated as fixed over the follow-up time. We expected the associations between continuous adjustment variables (time between cancer diagnosis and HADS completion, age at cancer diagnosis and deprivation score) and survival to be non-linear and therefore used restricted cubic splines with four knots to allow flexible parameterisation of these relationships. The models also included two-way interactions between the time interval between cancer diagnosis and completion of the HADS and the adjustment variables described above. This was because age at cancer diagnosis, social deprivation score and initial cancer treatment objective were all measured at the time of patients’ cancer diagnoses and it is plausible that the magnitude of their confounding associations with survival may change according to the time interval between cancer diagnosis and HADS completion. For each two-way interaction between this time interval and either age at cancer diagnosis or social deprivation score, all products of linear terms were included in the models. We used multiple imputation to deal with missing data on initial cancer treatment objective (2,533 patients) and social deprivation score (two patients). We used the substantive model compatible fully conditional specification (SMCFCS) method for each imputation in order to properly account for interactions and non-linear associations [7]. Imputation models could include extra variables that were found to be predictive of survival and missingness (see Appendix D for further details on the handling of missing data). We performed 20 imputations (separately for each cancer) using the final model. We fitted Cox regression models to each imputed dataset and combined the results using Rubin’s rules [8]. We then calculated predicted hazard ratios (HR) at all levels of depression and anxiety for each cancer. Imputations were carried out in R version 3.4.1 and all analysis models were fitted in Stata version 15 [9, 10]. Results We included data from 19,966 patients in the analysis (see Table 1 for their characteristics). The median time from HADS completion to death or censoring was 1.9 years (IQR: 1.1, 2.8). 5,884 patients died (from all causes) during the period of follow-up. Most (91.5%) of the deaths were recorded as being due to cancer (see Appendix E). The fitted associations of depression and anxiety with survival in males and females are shown in Figures 1 and 2 respectively (see also Appendix F). The figures (which are interpreted in detail in the next section) show plots for each of our seven groups (prostate cancer, colorectal cancer – males, lung cancer – males, breast cancer, gynaecological cancer, colorectal cancer – females, lung cancer – females). The plots on the left of each figure show predicted HRs for the association between depression and survival, without adjustment for anxiety (red lines) and then with adjustment for anxiety (blue lines). The plots on the right of each figure show predicted HRs for the association between anxiety and survival, without adjustment for depression (red lines) and then with adjustment for depression (blue lines). HRs refer to the hazard of mortality for patients with each HADS depression or HADS anxiety score relative to those with a score of zero. The bar chart below each plot shows the percentage of patients with each HADS depression or HADS anxiety score; around 90% of patients had depression scores of 10 or less and around 90% had anxiety scores of 12 or less. Association of depression with survival Greater depression was strongly associated with worse subsequent survival in all the seven groups (p<0.0001 in all groups). The HRs were sizeable, for example the HRs comparing patients with HADS depression scores of 10 and 0 varied from 4.30 (95% CI 2.63, 7.06) for prostate cancer to 1.81 (95% CI 1.48, 2.22) for lung cancer – females. The fitted relationships were not linear, typically being steeper at the lower end of the range than at the higher end. Association of depression with survival when adjusted for anxiety When we adjusted for anxiety, the association between depression and survival remained statistically significant for all seven groups (p<0.0001). The HRs comparing patients with HADS depression scores of 10 and 0 varied from 4.57 (95% CI 2.56, 8.16) for prostate cancer to 2.07 (95% CI 1.64, 2.61) for lung cancer – females. For the female groups depression was more strongly associated with survival after adjustment for anxiety (see figure 2 blue lines compared with red lines). Association of anxiety with survival Greater anxiety was also associated with worse subsequent survival in five of the seven groups (prostate cancer p=0.001, colorectal cancer – males p=0.0001, lung cancer – males p<0.0001, breast cancer p=0.022, gynaecological cancer p=0.040, colorectal cancer – females p=0.152, lung cancer – females p=0.066). The HRs observed were however smaller than those for depression, for example the HRs comparing patients with HADS anxiety scores of 10 and 0 varied from 1.89 (95% CI 1.31, 2.72) for prostate cancer to 0.96 (95% CI 0.67, 1.39) for colorectal cancer – females. Association of anxiety with survival when adjusted for depression When we adjusted for depression, the association of anxiety with survival changed markedly. For males, little or no association between anxiety and survival remained. For females, the association of anxiety with survival was typically in the opposite direction to that observed before we adjusted for depression. That is to say, greater anxiety was now associated with better survival (breast cancer p<0.0001, gynaecological cancer p=0.0002, colorectal cancer – females p=0.037, lung cancer – females p=0.019). The HRs comparing patients with HADS anxiety scores of 10 and 0 varied from 0.87 (95% CI 0.69, 1.10) for lung cancer - females to 0.58 (95% CI 0.39, 0.88) for colorectal cancer - females. The observed difference between the sexes in the association between anxiety and survival was clearest when comparing the plots for the sex-specific cancers. The same directional differences were also seen in the sex-specific analyses of the lung and colorectal cancer groups, however, formal interaction tests from models including both males and females were not statistically significant (colorectal p=0.111, lung p=0.095). Interaction between depression and anxiety in their associations with survival When fitting models with both depression and anxiety and an interaction between the two, there was some evidence of an interaction for those with breast cancer (p=0.025) but not for the other cancers. Results from this analysis are presented in Appendix G. We suggest caution in interpretation as this is only one statistically significant result of many tests done. Discussion Main findings We found that, as expected, greater depression was strongly associated with worse subsequent survival for both male and female patients for all the cancers we studied. After adjusting for anxiety, this association remained in males and became stronger in females. We also found that greater anxiety was associated with worse survival in most of the groups analysed. However, after adjusting for depression the relationship between anxiety and survival changed, disappearing in males and changing direction in females such that greater anxiety became associated with better subsequent survival. This negative association of greater anxiety and worse survival, coupled with the fact that depression and anxiety are highly associated, explains why the association between depression and survival became stronger in females after adjusting for anxiety. Other literature The finding that greater depression is associated with worse subsequent survival in people with cancer has been frequently reported [1–3], but is disputed on methodological grounds [11]. Our findings, from this large methodologically robust study, support this association. Although less studied, the finding that greater anxiety is associated with worse survival in people with cancer has also been reported [3, 12]. Our finding that, after adjustment for depression, this association effectively disappears in males (so that anxiety is no longer associated with survival) and actually reverses direction in females (so that greater anxiety is associated with better survival) is novel. We are not aware of any previous study of the associations of depression and anxiety with survival in patients with cancer that has examined these independent associations. There have however been a small number of relevant studies in other populations. In people with cardiac disease, a systematic review reported an association between greater anxiety and worse survival, but also that this association was weakened when severity of depression was adjusted for, suggesting that depression was the more important factor [13]. A large study of patients with suspected cardiac disease undergoing exercise testing found, as we did in patients with cancer, that after adjustment for depression, anxiety was associated with better, rather than worse survival [14]. Studies of the general population have also found that anxiety predicts better rather than worse life expectancy [15], and that when anxiety complicates depression the association between depression and worse survival is reduced [16]. These similar findings in non-cancer populations increase our confidence that our novel findings in patients with cancer are meaningful. Interpretation Our results suggest that, whatever the mechanism of the association between depression and worse survival in people with cancer, it is specific to depression [17]. Potential mechanisms for this association have been proposed, but none proven [18]. It is of interest that the relationship between depression and survival was not linear, typically being steeper at the lower end of the range than at the higher end. The explanation for this observation is unclear. However, we note that mild depression has been associated with worse survival in patients with heart disease [19], and small changes in that mild depression over time have been associated with improved survival [20]. Our findings in patients with cancer and these in patients with heart disease suggest that we should not focus solely on the association between severe depression and survival in patients with medical illnesses, but also consider mild and moderate depression. Anxiety appears to have a different relationship with survival than depression, with no association in males and an association with better, not worse, survival in females. This is most clearly seen in the female-specific cancers (breast and gynaecological). There are a number of potential mechanisms for the association of anxiety with better survival, but perhaps the most plausible is that anxiety leads to healthier behaviours, more medical care seeking and better adherence to medical treatments [21]. It is of interest that this is only clearly observed in female specific cancers and may reflect the importance of patient adherence to treatment recommendations in these cancers. Strengths and limitations The strengths of our study were: (a) the use of data from a large representative sample of patients with common cancers attending UK NHS cancer centres serving a geographically defined area; (b) the availability of continuous measures of depression and anxiety using a well-validated scale; (c) a cancer diagnosis and severity assessment done by oncologists; (d) an almost complete follow-up of the cohort using individually linked national registry data, including data on cause of death and (e) robust analysis of these unique data accounting for missing data and controlling for possible confounders, including not only age and sex, but also social deprivation (determined by the patient’s address) and initial cancer severity (determined by recorded treatment objective). Despite these strengths, our study also had limitations including: (a) findings that may not necessarily generalise to other populations (such as patients in different healthcare settings or those who were diagnosed with cancer many years ago and who no longer attend clinics); (b) the assessment of depression and anxiety using self-rating scales which unlike a diagnostic interview do not provide diagnoses, but rather a continuous measure of symptom severity; (c) some missing data on the HADS score and initial cancer treatment objective (which we addressed using multiple imputation in the analysis but we cannot rule out the possibility that these data were not missing at random); (d) the completion of the HADS at varying intervals after initial cancer diagnosis (although we did take account of this in our analysis); (e) a lack of information on the time-course of depression and anxiety either prior to or subsequent to the HADS completion; (f) follow-up data on patients for a mean of approximately two years from the time of HADS completion but not on all patients to the time of their death; (g) an inability to fully adjust for all potential confounders - in particular we had to rely on initial treatment objective as a measure of cancer severity as it was not possible to combine the different staging systems used for different cancer types in our analysis; (h) an inability to control for medical comorbidities, although it is unlikely that these were important in determining survival, as almost all the patient deaths were attributed to cancer. Conclusions Depression and anxiety have both been associated with the worse subsequent survival of people with common cancers. The findings presented here confirm that depression is associated with survival but also indicate that, when depression is adjusted for, anxiety is not. In fact anxiety may even predict better survival in females. The implication of these findings is that whatever the mechanism of the association of depression with worse survival, it is specific to depression. Depression and anxiety should not therefore be lumped together as ‘emotional distress’ but should be considered separately in future studies of the predictors of survival in people with cancer and indeed other illnesses. Supplementary Material Appendix Acknowledgements This analysis was funded by the UK National Institute for Health Research Collaboration for Leadership in Applied Health Research and Care Oxford at Oxford NHS Foundation Trust. MS is an NIHR Senior Investigator. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health. The original data collection was funded by the charity Cancer Research UK (grant no. C5547/A7375). The funders had no involvement in study design; collection, analysis and interpretation of data; writing of the report; or in the decision to submit the article for publication. Figure 1 Associations of survival with depression and anxiety for male patients with cancer Plots show predicted hazard ratios (hazard of mortality for patients with each HADS-D or HADS-A score relative to patients with a score of zero). Bar Charts show the percentage of patients with each HADS-D and HADS-A score. Red lines show unadjusted hazard ratios (prostate cancer HADS-D p<0.0001, HADS-A p=0.0013; colorectal cancer – males HADS-D p<0.0001, HADS-A p=0.0001; lung cancer – males HADS-D p<0.0001, HADS-A p<0.0001). Blue lines show the hazard ratios adjusted for the other symptom (depression or anxiety) of interest (prostate cancer HADS-D p<0.0001, HADS-A p=0.8401; colorectal cancer – males HADS-D p<0.0001, HADS-A p=0.4586; lung cancer – males HADS-D p<0.0001, HADS-A p=0.8798). Note that y-axis scales are different for HADS-D and HADS-A, but consistent across cancers. Figure 2 Associations of survival with depression and anxiety for female patients with cancer Plots show predicted hazard ratios (hazard of mortality for patients with each HADS-D or HADS-A score relative to patients with a score of zero). Bar Charts show the percentage of patients with each HADS-D and HADS-A score. Red lines show unadjusted hazard ratios (breast cancer HADS-D p<0.0001, HADS-A p=0.0220; gynaecological cancer HADS-D p<0.0001, HADS-A p=0.0398; colorectal cancer - females HADS-D p=0.0001, HADS-A p=0.1521; lung cancer - females HADS-D p<0.0001, HADS-A p=0.0656). Blue lines show the hazard ratios adjusted for the other symptom (depression or anxiety) of interest (breast cancer HADS-D p<0.0001, HADS-A p<0.0001; gynaecological cancer HADS-D p<0.0001, HADS-A p=0.0002; colorectal cancer – females HADS-D p<0.0001, HADS-A p=0.0371; lung cancer - females HADS-D p<0.0001, HADS-A p=0.0186). Note that y-axis scales are different for HADS-D and HADS-A, but consistent across cancers. Table 1 Demographics, depression and anxiety, and survival of patients included in the analysis. Prostate cancer Colorectal cancer-males Lung cancer-males Breast cancer Gynaecological cancer Colorectal cancer- females Lung cancer- females Total 1531a 1573 2299 8467 a 2910 a 1154 2041 Sex   Female 0 (0%) 0 (0%) 0 (0%) 8467 (100%) 2910 (100%) 1154 (100%) 2041 (100%)   Male 1531 (100%) 1573 (100%) 2299 (100%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) Age at cancer diagnosis [median years, IQR] 66 [62, 72] 65 [59, 72] 68 [61, 74] 57 [49, 66] 60 [50, 69] 65 [57, 72] 67 [60, 74] SIMD score quintile b   1 262 (17%) 281 (18%) 749 (33%) 1442 (17%) 617 (21%) 217 (19%) 700 (34%)   2 251 (16%) 304 (19%) 543 (24%) 1547 (18%) 617(21%) 231 (20%) 511 (25%)   3 254 (17%) 297 (19%) 365 (16%) 1539 (18%) 555 (19%) 195 (17%) 328 (16%)   4 334 (22%) 279 (18%) 309 (13%) 1630 (19%) 546 (19%) 204(18%) 249 (12%)   5 430 (28%) 411 (26%) 333 (14%) 2308 (27%) 575 (20%) 307 (27%) 253 (12%)   Missing 0 (0%) 1 (0%) 0 (0%) 1 (0%) 0 (0%) 0 (0%) 0 (0%) Initial cancer treatment objective   Curative 635 (41%) 1116 (71%) 566 (25%) 6533 (77%) 2010 (69%) 830 (72%) 546 (27%)   Palliative 634 (41%) 319 (20%) 1644 (72%) 466 (6%) 521 (18%) 231 (20%) 1391 (68%)   Missing 262 (17%) 138 (9%) 89 (4%) 1468 (17%) 379 (13%) 93 (8%) 104 (5%) Time interval between cancer diagnosis & HADS c completion [median years, IQR] 2.0 [0.8, 4.4] 1.0 [0.3, 2.5] 0.3 [0.1, 0.8] 2.0 [0.4, 5.2] 1.0 [0.4, 2.9] 1.0 [0.3, 2.6] 0.3 [0.1, 0.9] HADS   HADS-D (median, IQR) 3 [1, 6] 3 [1, 6] 6 [3, 9] 3 [1, 6] 4 [1, 7] 3 [1, 7] 6 [3, 9]   HADS-A (median, IQR) 4 [1, 7] 4 [1, 7] 5 [3, 9] 5 [3, 9] 5 [2, 8] 5 [2, 8] 7 [4, 10] Time from HADS completion to death or censoring [median years, IQR] 2.2 [1.7, 3.1] 1.8 [1.2, 2.8] 0.8 [0.3, 1.4] 2.3 [1.6, 3.0] 1.9 [1.2, 2.8] 1.8 [1.2, 2.7] 0.9 [0.4, 1.6] Died during study period 288 (19%) 518 (33%) 1603 (70%) 1000 (12%) 824 (28%) 328(28%) 1328 (65%) Data are n (%) unless stated otherwise. a 9 patients are included in this table twice because they were diagnosed with more than one primary cancer on the same day: 1 had breast & gynaecological cancers, 3 had colorectal & gynaecological cancers, 2 had breast & lung cancers, 1 had breast & colorectal cancers, 1 had colorectal & lung cancers (male), 1 had colorectal & prostate cancers. b Scottish Index of Multiple Deprivation quintile score: 1=most deprived, 5=least deprived. c Hospital Anxiety and Depression Scale: HADS-D=depression severity, HADS-A=anxiety severity Declaration of interests: None. [1] Satin JR Linden W Phillips MJ Depression as a predictor of disease progression and mortality in cancer patients: a meta-analysis Cancer 2009 115 22 5349 61 19753617 [2] Pinquart M Duberstein PR Depression and cancer mortality: a meta-analysis Psychol Med 2010 40 11 1797 810 20085667 [3] Wang YH Li JQ Shi JF Que JY Liu JJ Lappin JM Leung J Ravindran AV Chen WQ Qiao YL Shi J Depression and anxiety in relation to cancer incidence and mortality: a systematic review and meta-analysis of cohort studies Mol Psychiatry 2019 [4] Kircanski K LeMoult J Ordaz S Gotlib IH Investigating the nature of co-occurring depression and anxiety: Comparing diagnostic and dimensional research approaches Journal of Affective Disorders 2017 216 123 135 27554605 [5] Zigmond AS Snaith RP The hospital anxiety and depression scale Acta Psychiatrica Scandinavica 1983 67 361 370 6880820 [6] Harrell JFE Regression Modeling Strategies: With Applications to Linear Models Logistic and Ordinal Regression, and Survival Analysis 2015 [7] Bartlett J smcfcs: Multiple Imputation of Covariates by Substantive Model Compatible Fully Conditional Specification [8] Rubin D Multiple Imputation for Nonresponse in Surveys Wiley New York 1987 [9] R Core Team R: A language and environment for statistical computing Foundation for Statistical Computing Vienna Austria 2017 [10] StataCorp Stata Statistical Software: Release 15 StataCorp LLC College Station, TX 2017 [11] Miloyan B Fried E A reassessment of the relationship between depression and allcause mortality in 3,604,005 participants from 293 studies World Psychiatry 2017 16 2 219 220 28498573 [12] Shim EJ Lee JW Cho J Jung HK Kim NH Lee JE Min J Noh WC Park SH Kim YS Association of depression and anxiety disorder with the risk of mortality in breast cancer: A National Health Insurance Service study in Korea Breast Cancer Res Treat 2019 [13] Celano CM Millstein RA Bedoya CA Healy BC Roest AM Huffman JC Association between anxiety and mortality in patients with coronary artery disease: A meta-analysis American Heart Journal 2015 170 6 1105 1115 26678632 [14] Herrmann C Brand-Driehorst S Buss U Rüger U Effects of anxiety and depression on 5-year mortality in 5057 patients referred for exercise testing Journal of Psychosomatic Research 2000 48 4 455 462 10880666 [15] Colman I Kingsbury M Sucha E Horton NJ Murphy JM Gilman SE Depressive and anxious symptoms and 20-year mortality: Evidence from the Stirling County study Depression and anxiety 2018 35 7 638 647 29719088 [16] Mykletun A Bjerkeset O øverland S Prince M Dewey M Stewart R Levels of anxiety and depression as predictors of mortality: the HUNT study British Journal of Psychiatry 2009 195 2 118 125 [17] Spiegel D Giese-Davis J Depression and cancer: mechanisms and disease progression Biol Psychiatry 2003 54 3 269 82 12893103 [18] Bortolato B Hyphantis TN Valpione S Perini G Maes M Morris G Kubera M Köhler CA Fernandes BS Stubbs B Pavlidis N Depression in cancer: The manybiobehavioral pathways driving tumor progression Cancer Treatment Reviews 2017 52 58 70 27894012 [19] Bush DE Ziegelstein RC Tayback M Richter D Stevens S Zahalsky H Fauerbach JA Even minimal symptoms of depression increase mortality risk after acute myocardial infarction The American Journal of Cardiology 2001 88 4 337 341 11545750 [20] Lespérance F Frasure-Smith N Talajic M Bourassa MG Five-Year Risk of Cardiac Mortality in Relation to Initial Severity and One-Year Changes in Depression Symptoms After Myocardial Infarction Circulation 2002 105 9 1049 1053 11877353 [21] Ossola P Gerra ML DePanfilis C Tonna M Marchesi C Anxiety, depression, and cardiac outcomes after a first diagnosis of acute coronary syndrome Health Psychology 2018 37 12 1115 1122 30307271

PMC007xxxxxx/PMC7614810.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9502015 Nat Med Nat Med Nature medicine 1078-8956 1546-170X 33077958 7614810 EMS181445 10.1038/s41591-020-1076-0 Article Discovery and validation of a personalized risk predictor for incident tuberculosis in low transmission settings https://orcid.org/0000-0002-6257-1285 Gupta Rishi K. 1 Calderwood Claire J. 1 Yavlinsky Alexei 2 https://orcid.org/0000-0002-3982-642X Krutikov Maria 1 Quartagno Matteo 3 Aichelburg Maximilian C. 4 Altet Neus 56 Diel Roland 78 https://orcid.org/0000-0002-5460-0189 Dobler Claudia C. 910 Dominguez Jose 111213 Doyle Joseph S. 1415 Erkens Connie 16 Geis Steffen 17 https://orcid.org/0000-0002-1572-5421 Haldar Pranabashis 18 https://orcid.org/0000-0001-6569-0768 Hauri Anja M. 19 Hermansen Thomas 20 Johnston James C. 21 Lange Christoph 22232425 Lange Berit 26 https://orcid.org/0000-0002-5490-8968 van Leth Frank 242728 Muñoz Laura 29 Roder Christine 1415 Romanowski Kamila 21 Roth David 21 https://orcid.org/0000-0001-5482-0002 Sester Martina 2430 Sloot Rosa 31 https://orcid.org/0000-0002-1600-4474 Sotgiu Giovanni 2432 https://orcid.org/0000-0002-9253-6034 Woltmann Gerrit 18 Yoshiyama Takashi 33 Zellweger Jean-Pierre 2434 Zenner Dominik 1 Aldridge Robert W. 2 Copas Andrew 13 Rangaka Molebogeng X. 133536 Lipman Marc 3738 https://orcid.org/0000-0002-4774-0853 Noursadeghi Mahdad 39 https://orcid.org/0000-0002-0370-1430 Abubakar Ibrahim 1✉ 1 Institute for Global Health, University College London, London, UK 2 Institute of Health Informatics, University College London, London, UK 3 MRC Clinical Trials Unit, Institute of Clinical Trials and Methodology, University College London, London, UK 4 Department of Dermatology, Sozialmedizinisches Zentrum Ost-Donauspital, Vienna, Austria 5 Unitat de Tuberculosis, Hospital Universitari Vall d’Hebron-Drassanes, Barcelona, Spain 6 Unitat de TDO de la Tuberculosis ‘Servicios Clínicos’, Barcelona, Spain 7 Institute for Epidemiology, University Hospital Schleswig-Holstein, Campus Kiel, Kiel, Germany 8 Lung Clinic Grosshansdorf, Airway Research Center North (ARCN), Großhansdorf, Germany 9 Institute for Evidence-Based Healthcare, Faculty of Health Sciences and Medicine, Bond University, Gold Coast, Queensland, Australia 10 Department of Respiratory Medicine, Liverpool Hospital, Sydney, Australia 11 Institut d’Investigació Germans Trias i Pujol, Badalona, Barcelona, Spain 12 CIBER Enfermedades Respiratorias, Badalona, Barcelona, Spain 13 Universitat Autònoma de Barcelona, Badalona, Barcelona, Spain 14 Department of Infectious Diseases, The Alfred and Monash University, Melbourne, Australia 15 Disease Elimination Program, Burnet Institute, Melbourne, Australia 16 KNCV Tuberculosis Foundation, The Hague, The Netherlands 17 Institute for Medical Microbiology and Hospital Hygiene, Philipps University of Marburg, Marburg, Germany 18 Respiratory Biomedical Research Centre, Institute for Lung Health, Department of Respiratory Sciences, University of Leicester, Leicester, UK 19 Hesse State Health Office, Dillenburg, Germany 20 International Reference Laboratory of Mycobacteriology, Statens Serum Institut, Copenhagen, Denmark 21 British Columbia Centre for Disease Control, Vancouver, British Columbia, Canada 22 Division of Clinical Infectious Diseases, Research Center Borstel, Borstel, Germany 23 German Center for Infection Research (DZIF), Clinical Tuberculosis Center, Borstel, Germany 24 Tuberculosis Network European Trials Group (TBnet), Borstel, Germany 25 Department of Medicine, Karolinska Institute, Stockholm, Sweden 26 Department of Epidemiology, Helmholtz Centre for Infection Research, Braunschweig, Germany 27 Amsterdam Institute for Global Health and Development, Amsterdam, the Netherlands 28 Department of Global Health, Amsterdam University Medical Centres, Amsterdam, the Netherlands 29 Department of Clinical Sciences, University of Barcelona, Barcelona, Spain 30 Department of Transplant and Infection Immunology, Saarland University, Homburg, Germany 31 Department of Paediatrics and Child Health, Desmond Tutu TB Centre, Faculty of Medicine and Health Sciences, Stellenbosch University, Cape Town, South Africa 32 Clinical Epidemiology and Medical Statistics Unit, Department of Medical, Surgical and Experimental Sciences, Uniiversity of Sassari, Sassari, Italy 33 Research Institute of Tuberculosis, Tokyo, Japan 34 Swiss Lung Association, Berne, Switzerland 35 Wellcome Centre for Infectious Diseases Research in Africa, Institute of Infectious Diseases and Molecular Medicine, University of Cape Town, Cape Town, South Africa 36 Division of Epidemiology and Biostatistics, School of Public Health, University of Cape Town, Cape Town, South Africa 37 UCL-TB and UCL Respiratory, University College London, London, UK 38 Royal Free London NHS Foundation Trust, London, UK 39 Division of Infection & Immunity, University College London, London, UK ✉ i.abubakar@ucl.ac.uk 01 12 2020 19 10 2020 19 7 2023 26 7 2023 26 12 19411949 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. The risk of tuberculosis (TB) is variable among individuals with latent Mycobacterium tuberculosis infection (LTBI), but validated estimates of personalized risk are lacking. In pooled data from 18 systematically identified cohort studies from 20 countries, including 80,468 individuals tested for LTBI, 5-year cumulative incident TB risk among people with untreated LTBI was 15.6% (95% confidence interval (CI), 8.0-29.2%) among child contacts, 4.8% (95% CI, 3.0-7.7%) among adult contacts, 5.0% (95% CI, 1.6-14.5%) among migrants and 4.8% (95% CI, 1.5-14.3%) among immunocompromised groups. We confirmed highly variable estimates within risk groups, necessitating an individualized approach to risk stratification. Therefore, we developed a personalized risk predictor for incident TB (PERISKOPE-TB) that combines a quantitative measure of T cell sensitization and clinical covariates. Internal-external cross-validation of the model demonstrated a random effects meta-analysis C-statistic of 0.88 (95%CI, 0.82-0.93) for incident TB. In decision curve analysis, the model demonstrated clinical utility for targeting preventative treatment, compared to treating all, or no, people with LTBI. We challenge the current crude approach to TB risk estimation among people with LTBI in favor of our evidence-based and patient-centered method, in settings aiming for pre-elimination worldwide. pmcGlobally, TB accounts for the greatest number of deaths from a single pathogen, with an estimated 1.5 million deaths and 10 million incident cases in 20181. The World Health Organization’s End TB Strategy ambitiously aims for a 95% reduction in TB mortality and a 90% reduction in TB incidence by 20352. As part of this strategy, the priority for low transmission settings is to achieve pre-elimination (annual incidence of <1 per 100,000) by 20352. Preventative antimicrobial treatment for LTBI is considered critical for achieving this objective2,3. In the absence of an assay to detect viable M. tuberculosis bacteria, LTBI is currently clinically defined as evidence of T cell memory to M. tuberculosis, in the absence of concurrent disease and any previous treatment4,5. Individuals with LTBI are generally considered to have a lifetime TB risk ranging from 5% to 10%4, which is reduced by 65–80% with preventative treatment6. The positive predictive value (PPV) for TB using the current definition of LTBI is less than 5% over a 2-year period among risk groups, such as adult TB contacts7–9. This might lead to a large burden of unnecessary preventative treatment, with associated risks of drug toxicity to patients and excess economic costs to health services. The low PPV might also undermine the cascade of care, including uptake of preventative treatment among individuals in target groups, who perceive their individual risk of developing TB to be low10,11. In fact, the risk of TB among individuals with LTBI is highly variable between study populations, with incidence rates ranging from 0.3 to 84.5 per 1,000 person-years of follow-up7,12. Thus, quoting the 5–10% lifetime estimate is likely to be inaccurate for many people. Improved risk stratification is, therefore, essential to enable precise delivery of preventative treatment to those most likely to benefit5,13. Multiple studies have shown that the magnitude of the T cell response to M. tuberculosis is associated with incident TB risk, raising hope that quantitative tuberculin skin test (TST) or interferon gamma release assay (IGRA) results might improve predictive ability14,15. However, implementing higher diagnostic thresholds alone does not improve prediction on a population level owing to a marked loss of sensitivity with this approach16. In this study, we first sought to characterize the population risk of TB among people tested for LTBI using an individual participant data meta-analysis (IPD-MA). To study progression from LTBI to TB disease more accurately, we focused on settings with low transmission (defined as annual incidence ≤20 per 100,000 persons), where there is a minimal risk of reinfection during follow-up. We confirmed highly variable estimates of risk, necessitating an individual-level approach to risk estimation. Finally, we developed and validated a directly data-driven personalized risk predictor for incident TB (PERISKOPE-TB) that combines a quantitative T cell response measure with key clinical covariates. Results Systematic review Our systematic review identified 26 studies that aimed to assess the risk of progression to TB disease among individuals tested for LTBI in low TB transmission settings; corresponding authors of these studies were invited to contribute individual-level data (Extended Data Fig. 1). Of these, we received 18 individual-level data sets, including participants recruited in 20 countries. The pooled data set included a total of 82,360 individual records; of these individuals, 51,697 had evidence of LTBI, and 826 were diagnosed with TB. Of the received data, 80,468 participants (including 803 TB cases) had sufficient data for inclusion in the primary analysis (Extended Data Fig. 2). The characteristics of the included study data sets are summarized in Table 1 and Supplementary Table 1. Characteristics of the eight eligible studies for which IPD were not obtained were similar to those included in the analysis (Supplementary Table 2). Eight studies recruited adults only; the remainder recruited both adults and children. The target population was recent TB contacts in nine studies17–25, people living with HIV in two studies26,27, mixed immunocompromised groups in two studies28,29, transplant recipients in one study30, mixed population screening in two studies31,32, recent migrants in one study33 and a combination of recent contacts and migrants in one study9. Median follow-up of all participants was 3.7 years (interquartile range (IQR), 2.1–5.3 years). All contributing studies reported baseline assessments for prevalent TB through routine clinical evaluations, and all included culture-confirmed and clinically diagnosed TB cases in their case definitions. Four studies had a proportion of participants lost to follow-up of more than 5%18,24,27,28; baseline characteristics of those lost to follow-up were similar to those followed-up in each of these studies (Supplementary Table 3). All contributing studies achieved quality assessment scores of 6/6, 6/7 or 7/7 (Supplementary Table 4). Population-level analysis In the pooled data set, the 2-year cumulative risk of incident TB was estimated as 4.0% (95% CI, 2.6–6.3%) among people with LTBI who did not receive preventative therapy, 0.7% (0.4–1.3%) in people with LTBI who commenced preventative therapy and 0.2% (0.1–0.4%) in people without LTBI (Fig. 1 and Supplementary Table 5). The corresponding 5-year risk of incident TB among these groups was 5.4% (3.5–8.5%), 1.1% (0.6–2.0) and 0.3% (0.2–0.5%), respectively. Among untreated people with LTBI, 2-year risk of incident TB was 14.6% (95% CI, 7.5–27.4) among recent child (<15 years) contacts, 3.7% (2.3–6) among adult contacts, 4.1% (1.3-–12) among migrants and 2.4% (0.8–6.8) among people screened owing to immunocompromise (without an index exposure). Corresponding 5-year risk was 15.6% (8.0–29.2) among recent child contacts, 4.8% (3.0–7.7) among adult contacts, 5.0% (1.6–14.5) among migrants and 4.8% (1.5–14.3) among people screened owing to immunocompromise. Among recent child contacts, risk was markedly higher among those younger than 5 years old compared to those aged 5-14 years (2-year risk, 26.0% (9.4–60.1) versus 12.4% (5.7–25.6); Fig. 1). Among child contacts, 85.4% and 93.7% of cumulative risk was accrued in the first 1 and 2 years of follow-up, respectively. Among adult contacts and migrants, the annual risk also declined markedly with time. Of the cumulative 5-year risk, 58.2% and 77.6% were accrued in the first 1 and 2 years of follow-up for adult contacts, with corresponding values among migrants of 66.4% and 81.6%, respectively. There was a more even distribution of risk during follow-up in the immunocompromised group. TB incidence rates in years 0–2 and 2–5 of follow-up, stratified by LTBI result, commencement of preventative treatment and indication for screening, are shown in Extended Data Figs. 4 and 5. Within each of the risk groups assessed, incidence rates among untreated people with LTBI were markedly higher in the 0-2-year interval, compared to the 2–5-year interval, but were highly heterogeneous across studies (I2 statistics, representing the proportion of variance that is considered owing to between-study heterogeneity, ranged from 54% to 91% for incidence rates during the 0–2-year interval among untreated people with LTBI, when stratified by indication for screening; forest plots are shown in Extended Data Fig. 5). These findings suggest highly variable TB risk among people with LTBI, even within risk groups. Prediction model development The observed heterogeneity in TB incidence rates across studies, even after stratification by binary LTBI result, commencement of preventative treatment and indication for screening, suggests that an individual-level approach to risk stratification is required. We, therefore, developed a personalized risk prediction model using a subset of the received data (where sufficient individual-level variables were available), including 528 patients with TB among 31,721 participants from 15 studies (Extended Data Fig. 2). All of these data sets were used for model development and validation, using the internal-external cross-validation (IECV) framework34 described below. Characteristics of the studies included in prediction model development and validation were similar to those that were not (Table 1). Our modeling approach used a flexible parametric survival model with two degrees of freedom on a proportional hazards scale, because this showed the best fit in each imputed data set. From our list of a priori variables of interest, we evaluated nine candidate predictors, of which only previous Bacille Calmette–Guérin (BCG) vaccination and gender were omitted from the final model. The final prediction model included age, a composite ‘TB exposure’ variable (modeled with time-varying covariates to account for non-proportional hazards), time since migration for migrants from countries with high TB incidence, HIV status, solid organ or hematological transplant receipt, normalized LTBI test result and preventative treatment commencement. The final model coefficients and standard errors, pooled across multiply imputed data sets, are summarized in Supplementary Table 6, with visual representations of associations between each variable and incident TB risk shown in Fig. 2. IECV Next, we used the IECV framework, iteratively discarding one study data set from the model training set and using this for external validation, to concurrently validate the prediction model, explore between-study heterogeneity and examine generalizability34. Model discrimination and calibration parameters for 2-year risk of incident TB from the primary validation studies are shown in Fig. 3. We assessed discrimination using the C-statistic, which ranged from 0.78 (95% CI, 0.47–1.0) in a study of immunocompromised participants with a small number of incident TB cases29 to 0.97 (0.94-0–99) in a study of TB contacts18. The random effects meta-analysis estimate of the C-statistic was 0.88 (0.82–0.93). Calibration assesses agreement between predicted and observed risk. We assessed calibration visually using grouped calibration plots, supplemented by the calibration-in-the-large (CITL) and slope statistics (Fig. 3). Visual calibration plots suggested reasonable calibration in most studies (Extended Data Fig. 6). Because incident TB is an infrequent outcome, predictions were appropriately low, with average predicted risk less than 10% in all quintiles of risk. CITL and calibration slopes of 0 and 1 indicate perfect calibration, respectively. The pooled random effects meta-analysis CITL estimate was 0.14 (95% CI, –0.24 to 0.53), with evidence of systematic under-estimation of risk in one study (CITL, 1.02 (0.61–1.43)) and over-estimation in one study (CITL, –0.64 (–1.09 to 0.19)). The pooled random effects meta-analysis calibration slope estimate was 1.11 (0.83–1.38). Slopes appeared heterogeneous, although visual assessment of calibration plots suggested that these were prone to being extreme owing to the skewed distribution of predicted and observed risk, likely reflecting the relatively rare occurrence of incident TB events. Distribution of predicted risk and individual predictions Figure 4 shows the distributions of predicted TB risk among participants who did not commence preventative treatment from the pooled IECV validation sets, stratified by 1) binary LTBI test result and 2) indication for screening (among those with a positive test). The median predicted 2-year TB risk was 2.0% (IQR, 0.8–3.7%) and 0.2% (IQR, 0.1–0.3%) among participants with positive and negative binary LTBI test results, respectively. We then examined incident TB risk in four quartiles of predicted risk among untreated participants with positive LTBI tests from the pooled validation sets. Kaplan–Meier plots of the four quartiles showed clear separation of observed risk among these four groups (Fig. 4c), with illustrative predicted survival curves for one randomly sampled individual patient per quartile shown in Fig. 4d. Decision curve analysis Net benefit quantifies the tradeoff between correctly identifying true-positive patients (progressing to incident TB) and incorrectly detecting false positives, with weighting of each by the threshold probability35,36. The threshold probability corresponds to a measure of both the perceived risk:benefit ratio of initiating preventative treatment and the threshold of predicted risk above which treatment is recommended. How patients and clinicians weigh the relative costs of drug-related adverse events (as a result of inappropriate treatment) against the benefits of preventing a case of TB can be subjective. Among untreated participants with LTBI from the pooled validation sets in IECV, net benefit for the prediction model was greater than either treating all LTBI patients or treating none, throughout a range of threshold probabilities from 0% to 20% (reflecting a range of clinician and patient preferences) (Fig. 5). Sensitivity analyses We re-examined population-level TB risk without any exclusion of prevalent TB (cases diagnosed <42 d from testing), resulting in markedly higher cumulative risk for each risk group (Extended Data Fig. 3). Recalculation of model predictor parameters revealed similar directions and magnitudes of effect to the primary model when using shorter and longer definitions of prevalent TB (baseline risk was expectedly higher with shorter definitions) and when excluding participants who received preventative treatment (Supplementary Table 7). Model parameters were noted to be more extreme when using a complete case approach (for variables other than HIV, which was assumed negative when missing). The pooled random effects meta-analysis C-statistic from IECV when limiting to participants who did not receive preventative treatment was 0.89 (95% CI, 0.82–0.93), similar to the primary analysis (Extended Data Fig. 7a). The pooled random effects meta-analysis C-statistic, including only participants with a positive binary LTBI test, was 0.77 (0.70–0.83). This finding indicates good discrimination even among participants with a conventional diagnosis of LTBI, albeit lower than discrimination when also including participants with a negative binary LTBI test, likely owing to the high negative predictive value of LTBI tests when using standard cutoffs (Extended Data Fig. 7b). Finally, to assess model performance in situations where the quantitative test results are not available, we imputed an average quantitative positive or negative LTBI test result (based on the medians among the study population), according to the binary result in the validation sets. This analysis provided a pooled random effects meta-analysis C-statistic of 0.86 (0.76–0.93; Extended Data Fig. 7c), and net benefit appeared higher when using this model than the strategies of treating either all patients with evidence of LTBI or no patients, across the range of threshold probabilities. However, the model using a binary test result had a lower C-statistic and slightly lower net benefit across most threshold probabilities compared to the full model using quantitative test results (Extended Data Fig. 7d). Discussion In this study, we examined population-level incident TB risk in a pooled data set of more than 80,000 individuals tested for LTBI in 20 countries with low M. tuberculosis transmission (annual incidence ≤20 per 100,000 persons). We found cumulative 5-year risk of incident TB among people with untreated LTBI approaching 16% among child contacts and approximately 5% among recent adult contacts, migrants from high TB-burden settings and immunocompromized individuals. Most cumulative 5-year risk was accrued during the first year among risk groups with an index exposure, supporting previous data suggesting that risk of progressive TB declines markedly with increasing time since infection13. However, we noted substantial variation in incidence rates even within these risk groups, suggesting that an individual-level approach to risk stratification is required. Therefore, we developed the first directly data-driven model, to our knowledge, to incorporate the magnitude of the T cell response to M. tuberculosis with readily available clinical metadata to capture heterogeneity within risk groups and generate personalized risk predictions for incident TB in settings aiming for pre-elimination. Clinical covariates in the final model included age, recent contact (including proximity and infectiousness of the index case), migration from high TB-burden countries (and time since arrival), HIV status, solid organ or hematological transplant receipt and commencement of preventative treatment. The model was externally validated by quantifying the meta-analysis C-statistic for predicting incident disease over 2 years and by evaluating its calibration, using recommended methods37. Most importantly, the model showed clear clinical utility for informing the decision to initiate preventative treatment compared to treating all or no patients with LTBI. The personalized predictions from our model will enable more precise delivery of preventative treatment to those at highest risk of TB disease while concurrently reducing toxicity and costs related to treatment of people at lower risk. Moreover, the model will allow clinicians and patients to make more informed and individualized choices when considering initiation of preventative treatment. The model also challenges the fundamental notion of an arbitrary binary test threshold for diagnosis of LTBI. By incorporating a quantitative measure of immunosensitization to M. tuberculosis, we facilitate a shift from the conventional paradigm of LTBI as a binary diagnosis toward personalized risk stratification for progressive TB. This approach takes advantage of stronger T cell responses being a correlate of risk while guarding against a loss of sensitivity by arbitrarily introducing higher test thresholds programmatically16. The results of our analyses are consistent with and extend existing evidence. Recent analyses report similar population-level TB incidence rates among adult contacts12, with markedly higher risk among young children38. Moreover, these recent meta-analyses confirm highly heterogeneous population-level estimates, thus justifying an individual-level approach to risk estimation12,38. Previous models developed and validated in Peru, a high transmission setting, have generated individual or household-level TB risk estimates for TB contacts39–41. Another model, parameterized using aggregate data estimates from multiple sources, seeks to estimate TB risk after LTBI testing in all settings42. However, there are currently no publicly available validation data to support its use, and the model omits key predictor variables identified in the current study (including the magnitude of the T cell response and infectiousness of index cases)42. Strengths of the current study include the size of the data set, curated through comprehensive systematic review in accordance with Preferred Reporting Items for a Systematic Review and Meta-analysis of Individual Participant Data standards43 and with IPD obtained for 18 of 26 (69%) eligible studies. This allowed us to examine progression from LTBI to TB disease using the largest adult and pediatric data set available to date, to our knowledge. We conducted population-level analyses using both one- and two-stage IPD-MA approaches to present both cumulative TB risk and time-stratified incidence rates, respectively, with consistent results from both. We adhered to Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD)44 standards, using the recommended approach of IECV37, leading to a fully data-driven and validated model for personalized risk estimates after LTBI testing. The coefficients presented in the model are clinically plausible and have been made publicly available to facilitate further independent external validation. Moreover, the contributing data sets included heterogeneous populations of adults, children, recent TB contacts, migrants from high TB-burden countries and immunocompromised groups from 20 countries across Europe, North America, Asia and Oceania, thus making our results generalizable to settings aiming for pre-elimination globally. We also used a comprehensive approach to addressing missing data by using multi-level multiple imputation in the primary analysis, assuming missingness at random and in keeping with recent guidance34,45. This approach facilitated imputation of variables that were systematically missing from some included studies. Previous BCG vaccination and HIV status were noted to be missing from a large proportion of participants. This missingness might have reduced our power to detect an association between these variables and incident TB, and BCG vaccination was notably not included in the final prognostic model. Although increasing data support a role for BCG vaccination in reducing sensitization to M. tuberculosis46,47, additional data are required to further assess the association between BCG vaccination and incident TB risk after adjustment for other covariates, including quantitative T cell responses. We supported our primary multiple imputation approach using a complete case sensitivity analysis (for variables other than HIV, which was assumed to be negative when missing). This sensitivity analysis revealed similar findings to the primary analyses, although effect estimates were noted to be more extreme in the complete case approach, likely owing to a degree of bias in the latter, because complete cases analysis assumes no association between the pattern of missingness and the outcome (that is, incident TB) after adjusting for all other covariates48. Given that TB incidence and predictor missingness both varied according to contributing study, this assumption is unlikely to be valid in the current context. We also used a range of arbitrary definitions of prevalent TB in the primary and sensitivity analyses, because the aim of our prognostic model was to assess the risk of incident TB, after prevalent TB has been clinically ruled out, to inform risk:benefit decisions regarding preventative treatment initiation. With increasing recognition of the continuum of M. tuberculosis infection using novel diagnostics (including incipient and/or subclinical phases)49, the distinction between prevalent and incident disease is becoming increasingly blurred. Future studies could consider integration of our prognostic model with next-generation biomarkers, such as blood transcriptional signatures for incipient TB50,51. A limitation of this study is that its generalizability is restricted to low transmission settings (annual incidence ≤20 per 100,000 persons). The rationale for limiting to such settings was, first, to examine progression from LTBI to TB disease more accurately by reducing risk of re-infection with M. tuberculosis during follow-up. Second, most of the population in high transmission settings are likely to have a positive LTBI test result, further undermining test specificity for progression to TB disease52. Because the quantitative LTBI test result is a strong predictor in our model, a different prediction model might, therefore, be required in such settings. For example, a recent study developing a prediction model for TB among close contacts in Peru found that the TST result added no value to the model39. Future studies could test our model for use in high transmission settings, updating the parameters as necessary, to extend its application to these settings. A second limitation of the current study is that model calibration was observed to be imperfect during external validation. However, conventional metrics (such as the calibration slope) might not be entirely appropriate in this context, which has a highly skewed distribution of predicted and observed risk, reflecting the rare occurrence of incident TB events. Reassuringly, in decision curve analysis, which accounts for both discrimination and calibration performance in quantifying net benefit, the model showed clinical utility35. Future studies might evaluate the full health economic effect of programmatic implementation of the model. A further limitation is that, owing to a lack of data from contributing studies, other potential predictors that might be associated with incident TB risk (including diabetes, malnutrition, fibrotic chest x-ray lesions and other immunosuppression)4 were not evaluated. These unmeasured covariates might have contributed to imperfect discrimination and calibration, along with residual heterogeneity in model performance between data sets. As additional studies are published, the prognostic model can be prospectively evaluated and updated as required. We also note that offer and acceptance of preventative treatment might be more likely among people at higher risk of TB. We, therefore, accounted for preventative treatment provision in the model by including it as a covariate along with our other predictors of interest, as widely recommended53. However, residual confounding by indication cannot be excluded in observational studies. In addition, the present model is not applicable for patients commencing biologic agents, because no data sets were identified that examined the natural history of LTBI in the context of biologic therapy, in the absence of preventative treatment for TB. A ‘hybrid’ modeling approach, with mathematical parameterization of relative risk for any given biologic agent, might be required to extend its application to these therapies. Because the quantitative LTBI test result is a strong predictor in our model, predictions might also be attenuated in the context of advanced immunosuppression7. Reassuringly, performance appeared adequate in a data set of immunocompromised individuals during validation29. In summary, we present a freely available and directly data-driven personalized risk predictor for incident TB (PERISKOPE-TB; peris-kope.org). This tool will allow a programmatic paradigm shift for TB prevention services in settings aiming for pre-elimination globally by facilitating shared decision-making between clinicians and patients for preventative treatment initiation. Methods Systematic review and pooling of individual participant data We conducted a systematic review and IPD-MA, in accordance with Preferred Reporting Items for a Systematic Review and Meta-analysis of Individual Participant Data standards43, to investigate the risk of progression to TB disease among people tested for LTBI in low transmission settings. The study is registered with PROSPERO (CRD42018115357). We searched Medline and Embase for studies published from January 1, 2002, to December 31, 2018, using comprehensive MeSH and keyword terms for ‘TB’, ‘IGRA’, ‘TST’, ‘latent TB’ and ‘predictive value’, without language restrictions. Longitudinal studies that primarily aimed to assess the risk of progression to TB disease among individuals tested for LTBI and that were conducted in a low TB transmission setting (defined as annual incidence ≤20 per 100,000 persons at the midpoint of the study) were eligible for inclusion. The full search strategy and eligibility criteria are provided in Supplementary Tables 8 and 9. Titles and abstracts underwent a first screen; relevant articles were selected for the second screen, which included full text review. Both first and second screens were performed by two independent reviewers, with disagreements resolved through discussion and arbitration by a third reviewer when required. Corresponding authors of eligible studies were invited to contribute IPD. Received data were mapped to a master variables list, and the integrity of the IPD was examined by comparing original reported results with re-analyzed results using contributed data. Quality assessment was performed using a modified version of the Newcastle-Ottawa Scale for cohort studies54. Definitions Participants entered the cohort on the day of LTBI screening or diagnosis and exited on the earliest of censor date (last date of follow-up), active TB diagnosis date, date of death or date of loss to follow-up (where available). LTBI was defined as any positive LTBI test (TST or commercial IGRA), using TST thresholds as defined by the contributing study (a 10-mm cutoff was used for studies that assessed multiple thresholds). Quantitative IGRA thresholds were calculated according to standard manufacturer guidelines. IGRAs included three generations of QuantiFERON TB assays (QuantiFERON Gold-In-Tube, QuantiFERON Gold and QuantiFERON-TB Gold Plus; Qiagen), which were assumed to be equivalent25, and T-SPOT.TB (Oxford Immunotec). Microbiologically confirmed and/or clinically diagnosed TB cases were included, as per contributing study definitions. In the absence of a widely accepted temporal distinction between prevalent and incident disease, prevalent TB at the time of screening was arbitrarily defined as a TB diagnosis within 42 d of enrolment; these cases were omitted from the primary analysis. Alternative shorter and longer temporal definitions were tested as sensitivity analyses. Participants with missing outcomes or durations of follow-up were considered lost to follow-up. ‘Preventative treatment’ was defined as any LTBI treatment regimen recommended by the World Health Organization52. All contributing studies included regimens consistent with this guidance; the effectiveness of each regimen was assumed to be equivalent55. Population-level analysis Survival analysis In a one-stage IPD-MA approach, we used flexible parametric survival models, with a random effect intercept by source study to account for between-study heterogeneity, to examine population-level risk of incident TB, stratified by LTBI screening result (positive versus negative) and provision of LTBI treatment (commenced versus not commenced). We further examined progression risk among untreated participants with LTBI, stratified by indication for screening (recent child contacts (<15 years) versus adult contacts versus migrants versus immunocompromised), by separately fitting random effect flexible parametric survival models to each risk group. Child contacts were further stratified by age (<5 years versus 5–14 years). Incidence rates We also calculated TB incidence rates (per 1,000 person-years) in a two-stage IPD-MA approach stratified by LTBI screening result, provision of LTBI treatment and indication for screening. Rates were calculated separately for the 0–2-year and 2–5-year follow-up intervals. Pooled incidence rate estimates for each risk group and follow-up interval were derived using random intercept Poisson regression models, without continuity correction for studies with zero events, in the meta package in R56. Prediction model analysis Variables of interest We then developed and validated a personalized prediction model for incident TB, in accordance with TRIPOD guidance44. For this analysis, we included studies that reported quantitative LTBI test results, proximity and infectiousness (based on sputum smear status) of index cases for contacts and country of birth and time since entry for migrants, because we considered these variables fundamental a priori. Using this subset of the data, we examined the availability of a range of variables of interest, specified a priori, in the contributing data sets to determine eligibility for inclusion as candidate predictors in the model. We determined that the following predictors were available from a sufficient number of data sets for further evaluation: age, gender, quantitative LTBI test result, previous BCG vaccination, recent contact (including proximity and infectiousness of index case), migration from a high TB incidence setting, time since migration, solid organ or hematological transplant receipt, HIV status and TB preventative treatment commencement. Variable transformations Previous data showed that quantitative TST, QuantiFERON Gold-in-Tube and T-SPOT.TB results are associated with risk of incident TB16. However, each LTBI test was reported using different scales, and it has hitherto been unclear whether quantitative values of each test are equivalent with respect to incident TB risk. To assess this further, we examined a subpopulation of the entire cohort where all three tests were performed among the same participants in head-to-head studies. We normalized quantitative results for the TST, QuantiFERON Gold-in-Tube and T-SPOT.TB to a percentile scale using this head-to-head population and examined the association between normalized result and risk of incident TB using Cox proportional hazards models with restricted cubic splines. Because TST cutoffs are frequently stratified by BCG vaccination and HIV status57,58, we also examined whether these variables modified the association between quantitative TST measurement and incident TB risk in the head-to-head subpopulation. Because there was no evidence that including interaction terms for either BCG or HIV improved model fit (based on Akaike Information Criteria (AIC)), we used unadjusted TST measurements. This analysis revealed that the normalized percentile results for each test (unadjusted TST, QuantiFERON Gold-in-Tube and T-SPOT.TB) appeared to be associated with similar risk of incident TB (Extended Data Fig. 8). The LTBI tests implemented differed between contributing studies. From this point, all LTBI test results were, therefore, normalized to this percentile scale to enable data harmonization across studies, by transforming raw quantitative results to the relevant percentile using look-up tables derived from the head-to-head population (Supplementary Table 10). Because most people evaluated for LTBI under routine programmatic conditions have a single test performed, we included only one test result per participant in the prediction model. We preferentially included tests where quantitative results were available. Where quantitative results were available for more than one test, we preferentially included the QuantiFERON result (because this was the most commonly used test in the data set), followed by T-SPOT.TB and then the TST. Recent contacts were categorized as either ‘smear positive and household’ or ‘other’ contacts, because there was no evidence of separation of risk among additional subgroups of the ‘other’ contacts stratum during exploratory univariable analyses (Extended Data Fig. 8). Because we considered migration from a high TB-burden country (defined as annual TB incidence ≥100 per 100,000 persons at the year of migration) to be a proxy for previous TB exposure, we included this in a composite ‘TB exposure’ variable, which included four mutually exclusive levels: household contact of smear-positive index case; ‘other’ contact; migrant from country with high TB incidence, without recent contact; and no exposure. There was no evidence of separation of incident TB risk when stratified by TB incidence in country of birth above the binary country of birth threshold (TB incidence ≥100 per 100,000 persons) among migrants or when stratified by country of birth among recent contacts (Extended Data Fig. 8). Age and normalized test result variables were modeled using restricted cubic splines (using a default of five knots placed at recommended intervals59) to account for their nonlinear associations with incident TB. Multiple imputation A data dictionary and a summary of missingness of candidate predictor variables are provided in Supplementary Table 11. We performed multi-level multiple imputation to account for sporadically and systematically missing data (assuming missingness at random48) while respecting clustering by source study, in accordance with recent guidance45, using the micemd package in R60. We used predictive mean matching for continuous variables owing to their skewed distributions. We included all variables (including transformations) assessed in the downstream prediction model in the imputation model, along with auxiliary variables, to ensure congeniality. Multi-level imputation was done separately for contacts and non-contacts owing to expected heterogeneity between these groups. We generated ten multiply imputed data sets, with 25 between-imputation iterations. Model convergence was assessed by visually examining plots of imputed parameters against iteration number. All downstream analyses were done in each of the ten imputed data sets; model coefficients and standard errors were combined using Rubin’s rules61. No imputation was done for participants missing binary LTBI test results or for those lost to follow-up; these individuals were excluded. For recent TB contacts or people screened owing to HIV infection with missing data on transplant status, this was assumed to be negative owing to the very low prevalence of transplant receipt when observed among these risk groups (<0.5%). Variable selection and final model development We performed backward selection of the nine candidate predictors in each of the pooled imputed data sets using AIC. Variables that were selected in more than 50% of the imputed data sets were included in the final model. T cell responses to M. tuberculosis might be impaired in the context of immunosuppression (including among people with HIV or transplant recipients)7. We, therefore, also tested whether there was a significant interaction between HIV or transplant and the normalized percentile test result variable, to assess whether the association between the quantitative test result and incident TB risk varied according to HIV or transplant status. This analysis showed no evidence of effect modification, based on AIC; thus, these interaction terms were not included in the final model. We used flexible parametric survival models to facilitate estimation of baseline risk throughout the duration of follow-up62 using the rstpm2 package63. We examined a range of degrees of freedom for the baseline hazard, using proportional hazards and odds scales, and selected the final model parameters based on the lowest AIC across the imputed data sets. Visual inspection of survival curves suggested non-proportional hazards for the composite exposure category; we, therefore, assessed whether including this variable as a time-varying covariate (by including an interaction between the composite exposure covariate of interest and time) improved model fit64. Because the AIC for the time-varying covariate model was lower across all imputed data sets, this time-varying covariate approach was used for the final model. IECV After development of the final model, we used the IECV framework for model validation, allowing concurrent assessment of between-study heterogeneity and generalizability34. In this process, one entire contributing study data set is iteratively discarded from the model training set and used for external validation. This process is repeated until each data set has been used once for validation. The primary outcome for validation was 2-year risk of incident TB. We included data sets with a minimum of five incident TB cases, and where participants had been included regardless of LTBI test result, as the primary validation sets. We assessed model discrimination using the C-statistic for 2-year TB risk. Model calibration was assessed by visually examining calibration plots of predicted risk versus Kaplan–Meier-estimated observed 2-year risk in quintiles and using the calibration slope and CITL statistics65. Calibration slopes greater than 1 suggest under-fitting (predictions are not varied enough), whereas slopes less than 1 indicate over-fitting (predictions are too extreme). Slopes were calculated by fitting survival models with the model linear predictor as the sole predictor; the calculated coefficient for the linear predictor provides the calibration slope. CITL indicates whether predictions are systematically too low (CITL> 0) or too high (CITL < 0). We calculated CITL for each validation set by fixing all model coefficients from model development (including the baseline hazard terms) and re-estimating the intercept. The difference between the development model and recalculated validation model intercepts provided the CITL statistic66. Pooling of IECV parameters and random effects meta-analysis IECV was performed on each imputed data set. Validation set C-statistics, calibration slopes and CITL metrics were pooled for each study across imputations using Rubin’s rules61. We then meta-analyzed these metrics across validation studies with random effects, using logit-transformed C-statistics as previously recommended67, to derive pooled discrimination and calibration estimates. The IECV validation sets were also pooled, with averaging of the predicted 2-year risk of TB for each individual in the validation sets across imputations, for downstream decision curve analyses as described below. Decision curve analysis Decision curve analysis complements model validation parameters by assessing the potential clinical utility of a prediction model35,36. Net benefit quantifies the proportion of true-positive cases detected minus the proportion of false positives, with weighting of each by the ‘threshold probability’35. The ‘threshold probability’ reflects both the risk:benefit ratio of initiating preventative treatment and the percentage cut-point for the prediction model, above which treatment is recommended. We calculated net benefit across a range of clinically relevant threshold probabilities (to account for a range of clinician and patient preferences) in comparison to the default strategies of treating either all or no patients with a positive LTBI test. We analyzed net benefit using the stdca command from the ddsjoberg/dca package in R68, using the stacked validation sets of untreated participants with positive LTBI tests from IECV (to ensure that each individual for whom a prediction was generated had not been included in the model training set used to derive that prediction). Sensitivity analyses First, we re-examined population-level TB risk without exclusion of prevalent TB cases. Second, we recalculated prediction model parameters using alternative definitions of prevalent TB (ranging from diagnosis within 0–180 d of recruitment); a complete case approach (for all variables except for HIV status, which was assumed to be negative where this was missing); and exclusion of participants who received preventative treatment. Parameters for each of these models were compared with the primary model (without time-varying covariates to facilitate interpretation). We also examined IECV discrimination parameters for validation data sets when 1) restricted to participants with positive binary LTBI tests; 2) excluding those who received preventative treatment; and 3) imputing an average quantitative positive or negative LTBI test result (based on the medians among the study population), according to the binary result. The latter analysis was done to assess model performance in situations where the quantitative test result was not available. Ethics This study involved analyses of fully de-personalized data from previously published cohort studies, with data pooling via a safe haven. Ethical approvals for sharing of data were sought and obtained by contributors of individual participant data, where required. Extended Data Extended Data Fig. 1 Flow chart outlining systematic review process. The systematic search strategy and eligibility criteria are shown in Supplementary Tables 8 and 9. Extended Data Fig. 2 Flow chart showing inclusion of participants in the population-level and prediction modelling analyses. The systematic search strategy and eligibility criteria are shown in Supplementary Tables 8 and 9. Extended Data Fig. 3 Cumulative risk of prevalent and incident tuberculosis during follow-up. Risk is stratified by binary latent TB test result, provision of preventative treatment, and indication for screening among participants with untreated latent infection (total n = 80,468 participants). Cumulative risk is estimated using flexible parametric survival models with random effects for the intercept by source study, separately fitted to each risk group. Prevalent TB cases (diagnosed within 42 days of recruitment) are included in this sensitivity analysis. Each plot is presented as point estimates (solid line) and 95% confidence intervals (shaded area). PT = preventative treatment. Extended Data Fig. 4 Pooled TB incidence rates among adults, stratified by risk group. Pooled incidence rates are shown on log10 scale among participants with: latent TB infection (LTBI) with no preventative therapy (PT); LTBI commencing PT; and without evidence of LTBI. Rates are further stratified by follow-up interval (0–2 years vs. 2–5 years) and indication for screening (total n = 52,576 participants). Pooled incidence rate estimates were derived from random intercept Poisson regression models, without continuity correction for studies with zero events. Numeric results are shown for the subgroups with untreated latent TB infection in the forest plots in Extended Data Fig. 5. Plots show point estimates (filled circles) and 95% confidence intervals (vertical error bars). No pooled estimate could be calculated for child contacts without evidence of LTBI for the 2–5 year interval since there were no incident events. Extended Data Fig. 5 Forest plots showing incidence rates by source study among participants with untreated LTBI. Forest plots are stratified by follow-up interval (0–2 years vs. 2–5 years) and indication for screening (total n =52,576 participants). Pooled incidence rate estimates (shown as diamonds) were derived from random intercept Poisson regression models, without continuity correction for studies with zero events. Incidence rates per study are shown with a continuity correction of 0.5 for studies with zero events. Plots show study-level point estimates (grey squares) and 95% confidence intervals (CIs; horizontal error bars). Extended Data Fig. 6 Calibration plots from internal-external validation of prediction model, stratified by validation study. Data from nine primary validation studies are shown, from internal-external cross-validation of the model (developed among n = 31,090 participants; validated among 25,504 in this analysis). X-axis shows predicted risk, in quintiles, with corresponding Kaplan Meier 2-year risk of incident TB on the Y-axis (95% confidence intervals are shown by vertical error bars). Extended Data Fig. 7 Model validation sensitivity analyses. Forest plots showing recalculation of the C-statistics from internal-external cross validation, limiting validation sets to a, participants who did not receive preventative therapy (n = 23,060 participants); b, participants with a positive LTBI test (n = 9,063 participants); and c, binary LTBI test results (using an average quantitative positive or negative LTBI test result as appropriate, based on the medians among the study population; n = 25,504 participants). ‘TB’ column indicates number of incident TB cases within 2 years of study entry and ‘N’ indicates total participants per study included in analysis. Each forest plot shows point estimates (squares) and 95% confidence intervals (error bars). Pooled estimates are shown as diamonds. Panel d, shows decision curve analyses (n = 6,418 participants) when using the prediction model using a binary LTBI test result, compared to the full prediction model, ‘treat all’ and ‘treat none’ strategies across a range of threshold probabilities (x-axis). Net benefit appeared higher for the binary model than either the strategies of treating all patients with evidence of LTBI, or no patients, throughout the range of threshold probabilities. The full model had highest net benefit across most threshold probabilities. Extended Data Fig. 8 Data supporting assumptions underlying PERISKOPE-TB model. a, Quantitative results for the tuberculin skin test (TST), QuantiFERON Gold-in-tube (QFT-GIT) and T-SPOT.TB are normalised to a percentile scale using a head-to-head population among whom all three tests were performed from 3 studies including recent TB contacts, migrants and immunocompromised participants (n = 8,335; 158 TB cases). We examined the association between normalised test result and risk of incident TB using Cox proportional hazards models with restricted cubic splines. Normalised results for each test appeared to be associated with similar risk of incident TB. b, Kaplan Meier plots from pooled dataset showing cumulative risk of incident TB, stratified by proximity and infectiousness of index cases among contacts (n = 22,231 participants). There was no evidence of separation of risk of additional subgroups of the ‘other’ (non-smear positive household) contacts stratum. PTB = pulmonary TB; EPTB = extra-pulmonary TB. c, Kaplan Meier plots from pooled dataset showing cumulative risk of incident TB among people with positive latent TB tests, stratified by TB incidence in country of birth among migrants from high TB burden countries (n = 1,031 participants). P value represents Log-rank test. d, Kaplan Meier plots from pooled dataset showing cumulative risk of incident TB among people with positive latent TB tests, stratified by country of birth among recent contacts (n = 5,917 participants). P value represents Log-rank test. Supplementary Material Supplementary Material Reporting Summary Acknowledgements This study was funded by the National Institute for Health Research (NIHR) (DRF-2018–11-ST2-004 to R.K.G. and SRF-2011-04-001 and NF-SI-0616-10037 to I.A.), the Wellcome Trust (207511/Z/17/Z to M.N.) and NIHR biomedical research funding to University College London Hospitals. C.L. is funded by the German Center for Infection Research. J.S.D. receives salary support from the National Health and Medical Research Council (Australia). This paper presents independent research supported by the NIHR. The views expressed are those of the authors and not necessarily those of the National Health Service, the NIHR or the Department of Health and Social Care. The study funders had no role in the conceptualization, design, data collection, analysis, decision to publish or preparation of the manuscript. The authors would like to thank all of the research teams involved in the primary studies that contributed data for this analysis. Data availability The individual participant data pooled for this analysis are subject to data sharing agreements with the original study authors. The data might be shared with interested parties by the corresponding authors of the original studies, subject to data sharing agreements. Code availability The final prognostic model developed in this study has been made freely available to enable immediate implementation in clinical practice and independent external validation in new data sets (periskope.org). The code underlying the prediction tool is available at github.com/rishi-k-gupta/PERISKOPE-TB. Fig. 1 Population-level cumulative risk of incident TB during follow-up. Risk is stratified by binary latent TB test result, provision of preventative treatment (PT) and indication for screening among participants with untreated latent infection (total n=80,468 participants). Cumulative risk is estimated using flexible parametric survival models with random effects intercepts by source study, separately fitted to each risk group. Prevalent TB cases (diagnosed within 42 d of recruitment) are excluded. Each plot is presented as point estimates (solid line) and 95% CIs (shaded area). Child contacts are shown stratified by age (<5 years and 5-14 years). PT = preventative treatment. Numbers of participants, TB cases and numeric cumulative risk estimates for each plot are presented in Supplementary Table 5. Cumulative TB risk, including prevalent TB cases, is presented in Extended Data Fig. 3. Fig. 2 Visual representations of associations between predictors and incident TB. Illustrative estimates are shown for a 33-year-old migrant from a high TB-burden setting. The example ‘base case’ patient does not commence preventative treatment, is not living with HIV, has not received a previous transplant and has an ‘average’ positive latent TB test. We vary one of these predictors in each plot ((a) age; (b) normalized latent TB test result; (c) years since migration; (d) exposure to M. tuberculosis; (e) HIV status; (f) transplant receipt; and (g) preventative treatment). Each plot is presented as point estimates (solid line) and 95% CIs (shaded area). The model was trained on a pooled data set (n = 31,090 participants). Model parameters are provided in Supplementary Table 6. ‘Household smear + contact’ = household contact of sputum smear-positive index case; ‘Other contact’ = contact of non-household or smear-negative index case; ‘Migrant’ = migrant from high TB incidence country, without recent contact. Fig. 3 Forest plots showing model discrimination and calibration metrics for predicting 2-year risk of incident TB. Discrimination is presented as the C-statistic; calibration is presented as CITL and the calibration slope. Data from nine primary validation studies are shown, from IECV of the model (developed among n = 31,090 participants; validated among 25,504 participants in this analysis). ‘TB’ column indicates number of incident TB cases within 2 years of study entry, and ‘n’ indicates total participants per study included in analysis. Each forest plot shows point estimates (squares) and 95% CIs (error bars). Pooled estimates are shown as diamonds. Calibration slopes greater than 1 suggest under-fitting (predictions are not varied enough), whereas slopes less than 1 indicate over-fitting (predictions are too extreme). CITL indicates whether predictions are systematically too low (CITL>O) or too high (CITL<O). Dashed lines indicate line of no discrimination (C-statistic) and perfect calibration (CITL and slope), respectively. Fig. 4 Distribution of predictions and risk of incident TB in four quartiles of risk for people with positive latent TB tests. Distribution of risk from prediction model using pooled validation sets of people not receiving preventative therapy from IECV of the model (n = 27,511 participants), stratified by (a) binary latent TB test result and (b) indication for screening among untreated people with positive LTBI tests. c, Kaplan-Meier plots for quartile risk groups (1 = lowest risk) of untreated individuals with positive LTBI tests (n = 6,418 participants). Quartiles represent four equally sized groups based on predicted risk of incident TB, from the pooled validation sets derived from IECV of the prediction model. P value represents log-rank test (P = 1.137 × 10-40). d, Randomly sampled individual patients from each risk quartile. Patient 1 is a 22-year-old with no TB exposure and a normalized latent TB test result on the 68th percentile; Patient 2 is a 41-year-old migrant from a high TB-burden country (3.8 years since migration) with normalized latent TB test result on the 80th percentile; Patient 3 is a 51-year-old household contact of a smear-positive index TB case with a normalized latent TB test result on the 79th percentile; and Patient 4 is a 33-year-old household contact of a smear-positive index TB case with a normalized latent TB test result on the 94th percentile. All four example patients are HIV negative and are not transplant recipients. Equivalent values of normalized percentile test results for QuantiFERON, T-SPOT.TB and TST are shown in Supplementary Table 10. Plots (c, d) are presented as point estimates (solid line) and 95% CIs (shaded area). Fig. 5 Decision curve analysis. Shown as net benefit of the prediction model among untreated participants from the pooled validation sets with positive binary latent TB tests (n = 6,418 participants) compared to ‘treat all’ and ‘treat none’ strategies across a range of threshold probabilities (x axis). Net benefit quantifies the tradeoff between correctly identifying true-positive progressors to incident TB and incorrectly detecting false positives, with weighting of each by the threshold probability35. The threshold probability corresponds to a measure of both the perceived risk:benefit ratio of initiating preventative treatment and the percentage cutoff for the prediction model, above which treatment is recommended. Net benefit appeared higher than either the strategies of treating all patients with evidence of LTBI or no patients, throughout the range of threshold probabilities, suggesting clinical utility. For illustration, a patient who is very concerned about developing TB disease but not concerned regarding side effects of preventative treatment might have a low threshold probability (for example, 1%, which is equivalent to a risk:benefit ratio of 1:99—that is, the outcome of developing TB is considered to be 99 times worse than taking unnecessary preventative treatment). In contrast, a patient who is less concerned about developing TB but is very concerned about side effects of preventative treatment might have a higher threshold probability (for example, 10%, which is equivalent to a risk:benefit ratio of 1:9). The unit of net benefit is ‘true positives’35. For instance, a net benefit of 0.01 would be equivalent to a strategy where one patient per 100 tested was appropriately given preventative treatment, as they would otherwise have progressed to incident TB if left untreated. Table 1 Characteristics of contributing studies included in individual participant data meta-analysis Authors Publication Year Country n (total) Adults/children Population Follow-up years (median (IQR)) TB cases Loss to follow-up Included in prediction modeling NOSa Abubakar et al.9 2018 UK 10,045 Adults Contacts & migrants 4.7 (3.7–5.5) 147 10 (0.1%) Yes 7/7 Aichelburg et al.26 2009 Austria 830 Adults People with HIV 1.2 (0.7–1.4) 11 25 (3%) Yes 7/7 Altet et al.17 2015 Spain 1,339 Adults & children Contacts 4(4–4) 95 0 (0%) Yes 7/7 Diel et al.18 2011 Germany 1,414 Adults & children Contacts 3.5 (2.5–4.2) 19 381 (26.9%) Yes 7/7 Dobler & Marks19 2013 Australia 12,212 Adults & children Contacts 4.2 (2–6.9) 94 351 (2.9%) Nob 7/7 Doyle et al.27 2014 Australia 919 Adults People with HIV 2.9 (1.7–3.6) 2 47 (5.1%) Yes 7/7 Erkens et al.32 2016 Netherlands 14,241 Adults & children Mixed population screening 5.5 (3–7.4) 134 NA Nob 6/6 Geis et al.20 2013 Germany 1,283 Adults & children Contacts 0.8 (0.4–1.1) 33 62 (4.8%) Yes 6/6 Gupta et al.25 2020 UK 623 Adults Contacts 1.9 (1.6–2.2) 13 0 (0%) Yes 7/7 Haldar et al.21 2013 UK 1,411 Adults & children Contacts 1.9 (1.3–2.4) 37 30 (2.1%) Yes 7/7 Lange et al.28 2012 Germany 456 Adults Immunocompromised 2.8 (2–3.1) 1 42 (9.2%) Yes 7/7 Munoz et al.30 2015 Spain 76 Adults Transplant recipients 4.3 (3.6–4.8) 2 0 (0%) Yes 7/7 Roth et al.31 2017 Canada 22,949 Adults & children Mixed population screening 3 (1.8–4.3) 58 NA Subsetb 6/6 Sester et al.29 2014 Multiple European countries 1,464 Adults Immunocompromised 2.7 (1.5–3.5) 11 7 (0.5%) Yes 7/7 Sloot et al.22 2014 Netherlands 5,895 Adults & children Contacts 5.9 (3.6–7.7) 81 NA Yes 7/7 Yoshiyama et al.23 2015 Japan 625 Adults & children Contacts 1.8 (1.4–2) 12 0 (0%) Yes 6/7 Zellweger et al.24 2015 Multiple European countries 5,237 Adults & children Contacts 2.6 (1.9–3.5) 55 1339 (25.6%) Yes 7/7 Zenner et al.33 2017 UK 1,341 Adults Migrants 3.7 (3–4.8) 21 NA Nob 7/7 Total 82,360 3.7 (2.1–5.3) 826 2294 (2.8%) a Modified version of the Newcastle-Ottawa Scale for cohort studies. b Not included in prediction modeling owing to lack of data on proximity or infectiousness of index cases19 or absent quantitative LTBI test data32,33. A subset of the data set was included in the prediction model for the Roth et al. study31; contacts and migrants were excluded owing to no data being available on country of birth or infectiousness of index cases, respectively. Additional study characteristics are shown in Supplementary Table 1. Author contributions R.K.G. and I.A. conceived of the study and led the pooling of data. R.K.G., M.X.R., A.C, M.L., M.N. and I.A. wrote the study protocol and developed the analysis plan. R.K.G. conducted the analyses and wrote the first draft of the manuscript. R.K.G., C.J.C. and M.K. performed the systematic literature review. M.Q. and A.C. provided statistical and multiple imputation expertise. A.Y. and R.K.G. developed the website interface for the risk predictor tool. M.C.A., N.A., R.D., C.C.D., J.D., J.S.D., C.E., S.G., P.H., A.M.H., T.H., J.C.J., C.L.,B.L., F.v.L., L.M., C.R., K.R., D.R., M.S., R.S., G.S., G.W., T.Y., J.-P.Z. and D.Z. contributed primary data and assisted with interpretation. R.W.A contributed to data interpretation. All authors critically reviewed and approved the manuscript before submission. Competing interests J.S.D.’s institution receives investigator-initiated research grants and consultancy income from Gilead Sciences, AbbVie, Bristol Myers Squibb and Merck. The Burnet Institute receives funding from the Victorian Government Operational Infrastructure Fund. C.L. reports honoraria from Chiesi, Gilead, Insmed, Janssen, Lucane, Novartis, Oxoid, Berlin Chemie (for participation at sponsored symposia) and Oxford Immunotec (to attend a scientific advisory board meeting), all outside of the submitted work. M.S. reports receipt of test kits free of charge from Qiagen and from Oxford Immunotec for investigator-initiated research projects. I.A. reports receiving test kits free of charge from Qiagen for an investigator-initiated research project25. C.E. reports receiving test kits free of charge from Qiagen for investigator-initiated research projects outside of the submitted work. The authors declare no other conflicts of interest. Peer review information Alison Farrell is the primary editor on this article and managed its editorial process and peer review in collaboration with the rest of the editorial team. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article. Online content Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41591-020-1076-0. 1 World Health Organization Global Tuberculosis Report 2019 https://www.who.int/tb/publications/global_report/en/ 2 World Health Organization The End TB Strategy 2015 http://www.who.int/tb/strategy/End_TB_Strategy.pdf?ua=1 3 Getahun H Management of latent Mycobacterium tuberculosis infection: WHO guidelines for low tuberculosis burden countries Eur Respir J 2015 46 1563 1576 26405286 4 Getahun H Matteelli A Chaisson RE Raviglione M Latent Mycobacterium tuberculosis infection N Engl J Med 2015 372 2127 2135 26017823 5 Mack U LTBI: latent tuberculosis infection or lasting immune responses to M. tuberculosis? A TBNET consensus statement Eur Respir J 2009 33 956 973 19407047 6 Sterling TR Guidelines for the treatment of latent tuberculosis infection: recommendations from the National Tuberculosis Controllers Association and CDC, 2020 MMWR Recomm Rep 2020 69 1 11 7 Pai M Gamma interferon release assays for detection of Mycobacterium tuberculosis infection Clin Microbiol Rev 2014 27 3 20 24396134 8 Rangaka MX Predictive value of interferon-γ release assays for incident active tuberculosis: a systematic review and meta-analysis Lancet Infect Dis 2012 12 45 55 21846592 9 Abubakar I Prognostic value of interferon-γ release assays and tuberculin skin test in predicting the development of active tuberculosis (UK PREDICT TB): a prospective cohort study Lancet Infect Dis 2018 18 1077 1087 30174209 10 Gao J Knowledge and perceptions of latent tuberculosis infection among Chinese immigrants in a Canadian urban centre Int J Fam Med 2015 2015 546042 11 Alsdurf H Hill PC Matteelli A Getahun H Menzies D The cascade of care in diagnosis and treatment of latent tuberculosis infection: a systematic review and meta-analysis Lancet Infect Dis 2016 16 1269 1278 27522233 12 Campbell JR Winters N Menzies D Absolute risk of tuberculosis among untreated populations with a positive tuberculin skin test or interferon-γ release assay result: systematic review and meta-analysis BMJ 2020 368 m549 32156698 13 Behr MA Edelstein PH Ramakrishnan L Revisiting the timetable of tuberculosis BMJ 2018 362 k2738 30139910 14 Winje BA Stratification by interferon-γ release assay level predicts risk of incident TB Thorax 2018 73 652 661 15 Andrews JR Serial QuantiFERON testing and tuberculosis disease risk among young children: an observational cohort study Lancet Respir Med 2017 5 282 290 28215501 16 Gupta RK Quantitative IFN-γ release assay and tuberculin skin test results to predict incident tuberculosis: a prospective cohort study Am J Respir Crit Care Med 2020 201 984 991 31825645 17 Altet N Predicting the development of tuberculosis with the tuberculin skin test and QuantiFERON testing Ann Am Thorac Soc 2015 12 680 688 25699406 18 Diel R Loddenkemper R Niemann S Meywald-Walter K Nienhaus A Negative and positive predictive value of a whole-blood interferon-γ release assay for developing active tuberculosis: an update Am J Respir Crit Care Med 2011 183 88 95 20802162 19 Dobler CC Marks GB Risk of tuberculosis among contacts in a low-incidence setting Eur Respir J 2013 41 1459 1461 23728408 20 Geis S How can we achieve better prevention of progression to tuberculosis among contacts? Eur Respir J 2013 42 1743 1746 23900989 21 Haldar P Single-step QuantiFERON screening of adult contacts: a prospective cohort study of tuberculosis risk Thorax 2013 68 240 246 22956558 22 Sloot R Van Der Loeff MFS Kouw PM Borgdorff MW Risk of tuberculosis after recent exposure: a 10-year follow-up study of contacts in Amsterdam Am J Respir Crit Care Med 2014 190 1044 1052 25265362 23 Yoshiyama T Harada N Higuchi K Saitou M Kato S Use of the QuantiFERON-TB Gold in Tube test for screening TB contacts and predictive value for active TB Infect Dis 2015 47 542 549 24 Zellweger J-P Risk assessment of tuberculosis in contacts by IFN-γ release assays. A Tuberculosis Network European Trials Group study Am J Respir Crit Care Med 2015 191 1176 1184 25763458 25 Gupta RK Evaluation of QuantiFERON-TB Gold Plus for predicting incident tuberculosis among recent contacts: a prospective cohort study Ann Am Thorac Soc 2020 17 646 650 32083944 26 Aichelburg MC Detection and prediction of active tuberculosis disease by a whole-blood interferon-γ release assay in HIV-1-infected individuals Clin Infect Dis 2009 48 954 962 19245343 27 Doyle JS Latent tuberculosis screening using interferon-γ release assays in an Australian HIV-infected cohort: is routine testing worthwhile? J Acquir Immune Defic Syndr 2014 66 48 54 24457631 28 Lange B Vavra M Kern WV Wagner D Development of tuberculosis in immunocompromised patients with a positive tuberculosis-specific IGRA Int J Tuberc Lung Dis 2012 16 492 495 22325468 29 Sester M Risk assessment of tuberculosis in immunocompromised patients. A TBNET study Am J Respir Crit Care Med 2014 190 1168 1176 25303140 30 Munoz L Immunodiagnostic tests’ predictive values for progression to tuberculosis in transplant recipients: a prospective cohort study Transplant Direct 2015 1 e12 27500217 31 Roth DZ Impact of interferon-γ release assay on the latent tuberculosis cascade of care: a population-based study Eur Respir J 2017 49 1601546 28331032 32 Erkens CGM Risk of developing tuberculosis disease among persons diagnosed with latent tuberculosis infection in the Netherlands Eur Respir J 2016 48 1420 1428 27799389 33 Zenner D Loutet MG Harris R Wilson S Ormerod LP Evaluating 17 years of latent tuberculosis infection screening in north-west England: a retrospective cohort study of reactivation Eur Respir J 2017 50 1602505 28751410 34 Debray TPA Individual participant data (IPD) meta-analyses of diagnostic and prognostic modeling studies: guidance on their use PLoS Med 2015 12 e1001886 26461078 35 Vickers AJ van Calster B Steyerberg EW A simple, step-by-step guide to interpreting decision curve analysis Diagn Progn Res 2019 3 18 31592444 36 Zhang Z Decision curve analysis: a technical note Ann Transl Med 2018 6 19 29404365 37 Steyerberg EW Harrell FE Prediction models need appropriate internal, internal-external, and external validation J Clin Epidemiol 2016 69 245 247 25981519 38 Martinez L The risk of tuberculosis in children after close exposure: a systematic review and individual-participant meta-analysis Lancet 2020 395 973 984 32199484 39 Saunders MJ A score to predict and stratify risk of tuberculosis in adult contacts of tuberculosis index cases: a prospective derivation and external validation cohort study Lancet Infect Dis 2017 17 1190 1199 28827142 40 Saunders MJ A household-level score to predict the risk of tuberculosis among contacts of patients with tuberculosis: a derivation and external validation prospective cohort study Lancet Infect Dis 2020 20 110 122 31678031 41 Li R Two clinical prediction tools to improve tuberculosis contact investigation Clin Infect Dis 2020 10.1093/cid/ciz1221 42 Menzies D Gardiner G Farhat M Greenaway C Pai M Thinking in three dimensions: a web-based algorithm to aid the interpretation of tuberculin skin test results Int J Tuberc Lung Dis 2008 12 498 505 18419884 43 Stewart LA Preferred reporting items for a systematic review and meta-analysis of individual participant data JAMA 2015 313 1657 25919529 44 Collins GS Reitsma JB Altman DG Moons KGM Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement BMJ 2015 350 g7594 25569120 45 Audigier V Multiple imputation for multilevel data with continuous and binary variables Stat Sci 2018 33 160 183 46 Nemes E Prevention of M. Tuberculosis infection with H4:IC31 vaccine or BCG revaccination N Engl J Med 2018 379 138 149 29996082 47 Katelaris AL Effectiveness of BCG vaccination against Mycobacterium tuberculosis infection in adults: a cross-sectional analysis of a UK-based cohort J Infect Dis 2020 221 146 155 31504674 48 White IR Royston P Wood AM Multiple imputation using chained equations: issues and guidance for practice Stat Med 2011 30 377 399 21225900 49 Drain PK Incipient and subclinical tuberculosis: clinical review of early stages and progression of infection Clin Microbiol Rev 2018 31 e00021-18 50 Gupta RK Concise whole blood transcriptional signatures for incipient tuberculosis: a systematic review and patient-level pooled meta-analysis Lancet Respir Med 2020 8 395 406 31958400 51 Roe J Blood transcriptomic stratification of short-term risk in contacts of tuberculosis Clin Infect Dis 2019 70 731 737 52 World Health Organization Latent TB Infection: updated and consolidated guidelines for programmatic management 2018 http://www.who.int/tb/publications/2018/latent-tuberculosis-infection/en/ 53 Groenwold RHH Explicit inclusion of treatment in prognostic modeling was recommended in observational and randomized settings J Clin Epidemiol 2016 78 90 100 27045189 54 Wells G The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp 55 Zenner D Treatment of latent tuberculosis infection Ann Intern Med 2017 167 248 28761946 56 Balduzzi S Rücker G Schwarzer G How to perform a meta-analysis with R: a practical tutorial Evid Based Ment Health 2019 22 153 160 31563865 57 Seddon JA The impact of BCG vaccination on tuberculin skin test responses in children is age dependent: evidence to be considered when screening children for tuberculosis infection Thorax 2016 71 932 939 27335104 58 Cobelens FG Tuberculin skin testing in patients with HIV infection: limited benefit of reduced cutoff values Clin Infect Dis 2006 43 634 639 16886159 59 Harrell FE Biostatistical Modeling 2004 http://biostat.mc.vanderbilt.edu/wiki/pub/Main/BioMod/notes.pdf 60 Audigier V Resche-Rigon M micemd: multiple imputation by chained equations with multilevel data R package, version 1.6.0 2019 61 Rubin DB Multiple Imputation for Nonresponse In Surveys Wiley-Interscience 2004 62 Royston P Parmar MKB Flexible parametric proportional-hazards and proportional-odds models for censored survival data, with application to prognostic modelling and estimation of treatment effects Stat Med 2002 21 2175 2197 12210632 63 Clements M Liu X-R rstpm2: smooth survival models, including generalized survival models R package, version 1.5.1 2019 64 Bower H Capturing simple and complex time-dependent effects using flexible parametric survival models: a simulation study Commun Stat Simul Comput 2019 10.1080/03610918.2019.1634201 65 Steyerberg EW Vergouwe Y Towards better clinical prediction models: seven steps for development and an ABCD for validation Eur Heart J 2014 35 1925 1931 24898551 66 Westeneng H-J Prognosis for patients with amyotrophic lateral sclerosis: development and validation of a personalised prediction model Lancet Neurol 2018 17 423 433 29598923 67 Snell KI Ensor J Debray TP Moons KG Riley RD Meta-analysis of prediction model performance across multiple studies: which scale helps ensure between-study normality for the C-statistic and calibration measures? Stat Methods Med Res 2018 27 3505 3522 28480827 68 Sjoberg DD dca: decision curve analysis R package, version 0.1.0.9000 2020

PMC007xxxxxx/PMC7614811.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7909766 J Epidemiol Community Health J Epidemiol Community Health Journal of epidemiology and community health 0143-005X 1470-2738 32385128 7614811 10.1136/jech-2019-213412 EMS181447 Article Longitudinal associations between neighbourhood trust, social support and physical activity in adolescents: evidence from the Olympic Regeneration in East London (ORiEL) study Berger Nicolas Population Health Innovation Lab, Department of Public Health, Environments and Society, London School of Hygiene & Tropical Medicine, London, United Kingdom Lewis Daniel Daniel.Lewis@lshtm.ac.uk Population Health Innovation Lab, Department of Public Health, Environments and Society, London School of Hygiene & Tropical Medicine, London, United Kingdom Data Science Campus, Office for National Statistics, London, United Kingdom Quartagno Matteo m.quartagno@ucl.ac.uk MRC Clinical Trials Unit, University College London, London, United Kingdom, Department of Medical Statistics, London School of Hygiene & Tropical Medicine, London, United Kingdom Njagi Edmund Njeru Edmund.Njeru.Njagi@lshtm.ac.uk Department of Non-communicable Disease Epidemiology, London School of Hygiene & Tropical Medicine, London, United Kingdom Cummins Steven Steven.Cummins@lshtm.ac.uk Population Health Innovation Lab, Department of Public Health, Environments and Society, London School of Hygiene & Tropical Medicine, London, United Kingdom Corresponding author: Nicolas Berger, Department of Public Health, Environments and Society, London School of Hygiene & Tropical Medicine, Nicolas.Berger@lshtm.ac.uk, 15-17 Tavistock Place, London WC1H 9SH, UK 01 9 2020 08 5 2020 19 7 2023 26 7 2023 74 9 710718 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Background Most UK adolescents do not achieve recommended levels of physical activity (PA). Previous studies suggest that the social environment could contribute to inequalities in PA behaviours, but longitudinal evidence is limited. We examined whether neighbourhood trust and social support were longitudinally associated with four common forms of PA: walking to school, walking for leisure, outdoor PA, and pay and play PA. We further assessed whether gender moderated these associations. Methods We used longitudinal data from the Olympic Regeneration in East London (ORiEL). In 2012, 3,106 adolescents aged 11-12 were enrolled from 25 schools in four deprived boroughs of East London, UK. Adolescents were followed-up in 2013 and 2014. The final sample includes 2,664 participants interviewed at waves 2 and 3. We estimated logistic regression models using Generalised Estimating Equations (GEE) (pooled models) and proportional odds models (models of change) to assess associations between the social environment exposures and the PA outcomes, adjusting for potential confounders. Item non-response was handled using multi-level multiple imputation. Results We found that different aspects of the social environment predict different types of PA. Neighbourhood trust was positively associated with leisure-type PA. Social support from friends and family was positively associated with walking for leisure. There was some evidence that changes in exposures led to changes in the PA outcomes. Associations did not systematically differ by gender. Conclusion These results confirm the importance of the social environment to predict PA and its change over time in a deprived and ethnically diverse adolescent population. adolescents East London social environment trust social support PA walking social capital social cohesion pmcIntroduction As physical activity (PA) declines during childhood and adolescence [1], the majority of adolescents do not achieve the recommended level of PA in the UK [2]. Increasing and maintaining PA in this group is crucial because adolescence is an important period in the lifecourse during which life-long health behaviours start forming [3]. Among the many potential multi-level determinants of PA, features of the social environment have received particular scrutiny over recent years [4]. The social environment is defined as ‘the immediate physical surroundings, social relationships, and cultural milieus within which defined groups of people function and interact’ [5]. It encompasses a range of social constructs, among which social capital and social support are prominent tools to understand how the social context affects health [6]. Social capital is defined as the social resources that are accessed by individuals through their membership to a group or a network, including trust, norms of reciprocity and ability to undertake collective action [4,7]. Social capital is hypothesised to affect health behaviours through three primary mechanisms: social contagion, collective efficacy and informal social control [4]. The evidence to date in the UK and elsewhere has shown that aspects of social capital – including social cohesion and neighbourhood trust – were positively associated with total and recreational PA in adults and adolescents [8–14]. Social support describes resources provided from interpersonal relationships that can influence behaviour such as PA. These resources are diverse and include: psychological/emotional support (e.g. encouragement, praise), instrumental support (e.g. equipment, transport to a PA facility), co-participation (e.g. performing an activity with an adolescent), informational support (e.g. providing advice or instructions about an activity), and support as a role model [15]. Parents, family members and friends constitute the main sources of social support for PA in adolescents [16]. The growing literature on the benefits of social support for health behaviours has identified social support as one of the most consistent correlates of PA in young people. During adolescence, transportation, encouragement and role modelling are important types of social support provided by parents, while friends’ encouragement and co-participation in activities are the most salient resources of social support provided by friends [16–20]. However, longitudinal evidence on whether changes in social capital and social support are associated with changes in PA is still scarce, in particular in young people and outside of the USA [19,21,22]. The few available longitudinal studies are generally consistent with cross-sectional results, and found positive associations between baseline/change in the social environment and change in PA. Much of the literature, especially on social capital, captures total PA or leisure-based PA and does not explore whether a specific aspect of social context could differentially affect a range of forms of PA, such as outside play, structured activities or walking to school. There is also limited evidence as to whether the positive associations observed for the general population are consistent among ethnic minority and deprived populations. In this paper, we use data from the Olympic Regeneration in East London (ORiEL) study to explore how neighbourhood trust and social support from family, friends and significant others are longitudinally associated with four common forms of PA in a relatively deprived and ethnically diverse adolescent population. As Olympic-related regeneration accelerated ongoing transformations of East London, changes in the social environment are expected to have occurred before and during the ORiEL study period [23]. The ORiEL study therefore allows testing of hypotheses on 1) general associations between the social environment and PA; and 2) how short term changes in the social environment could immediately affect PA [17]. The PA outcomes we consider are walking to school, walking for leisure, outdoor PA and a composite measure of ‘pay and play’ PA. We additionally examine whether gender moderated these associations. Methods Study design and participants We used data from the ORiEL study, a prospective cohort study that aimed to assess the effect of urban regeneration following the London 2012 Olympic and Paralympic Games on health. Participants were enrolled from 25 schools in four boroughs of London: Hackney, Newham, Barking and Dagenham, and Tower Hamlets. These boroughs are characterised by the high ethnic diversity of their populations and have higher levels of social, environmental and economic deprivation than the English and London average [24]. Schools were selected using simple randomisation with refusals replaced by eligible schools from the same borough. More information on the data collection and study recruitment is detailed elsewhere [24]. The participants were in Year 7 of school at baseline (age 11-12 years; January to June 2012). They were followed-up for a first time in Year 8 of school (wave 2: age 12-13 years; January to June 2013) and for a second time in Year 9 of school (wave 3: age 13-14 years; January to June 2014). To reduce seasonality effects, timing of follow-up was matched by month for each school. Because the exposure variables (neighbourhood trust and social support) were not available at baseline, we restricted the analyses to adolescents who participated at wave 2 and wave 3. The final longitudinal cohort comprised all 2,664 adolescents of the 3,228 adolescents interviewed at wave 2, who were also followed-up at wave 3 (retention rate 82%). This analysis was conducted in accordance with the guidelines dictated in the Declaration of Helsinki. The project obtained ethical approval from the Queen Mary University of London Research Ethics Committee (QMREC2011/40), the London Boroughs Research Governance Framework (CERGF113), and the Association of Directors of Children’s Services (RGE110927). Head teachers provided written consent for the study to take place within their school, parents provided passive informed consent for their child to participate, and adolescent participants gave written informed assent. Measures Exposures Four exposure variables were used to capture respondent perceptions of their social environment: neighbourhood trust, as well as social support from friends, family and significant others. Neighbourhood trust is a single-item obtained from a broader set of age-adapted questions on trust in different groups of people [25]. The question asks whether the respondents ‘trust people in [their] neighbourhood’. The variable response is on a four category Likert scale, such that 1=’not at all’, 2=’a little’, 3=’some’, 4=’a lot’. The neighbourhood trust item was selected because it was expected to be particularly affected by Olympic-related regeneration. The social support measures are derived from the 12-item Multidimensional Scale of Perceived Social Support (MSPSS) [26]. It is a composite measure of social support which is non-specific to PA and captures more emotional than instrumental forms of support. MSPSS items were rated on seven-point Likert scale ranging from ‘agree very strongly’ to ’disagree very strongly’. We summed scores for each source of support (i.e. friends, family and significant others) and split them into tertiles (1=’low’, 2=’medium’, 3=’high’), owing to their skewed positive distribution. Ordinal exposure variables were treated as either discrete or continuous when there was indication of a dose-response relationship. In addition, change scores in each exposure variable were calculated as the difference between the two data collection points in the numeric values to which time-varying exposure variable is coded. Positive scores indicate improvement in the exposure variables over time. PA outcomes PA was measured using the Youth PA Questionnaire (Y-PAQ). Y-PAQ is a validated self-reported instrument that assesses the duration and frequency of a series of PA and sedentary activities over the past 7 days [27]. We computed four forms of PA expected to be differentially associated with the exposure variables: walking to school, walking for leisure, outdoor PA and pay and play PA. Outdoor PA aims to group physical activities that are mainly performed in open recreation areas such as parks, sport fields and other open spaces, and which are usually located in the residential neighbourhood of adolescents [28,29]. The measure is composed of football, rounders, basketball/volleyball (mainly outdoor), cricket, rollerblading, roller skating and rugby. Pay and play PA captures scheduled formal PA, usually performed in sport or leisure centres and for which adolescents might need to pay in order to participate. It includes aerobics, climbing, swimming, gymnastics, hockey, martial arts, netball and tennis. Owing to the skewed distribution of the total time spent on each form of PA and to the fact that no adequate transformation could be found, the four outcome variables were dichotomised as ‘activity reported at least once’ vs. not [30,31]. Measures of within individual change in the outcomes over time were also created and constructed as differences in the binary PA status between the two measurement points, resulting in ordinal variables with 3 responses categories (0= stopped reporting the PA outcome at wave 3; 1= no change; 2= started reporting the PA outcome at wave 3). Covariates We identified potential confounders a priori from existing literature. Those available at both measurement points (wave 2 and wave 3) were added to the adjusted models if we found evidence of association with the exposures and PA. The adjusted models included: gender, ethnicity (White UK, White Mixed, Indian, Pakistani, Bangladeshi, Black Caribbean, Black African, or Other), household composition (both parents vs. not), time lived in the neighbourhood (≤5 years vs. > 5); free-school meal status, family affluence (3 categories from the revised Family Affluence Scale II), and the presence of health condition (none vs. 1+ ; from the following conditions: mobility problems, longstanding illness, anaemia, asthma, diabetes, Chronic Fatigue Syndrome, hay fever, hearing and eyesight problems) [30,31]. Statistical analyses Missing values ranged from 0.0% to 21.0%. We investigated both the predictors of partially observed variables and the predictors of the probability of missingness using logistic regressions. These analyses indicated that the missing at random assumption was plausible. We used multilevel multiple imputation to impute missing data using the ‘jomo’ package in R (based on a joint multivariate normal modelling approach) [32]. Auxiliary variables were total physical activity (log-transformed), country of birth, language spoken at home, mental health (squared WEMWBS score), BMI z-score, self-rated health, parental involvement, and neighbourhood satisfaction. Our imputation model included two levels (adolescents nested in schools). The outcomes and covariates were included as fixed effects, so that a separate variable was used for each measurement occasion. We handled interaction terms between the exposure variables and gender by imputing the data for each gender separately [30,31]. The imputation model was compatible with the most saturated model of analysis [33]. We used a ‘burn in’ period of 4,050 iterations for boys and girls, and 1,000 between-imputation iterations to generate 20 imputed datasets. The convergence of the Markov Chain Monte Carlo chains was satisfactory. To assess general associations between exposures and the outcomes, we estimated unadjusted and adjusted logistic regression models with GEE (pooled models based on waves 2 and 3). This way of exploring associations provides information on whether differences in exposures, coming either from two different individuals or the same individual at different measurement points, lead to differences in PA. This modelling strategy is useful to capture evidence of association in situations with few repeated measurements or if exposures do not change much over time. We used GEE methods to account for the hierarchical structure of the data (although restricted to two levels of analysis), while preserving a population-average interpretation of the parameters [34]. The models accounted for the clustering due to repeated measurements on the same individuals (using unstructured working correlation structures). We did not adjust for all exposure variables together, given multicollinearity between the three sources of social support. We then explored whether within individual changes over time in exposure were associated with changes in outcomes using proportional odds models. Adjusted models included confounders as measured at the first measurement point (wave 2). None of the models accounted for clustering at school-level as it was negligible for our PA outcomes [30]. Finally, we explored whether gender was a moderator by running a series of fully adjusted models that included an interaction term between each exposure of interest and gender. Stratum-specific results were reported for the interactions with p-values <0.1. Analyses were conducted using Stata 15. Results Walking to school and outdoor PA were most prevalent (respectively, 77% and 76% at wave 2), and walking for leisure least prevalent (35% at wave 2) (Table 1). PA prevalence declined over time. While walking to school remained constant, walking for leisure and outdoor PA each decreased by about 5 percentage points at wave 3 and pay and play PA decreased by 13 percentage points. A majority of participants had at least some trust in their neighbours at each wave and high or medium levels of perceived social support from friends, family and significant others. While measures of social support had relatively stable distributions, those reporting no trust in their neighbours slightly increased at wave 3 (from 9.4% to 12.4%). The key socio-demographic characteristics of the sample are provided in Table 1. Overall, the sample was relatively deprived (37.3% received free school meals at wave 2; 40.7% had low family affluence) and ethnically diverse (16.6% were White UK and 36.3% were classified as Other). Walking to school The pooled models do not provide evidence of an association between neighbourhood trust and social support and walking to school (Table 2). Models of change in the exposures and outcomes confirm the absence of associations with walking to school (Table 3). The inclusion of interaction terms between each measure of the social environment and gender provides no evidence that gender moderates the associations, except for change in social support from significant others (Tables 2 and 3). For girls, there is weak evidence that improved social support from significant others over time might increase the odds of walking to school at wave 3 (OR=1.15; 95% CI: 0.98-1.35) (Table 4). Walking for leisure The pooled model indicates that measures of social support, unlike neighbourhood trust, are positively associated with walking for leisure (Table 2). The evidence appears to be stronger for social support from family (OR trend=1.15; 95% CI: 1.06-1.25). The model for change in the exposure and outcomes however only provides evidence that increased social support from friends increases the odds of walking for leisure at wave 3 (OR=1.11; 95% CI: 1.01-1.22) (Table 3). Gender does not moderate these associations (Tables 2 and 3). Outdoor PA Results from the pooled model provide evidence of a positive dose-response association between neighbourhood trust and outdoor PA (OR trend=1.10; 95% CI: 1.01-1.19) (Table 2). Social support was not associated with outdoor PA in the general model, however, stratified analyses indicate evidence of a positive dose-response relationship between social support from friends and PA in boys (OR trend=1.21; 95% CI: 1.04-1.42) (Table 5). The models for changes in neighbourhood trust or social support indicate no evidence of associations with change in outdoor PA (Table 3). Pay and play PA As for outdoor PA there is strong evidence from the pooled model of a positive dose-response relationship between neighbourhood trust and pay and play PA (OR trend=1.09; 95% CI: 1.02-1.17), but no evidence for social support (Table 2). The models for changes in neighbourhood trust or social support indicate no evidence of associations with change in pay and play PA (Table 3). Nonetheless, stratified results provide weak evidence that, for boys only, improved neighbourhood might improve the odds of pay and play PA at wave 3 (OR=1.13; 95% CI: 1.00-1.29) (Table 4). Discussion In this paper, we examined associations between two aspects of the social environment - neighbourhood trust and social support - and four PA outcomes in a deprived and ethnically diverse adolescent population. We found that perceived neighbourhood trust was positively associated with outdoor PA and with pay and play PA. There was also consistent evidence for an association between social support and walking for leisure. We did not find any evidence that walking to school was associated with any of the exposures. Neighbourhood trust The few studies which have investigated the associations between trust, or other aspects of social capital, and PA in adolescents have found positive associations with total and recreational PA though there is heterogeneity in the way exposures are measured [10–13]. For example, a recent cross-national study found consistent associations between collective efficacy and associated with objectively-measured total PA in 9-11 year old children in twelve high income countries [14]. The findings reported here are the first to investigate associations between trust and different forms of PA in the UK. We find evidence that neighbourhood trust was generally positively associated with leisure-time PA: walking for leisure, outdoor PA, and pay and play PA. This is consistent with previous results based on total or recreational PA as outcomes. Two plausible mechanisms could explain this. First, neighbourhood trust might favour autonomy and outside play by reducing fear of crime and increasing informal social control. Second, higher neighbourhood trust might also be indicative of stronger and better-organised communities, which might provide more opportunities for structured and unstructured PA [4]. In contrast, we found no evidence of an association with utilitarian walking, captured using walking to school. The prevalence of walking to school, unlike other forms of PA, did not decrease during the study period. This suggests that walking to school might be a more acceptable and pragmatic activity for adolescents, which is therefore less likely to be affected by the social environment. In this study, most of the evidence for an association between neighbourhood trust and PA comes from pooled longitudinal models, in which the association comes either from cross-sectional information or from within individual changes over time. Although, in the models focusing on within individual changes we found very limited evidence of associations. Interestingly, however, a model for boys indicated that an increase in neighbourhood trust over time was associated with an improvement in pay and play PA. This suggests that, despite the very short period of follow-up (1 year between wave 2 and wave 3), improvement in neighbourhood trust might have positive and relatively immediate consequences for leisure-time PA. Social support We found consistent associations between the three sources of social support (friends, family and significant others) and walking for leisure, but no consistent associations with walking to school, outdoor PA, or pay and play PA. These results contrast somewhat with the literature on social support and PA in young people which report consistent positive associations [16,19–21]. There are several possible reasons for this. First, we did not include a measure of total PA, for which a positive association might be observed, given the direction of the coefficients observed for walking for leisure and outdoor PA. Second, the MSPSS is generic tool which does not address PA specifically [20] and appears to be better at capturing emotional aspects of social support [35]. Instrumental support, co-participation and modelling, which were all shown to be relevant aspects of social support for PA [16,19,20], are not explicitly mentioned in the MSPSS instrument. Therefore, if adolescents were receiving non-emotional forms of support, it may be under-reported. In particular, more structured activities captured by pay and play PA typically require instrumental support from the parents, such as paying participation fees, buying equipment, and providing transportation [36]. The absence of an observed association might reflect the possibility that such aspects of social support are poorly captured by the MSPSS, and therefore might under-estimate associations between social support and PA investigated here. The main finding was to provide evidence supporting longitudinal associations, especially for social support from family and friends. These results are consistent with the few studies that have investigated changes in social support and changes in PA [17,37–39]. These generally show that increasing or maintaining general social support and encouragement from parents and friends during adolescence matters for PA. That we are able to observe improvements in PA from positive changes in social support over a short period of time (one year), suggest that interventions targeting social support might be beneficial for PA. We found some evidence that gender moderated observed associations for walking to school and for outdoor PA. In boys, we found strong evidence that higher social support from friends, and weak evidence that family social support, increases the odds of outdoor PA, and this association is consistent with the literature [16]. The fact that the association was only observed for boys is unexpected as boys and girls received similar amount of social support. One possible explanation is that the type of support received might differ for boys and girls and be more relevant to PA for boys (e.g. transportation, co-participation, encouragement). Strengths and limitations of this study To our knowledge, this is one of the first large-scale analyses of the longitudinal associations between social capital and social support with four measures of adolescent PA in the UK. We use statistical methods that account for non-independence of observations and item non-response. The Y-PAQ questionnaire allowed for the study of four common types of PA, and thus enabled us to explore how different aspects of PA were associated with measures of the social environment. Additional advantages of this study were the ethnic diversity of the sample; the high response rate (87% at wave 1) and the high retention rate (82% between wave 2 and wave 3). However, there are also some limitations. PA is self-reported and is subject to social desirability and recall biases [40]. In addition, the psychometric properties of neighbourhood trust, as a single-item, have not been established. The super-diversity of the ORiEL sample is both a strength and a weakness as over 200 ethnic categories were self-reported by respondents, which restricted the ability to examine ethnic-specific effects of the social environment. Although the ORiEL study is one of the few large longitudinal studies to explore the determinants of PA, its short period of follow-up for which data were available (1 year) may have limited the ability to find longitudinal associations. Nonetheless, the physical, economic and social transformation of East London occurring around the time of the 2012 Olympic and Paralympic Games are likely to have affected the aspects of the social environment investigated in this study [23]. It is therefore likely that the extent of change in exposure observed within a year of follow-up is larger than would naturally occur in other studies conducted in less dynamic urban settings. It should also be noted that due to age-dependent variations in (changes in) social support and trust as adolescent grow-up, our results should not be generalised to older adolescents. Finally, we were unable to assess causal relationships. Though the longitudinal design can help strengthen causal inference, reverse causality cannot be ruled out. Conclusion Our findings suggest that different aspects of the social environment predict different types of adolescent PA. This suggests that there may be no ‘one-size fits all’ strategy to improving the social environment to increase adolescent PA. Policymakers and practitioners rather need to consider tailoring social support and social capital interventions in order to increase different forms of PA. Funding NB was supported by the Economic and Social Research Council (Grant No. 1482460). The ORiEL study was funded by the NIHR Public Health Research Programme (Grant No. 09/3005/09 to SC). The funding source had no role in the writing of this article nor the decision to submit it for publication. Declarations Availability of data and materials Restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request. Please contact SC with data requests. What is already known on this subject? Previous studies suggest that the social environment could contribute to inequalities in physical activity behaviours. Social support and social capital (including social cohesion and neighbourhood trust) are positively associated with total and recreational physical activity. However, longitudinal evidence on whether changes in social support and social capital are associated with changes in physical activity is still scarce, particularly for young people outside of the USA. What this study adds? We found that different aspects of the social environment predict different types of physical activity, in a relatively deprived and ethnically diverse UK adolescent population. Neighbourhood trust was positively associated with leisure-type physical activity, while social support from friends and family was positively associated with walking for leisure. There was some evidence that changes in exposures led to changes in physical activity outcomes (1 year follow-up). Policymakers and practitioners should consider tailoring interventions to promote social support and social capital by physical activity type. Table 1 Characteristics of the ORiEL study participants, 2013-2014 (n=2,644) 2013 2014 % Missing Exposures Neighbourhood trust 13.9    Not at all 9.4 12.4    A little 27.7 27.6    Some 41.4 42.4    A lot 21.5 17.6 Social support – friends 20.7    Low 39.4 40.6    Medium 28.4 31.2    High 32.2 28.2 Social support – family 20.4    Low 29.6 33.4    Medium 27.3 30.1    High 43.1 36.5 Social support – significant others 21.0    Low 40.7 40.9    Medium 26.9 28.0    High 32.4 31.1 Outcomes % walking to school 76.6 75.7 3.5 % walking for leisure 35.0 30.1 6.1 % reporting outdoor PA 76.0 71.0 10.4 % reporting pay and play PA 64.1 51.3 9.5 Covariates % Girls 56.7 - 0.0 Ethnicity 0.1    % White: UK 16.6 -    % White: Mixed 8.3 -    % Asian: Indian 3.7 -    % Asian: Pakistani 3.9 -    % Asian: Bangladeshi 15.1 -    % Black: Caribbean 4.9 -    % Black: African 11.2 -    % Other 36.3 - % with health condition 39.7 41.4 14.4 % receiving free school meals 37.3 32.5 1.8 Family affluence 3.6    % Low 7.1 4.9    % Medium 50.6 51.5    % High 42.3 43.6 % not living with both parents 31.4 32.9 0.8 % living in the neighbourhood > 5 y 60.7 63.1 4.9 Table 2 General associations of the social environment with physical activity (n=2,644) Exposure Unadjusted Adjusted1 Gender-int.2 OR 95%CI p-value OR 95%CI p-value p-value Outcome: Walking to school Neighbour. trust Not at all 1.00 0.296 1.00 0.482 0.528 A little 1.02 [0.80,1.30] 0.99 [0.77,1.26] Some 1.17 [0.93,1.47] 1.10 [0.88,1.39] A lot 1.06 [0.81,1.38] 0.98 [0.75,1.28] Soc. sup. – friends Low 1.00 0.253 1.00 0.170 0.258 Medium 1.10 [0.93,1.29] 1.07 [0.91,1.27] High 0.95 [0.81,1.12] 0.90 [0.76,1.07] Soc. sup. – family Low 1.00 0.753 1.00 0.680 0.062 Medium 0.95 [0.80,1.14] 0.95 [0.79,1.15] High 0.94 [0.79,1.11] 0.93 [0.78,1.10] Soc. sup. – sig. oth. Low 1.00 0.916 1.00 0.934 0.265 Medium 0.97 [0.81,1.15] 0.97 [0.81,1.16] High 1.00 [0.85,1.17] 0.98 [0.83,1.16] Outcome: Walking for leisure Neighbour. trust Not at all 1.00 0.193 1.00 0.314 0.919 A little 1.28 [1.02,1.61] 1.24 [0.98,1.57] Some 1.25 [1.00,1.56] 1.24 [0.99,1.55] A lot 1.20 [0.94,1.52] 1.22 [0.95,1.57] Trend5 1.03 [0.96,1.11] 0.414 1.04 [0.97,1.12] 0.250 0.822 Soc. sup. – friends Low 1.00 0.001 1.00 0.079 0.464 Medium 1.24 [1.06,1.44] 1.17 [1.00,1.37] High 1.31 [1.12,1.53] 1.17 [0.99,1.38] Trend5 1.15 [1.06,1.24] 0.001 1.08 [1.00,1.18] 0.050 0.205 Soc. sup. – family Low 1.00 <0.001 1.00 0.004 0.641 Medium 1.20 [1.01,1.43] 1.19 [1.00,1.42] High 1.38 [1.18,1.62] 1.32 [1.12,1.56] Trend5 1.17 [1.08,1.27] <0.001 1.15 [1.06,1.25] 0.001 0.352 Soc. sup. – sig. oth. Low 1.00 [0.95,1.27] 0.001 1.00 0.055 0.474 Medium 1.18 [1.02,1.38] 1.11 [0.95,1.30] High 1.34 [1.14,1.56] 1.21 [1.03,1.43] Trend5 1.16 [1.07,1.25] <0.001 1.10 [1.02,1.20] 0.019 0.373 Outcome: Outdoor PA3 Neighbour. trust Not at all 1.00 <0.001 1.00 0.099 0.680 A little 1.03 [0.82,1.28] 0.97 [0.76,1.24] Some 1.16 [0.95,1.43] 1.08 [0.86,1.36] A lot 1.60 [1.24,2.05] 1.29 [0.97,1.70] Trend5 1.17 [1.09,1.26] <0.001 1.10 [1.01,1.19] 0.029 0.390 Soc. sup. – friends Low 1.00 0.164 1.00 0.748 0.027 Medium 0.91 [0.77,1.07] 1.06 [0.89,1.27] High 0.86 [0.73,1.01] 1.06 [0.89,1.26] Soc. sup. – family Low 1.00 0.844 1.00 0.815 0.179 Medium 1.01 [0.85,1.20] 1.05 [0.88,1.26] High 1.04 [0.89,1.22] 1.05 [0.88,1.25] Soc. sup. – sig. oth. Low 1.00 0.273 1.00 0.881 0.354 Medium 0.90 [0.76,1.06] 1.03 [0.86,1.23] High 0.89 [0.76,1.03] 1.04 [0.89,1.23] Outcome: Pay and play PA4 Neighbour. trust Not at all 1.00 0.005 1.00 0.025 0.695 A little 1.03 [0.82,1.29] 0.95 [0.75,1.20] Some 1.10 [0.89,1.35] 1.03 [0.82,1.27] A lot 1.40 [1.10,1.78] 1.27 [0.99,1.63] Trend5 1.12 [1.04,1.20] 0.001 1.09 [1.02,1.17] 0.013 0.850 Soc. sup. – friends Low 1.00 0.975 1.00 0.891 0.528 Medium 1.00 [0.86,1.15] 1.00 [0.86,1.17] High 1.01 [0.87,1.18] 0.97 [0.83,1.13] Soc. sup. – family Low 1.00 0.624 1.00 0.968 0.470 Medium 1.00 [0.86,1.17] 0.98 [0.84,1.15] High 1.07 [0.91,1.26] 0.98 [0.83,1.16] Soc. sup. – sig. oth. Low 1.00 0.761 1.00 0.867 0.847 Medium 0.99 [0.85,1.16] 0.96 [0.82,1.13] High 1.05 [0.90,1.22] 1.00 [0.85,1.17] Results are from logistic regression models estimated with Generalised Estimating Equations to account for the dependency across repeated measurements (unstructured working correlation matrix). 1 Adjusted for gender, ethnicity, health condition, free school meal status, family affluence, time lived in the neighbourhood, household composition and time. 2 The adjusted model was replicated for each outcome with an additional interaction term between gender and exposure. 3 Outdoor PA include: basketball (or volleyball), blading, cricket, football, rounders, rugby and roller skating. 4 Pay and play PA include: aerobics, climbing, swimming, gymnastics, hockey, martial arts, netball, and tennis. 5 Exposure modelled as a continuous variable when indication of a dose-response relationship. OR – Odds ratio, int. – interaction, Neighbour. – Neighbourhood, Soc. sup. – Social support, sig. oth. – significant others, Table 3 Associations of change in the social environment with change in physical activity (n=2,664) Exposure Unadjusted Adjusted1 Gender-interaction2 OR 95%CI p-value OR 95%CI p-value p-value Outcome: Walking to school Neighbour. trust 1.02 [0.92,1.14] 0.698 1.03 [0.92,1.15] 0.605 0.145 Soc. sup. – friends 0.97 [0.87,1.08] 0.598 0.97 [0.87,1.08] 0.575 0.157 Soc. sup. – family 1.01 [0.90,1.14] 0.831 1.00 [0.89,1.13] 0.956 0.529 Soc. sup. – sig. oth. 1.03 [0.92,1.14] 0.618 1.02 [0.92,1.14] 0.694 0.071 Outcome: Walking for leisure Neighbour. trust 1.07 [0.98,1.17] 0.127 1.07 [0.98,1.17] 0.156 0.876 Soc. sup. – friends 1.11 [1.01,1.22] 0.031 1.11 [1.01,1.22] 0.037 0.447 Soc. sup. – family 1.07 [0.97,1.19] 0.189 1.07 [0.97,1.19] 0.189 0.489 Soc. sup. – sig. oth. 1.05 [0.95,1.15] 0.351 1.04 [0.95,1.15] 0.364 0.167 Outcome: Outdoor PA3 Neighbour. trust 0.99 [0.90,1.09] 0.847 0.99 [0.90,1.09] 0.834 0.641 Soc. sup. – friends 1.01 [0.91,1.12] 0.893 1.01 [0.91,1.12] 0.867 0.315 Soc. sup. – family 0.97 [0.87,1.08] 0.582 0.97 [0.87,1.08] 0.543 0.556 Soc. sup. – sig. oth. 1.01 [0.91,1.12] 0.875 1.00 [0.90,1.12] 0.947 0.785 Outcome: Pay and play PA4 Neighbour. trust 1.06 [0.97,1.16] 0.170 1.06 [0.97,1.16] 0.201 0.099 Soc. sup. – friends 0.99 [0.91,1.09] 0.891 0.99 [0.91,1.08] 0.845 0.695 Soc. sup. – family 0.98 [0.89,1.09] 0.741 0.98 [0.88,1.09] 0.691 0.275 Soc. sup. – sig. oth. 0.98 [0.89,1.08] 0.672 0.98 [0.89,1.08] 0.672 0.340 Results are from proportional odds models. Results are displayed as ORs of change in PA status (contrasting increase vs. constant high/low or decrease; or increase or constant high/low vs. decrease) per unit increase in the original scale of neighbourhood trust or tertile change in social support. ORs > 1 indicate a positive change in the outcome as a response to an improvement in the exposure. 1 Adjusted for gender, ethnicity, health condition, free school meal status, family affluence, time lived in the neighbourhood and household composition at wave 2. 2 The adjusted model was replicated for each outcome with an additional interaction term between gender and exposure. 3 Outdoor PA include: basketball (or volleyball), blading, cricket, football, rounders, rugby and roller skating. 4 Pay and play PA include: aerobics, climbing, swimming, gymnastics, hockey, martial arts, netball, and tennis. OR – Odds ratio, Neighbour. – Neighbourhood, Soc. sup. – Social support, sig. oth. – significant others, Table 4 Associations of change in the social environment with change in walking to school and change in pay and play PA by gender (n=2,664) Exposure Unadjusted Adjusted1 OR 95%CI p-value OR 95%CI p-value Outcome: Walking to school Soc. sup. – sig. oth.     Boys 0.94 [0.80,1.09] 0.410 0.93 [0.80,1.09] 0.369     Girls 1.15 [0.98,1.34] 0.086 1.15 [0.98,1.35] 0.091 Outcome: Pay and play PA2 Neighbour. trust     Boys 1.14 [1.00,1.30] 0.040 1.13 [1.00,1.29] 0.055     Girls 0.97 [0.85,1.11] 0.665 0.98 [0.86,1.11] 0.718 Results are from proportional odds models. Results are displayed as ORs of change in PA status (contrasting increase vs. constant high/low or decrease; or increase or constant high/low vs. decrease) per unit increase in the original scale of neighbourhood trust or tertile change in social support. ORs > 1 indicate a positive change in the outcome as a response to an improvement in the exposure. 1 Adjusted for gender, ethnicity, health condition, free school meal status, family affluence, time lived in the neighbourhood and household composition at wave 2. 2 Outdoor PA include: basketball (or volleyball), blading, cricket, football, rounders, rugby and roller skating. OR – Odds ratio, Neighbour. – Neighbourhood, Soc. sup. – Social support, sig. oth. – significant others Table 5 General associations of the social environment with outdoor PA1 by gender (n=2,644) Exposure Adjusted OR2 95%CI p-value Adjusted OR2 95%CI p-value Boys Girls Soc. sup. – friends Low 1.00 0.039 1.00 0.509 Medium 1.22 [0.92,1.61] 0.93 [0.75,1.17] High 1.47 [1.07,2.02] 0.88 [0.70,1.10] Trend3 1.21 [1.04,1.42] 0.014 - - - Soc. sup. – family Low 1.00 0.166 1.00 0.733 Medium 1.18 [0.87,1.60] 0.98 [0.77,1.24] High 1.31 [0.99,1.74] 0.92 [0.74,1.15] Trend3 1.15 [0.99,1.32] 0.060 - - - Soc. sup. – sig. oth. Low 1.00 0.266 1.00 0.836 Medium 1.16 [0.87,1.53] 0.94 [0.74,1.19] High 1.24 [0.95,1.63] 0.94 [0.75,1.17] Trend3 1.12 [0.98,1.28] 0.106 - - - Results are from logistic regression models estimated with Generalised Estimating Equations to account for the dependency across repeated (unstructured working correlation matrix). 1 Outdoor PAs include: basketball (or volleyball), blading, cricket, football, rounders, rugby and roller skating. 2 Adjusted for ethnicity, health condition, free school meal status, family affluence, time lived in the neighbourhood, household composition and time. 3 Exposure modelled as a continuous variable when indication of a dose-response relationship. OR – Odds ratio, Neighbour. – Neighbourhood, Soc. sup. – Social support, sig. oth. – significant others Author contributions NB conceived of the study, designed and executed the statistical analyses, and drafted the manuscript. MQ and ENN advised the statistical analyses and contributed to the interpretation of data. DL and SC supervised all aspects of data processing and analysis, and guided the design of this study. All authors critically revised the manuscript. All authors read and approved the final manuscript. Competing interests The authors declare that they have no competing interests 1 Brodersen NH Steptoe A Boniface DR Trends in physical activity and sedentary behaviour in adolescence: Ethnic and socioeconomic differences Br J Sports Med 2007 41 140 4 10.1136/bjsm.2006.031138 17178773 2 Health and Social Care Information Centre Statistics on obesity, physical activity and diet England 2017 2017https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/613532/obes-phys-acti-diet-eng-2017-rep.pdf 3 Papas MA Alberg AJ Ewing R The built environment and obesity Epidemiol Rev 2007 29 129 43 17533172 4 Kawachi I Berkman LF Social cohesion, social capital, and health Berkman LF Kawachi I Glymour MM Social Epidemiology Oxford University Press 2014 290 319 5 Barnett E Casper M A definition of ‘social environment’ Am J Public Health 2001 91 465 6 Berkman LF Social Support, Social Networks, Social Cohesion and Health Soc Work Health Care 2000 31 3 14 10.1300/J010v31n02_02 11081851 7 Putnam RD The prosperous community Am Prospect 1993 4 35 42 8 McNeill LH Kreuter MW Subramanian SV Social environment and physical activity: a review of concepts and evidence Soc Sci Med 2006 63 1011 22 16650513 9 Lindström M Social capital and health-related behaviors Kawachi I Kim D Social Capital and Health Springer 2008 215 38 10 Carroll-Scott A Gilstad-Hayden K Rosenthal L Disentangling neighborhood contextual associations with child body mass index, diet, and physical activity: the role of built, socioeconomic, and social environments Soc Sci Med 2013 95 106 14 10.1016/j.socscimed.2013.04.003 23642646 11 Cradock AL Kawachi I Colditz GA Neighborhood social cohesion and youth participation in physical activity in Chicago Soc Sci Med 2009 68 427 35 10.1016/j.socscimed.2008.10.028 19036490 12 Franzini L Elliott MN Cuccaro P Influences of physical and social neighborhood environments on children’s physical activity and obesity Am J Public Health 2009 99 271 8 10.2105/AJPH.2007.128702 19059864 13 Kimbro RT Brooks-Gunn J McLanahan S Young children in urban areas: Links among neighborhood characteristics, weight status, outdoor play, and television watching Soc Sci Med 2011 72 668 76 10.1016/J.SOCSCIMED.2010.12.015 21324574 14 Sullivan SM Broyles ST Barreira TV Associations of neighborhood social environment attributes and physical activity among 9-11 year old children from 12 countries Health Place 2017 46 183 91 10.1016/j.healthplace.2017.05.013 28544991 15 Langford CP Bowsher J Maloney JP Social support: a conceptual analysis J Adv Nurs 1997 25 95 100 accessed 4 Dec 2017 9004016 16 Mendonça G Cheng L Mélo E Physical activity and social support in adolescents: a systematic review Health Educ Res 2014 29 822 39 accessed 18 May 2017 http://her.oxfordjournals.org/content/early/2014/05/07/her.cyu017.short 24812148 17 Lau EY Faulkner G Qian W Longitudinal associations of parental and peer influences with physical activity during adolescence: findings from the COMPASS study Heal Promot Chronic Dis Prev Canada Res Policy Pract 2016 36 235 42 accessed 12 Jan 2018 27882858 18 Sallis JF Bull F Guthold R Progress in physical activity over the Olympic quadrennium Lancet 2016 388 1325 36 10.1016/S0140-6736(16)30581-5 27475270 19 Laird Y Fawkner S Kelly P The role of social support on physical activity behaviour in adolescent girls: a systematic review and meta-analysis Int J Behav Nutr Phys Act 2016 13 79 10.1186/s12966-016-0405-7 27387328 20 Beets MW Cardinal BJ Alderman BL Parental social support and the physical activity-related behaviors of youth: a review Heal Educ Behav 2010 37 621 44 10.1177/1090198110363884 21 Yao CA Rhodes RE Parental correlates in child and adolescent physical activity: a meta-analysis Int J Behav Nutr 2015 12 10 10.1186/s12966-015-0163-y 22 Carrillo-Álvarez E Kawachi I Riera-Romaní J Neighbourhood social capital and obesity: a systematic review of the literature Obes Rev 2019 20 119 41 10.1111/obr.12760 30306717 23 Cummins S Clark C Lewis DJ Evaluating the impact of Olympic-related urban regeneration on physical activity and psychological health and wellbeing in adolescents and their parents The ORiEL Study 2017 24 Smith NR Clark C Fahy AE The Olympic Regeneration in East London (ORiEL) study: protocol for a prospective controlled quasi-experiment to evaluate the impact of urban regeneration on young people and their families BMJ Open 2012 2 e001840 10.1136/bmjopen-2012-001840 25 2000 Social Capital Community Benchmark Survey https://ropercenter.cornell.edu/featured-collections/2000-social-capital-community-benchmark-survey 26 Zimet GD Powell SS Farley GK Psychometric characteristics of the Multidimensional Scale of Perceived Social Support J Pers Assess 1990 55 610 7 10.1080/00223891.1990.9674095 2280326 27 Corder K van Sluijs EM Wright A Is it possible to assess free-living physical activity and energy expenditure in young people by self-report? Am J Clin Nutr 2009 89 862 70 10.3945/ajcn.2008.26739 19144732 28 D’Haese S Van Dyck D De Bourdeaudhuij I The association between the parental perception of the physical neighborhood environment and children’s location-specific physical activity BMC Public Health 2015 15 565 10.1186/s12889-015-1937-5 26088831 29 Esteban-Cornejo I Carlson JA Conway TL Parental and adolescent perceptions of neighborhood safety related to adolescents’ physical activity in their neighborhood Res Q Exerc Sport 2016 87 191 9 10.1080/02701367.2016.1153779 27030158 30 Berger N Lewis D Quartagno M Associations between school and neighbourhood ethnic density and physical activity in adolescents: Evidence from the Olympic Regeneration in East London (ORiEL) study Soc Sci Med 2019 112426 10.1016/J.SOCSCIMED.2019.112426 31387008 31 Berger N Lewis D Quartagno M Longitudinal associations between perceptions of the neighbourhood environment and physical activity in adolescents: Evidence from the Olympic Regeneration in East London (ORiEL) study BMC Public Health 2019 19 1760 10.1186/s12889-019-8003-7 31888573 32 Quartagno M Grund S Carpenter J jomo: A Flexible Package for Two-level Joint Modelling Multiple Imputation R J 2019 9 1 33 Carpenter JR Kenward MG Multiple imputation and its application John Wiley Sons 2012 34 Fitzmaurice GM Laird NM Ware JH Applied longitudinal analysis Wiley 2011 35 Dahlem NW Zimet GD Walker RR The Multidimensional Scale of Perceived Social Support: A confirmation study J Clin Psychol 1991 47 756 61 10.1002/1097-4679(199111)47:6<756::AID-JCLP2270470605>3.0.CO;2-L 1757578 36 Edwardson CL Gorely T Parental influences on different types and intensities of physical activity in youth: a systematic review Psychol Sport Exerc 2010 11 522 35 10.1016/J.PSYCHSPORT.2010.05.001 37 Zook KR Saksvig BI Wu TT Physical activity trajectories and multilevel factors among adolescent girls J Adolesc Heal 2014 54 74 80 10.1016/j.jadohealth.2013.07.015 38 Dowda M Dishman RK Pfeiffer KA Family support for physical activity in girls from 8th to 12th grade in South Carolina Prev Med 2007 44 153 9 10.1016/J.YPMED.2006.10.001 17157371 39 Davison KK Jago R Change in parent and peer support across ages 9 to 15 yr and adolescent girls’ physical activity Med Sci Sport Exerc 2009 41 1816 25 10.1249/MSS.0b013e3181a278e2 40 Prince SA Adamo KB Hamel M A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review Int J Behav Nutr Phys Act 2008 5 56 10.1186/1479-5868-5-56 18990237

PMC007xxxxxx/PMC7614812.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8303205 Soc Sci Med Soc Sci Med Social science & medicine (1982) 0277-9536 1873-5347 31387008 7614812 10.1016/j.socscimed.2019.112426 EMS181450 Article Associations between school and neighbourhood ethnic density and physical activity in adolescents: evidence from the Olympic Regeneration in East London (ORiEL) study Berger Nicolas Nicolas.Berger@lshtm.ac.uk Population Health Innovation Lab, Department of Public Health, Environments and Society, London School of Hygiene & Tropical Medicine, London, United Kingdom Lewis Daniel Daniel.Lewis@lshtm.ac.uk Population Health Innovation Lab, Department of Public Health, Environments and Society, London School of Hygiene & Tropical Medicine, London, United Kingdom, Data Science Campus, Office for National Statistics, London, United Kingdom Quartagno Matteo m.quartagno@ucl.ac.uk MRC Clinical Trials Unit, University College London, London, United Kingdom, Department of Medical Statistics, London School of Hygiene & Tropical Medicine, London, United Kingdom Njagi Edmund Njeru Edmund-Njeru.Njagi@lshtm.ac.uk Department of Non-communicable Disease Epidemiology, London School of Hygiene & Tropical Medicine, London, United Kingdom Cummins Steven Steven.Cummins@lshtm.ac.uk Population Health Innovation Lab, Department of Public Health, Environments and Society, London School of Hygiene & Tropical Medicine, London, United Kingdom Corresponding author: Nicolas Berger, Department of Public Health, Environments and Society, London School of Hygiene & Tropical Medicine, Nicolas.Berger@lshtm.ac.uk, 15-17 Tavistock Place, London WC1H 9SH, UK 01 9 2019 22 7 2019 19 7 2023 26 7 2023 237 112426112426 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. While most adolescents do not achieve the recommended level of physical activity in the UK, the risk of physical inactivity varies across ethnic groups. We investigated whether own-group school and neighbourhood ethnic density can explain ethnic differences in adolescent physical activity. We used longitudinal data from the Olympic Regeneration in East London (ORiEL) study. In 2012, 3,106 adolescents aged 11-12 were recruited from 25 schools in East London, UK. Adolescents were followed-up in 2013 and 2014. Own-group ethnic density was measured in 2012-2014 at school-level and in 2011 at neighbourhood-level, and calculated as the percentage of pupils/residents who were of the same ethnic group. Analyses were restricted to White British (n=382), White Mixed (n=190), Bangladeshi (n=337), and Black African groups (n=251). We estimated adjusted logistic regression models with generalised estimating equations for self-reported walking to school, walking for leisure, and outdoor physical activity. At school-level, there was consistent evidence that own-group ethnic density amplifies ethnic differences in walking to school. For each 10 percentage point increase in own-group ethnic density, there was evidence of increased probability of walking to school in Bangladeshi adolescents (OR=1.20; 95% CI 1.09-1.31) and decreased probability of walking to school in Black African (OR=0.58; 95% CI 0.45-0.75) and White Mixed adolescents (OR=0.51; 95%CI 0.35-0.76). Associations with walking for leisure and outdoor physical activity were in expected directions but not consistently observed in all ethnic groups. At neighbourhood-level, evidence was more restricted. Amplification of ethnic differences was found for walking to school in Bangladeshi adolescents (OR=1.31; 95% CI 1.14-1.51) and for outdoor physical activity in White British adolescents (OR=0.85; 95% CI 0.76-0.94). Our results suggest that own-group ethnic density contributes to explaining differences in physical activity by amplifying ethnic differences in some forms of physical activity. ethnicity race ethnic density place health behaviour walking England UK pmcIntroduction Most adolescents do not achieve the recommended level of physical activity in the UK (Health and Social Care Information Centre, 2017). Recent research, although limited, suggests that differences exist in children’s activity levels between ethnic groups in the UK. For example, data from the Child Heart and Health Study in England and the Millennium Cohort Study show that South Asian children were less active than the European White and Black African-Caribbean children (Griffiths et al., 2013; Owen et al., 2009). One of the very few studies investigating ethnic differences by type of activity reported that White European children were more likely to walk or cycle to school than ethnic minority groups (Owen et al., 2012). One explanation for ethnic differences in physical activity behaviour is ethnic-specific attitudes to different types of activities. Different ethnic groups might have differing norms with respect to socially acceptable health behaviours and activities, such as walking to school and playing outside (Bécares et al., 2011). These ethnic differences in physical activity norms might be reinforced for people living in areas with higher proportions of people of the same ethnicity, that is, areas with higher own-group ethnic density. Ethnic density has been hypothesised to influence other health behaviours by increasing civic engagement, increasing social capital and social support, and reducing exposure to racism and discrimination (Bécares and Nazroo, 2013; Shaw et al., 2012). A handful of studies have investigated associations between ethnic density and health behaviours in the UK, finding some protective effect for alcohol consumption in ethnic minorities (Bécares et al., 2011), and differential effects for smoking, which appear to vary depending on the prevalence of smoking in the ethnic group in question (Mathur et al., 2017). However, empirical research on other health behaviours remains limited. There are very few studies that have investigated the association between ethnic density and physical activity, and none in UK adolescents. Exploring the ethnic density hypothesis in adolescent health behaviours may help shed light on the relative importance of ethnic density in the residential and school settings (Astell-Burt et al., 2012). Teasing out the independent contributions of neighbourhood deprivation and ethnic density also remains an issue, given the correlation between the processes of ethnic and economic segregations (Karlsen and Nazroo, 2002). Focusing on homogeneously deprived but ethnically diverse areas might help better capture the ethnic density ‘effect’ itself (Uphoff et al., 2016). In this study we undertook a longitudinal analysis of a deprived adolescent population to address whether exposure to higher own-group density would be associated with physical activity, after adjusting for a number of potential confounders. Effects in residential and school settings were examined for four ethnic groups – White British, White Mixed, Bangladeshi and Black African – and for three physical activity outcomes – walking to school, walking for leisure and outdoor physical activity. Methods Study design and participants We analysed data from the ORiEL study, a prospective cohort study, a prospective cohort study aimed at assessing the health impact of urban regeneration following the London 2012 Olympic and Paralympic Games. Participants were recruited from 25 schools in four London boroughs: Tower Hamlets, Hackney, Barking and Dagenham, and Newham. The boroughs have highly ethnically diverse populations and higher levels of social, economic and environmental deprivation than the England average (McLennan et al., 2011; Office for National Statistics, 2013). Six schools per borough in Newham, Hackney and Barking & Dagenham, and seven schools in Tower Hamlets were selected using simple randomisation with refusals replaced by eligible schools from the same borough. Special-needs schools, pupil referral units and independent schools were excluded from the sampling frame. The sample consisted of both single and mixed-sex faith and non-denominational schools. Faith schools were affiliated to a range of religious denominations. Full details on study recruitment and data collection are described elsewhere (Smith et al., 2012). The participants, in year 7 at baseline (age 11-12 years: Jan-June 2012), were first followed-up in year 8 (wave 2, age 12-13 years: Jan-June 2013) and again in year 9 (wave 3, age 13-14 years: Jan-June 2014). Timing of follow-up for each school was matched by month to reduce seasonality effects. The longitudinal cohort comprised 2,260 adolescents who participated in all three waves, representing an overall retention rate of 73% (Figure 1). Measures Ethnicity Ethnicity was assessed by asking participants: “Which ONE category best describes you - this is your race or ethnic group?”, with 24 pre-defined categories available for selection. The question was adapted from the 2011 Census for England and Wales (Office for National Statistics, 2013). If the relevant category was not available respondents could write in free text their self-identified race/ethnicity. Due to statistical power issues, only the four largest ethnic groups were included in the analyses: “White British” (n=382), “White Mixed” (White and any other background; n=190), “Bangladeshi” (n=337) and “Black African” (n=251) (Figure 1). Own-group ethnic density exposures Ethnic density in school and residential settings were computed for each ethnic group and assigned to adolescents based on their self-reported ethnicity. The data sources used definitions of ethnicity compatible with the one used in this study. School-level prevalence of each ethnic group (i.e. ethnic density) was calculated in participating schools using ethnicity statistics from the Department for Education for the period 2012-2014 (Department for Education, 2014). Neighbourhood-level ethnic density was measured at the lower layer super output area (LSOA) using ethnic composition data from the 2011 UK Census Population. The LSOA has been suggested to be the best administrative area with available routine data to characterise ethnic density effects (Stafford et al., 2009). LSOA data were geo-coded to the home-address of the participants for each of the waves. Amongst adolescents belonging to one of the four main ethnic groups who reported a home address, some moved primary place of residence. As a result, 5.2% changed LSOA at wave 2, and another 5.9% changed LSOA wave 3. The neighbourhood-level ethnic density variable is therefore time-varying to account for changes in exposure due to residential mobility. Exposure variables were treated as continuous in the analyses, in the absence of established cut-off values in the literature (Shaw et al., 2012). Physical activity outcomes Physical activity was assessed using the Youth Activity Questionnaire (Y-PAQ). Y-PAQ is a validated self-reported tool that captures the frequency and duration of a range of physical and sedentary activities over the past 7 days (Corder et al., 2009). Three forms of physical activity expected to be differentially associated with the exposure variables were computed: walking to school, walking for leisure and outdoor physical activity. Outdoor physical activity aims to group physical activities that are mainly performed in open recreation areas such as parks, sport fields and other open spaces, which are usually located in the residential neighbourhood of the adolescents (D’Haese et al., 2015; Esteban-Cornejo et al., 2016). It combines basketball/volleyball (with the expectation that basketball is mainly reported in an outdoor court), (roller)blading, cricket, football, rounders, rugby and roller skating. Running was not included due to under-reporting which reflects that the activity was likely to have been understood as ‘running around’ by adolescents and not understood as a formal sporting activity. Owing to their non-normal distributions and to the fact that no adequate transformation could be found, the three outcome variables measuring forms of physical activity were dichotomised (e.g. activity reported at least once vs. not). Covariates Potential confounders available at baseline and for both follow-up surveys were identified a priori from existing literature. They were included in adjusted models if there was evidence of associations with physical activity and ethnic density. Gender; time lived in neighbourhood (≤ 5 years vs. > 5);household composition (both parents vs. none); family affluence score from the revised Family Affluence Scale II (low=0-2; medium=3-5; high=6-9) (Boyce et al. 2006); free-school meal status at baseline; health condition (none vs. 1+); and distance to school (for walking to school only) were selected. Country of birth was not associated with any of the physical activity outcomes and therefore omitted from analyses. Unlike previous studies, we were unable to adjust for area of deprivation because the study population was homogeneously deprived: 87% of adolescents’ residential LSOAs were classified below the 1st quintile of the Income Deprivation Affecting Children Index (IDACI) and 98% were below the 1st or 2nd quintiles. The full ORiEL questionnaire is available elsewhere (Cummins et al., 2018). Statistical analyses Prevalence of missing data for the outcomes and covariates were examined; missing values ranged from 0.0% to 13.7%. We explored both predictors of the probability of missingness and predictors partially observed variables through logistic regression modelling. Analyses suggested that data were not missing completely at random and that the missing at random assumption was plausible. Data were imputed using multilevel multiple imputation with the ‘jomo’ package in R, which uses a joint multivariate normal modelling approach through the Markov Chain Monte Carlo method (Quartagno et al., 2018). We imputed with 2 levels (first, adolescent; second, school) with all the outcomes and covariates as fixed effects using the data in the wide format, so that each measurement occasion was represented by a separate variable. Interaction terms between ethnicity and the ethnic density variables were handled by imputing the data separately for each ethnic group. The imputation model was chosen to be compatible with the most saturated model of interest; auxiliary variables were included to strengthen the missing at random assumption (Carpenter and Kenward, 2012). We used a ‘burn in’ period of 35,050 iterations and 5,000 between-imputation iterations to produce 20 imputed datasets. The Markov Chain Monte Carlo chains were examined to check for convergence. Unadjusted and adjusted logistic regression models were estimated using generalised estimating equations (GEE) in Stata 15 with the command “mi estimate: xtgee”. GEE methods were used to account for the hierarchical structure of the data at individual level (measurements nested within individuals), and have a convenient population-average interpretation of the parameters (Fitzmaurice et al., 2011). We were unable to specifically examine the effect of within-individual changes in ethnic density because of the restricted extent of change in residential LSOA over the study period. Preliminary analyses indicated no evidence of clustering at school- or neighbourhood-level, so that these additional levels of hierarchy were not taken into account in the final models. Lowess smoothers were used to explore the functional shape of the association between the logit of physical activity and the measures of ethnic density (Cleveland, 1979). For each outcome, separate logistic models were specified to test school-level and neighbourhood-level ethnic density effects by ethnic group. For each ethnic density variable, unadjusted models included time, exposure, ethnicity and ethnicity*exposure interaction terms. Partially adjusted models further included potential confounders. Finally, the fully adjusted models included time, ethnicity, potential confounders, the two exposures and their interaction with ethnicity. For sensitivity analyses purposes, models were also stratified by ethnic group instead of using interaction terms to allow confounding to differ by ethnic group; the exposure variables were modelled as tertiles to allow deviation from linearity; and an alternative working correlation structure was used to initiate the GEE models using exchangeable as opposed to unstructured correlation matrices (Molenberghs and Verbeke, 2005). Results Ethnic differences in physical activity prevalence differed by form of physical activity (Table 1). The prevalence of walking to school was highest in Bangladeshi (84.4%) and White British (80.8%) groups, and lowest in White Mixed (72.4%) and Black African (71.4%) groups. Walking for leisure was highest in the White British group (48.3%), intermediate in the White Mixed group (39.8%), and lowest in the Black African (28.5%) and Bangladeshi (24.4%) groups. Outdoor physical activity was highest in the Black African group (80.1%), intermediate in the White Mixed (75.1%) and Bangladeshi (74.8%) groups, and lowest in the White British group (71.4%). The vast majority of adolescents (96%) attended a local school located outside their residential LSOA (median distance to school was 1.6km). Own-group ethnic densities were highest for White British and Bangladeshi adolescents at both school- and neighbourhood-levels, and lowest for White Mixed and Black African adolescents (Table 1). Table 1 describes the key socio-demographic characteristics of the sample. In general, White British adolescents were less disadvantaged and were more likely to have lived in their neighbourhood for more than 5 years. Walking to school School-level own-group ethnic density (school-level ethnic density hereafter) is associated with walking to school, after adjustment for potential confounders (Table 2). A positive association is observed for the Bangladeshi group, indicating that a 10% increase in school-level ethnic density increases the odds of walking to school by 1.20 (95% CI: 1.09-1.31). In adjusted models, negative associations are observed for the White Mixed (OR: 0.51; 95% CI: 0.35-0.76) and Black African (OR: 0.58; 95% CI: 0.45-0.75) groups. The model using exposure tertiles (Supplementary Table 7) indicates a U-shaped relationship for the White British group such that the lowest odds of walking to school are observed for the 2nd tertile of ethnic density. Table 2 shows evidence of associations between neighbourhood-level own-group ethnic density (neighbourhood-level ethnic density hereafter) and walking to school. Compared to school-level measures, coefficients have the same signs but are mostly lower in magnitude. The strongest association is observed in the Bangladeshi group, where an increase in neighbourhood-level ethnic density by 10% increases the odds of walking to school by 1.31 (95% CI: 1.14-1.51). In fully adjusted model, which includes the two ethnic density exposures and potential confounders, school-level ethnic density remains a predictor of walking to school, whereas neighbourhood-level ethnic density coefficients are no longer statistically significant (Table 2). An increase in school-level ethnic density by 10% would decrease the odds of walking to school by a factor of 2.27 (=1/0.44, 95% CI: 1.43-3.57) for the White Mixed group and by 1.67 (=1/0.60, 95% CI: 1.43-3.57) for the Black African group. In the Bangladeshi group, coefficients of school-level and neighbourhood-level ethnic densities are attenuated in the fully adjusted model (ORs=1.13 and 1.15, respectively) and are no longer significant, which reflects an overlap between the two ethnic density measures for that group and the incapacity of the model to differentiate school-level from neighbourhood-level effects in this context. Walking for leisure There was no evidence of log-linear associations between ethnic density measures and walking for leisure for any ethnic group, before and after adjustment for potential confounders (Table 3). Results by tertile (Supplementary Table 8) confirm the lack of association with school-level ethnic density, with one possible exception. Tertile analysis indicates weak evidence of a negative dose-response relationship in the Bangladeshi group: as school-level ethnic density tertile increases, the odds of walking for leisure decreases. However, the fully adjusted model indicates that, in the presence of the two exposures and potential confounders, there is no evidence of association between ethnic density measures and walking for leisure (Supplementary Table 8). Outdoor physical activity Table 4 provides some evidence that school-level ethnic density is associated with outdoor physical activity in some ethnic groups, after adjustment for potential confounders. In particular, a negative association is observed for the White British group, indicating that an increase in school-level ethnic density by 10% decreases the odds of outdoor physical activity by 1.16 (=1/0.86; 95% CI: 1.03-1.30). The models using exposure tertiles suggest the presence of a bell-shaped relationship for the Black African group, such that estimated odds of outdoor physical activity are highest in the 2nd tertile of school-level ethnic density, and lowest in the 3rd tertile (Supplementary Table 9). There is evidence that school-level ethnic density is associated with outdoor physical activity in the White British group, such that an increase in neighbourhood-level ethnic density by 10% decreases the outdoor physical activity by 1.17 (=1/0.85; 95% CI: 1.06-1.32), after adjustment for potential confounders (Table 4). The fully adjusted model shows that, in the White British group, associations are attenuated but remain statistically significant at neighbourhood-level, but not at school-level (ORs are 0.87 (95% CI: 0.77-0.98) and 0.94 (95% CI: 0.82-1.08), respectively). Sensitivity analyses Additional analyses stratified by ethnic group and those based on different specifications of the working correlation structure in the GEE process indicated no differences in the interpretation of the results (Supplementary Tables 1-6). Analyses using ethnic density tertiles, as opposed to continuous scores, allowed us to obtain more correct estimates in the presence of non-linear relationships, as reported above. Non-linear relationship were observed between school-level ethnic density and walking to school in the White British group (Supplementary Table 7) and between school-level ethnic density and outdoor physical activity in the Black African group (Supplementary Table 9). Interpretations of other parameters remained unchanged (Supplementary Tables 7-9). Discussion We explored whether own-group ethnic density was associated with physical activity in an ethnically diverse and relatively deprived adolescent population, after controlling for individual socio-demographic characteristics. We found consistent evidence that school-level ethnic density is associated with walking to school. The direction of the associations are ethnic-specific but indicate that higher ethnic density amplifies the underlying ethnic-specific propensity to walk to school. A higher ethnic density appears to increase the propensity to walk to school in the Bangladeshi adolescents; conversely, it seems to decrease it in the White Mixed and Black African groups, which are groups with a lower prevalence of walking to school. No prior study has examined the association between ethnic density and physical activity in the UK (Bécares et al., 2012), but some studies on smoking have reported comparable results. In particular, a large study conducted using electronic health records of adults from the boroughs of Hackney, Lambeth, Newham and Tower Hamlets showed that the negative association between smoking and ethnic density was greater in ethnic minority groups where smoking was less socially accepted (Mathur et al., 2017). Another study conducted in a deprived population indicated that a higher South Asian density was associated with a lower probability of smoking during pregnancy in the Pakistani women, a group in which smoking is uncommon, whereas no protective effect was found amongst the White British women (Uphoff et al., 2016). There are three main theoretical pathways by which ethnic density might influence health and health-related behaviours (Bécares et al., 2009; Bécares and Nazroo, 2013; Das-Munshi et al., 2010; Halpern and Nazroo, 2000; Karlsen et al., 2012; Pickett and Wilkinson, 2008). Own-group ethnic density might increase civic engagement; increase social capital and social support; and reduce exposure to racism and discrimination. With respect to walking to school, the latter two processes are likely to be more salient. An increase in neighbourhood social capital and social support might in addition provide resources to cope better with experiences of racism and discrimination. As a result, experience of racism might not translate into a change in health behaviours. The three hypothesised pathways imply that higher ethnic density might provide greater opportunities to conduct ethnic-specific preferred health behaviours, which can lead to an amplification of ethnic differences if these cultural norms differ by ethnic group. Explaining observed associations in terms of amplification of ethnic-specific cultural norms seems plausible in this context. Previous studies have shown differences of knowledge, norms and expectations about health behaviours across ethnic minority groups (Koshoedo et al., 2015; Rawlins et al., 2013). In addition, studies have shown that ‘homophily’ or the tendency for friendships to form between those who are alike, is more frequent amongst ethnic minority groups, and that adolescents tend to adopt health behaviours that are similar to their friends’ behaviours (Lorant et al., 2016). These behaviours have been recognised as being both potentially positive and negative for health. Alternative explanations have been offered in the literature to explain ethnic differences (Nazroo, 2014) but these seem less consistent with the amplification phenomenon observed here. One of those alternative explanations is that observed associations might reflect the degree of acculturation, or the fact that ethnic minorities shift their behaviour over time and become more westernised so that health-related cultural differences between minority groups and the majority diminish (Bécares et al., 2011; Pickett et al., 2009). Acculturation might indeed confound the amplification phenomenon. In this study, however, we have found no evidence of association between the physical activity outcomes and either country of birth or language spoken at home in the ethnic group studied. Although acculturation might not be fully captured by the two variables (Bécares et al., 2011), these should at least have displayed some indication of an association if acculturation was playing a major role. Another alternative explanation for the results observed might come from differences in racism and discrimination across ethnic groups. Racism is considered as having a central role in the development of ethnic inequalities in health, and might affect perceived safety, fear of crime and health behaviours (Foster et al., 2014; Karlsen et al., 2012; Lorant et al., 2016; Rawlins et al., 2013). However, the experience of racism alone would not be enough to explain why the association with ethnic density is positive for some ethnic groups and negative for others. Therefore, it is plausible to explain these results in terms of amplification of ethnic-specific cultural norms, which might themselves, but not necessarily, have been the result of broader contextual and structural socio-economic inequalities (Karlsen and Nazroo, 2002; Nazroo, 1998). The associations observed for walking to school should be interpreted cautiously for the following reasons. First, despite being in the expected direction, associations are modest and not statistically significant in all ethnic groups. The strength of the association indicates that a 10 percent increase in ethnic density is estimated to increase the odds of walking to school by 0.44 to 1.10. Second, no clear associations were found with the other physical activity outcomes. The only other consistent evidence of an association was for the White British group, for whom a higher ethnic density decreases the odds of outdoor physical activity, which is less popular in that ethnic group compared to others. The reasons for inconsistent results relating to walking to school and outdoor physical activity are not clear. A possible explanation for outdoor physical activity might be the composite nature of the measure, which pools a series of activities with different levels of popularity across ethnic groups, and therefore dampens differences. We also compared the relative importance of school-level and neighbourhood-level ethnic density in explaining differences in physical activity. As expected, school-level density appears to matter more for walking to school, and neighbourhood-level ethnic density for outdoor physical activity. Where associations were observed, they were usually for both measures in partially adjusted models. However, in models adjusted for both ethnic density measures, only one of the measures would usually remain significant. A notable exception are Bangladeshi adolescents, for whom stronger associations between neighbourhood-level ethnic density and walking to school were observed, but no significant associations were found in the fully adjusted model. These results can be explained by the overlap between school-level and neighbourhood-level density measures in that group (r=0.69), and the fact that the ethnic density of Bangladeshi adolescents was very high in some schools (up to 80%), reaching a potential threshold above which an increase in ethnic density might not have any further effect. Astell-Burt et al. (2012) have also investigated the influences of neighbourhood and school-level densities in adolescents and reported negative associations with perception of racism, but the authors did not compare the relative influence of the two measures. Strengths and limitations of this study To our knowledge this is the first study to examine the association of ethnic density with physical activity in the UK, using validated instruments and appropriate statistical methods to account for non-independence of observations and item non-response. The Y-PAQ questionnaire allowed for the study of three common types of physical activity, and thus explored how different aspects of physical activity were associated with ethnic density. A further advantage of the current study was in the use of large-scale data of a representative sample of the ethnic diversity of East London, providing evidence from populations less studied in the physical activity research. Unlike previous studies of ethnic density, our study population was homogeneously deprived, which helped better capture the ethnic density ‘effect’ itself due to the absence of correlation between ethnic density and deprivation in our context (Uphoff et al., 2016). Results might nonetheless not be generalizable to other settings. The study had a high response rate (87% at baseline) and retention rate (71%), which is consistent with best practice in other school-based cohorts (Booker et al., 2011). This research also has limitations. Physical activity measured by the Y-PAQ is self-reported and might therefore be subject to recall and social desirability biases (Prince et al., 2008). However, the use of an objective physical activity measure was not practically possible given the size of the study. The Y-PAQ questionnaire does not have situational reference (Giles-Corti et al., 2005) and did not capture where the reported activity was taking place (e.g. garden, neighbourhood, parks). Such information would be valuable to better understand the relative contribution of school- and neighbourhood-level ethnic densities on more specific types of activities. As large-scale studies of ethnic minorities are rare in the field, especially in the UK, the ethnic diversity of the ORiEL study is a major strength. However, the super-diversity of the sample was a limiting factor because over 200 ethnic categories were self-reported for minor groups. Nonetheless,ethnic differences in the ethnic density could be analysed for four main ethnic groups and some promising results were found despite low statistical power. Although the ORiEL study is one of the few large longitudinal studies to investigate the determinants of physical activity, its short period of follow-up (3 waves; 2 years) restricted the ability to test the influence of time-change in ethnic density on physical activity, given the limited extent of residential mobility of the participants and the slow pace of change in the ethnic composition of their school and neighbourhood over time. Another weakness of this study is that we were unable to assess causal relationships. Reverse causality could have accounted for findings; it is plausible that families with preferences for certain lifestyles may choose to send their children to a school or live in a neighbourhood with a greater proportion of people of the same ethnic group. Conclusion This study suggests that own-group ethnic density contributes to explaining differences in physical activity in adolescents by amplifying ethnic differences, in particular for walking to school. Further research is needed to confirm these results in different populations and for different health behaviours. Supplementary Material Supplementary Material Figure 1 Data flowchart Table 1 Characteristics of the study participants by ethnic group, 2012-2014 White British (N=382) White Mixed (N=190) Bangladeshi (N=337) Black African (N=251) % Missing Exposure Median school-level ethnic density (10th- 90th percentiles) 22.7 (13.2-57.6) 14.2 (4.4-21.7) 63.3 (7.5-80.6) 19.3 (9.5-24.8) 0.0 Median neighbourhood-level ethnic density (10th- 90th percentiles) 40.5 (19.7-63.1) 12.8 (6.3-22.2) 22.3 (4.5-53.2) 13.6 (4.0-23.5) 8.2 Outcome Measures % walking to school 80.8 72.4 84.5 71.4 4.4 % walking for leisure 48.3 39.8 24.4 28.5 9.5 % reporting outdoor physical activity 71.1 75.1 74.8 80.1 13.7 Covariates % Girls 44.8 50.0 36.5 40.2 0.0 % with health condition 44.3 51.8 43.1 33.1 10.4 Family affluence 3.7     % Low 8.3 10.2 9.7 6.2     % Medium 43.0 48.5 62.9 57.4     % High 48.7 41.3 27.4 36.5 % receiving free school meals at baseline 29.2 44.2 45.3 41.4 1.7 % not living with both parents 33.1 50.2 13.8 33.3 2.7 % living in the neighbourhood > 5y 76.1 65.3 67.1 50.2 8.1 Median distance to school in km (10th- 90th percentiles) 1.6 (0.5-4.0) 2.1 (0.6-4.2) 1.2 (0.6-3.5) 2.2 (0.7-5.9) 8.5 Results are pooled across the 3 waves of data collection and obtained from 20 imputed datasets. Table 2 Association of increasing own-group ethnic density with walking to school. Values are odds ratios (95% confidence interval) Unadjusted Confounders Adjusted1 Fully Adjusted2 School-level ethnic density * White British 1.08 (0.96 to 1.21) 1.08 (0.96 to 1.21) 1.10 (0.94 to 1.30) White Mixed 0.53 (0.36 to 0.77) 0.51 (0.35 to 0.76) 0.44 (0.28 to 0.70) Bangladeshi 1.19 (1.09 to 1.31) 1.20 (1.09 to 1.31) 1.13 (0.96 to 1.32) Black African 0.58 (0.45 to 0.75) 0.58 (0.45 to 0.75) 0.60 (0.45 to 0.79) Neighbourhood-level ethnic density * White British 1.01 (0.88 to 1.17) 1.01 (0.88 to 1.16) 0.97 (0.81 to 1.15) White Mixed 0.95 (0.62 to 1.44) 0.94 (0.62 to 1.43) 1.33 (0.81 to 2.18) Bangladeshi 1.32 (1.14 to 1.52) 1.31 (1.14 to 1.51) 1.15 (0.91 to 1.46) Black African 0.80 (0.60 to 1.06) 0.80 (0.60 to 1.06) 0.91 (0.67 to 1.25) Results are from logistic regression models estimated with Generalised Estimating Equations to account for the dependency across repeated measurements. Missing data were handled using multilevel multiple imputation (20 datasets). * Assessed as change per 10 percentage points. 1 Adjusted for time, gender, health condition, family affluence, baseline free school meal status, household composition, time lived in the neighbourhood and distance to school. 2 Adjusted for time, gender, health condition, family affluence, baseline free school meal status, household composition, time lived in the neighbourhood, distance to school, the two ethnic density variables and their interaction with ethnicity. Table 3 Association of increasing own-group ethnic density with walking for leisure. Values are odds ratios (95% confidence interval) Unadjusted Confounders Adjusted1 Fully Adjusted2 School-level ethnic density * White British 0.99 (0.90 to 1.09) 0.99 (0.89 to 1.10) 0.96 (0.86 to 1.08) White Mixed 0.92 (0.66 to 1.29) 0.88 (0.62 to 1.25) 0.96 (0.65 to 1.40) Bangladeshi 0.94 (0.89 to 1.00) 0.95 (0.90 to 1.01) 0.97 (0.89 to 1.06) Black African 1.11 (0.83 to 1.49) 1.14 (0.86 to 1.51) 1.07 (0.78 to 1.47) Neighbourhood-level ethnic density * White British 1.03 (0.95 to 1.13) 1.02 (0.94 to 1.12) 1.04 (0.94 to 1.15) White Mixed 0.83 (0.57 to 1.19) 0.82 (0.57 to 1.18) 0.84 (0.56 to 1.25) Bangladeshi 0.92 (0.83 to 1.01) 0.93 (0.85 to 1.03) 0.97 (0.84 to 1.11) Black African 1.17 (0.90 to 1.52) 1.18 (0.91 to 1.54) 1.16 (0.86 to 1.55) Results are from logistic regression models estimated with Generalised Estimating Equations to account for the dependency across repeated measurements. Missing data were handled using multilevel multiple imputation (20 datasets). * Assessed as change per 10 percentage points. 1 Adjusted for time, gender, health condition, family affluence, baseline free school meal status, household composition, time lived in the neighbourhood. 2 Adjusted for time, gender, health condition, family affluence, baseline free school meal status, household composition, time lived in the neighbourhood, the two ethnic density variables and their interaction with ethnicity Table 4 Association of increasing own-group ethnic density with outdoor physical activity. Values are odds ratios (95% confidence interval) Unadjusted Confounders Adjusted1 Fully Adjusted2 School-level ethnic density * White British 0.86 (0.77 to 0.96) 0.86 (0.77 to 0.97) 0.94 (0.82 to 1.08) White Mixed 0.97 (0.66 to 1.43) 1.05 (0.68 to 1.62) 1.04 (0.65 to 1.67) Bangladeshi 1.05 (0.98 to 1.12) 1.02 (0.95 to 1.10) 1.04 (0.94 to 1.14) Black African 0.78 (0.57 to 1.08) 0.77 (0.58 to 1.04) 0.78 (0.56 to 1.09) Neighbourhood-level ethnic density * White British 0.84 (0.76 to 0.92) 0.85 (0.76 to 0.94) 0.87 (0.77 to 0.98) White Mixed 1.07 (0.73 to 1.57) 1.05 (0.70 to 1.57) 1.03 (0.66 to 1.61) Bangladeshi 1.03 (0.93 to 1.15) 1.01 (0.91 to 1.12) 0.97 (0.84 to 1.12) Black African 0.91 (0.66 to 1.22) 0.89 (0.67 to 1.18) 0.97 (0.71 to 1.32) Results are from logistic regression models estimated with Generalised Estimating Equations to account for the dependency across repeated measurements. Missing data were handled using multilevel multiple imputation (20 datasets). * Assessed as change per 10 percentage points. 1 Adjusted for time, gender, health condition, family affluence, baseline free school meal status, household composition, time lived in the neighbourhood. 2 Adjusted for time, gender, health condition, family affluence, baseline free school meal status, household composition, time lived in the neighbourhood, the two ethnic density variables and their interaction with ethnicity. Astell-Burt T Maynard MJ Lenguerrand E Harding S Racism, ethnic density and psychological well-being through adolescence: evidence from the Determinants of Adolescent Social well-being and Health longitudinal study Ethn Health 2012 17 71 87 10.1080/13557858.2011.645153 22332834 Bécares L Nazroo J Stafford M The ethnic density effect on alcohol use among ethnic minority people in the UK J Epidemiol Community Heal 2011 65 20 25 Bécares L Nazroo J Stafford M The buffering effects of ethnic density on experienced racism and health Health Place 2009 15 700 708 10.1016/j.healthplace.2008.10.008 Bécares L Nazroo JY Social capital, ethnic density and mental health among ethnic minority people in England: a mixed-methods study Ethn Health 2013 18 544 562 10.1080/13557858.2013.828831 23964679 Bécares L Shaw R Nazroo JY Stafford M Albor C Atkin K Kiernan K Wilkinson R Pickett K Ethnic density effects on physical morbidity, mortality, and health behaviors: a systematic review of the literature Am J Public Health 2012 102 e33 e66 10.2105/AJPH.2012.300832 Booker CL Harding S Benzeval M A systematic review of the effect of retention methods in population-based cohort studies BMC Public Health 2011 11 249 10.1186/1471-2458-11-249 21504610 Boyce W Torsheim T Currie C Zambon A The family affluence scale as a measure of national wealth: validation of an adolescent self-report measure Soc Indic Res 2006 78 3 473 87 Carpenter JR Kenward MG Multiple imputation and its application John Wiley Sons 2012 Cleveland WS Robust locally weighted regression and smoothing scatterplots J Am Stat Assoc 1979 74 829 836 10.1080/01621459.1979.10481038 Corder K van Sluijs EM Wright A Whincup P Wareham NJ Ekelund U Is it possible to assess free-living physical activity and energy expenditure in young people by self-report? Am J Clin Nutr 2009 89 862 870 10.3945/ajcn.2008.26739 19144732 Cummins S Clark C Lewis D Smith N Thompson C Smuk M Stansfeld S Taylor S Fahy A Greenhalgh T Eldridge S The effects of the London 2012 Olympics and related urban regeneration on physical and mental health: the ORiEL mixed-methods evaluation of a natural experiment, The effects of the London 2012 Olympics and related urban regeneration on physical and mental health: the ORiEL mixed-methods evaluation of a natural experiment NIHR Journals Library 2018 10.3310/PHR06120 D’Haese S Van Dyck D De Bourdeaudhuij I Deforche B Cardon G The association between the parental perception of the physical neighborhood environment and children’s location-specific physical activity BMC Public Health 2015 15 565 10.1186/s12889-015-1937-5 26088831 Das-Munshi J Bécares L Dewey ME Stansfeld SA Prince MJ Understanding the effect of ethnic density on mental health: multi-level investigation of survey data from England Br Med J 2010 341 c5367 20966012 Department for Education Schools, pupils and their characteristics: January 2014 2014 accessed 12.15.17 [WWW Document]. URL https://www.gov.uk/government/statistics/schools-pupils-and-their-characteristics-january-2014 Esteban-Cornejo I Carlson JA Conway TL Cain KL Saelens BE Frank LD Glanz K Roman CG Sallis JF Parental and adolescent perceptions of neighborhood safety related to adolescents’ physical activity in their neighborhood Res Q Exerc Sport 2016 87 191 199 10.1080/02701367.2016.1153779 27030158 Fitzmaurice GM Laird NM Ware JH Applied longitudinal analysis Wiley 2011 Foster S Knuiman M Hooper P Christian H Giles-Corti B Do changes in residents’ fear of crime impact their walking? Longitudinal results from RESIDE Prev Med 2014 62 161 166 24552845 Giles-Corti B Timperio A Bull F Pikora T Understanding physical activity environmental correlates: increased specificity for ecological models Exerc Sport Sci Rev 2005 33 175 181 16239834 Griffiths LJ Cortina-Borja M Sera F Pouliou T Geraci M Rich C Cole TJ Law C Joshi H Ness AR Jebb SA How active are our children? Findings from the Millennium Cohort Study BMJ Open 2013 3 e002893 10.1136/bmjopen-2013-002893 Halpern D Nazroo J The ethnic density effect: results from a national community survey of England and Wales Int J Soc Psychiatry 2000 46 34 46 10.1177/002076400004600105 14529077 Health and Social Care Information Centre Statistics on obesity, physical activity and diet England 2017 Karlsen S Bécares L Roth M Understanding the influence of ethnicity on health Craig G Atkin K Chattoo S Flynn R Understanding “Race” and Ethnicity: Theory, History, Policy Practice. Policy Press Bristol 2012 115 132 Karlsen S Nazroo JY Agency and structure: the impact of ethnic identity and racism on the health of ethnic minority people Sociol Health Illn 2002 24 1 20 10.1111/1467-9566.00001 Koshoedo SA Paul-Ebhohimhen VA Jepson RG Watson MC Understanding the complex interplay of barriers to physical activity amongst black and minority ethnic groups in the United Kingdom: a qualitative synthesis using meta-ethnography BMC Public Health 2015 15 643 10.1186/s12889-015-1893-0 26164652 Lorant V Soto Rojas V Bécares L Kinnunen JM Kuipers MAG Moor I Roscillo G Alves J Grard A Rimpelä A Federico B A social network analysis of substance use among immigrant adolescents in six European cities Soc Sci Med 2016 169 58 65 10.1016/j.socscimed.2016.09.031 27689513 Mathur R Schofield P Smith D Gilkes A White P Hull S Is individual smoking behaviour influenced by area-level ethnic density? A cross-sectional electronic health database study of inner south-east London ERJ Open Res 2017 3 00130 2016 28435842 McLennan D Barnes H Noble M Davies J Garratt E Dibben C The English indices of deprivation 2010 London 2011 Molenberghs G Verbeke G Models for discrete longitudinal data Springer Verlag New York, NY 2005 Nazroo JY Ethnic inequalities in health: addressing a significant gap in current evidence and policy If you could do one thing British Academy 2014 Nazroo JY Genetic, cultural or socio-economic vulnerability? Explaining ethnic inequalities in health Sociol Health Illn 1998 20 710 730 10.1111/1467-9566.00126 Office for National Statistics 2011 Census: QS211EW Ethnic group (detailed), local authorities in England and Wales 2013 Owen CG Nightingale CM Rudnicka A van Sluijs EMF Ekelund U Cook DG Whincup PH Travel to school and physical activity levels in 9-10 year-old UK children of different ethnic origin; Child Heart and Health Study in England (CHASE) PLoS One 2012 7 e30932 10.1371/journal.pone.0030932 22319596 Owen CG Nightingale CM Rudnicka AR Cook D Ekelund U Whincup P Ethnic and gender differences in physical activity levels among 9-10-year-old children of white European, South Asian and African-Caribbean origin: the Child Heart Health Study in England (CHASE Study). Int J Epidemiol 2009 38 1082 1093 10.1093/ije/dyp176 Pickett KE Shaw RJ Atkin K Kiernan KE Wilkinson RG Ethnic density effects on maternal and infant health in the Millennium Cohort Study Soc Sci Med 2009 69 1476 1483 10.1016/j.socscimed.2009.08.031 19765872 Pickett KE Wilkinson RG People like us: ethnic group density effects on health Ethn Health 2008 13 321 334 10.1080/13557850701882928 18701992 Prince SA Adamo KB Hamel M Hardt J Connor Gorber S Tremblay M A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review Int J Behav Nutr Phys Act 2008 5 56 10.1186/1479-5868-5-56 18990237 Quartagno M Grund S Carpenter J jomo: a package for multilevel joint modelling multiple imputation R J (under review) 2018 Rawlins E Baker G Maynard M Harding S Perceptions of healthy eating and physical activity in an ethnically diverse sample of young children and their parents: the DEAL prevention of obesity study J Hum Nutr Diet 2013 26 132 144 10.1111/j.1365-277X.2012.01280.x Shaw RJ Atkin K Bécares L Albor CB Stafford M Kiernan KE Nazroo JY Wilkinson RG Pickett KE Impact of ethnic density on adult mental disorders: narrative review Br J Psychiatry 2012 201 Smith NR Clark C Fahy AE Tharmaratnam V Lewis DJ Thompson C Renton A Moore DG Bhui KS Taylor SJC Eldridge S The Olympic Regeneration in East London (ORiEL) study: protocol for a prospective controlled quasi-experiment to evaluate the impact of urban regeneration on young people and their families BMJ Open 2012 2 e001840 10.1136/bmjopen-2012-001840 Stafford M Bécares L Nazroo J Objective and perceived ethnic density and health: findings from a United Kingdom general population survey Am J Epidemiol 2009 170 484 493 10.1093/aje/kwp160 19608764 Uphoff EP Pickett KE Crouch S Small N Wright J Is ethnic density associated with health in a context of social disadvantage? Findings from the Born in Bradford cohort Ethn Health 2016 21 196 213 10.1080/13557858.2015.1047742 26169185

PMC007xxxxxx/PMC7614813.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0413066 Cell Cell Cell 0092-8674 1097-4172 33861957 7614813 10.1016/j.cell.2021.03.001 EMS181487 Article Lack of evidence for a role of PIWIL1 variants in human male infertility Oud M.S. 1 Volozonoka L. 2 Friedrich C. 3 Kliesch S. 4 Nagirnaja L. 5 Gilissen C. 1 O’Bryan M.K. 67 McLachlan R.I. 89 Aston K.I. 10 Tüttelmann F. 3 Conrad D.F. 5 Veltman J.A. 11* 1 Department of Human Genetics, Donders Institute for Brain, Cognition and Behavior, Radboud university medical center, Nijmegen, The Netherlands 2 Scientific Laboratory of Molecular Genetics, Riga Stradins University, LV-1007, Riga, Latvia 3 Institute of Reproductive Genetics, University of Münster, Münster, Germany 4 Department of Clinical and Surgical Andrology, Centre of Reproductive Medicine and Andrology, University Hospital Münster, Münster, Germany 5 Division of Genetics, Oregon National Primate Research Center, Oregon Health & Science University, Portland, United States 6 School of Biological Sciences, Monash University, Clayton, Melbourne, Australia 7 School of BioSciences, Faculty of Science, The University of Melbourne, Parkville, Melbourne, Australia 8 Hudson Institute of Medical Research, Clayton, Melbourne, Australia 9 Department of Obstetrics and Gynecology, Monash University, Clayton, Melbourne, Australia 10 Division of Urology, Department of Surgery, University of Utah, Salt Lake City, Utah 11 Biosciences Institute, Faculty of Medical Sciences, Newcastle University, Newcastle upon Tyne, United Kingdom * Corresponding author: Joris.Veltman@newcastle.ac.uk 15 4 2021 20 7 2023 26 7 2023 184 8 19411942 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. PIWIL1 MIWI HIWI D-box male infertility azoospermia oligozoospermia exome sequencing single molecule molecular inversion probes piRNA clinical validity assessment pmcDear editor Infertility affects one in six couples, half of which is explained by a male factor (Tüttelmann et al., 2018). While thousands of genes are involved in spermatogenesis, there is a lack of diagnostically relevant genes. Critical evaluation of newly reported candidate genes is important before incorporating these into the diagnostic work-up. As part of a recently performed clinical validity assessment (Oud et al., 2019), an effort by the International Male Infertility Genomics Consortium (IMIGC, http://imigc.org), a closer look was taken at the quality of the evidence described in this journal by Gou et al., 2017 for involvement of Piwi Like RNA-Mediated Gene Silencing 1 (PIWIL1; also known as HIWI) in human male infertility. The mouse homologue of PIWIL1, Piwil1, plays a crucial role in producing the piRNA pool, and by extension shaping the mRNA pool, in haploid germ cells (Deng and Lin, 2002; Gou et al., 2014). Gou et al. (2017) expanded on Piwi1’s regulatory repertoire by showing its role in the remodeling of spermatid chromatin through a protein-protein interaction with RNF8. The authors also studied the role of PIWIL1 in human male infertility by investigating the conserved destruction box (D-box) region which was previously shown to be involved in ubiquitin-mediated PIWIL1 degradation in late spermatogenesis (Zhao et al., 2013). Sanger sequencing of the D-box element of PIWIL1 in a cohort of 413 idiopathic azoospermia patients revealed several heterozygous mutations in three patients (Figure S1A). While we do not dispute their mouse data, the human sequencing data are much less convincing with questionable Sanger peaks indicating mutations shown in Figure 1B of Gou et al. The mutations described in the human patients, if true, point to a mutational mechanism, which, to our knowledge, has never been reported (see Figure S1A and B). Three years since this high impact publication, no other group has reported a replication of these findings. Therefore, we tested whether variants in the D-box of PIWIL1 are associated with severe spermatogenic failure, by combining genomic data from IMIGC members. To investigate whether the PIWIL1 D-box region is a mutational hotspot in patients with severe spermatogenic failure, we studied the frequency of D-box mutations in patients (n=2,740) suffering from azoospermia (n=1,950) or severe oligozoospermia (n=790). Although we obtained high-quality data for the D-box region in all patients, no sequencing variants affected the D-box region (Figure S1C). If D-box variants were to explain 0.7% of our azoospermia cases, as reported in Guo et al., the binomial probability of sampling 0 D-box mutant cases in 1,950 azoospermic men is very small under standard assumptions about human mutation (p=6.70x10-07). If one were to assume the expected D-box mutation rate to be the same in azoospermia and oligozoospermia, the combined sampling probability for 0/2740 cases is even smaller (p=2.11x10-09). To test whether PIWIL1 variants in other domains intolerant to variation may cause infertility (Figure SID), we screened the same set of 2,740 patients as well as a control group of 3,347 men who conceived normally, for rare non-synonymous variation (<1% allele frequency in any subpopulation of gnomAD, Figure S1E-G). In both groups, we detected 19 rare non-synonymous variants in PIWIL1, which did not cluster to one of the variation intolerant domains (Figure S1E-G). There was no significance difference in the number of rare variants between the patient group and the control group (p=0.65; OR 1.22; 95% CI 0.65–2.31). Also, there was no significant clustering of rare missense variants in PIWIL1 (see Supplementary Methods; patients: qualifying variants = 14, |Xi–Xj|=754.31, δg=3.98x105, p=0.28; controls: qualifying variants = 17, |Xi–Xj|=974.46, δg=1.68x105, p=0.82)(Lelieveld et al., 2017), indicating that there are no spatially clustered gain-of-function or dominant-negative variants in critical domains of PIWIL1. Exome sequencing data from 185 patient-parent trios did not reveal de novo mutations in PIWIL1. In four of 2,740 patients, we identified a heterozygous loss-of-function (LoF) variant in PIWIL1 (Figure SIE). Based on the expected versus observed number of LoF variants found in 141,456 individuals of gnomAD version 2.1.1 (Lek et al., 2016), PIWIL1 is not predicted to be intolerant to LoF variation (pLi=0.0; oe=0.57 [lower and upper bound = 0.42–0.77]) and is, thus, unlikely to cause disease whenhaploinsufficient. This point is further strengthened by the discovery of two heterozygous LoF variants in fertile male controls (Figure S1F). There was no significant difference between the number of LoF variants in patient and control groups (p=0.51; OR 2.44; 95% CI 0.45–13.35). These findings are consistent with observations in mice that heterozygous knock-outs are fertile (Deng and Lin, 2002). There are several potential explanations for the previously reported results on the D-box region. Possibly, the quality of the DNA or sequencing was suboptimal or genomic rearrangements in studied individuals confounded their results. The authors provided us with the primers and PCR conditions used for Sanger sequencing. In silico PCR predictions and analysis of alignable scaffold-discrepant positions due to an alternative locus for PIWIL1 in the most recent genome build GRCh38/hg38, did not reveal possible confounding factors (data not shown). However, the presence of a different, as yet unrecognized alternate haplotype of PIWIL1 in the population sampled by Gou et al., could potentially explain the unusual patterns of D-box variation summarized in Figure S1A. Validation of the original primer sequences resulted in high-quality Sanger peaks in 14 of our own patients and 2 controls. DNA of the samples reported by Gou et al. was unfortunately not available. It thus remains unclear whether the experimental design or the sequencing methods used had any effect on the aberrant results. In conclusion, we express our concerns about the claimed role of the PIWIL1 D-box region mutations in human azoospermia as reported by Gou et al. Our data on 2,740 infertile men do not provide any evidence that variation in the D-box, or anywhere else in the PIWIL1 gene is causally linked to human male infertility. While that does not exclude a potential role for this gene in the disease, it is clear that PIWIL1 is not a frequently mutated male infertility gene. Supplementary Material S1 Acknowledgements The authors would like to thank Dr. Martin Jäger, Prof. Peter Robinson, Laurens van de Wiel, Roos Smits and Godfried van der Heijden for discussions, advice and comments. This work was supported by a VICI grant from The Netherlands Organization for Scientific Research (918-15-667 to J.A.V.), an Investigator Award in Science from the Wellcome Trust (209451 to J.A.V.), the German ResearchFoundation (DFG) Clinical Research Unit (CRU326) ‘Male Germ Cells’ (to F.T.), the National Institutes of Health (R01HD078641 to D.C. and K.A.), the National Health and Medical Research Council of Australia (to M.O.B., R.M.L., J.V., K.A. and D.C.). Author roles: Conceptualization, M.S.O. and J.A.V.; Investigation, M.S.O., L.V., C.F. and L.N.; Formal analysis, M.S.O. and C.G.; Resources, S.K., M.O.B., R.M.L, F.T., D.C. and J.A.V., Supervision, M.O.B., K.A., F.T., D.C. and J.A.V.; Writing – original draft, M.S.O. and J.A.V.; Writing – review and editing, M.S.O., L.V., C.F., S.K., L.N., C.G., M.O.B., R.M.L., K.A., F.T., D.C. and J.A.V. Declaration of interest: The authors declare no competing interest. Deng W Lin H miwi, a murine homolog of piwi, encodes a cytoplasmic protein essential for spermatogenesis Developmental cell 2002 2 819 830 12062093 Gou LT Dai P Yang JH Xue Y Hu YP Zhou Y Kang JY Wang X Li H Hua MM Pachytene piRNAs instruct massive mRNA elimination during late spermiogenesis Cell research 2014 24 680 700 24787618 Gou LT Kang JY Dai P Wang X Li F Zhao S Zhang M Hua MM Lu Y Zhu Y Ubiquitination-Deficient Mutations in Human Piwi Cause Male Infertility by Impairing Histone-to-Protamine Exchange during Spermiogenesis Cell 2017 169 1090 1104 e1013 28552346 Lek M Karczewski KJ Minikel EV Samocha KE Banks E Fennell T O’Donnell-Luria AH Ware JS Hill AJ Cummings BB Analysis of protein-coding genetic variation in 60,706 humans Nature 2016 536 285 291 27535533 Lelieveld SH Wiel L Venselaar H Pfundt R Vriend G Veltman JA Brunner HG Vissers L Gilissen C Spatial Clustering of de Novo Missense Mutations Identifies Candidate Neurodevelopmental Disorder-Associated Genes American journal of human genetics 2017 101 478 484 28867141 Oud MS Volozonoka L Smits RM Vissers L Ramos L Veltman JA A systematic review and standardized clinical validity assessment of male infertility genes Human reproduction (Oxford, England) 2019 34 932 941 30865283 Tüttelmann F Ruckert C Röpke A Disorders of spermatogenesis: Perspectives for novel genetic diagnostics after 20 years of unchanged routine Medizinische Genetik : Mitteilungsblatt des Berufsverbandes Medizinische Genetik eV 2018 30 12 20 Zhao S Gou LT Zhang M Zu LD Hua MM Hua Y Shi HJ Li Y Li J Li D piRNA-triggered MIWI ubiquitination and removal by APC/C in late spermatogenesis Developmental cell 2013 24 13 25 23328397

PMC007xxxxxx/PMC7614814.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101572033 Pediatr Obes Pediatr Obes Pediatric obesity 2047-6302 2047-6310 34114363 7614814 10.1111/ijpo.12818 EMS158108 Article Anthropometry-based prediction of body composition in early infancy compared to air-displacement plethysmography Olga Laurentya 1 van Beijsterveldt Inge ALP 2 Hughes Ieuan A 1 Dunger David B 13 Ong Ken K 134 Hokken-Koelega Anita CS 2 De Lucia Rolfe Emanuella 4 1 Department of Paediatrics, Cambridge Biomedical Campus Box 118, University of Cambridge, Cambridge, UK 2 Department of Pediatrics, Subdivision of Endocrinology, Erasmus University Medical Center-Sophia Children's Hospital, Rotterdam, The Netherlands 3 Institute of Metabolic Science, Cambridge Biomedical Campus, University of Cambridge, Cambridge, UK 4 MRC Epidemiology Unit, Cambridge Biomedical Campus Box 285, University of Cambridge, Cambridge, UK Corresponding author Emanuella De Lucia Rolfe, MRC Epidemiology Unit, University of Cambridge School of Clinical Medicine, Box 285 Institute of Metabolic Science, Cambridge Biomedical Campus, Cambridge CB2 0QQ United Kingdom, Telephone: +44 (0)1223 769194, Emanuella.De-Lucia-Rolfe@mrc-epid.cam.ac.uk 01 11 2021 10 6 2021 05 12 2022 26 7 2023 16 11 e12818e12818 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Background Anthropometry-based equations are commonly used to estimate infant body composition. However, existing equations were designed for newborns or adolescents. We aimed to (1) derive new prediction equations in infancy against air-displacement plethysmography (ADP-PEA Pod) as the criterion, (2) validate the newly developed equations in an independent infant cohort, (3) compare them with published equations (Slaughter-1988, Aris-2013, Catalano-1995). Methods Cambridge Baby Growth Study (CBGS), UK, had anthropometry data at 6 weeks (N=55) and 3 months (N=64), including skinfold thicknesses (SFT) at 4 sites (triceps, subscapular, quadriceps, and flank), and ADP-derived total body fat mass (FM) and fat-free mass (FFM). Prediction equations for FM and FFM were developed in CBGS using linear regression models and were validated in Sophia Pluto cohort, the Netherlands, (N=571 and N=447 aged 3 and 6 months, respectively) using Bland-Altman analyses to assess bias and 95% limits of agreement (LOA). Results CBGS equations consisted of sex, age, weight, length, and SFT from 3 sites, and explained 65% of the variance in FM and 79% in FFM. In Sophia Pluto, these equations showed smaller mean bias than the 3 published equations in estimating FM: mean bias (LOA) 0.008 (-0.489, 0.505) kg at 3 months and 0.084 (-0.545, 0.713) kg at 6 months. Mean bias in estimating FFM was 0.099 (-0.394, 0.592) kg at 3 months and -0.021 (-0.663, 0.621) kg at 6 months. Conclusions CBGS prediction equations for infant FM and FFM showed better validity in an independent cohort at ages 3 and 6 months than existing equations. pmcIntroduction Nutritional and growth patterns during early life have been associated with risks for obesity and cardiometabolic diseases later in life1–4. This association has been continuously reported even in the current studies and reviews5–7. Quantification of infant body composition enables accurate estimation of the effects of early life nutrition on growth and the putative developmental mechanisms leading to later co-morbidities. Weight-for-length and body mass index (BMI) are widely used as early adiposity screening tools8, however those parameters do not distinguish between fat mass (FM) and fat-free mass (FFM), the relative proportions of which vary markedly during infancy9. Moreover, in pediatric population, BMI often produces imprecise estimate of adiposity and it varies greatly with age and gender10,11. Several methods are available to assess infant body composition. These include: dual-energy X-ray absorptiometry (DXA)12; quantitative nuclear magnetic resonance (QMR)13; bioelectrical impedance analysis (BIA)14; total body electrical conductivity (TOBEC)15; stable isotope dilution; and air displacement plethysmography (ADP). BIA and TOBEC are non-invasive, safe, portable, inexpensive, and widely available but their use in infants is limited by poor accuracy14–18. Prediction studies in infants using BIA as the criterion are also scarce17,18. DXA and QMR provide more accurate estimates of infant body composition, however, they are often infeasible because they require infants to lie still, even with the use of sedative agents12,13. In addition, DXA uses ionising radiation and results could vary depending on the type of scans and softwares used19–22. Accordingly, the use of DXA in infants is limited and detailed body composition data in this population are not abundant17,18. ADP-PEA Pod is a non-invasive whole-body densitometry device to estimate infant body composition (total body FM and FFM). It is accurate and reliable in young infants when assessed against DXA23–25, although there was also a study reporting high correlation between those 2 instruments with significant difference19. Nevertheless, ADP-PEA Pod is limited to infants weighing between 1 to 8 kg, thus usually it cannot be used for infants older than 6 months. The equipment is relatively expensive, is not portable, and the process is often time-consuming, so is impractical to use in many large-scale population studies26. Furthermore, some parents report anxiety in leaving their young infants in the closed ADP-PEA Pod system for around 2 minutes26. Therefore, in research studies, estimates of infant body composition are often derived using anthropometry-based equations8,27. However, many of those equations include uncommonly collected measures (e.g. calf circumference28 and flank skinfolds29) that are not available in infant cohort studies. In this study, we aimed to develop new anthropometry-based equations for the prediction of total body FM and FFM in infancy against ADP-PEA Pod as the criterion, in a UK cohort, the Cambridge Baby Growth Study (CBGS). We also aimed to determine the accuracy of these new equations and three existing childhood anthropometry-based equations (Slaughter et al16, Aris et al23, and Catalano et al30, Table 1), in an independent birth cohort, Sophia Pluto study, The Netherlands, using ADP-PEA Pod as the reference method. While Aris et al23 and Catalano et al30 were derived among neonatal populations, Slaughter et al16 involved individuals aged 8-29 years old. Although the age range used in those 3 published equations was different from ours, they are frequently used in studies involving infants and children and were built using relevant anthropometry measures and skinfold sites. Subjects and Methods Derivation cohort The new anthropometry-based prediction equations were derived in CBGS, a longitudinal birth cohort study set up in 2001 at a single maternity hospital in Cambridge, UK, to investigate the prenatal and postnatal determinants of infancy weight gain, body composition, and adiposity31. To provide detailed growth measures in the first weeks of life, N=150 mother-infant pairs born between 2015-2018 underwent a more intensive measurement protocol. All infants were singleton, vaginally delivered at term, of normal weight mothers with no significant pregnancy comorbidities, and had normal birth weight. This analysis included a cohort subgroup of 77 infants with ADP-PEA Pod measurements. There were in total 119 measurements employed to derive the equations, N=55 at 6 weeks and N=64 at 3 months. There was no significant difference in 6 weeks and 3 months anthropometry between the subgroup and the whole cohort (data not shown). The study was approved by the Cambridge Local Research Ethics Committee, and all mothers gave written informed consent. Validation cohort The anthropometry-based equations developed in CBGS were validated in an independent birth cohort study, Sophia Pluto, a prospective study to collect longitudinal data on measured growth and body composition among large group of healthy infants born at term. Mothers were recruited between 2013-2018, from several maternity wards in and near Rotterdam, The Netherlands. Infant anthropometry Infant anthropometry data were collected by trained pediatric research nurses, following standard protocols. Weight was measured to the nearest 1 g using a Seca 757 electronic baby scale (Seca, Birmingham, UK). Length was measured to the nearest 0.1 cm using an Infantometer (Seca 416). Waist circumference was measured at the midpoint between the lowest rib margin and the iliac crest to the nearest 0.1 cm using a non-stretchable fibre-glass tape (Chasmors Ltd, London, UK) in CBGS and a measuring tape (Seca, Hamburg, Germany) in Sophia Pluto. Skinfold thickness (SFT) measures were taken in triplicate from the left side of the body at 4 sites, including triceps, subscapular, flank (suprailiac), biceps (Sophia Pluto only), and quadriceps (CBGS only) using a calibrated Holtain Tanner/Whitehouse Skinfold Caliper (Holtain, Crymych, UK) in CBGS and using a Skinfold caliper (Slimguide C-120, Creative Health) in Sophia Pluto. Infant body composition parameters (% body fat, FM and FFM) were estimated using ADP-PEA Pod (COSMED/Life Measurement Inc., Concord, CA, USA), which directly measures body volume and body weight to calculate body density. Infant % body fat was calculated from body density assuming the density of fat to be 0.9007 kg/L. Age- and gender-specific densities of FFM were computed using the data of Fomon et al32. FM and FFM were calculated from body weight and % body fat. ADP-PEA Pod was calibrated every day, according to the instructions of the manufacturer. In the CBGS, ADP-PEA Pod was conducted twice, at 6 weeks and 3 months old, while in Sophia Pluto it was conducted twice at 3 and 6 months. Statistical analysis In CBGS, stepwise multivariable regression models were performed to derive the optimal prediction of ADP-PEA Pod derived FM and FFM, using sex, age, length, weight, and skinfold thicknesses as independent variables. The equations involved 3 sites of skinfolds measurement which were commonly measured by both studies: triceps, subscapular, and flank (suprailiac). Quadriceps skinfold was omitted due to its unavailability in the validation cohort. In Sophia Pluto, FM and FFM values were predicted using newly developed equations and 3 other childhood prediction equations (Table 1). Agreement between predicted and ADP-PEA Pod measured FM and FFM values were assessed using one-sample paired Student’s t-test, bivariate correlation, linear regression analysis, and Bland-Altman analyses. In each Bland-Altman plot, the y-axis represents the difference or bias between equation-predicted and ADP-PEA Pod measured values with limits of agreement (LOA) described as the 95% confidence range (mean bias +1.96 standard deviations), while the x-axis represents the mean values of the 2 methods being compared (FM or FFM predicted from each corresponding equation and their absolute measured values from ADP-PEA Pod). The possibility of predicted results being affected by the magnitude of the measured values was assessed by running a correlation analysis between the mean (of the values measured by ADP-PEA Pod as the reference and each alternative equation) and the difference of values between the reference and each equation. Moreover, proportional bias was also calculated using linear regression, with the difference between measured and predicted FM/FFM acting as the dependent variable while the average of measured and predicted FM/FFM acting as the independent variable. Statistical analyses were performed using SPSS version 25.0 (IBM) and R version 1.0.136. A p value less than 0.05 was considered statistically significant. Results Baseline characteristics of derivation and validation cohorts are summarised in Table 2. At birth, CBGS infants were heavier than Sophia Pluto’s with comparable length. In contrast, both cohorts had similar weight average while CBGS infants were shorter at 3 months of age. In addition, 92.4% CBGS subjects were of Caucasian while Sophia Pluto included a more diverse population with 62.6% Caucasian (of which 93.8 % were white Caucasian and 6.2 % were Turkish/Moroccan), 27.1% of mixed ethnicities, and the remaining 10.3% of other ethnicities (Asian, African, Latin American). Derivation of anthropometry-based prediction equations Infant weight and length appeared as significant predictors of infant body composition while infant sex, gestational age (GA), and postnatal age at visit were not. However, since the equations were derived using the stepwise method with pragmatic approach, infant sex and age at visit were still included in the models. The proportion of variance explained by the derived prediction equations was greater in FFM than FM models. Infant weight, length, sex, and visit age explained 63 and 77% variance of the FM and FFM models, respectively (Table 3). The addition of SFTs only added a further 2% of variance proportion explained in both FM and FFM models. Furthermore, among the three SFT sites included, only flank SFT appeared as significant predictor of infant FFM. We also developed a number of other prediction equations for FM and FFM in CBGS using subsets of the available infant anthropometry parameters. While their predictive abilities are somewhat weaker than the above equations, these will allow a wider application in infant cohort studies that have collected limited anthropometric measurements (Supplementary Table 1). The final equations to be validated in the Sophia Pluto (Table 3) were chosen by taking R2 and root mean squared error (RMSE) into consideration. Independent validation in Sophia Pluto The performance of CBGS equations was assessed against ADP-PEA Pod-measured FM values as the criterion in Sophia Pluto infants and compared to 3 existing childhood equations. The CBGS equation was the only method that had the least significant difference (Table 4). Predicted values from CBGS equations did not differ with the absolute values from ADP-PEA Pod for FM at 3 months (p=0.402) and for FFM at 6 months (p=0.171). Accordingly, while all 4 equations predicted FM with strong positive correlations with ADP-PEA Pod measured FM values at both 3 and 6 months (Pearson coefficients >0.7), mean bias was lowest for CBGS-derived FM values (0.008 kg; LOA -0.489, 0.505) (Figure 1 and Table 5). All of the correlation analyses between the mean (of the values measured by ADP-PEA Pod as the reference and each alternative equation) and the difference of values between the reference and each equation resulted in significant negative correlations, with CBGS equations had the least negative Pearson correlation coefficients (Table 5). Negative proportional bias was also detected for predicted FM values derived from all 4 equations, but again the extent of this bias was smallest when using the CBGS equations (Table 5 and Supplementary Figure 1). FFM was predicted only using CBGS equations and these values were strongly correlated with FFM measured by ADP-PEA Pod at both time points (Pearson coefficients > 0.8). Similarly, FFM predicted by CBGS equations showed small mean bias compared to ADP-PEA Pod measured FFM (0.099 kg; LOA: -0.394, 0.592). Discussion In this study, we derived new anthropometry-based prediction equations for FM and FFM in UK infants aged 5-16 weeks using ADP-PEA Pod as the criterion method. In the CBGS, infant weight and length appeared as significant predictors of infant body composition, whereas infant sex, gestational age (GA), and postnatal age at visit were not. Using stepwise method with pragmatic approach to derive the prediction equations, infant sex and age at visit were still included in the models. Many studies have reported that there are sex differences in body composition33,34. These equations were then validated among Dutch infants aged 3 and 6 months in an independent cohort, Sophia Pluto. In the Sophia Pluto cohort, the CBGS-derived equations produced more accurate predictions of infant FM compared to the other existing FM prediction equations published by Slaughter et al, Aris et al, and Catalano et al. Based on paired T-test, predicted values from CBGS equations were the most accurate to ADP-PEA Pod results compared to the other published equations, in both FM and FFM. Of note, although the participants involved in Slaughter’s equations were much older than our infant population (Table 2), comparing our equation to theirs is still considered relevant. This is because Slaughter’s equations are frequently used in studies involving pediatric population, including those of younger groups12,23, especially when data harmonisation is needed across cohorts35. All equations produced significant negative proportional biases (Table 5), suggesting negative correlations between the mean and the difference of the predicted versus the actual FM/FFM values. This means that the performance of each equation depends on the magnitude of the actual values of FM/FFM and all equations tend to over- and underestimate FM/FFM in those with lower and higher measured values (by the ADP-PEA Pod), respectively. Compared to the other equations, CBGS equations had the smallest proportional biases. To our knowledge, this is one of the few studies testing the combinations of anthropometric parameters to build body composition prediction equations with the use of ADP-PEA Pod as the criterion method. We aimed to predict absolute FM and FFM, rather than relative or % body fat, since previous studies have reported better correlations between those absolute values with anthropometry33. The correlation coefficients of predicted and measured values of FFM were slightly higher than FM values, but the mean differences were similar. We observed that weight and length were the main contributors in predicting infant FM and FFM. Infant weight has been consistently reported in previous studies to be the most essential predictor of infant FM23,27,33. Apart from weight, Deierlein et al reported that other predictors included infant SFT (triceps, subscapular, and quadriceps), sex, age at measurement, and ethnicity27. Infant weight and sex were also described predictors of infant FM in a Singapore cohort (Aris et al), together with GA23. However, we did not find infant sex or GA to be significant contributors to our prediction equations. We postulate that this is due to the limited heterogeneity in ethnicity among CBGS infants, and the difference in age range covered by CBGS (5-16 weeks) compared to those other studies (1-3 days post-delivery)23,36. Nonetheless, although they were not statistically significant, we still included infant sex and age at measurement in the prediction equations as biologically plausible contributors. We found that SFTs contributed only modestly to the prediction of both FM and FFM. Lingwood et al also found that SFT did not improve their predictions equations beyond weight, length and sex33. Nonetheless, SFT were still included in the equations (Table 3) since they increased the R2, decreased the RMSE and therefore increased the precision of the equations, although not by much. Furthermore, of all 3 SFT sites included in the equations, only flank SFT appeared to be a significant independent predictor of infant FFM (Table 3). Since flank skinfold reflects central adiposity, this result could be speculatively interpreted as central fatness contributing more to the FFM estimation. However, if all SFT sites in CBGS cohort were considered, additional analyses showed that both flank and quadriceps SFTs were the most significant contributors to the prediction of FFM (Supplementary materials). Interestingly, flank SFT was also determined as the most significant FM predictor in Catalano’s equation30. Since Sophia Pluto did not measure quadriceps SFT, this parameter could not be included in the equations taken forward for validation. The proportional biases in the CBGS equations were smaller than those of the other equations, but they were all significant when compared to the criteria method (ADP-PEA Pod, Table 5 and Supplementary figure 1). Therefore, accurate body composition measurement during infancy should be pursued by ADP-Pea Pod or DXA, whilst equations can be employed as proxies to estimate fat-/fat-free mass where body composition instrument is not available. While the derivation sample included a wide distribution of % body fat (6.5-38.6%) and a relatively wide age range (38-112 days), we acknowledge some limitations. Firstly, the skinfold thickness measurements in the derivation and validation cohorts were conducted using different tools. However, despite the use of different calipers, the CBGS equations still produced smaller proportional biases compared to the other established equations. Second, since all CBGS infants were vaginally delivered with normal birth weight and born of healthy mothers with normal pre-pregnancy BMI, the equations might not be applicable in population with a high rate of Caesarean section and high variance of maternal pre-pregnancy BMI or infant’s birth weight. Third, both prediction and validation cohorts included only healthy and term infants thus our findings may not relevant for preterm infants. However, our validation cohort also included severe small for gestational age (SGA) infants with birth weight/length less than -2.5 z-score. Regarding ethnicity, although our derivation cohort was predominantly white Europeans, Sophia Pluto as the independent validation cohort included more diverse ethnicities with at least 37% of them were non-Caucasian. Although Aris et al did not find ethnicity to be significant in their FM equation derived in Asian infants23, a recent systematic review reported differences in infant body composition between ethnicities37. Therefore, the applicability of our equations to other ethnic populations remains in question. Conclusions We derived and validated new anthropometry-based equations for infant FM and FFM using simple parameters often measured in infant studies. These new equations appeared to be more robust in predicting infant FM and FFM when compared to other published childhood equations despite the presence of proportional bias. These equations are fit for use in longitudinal infant cohorts or trials, when reference methods, such as ADP-PEA Pod, are not feasible. Supplementary Material Supporting Information Acknowledgements CBGS The authors are grateful to the CBGS research nurses Suzanne Smith, Ann-Marie Wardell, and Karen Forbes, Addenbrooke’s Hospital, Cambridge. We thank all the families who contributed to the study, the staff at the NIHR-Wellcome Trust Clinical Research Facility, Cambridge the NIHR Cambridge Comprehensive Biomedical Research Centre, and the midwives at the Rosie Maternity Hospital, Cambridge, UK. Sophia Pluto The authors want to thank all parents for their participation in the Sophia Pluto study and greatly acknowledge the research nurses Janneke van Nieuwkasteele, Christel Bruinings-Vroombout, Marianne Huibregtse-Schouten, Esther Lems, Naomi Khieroe, Suzanne Besteman-Voortman, Jose Bontenbal-van de Wege. Funding This work was supported by the EU Commission to the JPI HDHL program ‘Call III Biomarkers’ for the project: BioFN - Biomarkers for Infant Fat Mass Development and Nutrition (Grant agreement No 696295), administrated in the Netherlands by the Netherlands Organisation for Health Research and Development (ZonMW) (grant number 529051013) and UK by the Biotechnology and Biological Sciences Research Council (BB/P028195/1). This study was also supported by the National Institute for Health Research (NIHR) Cambridge Biomedical Research Centre (IS-BRC-1215-20014). KKO is supported by Medical Research Council Epidemiology Unit (MC_UU_12015/2 and MC_UU_00006/2). The views expressed are those of the authors and not necessarily those of the funders. The Sophia Pluto study is an investigator-initiated cohort study, for which A.C.S.H.K. received an independent research grant by Danone Nutricia Research. All CBGS and Sophia Pluto study sponsors had no role in the study design, collection, analysis or interpretation of the data, the writing of the manuscript or the decision to submit it for publication. Figure 1 Bland-Altman plots showing mean bias (solid line) and limits of agreement (LOA, represented by 95% CI; dotted horizontal line) between FM values estimated by the CBGS equation (without skinfolds) versus values measured by ADP-PEA Pod as the criterion method among Sophia Pluto infants at age 3 (A) and 6 months (B) Table 1 Published anthropometry-based prediction equations for body composition in children First author (year) Prediction equation SFT site(s) Participants age Reference method Slaughter et al (1988)16 Boys: % BF = 1.21 x Sum SFT – 0.008 x (Sum SF)2 – 1.7 Girls: % BF = 1.33 x Sum SFT – 0.013 x (Sum SF)2 – 2.5 Triceps and Subscapular (mm) 8-29 years old Underwater weighing to measure body density and deuterium oxide dilution to measure body water Aris et al (2013)23 FM (kg) = -0.022 + 0.307 * Weight (kg) - 0.077 x Sex (1 = boy; 0 = girl) - 0.019 x GA (weeks)+ 0.028 x SFT Subscapular (mm) 1-3 days old ADP-PEA Pod Catalano et al (1995)30 FM (kg) = 0.54657 + 0.39055 x Weight (kg) + 0.0453 x SFT – 0.03237 x Length (cm) Flank (mm) 1-3 days old TOBEC SFT=skinfold thickness, FM=fat mass, FFM=fat-free mass, GA=gestational age, TOBEC=total body electrical conductivity Table 2 Baseline cohorts’ characteristics by sex Descriptive Cohort CBGS (N=77) Sophia Pluto (N=571) p value * Boys (55%) Girls (45%) Boys (54%) Girls (46%) Boys Girls Birth GA (weeks) 40.05 ± 1.17 40.13 ± 1.15 39.65 ± 1.24 39.77 ± 1.24 0.048 0.008 Weight (kg) 3.55 ± 0.47 3.47 ± 0.4 3.45 ± 0.49# 3.31 ± 0.5 0.211 0.072 Length (cm)^ 50.79 ± 1.89 50.21 ± 1.55 50.84 ± 2.18 49.91 ± 2.04 0.889 0.412 3 months Weight (kg) 6.29 ± 0.73# 5.65 ± 0.54 6.26 ± 0.70# 5.70 ± 0.69 0.795 0.682 Length (cm) 60.87 ± 2.12# 59.58 ± 1.74 61.95 ± 2.07# 60.23 ± 2.18 <0.010 0.089 FM (kg) 1.42 ± 0.37 1.29 ± 0.39 1.42 ± 0.41# 1.32 ± 0.4 1.000 0.675 FFM (kg) 4.78 ± 0.42# 4.42 ± 0.47 4.84 ± 0.47# 4.37 ± 0.42 0.430 0.516 FMI (kg/m2) 3.85 ± 0.98 3.62 ± 1.04 3.68 ±1.01 3.62 ± 1.02 0.313 0.96 FFMI (kg/m2) 12.95 ± 0.88# 12.41 ± 0.95 12.57 ±.0.91# 12.05 ±0.86 0.011 0.022 Values are mean ± SD GA=gestational age, FM=fat mass, FFM=fat-free mass ^ Birth length available in Sophia Pluto cohort: boys n=210, girls n=152 FMI=fat mass index, calculated by dividing FM (kg) by length squared (m2)26 FFMI=fat-free mass index, calculated by dividing FFM (kg) by length squared (m2)26 p values are based on independent T-test * p value between CBGS and Sophia Pluto of the same genders (i.e. Boys= CBGS boys vs Sophia Pluto boys, Girls=CBGS girls vs Sophia Pluto girls) # significantly different between boys and girls (p<0.005) in the same infant group (i.e. CBGS boys vs girls, Sophia Pluto boys vs girls) Table 3 CBGS-derived equations to predict ADP-PEA Pod measured infant FM and FFM combining 6 weeks and 3 months measurements Model Wt (Kg) L (cm) Sex Age (days) SFT-t (mm) SFT-s (mm) SFT-f (mm) Constant R2 RMSE Dependent variable: FM (Kg) 1 Wt+L+Sex +Age B±SE 0.624 ±0.06* -0.088 ±0.02* -0.07 ±0.05 0.001 ±0.002 2.87 ±0.79 0.63 0.262 2 Model 1 +SFT B±SE 0.512 ±0.08* -0.074 ±0.02* -0.037 ±0.05 0.002 ±0.002 0.041 ±0.02 0.008 ±0.03 0.011 ±0.02 2.167 ±0.86 0.65 0.258 Dependent variable: FFM (Kg) 1 Wt+L+Sex +Age B±SE 0.407 ±0.06* 0.067 ±0.02* 0.034 ±0.05 -0.002 ±0.002 -1.703 ±0.83 0.77 0.276 2 Model 1 +SFT B±SE 0.528 ±0.08* 0.052 ±0.02* -0.001 ±0.06 -0.002 ±0.002 -0.005 ±0.03 -0.014 ±0.02 -0.046 ±0.02* -0.954 ±0.9 0.79 0.271 FM = Fat mass, FFM = Fat-free mass, Wt = Weight, L = Length (cm), Sex (1=male, 0=female), SFT-t = Triceps Skinfold thickness, SFT-s = Subscapular, SFT-f = Flank (suprailiac) B = unstandardized beta, SE = standard error, RMSE = Root mean squared error * p<0.05 for statistically significant B Based on 119 infant measurements at ages 5-16 weeks Table 4 FM and FFM values predicted by anthropometry-based equations versus measured by ADP-PEA Pod among Sophia Pluto infants Slaughter et al Aris et al Catalano et al CBGS(no SFT) CBGS (with SFT) ADP-PEA Pod Age 3 months N = 571 (264 girls), mean age = 92.3 days FM (kg) Boys: 1.00±0.27 Girls: 0.88±0.23 1.23±0.24 1.25±0.30 1.29±0.34 1.39±0.35a All: 1.37±0.41 Boys: 1.42±0.41 Girls: 1.32±0.4 FFM (kg) NA NA NA 4.68±0.43 4.74±0.45 All: 4.62±0.5 Boys: 4.84±0.47 Girls: 4.37±0.42 Age 6 months N = 447 (211 girls), mean age = 183.4 days FM (kg) Boys: 1.28±0.32 Girls: 1.18±0.31 1.73±0.29 1.69±0.36 1.84±0.43 1.96±0.42 All: 1.86±0.51 Boys: 1.85±0.51 Girls: 1.86±0.51 FFM (kg) NA NA NA 5.62±0.49 5.77±0.52b All: 5.76±0.58 Boys: 6.02±0.53 Girls: 5.46±0.49 Values are mean±SD Paired t-test (compared to ADP-PEA Pod), all p<0.05, except ap=0.402 and bp=0.171 FM=fat-mass, FFM=fat-free mass, NA=not applicable Table 5 Bland and Altman and regression analyses of body composition values estimated by prediction equations against ADP-PEA Pod measurements Correlation Bland-Altman Proportional Bias Correlation between mean and difference* Pearson R p Mean Bias LOA (95% CI) B+SE p Pearson R p Age 3 months FM Slaughter et al (boys) 0.736 <0.001 -0.422 -0.987, 0.143 -0.470±0.043 <0.001 -0.535 <0.001 Slaughter et al (girls) 0.730 <0.001 -0.440 -1.000, 0.123 -0.613±0.045 <0.001 -0.645 <0.001 Aris et al 0.798 <0.001 -0.151 -0.675, 0.373 -0.570±0.026 <0.001 -0.676 <0.001 Catalano et al 0.800 <0.001 -0.131 -0.628, 0.366 -0.348±0.027 <0.001 -0.472 <0.001 CBGS-no SFT 0.785 <0.001 -0.093 -0.6, 0.414 -0.207±0.029 <0.001 -0.288 <0.001 CBGS-with SFT 0.794 <0.001 0.008 -0.489, 0.505 -0.189±0.028 <0.001 -0.271 <0.001 FFM CBGS-no SFT 0.867 <0.001 0.049 -0.453, 0.551 -0.178±0.022 <0.001 -0.318 <0.001 CBGS-with SFT 0.872 <0.001 0.099 -0.394, 0.592 -0.125±0.022 <0.001 -0.234 <0.001 Age 6 months FM Slaughter et al (boys) 0.722 <0.001 -0.589 -1.317, 0.139 -0.600±0.049 <0.001 -0.630 <0.001 Slaughter et al (girls) 0.733 <0.001 -0.677 -1.382, 0.029 -0.592±0.050 <0.001 -0.637 <0.001 Aris et al 0.755 <0.001 -0.144 -0.860, 0.572 -0.718±0.032 <0.001 -0.736 <0.001 Catalano et al 0.774 <0.001 -0.192 -0.852, 0.468 -0.469±0.032 <0.001 -0.568 <0.001 CBGS-no SFT 0.767 <0.001 -0.048 -0.702, 0.606 -0.275±0.034 <0.001 -0.361 <0.001 CBGS-with SFT 0.789 <0.001 0.084 -0.545, 0.713 -0.300±0.032 <0.001 -0.410 <0.001 FFM CBGS-no SFT 0.821 <0.001 -0.171 -0.833, 0.491 -0.261±0.029 <0.001 -0.390 <0.001 CBGS-with SFT 0.832 <0.001 -0.021 -0.663, 0.621 -0.188±0.029 <0.001 -0.299 <0.001 * Correlation between the mean (of the reference/ADP-PEA Pod and each alternative equation) and the difference between methods FM=fat mass, FFM=fat-free mass, SFT=skinfold thicknesses, B=unstandardized beta, LOA=limit of agreement, CI=confidence interval, SE=standard error of B Contributors LO, IAvB, EDLR and ACHK had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. DBD, KKO, and IAH were involved in the conduction of the CBGS and ACHK for Sophia Pluto. LO and IAvB were responsible for infant recruitment and clinic visit in the CBGS and Sophia Pluto, respectively. LO and IAvB performed statistical analyses. LO and EDLR drafted the manuscript. IAvB, ACHK, KKO, DBD, and IAH helped to improve the manuscript. All authors contributed to interpretation of data, critically revised the article for important intellectual content, and approved the final version. Conflict of interest The authors declare no conflict of interest. 1 Barker DJ Early growth and cardiovascular disease Arch Dis Child 1999 80 305 306 10.1136/adc.80.4.305 10086930 2 Hales CN Barker DJ Clark PM Fetal and infant growth and impaired glucose tolerance at age 64 BMJ 1991 303 6809 1019 1022 10.2337/dc10-1141 1954451 3 Hales CN Barker DJP Type 2 (non-insulin-dependent) diabetes mellitus: the thrifty phenotype hypothesis Int J Epidemiol 2013 42 5 1215 1222 10.1093/ije/dyt133 24159065 4 Hanson MA Gluckman PD Early developmental conditioning of later health and disease: physiology or pathophysiology? Physiol Rev 2014 94 Table 1 1027 1076 10.1152/physrev.00029.2013 25287859 5 Larqué E Labayen I Flodmark CE From conception to infancy — early risk factors for childhood obesity Nat Rev Endocrinol 2019 15 8 456 478 10.1038/s41574-019-0219-1 31270440 6 Gingras V Hivert MF Oken E Early-Life Exposures and Risk of Diabetes Mellitus and Obesity Curr Diab Rep 2018 18 89 1 10 10.1007/s11892-018-1050-0 29362940 7 Mameli C Mazzantini S Zuccotti GV Nutrition in the first 1000 days: The origin of childhood obesity Int J Env Res Public Heal 2016 13 838 1 9 10.3390/ijerph13090838 8 Freedman DS Horlick M Berenson GS A comparison of the Slaughter skinfold-thickness equations and BMI in predicting body fatness and cardiovascular Am J Clin Nutr 2013 98 3 1417 1424 10.3945/ajcn.113.065961.1 24153344 9 Breij LM Kerkhof GF De Lucia Rolfe E Longitudinal fat mass and visceral fat during the first 6 months after birth in healthy infants: support for a critical window for adiposity in early life Pediatr Obes 2017 12 4 286 294 10.1111/ijpo.12139 27072083 10 Tennefors C Forsum E Assessment of body fatness in young children using the skinfold techique and BMI vs body water dilution Eur J Clin Nutr 2004 58 3 541 547 10.1038/sj.ejcn.1601842 14985694 11 Bell KA Wagner CL Perng W Feldman HA Shypailo RJ Belfort MB Validity of Body Mass Index as a Measure of Adiposity in Infancy J Pediatr 2018 196 168 174 e1 10.1016/j.jpeds.2018.01.028 29551311 12 Schmelzle H Fusch C Body fat in neonates and young infants: validation of skinfold thickness versus dual-energy X-ray absortiometry Am J Clin Nutr 2002 February 76 1096 1100 12399284 13 Andres A Gomez-Acevedo H Badger TM Quantitative nuclear magnetic resonance to measure fat mass in infants and children Obesity 2011 19 10 2089 2095 10.1038/oby.2011.215 21779094 14 Tint MT Ward LC Soh SE Estimation of fat-free mass in Asian neonates using bioelectrical impedance analysis Br J Nutr 2016 115 6 1033 1042 10.1017/S0007114515005486 26856420 15 Fiorotto ML Cochran WJ Klish WJ Fat-free mass and total body water of infants estimated from total body electrical conductivity measurements1 Pediatr Res 1987 22 4 417 421 10.1203/00006450-198710000-00011 3684373 16 Slaughter M Lohman T Boileau R Skinfold equations for estimation of body fatness in children and youth Hum Biol 1988 60 5 709 723 3224965 17 Demerath EW Fields DA Body composition assessment in the infant Am J Hum Biol 2014 26 3 291 304 10.1002/ajhb.22500 24424686 18 Andrews ET Beattie RM Johnson MJ Measuring body composition in the preterm infant: Evidence base and practicalities Clin Nutr 2019 38 6 2521 2530 10.1016/j.clnu.2018.12.033 30737045 19 Fields DA Demerath EW Pietrobelli A Chandler-Laney PC Body composition at 6 months of life: Comparison of air displacement plethysmography and dual-energy X-ray absorptiometry Obesity 2012 20 11 2302 2306 10.1038/oby.2012.102 22522885 20 Koo WWK Hammami M Hockman EM Interchangeability of pencil-beam and fan-beam dual-energy X-ray absorptiometry measurements in piglets and infants Am J Clin Nutr 2003 78 2 236 240 10.1093/ajcn/78.2.236 12885703 21 Picaud JC Nyamugabo K Braillon P Dual-energy x-ray absorptiometry in small subjects: Influence of dual-energy x-ray equipment on assessment of mineralization and body composition in newborn piglets Pediatr Res 1999 46 6 772 777 10.1203/00006450-199912000-00011 10590038 22 Picaud JC Rigo J Nyamugabo K Milet J Senterre J Evaluation of dual-energy X-ray absorptiometry for body-composition assessment in piglets and term human neonates Am J Clin Nutr 1996 63 2 157 163 10.1093/ajcn/63.2.157 8561054 23 Aris IM Soh SE Tint MT Body fat in Singaporean infants: Development of body fat prediction equations in Asian newborns Eur J Clin Nutr 2013 67 9 922 927 10.1038/ejcn.2013.69 23549200 24 Fields DA Gilchrist JM Catalano PM Gianní ML Roggero PM Mosca F Longitudinal body composition data in exclusively breast-fed infants: A multicenter study Obesity 2011 19 9 1887 1891 10.1038/oby.2011.11 21311509 25 de Fluiter KS van Beijsterveldt IALP Goedegebuure WJ Longitudinal body composition assessment in healthy term-born infants until 2 years of age using ADP and DXA with vacuum cushion Eur J Clin Nutr 2020 74 4 642 650 10.1038/s41430-020-0578-7 32055012 26 Diet, Anthropometry and Physical Activity (DAPA) Measurement Toolkit https://dapa-toolkit.mrc.ac.uk/ 27 Deierlein AL Thornton J Hull H Paley C Gallagher D An anthropometric model to estimate neonatal fat mass using air displacement plethysmography Nutr Metab 2012 9 1 5 10.1186/1743-7075-9-21 28 de Bruin NC van Velthoven KAM Stijnen T Juttmann RE Degenhart HJ Visser HKA Body fat and fat-free mass in infants: New and classic anthropometric indexes and prediction equations compared with total-body electrical conductivity Am J Clin Nutr 1995 61 6 1195 1205 10.1093/ajcn/61.6.1195 7762517 29 Josefson JL Nodzenski M Talbot O Scholtens DM Catalano P Fat mass estimation in neonates: Anthropometric models compared with air displacement plethysmography Br J Nutr 2019 121 3 285 290 10.1017/S0007114518003355 30444206 30 Catalano PM Thomas AJ Avallone DA Amini SB Anthropometric estimation of neonatal body composition Am J Obstet Gynecol 1995 173 4 1176 1181 10.1016/0002-9378(95)91348-3 7485315 31 Prentice P Acerini CL Eleftheriou A Hughes IA Ong KK Dunger DB Cohort profile: The Cambridge Baby Growth Study (CBGS) Int J Epidemiol 2016 45 1 35 35g 10.1093/ije/dyv318 32 Fomon SJ Haschke F Ziegler EE Nelson SE Body composition of reference children from birth to age 10 years Am J Clin Nutr 1982 35 5 Suppl 1169 1175 10.1093/ajcn/35.5.1169 7081099 33 Lingwood BE Storm van Leeuwen AM Carberry AE Prediction of fat-free mass and percentage of body fat in neonates using bioelectrical impedance analysis and anthropometric measures: validation against the PEA POD Br J Nutr 2012 107 10 1545 1552 10.1017/S0007114511004624 21917194 34 Alexander T Conlon CA Gamble G Body composition of New Zealand-born term babies differs by ethnicity, gestational age and sex Early Hum Dev 2020 November 140 2 6 2019 10.1016/j.earlhumdev.2019.104924 35 Pastorino S Bishop T Crozier SR Associations between maternal physical activity in early and late pregnancy and offspring birth size: remote federated individual level meta-analysis from eight cohort studies BJOG 2019 126 4 459 470 10.1111/1471-0528.15476 30230190 36 Ellis KJ Yao M Shypailo RJ Urlando A Wong WW Heird WC Body-composition assessment in infancy: Air-displacement plethysmography compared with a reference 4-compartment model Am J Clin Nutr 2007 85 1 90 95 10.1016/S8756-5005(08)79133-9 17209182 37 Toro-Ramos T Paley C Pi-Sunyer FX Gallagher D Body composition during fetal development and infancy through the age of 5 years Eur J Clin Nutr 2015 69 12 1279 1289 10.1038/ejcn.2015.117 26242725

PMC007xxxxxx/PMC7614815.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0133660 Clin Biochem Clin Biochem Clinical biochemistry 0009-9120 1873-2933 31972148 7614815 10.1016/j.clinbiochem.2020.01.003 EMS114755 Article Circulating PCSK9 is associated with liver biomarkers and hepatic steatosis Paquette Martine M.Sc a Gauthier Dany B.Sc a Chamberland Ann B.Sc b Prat Annik PhD b De LuciaRolfe Emanuella PhD c Rasmussen Jon J. MD, PhD de Kaduka Lydia PhD f Seidah Nabil G. PhD b Bernard Sophie MD, PhD ag Christensen Dirk L. PhD ch Baass Alexis MD, M.Sc ai* a Lipids, Nutrition and Cardiovascular Prevention Clinic of the Montreal Clinical Research Institute (Montreal, Canada) b Laboratory of Biochemical Neuroendocrinology of the Montreal Clinical Research Institute (Montreal, Canada) c Medical Research Council Epidemiology Unit, University of Cambridge School of Clinical Medicine, Institute of Metabolic Science, Cambridge Biomedical Campus, (Cambridge, United Kingdom) d Centre of Endocrinology and Metabolism, Department of Internal Medicine, Copenhagen University Hospitals (Herlev and Gentofte, Denmark) e Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen (Copenhagen, Denmark) f Centre for Public Health Research, KEMRI (Nairobi, Kenya) g Department of Medicine, Division of Endocrinology, Université de Montreal (Montreal, Canada) h Department of Public Health, University of Copenhagen (Copenhagen, Denmark) i Department of Medicine, Divisions of Experimental Medicine and Medical Biochemistry, McGill University (Montreal, Canada) * To whom correspondence should be addressed: Alexis Baass, Lipids, Nutrition and Cardiovascular Prevention Clinic of the Montreal Clinical Research Institute, 110 avenue des Pins Ouest, Montreal, (Québec, Canada) H2W 1R7. Phone: 1-514-987-5650, Fax: 514-987-5689, alexis.baass@ircm.qc.ca 01 3 2020 20 1 2020 26 1 2021 26 7 2023 77 2025 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Background In parallel to the increasing prevalence of metabolic syndrome, the prevalence of hepatic steatosis has also increased dramatically worldwide. Hepatic steatosis is a major risk factor of hepatic cirrhosis, cardiovascular disease and type 2 diabetes. Circulating levels of proprotein convertase subtilisin/kexin type 9 (PCSK9) have been positively associated with the metabolic syndrome. However, the association between PCSK9 and the liver function is still controversial. Objective The objective of this study is to investigate the association between circulating PCSK9 levels and the presence of hepatic steatosis, as well as with liver biomarkers in a cohort of healthy individuals. Methods Total PCSK9 levels were measured by an in-house ELISA using a polyclonal antibody. Plasma albumin, alkaline phosphatase, ALT, AST, total bilirubin and GGT were measured in 698 individuals using the COBAS system. The presence of hepatic steatosis was assessed using ultrasound liver scans. Results In a multiple regression model adjusted for age, sex, insulin resistance, body mass index and alcohol use, circulating PCSK9 level was positively associated with albumin (β=0.102, P=0.008), alkaline phosphatase (β=0.201, P<0.0001), ALT (β=0.238, P<0.0001), AST (β=0.120, P=0.003) and GGT (β=0.103, P=0.007) and negatively associated with total bilirubin (β= -0.150, P<0.0001). Tertile of circulating PCSK9 was also associated with hepatic steatosis (OR 1.48, 95% CI 1.05-2.08, P=0.02). Conclusion Our data suggest a strong association between PCSK9 and liver biomarkers as well as hepatic steatosis. Further studies are needed to explore the role of PCSK9 on hepatic function. Liver hepatic steatosis PCSK9 metabolic syndrome insulin resistance pmc1 Introduction Non-alcoholic fatty liver disease (NAFLD) is a spectrum of progressive liver diseases that begins with an aberrant deposition of triglycerides (TG) in the hepatocytes (hepatic steatosis). Hepatic steatosis is defined as the presence of triglycerides droplets in more than ~ 5% of hepatocytes [1,2]. In some cases, the disease progresses to non-alcoholic steatohepatitis (NASH), which is characterized by hepatocyte ballooning and the presence of inflammation, with or without collagen deposition (fibrosis). NASH can then progress in up to 1/3 of cases to cirrhosis, which is a major risk factor for hepatocellular carcinoma and liver failure [3]. The worldwide prevalence of NAFLD has been estimated to be around 25-35% [4,5]. NAFLD represents the most frequent cause of liver enzymes elevation in plasma [6]. Hepatic steatosis occurs when there is an increased input or synthesis of triglycerides in the liver and/or a decreased output of triglycerides from the liver. The principal modifiable risk factors for the development of NAFLD include obesity, insulin resistance, diet (in particular, carbohydrate or fructose excess) and sedentary lifestyle [5, 7–9]. Hepatic steatosis is now recognized as the hepatic manifestation of the metabolic syndrome (MetS) [10–12]. Similar to the MetS, NAFLD has also been associated with an increased risk of cardiovascular disease (CVD) and type 2 diabetes [13]. To date, the sole effective treatment recommended for NAFLD remains weight loss via caloric restriction and exercise [14]. Proprotein convertase subtilisin/kexin type 9 (PCSK9) is a key modulator of the degradation of the LDL receptor (LDLR) [15]. Gain-of-function mutations in PCSK9 have been associated with familial hypercholesterolemia and increased cardiovascular risk whereas loss-of-function mutations have been associated with low LDL-C and cardiovascular risk [16]. Furthermore, plasma PCSK9 as been independently associated with each individual components of the lipid profile, the stronger associations being with IDL and triglycerides [17,18]. Since PCSK9 has been positively associated with the MetS, as well as with the individual MetS components and insulin resistance [17,19], it is logical to assume that PCSK9 would also be associated with the presence of NAFLD or with anomalies in liver enzymes. However, some controversies exist in the literature concerning the effect of PCSK9 on the liver function, both for studies conducted in mice [20–22] and in humans. Indeed, several small case report studies have previously linked the presence of PCSK9 loss-of-function mutation with the presence of liver steatosis [23–25], whereas other authors report a normal liver function or perhaps even protection against steatosis in PCSK9 loss-of-function carriers [26,27]. In PCSK9 inhibitors trials, no liver adverse effects have been reported to date. Furthermore, in FOURIER trial, there was no difference between the experimental and the control groups concerning the frequency of aminotransferase level > 3 times the upper limit of the normal range [28]. Similarly, in studies evaluating the association between circulating plasma PCSK9 levels and the presence of hepatic steatosis, some studies reported a positive association [29,30], whereas others did not detect any effect of plasma PCSK9 on hepatic steatosis or circulating liver enzymes [31]. The objectives of the present study is therefore to investigate the association between circulating PCSK9 and circulating liver biomarkers as well as the presence of hepatic steatosis in a large cohort of Kenyan individuals. 2 Materials and Methods 2.1 Study Population and Recruitment The cross-sectional Kenya Diabetes Study, conducted between August 2005 and January 2006, comprised a total cohort of 1459 Kenyans individuals from three rural districts and the capital city of Nairobi. The study participants were recruited on a voluntary basis, following information given at local community meetings. The local social mobilizers who recruited the participants described the study as a diabetes investigation. A more detailed description of the selection procedure has been presented elsewhere [32,33]. Inclusion criteria were ≥ 17 years of age and being Luo, Kamba, or Maasai in the rural districts whereas the urban population was of mixed ethnic origin. Exclusion criteria included pregnancy, serious illnesses (e.g.: malaria), inability to walk unassisted and severe mental disease. None of the participants were treated with lipid-lowering medication, oral hypoglycemic agents or insulin therapy. Ten participants were taking hypotensive medication. In this cohort of the Kenya Diabetes Study, ultrasound liver scans were performed in an opportunity sample of 875 participants. Of these, 767 liver scans were valid. Analyses of circulating PCSK9 was performed in a subgroup of 1338 participants based on the availability of both the biological samples and the clinical data. We then performed the liver biomarkers measurements in participants with both a valid liver scan result and a PCSK9 value, for a total of 698 participants included in the present manuscript. Information about the study was given to all participants before obtaining their informed consent (written or oral) in their preferred language (Kiswahili, English and/or other local tribal languages). The study was approved by both the National Ethical Review Committee of Kenya and the Danish National Committee on Biomedical Research Ethics. 2.2 Data Collection and Biochemical Analysis 2.2.1 PCSK9 Total circulating PCSK9 concentration was analysed by an in-house ELISA using a polyclonal antibody against human PCSK9 as previously described [34,35]. For this assay, defrosted samples were used. Measurements were performed at the Montreal Clinical Research Institute (Montreal, Canada). 2.2.2 Liver Biomarkers Albumin, ALP, ALT, AST, total bilirubin and GGT were measured by an automated analyzer (COBAS INTEGRA 400, Roche Diagnostic) on defrosted samples. Measurements were performed at the Montreal Clinical Research Institute (Montreal, Canada). 2.2.3 Hepatic steatosis The presence of hepatic steatosis has been assessed using ultrasound liver scans. This method is semi-quantitative and allows to distinguish between normal liver (score ≤ 4) or mild (score between 5-7), moderate (score between 8-10) and severe (score ≥ 11) steatosis according to standardised criteria. Interpretation of the ultrasound liver scans was done at University of Cambridge by a single reviewer (Cambridge, United Kingdom) [36]. The operator was blinded to all other study measures. In addition, the hepatic steatosis index (HSI) and the fatty liver index (FLI) were calculated as described previously [37,38], where HSI ≥ 36 and FLI ≥ 60 were indicator of hepatic steatosis. 2.2.4 Other Clinical Variables Method for the measurement of blood pressure, lipid profile, anthropometric measurements and glucose metabolism parameters has been described in details elsewhere [17]. All blood samples were collected in the morning following an 8-h overnight fast. The centrifuged samples were kept on ice in Kenya before to be shipped on dry ice and stored at -80 °C. Homeostasis model assessment of insulin resistance (HOMA-IR) was calculated using fasting glucose and fasting insulin values. The presence of metabolic syndrome was assessed using the International Diabetes Federation (IDF) 2009 consensus statement criteria [39]. Body mass index (BMI, kg/m2) was calculated. Information concerning alcohol consumption was self-reported by the participants. 2.3 Statistical Analysis The IBM SPSS Statistics version 25 (IBM Corp, Armonk, NY) was used for statistical analysis. A statistical significance level was established at P ≤ 0.05. P values are two-sided. Depending on the nature of the data (continuous or categorical), results are presented either as mean +/- standard deviation or n (percentage). For continuous variables with a skewed distribution, data are presented as median with interquartile range (Q1-Q3). Abnormally distributed variables were log-transformed prior to analysis. In order to compare the baseline characteristics between the participants with hepatic steatosis and those without, a Student’s t-test was performed for continuous variables, whereas a Chi2 test was used for categorical variables. The relationship between liver biomarkers and PCSK9 was assessed through linear regression models, whereas the association between the presence of hepatic steatosis and PCSK9 tertiles was determined via logistic regression models. The P value for the prevalence of hepatic steatosis by PCSK9 tertiles was obtained by a Chi2 analysis. 3 Results 3.1 Description of the study cohort The participants’ characteristics according to the presence or the absence of hepatic steatosis are presented in Table 1. The presence of hepatic steatosis on the ultrasound liver scan was observed in 104 participants (15%). For the remaining individuals of the study cohort, no sign of hepatic steatosis was observed (n=594 (85%)). The group with hepatic steatosis was five years older and had significantly lower proportion of men vs women (28% vs 44%, respectively) than the group without hepatic steatosis (P<0.05). In addition, we observed significant differences between groups for many metabolic parameters. The group with hepatic steatosis had higher BMI, waist circumference, blood pressure, fasting glucose, insulinemia, glycated hemoglobin (HbA1c), HOMA-IR, total cholesterol, LDL-C, triglycerides, PCSK9, prevalence of self-reported diabetes, prevalence of metabolic syndrome and average number of MetS criteria and lower HDL-C than the group without hepatic steatosis (P<0.05). There were significant differences between groups for only three liver biomarkers. Indeed, the group with hepatic steatosis had lower albumin, AST and total bilirubin levels than the group without hepatic steatosis (P<0.05). There was no significant difference between groups concerning alcohol consumption (P=0.46). 3.2 Relation between circulating PCSK9 levels and liver biomarkers The associations between PCSK9 and liver biomarkers according to several regression models are presented in Table 2. In an unadjusted model, only ALP, ALT and total bilirubin were significantly associated with PCSK9. When adjustment for age and sex (Model 2) or for age, sex, HOMA-IR, BMI and alcohol use (Model 3) were applied, all liver biomarkers were strongly associated with PCSK9 (In model 3: β=0.102 for albumin, β= 0.201 for ALP, β=0.238 for ALT, β=0.120 for AST, β=-0.150 for total bilirubin and β=0.103 for GGT, P values < 0.05). 3.3 Relation between circulating PCSK9 levels and the presence of hepatic steatosis The associations between PCSK9 tertiles and the presence of hepatic steatosis according to several regression models are presented in Table 3. PCSK9 was significantly associated with the presence of hepatic steatosis in the unadjusted model (OR 1.74, 95% CI 1.32-2.28, P<0.0001) as well as in the model adjusted for age, sex, HOMA-IR, BMI and alcohol use (OR 1.48, 95% CI 1.05-2.08, P=0.02). The prevalence of hepatic steatosis in the three tertiles was 8.7% in the lowest tertile, 14.0% in the intermediate tertile and 22.3% in the highest tertile (Figure 1A). Similar associations were observed for the prevalence of HSI ≥ 36 (P=0.008) and FLI ≥ 60 (P=0.02) between PCSK9 tertiles (Figure 1 B and C). The association between LDL-C and PCSK9 levels was statistically significant in the group without hepatic steatosis (β=0.224, P<0.0001), but not in subjects with hepatic steatosis (β=0.161, P=0.11) (Supplemental Table 1). 4 Discussion In the present study, including 698 participants from the Kenya Diabetes Study, we demonstrated that circulating PCSK9 levels were strongly associated with all circulating liver biomarkers as well as the presence of hepatic steatosis, as determined by ultrasound liver scans, HSI and FLI indexes. These associations were independent of insulin resistance status, although the strength of the association between PCSK9 and hepatic steatosis was somewhat decreased when corrected for HOMA-IR, BMI and alcohol use. This suggests that PCSK9 could have and indirect effect on hepatic steatosis via insulin resistance as well as a direct independent effect. In a previous publication in the same cohort, PCSK9 was also associated with the presence of metabolic syndrome, as well as with each individual components of the lipid profile [17]. Elevated level of liver enzymes is considered as an indicator of abnormal liver function. Thus, the screening for NAFLD stages using liver biopsy is usually reserved for individuals with elevated levels of circulating liver enzymes. However, liver enzymes are not systematically correlated with the level of liver fat and could be linked to other cardiometabolic pathways. For example, ALT levels would be normal in the majority (79%) of subjects who have increased liver fat [5]. In the present cohort, when the associations between the presence of hepatic steatosis and transaminases level were corrected for age, sex, HOMA-IR, BMI and alcohol use, these associations were non-significant (data not shown). Also, there exists some evidence suggesting that GGT itself would be associated with an increased incidence of new-onset metabolic syndrome and an increased incidence of CVD, independently of traditional risk factors [40]. However, the mechanisms underlying the action of GGT on metabolic risk remain unclear. Another group also reported a positive association between GGT and PCSK9 after adjustment for gender, age, type of diabetes, statin treatment, BMI, systolic blood pressure and HbA1c [41]. Further investigations should be done to improve our understanding of the independent role of PCSK9 on liver biomarkers. Ruscica et al (2016) observed an independent association between circulating PCSK9 and severity of steatosis in a cohort of 201 subjects who underwent liver biopsy for suspected NASH. The authors, as well as others, [42] suggested that this association would be explained by an activation of hepatic lipogenesis by PCSK9. However, they did not correct for the insulin resistance status or the presence of metabolic syndrome in the multivariate analysis. Furthermore, an association between circulating PCSK9 and ALT was found in univariate analysis, but the association was no longer significant in multivariate analysis [29]. In another study, Cariou et al [30] observed that a short-term high fructose diet was associated with a 27-93% increase in circulating PCSK9 levels in healthy volunteers and that PCSK9 concentration was positively associated with insulin resistance, liver steatosis and VLDL-TG. However, the authors performed Spearman’s correlation test, which is a univariate association. Therefore, whether the association between PCSK9 and hepatic steatosis is independent of insulin resistance in this study is not known. Furthermore, this association was not observed under baseline conditions [30]. On the other hand, in a study from the same group, there was no association between circulating PCSK9 and liver fat content, histological markers of NASH or transaminases level in a cohort of 478 high-risk patients with type 2 diabetes or metabolic syndrome and without lipid-lowering therapy [31]. The prevalence of hepatic steatosis found in our cohort is lower than the worldwide prevalence of 25-35% [4,5]. However, in a meta-analysis of NAFLD prevalence stratified by region, a pooled prevalence of 14% was found for Africa, which is in accordance with our own observations [4]. In the group with hepatic steatosis, 31% had MetS compared to 7% in the group without hepatic steatosis. The limitations of the present study include the method used to assess the presence of steatosis. Indeed, even if the ultrasound liver scan allows distinguishing between normal liver or mild, moderate and severe steatosis, the liver biopsy represents the gold standard for a proper diagnostic of the NAFLD stage. Also, the interpretation of the ultrasound scan has the disadvantage to be operator-dependent. Furthermore, our data concerning alcohol consumption are self-reported by the participants and under-reporting was suspected, especially among the women. Therefore, it was not possible to do adequate statistical correction for this variable. Similarly, we did not assess the presence of secondary causes of steatosis, such as the presence of other hepatic diseases that can cause steatosis and genetic predispositions. In addition, data on the presence of HIV/AIDS or other inflammatory or viral conditions that could influence plasma PCSK9 levels or hepatic function have not been systematically collected. Indeed, elevation of liver enzymes in HIV patients is frequent due to several reasons including antiretroviral therapy, coinfection with hepatitis B or C (up to 30% of cases), as well as concomitant alcohol, cocaine, or methamphetamine use [43]. Finally, because the study participants were recruited on a voluntary basis, and because only certain ethnic groups were selected, the sample probably does not represent the whole general Kenyan adult population. Therefore, the generalizability to other populations or ethnic groups needs to be confirmed. The major strength of our study is that the entire cohort was free of lipid-lowering therapy, oral hypoglycemic agents or insulin therapy. 5 Conclusions Results from our study confirm the association between PCSK9 and hepatic steatosis previously reported in other studies. Our study suggests that this effect could be mediated in part via the association between PCSK9 and insulin resistance as well as by a direct effect of PCSK9 on hepatic steatosis. Further research would be necessary to investigate the clinical implications of the association between PCSK9 and hepatic steatosis in the context of the use of PCSK9 inhibitors. Supplementary Material Table S1 6 Acknowledgements The authors want to thank the Montreal Clinical Research Institute (IRCM) research team and the nursing staff for their everyday help, support and implication. Furthermore, the authors are grateful to all participants, the local chiefs and sub-chiefs, the local elder councils, and district politicians as well as local laboratory technicians and assistants for excellent work throughout the data collection period. The authors acknowledge the permission by the Director of KEMRI to publish this article. 7 Funding Sources This work was supported by The Fondation Leducq Transatlantic Networks of Excellence [grant number 13CVD03], DANIDA [J. no.104.DAN.8–871, RUF project no. 91202], Cluster of International Health Grant (University of Copenhagen) and generous support for laboratory analyses from Steno Diabetes Center Copenhagen, Denmark and Section of Global Health, University of Copenhagen. The study funders had no role in the study design, in the collection, analysis and interpretation of data, in the writing of the report and in the decision to submit the article for publication. Figure 1 Prevalence of A) hepatic steatosis (ultrasound scan), B) HSI ≥ 36 and C) FLI ≥ 60 by tertile of circulating proprotein convertase subtilisin/kexin type 9 (PCSK9). HSI: hepatic steatosis index; FLI: fatty liver index. Table 1 Participants’ characteristics. Variables Absence of hepatic steatosis (n=594) Presence of hepatic steatosis (n=104) P value Age (y) 37 ± 11 42 ± 11 <0.0001 Sex (men (%)) 259 (44%) 29 (28%) 0.003 BMI (kg/m2) 20 (18-23) 27 (23-32) <0.0001 WC (cm) 74.7 (70.2-80.3) 93.2 (82.3-102.5) <0.0001 Systolic blood pressure (mmHg) 118 (110-127) 126 (116-139) <0.0001 Diastolic blood pressure (mmHg) 74 (67-80) 79 (73-86) <0.0001 Fasting glucose (mmol/L) 4.4 (4.0-4.7) 4.5 (4.1-5.0) 0.009 Fasting insulinemia (pmol/L) 22 (15-33) 39 (25-59) <0.0001 HbA1c (%) 5.0 (4.7-5.3) 5.4 (5.0-5.9) 0.001 HOMA-IR 0.6 (0.4-1.0) 1.2 (0.7-1.7) <0.0001 Total cholesterol (mmol/L) 3.83 ± 0.90 4.37 ± 1.11 <0.0001 LDL-C (mmol/L) 2.25 ± 0.74 2.74 ± 0.95 <0.0001 HDL-C (mmol/L) 1.15 ± 0.29 1.06 ± 0.25 0.003 Triglycerides (mmol/L) 0.82 (0.66-1.04) 1.02 (0.78-1.50) <0.0001 PCSK9 (ng/mL) 141.4 (118.4-170.4) 157.9 (131.7-189.5) 0.001 Self-reported diabetes (n(%)) 9/579 (2%) 5/103 (5%) 0.05 Self-reported CVD (n(%)) 37/562 (7%) 4/103 (4%) 0.38 MetS (n(%)) 42/577 (7%) 31/99 (31%) <0.0001 Number of MetS criteria 1.0 ± 0.9 2.1 ± 1.2 <0.0001 Albumin (g/L) 43.36 ± 4.33 42.36 ± 4.69 0.03 ALP (U/L) 74.6 (61.3-92.8) 78.7 (67.0-95.8) 0.16 ALT (U/L) 7.8 (6.0-9.9) 7.5 (5.6-10.6) 0.93 AST (U/L) 19.5 (16.7-23.4) 18.1 (14.9-22.1) 0.004 Total bilirubin (umol/L) 4.5 (3.0-7.4) 3.8 (2.3-5.9) 0.001 GGT (U/L) 17.6 (12.4-26.7) 20.2 (12.7-30.1) 0.16 Alcohol use (n(%)) 58/557 (10%) 8/100 (8%) 0.46 Data for continuous normally distributed variables are expressed as mean +/- SD, whereas continuous logarithmic variables are expressed as median (interquartile range). Categorical variables are expressed as frequency (n (%)). Bold type indicates P value ≤ 0.05. BMI, body mass index; CVD, cardiovascular disease; HbA1c, glycated hemoglobin; HDL-C, high-density lipoprotein cholesterol; HOMA-IR, homeostasis model assessment of insulin resistance; LDL-C, low-density lipoprotein cholesterol; MetS, metabolic syndrome; PCSK9: proprotein convertase subtilisin/kexin type 9; SD, standard deviation; WC, waist circumference. Table 2 Relation between circulating PCSK9 and liver biomarkers. Model 1 Model 2 Model 3 Liver biomarkers Standardized coefficient (β) P value Standardized coefficient (β) P value Standardized coefficient (β) P value Albumin -0.004 0.91 0.105 0.005 0.102 0.008 ALP 0.162 < 0.001 0.213 < 0.001 0.201 <0.0001 ALT 0.143 < 0.001 0.216 < 0.001 0.238 <0.0001 AST 0.041 0.28 0.085 0.03 0.120 0.003 Total bilirubin -0.242 < 0.001 -0.146 < 0.001 -0.150 <0.0001 GGT 0.052 0.18 0.104 0.005 0.103 0.007 P values for linear regression analysis. Bold type indicates P value ≤ 0.05. Model 1: Uncorrected Model 2: Corrected for age and sex Model 3: Corrected for age, sex, HOMA-IR, BMI and alcohol use. PCSK9 : proprotein convertase subtilisin/kexin type 9. Table 3 Relation between tertiles of circulating PCSK9 and hepatic steatosis. Prediction of hepatic steatosis Models OR 95% CI P value Model 1 1.74 1.32-2.28 <0.0001 Model 2 1.50 1.13-2.00 0.005 Model 3 1.48 1.05-2.08 0.02 P values for logistic regression analysis. Bold type indicates P value ≤ 0.05. Model 1: Uncorrected Model 2: Corrected for age and sex Model 3: Corrected for age, sex, HOMA-IR, BMI and alcohol use. PCSK9 : proprotein convertase subtilisin/kexin type 9. 8. Disclosures A.B. received research grants from Merck Frosst, Amgen, Sanofi, Astra Zeneca and the Fondation Leducq. He has participated in clinical research protocols from Pfizer, Regeneron Pharmaceuticals Inc., The Medecines Company, Amgen, Acasti Pharma Inc., Novartis, Sanofi, Ionis Pharmaceuticals, Inc., Astra Zeneca, Akcea and Merck Frosst. He has served on advisory boards and received honoraria for symposia from Amgen, Akcea and Sanofi. S.B. has participated in clinical research protocols from Akcea, The Medecines Company and Sanofi. She has served on advisory boards for Akcea, Novo Nordisk, Merck Frosst, Valeant Pharmaceuticals, Eli Lilly, Sanofi and Amgen and received honoraria for symposia from Akcea, Sanofi-aventis, Merck Frosst, Amgen, Novo Nordisk, Valeant Pharmaceuticals and Boehringer Ingelheim. D.L.C. has received consultancy payment from Novo Nordisk A/S. M.P., D.G., A.C., A.P., L.K., N.G.S., E.D.L.R. and J.J.R. have nothing to declare. The authors' contributions were as follows: All authors contributed to the discussion, analysis and interpretation of data and have reviewed the article for the intellectual content. M.P performed statistical analysis and has drafted the manuscript. All authors have approved the final article. A.B. had primary responsibility for final content. [1] Neuschwander-Tetri BA Caldwell SH Nonalcoholic steatohepatitis: summary of an AASLD Single Topic Conference Hepatology 2003 37 5 1202 19 12717402 [2] Szczepaniak LS Nurenberg P Leonard D Browning JD Reingold JS Grundy S Hobbs HH Dobbins RL Magnetic resonance spectroscopy to measure hepatic triglyceride content: prevalence of hepatic steatosis in the general population Am J Physiol Endocrinol Metab 2005 288 2 E462 8 15339742 [3] Argo CK Caldwell SH Epidemiology and natural history of non-alcoholic steatohepatitis Clin Liver Dis 2009 13 4 511 31 19818302 [4] Younossi ZM Koenig AB Abdelatif D Fazel Y Henry L Wymer M Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes Hepatology 2016 64 1 73 84 26707365 [5] Browning JD Szczepaniak LS Dobbins R Nuremberg P Horton JD Cohen JC Grundy SM Hobbs HH Prevalence of hepatic steatosis in an urban population in the United States: impact of ethnicity Hepatology 2004 40 6 1387 95 15565570 [6] Clark JM Brancati FL Diehl AM The prevalence and etiology of elevated aminotransferase levels in the United States Am J Gastroenterol 2003 98 5 960 7 12809815 [7] Cohen JC Horton JD Hobbs HH Human fatty liver disease: old questions and new insights Science 2011 332 6037 1519 23 21700865 [8] Semple RK Sleigh A Murgatroyd PR Adams CA Bluck L Jackson S Vottero A Kanabar D Charlton-Menys V Durrington P Soos MA Postreceptor insulin resistance contributes to human dyslipidemia and hepatic steatosis J Clin Invest 2009 119 2 315 22 19164855 [9] Zhu X Bian H Gao X The Potential Mechanisms of Berberine in the Treatment of Nonalcoholic Fatty Liver Disease Molecules 2016 21 10 E1336 [10] Marchesini G Brizi M Bianchi G Tomassetti S Bugianesi E Lenzi M McCullough AJ Natale S Forlani G Melchionda N Nonalcoholic fatty liver disease: a feature of the metabolic syndrome Diabetes 2001 50 8 1844 50 11473047 [11] Kotronen A Yki-Järvinen H Fatty liver: a novel component of the metabolic syndrome Arterioscler Thromb Vasc Biol 2008 28 1 27 38 17690317 [12] Bedogni G Miglioli L Masutti F Tiribelli C Marchesini G Bellentani S Prevalence of and risk factors for nonalcoholic fatty liver disease: the Dionysos nutrition and liver study Hepatology 2005 42 1 44 52 15895401 [13] Armstrong MJ Adams LA Canbay A Syn WK Extrahepatic complications of nonalcoholic fatty liver disease Hepatology 2014 59 3 1174 97 24002776 [14] Chalasani N Younossi Z Lavine JE Diehl AM Brunt EM Cusi K Charlton M Sanyal AJ The diagnosis and management of non-alcoholic fatty liver disease: practice Guideline by the American Association for the Study of Liver Diseases, American College of Gastroenterology, and the American Gastroenterological Association Hepatology 2012 55 6 2005 23 22488764 [15] Seidah NG Benjannet S Wickham L Marcinkiewicz J Jasmin SB Stifani S Basak A Prat A Chretien M The secretory proprotein convertase neural apoptosis-regulated convertase 1 (NARC-1): liver regeneration and neuronal differentiation Proc Natl Acad Sci U S A 2003 100 3 928 33 12552133 [16] Awan Z Baass A Genest J Proprotein convertase subtilisin/kexin type 9 (PCSK9): lessons learned from patients with hypercholesterolemia Clin Chem 2014 60 11 1380 9 25248569 [17] Paquette M Luna Saavedra YG Chamberland A Prat A Christensen DL Lajeunesse-Trempe F Kaduka L Seidah NG Dufour R Baass A Association Between Plasma Proprotein Convertase Subtilisin/Kexin Type 9 and the Presence of Metabolic Syndrome in a Predominantly Rural-Based Sub-Saharan African Population Metab Syndr Relat Disord 2017 15 8 423 9 28749722 [18] Kwakernaak AJ Lambert G Dullaart RP Plasma proprotein convertase subtilisin-kexin type 9 is predominantly related to intermediate density lipoproteins Clin Biochem 2014 47 7-8 679 82 24680982 [19] Awan Z Dubuc G Faraj M Dufour R Seidah NG Davignon J Rabasa-Lhoret R Baass A The effect of insulin on circulating PCSK9 in postmenopausal obese women Clin Biochem 2014 47 12 1033 9 24721682 [20] Rashid S Curtis DE Garuti R Anderson NN Bashmakov Y Ho YK Hammer RE Moon YA Horton JD Decreased plasma cholesterol and hypersensitivity to statins in mice lacking Pcsk9 Proc Natl Acad Sci U S A 2005 102 15 5374 9 15805190 [21] Zaid A Roubtsova A Essalmani R Marcinkiewicz J Chamberland A Hamelin J Tremblay M Jacques H Jin W Davignon J Seidah NG Proprotein convertase subtilisin/kexin type 9 (PCSK9): hepatocyte-specific low-density lipoprotein receptor degradation and critical role in mouse liver regeneration Hepatology 2008 48 2 646 54 18666258 [22] Demers A Samami S Lauzier B Des Rosiers C Ngo Sock ET Ong H Mayer G PCSK9 Induces CD36 Degradation and Affects Long-Chain Fatty Acid Uptake and Triglyceride Metabolism in Adipocytes and in Mouse Liver Arterioscler Thromb Vasc Biol 2015 35 12 2517 25 26494228 [23] Fasano T Cefalù AB Di Leo E Noto D Pollaccia D Bocchi L Valenti V Bonardi R Guardamagna O Averna M Tarugi P A novel loss of function mutation of PCSK9 gene in white subjects with low-plasma low-density lipoprotein cholesterol Arterioscler Thromb Vasc Biol 2007 27 3 677 81 17170371 [24] Cariou B Ouguerram K Zaïr Y Guerois R Langhi C Kourimate S Benoit I Le May C Gayet C Belabbas K Dufernez F PCSK9 dominant negative mutant results in increased LDL catabolic rate and familial hypobetalipoproteinemia Arterioscler Thromb Vasc Biol 2009 29 12 2191 7 19762784 [25] Baragetti A Balzarotti G Grigore L Pellegatta F Guerrini U Pisano G Fracanzani AL Fargion S Norata GD Catapano AL PCSK9 deficiency results in increased ectopic fat accumulation in experimental models and in humans Eur J Prev Cardiol 2017 24 17 1870 7 28758421 [26] Kotowski IK Pertsemlidis A Luke A Cooper RS Vega GL Cohen JC Hobbs HH A spectrum of PCSK9 alleles contributes to plasma levels of low-density lipoprotein cholesterol Am J Hum Genet 2006 78 3 410 22 16465619 [27] Di Filippo M Vokaer B Seidah NG A case of hypocholesterolemia and steatosis in a carrier of a PCSK9 loss-of-function mutation and polymorphisms predisposing to nonalcoholic fatty liver disease J Clin Lipidol 2017 11 4 1101 5 28711549 [28] Sabatine MS Giugliano RP Keech AC Honarpour N Wiviott SD Murphy SA Kuder JF Wang H Liu T Wasserman SM Sever PS Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease N Engl J Med 2017 376 18 1713 22 28304224 [29] Ruscica M Ferri N Macchi C Meroni M Lanti C Ricci C Maggioni M Fracanzani AL Badiali S Fargion S Magni P Liver fat accumulation is associated with circulating PCSK9 Ann Med 2016 48 5 384 91 27222915 [30] Cariou B Langhi C Le Bras M Bortolotti M Lê KA Theytaz F Le May C Guyomarc’h-Delasalle B Zaïr Y Kreis R Boesch C Plasma PCSK9 concentrations during an oral fat load and after short term high-fat, high-fat high-protein and high-fructose diets Nutr Metab (Lond) 2013 10 1 4 23298392 [31] Wargny M Ducluzeau PH Petit JM Le May C Smati S Arnaud L Pichelin M Bouillet B Lannes A Blanchet O Lefebvre P Circulating PCSK9 levels are not associated with the severity of hepatic steatosis and NASH in a high-risk population Atherosclerosis 2018 278 82 90 30261472 [32] Christensen DL Eis J Hansen AW Larsson MW Mwaniki DL Kilonzo B Tetens I Boit MK Kaduka L Borch-Johnsen K Friis H Obesity and regional fat distribution in Kenyan populations: impact of ethnicity and urbanization Ann Hum Biol 2008 35 2 232 49 18428015 [33] Christensen DL Friis H Mwaniki DL Kilonzo B Tetens I Boit MK Omondi B Kaduka L Borch-Johnsen K Prevalence of glucose intolerance and associated risk factors in rural and urban populations of different ethnic groups in Kenya Diabetes Res Clin Pract 2009 84 3 303 10 19361878 [34] Baass A Dubuc G Tremblay M Delvin EE O’Loughlin J Levy E Davignon J Lambert M Plasma PCSK9 is associated with age, sex, and multiple metabolic markers in a population-based sample of children and adolescents Clin Chem 2009 55 9 1637 45 19628659 [35] Dubuc G Tremblay M Paré G Jacques H Hamelin J Benjannet S Boulet L Genest J Bernier L Seidah NG Davignon J A new method for measurement of total plasma PCSK9: Clinical applications J Lipid Res 2010 51 140 9 19571328 [36] De Lucia Rolfe E Brage S Sleigh A Finucane F Griffin SJ Wareham NJ Ong KK Forouhi NG Validity of ultrasonography to assess hepatic steatosis compared to magnetic resonance spectroscopy as a criterion method in older adults PLoS One 2018 13 11 e0207923 30475885 [37] Lee JH Kim D Kim HJ Lee CH Yang JI Kim W Kim YJ Yoon JH Cho SH Sung MW Lee HS Hepatic steatosis index: a simple screening tool reflecting nonalcoholic fatty liver disease Dig Liver Dis 2010 42 7 503 8 19766548 [38] Bedogni G Bellentani S Miglioli L Masutti F Passalacqua M Castiglione A Tiribelli C The Fatty Liver Index: a simple and accurate predictor of hepatic steatosis in the general population BMC Gastroenterol 2006 6 33 17081293 [39] Alberti KG Eckel RH Grundy SM Zimmet PZ Cleeman JI Donato KA Fruchart JC James WP Loria CM Smith SC Jr International Diabetes Federation Task Force on Epidemiology and Prevention; Hational Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; International Association for the Study of Obesity Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity Circulation 2009 120 16 1640 5 19805654 [40] Lee DS Evans JC Robins SJ Wilson PW Albano I Fox CS Wang TJ Benjamin EJ D’Agostino RB Vasan RS Gamma glutamyl transferase and metabolic syndrome, cardiovascular disease, and mortality risk: the Framingham Heart Study Arterioscler Thromb Vasc Biol 2007 27 1 127 33 17095717 [41] Cariou B Le Bras M Langhi C Le May C Guyomarc’h-Delasalle B Krempf M Costet P Association between plasma PCSK9 and gamma-glutamyl transferase levels in diabetic patients Atherosclerosis 2010 211 2 700 2 20452593 [42] Tavori H Giunzioni I Predazzi IM Plubell D Shivinsky A Miles J Devay RM Liang H Rashid S Linton MF Fazio S Human PCSK9 promotes hepatic lipogenesis and atherosclerosis development via apoE- and LDLR-mediated mechanisms Cardiovasc Res 2016 110 2 268 78 26980204 [43] Pol S Lebray P Vallet-Pichard A HIV infection and hepatic enzyme abnormalities: intricacies of the pathogenic mechanisms Clin Infect Dis 2004 38 Suppl 2 S65 72 14986277

PMC007xxxxxx/PMC7614816.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9610576 Trop Med Int Health Trop Med Int Health Tropical medicine & international health : TM & IH 1360-2276 1365-3156 34704339 7614816 10.1111/tmi.13696 EMS144242 Article Hepatic steatosis is associated with anthropometry, cardio-metabolic disease risk, sex, age and urbanization, but not with ethnicity in adult Kenyans Kastberg Sophie E. 1 Lund Helene S. 1 de Lucia-Rolfe Emanuella 2 Kaduka Lydia U. 3 Boit Michael K. 4 Corpeleijn Eva 5 Friis Henrik 6 Bernard Sophie 7 Paquette Martine 7 Baas Alexis 78 Rasmussen Jon J. 9 Christensen Dirk L. 1 1 Department of Public Health, University of Copenhagen, Copenhagen, Denmark 2 NIHR Cambridge Biomedical Research Centre-Diet, Anthropometry and Physical Activity Group, MRC Epidemiology Unit, Institute of Metabolic Science, University of Cambridge, Cambridge, United Kingdom 3 Centre for Public Health Research, KEMRI, Nairobi, Kenya 4 Department of Physical, Exercise and Sport Science, Kenyatta University, Nairobi, Kenya 5 University of Groningen, University Medical Center Groningen, Department of Epidemiology, The Netherlands 6 Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark 7 Lipids, Nutrition and Cardiovascular Prevention Clinic of the Montreal Clinical Research Institute, Montreal, Canada 8 Department of Medicine, Divisions of Experimental Medicine and Medical Biochemistry, McGill University, Montreal, Canada 9 Department of Endocrinlogy, Copenhagen University Hospital (Rigshospitalet), Copenhagen, Denmark Correspondence: Dirk L. Christensen, University of Copenhagen, Department of Public Health, Section of Global Health, CSS Campus, Building 9, Oester Farimagsvej 5, 1014 Copenhagen K, Denmark. Phone: +45 214 268 46, dirklc@sund.ku.dk 01 1 2022 07 11 2021 08 4 2022 26 7 2023 27 1 4957 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Objectives We aimed to determine the associations of non-alcoholic fatty liver disease (NAFLD) with cardio-metabolic risk factors for diabetes in adult Kenyans. Methods A cross-sectional study was undertaken among rural and urban Kenyans of different ethnic origin. Ultrasonography scanning (USS) methods were used for the assessment of hepatic fat accumulation for NAFLD assessment and abdominal fat distribution, and simple anthropometry measurements were performed. All participants underwent a 2-h oral glucose tolerance test, and biochemical, haemodynamic and lifestyle data were obtained. Multivariate logistic regression analyses were used to assess sex, age, residency and ethnic differences in the association between NAFLD and various metabolic parameters. Results In total, 743 individuals (59.1% women) with a mean age of 38.0 (range 18-68) years participated in the study. Overall, 118 individuals (15.9%) had NAFLD, of whom 94.1% had mild steatosis. Age >40 years was significantly associated with having NAFLD compared to <30 years of no difference found in NAFLD between ethnic groups (Luo, Kamba, Maasai). All body composition and clinical measurements were associated with NAFLD (p<0.045 for OR). Conclusions Finding lower odds for NAFLD in men was unexpected, as was the lack of differences in NAFLD among the ethnic groups, while higher odds for NAFLD with increasing age and in urban vs. rural populations was expected. Especially the sex-specific results warrant further studies in black African populations on biology of body composition for having NAFLD, and whether this translates into insulin resistance and higher risk of diabetes and consequently cardiovascular disease in black African women. Non-alcoholic fatty liver disease sub-Saharan Africa cardio-metabolic risk fatty liver index Sustainable Development Goals Good Health and Wellbeing Reduced inequalities pmcIntroduction Non-communicable diseases (NCDs) are becoming a significant burden in sub-Saharan Africa (SSA) (1). Two decades ago, the “Global Burden of Disease Study” showed that 20% of deaths in SSA were caused by NCDs (2). More recently, in rural and urban populations in Kenya, we showed that obesity (3), diabetes (4), as well as hypertension and dyslipidemia (5) are moderately prevalent in adults with a strong positive rural-urban and age gradient of obesity and diabetes prevalence independently of sex and ethnicity. However, central obesity as measured by visceral adipose tissue (VAT) was higher in men than women, and in Maasai compared to Luo and Kamba (3). The diets of these ethnic groups are characterised by mainly milk and cereals, fish and cereals, and cereals, respectively, as staples in traditional, rural societies (6). Furthermore, we have also shown that proprotein convertase substilisin/kexin type 9 – a key modulator of the degradation of the low-density lipoprotein cholesterol receptor – is associated with hepatic steatosis in adult Kenyans, and to some degree mediated by hepatic insulin resistance (7). Thus, it is likely that hepatic steatosis will become more prevalent in SSA and contribute to the development of cardio-metabolic disease, especially type 2 diabetes in Kenyan as well as SSA populations in the light of increasing urbanization. Non-alcoholic fatty liver disease (NAFLD) covers a broad range of histologic manifestations ranging from simple deposition of adipose tissue to more progressive steatosis with associated hepatitis, fibrosis, cirrhosis and in some cases hepatocellular carcinoma (8). The term NAFLD is comprised of non-alcoholic fatty liver (NAFL) and nonalcoholic steatohepatitis (NASH) (9). NAFL is characterized by steatosis of the liver that involves more than 5% of the parenchyma with no evidence of hepatocyte injury (10). NASH is defined by a necro-inflammatory process where the liver cells become injured due to steatosis (10). Even though simple steatosis follows a more benign course, NASH has the potential to progress into cirrhosis and thereby increase the risk of developing hepatocellular carcinoma (11;12). To capture the hepatic steatosis a liver biopsy is the gold standard (8). A liver biopsy is an invasive technique, which may be associated with clinical complications. Ultrasonography is a cheaper, non-invasive and more practical solution and is validated and approved with regards to assessment of NAFLD (13). In a recent meta-analysis of studies either using imaging or liver biopsy, the regional prevalence of NAFLD was estimated to be 31.8% in the Middle-East, 30.5% in South America, 24.1% in North America, 23.7% in Europe, 23.4% in Asia and 13.5% in Africa (14). It is noteworthy that the low prevalence in Africa is based on two studies with small sample sizes (15;16) hardly representative of SSA populations. Globally, the estimated prevalence of NAFLD is 25% and 23% of the affected individuals also have type 2 diabetes, hyperlipidemia (69%), hypertension (39%), or metabolic syndrome (43%) (14). A prospective study of 1,051 middle-aged participants from the Framingham Heart Study found a bidirectional relationship between NAFLD and cardiovascular disease risk factors even if the underlying mechanisms of NAFLD and cardiovascular risk factors are not fully understood (17). We aimed to identify anthropometric and cardio-metabolic correlates of NAFLD by sex, age, ethnicity and urban vs. rural residency in adult Kenyans. Based on current evidence, we hypothesised that NAFLD is more prevalent in urban populations, with increasing age, in men vs. women and in Maasai vs. Luo and Kamba ethnic populations. Methods Study area and population A community-based cross-sectional study was conducted in 2005-2006 in rural and urban Kenya. Inclusion criteria for the study were ≥18 years of age and Luo, Kamba, or Maasai ethnicity or culturally related ethnic groups. Exclusion criteria were pregnancy, serious illnesses such as malaria, inability to walk unassisted and severe mental disease (3). Liver scans were performed in 799 individuals. All participants gave written or thumb print informed consent following information given at local community meetings. Ethical permission was obtained from the National Ethical Review Committee in Kenya (SSC Protocol No. 936), and consultative approval was given by the Danish National Committee on Biomedical Research Ethics in Denmark. For more details concerning the selection procedure and the study participants, see Christensen et al. (3). Liver scan, scoring methods and fatty liver index Ultrasonography scan (USS) was used for the assessment of hepatic tissue with regards to steatosis using a portable scanner (Aquila Basic Unit, Pie Medical Equipment; Esaote, Masstricht, the Netherlands) with a 3C.RS 3.5/5.0 MHz curved transducer (Probe Article no. 410638 Curved Array HiD probe R40; Pie Medical Equipment, Maastricht, the Netherlands). The abdominal ultrasound investigation was performed with the participant holding his/her breath. Multiple images were acquired on both sides of the main liver lobes by two trained operators. A sub-costal approach was used, and four sweeps were performed using standard imaging protocol (13). A semi-quantitative grading system was used to define normal, mild, moderate and severe steatosis. The liver scoring criteria were (13): 1: Increased echogenicity of liver parenchyma (bright liver in comparison with the kidney) 2: Decreased visualization of intra-hepatic vasculature 3: Attenuation of the ultrasound beam. The different scoring criteria were summed to derive a liver steatosis score. A score of ≤4 was classified as normal, score 5-7 for mild steatosis, score 8-10 for moderate steatosis and score ≥11 for severe steatosis. The usage of ultrasonography versus magnetic resonance spectroscopy was validated and approved with regards to assessment of NAFLD (13). Fatty liver index (FLI) was calculated as previously described by Bedogni et al., and categorized as grade I (FLI<30), grade II (FLI≥30-59), and grade III (FLI≥60) according to the formula: (e ^ 0.953*loge (triglycerides) + 0.139*bmi + 0.718*loge (ggt) + 0.053*waist circumference - 15.745) / (1 + e ^ 0.953*ln(triglycerides) + 0.139*bmi + 0.718*loge (ggt) + 0.053*waist circumference - 15.745)*100 (18). We used the ultrasound-based scores to define NAFLD, while FLI was used as supplementary information. Anthropometry and ultrasonography for abdominal fat distribution Anthropometric measurements were taken with the participants standing barefoot and wearing light clothing. Body mass index (BMI, kg/m2) was calculated and waist circumference (WC) was measured with a body tape (WM02 Body Tape; Chasmors, Haechstmass, Germany) midway between the iliac crest and the costal margin following a quiet expiration. For further details, see Christensen et al. (3). Abdominal fat distribution, i.e. VAT and subcutaneous adipose tissue (SAT), was measured in cm using USS (Aquila Basic Unit, Esaote, Pie Medical Equipment, Maastricht, the Netherlands) with a 3.5/5.0MHz transducer (Probe Article no. 410638 Curved Array HiD probe R40 Pie Medical Equipment). We followed the protocol by Stolk et al. which has been validated in relation to computed tomography and magnetic resonance imaging (19). Laboratory tests Following an overnight fast (≥8-h), all participants had a blood test taken and subsequently underwent a standard 75-g oral glucose tolerance test (OGTT). Venous blood glucose was determined by the glucose dehydrogenase method using haemolysation and deproteinisation on a HemoCue B-Glucose 201+ device (HemoCue AB, Ängelholm, Sweden) with 5 mL of blood. Standard procedures were followed to centrifuge blood, which was stored as plasma and serum in cryotubes at -20°C at the nearest health facility while in the field, and later at -80°C at KEMRI, Nairobi, Kenya, before being shipped to Steno Diabetes Center Copenhagen in Denmark for insulin and standard lipid profile analyses; for further details, see (5). Later, plasma samples were shipped to Canada for analysis of a standard liver enzyme profile at Montreal Clinical Research Institute; for further details, see (7). Insulin resistance was calculated by the homoeostasis model assessment of insulin resistance by computer model (HOMA-IR) (20). Based on Matsuda et al., HOMA-IR was calculated according to the formula: (fasting insulin (pmol/l) x fasting glucose (mmol/l)/135 (21). Insulin resistance was defined as follows: 1) BMI >28.9 kg/m2, or 2) HOMA-IR >4.65, or 3) HOMA-IR >3.60 if BMI >27.5 kg/m2 based on Stern et al. (22). Dyslipidemia was regarded as high triglyceride (>1.7 mmol/L) and/or low high-density lipoprotein cholesterol (HDL-C) levels (<1.03 for men, and <1.29 for women) (23). Other clinical variables Blood pressure was measured with each participant sitting upright and determined as the average of two measurements using a full-automatic device (Omron M6, HEM-7001-E, Kyoto, Japan). Hypertension was defined as ≥140 and/or 90 mmHg of systolic and diastolic blood pressure, respectively (24). Blood haemoglobin was determined on site using a standard Coulter counter technique (model KX-21N, Sysmex Corporation, Kobe, Japan). Information concerning alcohol consumption and tobacco use was self-reported by the participants via an interactive Medical Health Assessment Form. Excessive alcohol consumption threshold was 20 g/d and 30 g/d for women and men, respectively (25). Statistical analysis Continuous variables were tested for normality using histogram and normal quantile plot. Skewed data were log-transformed prior to analysis. Differences between non-NAFLD and NAFLD groups were determined by using Student’s t-test, while the chi2 test was used to test for differences in proportions. Regression diagnostics showed that the associations between the outcome and exposure variables were linear, the residuals were normally distributed, and the requirement for homoscedasticity was fulfilled. Logistic regression analyses adjusted for age and sex were used to assess the association between the binary outcome for not having or having hepatic steatosis in relation to age categories, sex, residence, ethnicity, waist circumference (per cm increase), VAT (per cm increase), SAT (per cm increase), HOMA-IR (per unit increase), glucose metabolism categories, HDL-C (per mmol/L increase), dyslipidemia, and hypertension expressed as odds ratio (OR) (95% CI). Parameters including sex, NAFLD, HOMA-IR, DM, IGT and IFG, dyslipidemia, and hypertension were categorized by FLI categories. In order to establish to what extent the differences in proportions of IFG, IGT and DM were driven by NAFLD, analyses of associations between IFG, IGT and DM without having and with NAFLD were carried out. Data that showed normal distribution are presented as means (SD) for continuous variables, and n (%) for categorical variables. Data with skewed distribution are presented as geometric means (SD). Statistical analyses were performed using Stata/MP 14.0 (Stata Corp, College Station, USA). Results The study population included 799 Kenyans. Of these, potential participants were excluded due to not reaching the 18-years-old age-limit (n=2), missing values for age, height, weight, sex, BMI and steatosis score (n=28), or if they exceeded the criteria for AFLD (25), i.e. excessive alcohol consumption (n=26). Thus, 743 participants with a mean age (SD) of 38.0 (11.0) (range 18-68) years participated in the study whereof 439 (59.1%) were women. In total, 587 (79%) were rural residents, and ethnic distribution was 68 (9.2%) Luo, 353 (47.5%) Kamba, and 307 (41.3%) Maasai, respectively while 15 (2%) belonged to other ethnic groups. Eight (1.1 %) participants were diagnosed with DM during the study, while 16 (2.2 %) had known DM, of whom 11 took medication. Background characteristics of participants without and with NAFLD are presented in Table 1. Based on ultrasound scanning, 118 (15.9%) had NAFLD, with 111 (94.1%) and 7 (5.9%) having mild and moderate steatosis, respectively. None had severe steatosis. The proportion of NAFLD was 13.3% (n=78) and 25.6% (n=40) of the rural and urban participants, respectively (p<0.001), respectively. Among those with NAFLD, 21 (17.8 %) and 66 (55.9 %) were women <50 and ≥50 years of age, and 31 (26.3%) were men, respectively. Men were less likely to have NAFLD than women (OR=0.48 (95% CI: 0.31, 0.75), p=0.001). When further adjusted for WC or VAT, the results for sex difference remained statistically significant and more pronounced (0.39, 95% CI: 0.23, 0.65 and 0.29, 95% CI: 0.16;0.53, respectively, both p<0.001). After further adjustment for BMI or SAT, there was no longer a difference in OR between women and men for having NAFLD (OR=0.78 95% CI: 0.47, 1.30, p=0.34)) and (OR=1.35 95% CI: 0.79, 2.30, p=0.28), respectively. For individuals having NAFLD, mean BMI in women <50 and ≥50 years was 29.7 (5.8) and 26.9 (5.5) kg/m2, respectively, and for men it was 27.0 (6.4) kg/m2. Mean SAT was 3.56 (1.00), 2.84 (1.50), and 2.40 (1.19) cm, for women <50 years, women ≥50 years, and men, respectively. Among those with NAFLD, WC and VAT was higher in men, while BMI and SAT was higher in women (results not shown). All body composition, biochemical and haemodynamics data were positively associated with NAFLD (p<0.045) (see Table 2). Except for sex distribution and the distribution of IFG between the three FLI grade groups, all cardio-metabolic diseases showed significant associations across FLI grade groups going from grade I to grade III (p<0.001) (see Supplemental Table). The proportion of IFG, IGT and DM in the participants with NAFLD was 4.2, 2.0 and 3.8 times higher, respectively compared to those without NAFLD (Figure 1). Blood glucose levels at 0, 30 and 120 min. were significantly higher in the participants with NAFLD (p<0.002) (Figure 2). Discussion In Kenyan adults, we found lower OR (0.48) for having NAFLD in men than women with a female-to-male ratio of 2.8 to 1. Furthermore, being age >40 years showed an OR of ~2.5 for NAFLD versus being <30 years of age, and the urban population had a ~3-fold higher risk of NAFLD compared to the rural population. Cardio-metabolic risk parameters were all significantly associated with NAFLD (OR >1.5), with hypertension and DM having the highest OR of 3.4 and 3.8, respectively. The lower OR for having NAFLD in men than women that we found in the current study is in contrast to previous findings (26;27). Furthermore, this result is despite men in the current cohort having higher abdominal fat accumulation measured as WC and VAT, i.e. metabolically active fat tissue contributing to venous drainage of fat tissue going directly to the liver via the portal vein resulting in ectopic fat accumulation and insulin resistance (28). When adjusting for WC and VAT accumulation, the OR estimates became even more pronounced. This may be explained by women being more prone to NAFLD for a given amount of intra-abdominal fat accumulation, even if at this stage this suggestion remains pure speculation. Interestingly, when adjusting for BMI and SAT accumulation, both of which were higher in women, the sex-specific difference for having NAFLD disappeared. In a systematic review, Pang et al. showed that WC and VAT had a stronger association with NAFLD than BMI, even though all three obesity measures were independently associated with increased risk of NAFLD (29). Furthermore, Fan et al. (30) found a dose-response relationship between BMI and liver steatosis risk in an approximate J-shaped fashion. In the current cohort, but including a higher number of participants (n=1459), we previously showed that in women, BMI had the highest OR for glucose intolerance risk with increasing obesity as compared to WC, VAT, and SAT, while for men VAT showed the highest OR for glucose intolerance (4). Thus, general obesity appears to be more detrimental to metabolic disorders in Kenyan women compared to men. Nevertheless, SAT also explained the difference in sex-specific OR for having NAFLD. In the NAFLD individuals, women regardless of age group had higher SAT than men; furthermore, women <50 years (presumably premenopausal), had higher SAT accumulation than women ≥50 years (presumably postmenopausal). This difference could be explained by the high-energy demands of lactation in young women as SAT acts as an energy reservoir for women within reproductive age (31). SAT in general is a depot for surplus triglycerides (32), and is regarded as protective in relation to cardio-metabolic disorders (33). However, the USS technique does not distinguish between superficial and deep SAT divided by the Scarpa’s fascia (34). Deep SAT adipocytes have been shown to have higher lipolytic activity compared to superficial SAT adipocytes (35), and contain more saturated fat than superficial SAT (36). Thus, deep SAT to some extent resembles VAT, and may therefore be responsible for ectopic fat accumulation, including hepatic steatosis. It is important to note that more than half of the individuals with NAFLD in the current study were postmenopausal women when using age ≥50 years as a proxy measure. Prevalence and incidence of NAFLD are higher in men and postmenopausal women compared to premenopausal women (37;38). For the postmenopausal women, this is due to the loss of oestrogens, which among others regulate liver lipogenesis (39). Nevertheless, Naran et al. have shown that SAT is a significant negative determinant of hepatic steatosis as measured by computed tomography in a small sample (n=29) of black, South African women (40). Cardio-metabolic disorders were significantly higher in the current study population with NAFLD whether measured as absolute numbers, by OR for each unit increase, or by OR for standard cut-offs for disease. Increased HOMA-IR is regarded as the metabolic manifestation of hepatic steatosis, and as this feature progresses may result in DM (41). We found significantly higher proportions of DM as well as pre-diabetes (IFG and IGT) in those with NAFLD. The results are based on a fasting blood test followed by an OGTT, and this indicates not only hepatic IR, but also peripheral IR as demonstrated by higher 30 min and 2-h whole blood glucose levels in NAFLD individuals. Furthermore, DM showed the highest OR for having NAFLD at 3.80 compared to other cardio-metabolic and anthropometric correlates of hepatic steatosis. This is in line with previous studies showing that excess hepatic fat deposition is uniformly present before the onset of “classical” type 2 diabetes (42;43), assuming that the DM subtype in the present study was predominantly if not entirely type 2 diabetes. We found no difference in OR for having hepatic steatosis between the three ethnic groups studied, i.e. the Luo, Kamba and Maasai. Based on results from the same but larger groups of participants, the Maasai had higher VAT accumulation (n=1430), and higher hepatic insulin resistance based on HOMA-IR (n=1087), respectively, compared to the other ethnic groups (3;44). Therefore, a higher proportion of hepatic steatosis in the Maasai was expected. Thus, this adds to the evidence of a lack of linear relationship between VAT accumulation, insulin resistance and hepatic steatosis, but points to a more complex interaction between these interrelated parameters. FLI is an algorithm used as a prediction for liver steatosis (16), which is often employed in clinical practice as the parameters (triglycerides, BMI, GGT, and WC) included in the algorithm are routine measurements, albeit not in public clinics or hospitals in Kenya or SSA in general. According to Bedogni et al., FLI ≥60 is an indicator of hepatic steatosis when using USS as reference (18). We found a significant association for having NAFLD across the three grades of FLI. Nevertheless, FLI may underestimate hepatic steatosis in black Africans due to low triglyceride levels in black African populations in general, which we also found in the current study population (5). Importantly, HOMA-IR had an overlap of 83.8% when comparing FLI ≥60 with USS-derived diagnosis. Therefore, FLI seems to be relevant for cardio-metabolic disease risk in black African populations, and this algorithm may be easier to implement in clinical practice compared to analysing USS-captured images which is relatively time consuming. The prevalence of NAFLD in the current study was 15.9%, which is slightly higher than the prevalence reported by Younossi et al. based on a meta-analysis (14) based on two very small studies in SSA. It is of note, that we only found mild and moderate steatosis, and that USS technique is limited by lower sensitivity for detecting milder degrees (<30%) of steatosis as compared to magnetic resonance imaging and magnetic resonance spectroscopy (45). Thus, we may have underestimated the prevalence of NAFLD. Our results show that NAFLD is not only an important factor for adverse cardio-metabolic traits, but also that hepatic steatosis becomes more prevalent with increasing age and urbanisation. The latter two factors are important to consider in Kenya and other SSA nations where life expectancy and urbanisation have been increasing over the past few decades – a trend which is expected to continue (46;47). Thus, NAFLD prevalence is likely to increase, and along with this development cardio-metabolic disease risk is expected to increase as well. Special attention may be needed for women in this context, especially if NAFLD translates into higher incidence of diabetes and cardiovascular diseases in the female part of the population. Strengths of the current study were the large sample size in an African context, inclusion of rural and urban participants, and a comprehensive panel of USS, simple anthropometric, biochemical and haemodynamic measurements across a population with a wide age range (18 to 68 years). We acknowledge several limitations to this study. Neither intra- nor inter-observer validity USS were carried out, and USS is not a gold standard technique, all of which could have compromised the quality of the data. However, NAFLD screening using USS at the population level in resource-poor settings may be the most cost-effective approach in a public health context. Furthermore, this is a cross sectional study, and therefore the interpretation of the results from a cause-relationship perspective, must be done with caution. We also want to stress that using logistic regression analysis for association with frequent disease (>10% prevalence) such as hypertension or dyslipidaemia may inflate the estimates. It is of note that 10 subjects in the cohort were treated with hypotensive drugs, and none of them were on lipid-lowering, oral hypoglycemic agents or insulin therapy. Finally, we did not collect data on HIV including anti-retroviral drugs, hepatitis B or C or other biomarkers for infections, which could have affected the results. In conclusion, NAFLD is more common in Kenyan women than men, which may be explained by general obesity (BMI) and SAT accumulation. Furthermore, increasing age and urbanisation are both associated with NAFLD, while ethnicity comparing Luo, Kamba and Maasai people showed no difference in NAFLD. Especially the sex-specific results warrant further studies in black African populations on biology of body composition for having NAFLD, and whether this translates into insulin resistance and higher risk of diabetes and consequently cardiovascular disease in black African women. Supplementary Material Supplementary Tables Acknowledgements We are grateful to all study participants, the local chiefs and sub-chiefs, the local elder councils and district politicians. We kindly thank all local assistants for their excellent work mobilizing participants and measurements of anthropometry and body composition. A special thanks to Andreas W. Hansen who performed parts of the abdominal ultrasound scans. We are grateful to funding to EDLR, who is supported by the NIHR Cambridge Biomedical Research Centre (IS-BRC-1215-20014. Likewise, we are grateful to the general contributions to the Kenya Diabetes Study by Professor Knut Borch-Johnsen, Professor Inge Tetens, and Dr. David L. Mwaniki. Funding was received from DANIDA (J. no.104.DAN.8–871, RUF project no. 91202), University of Copenhagen (Cluster of International Health), Steno Diabetes Center Copenhagen, Beckett Foundation, Dagmar Marshall’s Foundation, Dr Thorvald Madsen’s Grant for the Advancement of Medical Sciences, Kong Christian den Tiende’s Foundation, and Brdr. Hartmann Foundation. Figure 1 Prevalence of impaired fasting glucose (IFG*, 6.1 to 6.9 mmol/L** and <7.8 mmol/L***), impaired glucose tolerance (IGT*, <7.0 mmol/L** and ≥7.8 and <11.1 mmol/L***) and diabetes mellitus (DM*, ≥7.0 mmol/L** or ≥11.1 mmol/L***) among participants with non-NAFLD and NAFLD. *Plasma values; **fasting glucose test; ***2-h glucose tolerance test. Figure 2 Whole blood glucose profile in participants with NAFLD and non-NAFLD during 2-h oral glucose tolerance test. Presented as geometric means. Table 1 General characteristics of participants without and with non-alcoholic fatty liver disease (n=743) presented as mean (SD) and proportion (%).   Non-NAFLD NAFLD   Variable n=625 n=118 p-value Age (years) 36.7 (10.8) 41.6 (11.4) 0.001 Sex (F: M %) 352:273 (56:44) 87:31 (74:26) 0.001 Urban residence, n(%) 116 (19) 40 (34) 0.001 BMI (kg/m2) 21.3 (3.8) 27.4 (5.8) 0.001 Waist circumference (cm) 76.8 (9.3) 91.5 (14.2) 0.001 VAT (cm) 6.3 (1.31) 7.7 (2.0) 0.001 SAT* (cm) 1.4 (1.0) 2.9 (1.4) 0.001 Fasting plasma glucose   (mmol/L)* 4.5 (1.4) 4.9 (1.8) 0.012 Fasting serum insulin   (pmol/L) 27.6 (20.5) 48.6 (41.5) 0.001 HOMA-IR 0.82 (0.76) 1.65 (2.7) 0.001 DM, n (%) 14 (2.2) 10 (8.5) 0.001 IGT, n (%) 45 (7.2) 17 (14.4) 0.010 IFG, n (%) 6 (1.0) 5 (4.2) 0.007 Total cholesterol (mmol/L) 3.9 (0.9) 4.4 (1.1) 0.001 LDL-C (mmol/L) 2.2 (0.9) 2.7 (1.0) 0.001 HDL-C (mmol/L) 1.15 (0.29) 1.08 (0.28) 0.015 Triglycerides (mmol/L) 0.9 (0.5) 1.2 (0.65) 0.001 Dyslipidemia, n (%) 167 (26.7) 45 (38.1) 0.017 ALP (U/L) 80.4 (31.6) 84.1 (26.5) 0.264 ALT (U/L) 8.7 (5.1) 8.6 (4.4) 0.750 AST (U/L) 21.4 (10.2) 19.6 (7.79) 0.107 Total bilirubin (μmol/L) 4.7 (3.8) 6.5 (5.93) 0.003 GGT (U/L) 22.5 (19.6) 27.4 (31.8) 0.040 Systolic BP (mmHg) 119 (16) 127 (17) 0.001 Diastolic BP (mmHg) 74 (10) 78 (11) 0.001 Hypertension, n (%) 60 (9.6) 30 (25.4) 0.001 Haemoglobin (g/dL) 13.7 (2.1) 13.4 (2.2) 0.145 Tobacco use, n (%) ** 76 (12.2) 2 (1.7) 0.001 Alcohol, n (%) *** 46 (7.4) 7(5.9) 0.543 * Presented as geometric mean (SD) ** For most Maasai smoking is sniff and chewing tobacco *** Participants with an alcohol-consumption that does not exceed the criteria for NAFLD Data are presented as means (SD) for continuous variables and n (%) for categorical variables. BMI: body mass index; VAT: visceral adipose tissue; SAT: subcutaneous adipose tissue; HOMA-IR: homeostasis model assessment was calculated as (fasting plasma insulin X fasting plasma glucose)/22.5; LDLC: low-density lipoprotein; HDL-C: high-density lipoprotein; Systolic BP: systolic blood pressure; Diastolic BP: diastolic blood pressure; DM: diabetes mellitus; IGT: impaired glucose tolerance; IFG: impaired fasting glycaemia. Waist circumference: n=741; VAT: n=741; SAT: n=741; Fasting serum insulin: n=722; HOMA-IR: n=722; Total cholesterol: n=711; LDL-C: n=734; HDL-C: n=711; Triglycerides: n=711; Systolic BP: n=731; Haemoglobin: n=699; Smoking: n=706; Alcohol: n=699. Table 2 Odds ratio for univariate and multivariate logistic regression for having nonalcoholic fatty liver disease Variable Univariate analysis Odds ratio (95%CI) p-value Multivariate analysis Odds ratio (95%CI) p-value Age (years)1 Age ≤ 30 1.00   1.00   30 < Age ≤ 40 1.31 (0.72, 2.40) 0.383 1.22 (0.66, 2.22) 0.527 Age > 40 2.67 (1.58, 4.50) 0.001 2.48 (1.46, 4.21) 0.001 Men2 0.47 (0.29, 0.71) 0.001 0.48 (0.31, 0.75) 0.001 Urban residence 2.25 (1.12, 3.46) 0.001 2.85 (1.81, 4.50) <0.001 Ethnicity Luo 1   1   Kamba 0.68 (0.45, 1.03) 0.069 0.82 (0.53, 1.27) 0.317 Maasai 0.70 (0.15, 3.16) 0.641 1.13 (0.24, 5.38) 0.872 Body mass index (per kg/m2increase) 1.28 (1.22, 1.34) <0.001 1.27 (1.21, 1.33) <0.001 Waist circumference (pr cm increase) 1.11 (1.09, 1.13) <0.001 1.11 (1.09, 1.14) <0.001 VAT (pr cm increase) 1.70 (1.46, 1.98) <0.001 1.81 (1.53, 2.14) <0.001 SAT (pr cm increase) 2.57 (2.15, 3.10) <0.001 2.65 (2.16, 3.27) <0.001 HOMA-IR (per unit increase) 1.83 (1.50, 2.26) <0.001 1.77 (1.43, 2.19) <0.001 Glucose metabolism NGT 1   1   Prediabetes (IFG/IGT) 2.64 (1.53, 4.55) <0.001 2.43 (1.39, 4.24) 0.002 Diabetes mellitus 4.63 (1.99, 10.7) <0.001 3.82 (1.59, 9.17) 0.003 Total cholesterol (per mmol/L increase) 1.64 (1.34, 2.00) <0.001 1.57 (1.26, 1.95) <0.001 HDL-C (per mmol/L increase) 0.42 (0.21, 0.85) 0.016 0.41 (0.20, 0.83) 0.014 LDL-C (per mmol/L increase) 1.18 (144, 2.29) <0.001 1.73 (1.36, 2.03) <0.001 Triglycerides (per mmol/L increase) 2.26 (1.61, 3.14) <0.001 2.17 (1.51, 3.13) <0.001 Dyslipidemia 1.65 (1.01, 2.34) 0.018 1.51 (0.99, 2.31) 0.057 Hypertension 3.22 (1.97, 5.28) <0.001 3.34 (1.96, 5.70) <0.001 All multivariable analyses adjusted for age and sex except for 1 only adjusted for sex and 2 only adjusted for age. VAT: visceral adipose tissue; SAT: subcutaneous adipose tissue; HOMA-IR: homeostasis model assessment was calculated as (fasting plasma insulin x fasting plasma glucose)/135; HDL-C: high-density lipoprotein; LDLC: low-density lipoprotein; NGT: Normoglycaemia; IGT: impaired glucose tolerance; IFG: impaired fasting glycaemia. * Excludes 15 participants of other ethnicities than Luo, Kamba, Maasai (1) Baingana FK Bos ER Changing Patterns of Disease and Mortality in Sub-Saharan Africa: An Overview 2006 (2) The International Bank for Reconstruction and Development / The World Bank Global Burden of Disease and Risk Factors Oxford University Press New York 2006 (3) Christensen DL Eis J Hansen AW Larsson MW Mwaniki DL Kilonzo B Obesity and regional fat distribution in Kenyan populations: impact of ethnicity and urbanization Ann Hum Biol 2008 Mar 35 2 232 49 18428015 (4) Christensen DL Friis H Mwaniki DL Kilonzo B Tetens I Boit MK Prevalence of glucose intolerance and associated risk factors in rural and urban populations of different ethnic groups in Kenya Diabetes Res Clin Pract 2009 Jun 84 3 303 10 19361878 (5) Christensen DL Faurholt-Jepsen D Birkegaard L Mwaniki DL Boit MK Kilonzo B Cardiovascular risk factors in rural Kenyans are associated with differential age gradients, but not modified by sex or ethnicity Ann Hum Biol 2016 43 1 42 9 26073640 (6) Hansen AW Christensen DL Larsson MW Eis J Christensen T Friis H Mwaniki DL Kilonzo B Boit MK Borch-Johnsen K Tetens I Dietary patterns, food and macronutrient intakes among adults in three ethnic groups in rural Kenya Public Health Nutr 2011 Sep 14 9 1671 9 21299918 (7) Paquette M Gauthier D Chamberland A Prat A De Lucia RE Rasmussen JJ Circulating PCSK9 is associated with liver biomarkers and hepatic steatosis Clin Biochem 2020 Mar 77 20 5 31972148 (8) Benedict M Zhang X Non-alcoholic fatty liver disease: An expanded review World J Hepatol 2017 Jun 8 9 16 715 32 28652891 (9) Sayiner M Koenig A Henry L Younossi ZM Epidemiology of Nonalcoholic Fatty Liver Disease and Nonalcoholic Steatohepatitis in the United States and the Rest of the World Clin Liver Dis 2016 May 20 2 205 14 27063264 (10) Kanwar P Kowdley KV The Metabolic Syndrome and Its Influence on Nonalcoholic Steatohepatitis Clin Liver Dis 2016 May 20 2 225 43 27063266 (11) Marrero JA Fontana RJ Su GL Conjeevaram HS Emick DM Lok AS NAFLD may be a common underlying liver disease in patients with hepatocellular carcinoma in the United States Hepatology 2002 Dec 36 6 1349 54 12447858 (12) Teli MR James OF Burt AD Bennett MK Day CP The natural history of nonalcoholic fatty liver: a follow-up study Hepatology 1995 Dec 22 6 1714 9 7489979 (13) De Lucia RE Brage S Sleigh A Finucane F Griffin SJ Wareham NJ Validity of ultrasonography to assess hepatic steatosis compared to magnetic resonance spectroscopy as a criterion method in older adults PLoS One 2018 13 11 e0207923 30475885 (14) Younossi ZM Koenig AB Abdelatif D Fazel Y Henry L Wymer M Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes Hepatology 2016 Jul 64 1 73 84 26707365 (15) Almobarak AO Barakat S Khalifa MH Elhoweris MH Elhassan TM Ahmed MH Non alcoholic fatty liver disease (NAFLD) in a Sudanese population: What is the prevalence and risk factors? Arab J Gastroenterol 2014 Mar 15 1 12 5 24630507 (16) Onyekwere CA Ogbera AO Balogun BO Non-alcoholic fatty liver disease and the metabolic syndrome in an urban hospital serving an African community Ann Hepatol 2011 Apr 10 2 119 24 21502672 (17) Ma J Hwang SJ Pedley A Massaro JM Hoffmann U Chung RT Benjamin EJ Levy D Fox CS Long MT Bi-directional analysis between fatty liver and cardiovascular disease risk factors J Hepatol 2017 Feb 66 2 390 397 27729222 (18) Bedogni G Bellentani S Miglioli L Masutti F Passalacqua M Castiglione A The Fatty Liver Index: a simple and accurate predictor of hepatic steatosis in the general population BMC Gastroenterol 2006 Nov 2 6 33 17081293 (19) Stolk RP Wink O Zelissen PM Meijer R van Gils AP Grobbee DE Validity and reproducibility of ultrasonography for the measurement of intra-abdominal adipose tissue Int J Obes Relat Metab Disord 2001 Sep 25 9 1346 51 11571598 (20) Matthews DR Hosker JP Rudenski AS Naylor BA Treacher DF Turner RC Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man Diabetologia 1985 Jul 28 7 412 9 3899825 (21) Matsuda M DeFronzo RA Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp Diabetes Care 1999 Sep 22 9 1462 70 10480510 (22) Stern SE Williams K Ferrannini E DeFronzo RA Bogardus C Stern MP Identification of individuals with insulin resistance using routine clinical measurements Diabetes 2005 Feb 54 2 333 9 15677489 (23) Alberti KG Eckel RH Grundy SM Zimmet PZ Cleeman JI Donato KA Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity Circulation 2009 Oct 20 120 16 1640 5 19805654 (24) Whitworth JA 2003 World Health Organization (WHO)/International Society of Hypertension (ISH) statement on management of hypertension J Hypertens 2003 Nov 21 11 1983 92 14597836 (25) EASL-EASD-EASO Clinical Practice Guidelines for the management of non-alcoholic fatty liver disease J Hepatol 2016 Jun 64 6 1388 402 27062661 (26) Lonardo A Suzuki A Nonalcoholic fatty liver disease: does sex matter? Hepatobiliary Surg Nutr 2019 Apr 8 2 164 6 31098369 (27) Lonardo A Nascimbeni F Ballestri S Fairweather D Win S Than TA Sex Differences in Nonalcoholic Fatty Liver Disease: State of the Art and Identification of Research Gaps Hepatology 2019 Oct 70 4 1457 69 30924946 (28) Jensen MD Role of body fat distribution and the metabolic complications of obesity J Clin Endocrinol Metab 2008 Nov 93 11 Suppl 1 S57 S63 18987271 (29) Pang Q Zhang JY Song SD Qu K Xu XS Liu SS Central obesity and nonalcoholic fatty liver disease risk after adjusting for body mass index World J Gastroenterol 2015 Feb 7 21 5 1650 62 25663786 (30) Fan R Wang J Du J Association between body mass index and fatty liver risk: A dose-response analysis Sci Rep 2018 Oct 15 8 1 15273 30323178 (31) Wells JC The evolution of human fatness and susceptibility to obesity: an ethological approach Biol Rev Camb Philos Soc 2006 May 81 2 183 205 16677431 (32) Al-Sulaiti H Diboun I Banu S Al-Emadi M Amani P Harvey TM Triglyceride profiling in adipose tissues from obese insulin sensitive, insulin resistant and type 2 diabetes mellitus individuals J Transl Med 2018 Jun 26 16 1 175 29940972 (33) Snijder MB Visser M Dekker JM Goodpaster BH Harris TB Kritchevsky SB Low subcutaneous thigh fat is a risk factor for unfavourable glucose and lipid levels, independently of high abdominal fat. The Health ABC Study Diabetologia 2005 Feb 48 2 301 8 15660262 (34) Lancerotto L Stecco C Macchi V Porzionato A Stecco A De CR Layers of the abdominal wall: anatomical investigation of subcutaneous tissue and superficial fascia Surg Radiol Anat 2011 Dec 33 10 835 42 21212951 (35) Monzon JR Basile R Heneghan S Udupi V Green A Lipolysis in adipocytes isolated from deep and superficial subcutaneous adipose tissue Obes Res 2002 Apr 10 4 266 9 11943835 (36) Marinou K Hodson L Vasan SK Fielding BA Banerjee R Brismar K Structural and functional properties of deep abdominal subcutaneous adipose tissue explain its association with insulin resistance and cardiovascular risk in men Diabetes Care 2014 37 3 821 9 24186879 (37) Pan JJ Fallon MB Gender and racial differences in nonalcoholic fatty liver disease World J Hepatol 2014 May 27 6 5 274 83 24868321 (38) Yang JD Abdelmalek MF Pang H Guy CD Smith AD Diehl AM Gender and menopause impact severity of fibrosis among patients with nonalcoholic steatohepatitis Hepatology 2014 Apr 59 4 1406 14 24123276 (39) Palmisano BT Zhu L Stafford JM Role of Estrogens in the Regulation of Liver Lipid Metabolism Adv Exp Med Biol 2017 1043 227 56 29224098 (40) Naran NH Haagensen M Crowther NJ Steatosis in South African women: How much and why? PLoS One 2018 13 1 e0191388 29351564 (41) Taylor R Pathogenesis of type 2 diabetes: tracing the reverse route from cure to cause Diabetologia 2008 Oct 51 10 1781 9 18726585 (42) Bugianesi E Gastaldelli A Vanni E Gambino R Cassader M Baldi S Insulin resistance in non-diabetic patients with non-alcoholic fatty liver disease: sites and mechanisms Diabetologia 2005 Apr 48 4 634 42 15747110 (43) Targher G Bertolini L Padovani R Rodella S Tessari R Zenari L Prevalence of nonalcoholic fatty liver disease and its association with cardiovascular disease among type 2 diabetic patients Diabetes Care 2007 May 30 5 1212 8 17277038 (44) Christensen DL Faurholt-Jepsen D Faerch K Mwaniki DL Boit MK Kilonzo B Insulin resistance and betacell function in different ethnic groups in Kenya: the role of abdominal fat distribution Acta Diabetol 2014 Feb 51 1 53 60 23563691 (45) Bohte AE van Werven JR Bipat S Stoker J The diagnostic accuracy of US, CT, MRI and 1H-MRS for the evaluation of hepatic steatosis compared with liver biopsy: a meta-analysis Eur Radiol 2011 Jan 21 1 87 97 20680289 (46) Saghir J Santoro J Urbanization in sub-Saharan Africa: Meeting Challenges by Bridging Stakeholders Center for Strategic & International Studies Washington D.C 2018 (47) World Bank Life Expectancy at Birth, Total (Years) – sub-Saharan Africa World Bank Group Washington D.C 2021

PMC007xxxxxx/PMC7614817.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 100888143 Hum Fertil (Camb) Hum Fertil (Camb) Human fertility (Cambridge, England) 1464-7273 1742-8149 34190021 7614817 10.1080/14647273.2021.1946173 EMS181577 Article Screening by single-molecule Molecular Inversion Probes targeted sequencing panel of candidate genes of infertility in azoospermic infertile Jordanian males Batiha Osamah 1* Burghel George J 2 Alkofahi Ayesha 1 Alsharu Emad 3 Smith Hannah 4 Alobaidi Bilal 4 Al-Smadi Mohammad 3 Awamlah Nour 5 Hussein Lama 5 Abdelnour Amid 5 Sheth Harsh 46 Veltman Joris A 4 1 Department of biotechnology and genetic engineering, Jordan university of science and technology, Irbid, Jordan 2 The Manchester Centre for Genomic Medicine, University of Manchester NHS foundation trust, Oxford Road, Manchester, UK 3 Reproductive endocrinology and IVF unit, King Hussein medical center, Amman, Jordan 4 Biosciences Institute, Faculty of Medical Sciences, Newcastle University, Newcastle upon Tyne, United Kingdom 5 Biolab Diagnostic Laboratories, Amman, Jordan 6 FRIGE's Institute of Human Genetics, FRIGE House, Ahmedabad, India * Corresponding author: oybatiha@just.edu.jo, Tel. +962 2 7201000 ext. 23466, Fax. +962 2 7201071 01 12 2022 30 6 2021 21 7 2023 26 7 2023 25 5 939946 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Infertility is a common health problem that affects around 1 in 6 couples in the United States, where half of these cases are attributed to male factors. Genetics play an important role in infertility and it is estimated that up to 50% of cases are due to genetic factors. Despite this, many male infertility cases are still idiopathic. This study aimed to identify the presence of possibly pathogenic rare variants in a set of candidate genes related to azoospermia in a Jordanian cohort composed of 69 cases using a next-generation sequencing-based panel covering more than a hundred male infertility related genes. A total of 9 variants were found and validated. Among them, two variants included reported pathogenic variants in CFTR and one novel pathogenic variant in the USP9Y gene. We also report the detection of 6 other variants with uncertain significance in other genes. Interestingly, male cases with CFTR variants did not show the expected cystic fibrosis phenotypes except for infertility. This work helps to uncover the contribution of additional genetic factors to the etiology of male infertility and highlights the importance to obtain more reliable information about the presence of genetic variation in the Jordanian population. Male infertility Azoospermia single molecule Molecular Inversion Probes next generation sequencing CFTR mutations pmcIntroduction Infertility is a worldwide problem, defined as a disease of the reproductive system, results in the failure to achieve a pregnancy after 12 months of regular unprotected sexual intercourse (World Health Organization, 2018). Fertility care is a reproductive right, health equity, and gender equality issue. Approximately, 1 in every six couples in the United States are infertile, and among them, male factor infertility accounts for approximately 50% of causes (Thoma et al., 2013; Zorrilla & Yatsenko, 2013). Despite the high burden, couples who desire but are unable to achieve and maintain a pregnancy, have needs that are not being addressed, especially in lower resource settings worldwide. Yet, the field of reproductive medicine and endocrinology is rapidly growing (Agarwal et al., 2015; World Health Organization, 2018). Male infertility is a multifactorial disease encompassing a wide variety of disorders and can be initially diagnosed by semen fluid analysis (Poongothai et al., 2009). Genetic factors play a major role in idiopathic male infertility (Mazhar S. Al Zoubi et al., 2020; Mazhar Salim Al Zoubi et al., 2020; Plaseska-Karanfilska et al., 2012). The main genetic cause of male infertility is chromosomal abnormalities, which accounts for ~5% of infertile males, and the prevalence increases to 15% in the azoospermic males (Krausz & Riera-Escamilla, 2018; Zorrilla & Yatsenko, 2013). Men with non-obstructive azoospermia have a high prevalence of aneuploidy, particularly in their sex chromosomes. The second most common genetic cause of male infertility is Y chromosome microdeletions affecting the azoospermia factor (AZF) region (Batiha et al., 2012; Zorrilla & Yatsenko, 2013). Microdeletions in this region cause defects in spermatogenesis that lead to the development of azoospermia and oligozoospermia (Krausz & Riera-Escamilla, 2018). There are also male infertility cases caused by defects in single genes including CFTR, DDX3Y, SYCP3, TEX11, AURKC, and DPY19L2, but the number of genes confidently linked to male infertility remains very low (Oud et al., 2019). The poor genetic diagnosis makes a large proportion of infertile males falling in the "idiopathic infertility" category with no obvious reasons explaining their infertility problem. Recent advances in molecular biology technologies such as next-generation sequencing (NGS) has enabled rapid and relatively cost-effective whole-exome and whole-genome sequencing. This, in turn, allowed rapid, sensitive, and efficient detection of the genetic etiologies of many diseases (Gilissen et al., 2011). The development of such technologies is promising to revolutionize the hunt for new genetic markers for many diseases, including male infertility. Recently, we conducted a study on a group of azoospermic Jordanian infertile males and analyzed them for common mutations in the Y chromosome, in addition to the androgen receptor (AR) gene (Batiha et al., 2018). We detected microdeletions in the Y chromosome in only 5% of samples and found no significant difference of a common mutation in the AR gene (Batiha et al., 2018). To further characterize and identify additional causative genetic variants we sequenced these cases using single molecule Molecular Inversion Probes (smMIPs) NGS-based panel covering more than a hundred male infertility-related genes (Oud et al., 2017). Materials and methods Patients The cohort was described previously (Batiha et al., 2018). In summary, 142 unrelated, idiopathic, azoospermic Jordanian Arab males who were previously tested for AZF deletion/duplications and AR-CAG gene repeats were included. Samples with congenital bilateral absence of the vas deferens (CBAVD) was excluded when possible. Patients’ age ranges from 20 and 50 years, with an average of 32.4 ± 6 years. The Institutional Review Board (IRB) at Jordan University of Science and Technology and the ethical committee at King Hussein Medical Center approved the study. Participants were informed about the goal of the study, and signed a written consent form. Samples with Y-chromosome microdeletions, and samples with low DNA concentration and/or poor quality were excluded. The remaining 69 DNA samples were qualified for smMIP sequencing. smMIP based targeted sequencing smMIPs targeting 134 genes were previously described (Oud et al., 2017), with modifications to add recently published infertility genes (Oud et al., 2019). All 69 samples were sequenced in three runs on the Illumina NextSeq platform at an average 1200x depth per amplicon/sample. Genomic regions of interest were captured in a reaction containing a molecule ratio between patients’ gDNA and smMIP of 1:1000. The conditions of smMIP were: for denaturation of DNA ten minutes at 95 ° C, then incubation period for 23 hours, next all non-circular targets were amplified with primers containing barcoded reverse primers by the next PCR conditions: denaturation at 98° c for 30 sec, then 17 cycles of 10 sec at 98 °c, 30 sec at 60 °c, and 30 sec at 72 ° c and finally 2 min. at 72° c. Bioinformatic analysis Data was analyzed as previously described using an in-house smMIP-pipeline (Oud et al 2017). The files produced via the pipeline were run through a custom script in RStudio v3.5.1 (RStudio Team. (2015). RStudio: Integrated Development Environment for R. Boston, MA. Retrieved from http://www.rstudio.com/) which filtered for only variants which showed more than 10 reads at the potential variant loci; this was to ensure reliability in the results (Oud et al., 2017). These variants were then segregated into two separate files based on their inheritance pattern, those with 20% to 80% of all the reads at a given locus being mutated were sorted into the heterozygous file and those with > 80% were then classed as homozygous. These files were then passed through a custom Linux script. This filtered only variants with an allele frequency in the general population of less than 1% (ExAC, gnomAD, and 1000 genomes). The remaining variants were only kept when they occurred in exonic regions such as frameshifts, missense, and stop gained as well as splice site donor and acceptor region mutations. Variant interpretation The filtered variants were prioritized using Alamut® Visual V.2.11 (Interactive Biosoftware, Rouen, France) based on pathogenicity scores provided in the annotated files and additional information on the affected gene itself. The pathogenicity scores highlight how damaging the change in an amino acid is to the overall protein function (Sorts Intolerant from Tolerant (SIFT) and Polymorphism Phenotyping (PolyPhen)). The Combined Annotation Dependent Depletion, Phred scale (CADD Phred) score measures deleteriousness of the variant using the observed variant frequency as the basis for its calculation, the score ranges from 1 to 100 with scores over 10 being classed as being the 10% most deleterious substitutions and scores > 20 being in the top 1% (Rentzsch et al., 2019). All identified variants were also interpreted using the American College of Medical genetics (ACMG) recommendations (Richards et al., 2015). Sanger sequencing University of California, Santa Cruz (UCSC) genome browser, Primer 3, OligoCalc and UCSC in silico PCR tools were all used to design the primers. Sanger sequencing was used to validate all variants classified as pathogenic, likely pathogenic and of uncertain significance. Results DNA samples from 69 azoospermic infertile Jordanian males were analyzed using SmMIP targeted sequencing NGS panel for 134 known and candidate male infertility genes. After filtering for rare likely pathogenic variants, a total of 9 variants were prioritized. Three variants were classified as pathogenic and likely pathogenic (Table 1), and 6 variants were classified as variants of uncertain clinical significance (Table 2). All detected variants were confirmed by Sanger sequencing. We detected a homozygous c.3909C>G p.(Asn1303Lys) variant in the cystic fibrosis transmembrane conductance regulator (CFTR) gene in one of the samples (J34). This is a previously reported pathogenic variant in CFTR (https://www.ncbi.nlm.nih.gov/clinvar/RCV000007556/). The patient is 34 years old and does not show any cystic fibrosis phenotype. Both endo-rectal and scrotal ultrasonography performed did not show any abnormality or congenital bilateral absence of vas deferens (CBAVD). We also detected a homozygous c.3454G>C p.(Asp1152His) variant in the CFTR gene in another sample (SI008). This is also a known and previously reported pathogenic variant in CFTR. The patient history did not indicate a cystic fibrosis phenotype, but we could not re-examine the patient to exclude CBAVD, however, it was a recruitment criterion. Additionally, a novel hemizygous c.6537T>A p.(Tyr2179Ter) likely pathogenic nonsense variant in the ubiquitin-specific protease 9 Y-linked (USP9Y) gene was detected in case J93. Homozygous variants in the minichromosome maintenance 8 homologous recombination repair factor (MCM8) and the lysine-specific demethylase 5D (KDM5D) genes were detected but with uncertain pathogenicity. Both are missense mutations with a predicted p.(Gly333Arg) and p.(Arg1468Trp) amino acid substitutions, respectively. Four other heterozygous variants were detected in the microtubule-associated serine/threonine kinase 2 (MAST2), a meiosis-specific protein with OB domains (MEIOB), UTP14C small subunit processome component (UTP14C) and dynein axonemal heavy chain 6 (DNAH6) genes with uncertain pathogenicity. These variants are predicted to cause either missense or frameshift deletions in the encoded proteins. Discussion Infertility is a major health problem that harms both social and economic levels. In this study, a total of 69 DNA samples from azoospermic infertile Jordanian men were analyzed by MIPs with 6014 probes targeting 134 genes associated with male infertility. A total of 9 variants were found using MIPs and confirmed by Sanger sequencing, these variants include both reported pathogenic and novel variants. Three pathogenic and likely pathogenic variants were found in CFTR and USP9Y genes. Pathogenic variants in the CFTR gene have been associated with different forms of male infertility (Chen et al., 2012). In this study, two homozygous variants in the CFTR gene have been found in two different patients; N1303K and D1152H, none of the patients had cystic fibrosis (CF). CFTR c.3909C>G p.(Asn1303Lys) has been reported to be a pathogenic variant causing CF and is linked to CBAVD (De Braekeleer & Ferec, 1996; Van Hoorenbeeck et al., 2007). However, the patient in this study was normal and confirmed to have the vas deferens by ultrasonography, and none of his family members had CF. Consistent with this, some studies have shown that CFTR pathogenic variants may also cause other non-CBAVD Azoospermia (Chen et al., 2012; Dohle, 2002; Smits et al., 2019), in addition, phenotypic heterogeneity of CF patients with N1303K variant could be explained by the presence of specific haplotypes (Cordovado et al., 2012; Osborne et al., 1992). The contribution of CFTR variants to male infertility has not yet been assessed in Jordan, moreover, the N1303K mutation and its association with cystic fibrosis have not been studied yet in the Jordanian population. On the other hand, D1152H (3454G>C) has been initially linked to CBAVD and CF (Feldmann et al., 2003; Highsmith et al., 2005). Recently, it has shown that this mutation is associated with pancreatitis but not CF (LaRusch et al., 2014), while the CFTR2 database classifies the D1152H as a mutation with variable penetrance (http://www.http.com//www.cftr2). The third patient had a pathogenic novel mutation in the USP9Y gene (c.6537T>A) with unknown inheritance pattern. This mutation causes a premature stop codon in the USP9Y gene located in the azoospermia factor a region (AZFa) of Y-chromosome which is known to cause azoospermia when deletions or premature stop codons occurs (Luddi et al., 2009; Online Mendelian Inheritance in Man, 2019). Furthermore, USP9Y is linked to spermatogenic failure and azoospermia (Krausz et al., 2006; Sun et al., 1999). Early reports have suggested a role for USP9Y in spermatogenic failure associated with azoospermia and male infertility, where point mutations in USP9Y were found in males with spermatogenic failure and were absent in their fertile male siblings or in the control fertile group (Brown et al., 1998; Hall et al., 2003; Sun et al., 1999), however recent papers have challenged this view (Krausz et al., 2006; Luddi et al., 2009). Deletions in USP9Y were found in a normozoospermic male, and both in his fertile father and brother (Luddi et al., 2009), while another paper found that USP9Y does not perform an essential function during spermatogenesis (Krausz et al., 2006), opposite to what has been suggested before. The homozygous substitution mutation –found in this study- could explain the azoospermic phenotype manifested in the patient, or this phenotype could be linked to other genetic or non-genetic factors. In addition to the previously mentioned variants, six novel variants with uncertain pathogenicity have been found, two homozygous missense mutations in MCM8 and KDM5D, two heterozygous frameshift mutations in MAST2 and MEIOB, and two heterozygous missense mutations in UTP14C and DNAH6. MCM8 is crucial for gametogenesis and gonadal development (Tenenbaum-Rakover et al., 2015). Mutations in MCM8 have been shown to cause gonadal and spermatogenesis failure (Lutzmann et al., 2012; Tenenbaum-Rakover et al., 2015). KDM5D is one of the AZFb region genes on Y-chromosome, similar to USP9Y, it has been linked to azoospermia and spermatogenic failure (Rastegar et al., 2015; Yu et al., 2015). The heterozygous frameshift mutations have been found in MAST2 and MIEOB. MAST2 functions in spermatids maturation (PubChem database. National Center for Biotechnology Information, 2016). Mutations in MAST2 have is associated with non-obstructive azoospermia (Huang et al., 2015). MIEIOB has been linked to spermatogenic failure and male infertility associated with oligospermia or azoospermia (GeneCards. The human gene database)(Gershoni et al., 2017). A recent study linked a frameshift mutation in MIEOB with azoospermia (Gershoni et al., 2019). The last two heterozygous missense mutations are found in UTP14C and DNAH6. UTP14C is essential for spermatogenesis (GeneCards.The human gene database). Mutations in UTP14C gene have been linked to spermatogenic arrest and male infertility (Rohozinski et al., 2006). Furthermore, mutations in DNAH6 were shown to be associated with spermatogenic abnormalities and male infertility (Gershoni et al., 2017; Li et al., 2018; Tu et al., 2019). We predict that heterozygous variants have a dominant pathogenic pattern of inheritance with reduced penetrance and could be inherited maternally, or had arisen from denovo mutations in the germline cells. Finally, it is interesting to note that the mutation pick-up rate in this study was significantly lower than the original cohort where this panel was developed and validated (Oud et al., 2017), which shows that genetic variants responsible for infertility in the Jordanian populations may be very different. In conclusion, this work provided the first insight into monogenic causes of male infertility in Jordan and highlighted a different spectrum of genotype-phenotype correlation of known pathogenic CFTR variants in the Jordanian population. For clinical implementation, it is important to obtain more reliable information about the presence of genetic variation in the Jordanian population. Also, it is clear that these kinds of genetic studies should go hand in hand with detailed clinical phenotyping to be able to interpret these findings in a clinical setting. Acknowledgment The authors would like to thank all volunteers who participated in this study. Many thanks to the Royal Medical services for approving and participating in the study. This study was funded by the deanship of research at Jordan University of science and technology (grant # 12/2019). J.A.V. is supported by grants from The Netherlands Organization for Scientific Research (918-15-667) as well as an Investigator Award in Science from the Wellcome Trust (209451). Author biography Osamah Batiha: assistant professor of cell & developmental genetics, Jordan University of science and technology, Irbid, Jordan George J Burghel: Principal Clinical Scientist in cancer pharmacogenetics, Manchester Centre for Genomic Medicine, Manchester University Hospital, Manchester, UK Emad Alsharu: Senior specialist IVF /OBGYN, Royal medical services, Amman, Jordan Joris A Veltman: Jacobson chair of Personalized Medicine, Dean Biosciences Institute, Faculty of Medical Sciences, Newcastle University, Newcastle upon Tyne, United Kingdom Harsh Sheth: Assistant Professor and Head of Advanced Genomic Technologies Division at FRIGE's Institute of Human Genetics Table 1 Summary of the variants highlighted and validated as being Pathogenic or Likely pathogenic in 3 patients. Sample ID Gene Variant Protein Change Consequence Zygosity SIFT PolyPhen CADD Phred ACMG-AMP classification Gnomad Population frequency ClinVar J34 CFTR c.3909C>G p.(Asn1303Lys) Missense Homozygous 0 1.00 27.9 Class 5: Pathogenic - known CFTR mutation 1.39x10^-4 10 pathogenic entries SI008 CFTR c.3454G>C p.(Asp1152His) Missense Homozygous 0 0.80 31.0 Class 4: Likely pathogenic 3.75x10^-4 15 path/LP entries on ClinVar J93 USP9Y* c.6537T>A p.(Tyr2179Ter) Stop gained Hemizygous - - - Class 5: Pathogenic Absent Absent Table 2 Summary of the 6 variants highlighted and validated as being Uncertain in significance in 5 patients. Sample ID Gene name Variant Protein Change Consequence Zygosity SIFT PolyPhen CADD Phred ACMG-AMP classification Gnomad ClinVar SI002 MCM8 c.997G>A p.(Gly333Arg) Missense Homozygous 0 0.99 31.0 Class 3: Uncertain significance Absent Absent J53 KDM5D c.4402C>T p.(Arg1468Trp) Missense Hemizygous 0 0.80 - Class 3: Uncertain significance 2/57279 (South Asian) Absent SI004 MAST2 c.3039delC p.(Thr1014GlnfsTer34) Frameshift deletion Heterozygous - - - Class 3: Uncertain significance Absent Absent SI004 MEIOB c.1098delC p.(Leu367TrpfsTer13) Frameshift deletion Heterozygous - - - Class 3: Uncertain significance Absent Absent SI008 UTP14C c.268C>T p.(Leu90Phe) Missense Heterozygous 0 0.99 24.9 Class 3: Uncertain significance Absent Absent SI010 DNAH6 c.2837T>C p.(Leu946Pro) Missense Heterozygous 0 0.99 26.0 Class 3: Uncertain significance Absent Absent Disclosure Statement The authors report no conflict of interest Agarwal A Mulgund A Hamada A Chyatte MR A unique view on male infertility around the globe Reproductive Biology and Endocrinology 2015 10.1186/s12958-015-0032-1 Al Zoubi Mazhar S Al-Batayneh K Alsmadi M Rashed M Al-Trad B Al Khateeb W Aljabali A Otoum O Al-Talib M Batiha O 4,977-bp human mitochondrial DNA deletion is associated with asthenozoospermic infertility in Jordan Andrologia 2020 10.1111/and.13379 Al Zoubi MazharSalim Bataineh H Rashed M Al-Trad B Aljabali AAA Al-Zoubi RM Al Hamad M Issam AbuAlArjah M Batiha O Al-Batayneh KM CAG Repeats in the androgen receptor gene is associated with oligozoospermia and teratozoospermia in infertile men in Jordan Andrologia 2020 10.1111/and.13728 Batiha O Al-Ghazo MA Elbetieha AM Jaradat SA Screening for deletions in the AZF region of Y chromosome in infertile Jordanian males Journal of Applied Biological Sciences 2012 6 2 Batiha O Haifawi S Al-Smadi M Burghel GJ Naber Z Elbetieha AM Bodoor K Sumadi A Al, Swaidat S Jarun Y Abdelnour A Molecular analysis of CAG repeat length of the androgen receptor gene and Y chromosome microdeletions among Jordanian azoospermic infertile males Andrologia 2018 10.1111/and.12979 Brown GM Furlong RA Sargent CA Erickson RP Longepied G Mitchell M Jones MH Hargreave TB Cooke HJ Affara NA Characterisation of the coding sequence and fine mapping of the human DFFRY gene and comparative expression analysis and mapping to the Sxrb interval of the mouse Y chromosome of the Dffry gene Human Molecular Genetics 1998 10.1093/hmg/7.1.97 Chen H Ruan YC Xu WM Chen J Chan HC Regulation of male fertility by CFTR and implications in male infertility Human Reproduction Update 2012 18 6 703 713 10.1093/humupd/dms027 22709980 Cordovado SK Hendrix M Greene CN Mochal S Earley MC Farrell PM Kharrazi M Hannon WH Mueller PW CFTR mutation analysis and haplotype associations in CF patients ☆ Molecular Genetics and Metabolism 2012 105 2 249 254 10.1016/j.ymgme.2011.10.013 22137130 De Braekeleer M Ferec C Mutations in the cystic fibrosis gene in men with congenital bilateral absence of the vas deferens Molecular Human Reproduction 1996 2 9 Dohle GR Genetic risk factors in infertile men with severe oligozoospermia and azoospermia Human Reproduction 2002 17 1 13 16 10.1093/humrep/17.1.13 11756355 Feldmann D Couderc R Audrezet MP Ferec C Bienvenu T Desgeorges M Claustres M Mittre H Blayau M Bozon D Malinge MC CFTR genotypes in patients with normal or borderline sweat chloride levels Human Mutation 2003 10.1002/humu.9183 GeneCards The human gene database MEIOB (n.d.) GeneCards The human gene database UTP14C Gene (n.d.) Gershoni M Hauser R Barda S Lehavi O Arama E Pietrokovski S Kleiman SE A new MEIOB mutation is a recurrent cause for azoospermia and testicular meiotic arrest Human Reproduction 2019 34 4 666 671 10.1093/humrep/dez016 30838384 Gershoni M Hauser R Yogev L Lehavi O Azem F Yavetz H Pietrokovski S Kleiman SE A familial study of azoospermic men identifies three novel causative mutations in three new human azoospermia genes Genetics in Medicine 2017 19 9 998 1006 10.1038/gim.2016.225 28206990 Gilissen C Hoischen A Brunner HG Veltman JA Unlocking Mendelian disease using exome sequencing Genome Biology 2011 10.1186/gb-2011-12-9-228 Hall NM Brown GM Furlong RA Sargent CA Mitchell M Rocha D Affara NA Usp9y (ubiquitin-specific protease 9 gene on the Y) is associated with a functional promoter and encodes an intact open reading frame homologous to Usp9x that is under selective constraint Mammalian Genome 2003 10.1007/s00335-002-3068-4 Highsmith WE Friedman KJ Burch LH Spock A Silverman LM Boucher RC Knowles MR A CFTR mutation (D1152H) in a family with mild lung disease and normal sweat chlorides [1] Clinical Genetics 2005 10.1111/j.1399-0004.2005.00459.x Huang N Wen Y Guo X Li Z Dai J Ni B Yu J Lin Y Zhou W Yao B Jiang Y A Screen for Genomic Disorders of Infertility Identifies MAST2 Duplications Associated with Nonobstructive Azoospermia in Humans1 Biology of Reproduction 2015 93 3 10.1095/biolreprod.115.131185 Krausz C Degl’Innocenti S Nuti F Morelli A Felici F Sansone M Varriale G Forti G Natural transmission of USP9Y gene mutations: A new perspective on the role of AZFa genes in male fertility Human Molecular Genetics 2006 10.1093/hmg/ddl198 Krausz C Riera-Escamilla A Genetics of male infertility Nature Reviews Urology 2018 10.1038/s41585-018-0003-3 LaRusch J Jung J General IJ Lewis MD Park HW Brand RE Gelrud A Anderson MA Banks PA Conwell D Lawrence C Mechanisms of CFTR Functional Variants That Impair Regulated Bicarbonate Permeation and Increase Risk for Pancreatitis but Not for Cystic Fibrosis PLoS Genetics 2014 10 7 10.1371/journal.pgen.1004376 Li L Sha Y-W Xu X Mei L-B Qiu P-P Ji Z-Y Lin S-B Su Z-Y Wang C Yin C Li P DNAH6 is a novel candidate gene associated with sperm head anomaly Andrologia 2018 50 4 e12953 10.1111/and.12953 Luddi A Margollicci M Gambera L Serafini F Cioni M De Leo V Balestri P Piomboni P Spermatogenesis in a man with complete deletion of USP9Y New England Journal of Medicine 2009 10.1056/NEJMoa0806218 Lutzmann M Grey C Traver S Ganier O Maya-Mendoza A Ranisavljevic N Bernex F Nishiyama A Montel N Gavois E Forichon L MCM8- and MCM9-Deficient Mice Reveal Gametogenesis Defects and Genome Instability Due to Impaired Homologous Recombination Molecular Cell 2012 47 4 523 534 10.1016/j.molcel.2012.05.048 22771120 Online Mendelian Inheritance in Man, O OMIM: 415000 2019 Osborne L Santis G Schwarz M Klinger K Dörk T McIntosh I Schwartz M Nunes V Macek M Reiss J Incidence and expression of the N1303K mutation of the cystic fibrosis (CFTR) gene Human Genetics 1992 89 6 653 658 10.1007/bf00221957 1380943 Oud MS Ramos L O’Bryan MK McLachlan RI Okutman Ö Viville S de Vries PF Smeets DFCM Lugtenberg D Hehir-Kwa JY Gilissen C Validation and application of a novel integrated genetic screening method to a cohort of 1,112 men with idiopathic azoospermia or severe oligozoospermia Human Mutation 2017 10.1002/humu.23312 Oud MS Volozonoka L Smits RM Vissers LELM Ramos L Veltman JA A systematic review and standardized clinical validity assessment of male infertility genes Human Reproduction (Oxford, England) 2019 10.1093/humrep/dez022 Plaseska-Karanfilska D Noveski P Plaseski T Maleva I Madjunkova S Moneva Z Genetic causes of male infertility Balkan Journal of Medical Genetics 2012 10.2478/v10034-012-0015-x Poongothai J Gopenath TS Manonayaki S Poongothai S Genetics of human male infertility Singapore Medical Journal 2009 PubChem database. National Center for Biotechnology Information MAST2 - microtubule associated serine/threonine kinase 2 (human) 2016 https://pubchem.ncbi.nlm.nih.gov/gene/MAST2/human Rastegar DA Tabar MS Alikhani M Parsamatin P Sahraneshin Samani F Sabbaghian M Sadighi Gilani MA Mohammad Ahadi A Mohseni Meybodi A Piryaei A Ansari-Pour N Isoform-level gene expression profiles of human y chromosome azoospermia factor genes and their X chromosome paralogs in the testicular tissue of non-obstructive azoospermia patients Journal of Proteome Research 2015 14 9 3595 3605 10.1021/acs.jproteome.5b00520 26162009 Rentzsch P Witten D Cooper GM Shendure J Kircher M CADD: Predicting the deleteriousness of variants throughout the human genome Nucleic Acids Research 2019 10.1093/nar/gky1016 Richards S Aziz N Bale S Bick D Das S Gastier-Foster J Grody WW Hegde M Lyon E Spector E Voelkerding K Standards and guidelines for the interpretation of sequence variants: A joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology Genetics in Medicine 2015 10.1038/gim.2015.30 Rohozinski J Lamb DJ Bishop CE UTP14c Is a Recently Acquired Retrogene Associated with Spermatogenesis and Fertility in Man1 Biology of Reproduction 2006 74 4 644 651 10.1095/biolreprod.105.046698 16354793 Smits RM Oud MS Vissers LELM Lugtenberg D Braat DDM Fleischer K Ramos L D’Hauwers KWM Improved detection of CFTR variants by targeted next-generation sequencing in male infertility: a case series Reproductive BioMedicine Online 2019 10.1016/j.rbmo.2019.08.005 Sun C Skaletsky H Birren B Devon K Tang Z Silber S Oates R Page DC An azoospermic man with a de novo point mutation in the Y-chromosomal gene USP9Y Nature Genetics 1999 10.1038/70539 Tenenbaum-Rakover Y Weinberg-Shukron A Renbaum P Lobel O Eideh H Gulsuner S Dahary D Abu-Rayyan A Kanaan M Levy-Lahad E Bercovich D Minichromosome maintenance complex component 8 (MCM8) gene mutations result in primary gonadal failure Journal of Medical Genetics 2015 52 6 391 399 10.1136/jmedgenet-2014-102921 25873734 Thoma ME McLain AC Louis JF King RB Trumble AC Sundaram R Buck Louis GM Prevalence of infertility in the United States as estimated by the current duration approach and a traditional constructed approach Fertility and Sterility 2013 10.1016/j.fertnstert.2012.11.037 Tu C Nie H Meng L Yuan S He W Luo A Li H Li W Du J Lu G Lin G Identification of DNAH6 mutations in infertile men with multiple morphological abnormalities of the sperm flagella Scientific Reports 2019 9 1 10.1038/s41598-019-52436-7 Van Hoorenbeeck K Storm K van den Ende J Biervliet M Desager KN N1303K and IVS8-5T, clinical presentation within a family with atypical cystic fibrosis Journal of Cystic Fibrosis 2007 6 3 220 222 10.1016/j.jcf.2006.10.002 17127107 World Health Organization Infertility definitions and terminology 2018 WHO https://doi.org/http://www.who.int/reproductivehealth/topics/infertility/definitions/en/ Yu XW Wei ZT Jiang YT Zhang SL Y chromosome azoospermia factor region microdeletions and transmission characteristics in azoospermic and severe oligozoospermic patients International Journal of Clinical and Experimental Medicine 2015 8 9 14634 14646 E-Century Publishing Corporation 26628946 Zorrilla M Yatsenko AN The Genetics of Infertility: Current Status of the Field Current Genetic Medicine Reports 2013 1 4 247 260 10.1007/s40142-013-0027-1

PMC007xxxxxx/PMC7614818.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9508166 Neurobiol Learn Mem Neurobiol Learn Mem Neurobiology of learning and memory 1074-7427 1095-9564 36174888 7614818 10.1016/j.nlm.2022.107685 EMS177467 Article Memory recall: New behavioral protocols for examining distinct forms of context specific recall in animals Prodan A. * Davies H. * Eneqvist H. Mastroberardino G. Wijayathunga H. Wardlaw K. Morris R.G.M. Laboratory for Cognitive Neuroscience, Centre for Discovery Brain Sciences, Edinburgh Neuroscience, University of Edinburgh, 1 George Square, Edinburgh, EH8 9JZ, United Kingdom * Joint first authors 01 11 2022 26 9 2022 05 7 2023 26 7 2023 195 107685107685 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. This study outlines two novel protocols for examining context specific recall in animals prior to embarking on neurobiological studies. The approach is distinct from and contrasts with studies investigating associative familiarity that depend upon procedural variations of the widely used novel object recognition task. It uses an event arena in which animals are trained across numerous sessions to search for, find and dig up reward from sandwells during sample and choice trials – a prominent spatial event for a rodent. The arena could be laid out as either of two highly distinct contexts with which the animals became fully familiar throughout training. In one protocol, the location of the correct sandwell in each context remained stable across days, whereas in the other, the correct digging location varied in a counterbalanced manner across each successive session. Thus, context-specific recall of the spatial location of successful digging during choice trials was either from a stable long-term memory or could reflect context specific spatial recency of the location where reward had been available that session. Both protocols revealed effective memory recall in choice and probe tests which, at the point of test, were procedurally identical in both cases. context specific recall episodic memory event arena allocentric spatial representation context-specific recency pmcIntroduction The concept of ‘context-specificity’ in memory is discussed extensively in experimental psychology and neuroscience (Nadel & Maurer, 2020). It can carry different meanings, ranging from remembering where objects are located within a context or where events happen, through to non-spatial facets such as social context. When context is spoken about colloquially, it is generally in the domain of recall rather than recognition, as in “Where do I go to find the Empire State building?” or “When or where did something happen?”. The answer is likely to involve recall, with a reply to the former involving recall from a stable long-term spatial memory but a reply to the latter having a more episodic character. Context has been examined experimentally in neurobiological animal studies of memory for many years. Prominent examples include its role in access to reward (or punishment). Cues are present associated with recognizing a context or another, and the task is to associate a biologically significant event (e.g. shock) with one or another recognised context. This is the simplest form of context discrimination of which the paradigmatic example is context fear conditioning (Anagnostaras et al., 1999; Kim et al., 1991). This protocol is now used in numerous ‘gain-of-function’ optogenetic studies pointing to a role for the dentate gyrus in contextual identification (Josselyn & Tonegawa, 2020). In other studies, a context sets the occasion for some other stimulus to be rewarded but that same stimulus goes unrewarded in a different context, and vice versa — a context-specific stimulus discrimination task in which one stimulus was repeatedly rewarded in one context and another non-rewarded, and these reward assignments reversed in the other context (Good & Honey, 1991). Learning such a task has been shown to depend on the integrity of the hippocampus. Context may also define the presence of objects that animals can recognize and investigate, such as discriminable familiar and novel objects during an initial exploratory sample trial in a context to which the animals have previously been habituated (Ennaceur & Aggleton, 1994; Ennaceur & de Souza Silva, 2018). In a subsequent choice trial, one (or more) of these objects are either temporarily absent, replaced by another novel object, moved to a different context, or moved to a different location in the same or a different context (Dix & Aggleton, 1999). Such procedural manipulations alter the ‘associative’ familiarity or novelty of these objects relative to the sample trial. Some detailed novel object protocols have examined ‘context-specific’ novelty (Langston & Wood, 2010) and its relevance to, for example, neurodevelopmental disorders (Asiminas et al., 2019), but have tended to do so over relatively short memory intervals (e.g., 3 min for Langston and Wood, 2010). Examining the contribution of context to recognition memory is central to a wide range of work, the typical result being that the experimental animal engages in renewed exploratory behavior in the incongruous setting relative to the congruent one, but at the expense of using only relatively short memory retention intervals. Spontaneous exploration has, nonetheless, revealed much about the anatomical organisation of recognition memory (Aggleton & Nelson, 2020; Aggleton & Pearce, 2001). In contrast, our interest in the present work centers initially on the ability to recall rather than merely recognise in a context-specific manner. Recall about where something is located or what happened in a specific context is not only frequent in everyday life, but also more common than recognition in ordinary discourse; a casual question from someone about whether one can remember something is rarely accompanied by a forced choice test of alternatives, except perhaps in a police line-up. A second key idea is that recall from a stable long-term memory of facts (‘semantic-like’ recall) is likely different from remembering something that happened recently (‘episodic-like’ recall). For example, you likely recall where and when you last saw someone. Episodic recall engages a process of something ‘coming to mind’ from the past, but it is important to realize that recall may not always be ‘episodic’. It may also be from a stable long-term memory – more akin to recall from semantic memory. This important distinction pertains to animal tasks as well. Navigation in the standard watermaze task involves recall but it is not episodic and does not require mental time travel (Morris et al., 1982). The hidden escape platform is generally in a fixed location across trials and the animal’s swim trials start at each of the four cardinal locations around the pool (N, S, E and W) in quasi-random sequence. After approximately 10 days of training, the animal can be placed in the pool at any location, and it will navigate using a relatively direct path to the escape location. There is no local cue associated with the hidden escape platform – in effect nothing to ‘recognise’ beyond the context which itself provides no explicit guidance cues about where to go. The watermaze limits the use of path-integration by animals (because there is no return path) and, with this type of standard training from multiple start locations, the spatial representation is definitively allocentric. Note that there is no need for the animal at the start of a trial to remember what it did in the pool on some prior occasion. With respect to context specificity, watermaze navigation to a stable planned destination may also be achieved by a learned and relatively direct route that differs from one context to another (Bannerman et al., 1995). In these training protocols, it is not an ‘episodic-like’ memory task. Other protocols in the watermaze are, however, more ‘event-like’ as outlined and discussed by Steele and Morris (1999). The key change in that study was to move the correct location of the hidden platform each day such that the animal could not know the correct location of the first daily trial. However, the same study showed that the new daily location could be learned in one trial. This study was later followed by similar protocols in the event arena (Bast et al., 2005). Memory recall has, however, been more exactingly investigated in ‘episodic-like’ memory tasks such as ‘what-where-when’ food choice after relatively long memory delays in corvids (Clayton & Dickinson, 1998) and rats using carefully designed radial-maze tasks (Babb & Crystal, 2006; Crystal, 2021). There are also tasks testing whether absolute time of memory encoding or the passage of time is recalled at the point of retrieval with data indicating that either may be used depending on procedure (Roberts et al., 2008; Zhou & Crystal, 2009). These tasks may, however, not be the only way of thinking about temporal ‘recency’. Another concept, first introduced in the human literature by Morton and colleagues with reference to the concept of “headed records”, explores the memory representations of events for the last meeting between two individuals – whether be it recently or much longer ago (Morton et al., 1985). It is this sense of ‘recency’ that is explored in this study. We here outline two novel ‘recall’ protocols, as a prelude to later neurobiological studies with lesions or drugs, to examine these distinct facets of context-specific recall. Specifically, we look in Phase 1 at the learning and then recall of a stable, allocentrically-defined context-specific target location, unmarked by local cues, to which an animal must go to secure reward (non-episodic); and in Phase 2 at the recall of the most recent location of a target whose position in a test arena changes daily in an apparently unstable manner (episodic-like). Although previously developed paradigms involving the use of classical and instrumental conditioning, or spatial learning protocols such as the watermaze, radial arm maze, or T-maze have been used to address certain facets of associative learning and spatial navigation, their utility in investigating episodic-like memory is limited because there is no demonstration of either information in context, nor that this information is changing over time. This motivated the development of our novel behavioral protocols using the event arena, a specifically designed customizable platform that permits disentangling event encoding and recall as well as long-term (multiple days) and short-term (several hours) context-specific object-location associations. The animals are trained in these sample trials to approach a cryptic sandwell containing hidden reward from which they can retrieve and then carry food to an allocentrically defined homebase to eat (Fig. 1A; Broadbent et al., 2020). The sample trials occur in rapid succession but do not involve discrimination between correct and incorrect locations; memory of where digging is occurring on each trial may nonetheless be incidentally encoded. The sample trials are followed, circa 1.5 h later, by a ‘choice trial’ in each context in which there are 6 sandwells with only the correct sandwell in that context being rewarded (6-alternative forced-choice, win-stay strategy; Fig. 1B). The choice trial provides a measure of memory recall. In a two-phase study, animals began Phase 1 by performing 2 sample trials (from either S, E or W) to a stable location in each context for 20 training sessions and 3 probe tests. The animals then performed one choice trial in each context per session, with the rewarded location in context A and that in context B being different but stable across sessions. An additional purpose of this phase was to provide clear evidence that animals could distinguish the contexts, a point often assumed but not directly tested in other studies. In the second Phase of the study, also consisting of 20 training sessions, we introduced the ‘recency’ dimension into the protocol. After demonstrating successful context-specific recall in Phase 1, the animals now had to remember in each session which was the most recently rewarded sandwell location in context A or context B. This location was randomly changed across sessions (Fig. 1C). By virtue of the animals now needing to recall where the correct sandwell was last found in each context, we conducted only one memory recall choice trial at the end of each day – and at a time-interval well beyond the domain of short-term memory. Whereas the former protocol is non-episodic, the latter is ‘episodic-like’ in Clayton and Dickinson’s (1998) terminology. A distinctive feature of the choice trials is that they are procedurally identical in both protocols; what differs is only the type of memory recall required of the animals. Results The animals (N = 14; cohort 1 n = 8; cohort 2 n = 6) gradually learned the ‘stable’ Phase 1 recall protocol, beginning from a near chance level of choice trial performance (block 1) and rising to >80% correct over 10 blocks of training (blocks 7–10; 2 sessions/block; Fig. 2A). Individual animals learned to discriminate the two contexts and selectively approach and dig at the rewarded location in context A and the one in context B. A repeated measures ANOVA showed the increase in performance across sessions was highly significant (F(7.2, 194.3) = 12.48, p < 0.001, Greenhouse-Geisser correction). Furthermore, by blocks 6–10, the performance was highly significantly above chance (t(27) = 13.77, p < 0.001) with, on block 10, excellent performance and low variability (mean ± SEM; 85.71 ± 2.59%). This would be good performance in a 2-alternative forced-choice task and is exceptional for a 6-alternative task, indicating the accumulation of spatial knowledge across sessions (z score = 13.78). There may have been a trend to learn context B slightly earlier than context A but, overall, there was no significant interaction between context and blocks following asymptotic learning (S13–23; F(4, 108) = 0.734, p > 0.05; Fig. 2B). The final performance index score for contexts A and B over sessions 13–22 did not differ (Fig. 2B left; t(13) = 1.84, p > 0.05). Three separate probe tests (food reward absent, time spent digging measured over 60 s) were conducted at the start, mid-point, and end of training (sessions 1, 12, and 23, respectively). The probe tests were scheduled 1 h after the sample trials. The figure shows the striking growth of a significant interaction between probe trial and digging location (F(2.47, 32.23) = 9.38, p < 0.001) with a steady increase in accurate recall of the context-specific location discrimination. By PT3, the overwhelming proportion of time was spent digging at the contextually-specific correct location (68.23 ± 7.37% relative to chance at 16.67%; correct vs. chance, t(13) = 6.99, p < 0.001). There was a significant difference between digging time (%) in the correct sandwell and the incorrect sandwell(s) in PT2 (mean difference ± SE = 29.17 ± 7.65; 95% CI = 8.17– 50.16%, p < 0.01, Bonferroni; Fig. 2C) and PT3 (mean difference ± SE = 64.54 ± 8.62, 95% CI = 40.87–88.22%, p < 0.001, Bonferroni). While digging errors could occur to any of the other 5 sandwells, these errors could be categorically distinguished with respect to whether an error was to a sandwell location that was always incorrect or one that was correct in the other context, and the figure captures that distinction. In PT3, there was a significant difference between digging time (%) in the correct sandwell compared to that in the sandwell rewarded in the other context (mean difference ± SE = 51.21 ± 10.03, 95% CI = 23.66–78.75%, p < 0.001, Bonferroni). Moreover, when the data was examined with respect to performance in the opposite context, the sandwell correct in the other context was never above chance. There were, however, significant differences in time spent digging at the correct location in the other context and other incorrect locations (PT2: mean difference ± SE = 12.18 ± 4.34, p < 0.05; PT3: mean difference ± SE = 13.34 ± 2.85, p < 0.01). Observation of the digging time at the incorrect locations in the stable location Phase 1 probe trials reveals that the animals spent successively less time digging at locations that were never rewarded. Only two spatial locations were ever rewarded per animal, but the key point is that the choice of the correct location in the other context is never above chance, whereas the choice of the correct location in the appropriate context improved across probe tests reflecting the binding of spatial location to context in this task. In the last session of Phase 1, a control procedure was conducted in which choice performance in the regular daily training protocol was tested in a single ‘non-encoding’ (NE) session. This involved not scheduling the normal sample trials prior to the choice trials (Fig. 2A). Despite the absence of sample trials, the animals’ choice performance in this NE control was excellent (mean = 88.57 ± 2.75%) and indistinguishable from that of block 10 and PT3 (Fig. 2A). This indicates that the animals had developed a long-term context-specific location memory that they could recall without needing the ‘reminder’ of the daily sample trials. We then turned to Phase 2, the episodic-like ‘recency’ protocol. With two exceptions, the task performed by the animals in this second Phase was identical to that of the earlier stable Phase (sample trials followed by choice). First, while we began by also having 2 sample trials, we later increased this to 6. Our findings indicated above-chance spatial recency memory with 2 sample trials (61.99% ± 2.08%; t(7) = -5.77, p < 0.001; Fig. 3A) and better absolute performance with 6 sample trials (73.83 ± 3.98% (t(5) = 5.99, p < 0.001; Fig. 3A). This improvement was significant (t(5) = 3.90, p < 0.01). The second exception was that on any one day/session, the choice trial was only tested in one of the two contexts but not both as in Phase 1 (the reason for this latter change being because, when choice trials were performed in both contexts in Cohort 1, performance in the choice trial in the second-tested context was influenced by the performance shown by the animals in the first context – likely a non-specific absolute recency effect. As we only wanted to measure context-specific recall of the most recent context-sample location association, choice trials were thereafter performed for only one context, with context A and context B being chosen in a random but counterbalanced order across sessions. A consideration is: what do these data really represent? We considered three alternative explanations other than context-specific spatial recency. First, while we expected the scores reached at asymptote to be lower than that reached during the stable protocol (85%), both scores (62% and 74%) were good, above chance, and had lower inter-animal variability. However, performance might have been good in one context but not the other. The choice trial data for the 6-sample condition are therefore plotted as a function of the context in which recall was being expressed, revealing no difference between contexts A and B (t(5) = 0.07, p > 0.05; Fig. 3B and 3C-left). A second possibility was that animals were only displaying ‘absolute’ recency over time (Roberts et al., 2008), rather than ‘context-specific’ recency. A typical sequence of daily trials for an individual animal could have been sample trials (context A), sample trials (context B), choice trial (context A); or sample trials (context A), sample trials (context B), choice trial (context B). With the order of the contexts also counterbalanced across the sample trials, performance in the choice trial might have been dominated by high scores only when the context for the choice trial was the same as the immediately preceding context of the second set of sample trials, i.e. absolute recency, with poorer performance in recall of the earlier different context, that collectively average to above chance. In practice, we observed no difference in average performance between sessions in which the choice trial tested was in the ‘same’ context as the one used for the immediately preceding sample trials, i.e. ~15 min earlier, or in those given in the ‘different’ context of approximately 1.5 h earlier (Fig. 3C-middle; t(71) = -1.11, p > 0.05). This indicates that performance was guided by the recall of a context-specific recency memory rather than absolute recency. Third, we also considered whether alterations in the relative familiarity of the two intra-arena cues might contribute to apparent recall, guiding the animal to the vicinity of the correct sandwell. This seems unlikely as the animals were now familiar with all four intra-arena cues (two in each context) over as many as 23–43 sessions. Still, the prediction from a familiarity perspective might be that performance would be better for locations 3 and 4 in the arena (Fig. 1B), where the correct sandwell was directly beside an intra-arena cue, than for the other four more remote locations. We observed no such effect (Fig. 3C-right; t(5) = 1.17, p = 0.29). Performance during the probe tests offered a further way to examine the context-specificity of recall. The measure is percent time digging at (a) the context-specifically correct sandwell, (b) the sandwell correct in the other context, or (c) the other sandwells that were incorrect on that session (but may be correct on other sessions). No accessible food was available. There was a significant difference between digging time (%) in the sandwells as a function of context-specific recency across all probe trials (6 sample condition: (F(2, 10) = 16.72, p < 0.001; Fig. 3D). Further, the digging time (%) in the correct sandwell was significantly different from that in the correct sandwell for the other context (mean difference ± SE; 14.63 ± 3.93, p < 0.001), and that in the incorrect sandwells for that session (mean difference ± SE; 24.23 ± 5.21, p < 0.001). However, as shown in Fig. 3D, no difference in digging time was observed between that in the correct sandwell for the other context and that in the incorrect sandwells (mean difference ± SE; 9.60 ± 3.29, p > 0.05). In this protocol, the incorrect sandwells are, of course, sometimes rewarded and thus the pattern of declining digging time in “incorrect” sandwells across training shown in Phase 1 does not occur. Finally, as an internal control and unexpected test, we also conducted an KE session. Whereas in Phase 1, the prediction was that the daily sample trial would be unnecessary, the opposite prediction prevails for the recency Phase 2 as the correct daily location is defined by what happens in terms of memory encoding on the sample trials. As predicted, choice performance fell to chance (one sample t-test, NE vs chance; p > 0.05; Fig. 3A, orange shading, labelled NE). Moreover, an independent-samples t-test indicated a significant difference between the performance in the NE session in Phase 1 (mean ± SEM; 88.57 ± 2.75%) and that in Phase 2 (56.67 ± 14.06%; t(18) = -3.25, p < 0.01). Discussion These novel protocols show that rats can readily (1) learn and later recall a stable context-specific spatial location of reward revealed as preferential approach to a different sandwell in each of two contexts; and separately (2) recall successfully the most recent location to approach in each context when the correct digging location is varied from day to day. In the latter case, access to a stable long-term memory provides allocentrically encoded information about the nature of each context, but the animal’s memory recall on the choice trial was of the most recent event of digging successfully for food in the contextually appropriate location. One interest in developing these protocols was in exploring the transition from novelty-associated ‘recognition’ memory tasks to new ones based on ‘recall’. Recognition memory has been imaginatively modelled with ever more complex protocols that capture different facets of recognition – the absolute familiarity of an object, the relative familiarity/novelty of an object’s location, the formation of object-location associations and recognition of when these are misplaced in the same or different contexts (Aggleton & Nelson, 2020). In these tasks, the animal does not express either ‘correct’ or ‘incorrect’ behavior – it investigates an object and the extent to which it does so is measured quantitatively. The differential investigation of objects is then taken as a measure of the different forms of recognition memory dissociated largely through lesion studies complemented by studies measuring immediate early-gene activation (such as c-fos). It is claimed that the more complex versions of these tasks model facets of ‘episodic-like’ memory; they may do, but they may reflect no more than different forms of associative familiarity with no re-experiencing of a past event. In contrast, in a recall task, the animal must express behavior that reflects what it knows (non-episodic), or it remembers by re-experiencing a past event (episodic) in the absence of distinctive objects directly associated with what is being recalled. The former includes approaching one cryptic location rather than another in a spatial task. This is striking in the watermaze (Morris et al., 1982) because, from any cardinal starting position around the pool and using allocentric coding, a trained animal heads relatively directly to the correct stable position and lingers there searching for the platform until found, or if absent in a probe test, in a more extended but localised search. In effect, the animal heads directly for an object it cannot see, hear, or smell, and cannot feel until it has got there. The watermaze has, however, major limitations in that motivation cannot readily be manipulated nor is it suitable for electrophysiological, optical imaging or optogenetic procedures. In contrast, not only is the event arena much easier to use for these additional neuroscience techniques, but the level of deprivation of the animals and the reward incentive can be manipulated. In one study (Wang et al., 2010), a contrast is made between receiving a 1-pellet reward and a 3-pellets reward; in the former case there is clear forgetting over 24 h, whereas in the latter, memory of a specific daily location was above chance at 24 h. A distinctive feature of the present protocols, which are based in part on the watermaze, is that the appearance of the apparatus at the start of any choice trial is identical – irrespective of whether the animal is in the stable target or recency target task. Extra-arena and two intra-arena cues stably define one context from the other across all sessions. The animal is in any of 3 start locations (E, S, or W) but, when the door opens to reveal the arena, the appearance of the arena and its intra-arena cues is unchanged apart from the ‘scene’ being rotationally different from each cardinal direction. The two tasks require carrying the reward from the correct sandwell to the north sidewall ‘homebase’; such carrying has been shown to be a natural behavior of rats (Whishaw et al., 1995) and the home-base definitively renders the task allocentric (Broadbent et al., 2020). What the animal must then do depends on the memory strategy being followed: it may recall the stable sandwell location appropriate to that context (Phase 1; ‘non-episodic’ memory); or it must go to the last location at which it found a rewarded sandwell in that context (Phase 2; a component of ‘episodic-like’ memory). Note that the startbox location used on a choice or probe trial is different from some of the starting locations of the sample trials. Thus, remembering an egocentrically directed path will not work. There are other ingenious protocols for animals that examine recall. One example is a study by Eacott et al.. (2005) in which rats had to recall whether to turn left or right in a T-maze to approach a novel object that could not be seen (heard or smelled) at the choice point of either of two contexts. The animals could perform this ‘what-where-which’ task, and performance was disrupted by fornix lesions. A follow up study dissociated object familiarity from the ability to recall (Easton et al., 2009). Other similar examples include studies from Crystal Laboratory of ‘episodic-like’ memory in rodents based on studies of ‘foodstuff-location-time’ associations in the radial maze (Babb & Crystal, 2006; Crystal & Smith, 2014). Not only do these studies involve recall, but they also meet certain additional criteria that characterize a definitive animal model of episodic memory, including ‘what-where-which’ in the case of the Eacott studies. Our approach here shows definitively that the animals can successfully distinguish the two contexts (something that is implied but not shown directly in these other studies), that the animals can successfully recall the location in allocentric space where events have occurred in a specific context, and that the animals can do the episodic-like recency task repeatedly across sessions without interference. Our approach is, however, so far limited with respect to meeting ‘www’ criteria but follows the work of Howard Eichenbaum in using digging to secure reward (Fortin et al., 2004). This is more ‘event-like’ than merely inspecting an object, and the digging technique has been successfully shown to distinguish recollection and familiarity aspects of memory retrieval (Eichenbaum et al., 2007). We here meet a ‘context-location-event’ criterion. Another issue concerns the use of a single-event versus multiple events. More recent work from the Crystal laboratory zeroes in elegantly addressing this issue and shows that multiple single events can be successfully encoded with respect to the context in which they occur (Panoz-Brown et al., 2016), and later recalled using a protocol that successfully distinguished between familiarity-based recognition of novelty from episodic-recall of item-in-context (Easton et al., 2009). The use of multiple single events as part of the daily “episode” in Phase 2 of our study (either 2 or 6 sample events) might be thought to undermine the status of our recency task in requiring episodic-like memory. We think this is unlikely for three reasons. First, certain tasks such as context fear conditioning are claimed to be “episodic” on the grounds that conditioning takes place in a single 3 min trial. In our view, this claim is a misunderstanding as Pavlovian conditioning can occur in a single trial and tests of this form of conditioning do not include any measures showing that the animals recall the conditioning experience; they merely freeze. Thus, having only a single trial is not a sufficient condition for episodic memory. Second, episodes often comprise multiple different events (e.g., remembering a children’s birthday party), but what makes the memory of such an event episodic is that the person has the experience of remembering it having happened before, in a particular context and consisting of a number of events (games, cake, candles, secret wishes etc.). Third, as emphasised by Crystal (2021), the encoding of an experience and the later recall of an episodic memory is not the expression of a learned behavior, but an experience of recalling something from the past. Our task meets this criterion because, on a choice trial in Phase 2, the animal is placed in either Context A or Context B and must first detect in which context it is located and then recall the cryptic spatial location where the event of digging for food has happened most recently. Such a situation might, in later experiments, allow for the possibility of an “unexpected test” (Zhou et al., 2012). The closest we got to this were our two unexpected NE choice tests in which the sample trials were not performed. Specifically, in our non-episodic recall task (Phase 1), the animal’s performance was excellent and unchanged, whereas performance fell to chance in the episodic task (Phase 2). The use of a ‘recency’ protocol here might, nonetheless, be thought to include a cryptic familiarity component. One might argue that one area of space within each context becomes slightly more familiar during the 6 sample trials. In this case, upon exiting the start box, the animals in Phase 2 would not necessarily have to remember back to the prior events of those sample trials, but merely approach what feels like the most familiar part of the arena. This explanation is, however, unlikely for several reasons. First, the context of learning is completely familiar across more than 40 sessions of exposure, a situation that applies to both the extra-arena and intra-arena cues. Familiarity is usually studied using a novel object that is seen only once and this experience contrasted with a second trial in which the object is presented again together with a new ‘never-seen-before’ object. Rodents preferentially explore the novel object, becoming familiar with earlier novel objects in 1 or 2 exposures. Third, familiarity is a term that is usually applied to a specific stimulus, such as an object that can be approached; however, a location within a well learned, indeed, overtrained context representation cannot be considered a specific stimulus. There are no local cues and the correct sandwell is identical in visual appearance and olfactory cues to an incorrect one. The animals might, however, have adjusted the relative familiarity of a nearby local cue such as one of the two intra-arena cues. This is also unlikely as the intra-arena cues would have been seen as many as 92 times in each context (by the end of Phase 1). However, to test this, we measured and found no difference in the performance index when we compared trials in which local cues were adjacent to the correct sandwell (locations 3 and 4) to those in which they were more distant (locations 1, 2, 5 and 6; the apparatus was 1.6 m × 1.6 m; Fig. 3C-right). Moreover, familiarity is thought to decline with time, and our data shows (Fig. 3C-middle) that absolute familiarity did not mediate performance. We therefore argue that an episodic-like solution is the more reasonable interpretation: when the start door opens on a choice or probe trial, the animal must first judge his allocentric location at the edge of the context in front of him and then remember where he dug for food approximately 1.5–3 h earlier in that context. The human analogy would be to recognise that, in going from the kitchen to search for one’s glasses in the living room, it is not as if that location in the living room feels more familiar at the point of entering the room; one simply recalls an earlier event sitting on the sofa or at a table and searches for the glasses appropriately. Interestingly, the concept of “headed records” which we alluded to in the Introduction raises the interesting possibility that a record at the top of the pile may decay very little over time (Morton et al., 1985). If you haven’t seen a good friend for several years, and then meet up, you likely remember the place and circumstances of your last meeting however many days, weeks or years it was earlier. It would be interesting to test whether our context-specific spatial recall could successfully last a week, a month or longer. The protocols described here are only a first step along the path to meeting the fuller ‘re-experiencing’ criteria of ‘episodic-like’ memory, with a distinctive feature so far being our ability to probe either non-episodic-like or episodic-like memory with no change of behavioral procedure on critical choice trials. Moreover, unlike spontaneous exploration studies of recognition memory, the control of inter-animal variability is arguably much better. In Phase 2 of this study, no less than Phase 1, every animal performed at an above chance level. There are several possible next steps. From a strictly behavioral perspective, it would as noted be valuable to know how long the stable and recency memory traces last. The prediction is that the latter might fade – but this time could still be quite long. A lesion study might reveal that both tasks are hippocampus-dependent, but an intriguing alternative would be if both required the hippocampus to be intact for learning, but only the recency task required an intact hippocampus for recall. There is the interesting possibility that so-called ‘hippocampal-dependent’ tasks are episodic-like in character during the initial stages of learning, but the temporal attributes become stripped from the memory traces as stability develops permitting systems consolidation and the mediation of recall by the neocortex without recollection. If the neocortex cannot ordinarily learn quickly, as the complementary learning systems idea supposes (McClelland et al., 1995), it is possible that the integrity of the hippocampus would not be required after consolidation unless the animal is seeking to update its memory or keep track of recency. A pharmacological study would open the opportunity to look at drugs that might affect memory encoding but not recall (e.g., intrahippocampal infusion of an NMDA receptor antagonist; (Morris, 1989)) and conversely, a drug that would limit recall even if only given just before the choice trial (e.g., an AMPA receptor antagonist (Bast et al., 2005)). Investigation of other brain areas would also be valuable (e.g., medial prefrontal cortex, retrosplenial cortex, and thalamus). For now, we seek to establish the viability of examining the context specificity of spatial memory in the domain of recall using two different protocols distinct from those used to study recognition memory. Materials and Methods Animals Phase 1 of the study was conducted in two “replications” to provide an internal replication of stable context-specific spatial memory. There were 14 male Lister-hooded rats (Charles River, UK), aged 10–18 weeks and weighing 250–500 g (n = 8, cohort 1; n = 6, cohort 2). Phase 2 was conducted using the animals of cohort 2. The animals were maintained on a 12/12 h light-dark cycle, with behavioral training carried out in the light Phase at a controlled light intensity (110–120 lx) and temperature (20 ± 1 °C). During the training, the animals had ad libitum access to water, but access to food was restricted to maintain 85–90% of free-feeding body weight in reference to a normal growth curve. The study was conducted under a Project Licence to RM (P7AA53C3F) under the UK Animals (Scientific Procedures) Act 1986 and in compliance with University of Edinburgh regulations. Apparatus Event arena Training and behavioral tests were conducted in an “event arena” (an arena where events happen), consisting of a 160 cm2 open field with a floor comprised of a 7 × 7 grid of white removable tiles (20 cm2; Fig. 1B). Three black Plexiglas start-boxes (25 cm3), each with an automated door that was operable via an automated air pressure release system, were positioned centrally on three walls of the arena: located at East (E), South (S), and West (W; shaded orange and white in Fig1A). The goalbox located at N, which differed from the other start-boxes by having a closed ceiling (shaded blue), was used as a stable home-base to promote allocentric navigation (Broadbent et al., 2020). The event arena could be configured as two distinct contexts (A and B), distinguishable by unique intra- and extra-arena cues (Fig. 1B). Context A consisted of a rabbit statue and lighthouse as the intra-arena cues; and 3D hanging objects, a grey blind, and a coloured photograph as extra-arena cues. Further, the arena walls (30 cm tall) were transparent in context A. Context B consisted of multi-coloured tennis balls and a fish statue as the intra-arena cues; occluding the extra-arena cues of context A, five grey blinds were drawn and used as the extra-arena cues in context B, as well as a blue cylindrical object in the N-W corner. Additionally, the grey blind drawn in context A was raised, leaving a white wall as an extra-arena cue in context B (Fig. 1B). There were black Plexiglas walls (30 cm tall) in context B. Sandwells Unflavored food rewards (0.5 g pellets; BioServ, UK) available as reward in the trials were buried in sandwells (described in detail in Broadbent et al., 2020) which could be inserted into one of six positions in the event arena. The sandwells had one accessible and one inaccessible compartment, with holes between the two allowing the passage of sand and odours. Irrespective of whether a sandwell was rewarded or non-rewarded during a trial, 16 pellets (16 in the inaccessible compartment, unrewarded; 12 in the inaccessible compartment and 4 in the accessible compartment, rewarded) were placed in the sandwells to radically reduce the likelihood of the animals using olfaction to guide performance. Sandwells were filled with sand and garam masala (4 kg: 6.5 g) to mask the scent of the pellets. Experimental protocol The timeline over several weeks consisted of habituation, Phase 1 training (stable context-specific location task); Phase 2 training (context-specific spatial recency memory), and control sessions. Habituation After an acclimatisation period of 7 d, habituation sessions were conducted to familiarise the rats with the event arena, the presence of intra- and extra-arena cues, start-boxes, and specifically the sandwells in which they were taught to dig for food pellets. On habituation day 1 (HD1), the animals were given a food pellet in one of the start-boxes before exploring the arena for 5 min. On HDs 2–4, after receiving a pellet in a start-box, the rats were guided to locate a sandwell in which they dug for a pellet. After retrieving the pellet, the rats were guided to N to consume the food. On successive habituation days (HDs 5–8), the animals could retrieve two consecutive pellets from the sandwell, consuming both in N. The position of the sandwell changed daily and pellets were buried deeper to encourage digging behavior. All animals completed habituation successfully. Phase 1: Context-specific location memory Training sessions in the Phase 1 protocol began with 2 sample trials per context (as shown in Fig. 1A). A rewarded sandwell was placed at one of the 6 potential positions. Animals were placed in a start-box (E, S or W) with one 0.5 g pellet to eat as a cue signalling food availability in the arena. The door of the start-box was opened after 40 s allowing the rats to enter the arena. Once they had dug in the single available sandwell and retrieved a pellet (which Whishaw et al., (1995) have shown is spontaneously carried to a place of safety if above a certain size), the home-base door at north (N) was opened to permit entry with the reward pellet to be consumed there, the entrance door being closed behind the animal. After 30 s, the door of the home-base was re-opened, and the animals left to retrieve a second pellet from the same sample sandwell, confirming the encoding location, and again returned to N. This process of retrieving a pellet, entering N, re-entering the arena for a second pellet, and re-entering N was universal across the different types of trials. A 2nd sample trial starting from a different start-box (e.g., E or W if the first was from S) was then conducted in the same context. After all animals had completed both sample trials in one context, the contextual configuration of the event arena was changed from A to B (or vice versa). The entire sequence of sample trials was then repeated in the other context. In this Phase 1 protocol, each animal was trained to retrieve food from only one location per context throughout their training (e.g., location 1 in context A and location 6 in context B). The animals were trained in groups such that all animals completed their sample trials in one context before being trained in the other, creating an interval of approximately 90 min between their first 2 sample trials in one context, and then their second 2 sample trials in the second context. Choice trials then commenced following the daily completion of the sample trials in sessions 2–11 and S13–22. During the daily choice trials, six sandwells (one rewarded and five non-rewarded) were placed in the arena (positions 1–6). The rewarded sandwell location corresponded with the sandwell location trained in the sample trials, specific to each context and counterbalanced with respect to locations across animals. The order of the contexts used during the choice trials was counterbalanced across sessions. Performance measures Each rat completed one choice trial per context. The time taken for the animal to dig in the correct sandwell (latency, s) and the number of errors were recorded for the retrieval of both pellets. In contrast, during probe trials (sessions 1, 12, and 23), six non-rewarded sandwells were placed in the arena. After consuming the cued pellet in the start-box, the animals were released from the start-box and the times spent digging in the correct and incorrect sandwell(s) were recorded during the first 120 s following the first dig (MultiTimer and LabView software). After 120 s, two pellets were placed in the correct sandwell for the animals to retrieve, helping to avoid any ‘extinction’ of the location of the correct sandwell. As in regular training, a session with probe trials included two sample trials in each context, followed by one probe trial per context. Phase 2: Context-specific recency The daily training protocol was identical to Phase 1 with two exceptions: (a) the location of the sandwell in the sample trial in each context changed every session, and thus the correct sandwell in the daily choice trial also changed each day; (b) following pilot work on the protocol, we switched from 3 sample trials/session/context to 6 sample trials/session/context (Fig. 1C). The animals’ task now was to find and remember the location of the sample trial sandwell and then recall its location during the daily probe test. Training sessions continued across a further 26 sessions. This protocol is procedurally every similar to Phase 1, but the task is episodic in character. Control sessions These sessions were non-encoding (NE) sessions in which the sample trials were not scheduled prior to the choice trials. They provided an internal check on our protocols. For Phase 1, the NE session served as a test of long-term context-specific location memory, as the rewarded sandwells were maintained at the same location across sessions. It was predicted that the animals would choose correctly during Phase 1. In contrast, the NE session for the recency protocol served as a test of whether the animals were inappropriately using another method other than recency memory to complete the task. One possibility would be the use of any olfactory cues from the rewarded sandwell, despite our efforts through sandwell design and masking to prevent this from happening. It was predicted that performance would be at chance for such a test during Phase 2. Acknowledgements This work was supported by an Advanced Investigator Award from the Wellcome Trust (Grant 207481/Z/17/Z) and conducted according to the requirements of the UK Animals (Scientific Procedures) Act of 1986 (Project Licence P7AA53C3F). We are indebted to Patrick Spooner for assistance in designing and building the apparatus and video-monitoring software. For the purpose of open-access, the corresponding author has applied a CC BY public copyright licence to the Authors Accepted Manuscript arising from this submission. Highlights Rats can learn context-specific allocentric representations of spatial memory Rats can recall stable context-specific spatial locations of an event Rodents can also show context-specific recall of recent varying event locations Identical choice trials can distinguish between long-term and recency memory Figure 1 The apparatus and protocol concept. A) A set of 2 sample trials are scheduled each for each of two distinctive contexts followed by 2 choice trials, one in each context (and conducted in a counterbalanced order). As shown, the rewarded sample sandwell for context A (white) is in the north-east corner, and that for context B (grey) is in a south-east location. The startbox for each trial is shaded orange and the animal carries the food it has dug up to the home-base goal-box (blue). During each sample and choice trial, the animal retrieves one 0.5 g pellet from the rewarded sandwell (continuous line), which is of a size that the animal spontaneously chooses to carry back to the safety of the home-base (blue) where it is consumed. The rat is then given the opportunity to retrieve a second pellet, which is also brought back to the home-base (dotted line). This repetition serves only to confirm the encoding of where the food is to be found in each context. B) Photos and layout of the two contexts. Note distinct extra- and intra-maze cues whose identity and location remains stable for all 49 sessions. C) The second Phase of training involved a similar training protocol of sample trials followed by a choice trial, with the key difference being that the location of the rewarded sandwell (sample and choice) changed across successive sessions (N, N+1, etc.). For the second Phase of training, each session comprised of either 2 or 6 sample trials per context, each trial allowing the animal to retrieve one pellet from the rewarded sandwell, followed by the confirmatory run from the home-base to aid encoding (as described for A). Figure 2 Context specificity with a stable sandwell location across sessions. A) Acquisition of the performance index for the choice trials for which 50% is chance performance and 100% represents no errors. A significant difference to chance developed early (by session 2) but the level of significance plotted only for the second half of training. Note steady gradual improvement of context-specificity. The non-encoding (NE) control session (no sample trials) is plotted with orange shading. B) There may have been a slight trend for context B to have been learned slightly faster than context A, but none of the quantitative statistics comparing the two contexts were significant. The significance level of performance during the last 10 sessions is shown for contexts A and B. Both are highly significantly above chance. C) Preferential digging at the contextually-specific correct sandwell rose from chance (PT1) to highly significant values in PT3. At no point was digging at the sandwell that would be correct in the other context ever above chance. Digging at the sandwells that were never rewarded gradually declined across sessions. * p < 0.05, ** p < 0.01, *** p < 0.001. Mean ± SEM. Figure 3 Context specificity of spatial recency recall across sessions. A) The now task-experienced animals were, on average, above chance in choosing the contextually-specific recent location of the sandwell as measured using the performance index for both the 2- and 6-sample protocol. The full data set is plotted across sessions for the 6-sample protocol. As in Phase 1, levels of significance are plotted for the second half of Phase 2 training. The non-encoding (NE) control session (no sample trials) is plotted with orange shading. B) No difference was observed in performance between contexts A and B. C) The full data was averaged across all sessions showing a performance index of circa 70% with minimal variability in both contexts (left). When performance in choice trials was plotted as a function of whether the animals were tested in the most recent sample-trials context or the least recent, no significant difference was observed (middle). When these same data were plotted as a function of sandwell proximity to the intra-arena cues, no significant difference was observed for near vs remote. D) In the probe trials, the percent time digging in the contextually-specific location of the earlier sample trials was well above chance. Again, percentage time digging in the correct location of the other context was at chance. * p < 0.05, ** p < 0.01, *** p < 0.001. Mean ± SEM. Contributions Alex Prodan: Methodology, Investigation, Contributed to writing and figures for Cohorts 1 and 2; Hannah Davies: Methodology, Investigation, Contributed to writing and figures for Cohorts 1 and 2; Hanna Eneqvist and Giulia Mastroberadino: Methodology, Investigation for Cohort 1; Harin Wijayathunga and Kyle Wardlaw: Methodology, Investigation for Cohort 2; Richard Morris: Conceptualization, Methodology, Formal analysis, Supervision, Project administration, Funding acquisition, Writing – review & editing. Aggleton JP Nelson AJD Distributed interactive brain circuits for object-in-place memory: A place for time? Brain Neurosci Adv 2020 4 2398212820933471 10.1177/2398212820933471 32954003 Aggleton JP Pearce JM Neural systems underlying episodic memory: insights from animal research Philosophical Transactions of the Royal Society of London Series B: Biological Sciences 2001 356 1413 1467 1482 Retrieved from http://research.bmn.com/medline/search/results?uid=MDLN.21455695 11571037 Anagnostaras SG Maren S Fanselow MS Temporally graded retrograde amnesia of contextual fear after hippocampal damage in rats: within-subjects examination J Neurosci 1999 19 3 1106 1114 Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=9920672 9920672 Asiminas A Jackson AD Louros SR Till SM Spano T Dando O Kind PC Sustained correction of associative learning deficits after brief, early treatment in a rat model of Fragile X Syndrome Sci Transl Med 2019 11 494 10.1126/scitranslmed.aao0498 Babb SJ Crystal JD Episodic-like memory in the rat Curr Biol 2006 16 13 1317 1321 10.1016/j.cub.2006.05.025 16824919 Bannerman DM Good MA Butcher SP Ramsay MF Morris RG Distinct components of spatial learning revealed by prior training and NMDA receptor blockade Nature 1995 378 6553 182 186 7477320 Bast T Da Silva BM Morris RGM Distinct contributions of hippocampal NMDA and AMPA receptors to encoding and retrieval of one-trial place memory J Neurosci 2005 25 25 5845 5856 Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=15976073 15976073 Broadbent N Lumeij LB Corcoles M Ayres AI Bin Ibrahim MZ Masatsugu B Tse D A stable home-base promotes allocentric memory representations of episodic-like everyday spatial memory Eur J Neurosci 2020 51 7 1539 1558 10.1111/ejn.14681 31944427 Clayton NS Dickinson A Episodic-like memory during cache recovery by scrub jays Nature 1998 395 6699 272 274 Retrieved from http://journals.bmn.com/medline/search/record?uid=MDLN.98421673 9751053 Crystal JD Evaluating evidence from animal models of episodic memory J Exp Psychol Anim Learn Cogn 2021 47 3 337 356 10.1037/xan0000294 34618532 Crystal JD Smith AE Binding of episodic memories in the rat Curr Biol 2014 24 24 2957 2961 10.1016/j.cub.2014.10.074 25466681 Dix SL Aggleton JP Extending the spontaneous preference test of recognition: evidence of object-location and object-context recognition Behav Brain Res 1999 99 2 191 200 10.1016/s0166-4328(98)00079-5 10512585 Eacott MJ Easton A Zinkivskay A Recollection in an episodic-like memory task in the rat Learn Mem 2005 12 3 221 223 Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=15897259 15897259 Easton A Zinkivskay A Eacott MJ Recollection is impaired, but familiarity remains intact in rats with lesions of the fornix Hippocampus 2009 19 9 837 843 10.1002/hipo.20567 19235228 Eichenbaum H Yonelinas AP Ranganath C The medial temporal lobe and recognition memory Annu Rev Neurosci 2007 30 123 152 10.1146/annurev.neuro.30.051606.094328 17417939 Ennaceur A Aggleton JP Spontaneous recognition of object configurations in rats: effects of fornix lesions Exp Brain Res 1994 100 1 85 92 Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=7813656 7813656 Ennaceur A de Souza Silva M Handbook of Object Novelty Recognition; Handbook of Behavioral Neuroscience Blackwells (Wiley) 2018 26 Fortin NJ Wright SP Eichenbaum H Recollection-like memory retrieval in rats is dependent on the hippocampus Nature 2004 431 7005 188 191 Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=15356631 15356631 Good M Honey RC Conditioning and Contextual Retrieval in Hippocampal Rats Behavioral Neuroscience 1991 105 499 509 1930720 Josselyn SA Tonegawa S Memory engrams: Recalling the past and imagining the future Science 2020 367 6473 10.1126/science.aaw4325 Kim JJ DeCola JP Landeira-Fernandez J Fanselow MS N-methyl-D-aspartate receptor antagonist APV blocks acquisition but not expression of fear conditioning Behavioral Neuroscience 1991 105 1 126 133 10.1037//0735-7044.105.1.126 1673846 Langston RF Wood ER Associative recognition and the hippocampus: differential effects of hippocampal lesions on object-place, object-context and object-place-context memory Hippocampus 2010 20 10 1139 1153 10.1002/hipo.20714 19847786 McClelland JL McNaughton BL O’Reilly RC Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory Psychol Rev 1995 102 3 419 457 Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=7624455 7624455 Morris RGM Synaptic plasticity and learning: selective impairment of learning rats and blockade of long-term potentiation in vivo by the N-methyl-D-aspartate receptor antagonist AP5 Journal of Neuroscience 1989 9 9 3040 3057 Retrieved from http://www.jneurosci.org/cgi/content/abstract/9/9/3040 2552039 Morris RGM Garrud P Rawlins JN O’Keefe J Place navigation impaired in rats with hippocampal lesions Nature 1982 297 5868 681 683 Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=7088155 7088155 Morton J Hammersley RH Bekerian DA Headed Records: A model for memory and its failures Cognition 1985 20 1 23 4017518 Nadel L Maurer AP Recalling Lashley and reconsolidating Hebb Hippocampus 2020 30 8 776 793 10.1002/hipo.23027 30216593 Panoz-Brown D Corbin HE Dalecki SJ Gentry M Brotheridge S Sluka CM Crystal JD Rats Remember Items in Context Using Episodic Memory Curr Biol 2016 26 20 2821 2826 10.1016/j.cub.2016.08.023 27693137 Roberts WA Feeney MC Macpherson K Petter M McMillan N Musolino E Episodic-like memory in rats: is it based on when or how long ago? Science 2008 320 5872 113 115 10.1126/science.1152709 18388296 Steele RJ Morris RGM Delay-dependent impairment of a matching-to-place task with chronic and intrahippocampal infusion of the NMDA-antagonist D-AP5 Hippocampus 1999 9 2 118 136 Retrieved from http://journals.bmn.com/medline/search/results?uid=MDLN.99243360 10226773 Wang SH Redondo RL Morris RG Relevance of synaptic tagging and capture to the persistence of long-term potentiation and everyday spatial memory Proc Natl Acad Sci U S A 2010 107 45 19537 19542 10.1073/pnas.1008638107 20962282 Whishaw IQ Coles BL Bellerive CH Food carrying: a new method for naturalistic studies of spontaneous and forced alternation J Neurosci Methods 1995 61 1-2 139 143 10.1016/0165-0270(95)00035-s 8618411 Zhou W Crystal JD Evidence for remembering when events occurred in a rodent model of episodic memory Proc Natl Acad Sci U S A 2009 106 23 9525 9529 10.1073/pnas.0904360106 19458264 Zhou W Hohmann AG Crystal JD Rats answer an unexpected question after incidental encoding Curr Biol 2012 22 12 1149 1153 10.1016/j.cub.2012.04.040 22633809

PMC007xxxxxx/PMC7614819.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8918110 Eur J Neurosci Eur J Neurosci The European journal of neuroscience 0953-816X 1460-9568 34672048 7614819 10.1111/ejn.15500 EMS177466 Article Experiential modulation of social dominance in a SYNGAP1 rat model of ASD Harris E. 1* Myers H. 1* Saxena K. 12* Mitchell-Heggs R. 1 Kind P. 12 Chattarji S 23 Morris R.G.M. 12 1 Edinburgh Neuroscience, Centre for Discovery Brain Sciences, 1 George Square, The University of Edinburgh, Edinburgh, EH8 9JZ, U.K. 2 Simons Initiative for the Developing Brain, The University of Edinburgh, Edinburgh, EH8 9XD, U.K. 3 Centre for Brain Development and Repair, National Centre for Biological Sciences and Institute for Stem Cell Science & Regenerative Medicine, Bangalore 560065, India Corresponding author: R.G.M. Morris (r.g.m.morris@ed.ac.uk) * Shared first authors (Harris, Myers and Saxena) 01 11 2021 02 11 2021 05 7 2023 26 7 2023 54 10 77337748 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Advances in the understanding of developmental brain disorders such as autism spectrum disorders (ASD) is being achieved through human neurogenetics such as, for example, identifying de novo mutations in SYNGAP1 as one relatively common cause of ASD. A recently developed rat line lacking the calcium/lipid binding (C2) and GTPase activation protein (GAP) domain may further help uncover the neurobiological basis of deficits in children with ASD. This study focused on social dominance in the tube test using Syngap+/Δ-GAP (rats heterozygous for the C2/GAP domain deletion) as alterations in social behaviour are a key facet of the human phenotype. Male animals of this line living together formed a stable intra-cage hierarchy, but they were submissive when living with WT cage-mates thereby modelling the social withdrawal seen in ASD. The study includes a detailed analysis of specific behaviours expressed in social interactions by WT and mutant animals, including the observation that when the Syngap+/Δ-GAP mutants which had been living together had separate dominance encounters with WT animals from other cages, the two higher ranking Syngap+/Δ-GAP rats remained dominant whereas the two lower ranking mutants were still submissive. While only observed in a small subset of animals, these findings support earlier observations with a rat model of Fragile-X indicating that their experience of winning or losing dominance encounters has a lasting influence on subsequent encounters with others. Our results highlight and model that even with single-gene mutations, dominance phenotypes reflect an interaction between genotypic and environmental factors. Social neuroscience autism spectrum disorders social dominance tube-test cognitive compensation pmcIntroduction According to the World Health Organisation, as of 2019, 1 in 160 children worldwide develop autism spectrum disorder (ASD) (https://www.who.int/news-room/fact-sheets/detail/autism-spectrum-disorders). This neurodevelopmental disorder is characterised by early onset of impairments in social interaction and communication, limited interest in others, and the presence of repetitive or stereotypical behaviours (Parishak et al, 2013; Yoo, 2015). Multiple studies have identified de novo mutations in the Synaptic Ras GTPase-activating protein 1 (Syngap1) gene as a risk factor for ASD (Hamdan et al. 2011; O’Roak et al. 2014; Berryer et al. 2012). The SYNGAP1 gene codes for a postsynaptic density protein primarily expressed in excitatory neurons (Walkup et al, 2016). The SYNGAP protein interacts with N-methyl-D-aspartate (NMDA) receptors and negatively regulates both Ras and Rap GTPase. Ras signalling activates the MAPK/ERK cascade, important for the induction and maintenance of long-term potentiation (LTP) via insertion of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors into the postsynaptic membrane. The Rap pathway mediates long-term depression (LTD) via p38MAPK. Hence, SYNGAP is expected to play an essential role in normal synaptic function and plasticity (Komiyama et al. 2002; Kim et al. 2003). Heterozygous Syngap+/Δ-GAP mice display strong excitatory/inhibitory imbalance in hippocampal and forebrain neural networks (Clement et al. 2012; Ozkan et al. 2014), an imbalance that has been proposed to be a contributor to the social deficits seen in ASD and other neuropsychiatric disorders (Yizhar et al. 2011). These neurobiological disruptions are associated with behavioural and cognitive phenotypes in murine models that mimic symptoms of the human condition such as elevated locomotor activity (hyperexcitability), impaired working and spatial memory, and a decreased sensitivity to painful stimuli (Guo et al. 2009; Muhia et al. 2010; Clement et al. 2012; Nakajima et al. 2019). The implicit supposition behind this animal work is that a single-gene mutation would primarily have a genetically deterministic effect on phenotype. Few studies have addressed the altered sociability of Syngap mutations in rodents. A study looking at schizophrenic-like symptoms in mutant mice found Syngap+/- animals had reduced social memory (Guo et al, 2009). Specifically, while they gave similar results to wildtype (WT) mice in the three-chambered sociability test, they failed to distinguish between a familiar and novel conspecific mouse. Furthermore, in the laboratory, Syngap mutants are reported to spend more time alone than interacting with other animals (novel or familiar), and less social interaction in both novel and home environments compared to WT animals (Guo et al. 2009; Nakajima et al. 2019). As ASD is primarily a social communication disorder, further characterisation of social interaction in animals modelling SYNGAP mutations would be valuable. This new study is conducted using rats rather than mice, that are central to a wide-ranging Simons funded programme of research at Edinburgh. The primary aim was to establish whether Syngap+/Δ-GAP rats are socially submissive and display a distinct profile of specific behaviours in social dominance interactions than WTs. Establishing a dominance hierarchy requires recognising social cues. Once established, a hierarchy usefully determines access to resources while minimising the need for aggressive conflict (Cummins, 2000; Fan et al. 2019). We used the dominance tube-test which is a behavioural assay of social dominance (Wang et al. 2014). It has been widely used as an assay of social hierarchy, such as to examine the relative dominance of different mouse strains (Kunkel & Wang, 2018), the neural circuits underlying dominance behaviour (Wang et al. 2011; Zhou et al. 2017), neuropsychiatric disorders such as major depression (Yang et al. 2014) and ASD-like syndromes (Huang et al. 2018; Saxena et al. 2018). With analogous results to other dominance tests, with which rankings correlate, it has been suggested as an accurate measure of social dominance (Wang et al. 2011). Mouse models have been used predominantly, but even they have been shown to reveal similar weanling dominance patterns to those of children (Chou et al, 2021). A secondary facet of the study concerns the opportunity in social dominance interactions to learn about and remember the other animal/person - a dimension of social experience. One might then expect deficits arising from failures of social memory (e.g. about the identity or social status of another animal). However, that very deficit may indirectly foster the development of motor habits in social interactions that are less demanding on day-to-day memory but reflect learned patterns of dominance or submissive behaviour. Such habits are likely more inflexible such that, once learned, they would be expressed repetitively even in inappropriate situations. Accordingly, after the initial phase of testing interactions within each cage, we examined contests between animals from different cages that had by then assumed a particular intra-cage rank. Rats are a species that is inherently social which has evolved a complex social repertoire (Lore and Flannelly 1977), and the outbred nature of the rats we used more closely mimics the genetic variation seen in humans. They may, therefore, be an appropriate model for the study of human psychiatric disorders characterised by deficits in social cognition (Ellenbroek & Youn, 2016). Saxena et al. (2018) studied FMR1 knockout (Fmr1-/y) rats as a model for Fragile-X Syndrome (FXS), another monogenic cause of autism that shares many of the behavioural deficits seen in Syngap+/- animals (Spencer et al. 2005; Kazdoba et al. 2014; Ding et al. 2014; McNaughton et al. 2008). As expected, male knockout rats were submissive to WT animals in mixed-line groups living together, a finding that makes sense in terms of model “validity”. However, a small number of high- and low-ranking FXS mutants who had lived together and formed an intra-cage hierarchy went on to display the same phenotype in inter-cage contests. Specifically, they won (or lost) social dominance contests against stranger animals regardless of stranger rank. This study explored social dominance in Syngap+/Δ-GAP mutants using the same experimental protocols as in our earlier study of FXS mutants (Saxena et al. 2018). Materials and Methods Subjects Adult (>12 weeks, n=16) male Long-Evans hooded rats were used, weighing 500g to 600g. The rats were cohoused in groups of 4 per cage, from weaning, with ad libitum food/water and a 12h light/dark cycle. The cages contained a 25 cm long section of Perspex tube similar to the one used in the actual tube-test to allow the animals to become used to being inside such a tube. The colony founders were produced by Sigma Advanced Genetic Engineering (SAGE) labs (St. Louis, MO, US) using the zinc finger nuclease mediated deletion (Gurts et al. 2009) of the GAP domain of Syngap. Later rats for the experiments were bred in-house, the Syngap+/Δ-GAP rats were generated by mating female Syngap heterozygous rats with male WT Long-Evans hooded rats acquired from Charles River Labs, hereafter called ‘Het’. The WT animals used as controls were littermates. There was one WT single-line cage (n=4), one Het cage (n=4), and two mixed cages (nWT=2, nHet=2 per cage; overall n=8). All experiments were done blind to genotype, with animals being given a cage number on their tail and a coloured spot on their fur (using animal paint, red, green, blue, purple) to identify them, the colouring is random. The code was retained by someone independent of the study and only broken when all procedures had been completed. The genotype of the animals was confirmed externally before and after finishing the experiments by a company (Transnetyx). Ethics and Legal Statement The studies conducted were all behavioural and did not involve surgery or the administration of drugs. We monitored the animals carefully at all stages of handling and experimentation, particularly in Phase 2 (below) when animals from different cages were tested together. The study was conducted according to the regulations of the Animals (Scientific Procedures) Act 1986, a U.K. Project Licence held by RGMM (I49398628), and under the supervision of the named veterinary surgeons of the University of Edinburgh. Apparatus The tube-test assay was the same as used in Saxena et al. (2018). A transparent perspex tube, 1 m long and 7 cm diameter, served to connect two holding boxes (Fig. 1A). In each box, bedding was placed from the home cage of the animals to help reduce anxiety (Fig. 1B). The tube was large enough for the rats to move freely, but not to cross past each other or turn around. A removable metal grid was placed in the middle of the tube, this being lifted to start an encounter that ended when one animal retreated (Fig. 1C). A camera provided a direct view of the tube to record the trials using OBS recording software (https://obsproject.com/). The entire apparatus was connected to custom-made-Arduino-based hardware, and we used its serial reader functionality for reading the button-presses denoting the start/stop times of the trials. Behaviour Protocols The training protocol consisted of Habituation, Phase 1 Contests between animals within a cage (intra-cage), and a Phase 2 Competition involving all animals in one cage against all the other cages (Fig. 1D; inter-cage). The experiments were performed between 9 am to 4 PM. There were a total of 2640 competitions between individual animals. Habituation Each animal was handled for 3 days and allowed to run freely in the apparatus (alone) for at least 10 min per day. See details of procedure in Saxena et al (2018). Phase 1 Intra-Cage Contests The basic tube-test consisted of two rats being placed in the holding boxes, one on each side. The rats then entered the tube and met in the middle with the metal grid present and acting as a barrier. The trial started when the barrier was removed. During the trial, the rats competed for dominance during which a variety of behaviours were observed. Typically, the animals were together, their heads side-by-side and relatively still for a few sec. Thereafter, either one rat pushed the subordinate to retreat out of the tube (dominant), or the other rat withdrew of its own accord (subordinate). A trial was defined as ending when one rat backed out into the holding box from which it started. This rat was recorded as the “loser” and the other as the “winner”. Each pair of rats underwent 5 trials each session to obtain a secure measure of dominance (3:2, 4:1 or 5:0), alternating their starting positions between left or right. All cages were tested on the same day, with the 10 sessions of 5 trials on consecutive days. The order in which the cages were trained each day was randomised in counterbalanced order. To observe and measure intra-cage hierarchies, each animal competed against all the other animals in its cage (5 runs per pair, 6 competitions/cage). Cage ranks were determined by adding up the number of wins or losses for each animal across all six encounters (of 5 trials) within a cage. The animal with the most wins was denoted as 1st rank. The following ranks correspond to increasing number of losses with the 4th rank animal being the one losing the most. If two animals had the same number of losses, the relative rank was determined by which animal lost most of that pair. This daily tube-test assay was repeated over 10 sessions. There were, therefore, 4 cages x 6 encounters x 5 runs x 10 sessions (= 1200). Trial latency was taken as the time (sec) to complete one trial. Phase 2 Inter-Cage competitions The procedure to conduct tube-tests was exactly as above, but the animals now competed against animals in the other cages who were, effectively, stranger opponents. Each animal competed against all four animals from the other 3 cages. Each trial between a pair was repeated 5 times and all trials were repeated over 3 sessions. This comes up to a total of 1440 trials (6 cage interactions × 16 encounters × 5 runs × 3 sessions). We analysed only the 1st and the 3rd sessions for the detailed video analysis of behaviour, giving 6 × 16 × 5 × 2 contests (= 960). The sequence of pairs and cages was randomised across the 3 sessions. Video Analysis Recordings from sessions 1 and 10 of Phase 1 and sessions 1 and 3 of Phase 2 were analysed, on a frame-by-frame basis (1/20th sec resolution per frame) using Behavioural Observation Research Interactive Software (BORIS) Friard & Gamba (2016), with these sessions chosen for this detailed analysis as the beginning and end of each phase. Behaviours were logged for each animal as follows: PUSH (included any form of pushing, with paws, nose or body and biting), RESIST (resisting a push from an opponent), MOVE FORWARD, STILLNESS (when no displacement of the body was seen, but including other stationary activities such as sniffing or grooming), RETREAT (backing out of the tube by being pushed) and WITHDRAWAL (voluntarily backing out of the tube). It proved easy to distinguish PUSH and RESIST despite being quite similar in terms of effort. Statistical analysis For all statistical analyses, we chose tests based on the data secured (wins, latency, behaviour occurrences). Values computed include total number (e.g. of wins), from which rank was derived, and means (latency, rank), occurrences of specific behaviours, and measures of variability of rank (stability/variance). Stability was computed by counting the number of times an individual rat changed rank during phase 1 without regard to whether it changed by one position of rank or more, and these data were normalised (100% stable meaning no changes in rank across the 10 sessions, 0% meaning a change every session). Variance was a similar measure and computed as the true variance of the rank scores across all 10 sessions of phase 1. For latency and behaviour occurrences, we computed mean and standard error of the mean. For all parametric tests, the data fitted the assumptions of equal variance and were normally distributed (ANOVA and t-tests). For non-parametric tests, Mann-Whitney, Chi-squared or a Fisher’s exact test were used. The behavioural occurrences analysis for both Phase 1 and 2 required a Two-way repeated measures ANOVA, with the different behaviours treated as a ‘repeated measures’ within subject variable. This was paired with a Sidak’s multiple comparison test. The value computed in each test (e.g. F-ratio), significance levels and degrees of freedom are reported. GraphPad Prism v7 was used for preparing the graphs and statistics, and then displayed using Adobe Illustrator. Results Qualitatively, all animals explored the tube readily during habituation, and walked through it without hesitation during dominance contests. We anticipated that frequent losers might become hesitant to enter or walk through the tube over time but, surprisingly, this did not happen. We observed no signs of overt inter-animal aggression, even in contests across cages, and its absence may have contributed to the success in training of both phases of the study. All encounters were conducted and evaluated “blind” to genotype, and the occupants of each cage was unknown to the experimenters. Although the number of animals in the study was modest (n=16; WT, n=8, Syngap+/Δ-GAP n=8), the number of contests observed was extensive (n=2160). Phase 1 Contests The first question was whether single-line WT animals living together, and Syngap+/Δ-GAP animals likewise (hereafter called ‘Het’), would each form a hierarchy. Ranks of individual animals in single-line cages are shown in Fig. 2A, with both the WTs (as expected) and Hets forming a hierarchy. The 1st ranked Het rat stayed at the top over all 10 sessions, all other ranks stabilising by session 7. The mean rank was calculated from the ranks of each animal over all 10 sessions. A clear hierarchy was seen in both cages (Fig. 2B), there was a statistically significant difference between ranks as determined by one-way ANOVA (WT: F(3,39) = 33.45, p = 0.0001; Het (F(3,39) = 27.89), p = 0.0002). No differences were observed in normalised stability (unpaired t-test, t=0.134, df=6 p=0.898, Fig. 2C) or variance of rank between WT and Het animals over the 10 sessions (Mann-Whitney test, p=0.829). The next step was to examine the hierarchies formed in each of the two mixed-line cages. Both cages showed an overall hierarchy (Fig. 3a). Het rats in mixed cages won less contests than the WT animals (Chi-squared test of independence, X2=68.06, df=1, p<0.0001; Fig.3B). Het animals also had a lower average rank of 3.18 ± 0.15 over 10 sessions compared to WT animals with an average rank of 1.83 ± 0.13 (unpaired t-test, t=6.69, df=78, p<0.0001, Fig. 3B; the highest rank score = 1). Both comparisons were highly significant. Averaged across the two cages, there was no overlap of mean rank between the WT and Het rats (Fig. 3A). With respect to trial-to-trial variability, no significant difference was detected between WT and Het animals on measures of rank stability (unpaired t-test, t=1.32, df=6, p=0.235, Fig. 3C) or variance (unpaired t-test, t=1.12, df=6, p=0.304, Fig. 3C). The pair-wise competitions took time (Fig. 4). Competitions took longer at the beginning of the series of 10 sessions, averaging 19.5 ± 4.84 s from the moment the barrier between the two animals in the tube was raised and the point when one of them fully retreated, but this time declined to less than 10 s by session 3 (8.27 ± 0.74 s; Fig. 4A) and then stabilised. This overall decline was significant (One-way repeated measures ANOVA, F9.207=4.92, p=0.0094, Fig. 4a). The pattern for latencies for the 50% subset of mixed-line cages was much the same (One-way repeated measures ANOVA, F9.99=5.08, p=0.032, Fig. 4b). For the single-line cages, Het animals displayed longer latencies than WT (Two-way repeated measures ANOVA, F1.6=12.3, p=0.0056, Fig. 4c). A key next question was whether the specific patterns of behaviour (e.g. PUSH, RETREAT) displayed during dominance encounters differed as a function of dominance or genotype. The video analysis showed a clear interaction between rank and behaviour occurrences (Two-way repeated measures ANOVA, F5.70=4.54, p=0.0012, Fig. 5a). The trend shows High-Rank animals were more likely to PUSH and MOVE FORWARD compared to the Low-Rank animals who executed more RETREAT and WITHDRAWAL. In certain respects, such a pattern must occur by definition. However, the high occurrences of STILLNESS in High-Rank animals was not expected a priori. One complication in quantifying the “occurrences” of such behaviours is that PUSH behaviour may occur several times during a trial, often met by RESIST behaviour from the other animal. However, RETREAT happens less frequently and may, in the limit, happen only once to resolve a contest. Attempting to “normalise” their relative behaviours to create a numerical “level playing field” by examining occurrences over durations of time in order to calculate a measure of frequency does not help, as pushing can start and stop throughout a trial and thus the full duration of the trial would have to be considered the duration; but equally, retreat can happen at any time, even if only once, and thus the duration of time for calculating frequency is the same. However, the overall higher behaviour occurrences of the single-line Het cage than the single-line WT cage (Fig. 5b,c) is consistent with the latency data of Fig. 4. Moreover, the significant interaction between rank and behaviour (Two-way repeated measures ANOVA, F5.30=4.46, p=0.0037, Fig. 5d) was observed only in the mixed-line cages in which the high-rank animals were, in practice, WT rats. With respect to STILLNESS behaviour, frequently occurring while the two animals are in close contact and during which dominance “decisions” may be being made, single-line Het rats animals had a significantly higher STILLNESS occurrences than single-line WT animals (Sidak’s multiple comparison test, p=0.0131; compare Fig 5c with 5b). This was not observed in the mixed cages (unpaired t-test, t=0.732, df=6, p=0.492; data not shown), for which the primary determinant of STILLNESS was rank (Fig. 5d). Phase 2 Inter-Cage Competitions The second phase of testing was an inter-cage tournament in which the animals underwent tube-test competitions against all animals from the other cages. In contrast to Phase 1 in which WT animals were clearly dominant, the phenotype now reflected some facets of the rank of the animals in Phase 1. Intriguingly, Het animals now won as equally often as WT rats (Fisher’s exact test, p=0.553, Fig. 6a). In fact, the Hets in single-line cages won significantly more competitions against single-line WT animals from different cages (Fisher’s exact test, p=0.0321, Fig. 6b), whereas Mixed-cage Het rats lost more contests against stranger mixed-cage WT animals (Fisher’s exact test, p=0.0152, Figure 6c). These changes are identical to what was observed by Saxena et al (2018) in a rat model of FXS. One possibility is that rank was simply unstable between Phases 1 and 2 and, that there are major reliability issues with the measures being quantified and the experimental approach being taken. However, rank in Phase 1 (regardless of genotype) was predictive of winning/losing in Phase 2 (Fig. 6d). That is, the overall number of wins in Phase 2 by animals that were High-Rank in Phase 1 was significantly higher than those of Phase 1 Low-Rank animals (Fisher’s exact test, p<0.0001). This indicates that previous experience of winning or losing can be predictive of the outcome of competing against a stranger. Logic then required us to distinguish high-rank and low-rank animals as a function of genotype. In a single-line cage, there will be two animals in ranks 1 and 2 which we shall refer as “high rank” and the other two animals (ranks 3 and 4) as “low-rank”; this is also true of the Het single-line cage than of the WT single-line cage. One possibility is that a Het winner may “get used” to winning and a Het loser “get used” to losing - i.e. their social dominance interactions become habits, possibly because of a deficit in social perception or memory. The ten sessions of 5 encounters per session in Phase 1 provided 50 trials between a pair in which to develop habits in the tube-test, such as habits of pushing or retreating etc. Interestingly from the perspective of habit vs. memory, High-Rank Het animals won significantly more competitions against High-Rank WT rats (Fisher’s exact test, p=0.0009, Fig. 6e), whereas Low-Rank Hets performed equivalently to Low-Rank WT rats when competing against them (Fisher’s exact test, p>0.99, Fig. 6F). By the same argument, Het animals living in mixed-cages (generally in ranks 3 and 4) would be expected to be submissive to single-line WTs - which they were (Fisher’s exact test, p=0.0321), having a developed a habit of withdrawal and retreat. These findings in Phase 2 are statistically significant, but modest, as they entail comparisons between subsets of animals with an “n” of only 2, 4 and 6. Nonetheless, they raise the possibility that there may be a difference between winning when you can process social cues effectively and winning when you cannot. Variations in the occurrences of distinct behaviours in the tube may reflect these differing states of affairs. Overall, behaviour and rank interacted significantly to affect the behavioural occurrences (Two-way repeated measures ANOVA, F5.70=4.40, p=0.0015, Fig. 7a), but there was no significant interaction between behaviour and genotype (Two-way repeated measures ANOVA, F5.70=0.39, p=0.854, Fig. 7b). This data suggests that, in the inter-cage contests of Phase 2, previous experience determines future behaviour more than genotype. As the data on combined effect of rank and genotype was revealing with respect to dominance, it might also have affected behaviour occurrences. For trials between high-ranking animals, genotype interacted significantly with behaviour (Two-way repeated measures ANOVA, F5.30=8.81, p<0.0001, Fig. 7c), with genotype accounting for a significant amount of the variance seen in the behavioural occurrences (Two-way repeated measures ANOVA, F1.6=42.57, p=0.006). A trend reveals that high-ranking Het animals had a higher occurrence of PUSH, MOVE FORWARD, STILLNESS and RESIST behaviours compared to High-Rank WT. This indeed complements the findings in Fig. 6e as an increase in these behaviours would explain the increased number of wins. For competitions between low-ranking rats, there was also a significant interaction between genotype and behaviour but in the opposite direction (Two-way repeated measures ANOVA, F5.30=19.0, p<0.0001, Fig. 7d). The overall behavioural occurrence was now relatively higher in WTs (Two-way mixed-model ANOVA, F1.6=78.7, p=0.0001, Figure 7d), and specifically, Low-Rank WT animals had a higher occurrence of PUSH behaviour (mean ± SE = 68.0 ± 1.41) than Low-Rank Het animals (mean ± SE = 21.7 ± 2.26) (Sidak’s multiple comparison test, p<0.0001, Fig. 7d). This would suggest the WT should be dominant over the Het animals but, as shown in Fig. 6f was not the case. The reason may be that Low-Rank WT animals also had a high RETREAT occurrence (Sidak’s multiple comparison test, p=0.0091). Discussion The aim of this study was to examine the generality of the idea that ASD model animals would show consistent changes in social dominance relative to WT animals. Using a social dominance tube-test paradigm, we observed that male Het Syngap+/Δ-GAP animals living together form a stable hierarchy but, when living with WT animals, have a submissive phenotype compared to their WT cage-mates. This models social withdrawal and is analogous to what we observed with FXS mutants (Saxena et al, 2018). Specific behaviours exhibited during the tube-test included expected facets (such as greater PUSH behaviour by dominant animals), but also a striking increase in STILLNESS behaviour by the Het animals housed in the single-line cage and a higher latency to resolve conflicts when two Het animals competed, both suggestive of a social processing deficit. We also found that social dominance experience affects subsequent interactions, interacting in a subset of animals in a surprising way with genotype. Het animals (n=2) that were previously dominant in single-line cages in the intra-cage analyses (Phase 1) of the study were, in the inter-cage (Phase 2) tournaments, also dominant against all animals from other cages including previously dominant WT rats (n=6). Het animals (n=6) that were previously submissive continued to be largely submissive. These observations are subject to the qualification of small “n” (2 and 6 respectively), but they do replicate in another ASD line the paradoxical reversal of phenotype observed in a much larger number of FXS mutants by Saxena et al (2018). They also add to earlier observations of some similarities between Syngap and FXS mutant mice despite various differences (Barnes et al, 2015). We shall argue that these data collectively point to a reduced ability of Syngap+/Δ-GAP animals to process social cues. The patterns we observed resemble behaviours observed in ASD children. More severely impaired individuals show rigid operant learning in which they continue to use a previously learned strategy, even across different contexts (Stanfield, A. pers. comm). Moreover, children with autism are both more likely to be “bullied” by their siblings but also to “bully” them back (Toseeb et al. 2018). Perhaps these behaviours, once they develop, become inflexible habits. The dual phenotype of Syngap+/Δ-GAP behaviour we observed, from the submissive Het animals living in the mixed cages through to the suggestion of dominant single-line Hets winning novel encounters, may model this facet of the apparently opposing behaviours seen in ASD children. In Katsanevaki et al’s (2020) original study of this line of Syngap+/Δ-GAP rats, they observed a failure to extinguish a conditioned fear response despite many sessions of “extinction”. They saw normal sociability as measured by proximity, but the mutants showed a striking decrease in active sniffing of the other animal. Interestingly, in a study of both children and weanling mice, Chou et al (2021) show that (normal) children who are less persistent in games, have low emotional intensity and withdraw from social encounters easily, are more likely to be subordinate. They go on to show, in a demonstration of face-validity, that tube-test contests between mice tend to be resolved most often by loser withdrawal. Collectively, these results highlight and model the current belief that ASD reflects environmental interactions with a genetic predisposition (Chaste & Leboyer, 2012). Thus, depending on their rearing environment and previous social history in competitive interactions, children and animals with an ASD mutation (FMR1, SYNGAP1) may present different behavioural phenotypes. The pattern of behavioural findings Syngap1 mutations in mice lead to severe cognitive impairment, including deficits in social memory (Guo et al. 2009; Komiyama et al. 2002; Muhia et al. 2010; Ozkan et al. 2014). The Syngap+/Δ-GAP animals living in mixed cages were clearly submissive to their WT cage-mates, and thus both WTs were in ranks 1 and 2. This is a similar phenotype to the Fmr1-/y rats of Saxena et al (2018). We also reasoned that Syngap+/Δ-GAP rats would be able to form a hierarchy but that it would be less stable than WT animals, as seen in the Fmr1-/y model of ASD (Saxena et al. 2018). We found, however, that the Het rats were not only able to form a hierarchy, but did so with statistically similar stability and variance of rank to WT. Although there are some exceptions (e.g. the purple rat in the WT single line cage that changed its rank 7 out of 9 times, and the red animal in the Mixed cages 1 and cage 2), some of the Het rats were more stable than the WTs in the mixed cages. These results suggest that Het animals may have enough social memory to distinguish between the ranks of the conspecifics with whom they are living. It is possible that rank stability observed in Hets reflects a lack of motivation, or a tendency to “give up” easily by rats in the lower order of hierarchy. The Syngap+/Δ-GAP animals living in mixed cages being submissive to their WT cage-mates, during encounters against novel conspecifics (the inter-cage assessments in Phase 2), we found that High-Rank Syngap+/Δ-GAP animals won over the High-Rank WTs whereas Low-Rank Syngap+/Δ-GAP animals showed a different pattern. In presenting these results, we noted the qualification that the numbers of animals in these comparisons (2 and 6) were small, but it should be recognised that the number of competitions they undertook was quite large. What is suggestive is that these are very similar results to Saxena et al’s (2018) FXS model, raising the possibility that both Fmr1-/y and Syngap+/Δ-GAP rats are poorer in detecting social cues relating to the dominance status of an opponent. Faced with this cognitive deficit, they may therefore develop habits of repetitive behaviour during Phase 1 that serve them well. Specifically, High-Rank Syngap+/Δ-GAP animals would have won more frequently against Low-Rank animals in their single-line cage (10 sessions of 5 trials per inter-animal encounter (i.e. 100 trials for the 2 animals in each sub-group). This training may have been sufficient to develop habits such as extensive pushing, or resisting against pushes by the other animal, and that once learned, these behaviours would have continued into Phase 2. Such habitual patterns might also have been acquired by dominant WT animals but they could, each time, better appraise their opponent by providing and receiving social cues. Repetitive behaviours have been observed in Syngap+/Δ-GAP mice, with the animals having a higher stereotypic count in an open field test than WT (Guo et al. 2009). In our study, High-Rank Syngap+/Δ-GAP animals PUSHED significantly more than the High-Rank WT suggesting they simply repeated a learned PUSH behaviour, whereas the WT rats adapt flexibly to the changing social environment. An additional point is that Syngap+/Δ-GAP animals in single-line cages showed significantly increased STILLNESS occurrence (Fig 5d), a pattern that carried over to the inter-cage tournaments for the high-rank animals. Alterations in these animals’ ability to detect social cues could be a cause of both the high STILLNESS occurrence and the longer latencies to resolve competitions when confronted by another Syngap+/Δ-GAP animal. We recognise that it is speculative to relate these findings to the work on bullying by Toseeb et al (2018), but worth noting nonetheless. Neurobiological considerations The medial prefrontal cortex (mPFC) and adjacent anterior cingulate gyrus have been implicated in decision-making during dominance encounters (Zhou et al, 2017; Nelson et al, 2020). Decreased connectivity between mPFC and the primary somatosensory cortex has been previously observed in Het animals (Aceti et al. 2015). A separate study noted Syngap+/Δ-GAP mice had lowered excitability in the upper lamina somatosensory neurons which encode touch-information (Michaelson et al. 2018). Syngap+/Δ-GAP animals also have altered volume of cortical areas related to visual system processing (Kilinc et al. 2018). These findings suggest that Syngap mutants may have decreased sensory perception and processing that affects their social communication. Accelerated maturation during development leads to enlarged mushroom shaped spines with clusters of AMPA receptors (Kim et al. 2003; Vazquez, 2004; Aceti et al. 2015; Clement et al. 2012). In addition, they have a more than 50% reduction in the SYNGAP protein known to be important in synaptic transmission (Komiyama et al. 2002, Jeyabalan & Clement, 2016). Interestingly, striking observations have been made in hIPSC derived human neurons in which Syngap1 was deleted using CRISPR/Cas9 technology (Llamosas et al, 2020). Previous studies in normal WT mice have suggested that increased synaptic strength in the mPFC leads to more dominant behaviour and that it may be the “neurobiological foundation for dominance-associated personality traits” (Wang et al. 2011, McMahon et al, 2012, Zhou et al. 2017). We therefore wonder whether Syngap+/Δ-GAP rats have defective synapses in association with their submissive phenotype. It cannot, however, be a simple reduction in strength or efficacy as Syngap+/Δ-GAP adult mice have been found to have higher occurrence and amplitude of miniature excitatory post-synaptic currents (mEPSCs) in layer 2/3 mPFC neuron slices, pointing to an increase in unitary synaptic strength (Ozkan et al. 2014). Were this to be seen also in mPFC, Zhou et al’s (2017) data predicts these animals might even be more dominant. Based on these findings, Syngap+/Δ-GAP animals may sometimes be capable of displaying a more dominant phenotype. One possible explanation for this contradiction is that the submissive Syngap+/Δ-GAP phenotype is caused by reduced LTP at mPFC synapses. Wang et al. (2011) found that transgenic manipulations that increase AMPA receptor trafficking to the post synaptic membrane led to an increase in rank in the tube-test whereas reducing it led to a decrease. Subsequently, Zhou et al. (2017) described the “winner effect” wherein repeated winning in the dominance tube-test caused strengthening of mPFC synapses via LTP and changed the rank of an animal, an effect that could be mimicked by optogenetic activation of thalamic inputs to mPFC. This suggests that an LTP-like process can be important in establishing dominance in normal animals. Multiple studies have found reduced LTP in CA1 hippocampal regions of Syngap+/Δ-GAP rodents, associated with elevated basal levels of Ras signalling in Syngap mutants which prevents further Ras activation upon synaptic stimulation and thereby inhibits LTP (Komiyama et al. 2002; Kim et al. 2003; Ozkan et al. 2014; Kilinc et al. 2018). If these findings from hippocampus synapses generalise to other brain regions, Syngap+/Δ-GAP animals could have reduced LTP in other regions including the mPFC. This deficit could limit their ability to adjust their behaviour flexibly in response to experience and so cause them to fall back on well-learned habits. Limitations and future directions This study builds upon previous research addressing Syngap mutant rodents as models for ASD, but is not without limitations. First, it would be valuable to train additional cohorts of Syngap animals as they become available. Second, in keeping with the 3Rs (reduction, refinement and replacement), we used a small but not inappropriate number of animals in the present study which were all male. Although the full data set was large and many facets of the statistical findings are robust, reproducing these experiments with both genders and multiple cages might allow higher confidence in the findings. They are, nonetheless, in line with the behavioural phenotype of Syngap mutant mice in revealing social withdrawal, repetitive behaviour and hyperactivity. Third, additional dominance tests, using different sensory or motor properties, need to be done to be conclusive about the generality of the dominance phenotype of Syngap+/Δ-GAP rats (Wang et al. 2014; Fan et al. 2019). Such investigations should include examining the development of habits that compensate for the loss of social flexibility. Fourth, we have relied on published data in mice with respect to three-chamber tests of social novelty to claim that Syngap+/Δ-GAP rats may have a deficit in social memory. Clearly this should be tested directly in rats also, but the test should not only include this classic test of social memory (Crawley, 2006) but also some way of assessing whether animals can encode and remember the dominance status of a novel opponent (and not just its identity). Finally, one interesting future test would be to examine whether changes in rank can be induced by artificial induction of winning in subordinate Syngap+/Δ-GAP animals. This could be done by optogenetic activation in the thalamic to prefrontal cortex pathway, as seen in Zhou et al’s (2017) study. Such a study might be conducted in conjunction with a therapeutic intervention, such as by the reintroduction of Syngap protein in a Syngap+/Δ-GAP line of rats, in the manner of classic “genetic rescue” studies of Rett syndrome (Guy et al, 2007). Alternatively, it might be done pharmacologically such as by restoring Ras signalling which is reported to improve cognitive deficits in mouse models associated with ASD (Ogden et al. 2016; Asiminas et al, 2019). Acknowledgements The authors thank Patrick Spooner for building the apparatus and to the Simons Foundation for the Developing Brain (Edinburgh) for funding. Data availability Raw data for all figures (Excel), the details of all of the statistics done along with an example video is uploaded in a data repository (https://datashare.ed.ac.uk/submit?workspaceID=6884). Abbreviation List ASD Autism Spectrum Disorder BLIND when experiments are conducted without the experimenter knowing the genotype or other information about the subjects or protocol that could bias their observations (an aspect of unconscious bias) BORIS Behavioural Observation Research Interactive Software C2 calcium lipid binding protein FMR1 The gene Fragile C mental retardation 1 FXS Fragile X GAP GTPase activation protein. GTPases are a large family of hydrolase enzymes that bind to the nucleotide guanosine triphosphate (GTP) and hydrolyze it to guanosine diphosphate (GDP) Het heterozygous knock-out animal (in this case Syngap+/Δ-GAP) LTP Long term potentiation MAPK/ERK Mitogen-activated protein kinases and extracellular signal-regulated kinases mEPSC - miniature excitatory post-synaptic current mPFC medial prefrontal cortex NMDA N-methyl-D-aspartate Rap A protein named after “Rat sarcoma virus” that belongs to a class of protein called small GTPase which are involved in transmitting signals within cells (cellular signal transduction) Ras also a GTPase which is similar in structure to Rap SAGE Sigma Advanced Genetic Engineering SYNGAP1 a gene that makes a protein called SynGAP which is found at the junctions (aka ‘gaps’) between synapses Syngap+/Δ-GAP (rats heterozygous for the C2/GAP domain deletion) WHO World Health Organisation WT wild-type Figure 1 Tube Test Apparatus and Protocol A, B) The plastic, transparent tube in which dominance interactions took place (arrow 1). A small metal grid barrier inserted at the tube centre point is used to separate the rats before the start of every trial (arrow 2). One of two holding boxes, filled with bedding from the rats’ home cages (3). C) Rats were placed at either end of the tube and they readily moved towards the centre point. When both rats reached the central grid barrier, the barrier was lifted and the start button pressed to begin the trial. The rats competed for dominance until the head of the “losing rat” moved past the entrance at either end of the tube. See supplementary movie. D) The two-phases of training. Phase 1 involved contests between all animals in each cage, this being repeated across 10 sessions to secure a mean measure of within-cage ranking. Cartoon depicts all 6 inter-animal contests. Phase 2 consisted of competitions between cages and involved all the animals of one cage and all the animals of the other cage (4 animals (cage1) x 4animal (cage 2) = 16 animal pairs). For clarity, the cartoon depicts only one exemplar rat (green) competing against all rats of the other cage. In phase 2, there were in 4 cages in total that competed against each other. Figure 2 Phase 1 - Establishment of intra-cage hierarchy in single line cages A) Cage hierarchies: Plot of individual animal rank across all 10 sessions for WT and Het single-line cages. The 4 colours represent each cage individual, the colour code plotted in accordance with average rank across all 10 sessions. B) Mean ranks and individual data points (for each session) of the WT cage (blue) and Syngap ‘Het’ cage (Orange) showed significant hierarchies in both cages (WT: F (1.378, 12.40)=33.45, p=<0.0001, with Greenhouse-Geisser correction for degrees of freedom; Het: F (1.138, 10.24)=27.89, p=0.0002). C) No significant differences were observed between WT and Het lines with respect to either Stability (two-tailed, unpaired t-test, t(6)=0.195, p=0.852, n=4) or Variance. (t-test, t(6)=0.193, p=0.853, n=4). Means ± 1 SEM. * indicates p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001, in Tukey’s multiple comparison test. See main text for details. Figure 3 Phase 1 - Establishment of intra-cage hierarchy in mixed-line cages A) Cage hierarchies: Plot of individual animal rank across all 10 sessions for WT and Het single-line cages. The 4 colours represent each cage individual, the colour code plotted in accordance with average rank across all 10 sessions. There was no overlap between the more dominant WT (blue) and more submissive Het animals (Orange), although the three submissive WTs were broadly comparable to the most dominant Het. B) Contest wins and average rank in the mixed cages: There were highly significant differences between the number of wins (X2=68.06, df=1, p<0.0001) and average rank of WT and Het animals (t=6.73, df=78, p<0.0001). C) As in the single-line cages, there was no differences were observed with respect to either Stability (unpaired t-test, t(6)=1.321, p=0.235) or Variance (t(6)=1.124, p=0.304). Means ± 1 SEM. **** p<0.0001. Figure 4 Phase 1 - Time taken to complete individual contests (Latency) A) Average latency of trials (s) from all pairs of rats from all cages over 10 sessions (n=16). B,C) Average latency for mixed cages (n=8) and single line cages (n=8). Note longer time taken by Het animals in the single-line cages. For all graphs, individual latency data points for each pairing are not shown in order to increase clarity of results. Means ± 1 SEM. Figure 5 Phase 1 - Patterns of behaviour observed during contests A) Sum of occurrences of various behaviours by individual animals plotted as a function of overall rank, together with individual animal data (n=16). High rank is animals in Ranks 1 or 2 of a cage (green), Low-Rank is animals in Ranks 3 and 4 of a cage (red). B) Sum of occurrences of behaviours by WT animals living together (n=4). C) Sum of occurrences of behaviours by Het animals living together (n=4). Note high occurrence of “STILLNESS” in this cage (>40) but not the WT cage (<20). D) Sum of occurrences of behaviours by high- and low-rank WT and Het animals living together in mixed cages (n=8). Note that STILLNESS is again high, but in these cages restricted to the High-Rank animals (i.e. the WT rats). For all graphs, behavioural occurrences were taken as an overall count of each behaviour for each animal over 240 trials from Phase 1, sessions 1 and 10. Means ± 1 SEM. Figure 6 Phase 2 - Competitions between animals from different cages as a function of genotype and rank predicted from Phase 1 A) The overall phenotype of WT dominance phenotype of Phase 1 was lost in Phase 2 - WT and Het animals won an equivalent number of contests. B, C) Contests won by WT and Het animals that had lived in single-line cages (B, n=4 per cage) or Mixed-Line cages (C, n=4 per cage). Note that Het animals from single-line cages won more contests in Phase 2, whereas WT animals from Mixed-line cages won more contests. D) Predictive rank: when wins and losses in Phase 2 are plotted as a function of Rank secured by animals in Phase 1 (High vs. Low), the phase 1 rank is predictive of the outcome of contests in Phase 2. E) In contests between High-rank animals (n=6, WT; n=2 Het), the Het animals won far more contests. F) In contests between Low-rank animals (n=2 WT, n=6 Het), the total number of wins was equivalent. Means ± 1 SEM. * p<0.05, *** = p<0.001, indicates p<0.0001. See text for details and comments on numbers of animals. Figure 7 Phase 2 - Patterns of behaviour observed during contests (sessions 1 and 3) A) In Phase 2, there was modest but a significant difference between rank and behaviour subtypes. The trends suggest high-ranking animals showed more moving forward and stillness but less retreat. B) No significant difference in behaviour subtypes were observed when WT animals were compared with Het. C) Analysis of only the contests between high-ranking WT and Het animals revealed striking interaction was found between high-ranking WT and Het animals and social behaviour subtype (ANOVA; F(5,30)=8.807, p=<0.001). The data reveals high-ranking Het displayed more of every behaviour subtype except withdrawal. D) The opposite pattern prevailed for trials only involving the contests between low-ranking WT and Het animals: (ANOVA; F(5,30)=18.99, p=<0.001). For all graphs, behavioural occurrences were taken as an overall count of each behaviour for each animal over the 960 trials of sessions 1 and 3 of Phase 2. Means ± 1 SEM. ** indicates p<0.01, **** indicates p<0.0001. Conflict of Interest All authors declare no conflicts of interest regarding this work. Roles Emma Harris – conducted the study and prepared first drafts of figures. Honor Myers – conducted the study and prepared first drafts of figures. Kapil Saxena – set up the apparatus, secured the animals and provided day-to-day supervision. Rufus Mitchell-Heggs – advised on suitable analyses and contributed to the text. Peter Kind – grant holder and commented on drafts of the text. Shona Chattarji – grant holder and commented on drafts of the text. Richard Morris – laboratory principal investigator, designed the study, modified the figures, and wrote the manuscript. Supplementary Video The videos are accessible at: https://datashare.ed.ac.uk/handle/10283/4005 Aceti M Creson T Vaissiere T Rojas C Huang W Wang Y Petralia R Page D Miller C Rumbaugh G Syngap1 Haploinsufficiency Damages a Postnatal Critical Period of Pyramidal Cell Structural Maturation Linked to Cortical Circuit Assembly Biological Psychiatry 2015 77 9 805 815 25444158 Asiminas A Jackson AD Louros SR Till SM Spano T Dando O Bear MF Chattarji S Hardingham GE Osterweil EK Wyllie DJA Sustained correction of associative learning deficits after brief, early treatment in a rat model of Fragile X Syndrome Science Translational Medicine 2019 11 eaao0498 10.1126/scitranslmed.aao0498 31142675 Berryer M Hamdan F Klitten L Møller R Carmant L Schwartzentruber J Patry L Dobrzeniecka S Rochefort D Neugnot-Cerioli M Lacaille J Mutations inSYNGAP1Cause Intellectual Disability, Autism, and a Specific Form of Epilepsy by Inducing Haploinsufficiency Human Mutation 2012 34 2 385 394 23161826 Chaste P Leboyer M Autism risk factors: genes, environment, and gene-environment interactions Dialogues in clinical neuroscience 2012 14 3 281 292 23226953 Chou Y-J Lu Y-H Ma Y-K Su Y-S Kuo T-H The decisive role of subordination in social hierarchy in weanling mice and young children iScience 2021 24 102073 10.1016/j.isci.2021.102073 33604524 Clement J Aceti M Creson T Ozkan E Shi Y Reish N Almonte A Miller B Wiltgen B Miller C Xu X Pathogenic SYNGAP1 Mutations Impair Cognitive Development by Disrupting Maturation of Dendritic Spine Synapses Cell 2012 151 4 709 723 23141534 Crawley JN What’s wrong with my mouse? Second Edition Wiley New York 2006 Creson T Rojas C Hwaun E Vaissiere T Kilinc M Jimenez-Gomez A Holder J Tang J Colgin L Miller C Rumbaugh G Re-expression of Syngap protein in adulthood improves translatable measures of brain function and behavior eLife 2019 8 Cummins D How the Social Environment Shaped the Evolution of Mind Synthese 2000 122 1/2 3 28 Ding Q Sethna F Wang H Behavioral analysis of male and female Fmr1 knockout mice on C57BL/6 background Behavioural Brain Research 2014 271 72 78 24886775 Ellenbroek B Youn J Rodent models in neuroscience research: is it a rat race? Disease Models Mechanisms 2016 9 10 1079 1087 27736744 Fan Z Zhu H Zhou T Wang S Wu Y Hu H Using the tube test to measure social hierarchy in mice Nature Protocols 2019 14 3 819 831 30770887 Friard O Gamba M BORIS: a free, versatile open-source event-logging software for video/audio coding and live observations Methods in Ecology and Evolution 2016 7 1325 1330 Gurts AM Cost GJ Freyvert Y Zeitler B Miller JC Choi VM Jenkins SS Wood A Cui X Meng X Vincent A Knockout rats via embryo microinjection of zinc-finger nucleases Science (New York, NY) 2009 325 433 Guo X Hamilton P Reish N Sweatt J Miller C Rumbaugh G Reduced Expression of the NMDA Receptor-Interacting Protein Syngap Causes Behavioral Abnormalities that Model Symptoms of Schizophrenia Neuropsychopharmacology 2009 34 7 1659 1672 19145222 Guy J Gan J Selfridge J Cobb S Bird A Reversal of neurological defects in a mouse model of Rett syndrome Science 2007 315 1143 7 17289941 Hamdan F Daoud H Piton A Gauthier J Dobrzeniecka S Krebs M Joober R Lacaille J Nadeau A Milunsky J Wang Z De Novo SYNGAP1 Mutations in Nonsyndromic Intellectual Disability and Autism Biological Psychiatry 2011 69 9 898 901 21237447 Huang W Wang D Allen W Klope M Hu H Shamloo M Luo L Early adolescent Rai1 reactivation reverses transcriptional and social interaction deficits in a mouse model of Smith-Magenis syndrome Proceedings of the National Academy of Sciences 2018 115 42 10744 10749 Jeyabalan N Clement J SYNGAP1: Mind the Gap Frontiers in Cellular Neuroscience 2016 10 1 32 26858601 Kalueff A Stewart A Song C Berridge K Graybiel A Fentress J Neurobiology of rodent self-grooming and its value for translational neuroscience Nature Reviews Neuroscience 2015 17 1 45 59 26675822 Katsanevaki D Till SM Buller-Peralta I Watson TC Nawaz MS Arkell D Tiwari S Kapgal V Biswal S Smith JAB Anstey NJ Heterozygous deletion of SYNGAP enzymatic domains in rats causes selective learning, social and seizure phenotypes bioRxiv 2020 2020.2010.2014.339192 Kazdoba T Leach P Silverman J Crawley J Modeling fragile X syndrome in the Fmr1 knockout mouse Intractable Rare Diseases Research 2014 3 4 118 133 25606362 Kilinc M Creson T Rojas C Aceti M Ellegood J Vaissiere T Lerch J Rumbaugh G Species-conserved SYNGAP1 phenotypes associated with neurodevelopmental disorders Molecular and Cellular Neuroscience 2018 91 140 150 29580901 Kim J Lee H Takamiya K Huganir R The Role of Synaptic GTPase-Activating Protein in Neuronal Development and Synaptic Plasticity The Journal of Neuroscience 2003 23 4 1119 1124 12598599 Komiyama N Watabe A Carlisle H Porter K Charlesworth P Monti J Strathdee D O’Carroll C Martin S Morris R O’Dell T Syngap Regulates ERK/MAPK Signaling, Synaptic Plasticity, and Learning in the Complex with Postsynaptic Density 95 and NMDA Receptor The Journal of Neuroscience 2002 22 22 9721 9732 12427827 Kunkel T Wang H Socially dominant mice in C57BL6 background show increased social motivation Behavioural Brain Research 2018 336 173 176 28859999 Llamosas N Arora V Vij R Kilinc M Bijoch L Rojas C Reich A Sridharan B Willems E Piper DR Scampavia L SYNGAP1 Controls the Maturation of Dendrites, Synaptic Function, and Network Activity in Developing Human Neurons The Journal of Neuroscience 2020 40 7980 7994 32887745 Lore R Flannelly K Rat Societies Scientific American 1977 236 5 106 118 McMahon A Barnett M O’Leary T Stoney P Collins M Papadia S Choudhary J Komiyama N Grant S Hardingham G Wyllie D Syngap isoforms exert opposing effects on synaptic strength Nature Communications 2012 3 900 1 9 McNaughton C Moon J Strawderman M Maclean K Evans J Strupp B Evidence for social anxiety and impaired social cognition in a mouse model of fragile X syndrome Behavioral Neuroscience 2008 122 2 293 300 18410169 Michaelson S Ozkan E Aceti M Maity S Llamosas N Weldon M Mizrachi E Vaissiere T Gaffield M Christie J Holder J Miller C SYNGAP1 heterozygosity disrupts sensory processing by reducing touch-related activity within somatosensory cortex circuits Nature Neuroscience 2018 21 12 1 13 Miczek KA de Boer SF Aggressive, Defensive, and Submissive Behavior Whishaw IQ Kolb B The behavior of the laboratory rat: A handbook with tests Oxford University Press 2005 344 352 Muhia M Yee B Feldon J Markopoulos F Knuesel I Disruption of hippocampus-regulated behavioural and cognitive processes by heterozygous constitutive deletion of Syngap European Journal of Neuroscience 2010 31 3 529 543 20105235 Nakajima R Takao K Hattori S Shoji H Komiyama N Grant S Miyakawa T Comprehensive behavioral analysis of heterozygous Syngap1 knockout mice Neuropsychopharmacology Reports 2019 39 3 223 237 31323176 Nelson AC Kapoor V Vaughn E Gnanasegaram JA Rubinstein ND Murthy VN Dulac C Molecular and Circuit Architecture of Social Hierarchy bioRxiv 2020 10.1101/838664 Ogden K Ozkan E Rumbaugh G Prioritizing the development of mouse models for childhood brain disorders Neuropharmacology 2016 100 2 16 26231830 O’Roak B Stessman H Boyle E Witherspoon K Martin B Lee C Vives L Baker C Hiatt J Nickerson D Bernier R Recurrent de novo mutations implicate novel genes underlying simplex autism risk Nature Communications 2014 5 1 1 6 Ozkan E Creson T Kramár E Rojas C Seese R Babyan A Shi Y Lucero R Xu X Noebels J Miller C Reduced Cognition in Syngap1 Mutants Is Caused by Isolated Damage within Developing Forebrain Excitatory Neurons Neuron 2014 82 6 1317 1333 24945774 Parikshak Neelroop N Luo R Zhang A Won H Lowe Jennifer K Chandran V Horvath S Geschwind Daniel H Integrative Functional Genomic Analyses Implicate Specific Molecular Pathways and Circuits in Autism Cell 2013 155 1008 1021 24267887 Saxena K Webster J Hallas-Potts A Mackenzie R Spooner P Thomson D Kind P Chatterji S Morris R Experiential contributions to social dominance in a rat model of fragile-X syndrome Proceedings of the Royal Society B: Biological Sciences 2018 285 1880 20180294 Spencer C Alekseyenko O Serysheva E Yuva-Paylor L Paylor R Altered anxiety-related and social behaviors in the Fmr1 knockout mouse model of fragile X syndrome Genes, Brain and Behavior 2005 4 7 420 430 16176388 Spigel I Trivett S Fraser D Grooming behavior and competitive dominance in the albino rat Journal of Comparative and Physiological Psychology 1972 78 3 409 411 Toseeb U McChesney G Wolke D The Prevalence and Psychopathological Correlates of Sibling Bullying in Children with and without Autism Spectrum Disorder Journal of Autism and Developmental Disorders 2018 48 7 2308 2318 29423609 Vazquez L Syngap Regulates Spine Formation Journal of Neuroscience 2004 24 40 8862 8872 15470153 Walkup WGIV Mastro TL Schenker LT Vielmetter J Hu R Iancu A Reghunathan M Bannon BD Kennedy MB A model for regulation by SynGAP-α1 of binding of synaptic proteins to PDZ-domain ‘Slots’ in the postsynaptic density eLife 2016 5 e16813 27623146 Wang F Kessels H Hu H The mouse that roared: neural mechanisms of social hierarchy Trends in Neurosciences 2014 37 11 674 682 25160682 Wang F Zhu J Zhu H Zhang Q Lin Z Hu H Bidirectional Control of Social Hierarchy by Synaptic Efficacy in Medial Prefrontal Cortex Science 2011 334 6056 693 697 21960531 Wesson D Sniffing Behavior Communicates Social Hierarchy Current Biology 2013 23 7 575 580 23477727 Yang C Bai Y Ruan C Zhou H Liu D Wang X Shen L Zheng H Zhou X Enhanced Aggressive Behaviour in a Mouse Model of Depression Neurotoxicity Research 2014 27 2 129 142 25367807 Yizhar O Fenno L Prigge M Schneider F Davidson T O’Shea D Sohal V Goshen I Finkelstein J Paz J Stehfest K Neocortical excitation/inhibition balance in information processing and social dysfunction Nature 2011 477 7363 171 178 21796121 Yoo H Genetics of Autism Spectrum Disorder: Current Status and Possible Clinical Applications Experimental Neurobiology 2015 24 4 257 272 26713075 Zhou T Zhu H Fan Z Wang F Chen Y Liang H Yang Z Zhang L Lin L Zhan Y Wang Z History of winning remodels thalamo-PFC circuit to reinforce social dominance Science 2017 357 6347 162 168 28706064

PMC007xxxxxx/PMC7614820.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8918110 Eur J Neurosci Eur J Neurosci The European journal of neuroscience 0953-816X 1460-9568 31944427 7614820 10.1111/ejn.14681 EMS177465 Article A stable home-base promotes allocentric memory representations of episodic-like everyday spatial memory Broadbent Nicola 1* Lumeij Lucas Berend 2* Corcoles Marta 2 Ayres Alice I 2 Ibrahim Mohammad Zaki Bin 2 Masatsugu Brittany 1 Moreno Andrea 2 Carames Jose-Maria 2 Begg Elizabeth 2 Strickland Lauren 2 Mazidzoglou Theofilos 2 Padanyi Anna 2 Munoz Monica 2 https://orcid.org/0000-0002-9981-4260 Takeuchi Tomonori 23^ Peters Marco 4^ https://orcid.org/0000-0001-8661-1520 Morris Richard G M 2^ https://orcid.org/0000-0003-2263-8139 Tse Dorothy 2^ 1 Dart Neuroscience 2 Centre for Discovery Brain Sciences, Edinburgh Neuroscience, University of Edinburgh 3 Danish Research Institute of Translational Neuroscience (DANDRITE), Department of Biomedicine, Aarhus University, Denmark 4 Department of Neurobiology and Behavior and Center for the Neurobiology of Learning and Memory, University of California, Irvine, Irvine, CA, USA ^ Corresponding authors: Dorothy Tse (Dorothy.Tse@ed.ac.uk) and Richard G M Morris (r.g.m.morris@ed.ac.uk) * First co-authors 01 4 2020 20 3 2020 05 7 2023 26 7 2023 51 7 15391558 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. A key issue in neurobiological studies of episodic-like memory is the geometric frame of reference in which memory traces of experience are stored. Assumptions are sometimes made that specific protocols favour either allocentric (map-like) or egocentric (body-centered) representations. There are, however, grounds for suspecting substantial ambiguity about coding strategy, including the necessity to use both frames of reference occasionally, but tests of memory representation are not routinely conducted. Using rats trained to find and dig up food in sandwells at a particular place in an event-arena (episodic-like 'action-where' encoding), we show that a protocol previously thought to foster allocentric encoding is ambiguous but more predisposed towards egocentric encoding. Two changes in training protocol were examined with a view to promoting preferential allocentric encoding - one in which multiple start locations were used within a session as well as between sessions; and another that deployed a stable home-base to which the animals had to carry food reward. Only the stable home-base protocol led to excellent choice performance which rigorous analyses revealed to be blocked by occluding extra-arena cues when this was done after encoding but before recall. The implications of these findings for studies of episodic-like memory are that the representational framework of memory at the start of a recall trial will likely include a path-direction in the egocentric case but path-destination in the allocentric protocol. This difference should be observable in single-unit recording or calcium-imaging studies of spatially-tuned cells. event arena rats hippocampus frames of reference path-integration pmcIntroduction Humans automatically form memories of attended experience, a process thought to involve the hippocampus (Marr, 1971). If later asked about our day, we readily recall some events and where they happened, whereas we forget others with many parameters of remembering and forgetting now well understood (Wixted, 2004; Richards & Frankland, 2017). In the case of spatial memory, for which recall as well as recognition can be tested in animals, it is important to establish in which geometric frame of reference such event-memory traces are encoded. Is it allocentric or egocentric? That is, is the stored representation within a frame of reference that is independent of the actor or observer (allocentric), or in a body-centered frame of reference (egocentric)? In a classic study, Packard & McGaugh (1992) showed that, faced with ambiguity about potential representations in a T-maze, rats favour allocentric representations initially but this shifts over time to an egocentric strategy. However, this shift does not always occur for, in the delayed matching-to-place protocol in the watermaze in which a new escape location has to be learned each day, rats consistently use an allocentric representations for as long as they are trained (Steele and Morris, 1999). Bast et al., (2005) and Wang et al., (2010) have outlined a potentially powerful episodic-like ‘everyday memory' task for animals, conceptually similar to the ‘delayed-matching-to-place’ protocol in the watermaze (Steele & Morris, 1999). Procedurally an appetitive task, rats (or mice) are trained to find and dig up food reward in an event-arena over several weeks, entering the arena from one of four start locations whose location varies across days (North, S, E or W). The food is hidden in an odour-masked sandwell whose location also varies over days. They are later tested for the accuracy of their recall of where the action of finding food occurred most recently. The animals learn to do this well in about 10 days and then, successfully remember each day the location where food was dug up after only a single reward. Use of multiple reward pellets, each spontaneously carried back to the start location one by one, enhances memory retention. This recency recall is good for several hours, but then typically decays to chance levels over 24 h (Whishaw, 1998; Bast et al., 2005; Wang et al., 2010; Takeuchi et al., 2016; Nonaka et al., 2017). This task is analogous to aspects of human everyday memory in that testing takes place in a familiar environment (like a room in one's house), but the events that happen within it vary with respect to their location on a day-to-day basis (such as the action of putting down one's glasses somewhere). In a recent Technical Note published in this journal, various determinants of memory retention for this protocol were identified, including trial-spacing and peri-event novelty, along with certain molecular markers of enhanced retention derived from RNAseq (Nonaka et al., 2017). The suitability of this translationally relevant protocol to test novel cognitive enhancers related to neuronal plasticity was confirmed with a Phosphodiesterase Type 4 (PDE4) inhibitor. Subsequent to the Nonaka et al., (2017) publication, we discovered an unsuspected ambiguity with respect to the frame of reference of memory encoding. This emerged during tests in which the location from which the animals entered the arena was changed, within each session, between the memory encoding trial and the subsequent recall trial. Performance was very good when, as shown in Figure 1A (Protocol 1), the starting locations were the same for encoding and recall (even though these changed across sessions). However, it declined to chance levels when shifted by 90° or 180° between these two daily trial types. Experiment 1A in San Diego, replicated as Expt. 1B in Edinburgh, document this phenomenon. We were surprised by this finding as we had believed that the use of changing start locations across daily sessions in a stable environment would, as in the watermaze, promote allocentric representations. To the contrary, it seemed that being able to run back to the start-box promoted an egocentric path-integration solution that was severely disrupted by changing the start box location between the encoding and recall trials of a given session. Accordingly, we set about examining two different ways in which to promote allocentric encoding. In Protocol 2 (Figure 1B), multiple start locations were used for the several trials of daily memory encoding, while in Protocol 3 (Figure 1C), there was also a stable 'home-base' to which the animals were trained to carry the food reward. The latter protocol successfully precluded the use of path-integration. Thus, principled changes were made, step-by-step, between the successive protocols 1-3. There is a further important procedural detail relevant to the allocentric vs. egocentric representation issue. In Protocol 1 (Nonaka et al, 2017), the animals were typically given the opportunity to retrieve 3 food pellets (0.5 gm each), one by one, within each trial. In Figure 1A, a dotted line represents the one trial with three pellets collected during memory encoding. Specifically, upon digging for a few seconds successfully in a sandwell, the animals carried each pellet back to the (dark) original startbox (orange) where it was eaten. The animals then returned to the encoding sandwell to collect pellets 2 and 3 in turn. The resulting pattern of a path-out (black) followed by a return-home (green) is a component of hoarding behaviour in laboratory-based tasks studied intensively by Whishaw in a series of papers that collectively pointed to the importance of dead-reckoning/path-integration in rodent homing (Whishaw et al., 1995; Whishaw, 1998; Whishaw et al., 2001). His group observed that, provided the food-pellet takes longer to eat than the likely travel time back to a safe place to eat, the animals will generally run accurately back to the startbox to do so (as we also observed). A consequence of this 3-pellet schedule was that, even though the finding of the goal location for the first pellet typically involved a circuitous path, finding the second and third pellet of the memory encoding trial generally involved relatively direct paths from and then back to the start location. These are conditions that could inadvertently encourage a cumulative egocentric representation of path distance and direction, using self-monitored path integration (Redish, 1999; McNaughton et al., 2006). Note that this running back and forth does not occur in the watermaze (Morris et al., 1982; Sutherland et al., 1983) for which the encoding of goal-location happens on the escape platform with no return to the starting location. Memory representation in the standard and many other protocols of the watermaze is allocentric, although some experiments have been devised that require the use of local landmark heading vectors (Pearce et al., 1998). Keen to use a dry-land task for physiological and optogenetic studies later, the challenge before us was to establish a protocol affording true allocentric memory representations for the land-based event arena on the grounds that, to be a valid model, it should mimic our ability to remember where a daily event happened rather than merely memory of how to get there. Methods Subjects At Dart Neuroscience (DN), 11 adult male Long-Evans (Envigo Laboratories Huntingdon, UK; 300-400g at start of study) were used in Expt. 1A. Rats were housed 2/cage, food-restricted to 85-90% of the free-feeding weight (adjusted for growth), had free access to water and maintained on a 12:12 h light/dark cycle with training conducted during the light phase. All experimental methods were approved by the DN Institutional Animal Care and Use Committee, and followed the guidance of the National Research Council Guide for the Care and Use of Laboratory Animals Studies (2011). At Edinburgh (EDI), a total of 32 adult male Lister-hooded rats were used in Expt. 1B (n=7; Protocol 1), Expt. 2 (n=8; Protocol 2) and Expts. 3A and 3B (n=17; Protocol 3). As described in detail in Nonaka et al. (2017), they were group housed (3-4/cage; and similarly food-deprived to 85-90% of the free-feeding weight against a growth curve, with free-access to water, 12:12 h light-dark cycle with training in the light phase). Care of the animals complied with the UK Animals (Scientific Procedures) Act conducted under a Project Licence (PPL 60/4566). Blinding and replicability Growing interest in the replicability of biomedical studies has led us (and others) to be explicit about blinding and other procedures CAMARADES (Camarades). As the animals were not, except in Expt. 3B, in separate groups, caution regarding artefacts and replicability took the form of adherence to specific procedures. These included careful design of the sandwells to prevent olfactory artefacts, counterbalancing of the sequencing of trials to prevent an animal from merely following the path of a previously tested animal, the experimenter(s) scoring probe tests without knowledge of the correct sandwell of the 5 or 6 being used, independent scoring from video data by two independent experimenters with cross-correlation of their data, and so on. We did not rotate the arena between trials or sessions, but instead cleaned the perspex floor surface with alcohol-impregnated wipes between all trials. In Expt. 3B, the experimenters were also blind with respect to whether an individual animal had received drug or vehicle. Apparatus All experiments were conducted using each of two identical ‘event-arenas’ in both Edinburgh and San Diego. Designed and made in Edinburgh, their construction and appearance is fully described in Nonaka et al (2017). Of note is that numerous precautions were taken to avoid olfactory artefacts, with behavioural checkpoints used to ensure that these worked. Specifically, Plexiglas sandwells (6 cm diameter, 4 cm depth) that contained the hidden reward pellets were placed in one or a subset of the floor panels with holes. To mask the smell of the food, the sandwells were filled with bird sand mixed with Garam Masala (P&B Foods, Bradford, UK), 150 g/5 kg sand initially (and replenished daily). Each sandwell had a spherical plastic bowl within it in which one or more reward pellets (0.5 g) were placed and thereby accessible. This plastic bowl also made it possible for an equal number of reward pellets to be placed underneath, and thereby inaccessible. The plastic bowls had holes and so were porous to odours, ensuring that the rewarded and non-rewarded sandwells contained the same number of reward pellets at approximately the same depth in the sand and thus should exude the same smell. Extensive randomising and counterbalancing was also arranged to minimize olfactory artefacts: (a) the same sandwells used in the encoding trial were not used for the recall trial of the same session; (b) all sandwells were used a rewarded or non-rewarded sandwell across days; (c) the arena floor was regularly wiped with a 70% alcohol impregnated towel between trial, and before recall and probe trials. The sandwell arrangement is shown in Figure 1B of Nonaka et al (2017). Procedure The arena consists of four different startboxes (on the video screen but not magnetically at North, S, E and W) that were located just outside the perimeter of a large 1.6 x 1.6 m arena. Within it, there is a 7 x 7 grid of possible sandwell locations, with one or more sandwells containing either accessible food (rewarded) or non-accessible food (non-rewarded). The arenas were each set within a laboratory room with stable extra-arena cues. The animals entered the arena from a startbox at either N, S, E or W in random sequence (depends on protocol) across successive sessions of training, continuing across many weeks. New memories were formed in each session, and usually forgotten within 24 hr. The apparently automatic one-trial encoding of where the food-digging event happened leading to one-shot memory in both recall choice-trials and recall-probe tests after daily encoding pointed to an episodic-like memory representation within an allocentric map-like framework (Tolman, 1948; O'Keefe and Nadel, 1978). Habituation In all protocols, the rats were first taught to dig for food in sandwells inside their home cages. In a first habituation session in the arena(s), containing no sandwells, the rats were permitted to explore the arena with two intra-arena cues and surrounding extra-arena cues for 10 min. They were then given five sessions of daily habituation, starting by being put in a startbox and given a 0.5 g ‘cue’ food pellet to eat. When the pellet was eaten (typically around 30 s), the rats were allowed 10 min access to the arena. Rats started exploration from a different startbox in each session and were trained to search and dig for control food pellets in sandwells in the various locations in the event-arena. On habituation session 2, one 0.5 g pellet was placed on top of the sandwell; rats collected the pellet and took it back to the startbox. On habituation sessions 3 & 4, one 0.5 g pellet was placed on top of the sandwell and another was buried in the middle of the sandwell. On habituation session 5, three 0.5 g pellets were buried at the bottom of the sandwell. By the end of habituation, all rats were running quickly into the arena, collecting pellets and returning to the startbox to eat each pellet. Habituation normally lasted for 6 sessions. Protocol 1 - training - encoding and recall choice trials The Protocol 1 experiments constituted a collaboration between two independent laboratories with only minor differences of procedure between them. The key feature of Protocol 1 is that the animals run back and forth between the correct sandwell and the startbox three times. In Expts 1A and 1B, there were 2 sandwells available during memory encoding trials, one rewarded and one non-rewarded, and 6 sandwells during memory recall trials (one sandwell with accessible food reward, and 5 sandwells with only inaccessible food). Later studies used either 1 (S+) or 2 sandwells (S+ and S-) at encoding. It makes little difference whether there is a choice on the encoding trial or not, and thus data from both procedures within each protocol are described. A training session consisted of an encoding trial followed ~ 60 min later by a recall trial. On encoding trials, each rat was placed in the startbox designated for that session (N, S, E, or W, counterbalanced across sessions) and given a 0.5 gm flavoured pellet. Once the experimenter had exited the testing room, the startbox door was opened remotely and the rat allowed to explore the arena and sandwell(s), one of which contained accessible food (S+). The encoding trial ended once the rat had retrieved the pellets from the rewarded sandwell and returned to the startbox. The door to the arena was closed. On recall trials, the rat was returned to the original startbox and then presented in the arena with six sandwells which included the one rewarded and now five unrewarded sandwells. The focus in the recall trial was whether a rat would preferential choose the rewarded sandwell from the encoding trial(s), and ended once the rat retrieved 3 pellets from it and returned to the startbox. Each training session used a different six sandwell 'configuration', requiring the rats to learn a new rewarded sandwell location each session. The configuration map was used for all rats in each session (i.e. locations A, B, C, D, E and F); however, half of the rats were trained on a different pair of rewarded and nonrewarded encoding locations (i.e. A and B) from the rest (i.e. C and D) to control for any potential location or response biases. We saw no indication that any rat was following the path of a previously trained which would, anyway, have been an unsuccessful strategy. We also always cleaned the arena with 70% alcohol wipes between trials. Memory for location during encoding was calculated in two separate ways. The first measure, used during training, was choice performance during recall trials - called Performance Index (PI). For clarity and comparison to 2-alternative forced choice data, this index was computed to ensure a 100% score implied perfect memory (minimal errors) whereas a 50% score implied chance [Performance Index (PI) = (maximum number of errors that can be made - number of errors made on this trial)/Maximum number of errors that can be made) x 100]. The second measure of memory, which is very sensitive, was the proportion of time spent digging at the correct or other sandwells during recall trials when no accessible food was available. The first trial type is called a "recall choice trial"; the second a "recall probe trial". Other measures during training trials included latency before digging at the correct sandwell, and qualitative aspects that we noted such as paths taken, returns to an inappropriate startbox, etc. Protocol 1: Memory recall probe trials Multiple recall probe trials were used to test the impact of different conditions relevant to the allocentric vs. egocentric coding issue (e.g. time delay, arena rotation, drug infusion). These probe tests consisted of a recall phase with only non-rewarded sandwells present (typically 5 or 6). The rats were allowed to search for the correct sandwell for 60 s from the first dig at any sandwell, with the time spent digging monitored carefully. After 60 s, the experimenter placed pellets in the correct sandwell (to prevent extinction) and the animal allowed to find them. Dig time during the 60 s recording period was measured, and the relative proportion of time at the correct and incorrect sandwells was calculated as percentage dig times. Protocol 1: Experiment 1A (San Diego) Once the rats (n=11) were consistently performing with a PI above 80% on the daily sessions, test sessions were interleaved periodically. The consistent performance across days in this longitudinal paradigm is critical for allowing different tests at different times with data that can be compared. One series of tests investigated the effects of varying the retention interval on memory. These test sessions consisted of a standard encoding trial, with the animals only retrieving 1 pellet from the rewarded sandwell on the encoding trial. Following a variable delay (0.5 – 72 h), memory recall for the event location was assessed using a probe test. The impact of rotating the starting location between encoding and recall trials by 180° during four sessions (with intra-arena cues removed) was then examined. For example, rats released from the N startbox on the encoding trial would, on the recall trial, be released from the N startbox (control procedure) or from the S startbox (rotation condition). The impact of occluding spatial cues was also assessed. For these tests, there was a standard encoding trial, a memory delay of 60-90 min, and then half the rats received a recall trial with intra-arena cues removed and extra-arena cues occluded (by curtains around the arena). The other half of the rats received a standard recall trial with all cues visible. These two conditions were examined within-subjects in counterbalanced order, interleaved with additional training sessions. Protocol 1: Experiment 1B (Edinburgh) This study of the impact of startbox rotation was conducted at the end of the Nonaka et al. (2017) study using the same animals (n=7; Figure 2). By that time, they had completed 100+ training sessions over 5 months. Using essentially the same procedure in Edinburgh as in San Diego, standard control sessions were interleaved with 'rotation' sessions in which the startbox location was rotated for either a 180° rotation (as in San Diego) or a 90° rotation. These rotations could potentially have a different impact as the 180° rotation varies the relation between the goal-sandwell and the startbox in both distance and direction (for example, from 'near and to the left of during encoding to a location that was 'far and to the right of' during testing); whereas in the 90° rotation (Figure 2B) achieves a symmetrical flip between 'near vs. far' (with respect to proximity between the goal location and startbox), but with the direction (right or left) unaltered. Protocol 2: Experiment 2 (Edinburgh) In Protocol 2 (Expt. 2; Figure 1B), using new experimentally naive subjects (n=8), we made one key change. This involved having multiple start locations during the memory encoding trials rather than the single start location of Protocol 1 that had inadvertently encouraged egocentric encoding. In Protocol 2, there were also 3 encoding trials but now from 3 separate start locations within each session (and in a counterbalanced sequence across sessions). However, as in Protocol 1, the location of the rewarded sandwell location continued to vary between sessions to ensure this was an episodic-like task in which the animals had to remember where digging up food occurred most recently. We began with 10 sessions of 'pre-training' using a 5-alternative-choice (5-AFC) sandwell discrimination protocol in which all 5 sandwells were available on the 3 encoding trials. This procedure was not of primary interest, but was included to encourage learning that one sandwell was rewarded but the others not. For main Protocol 2 model, intended to simplify the memory demands at encoding, only a single rewarded sandwell was used from session 11. The expectation was that performance would improve and reflect allocentric coding, based on the successful DMP procedure in the watermaze in which multiple start locations are used within a single session of training to a single hidden escape platform in an otherwise featureless pool (Steele & Morris, 1999). The fatal complication that rapidly emerged from using this protocol, which does not arise in the watermaze, is that the animals had to remember from which startbox they had started and thus to which they should return with their single reward pellet. This proved very problematic, with striking interference building up within each session. This problem is captured graphically in Figure 1B with the return paths of the animals from the encoding sandwell (green) reflecting the confusion about where to go. From session 17, training was increased to as many as 9 daily encoding trials from the 3 different starting locations, ending with the recall trial on the 10th trial from a novel start location. In this final attempt to get Protocol 2 to work, all startbox doors were open with the carrying of the food reward to any startbox being permitted. Protocol 3: Experiments 3A and 3B (Edinburgh) In Protocol 3, a further conceptual change in protocol was added. The standard procedure in food hoarding tasks is that the return after foraging is to the starting location (as in Whishaw's studies). If this anchoring promotes egocentric encoding, perhaps the demand to carry the reward pellets to a fixed 'home-base' might change things to favour allocentric encoding. The logic behind this likely change in preferred strategy is that cumulative idiothetic path-integration could get the animal back to the start location, but could not direct an animal to the home-base from which the animals did not start on that trial. But it was the home-base to which the animals ran with the reward. In effect, the animals had no alternative but to do the task in a different way. The home-base (North) was never used as a startbox, only as the place to go with the 0.5 g food rewards. This shift ensured that carrying was never along a direct path from the rewarded sandwell back to the startbox, thereby likely precluding path-integration (Figure 6A). Instead of 3 daily encoding trials (as Protocol 1 and Protocol 2), we used only 2 encoding trials that could begin at either E, W or S (in random sequence across days), with the door to the home-base (at N) opened once the animal had dug up food at the rewarded sandwell. Access to W, S and E was simultaneously disallowed by closing the door of whichever startbox door had been used on that trial. Once the animal had successfully carried food to the home-base (green paths in Figure 1C), a second run was allowed from that home-base to the rewarded sandwell (black path), enabling pellet 2 to be secured, followed by a return to the home-base (red path) whereupon the door was then closed. The second encoding trial was given approximately 1-2 min later, but from a different start-box location. After a short delay (24 min), a recall trial was scheduled using 6 sandwell protocol (of which only the encoding location contained accessible food). This trial began at the one remaining unused starting location for that trial (red path). Again, the animal had to carry the food to the home-base. Protocol 3: Rigorous procedures for and analyses of allocentric encoding Having established that the rats (n=8) could successfully learn this allocentric protocol, we then examined spatial memory over two retention delays (24 min vs. 24 h), predicting the same decay of memory to near-chance levels as observed by Nonaka et al. (2017). We also examined the impact of removal or alteration of the intra- and extra-arena cues between the encoding and recall trials, predicting that doing this would now have a deleterious effect on performance to chance levels. The analysis of allocentric encoding began with a post-hoc video-analysis conducted to examine the qualitative paths taken on encoding and recall trials. The reasons for doing this is because we wondered if a cryptic egocentric strategy might yet possible. Specifically, an independent observer (J-MC) blind to the experimental conditions on any trial viewed all videos ad hoc and recorded the number of times animals went directly to two different locations: one location was to the correct sandwell from the startbox (necessarily using an allocentric strategy); the other was to the home-base before going to the correct sandwell (a strategy which could potentially be permissive for egocentric encoding with a switch to path-navigation from the home-base). A direct approach beyond a 45° angle to the side-walls (Figure 7A) was used as the criterion for identifying approach to the home-base, thereby excluding occasions when the animal merely ran around the perimeter (such trials were in practice very rare). The frequency of the distribution of these distinct paths across 4 sessions was monitored. Experiment 3B - Electrophysiology In Experiment 3B, male Lister-Hooded rats, weighing 250+ g, n=5 per drug (CNQX) 6-cyano-7-nitroquinoxaline-2,3-dione, 3 mM; Tocris, Abingdon, UK and artificial cerebrospinal fluid (aCSF, Sigma, Irvine, UK) were used in the non-recovery electrophysiology studies. These animals were prepared for acute surgery in a stereotaxic apparatus (David Kopf Instruments) under non-recovery urethane anaesthesia (1.3 g/kg body weight; Sigma-Aldrich, USA), with the first intraperitoneal injection given during brief isofluorane anaesthesia (4% isoflurane in 0.8 l/min O2). These studies typically lasted 6–8 hr, with the initial 2 hr being spent securing accurate placement of the stimulating and recording electrodes and cannula, and the subsequent 4 hr monitoring field-potential baseline and the impact of intrahippocampal drug infusions. Stimulating and recording electrode positions are shown in Figure 8A. The stimulating electrode was a twisted bipolar Teflon-coated platinum-iridium electrode (20 μm diameter, 400 μm coated diameter for each of the two single strands) aimed at the angular bundle of the perforant path (anterior-posterior (AP) 0.0 mm from lambda; mediolateral (ML) 4.2 mm; dorsoventral (DV) 2.15 mm from dura). The recording electrode was a single Teflon coated platinum-iridium wire targeting the hilus of the dentate gyrus (AP 4.08 mm from bregma; ML 2.5 mm; DV 3.5 mm). The drug cannula was a 28 gauge stainless steel tube whose tip was stereotaxically located at least 0.5 to 1.0 mm (± 0.3) mm away from the recording electrode (AP 3.6 mm from bregma; ML 2.6 mm; DV 3.5 mm). Conventional field-potential recordings were made, with stimulation every 20 s, and these monitored and calculated on-line using EPS software (in house). In response to biphasic 200 μs stimulus pulses of circa 600–800 μA, we measured both the early-rising slope of the evoked potential by linear regression over several points, and the amplitude of the evoked dentate population spike. The stimulus intensity was adjusted to secure initial population spike amplitudes of circa 3–6 mV. Once acquired using suitable electrode placements, potentials typically remained relatively stable over periods of up to 3–4 hr, with a small upward drift of the population spike (but not fEPSP) that rarely exceeded 15% over this long period. Animals for which the potentials were unstable were discarded. The same long time-period stability was observed when aCSF, was infused into the dorsal hippocampal formation at a depth targeting a region encompassing the stratum moleculare of area CA1. A volume of 2 μl aCSF (as vehicle) (in mM: 150 Na+, 3 K+, 1.4 Ca2+, 0.8 Mg2+, 155 Cl-, 0.2 H2PO4-, 0.8 HPO42-, pH 7.2) or CNQX was infused (0.5 μl/min) that, on the basis of previous autoradiographic data (Riedel et al., 1999), would be expected to diffuse throughout the entire CA1, CA3 and dentate gyrus regions of the septal (dorsal) hippocampus. After the infusion, electrophysiological recordings, measuring the same parameters and under the same conditions per animal, last for 180 min. Protocol 3 - Experiment 3B - Impact of hippocampal inactivation Finally, we explored the impact of temporary inactivation of the dorsal hippocampus through microinfusion of the AMPA receptor antagonist CNQX via intracerebral cannulae. Rats (n=9) were anaesthetised with 2-5% isoflurane (Abbott, UK) and positioned in a stereotaxic frame (Kopf instruments, CA, USA). Guide cannulae (26-gauge; Plastics One, UK) were implanted bilaterally into the dorsal hippocampus (coordinates relative to skull at Bregma: AP = -4.5 mm; ML = 3.0 mm; DV from dura = -2.5 mm). Dental cement (Kemdent, Purton, UK) was then sculpted around the guide cannulae. Solid stainless steel (“dummy”) cannulae were inserted into the implanted guide cannulae to prevent infection or blockages. The dummy cannulae were 33 gauge with a 0.5 mm protrusion from the end of guide cannulae when inserted. All rats were allowed a recovery period of at least 10 days to allow them to regain their pre-surgery weights before food restriction and behavioural testing commenced. Phosphate-buffered aCSF were used as infusion vehicles and for dissolving drugs. Drug concentration for infusions was 0.89 mg/ml (3 mM) of the competitive AMPA/kainate receptor antagonist CNQX. The pH of the CNQX solution was adjusted to 7.2 by the addition of concentrated phosphoric acid. These volumes and concentrations were calibrated in a study with electrophysiological monitoring of the extent and duration for which excitatory post-synaptic potentials and population spikes were blocked by this concentration and volume of CNQX (Rossato et al., 2018). One day before drug infusions, a mock infusion was used to habituate rats to the drug infusion conditions. Rats were restrained manually and their dummy cannulae removed. Injection cannulae were placed into the guide cannulae (for 5 min) but no solutions were infused into the rats’ brains. Thereafter, the rats were restrained manually and infusions into both hemispheres were performed simultaneously in a control testing room. Prior to infusion, the injection cannulae tips were dipped into 70% alcohol and then rinsed in saline. The tips of these infusion cannulae protruded 0.5 mm from the ends of the guide cannulae within the brain, and were connected to microsyringes (SGE brand, World Precision Instruments, FL, USA) on a microinfusion pump (World Precision Instruments, USA) via flexible polyvinyl chloride tubing (Plastics One, UK). The flexible tubing was rinsed through with bottled water for injections (Hameln, Gloucester, UK). CNQX (1 μl/hemisphere) was infused at a rate of 0.2 μl/min over 5 min, after which the infusion cannulae were left in place for a further 2 min to ensure all droplets of solution entered the brain. Dummy cannulae were then rinsed with alcohol and saline and placed back into the guide cannulae. Finally, all rats were terminally anaesthetised with Euthanal (Merial, Roydon, UK) and then perfused intracardinally with 0.9% saline, followed by 4% formalin in saline. The brains were removed and stored in 4% formalin for several days. Coronal sections (30 μm) were cut using a cryostat for histological analysis and were mounted on slides, stained with cresylviolet, and coverslipped using DPX. The sections were examined under a light microscope with 20-fold magnification to verify cannulae placements. For each brain, the infusion site was plotted by determining the deepest point (Figure 8D) at which tissue damage was evident and marking this location on the appropriate coronal sections from a standard rat brain atlas (Paxinos & Watson, 1998). Results Protocol 1: Experiment 1A In Protocol 1, rats run back and forth between the start box and the correct sandwell during the sample trial, and hopefully chose the correct sandwell in the choice trial. It was observed that the rats rapidly acquired the standard task of running into the arena from a different daily startbox, finding and digging up a single pellet of food 3 times during a single memory encoding trial (3 pellet encoding; Figure 2A). During the recall trial, they exhibited similar levels of memory recall in San Diego to that of the Nonaka et al. (2017) study conducted in Edinburgh. In a series of recall probe tests, the animals showed time-dependent forgetting of the location of the rewarded sandwell characteristic of everyday memory (Figure 2B). When the delay interval was short (0.5-3 h), rats spent significantly more percent time digging at the correct sandwell compared to chance (0.5 h = 39.6 ± 6.4%, 3 h = 35.4 ± 7.8%, one sample t-test vs. chance (16.7%), ts(10) > 2.40, ps < 0.05). Retention intervals of 24 h or longer resulted in chance performance (24 h = 26.5 ± 5.4%, 48 h = 21.9 ± 5.1%, and 72 h = 19.9 ± 3.8; ts(10) = 0.85-1.82, n.s.). Figure 2C,D shows the effects of the 180° rotation in the starting location between encoding and recall choice trials. This resulted in a dramatic decline in the average PI score from circa 85% to chance levels (chance = 50%, before rotation: ts(10) = 7.67-18.81, p < 0.0001; after rotation: ts(10) = 0.46-2.32, n.s.). Performance did not improve across further sessions of training, although two representative animals (Nos 3 and 7) illustrate within group variability in reaction to the rotation test. The poor mean performance across the startbox-rotation sessions contrasts with that observed for the initial acquisition of the standard no-rotation version of the task. As shown in Figure 2E,F, the rats surprisingly exhibited unimpaired performance when the intra- and extra-arena cues were obscured during a choice trial (t(8) = 0.82, n.s.; 2 animals excluded for failing to dig in the sandwells effectively). This outcome suggests that the 10 animals included were largely using a egocentric or path navigation strategy to find the correct sandwell. Although they could find the correct well with high levels of accuracy on trials where the spatial cues were obscured, they were significantly slower to make the first dig in any sandwell on these trials (latency to first dig on trials with extra-arena cues = 11.5 ± 4.7 s, latency to first dig on trials without extra-arena cues = 79.1 ± 12.3 s; t(8) = 3.9, p <0.01), indicating that the rats were at least aware of the change in the extra-arena environment. Protocol 1: Experiment 1B This surprising "Houston, we have a problem" finding prompted a replication, conducted in Edinburgh. This used animals that were already running well in the arena according to the identical training protocol and showing a comparable level of efficiency with a PI of >80% (as in Nonaka et al., 2017). To test the impact of startbox rotation between encoding and recall, 4 encoding/recall trials were given across 12 sessions with either no rotation between encoding and recall (2 tests) or either a 180° or a 90° rotation (2 tests; Figure 3A,C). These tests were interspersed with 8 sessions of regular training. The change in startbox location caused an immediate reduction in recall to chance levels (Figure 3B,D: 180°: t (6) = 3.1, p < 0.05; 90°: t (6) = 3.2, p < 0.05). The data is plotted with means and SEMs, together with individual data points that graphically display the increased variability of the PI scores on rotation trials, some animals showing little change in the PI score from circa 85% whereas most show a decline to as many as 5 errors per trial (i.e. to a PI = 0%). This variability may, respectively, reflect a subset using an egocentric encoding strategy (massive disruption in performance) and a smaller subset that were predominantly using an allocentric strategy (little change). The implication appears to be that the 'standard' training protocol fosters an ambiguous outcome with respect to coding strategy. Protocol 2: Experiment 2 Our next step towards trying to find an effective allocentric strategy was to supplement daily between-session changes in startbox location with within-session changes also (as in Bast et al, 2005). Specifically, in Protocol 2, there were three sample trials but each begun in different startbox locations. Specifically, instead of allowing 3 pellets to be collected and taken one-by-one back to the same startbox to eat, Protocol 2 permitted (a) only 1 pellet for each startbox location, and (b) scheduled 3 separate memory encoding trials from different starting locations (Figure 4A,B). A fourth startbox location was used for the memory recall trial. This protocol was begun with new animals (n = 8) in which pretraining consisted of a 10-session sandwell 5-AFC discrimination protocol (Figure 1B). Performance improved from chance levels to show a trend for above chance memory recall (Figure 4; yellow shading), but was characterised by considerable within-animal variability with only 2/8 animals showing less than 3 errors on every recall trial at some point from session S4 to S10. Average performance over these sessions was 70.1 +/- 3.1%, above chance, but it was unstable on a day-to-day basis and at only 62.50 +/- 10.6% on session 10 (N.S. compared to chance). The main single sandwell encoding procedure of Protocol 2 training was begun at session 11 with, initially and encouragingly, performance well above chance (84.4 +/- 6.6% correct; t(7) = 5.23, p = 0.001; Figure 4; green shading). However, instead of being sustained, the within-animal variability across sessions continued to be problematic and average performance steadily declined across the next 5 sessions. We wondered if this poor performance might be overcome by additional daily encoding trials. Accordingly, over sessions 16-18, these were increased from 3 to 9 encoding trials by repeating three times the sequence of a single pellet from each of the 3 initial startbox locations. However, no change in performance was observed (Figure 4; blue shading). The PI on session 18 was at chance (t(7) = 0.42, n.s.). Observation of the behaviour of the animals in the arena revealed the problem. Specifically, there was 'confusion' about where the animals should go with their reward pellet, triggering such behaviour as patrolling around the perimeter of the arena (see Figure 1B). This would likely have caused interference in working memory, limiting effective memory of that day's target sandwell location. In short, increasing the number of start locations into the arena from 1 to 3 did not help. As this failure could have been due to a poor batch of rats, we sought to check that these same animals could nonetheless learn an egocentric strategy. We therefore continued training these animals beyond S18 using a single sandwell during memory encoding and allowing the animals to return repeatedly to a single start location (i.e. a return to Protocol 1). Performance improved dramatically and stabilised at levels of around 75% or better throughout (reliably above chance at 79.8 +/- 2.4 % with t(7) values ranging from 2.50 to 5.70, ps < 0.05; data not shown). There was therefore nothing odd about this batch of animals. Protocol 3: Experiment 3A&B The turning point of this work occurred in two studies, conducted with new cohorts of animals (n=17 total; ns = 8 and 9 respectively). In Protocol 3, the key change was to assign a stable "home-base" to which the animals should carry any reward pellets they had dug up (Figure 1C; home-base in blue). Specifically, having dug up food during either a sample or choice trial, the food was not to be carried back to the start, but always to the home-base (which may be to the left, right or straight ahead of the animal's then location, but always in the same allocentric location. Following this change, Expt. 3A established the successful use of allocentric coding, while Expt. 3B revealed hippocampal-dependence. Even though experimentally naïve, the animals of both experiments were reluctant to approach the home-base at the outset, making frequent attempts to re-enter the startbox from which any trial had commenced (entry door was now closed). After a few sessions of training, with 2 encoding trials each from 2 different startboxes per session (varying in location across sessions), they began to more readily enter the stable home-base willingly and settled into a routine of doing this routinely by sessions 6-8. The PI measure rose to a high level quickly, was stably elevated across successive sessions, and significantly above chance. Representative paths taken by one exemplar animal on 2 encoding trials and a later recall choice trial is shown in the supplementary movie file (See Movie S1). Whereas trial 1 was characterised by exploration all over the arena until the location of the sandwell was found, encoding trial 2 from a different startbox position, shows a typically direct approach to the correct location of the sandwell (see movie). The recall trial shows good performance with one proximal digging error before the direct approach to the target (i.e. a PI of 80%). Averaged across all animals, the mean PI on the recall trial stabilised across sessions, with ever more direct paths from the rewarded sandwell to the home location. Interestingly, as they did so, signs of hesitation about leaving the startbox on the initial encoding trial of the day tended to increase, with the animals apparently inspecting the arena and extra-arena cues before venturing out. Unfortunately, this pausing behaviour was not timed (but will be in future studies). Non-encoding control trials (early and late in training), to check for any olfactory artefacts, showed choice trial performance fall exactly to chance levels when the initial encoding trials were not given (S18 and S68; Figure 5A,C) Thus, the animals were not following any differential smell cues emanating from a rewarded sandwell. They did, however, notice the change in non-encoding procedure (i.e. a recall trial without prior encoding trials) with much longer latencies to dig at the correct sandwell on ‘non-encoding’ sessions (Figure 5C). In a critical test of 'everyday' memory, statistically significant forgetting characteristic of ‘everyday’ memory was observed over 24 h (t(7) = 2.85, p < 0.05; Figure 5B). Additionally, there was significantly lower total digging in all sandwells in the 24 h condition (t(7) = 2.97, p<0.05). Overnight forgetting is an important characteristic of episodic-like everyday memory. Two independent observers showed a close correlation in their scoring of the dig times at all sandwells during probe trials (Figure 5D), pointing to the objectivity of our 'blind' data scoring. We then conducted two tests of allocentric encoding - one procedural, the other analytic. First, we examined whether memory recall was affected by limiting access to the extra-arena cues during the choice trials (Figure 6A). There was a clear sensitivity to the occlusion of intra- and extra-arena cues. Performance declined to chance in a statistically significant manner (t(7) = 3.37,p < 0.05; Figure 6B). The total time spent digging at all sandwells was also significantly lower when cues were occluded (t(7)=3.70, p<0.05) indicating, as in Expt. 1, that the animals noticed the change in contextual cues (Data not shown). Second, we analytically addressed the unlikely possibility that, instead of the stable home-base (at North) aiding allocentric encoding, it was used as an 'anchor point' for a dead-reckoning-like accumulation of distances and rotations that could potentially mediate a cryptic egocentric path-navigation route to the correct sandwell. On this alternative view, the animals would have to first use allocentric memory to go from any startbox to the north location (using reference memory), and then switch to an egocentric strategy while exploring from this home-base anchor to the rewarded sandwell. We therefore monitored approaches to the home-base (Figure 7A; see Methods). The importance of this analysis derives from the fact that, if the animals did this, they would also fail on the arena cue-occlusion test (above) because they would be unable to locate the home-base (Figure 6B); they would fall to chance on the non-encoding trial (Figure 5A); and would most likely also show the overnight forgetting characteristic of everyday memory (Figure 5B). This critical additional analysis hinges upon whether (a) the animals approach the home-base preferentially on encoding or recall trials or both, and (b) display an increasing proportion of approaches to the home-base across the three daily trials. Video files were monitored across 4 sessions for each of two sub-groups of trained animals (total n=17) to identify the frequency of approaches from any startbox to the home-base prior to approaching the correct sandwell at which to dig. These showed a declining percentage of approaches to a level of 27% on the recall trial (Figure 7B, black triangles, true data), precisely the opposite of the prediction one would make for a cryptic egocentric strategy for which it should increase (Figure 7B, red symbols, theoretical data). When the sub-set of animals (n=5) that did sometimes preferentially approach the home-base were compared with those going directly to the correct sandwell before carrying the reward to the home-base (n=12), there was no difference in PI score between the two sub-groups (Figure 7C; t (15) = 0.051, n.s.). That approaches are made to the home-base on encoding trial 1 (47%) is not itself surprising, as it likely reflects a combination of (a) exploration on trial 1 of the day (with searching all over the arena including to all of the startboxes), and (b) the north box being the place where a total of 6 food pellets are eaten each session. The home-base would thereby have acquired secondary reinforcing properties through Pavlovian context conditioning. The frequency of the different combinations of preferential approach to the home-base across encoding and recall trials before the animal went to the rewarded sandwell are shown in Figure 7D. Protocol 3: Experiment 3B Finally, using new animals (n=9), hippocampal-dependence of both memory encoding and memory recall were examined. In Expt. 3B, the quantitative characteristics of PI performance were very similar to those of Expt. 3A (data not shown). The critical measure was the impact of bilateral micro-infusions (2 ul) of CNQX (6-cyano-7-nitroquinoxaline-2,3-dione; 3 mM), an antagonist for AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazole propionate)/kainate-type glutamate receptor, into the dorsal and intermediate hippocampus. The first step was to examine the impact of aCSF or CNQX on perforant path evoked dentate field-potentials (fEPSPs and PS). These showed a clear drug and defined times (blue and grey shading) when relevant tests of the impact on encoding and recall could be tested (Figure 8A). In the behavioural studies, drug infusions were given 15 min prior to memory encoding or 15 min before memory recall in separate tests, and the on memory recall in a probe test at 3 h. The dose and volume used has been shown in the Edinburgh lab (Rossato et al., 2018) to cause a massive disruption of the both EPSPs and population spikes in evoked field-potentials e.g. (Rossato et al., 2018). As shown in Figures 8B,C bilateral CNQX infusion caused a complete blockade of both memory encoding and memory recall (ts(8) > 2.31, p<0.05). There were also no differences in total dig time across drug conditions (ts(8) < 0.82, n.s; Data not shown). Histological analysis of the tips of the bilateral cannulae were located in the dorsal hippocampus (Figure 8D). Discussion The important new finding of this study is that the provision of a stable home-base to which food should be carried on each trial in an appetitively motivated open-field foraging task favours allocentric encoding. This observation is new because previous studies of food-carrying have pointed to egocentric encoding as the predominant coding strategy when animals carry food back to the start position. Allocentric encoding emerges with the provision of a stable home-base that is distinct from a varying starting location. Detailed investigation of the first and the third training protocol showed that animals preferentially adopted an egocentric or an allocentric form of memory encoding respectively. Frequent returns to an initial startbox favoured but did not absolutely enforce egocentric encoding (Protocol 1). The use of multiple startbox locations (Protocol 2) caused confusion about where food was to be carried within a session in which both egocentric and allocentric encoding strategies could be adopted and interfere. Critically, carrying the reward to a fixed, stable home location favours allocentric coding, even when coupled to varying startbox locations within a single session of training (Protocol 3). These findings have implications for both behavioural and cognitive tests of episodic-like memory and, likely, also for physiological single-unit recording studies of the spatial localization and navigation system. Our starting point was that unexpectedly poor spatial memory was observed, in independent across-laboratory studies, when there was a shift of startbox location between encoding and retrieval (Expts. 1A,1B). First observed in San Diego (Expt. 1A), and replicated in Edinburgh (Expt. 1B), Protocol 1 led to a relatively dominant use of egocentric encoding or a path-integration strategy to get to the correct sandwell. The across laboratory replication, which adds rigor to our observations, also revealed a small sub-set of animals in both laboratories that were relatively unaffected by within-session changes in startbox location. The change made for Protocol 2 to multiple within-session startbox locations, intended to promote allocentric memory encoding, was not successful (Expt. 2). It became apparent that the demands of remembering the different locations of where food reward was to be carried once it had been secured caused considerable interference such that effective memory processing of the goal-location was poor (Figure 1B). Interestingly, the use of multiple start positions within a daily session is standard in watermaze experiments, but there is no ‘return to the start position' in such experiments - the animal waits on the escape platform until removed. Thus, there may have been a modest beneficial effect of multiple start locations for allocentric encoding in the event arena, but this benefit was obscured by our requirement that the animals carry the food to varying safe places to eat within a session. The novel finding of the study emerged from the use of Protocol 3 requiring that the food reward be carried always to a stable allocentrically-defined home-base at the perimeter. This was successful in realising effective memory recall and switching animals to a truly allocentric memory representation (Expts. 3A and 3B). Performance on recall trials is (A) stably above chance across successive sessions; (B) shows gradual forgetting of everyday memory over 24 h on non-rewarded probe trials; and (C) a memory representation that is sensitive to the occlusion of intra- and extra-arena cues. However, as an extra precaution, we note that even this latter and ostensibly definitive test of allocentric encoding is ambiguous because a potential cryptic path-integration strategy might nonetheless have been deployed. This strategy would have been to run to the home-base and then use it as the anchor-point for subsequent dead-reckoning. Blind analysis of videos of the paths taken by the animals in Protocol 3 showed, however, that such a strategy was not used by the animals. Additionally, (D) hippocampal-dependence was established pharmacologically by showing that blockade of fast synaptic transmission with a selective AMPA receptor antagonist at the time of memory encoding or separately at memory recall itself impaired memory at 24 h. Thus, the key new concept to emerge from these studies is that, while the Whishaw procedure of food-carrying by the animals back to a start location usually encourages an egocentric/ path-integration strategy (Whishaw et al., 1995; Whishaw, 1998; Redish, 1999; Whishaw et al., 2001), allocentric encoding is dominant when a safe, fixed, allocentrically defined home-base location is used that is separate from the start locations. Automatic encoding of everyday memory The ‘episodic-like’ feature of this everyday memory task in its various forms is based on the concept shared by other tasks that much memory encoding for single events happens automatically in the course of everyday life. Tasks such as novel object recognition, object place memory and object-place-context memory also reveal 'automatic' memory encoding associated with experiences that are not rewarded (Ennaceur & Delacour, 1988; Aggleton & Pearce, 2001; Eacott & Easton, 2007). However, these are recognition tasks in which novelty-induced and context-specific novelty-induced exploration (Morris, 1983) can be triggered in various ways - by objects that are absolutely novel, by familiar objects in novel locations, or by objects in contexts different to those used during initial memory encoding (Langston & Wood, 2010). There is always something visually different in such object- or location change protocols, changes that can be recognised, but there is usually no demonstration of memory recall in such tasks. Recollection of 'what-where-which' has been successfully modelled in an object exploration task (Eacott et al., 2005), but the level of performance attained, while significantly above chance, was modest at even the shortest memory delay. In contrast, the event-arena shares with the watermaze that it is a 'memory recall' task. Our supposition is that the animal recollects where it performed the action of digging up food most recently, and indexes that recollection by navigating effectively to it from any starting position (Morris et al., 1982; Steele & Morris, 1999; Morris, 2006). We have no direct evidence that the animal remembers the act of digging, but it is a very natural behaviour and it seems likely that they would. Nor in these studies is there direct evidence that the animals recall what food is to be found in the target location, although food-specific memory has been observed in studies of paired-associate learning involving schemas (Tse et al, 2007). The event arena has two potentially powerful advantages over the watermaze. First, although not used in these studies, it is permissive for electrophysiological and optical recording of brain activity during behaviour. Second, as the animals only spend time at sandwells, irrespective of where they may have been located on a previous session, memory encoding can be studied in the absence of extinction by simply moving a sandwell from one location to another. As noted, a limitation of the event-arena to date is that variation in the 'what' or 'which' components of recall have not yet been investigated in the episodic-like protocols, but this could be examined by varying the target flavour of food to be secured (as in Tse et. al, 2007), or the context cues in which an arena is placed. The latter test would be analogous to manipulations examined in the context of novel object recognition memory ('object-context' and 'object-place-context'). To date, the focus in the everyday memory task has been primarily on the recency of where the discrete action of digging up food occurred. This 'action-where' conjunction ensures that, at the time of recall, there are no perceptual affordances that could mediate recognition of the correct location over any other. On choice and probe trials, the arena always looks the same. Recency memory is conceptually related to automatic encoding (Marr, 1971; Morris, 2006). One consequence of automatic encoding is the risk of too much information being encoded in the course of a day creating potential saturation and interference. Thus, forgetting in the form of retention selectivity is, we believe, an essential feature of this form of memory. A human example might be remembering where one has parked one's bicycle at the station on the daily commute, or recalling where one's glasses have recently been mislaid around the house. Such memory is useful for a few hours, but generally not necessary for longer periods. Such memory must, almost by definition, fade over time. It is precisely this kind of everyday recollective memory that is at risk in older individuals and those in the early stages of neurodegenerative diseases that target memory formation, and for which palliative cognitive enhancement could be so useful. Egocentric or allocentric encoding A key issue in this study concerns the frame of reference in which the 'where' component of recency memory is encoded. In Protocol 1, there was a clear dominance of egocentric encoding from the starting position. Terrestrial rodents (e.g. the Norway Rat) do create foraging trails from their burrows that are helpful in mediating a route home for themselves and, potentially, other rats (Galef & Buckley, 1996). In laboratory settings, and in the absence of odour trials, rats may also keep track of the path they have taken using path-integration to encode distance moved, radial turning etc., as described in both quantitative experimental work (Whishaw, 1998; Whishaw & Brooks, 1999) and formal path-integration models of navigation (McNaughton et al., 1991; McNaughton et al., 1996; Redish, 1999). Egocentric encoding may have been encouraged in our first protocol by allowing the animals to repeatedly carry the reward back to the same startbox from which they emerged at the start of the first encoding trial - for which they would have become more accurate across the 3 reward pellets used in encoding. The use of such a strategy would not preclude the animals also forming an allocentric representation, but our original protocol likely allowed an egocentric representation to display trace dominance (Dudai, 2012) when the two types of memory representation were put in competition. The variability of the data secured in both the initial San Diego and the Edinburgh experiments reflected precisely that ambiguity. However, our interest is not primarily in the navigational path the animal takes than in how its accuracy tells us something about episodic-like everyday memory encoding. The procedural change of creating a home-base in Expts. 3A and 3B disposed the animals towards a "where is it?" memory representation. In Protocol 3, our animals did continue to carry food, as in the experiments of Whishaw, but to a dark, safe, well-learned home-base rather than to the varying starting location. Eilam & Golani, (1989) have noted that, even in open-arenas, rats create a stable home-base for themselves. Critically, our 'north' home-base was encoded allocentrically and stored in long term memory, thereby obviating regular updating via working memory and so limiting interference with newly encoded information. Frustration on the part of the animals about where to go in the early sessions was reflected in frequent attempts to get back to the original startbox of the day, but they gradually settled into running directly to the home-base with their large 0.5 g reward pellet. It is also noteworthy that, upon opening the door of the startbox at the beginning of a memory encoding trial, the animals in the home-base protocol would generally pause and inspect the arena and surrounding cues, as if to identify their location within it and work out where to go. A representation of self-location, likely mediated by hippocampal encoding using place-cells, complemented by prefrontal activity contributing to vectorial representations of goal-location and then trajectory (Ito et al., 2015; Sarel et al., 2017), should also be sensitive changes in the intra- and extra-arena cues between encoding and retrieval. We observed a decline of performance to chance levels in Expt. 3A when these cues were occluded by curtains, in contrast to what was observed in Expt. 1A. Conclusion and implications for single-unit recording studies Does any of this matter for single-unit recording studies? We suspect it does because, for example, goal-location recall is generally not necessary in any task in which an animal deploys a praxic egocentric strategy. To the contrary, the navigational system need only keep track of the animal's movements and compute - using path-integration - a 'return vector' that would later enable the action system to carry it out. Interestingly, there is now considerable interest in the single-unit recording community about the possibility that self-location is encoded egocentrically in the medial and lateral entorhinal cortex using the metric of grid- and landmark-vector cells (Moser et al., 2008; Knierim et al., 2014). Head-direction cell firing likewise implies representation of head orientation within an environment that is perceived as polarized (Dudchenko, 2015). In such coding frameworks, goal-location encoding does not matter - only the representation of how to get there. Like tourists lost in Manhattan who are told by a local resident to walk 5 blocks north and then take a left and walk 3 blocks west, they arrive at their destination without ever knowing where it is. Our analysis suggests that in situations in which there is a single start location to which the animal returns frequently, egocentric coding will gradually come to prevail as training continues (Packard & McGaugh, 1992). The gradual shift over learning of the receptive fields of CA1 place cells to reflect reward locations observed by Boccara et al., (2019), using a single start location, may be a similar phenomenon to the dominance of routes over goals observed by Grieves et al., (2016). In contrast, place cells recorded when there is no explicit task requiring directed navigation display allocentric encoding of self-location that is sensitive to cue-card rotation (Muller et al., 1987). Reward related distortions of the metric of space may still occur in a manner that reflects aspects of the navigational task underway (Butler etal., 2019), but the coding of goals could nonetheless be allocentric. It remains a much harder task to identify the neurobiological mechanisms by which the nervous system identifies the location of goal place G from a remote start place S (a task that Sachin Deshmukh (pers. comm.) has amusingly identified as "someone else's problem"). However, the directed performance of animals in the watermaze from any point on the circumference of the pool to the hidden platform (Morris, 1984), of rats on successive hexagons of the honeycomb maze (Wood et al., 2018), and of the animals trained in the home-location protocol in the present study, collectively indicate that spatial memory recall can be realised from a remote location. The location recalled can surely much further away than the several theta cycles of distance observed in the important studies of vicarious trial and error behaviour by (Johnson & Redish, 2007) but dissociations of remote allocentric goal-identification independent of egocentric path-directionality have not to our knowledge yet been conducted. The home-base event arena protocol may yet lend itself to such a study. In conclusion, our findings qualify but do not invalidate our recent observations of the determinants of selective everyday memory and forgetting (Nonaka et al., 2017). Rather, they have led us to a modification of the protocol that can hopefully serve as an effective test-bed for further examining the impact of parameters such as trial-spacing, unexpected novelty and neuromodulation on memory retention using both behavioural and physiological techniques. Acknowledgements We thank Patrick Spooner for the event arena construction and software support, and thank the two anonymous reviewers for their constructive suggestions. Funding This work was supported by intra- and extramural funding from Dart Neurosciences, and by grants to RGMM from the European Research Council (NEUROSCHEMA - 268800) and the Wellcome Trust (207481/Z/17/Z). Data accessibility The database containing the primary data used in this study title Broadbent_2019_database.sav will be made available on Figshare by Wiley. Figure 1 Three distinct protocols. (A) Protocol 1: In our standard protocol (as used in Nonaka et al, 2017), there was a single encoding trial consisting of three runs (black path) out to either of two sandwells (one of which was rewarded) followed by returns to the same startbox (green paths). After a memory delay, a choice trial included a run out (red path) to choose amongst 6 sandwells (only one correct) and then a return once again to same startbox. (B) Protocol 2: The primary modification was the use of four different startboxes within a session, thereby changing from 3 rewards at encoding within a single trial (Protocol 1) to three encoding trials each with a single reward pellet. The green return paths are representative in displaying the confusion of the animals about the location to which to return. The choice trial was from the fourth location of the day. (C) Protocol 3: The primary further modification was the use of a fixed 'home-base' (blue) to which the animals had to carry the food that they had dug up (green paths). Encoding was now divided into only 2 trials, but included the opportunity to run out of the home-base back to the sandwell on each trial (i.e. 4 rewards during encoding). The recall choice trial (and any probe trial) was started from a novel location on that session, as in Protocol 2. Figure 2 Protocol 1 - Experiment 1A (San Diego) - Time-dependent decay of every-day memory and the effect of 180 Startbox rotations on recall trial performance. (A,B) Sandwell arrangement for the arena used in San Diego for the study of memory retention across delays (30 min to 72 h). Memory for the correct location was significant at short retention intervals of 30 min and 3 h, consistent with Nonaka et al (2017), but memory returned to chance within 24 h. (C, D) Impact of a 180° startbox rotation in well-trained rats. The startbox location varied across successive sessions (not shown). In the recall trial, rats were either started from the same daily location, or the startbox was rotated by 180°. Rotation resulted in a significant decline of the performance index (PI). Note that individual animals vary, with one representative animal (no. 7) displaying only temporary disruption of performance, while another (no. 3) was consistently affected over 10+ sessions. (E, F) Occlusion of extramaze cues and removal of intramaze cues had no effect on performance. Means +/- 1 SEM. Figure 3 Protocol 1 - Experiment 1B (Edinburgh) - Startbox rotations impair recall trial performance. Using the same animals as in Nonaka et al (2017), we examined the impact of 180° (A, B) or 90° (C, D) rotations of the startbox used between the encoding and memory recall choice trial. The design allowed counterbalancing for “near” vs “far” (A), and “left” vs “right”(C) in the separate tests. Repeated measures data allowed comparison of an individual animals' scores on rotated and non-rotated trials. Both rotations resulted in a significant decline of the performance index (PI) on recall to chance level. Means +/- 1 SEM and individual animal data plots. Figure 4 Protocol 2 - Experiment 2 (Edinburgh) - Different startbox locations across successive encoding trials. Pretraining consisted of a 5-alternative discriminative choice procedure in which all 4 trials of a session started from different startboxes (yellow shading). From session 11 onwards, a single rewarded sandwell was used on each of 3 encoding trials from three different startbox locations (green shading). On the recall trial, there were 5 sandwells, with the obligation to choose the correct sandwell and then return with the food reward to the startbox location of that trial. From sessions 16-18, there were 9 encoding trials (blue shading, 1 pellet each). Performance was initially good upon transfer to the main encoding-choice protocol (session 11), but declined across further training sessions. Means +/- 1 SEM. Figure 5 Protocol 3 - Experiment 3A - impact of a stable home-base to which food-reward should be carried. (A) Rapid acquisition of effective performance, with stable above chance performance from session 16, this maintained through to session 70. Two non-encoding control sessions were conducted at the start and end of training (s18, s68) both show performance dropping to chance. Extended regular training provided a stable >80% PI baseline permitting various memory probes tests including a test of retention over 24 h and the impact of withdrawing spatial cues (Figure 6). (B) Memory retention with 3 pellet reward declines from well above chance at 24 min to a lower but still above chance level at 24 h. (C) latency data for the time taken to dig at correct sandwell. Note massive increase in this. time on non-encoding choice-trial sessions as predicted. (D) Inter-experimenter correlation of blind probe test scoring of two experimenters (AA and TT). Means +/- 1 SEM and individual animal data plots (B). Figure 6 Protocol 3 - Experiment 3A - impact of a stable home-base on performance and its decline upon removal of spatial cues. (A) Curtains occluding spatial cues were either drawn around the arena, and intra-maze cues removed, or these cues were fully available. (B) In a recall probe test at 24 min, performance was very good with cues available (>70%) or fell to chance (without cues). Means +/- 1 SEM and individual animal data plots. Figure 7 Protocol 3: Experiment 3A - detailed analysis of paths taken in the arena. (A) The criterion for identifying whether the rats approached the home-base or not before they reached the correct reward sandwell on successive daily trials. An approach was at an angle > 45°, whereas a by-pass was at < 45°. (B) A possible cryptic 'egocentric' strategy with the 'home-base' might be to run to it and then use it as an anchor-point for the start of a path-integration-associated accumulation of information. This view predicts that approaches to the home-base location would increase within each session, and be high on recall trials (red symbols and shading). In fact, the actual data (black-symbols) shows the opposite trend. Some animals visited the home-base location on encoding trial 1, but this declined as the animals learned the allocentric location of that session 's rewarded sandwell. (C) There was no difference in PI score between a subset of animals that approached the home-base first (grey) and those first visiting the correct sandwell (green). (D) The frequency of different combinations of preferential approach to the home-base before visiting the correct rewarded sandwell. The left is more egocentric, while right is more allocentric. The most egocentric category implies the rats would always visit the home-based before digging at the correct sandwell, while the most allocentric category implies they should visit the correct sandwell directly. Figure 8 Experiment 3B - Electrophysiology, Histology and. Behaviour - Impact of bilateral inactivation of the dorsal hippocampus. (A) Experimental design for electrophysiology. Dentate field potentials (fEPSP and PS) were measured over a period from -30 min to +180 min after 2 ul drug infusion of aCSF or CNQX. In the later behavioural study, the blue shading marks time-point after infusion when encoding trials were given (Data in B) and the grey shading marks the time point when recall trials were given (C). Note that both the fEPSP and the PS decline to near zero from a point about 15 min after infusion until 90 min, whereupon both measures return to baseline. (B) The hippocampus was inactivated by bilateral microinfusion of CNQX 15 min before the first of two encoding trials. The recall probe test was conducted 2.5 h later after the effects of CNQX would have dissipated (and thus hippocampal activity should be back to normal). Memory was poor after CNQX but normal after vehicle infusions. (C) The hippocampus was inactivated by bilateral microinfusion of CNQX 15 min before the recall probe test, encoding having been conducted in the absence of the drug. Memory was again poor after CNQX. Means +/- 1 SEM and individual animal data plots. (D) Histological verification of locations of the tips of the bilateral guide cannulae used for drug infusions. Author Contributions NB supervised Expt. 1A by BM in San Diego, with the study coordinated by MP. MC conducted the replication (Expt. 1B) in Edinburgh. LL and DT organised follow-up studies in Edinburgh (Expts. 2-3), assisted by ZI. RM conceived and, with TT, conducted the first stable home-location study; a full replication was conducted by AA and TT (Expts. 3A). DT organised various follow-up and confirmatory replication studies with the assistance of students and interns in the laboratory - EB, LS, TM and AP -collectively establishing that allocentric encoding depended on the integrity of hippocampal function (Expt. 3B). RM and DT did the surgeries to implant cannulae in dorsal hippocampus (Exp. 3B). JC did the detailed video-analysis of the animal paths in Expts. 3. MM assisted with the figures and provided a critical appraisal of the experiment and manuscript. The manuscript was written by RM, MP, NB, LL, MM and DT. Competing interest: None. Ethics The studies were conducted to comply with animal experimentation regulations of DART Neuroscience, NIH and the United Animals (Scientific Procedures) Act. Aggleton JP Pearce JM Neural systems underlying episodic memory: insights from animal research Philosophical Transactions of the Royal Society of London Series B: Biological Sciences 2001 356 1467 1482 11571037 Bast T da Silva BM Morris RGM Distinct contributions of hippocampal NMDA and AMPA receptors to encoding and retrieval of one-trial place memory J Neurosci 2005 25 5845 5856 15976073 Boccara CN Nardin M Stella F O’Neill J Csicsvari J The entorhinal cognitive map is attracted to goals Science 2019 363 1443 1447 30923221 Butler WN Hardcastle K Giocomo LM Remembered reward locations restructure entorhinal spatial maps Science 2019 363 1447 1452 30923222 Camarades Review of animal data from experimental studies 2014 http://www.dcn.ed.ac.uk/camarades/ Dudai Y The restless engram: consolidations never end Annu Rev Neurosci 2012 35 227 247 22443508 Dudchenko PA Spatial cognition: a tabula rasa for the sense of direction Curr Biol 2015 25 R143 144 25689906 Eacott M Easton A On familiarity and recall of events by rats Hippocampus 2007 17 890 897 17604354 Eacott MJ Easton A Zinkivskay A Recollection in an episodic-like memory task in the rat Learn Mem 2005 12 221 223 15897259 Eilam D Golani I Home base behavior of rats (Rattus norvegicus) exploring a novel environment Behav Brain Res 1989 34 199 211 2789700 Ennaceur A Delacour J A new one-trial test for neurobiological studies of memory in rats. 1: Behavioral data Behav Brain Res 1988 31 47 59 3228475 Galef BG Buckley LL Use of foraging trails by Norway rats Animal Behaviour 1996 51 765 771 Grieves RM Wood ER Dudchenko PA Place cells on a maze encode routes rather than destinations Elife 2016 5 Ito HT Zhang SJ Witter MP Moser EI Moser MB A prefrontal-thalamo-hippocampal circuit for goal-directed spatial navigation Nature 2015 522 50 55 26017312 Johnson A Redish AD Neural ensembles in CA3 transiently encode paths forward of the animal at a decision point J Neurosci 2007 27 12176 12189 17989284 Knierim JJ Neunuebel JP Deshmukh SS Functional correlates of the lateral and medial entorhinal cortex: objects, path integration and local-global reference frames Philos Trans R Soc Lond B Biol Sci 2014 369 20130369 24366146 Langston RF Wood ER Associative recognition and the hippocampus: differential effects of hippocampal lesions on object-place, object-context and object-place-context memory Hippocampus 2010 20 1139 1153 19847786 Marr D Simple memory: A theory for archicortex Philosophical Transactions of the Royal Society of London Series B: Biological Sciences 1971 262 23 81 4399412 McNaughton BL Barnes CA Gerrard JL Gothard K Jung MW Knierim JJ Kudrimoti H Qin Y Skaggs WE Suster M Weaver KL Deciphering the Hippocampal Polyglot: The Hippocampus as a Path Integration System J Exp Biol 1996 199 173 185 8576689 McNaughton BL Battaglia FP Jensen O Moser EI Moser MB Path integration and the neural basis of the ‘cognitive map’ Nat Rev Neurosci 2006 7 663 678 16858394 McNaughton BL Chen LL Markus EJ “Dead Reckoning”, Landmark Learning, and the Sense of Direction: A Neurophysiological and Computational Hypothesis Journal of Cognitive Neuroscience 1991 3 190 202 23972093 Morris RGM Neural subsystems of exploration in rats Archer J Birke L Exploration in animals and humans Van Nostrand Reinhold London 1983 117 146 Morris RGM Developments of a watermaze procedure for studying spatial learning in the rat Journal of Neuroscience Methods 1984 11 47 60 6471907 Morris RGM Elements of a neurobiological theory of hippocampal function: the role of synaptic plasticity, synaptic tagging and schemas Eur J Neurosci 2006 23 2829 2846 16819972 Morris RGM Garrud P Rawlins JNP O’Keefe J Place navigation impaired in rats with hippocampal lesions Nature 1982 297 681 683 7088155 Moser EI Kropff E Moser MB Place cells, grid cells, and the brain’s spatial representation system Annu Rev Neurosci 2008 31 69 89 18284371 Muller RU Kubie JL Ranck JBJ Spatial firing patterns of hippocampal complex-spike cells in a fixed environment Journal of Neuroscience 1987 7 1935 1950 3612225 Nonaka M Fitzpatrick R Lapira J Wheeler D Spooner PA Corcoles-Parada M Munoz-Lopez M Tully T Peters M Morris RGM Everyday memory: towards a translationally effective method of modelling the encoding, forgetting and enhancement of memory Eur J Neurosci 2017 46 1937 1953 28677201 Packard MG McGaugh JL Double Dissociation of Fornix and Caudate Nucleus Lesions on Acquisition of Two Water Maze Tasks: Further Evidence for Multiple Memory Systems Behavioral Neuroscience 1992 106 439 446 1616610 Paxinos G Watson C The Rat Brain in Stereotaxic Coordinates Academic Press Cambridge, MA 2013 Pearce JM Roberts ADL Good M Hippocampal lesions disrupt navigation based on cognitive maps but not heading vectors Nature 1998 396 75 77 9817202 Redish AD Beyond the Cognitive Map: From place cells to episodic memory MIT Press Cambridge, MA, USA 1999 Richards BA Frankland PW The Persistence and Transience of Memory Neuron 2017 94 1071 1084 28641107 Riedel G Micheau J Lam AG Roloff E Martin SJ Bridge H Hoz L Poeschel B McCulloch J Morris RGM Reversible neural inactivation reveals hippocampal participation in several memory processes Nature Neuroscience 1999 2 898 905 10491611 Rossato JI Moreno A Genzel L Yamasaki M Takeuchi T Canals S Morris RGM Silent Learning Curr Biol 2018 28 3508 3515 e3505 30415706 Sarel A Finkelstein A Las L Ulanovsky N Vectorial representation of spatial goals in the hippocampus of bats Science 2017 355 176 180 28082589 Steele RJ Morris RGM Delay-dependent impairment of a matching-to-place task with chronic and intrahippocampal infusion of the NMDA-antagonist D-AP5 Hippocampus 1999 9 118 136 10226773 Sutherland RJ Whishaw IQ Kolb B A behavioural analysis of spatial localization following electrolytic, kainate-or colchicine-induced damage to the hippocampal formation in the rat Behav Brain Res 1983 7 133 153 6830648 Takeuchi T Duszkiewicz AJ Sonneborn A Spooner PA Yamasaki M Watanabe M Smith CC Fernandez G Deisseroth K Greene RW Morris RGM Locus coeruleus and dopaminergic consolidation of everyday memory Nature 2016 537 357 362 27602521 Tse D Langston RF Kakeyama M Bethus I Spooner PA Wood ER Witter MP Morris RGM Schemas and memory consolidation Science 2007 316 76 82 17412951 Wang SH Redondo RL Morris RGM Relevance of synaptic tagging and capture to the persistence of long-term potentiation and everyday spatial memory Proc Natl Acad Sci U S A 2010 107 19537 19542 20962282 Whishaw IQ Place learning in hippocampal rats and the path integration hypothesis Neurosci Biobehav Rev 1998 22 209 220 9579312 Whishaw IQ Brooks BL Calibrating Space: Exploration Is Important for Allothetic and Idiothetic Navigation Hippocampus 1999 9 659 667 10641759 Whishaw IQ Coles BL Bellerive CH Food carrying: a new method for naturalistic studies of spontaneous and forced alternation J Neurosci Methods 1995 61 139 143 8618411 Whishaw IQ Hines DJ Wallace DG Dead reckoning (path integration) requires the hippocampal formation: evidence from spontaneous exploration and spatial learning tasks in light (allothetic) and dark (idiothetic) tests Behav Brain Res 2001 127 49 69 11718884 Wixted JT The psychology and neuroscience of forgetting Annu Rev Psychol 2004 55 235 269 14744216 Wood RA Bauza M Krupic J Burton S Delekate A Chan D O’Keefe J The honeycomb maze provides a novel test to study hippocampal-dependent spatial navigation Nature 2018 554 102 105 29364869

PMC007xxxxxx/PMC7614821.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101495384 Atten Percept Psychophys Atten Percept Psychophys Attention, perception & psychophysics 1943-3921 1943-393X 36253588 7614821 10.3758/s13414-022-02584-2 EMS158840 Article Working memory is updated by reallocation of resources from obsolete to new items Taylor Robert 1 Tomić Ivan 12 Aagten-Murphy David 1 Bays Paul M. 1 1 University of Cambridge, Department of Psychology, Cambridge, UK 2 University of Zagreb, Department of Psychology, Zagreb, CRO Correspondence should be addressed to: Ivan Tomic, Department of Psychology, University of Zagreb, Ivana Lucica 3, 10000, Zagreb, CRO; ivn.tomic@gmail.com 17 10 2022 17 10 2022 22 12 2022 26 7 2023 10.3758/s13414-022-02584-2This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Visual working memory (VWM) resources are limited, placing constraints on how much visual information can be simultaneously retained. During visually-guided activity, stored information can quickly become outdated, so updating mechanisms are needed to ensure the contents of memory remain relevant to current task goals. In particular, successful deallocation of resources from items that become obsolete is likely to be critical for maintaining the precision of those representations still in memory. The experiments in this study involved presenting two memory arrays of coloured disks in sequence. The appearance of the second array was a cue to replace, rehearse, or add a new colour to the colours in memory. We predicted that successful resource reallocation should result in comparable recall precision when an item was replaced or rehearsed, owing to the removal of pre-replacement features. In contrast, a failure to update WM should lead to comparable precision with a condition in which a new colour was added to memory. We identified a very small proportion (~ 5%) of trials in which participants incorrectly reported a feature from the first array in place of its replacement in the second, which we interpreted as a failure to incorporate the information from the second display into memory. Once these trials were discounted, precision estimates were consistent with complete redistribution of resources in the case of updating a single item. We conclude that working memory can be efficiently updated when previous information becomes obsolete, but that this is a demanding active process that occasionally fails. Visual working memory short term memory memory updating resource reallocation intrusion error pmcWorking memory is constrained in how much information can be actively maintained at any given time, which in combination with its central role in supporting cognition means that memory contents are continually in flux. Consider the task of keeping a shopping list in memory while going to a store, then receiving a call to buy apple juice instead of orange juice: this requires discarding an irrelevant shopping item and reallocating freed memory capacity to the new item. Updating a spatial memory representation of nearby cars while driving is another example, although one with less trivial consequences of updating failure. It has been proposed that solving such tasks requires updating mechanisms to ensure memory resources are reorganized effectively and the contents of memory remain relevant to current task goals (e.g., Oberauer, 2001, 2018). The efficiency of these mechanisms may be crucial in dynamic environments where information can quickly become obsolete. In human vision, the limited capacity of working memory has been observed as a decline in recall fidelity as the number of items in memory increases (Bays & Husain, 2008; Fougnie, Suchow, & Alvarez, 2012; Palmer, 1990; Schneegans, Taylor, & Bays, 2020; van den Berg, Shin, Chou, George, & Ma, 2012; Zhang & Luck, 2008). Successful deallocation of resources from obsolete items is therefore likely to be critical for maintaining the precision of task-relevant representations, while storage of new information requires resources to be reorganized. In retrospective cuing paradigms (Griffin & Nobre, 2003; Landman, Spekreijse, & Lamme, 2003; Maxcey-Richard & Hollingworth, 2013; Pertzov, Bays, Joseph, & Husain, 2013; Souza, Rerko, Lin, & Oberauer, 2014), an informative spatial cue presented during the delay interval directs attention toward an item that is no longer visible. Despite the observer only having access to the memory representation of the cued item, the retro-cue nonetheless improves recall of that item relative to non-cued items. The exact mechanisms responsible for the retro-cue benefit are debated (Souza & Oberauer, 2016). One recent study captured the effects of retro-cueing as an elevated amplitude of activity for the cued item relative to non-cued items in a neural population representation (Bays & Taylor, 2018). The increased strength of signal may protect the cued item representation in multiple ways: against passive loss of precision over time due to diffusion of stored values (Schneegans & Bays, 2018); against confusion with other items in memory (Oberauer & Lin, 2017; Schneegans & Bays, 2017); and against disruption by subsequent visual input (Souza & Oberauer, 2016; Tabi, Husain, & Manohar, 2019). Often, changes in the environment, e.g., when visiting a store or driving a car, call for a rapid erasure of outdated information. Such requirements may not be well served by passive forms of information loss but instead require active processes that selectively remove irrelevant representations currently held in memory (e.g., Oberauer, 2001, 2018). This rapid erasure of outdated information subsequently frees up memory resources for the remaining task-relevant items (Souza, Rerko, & Oberauer, 2014) and has been shown to attenuate the effect of load on memory (Souza, Rerko, Lin, & Oberauer, 2014). In addition, several studies have demonstrated that individuals can simply forget information they no longer need when cued to do so (Williams & Woodman, 2012), to the point that any information pertaining to the removed items may become completely irretrievable (Williams, Hong, Kang, Carlisle, and Woodman 2013; see Lewis-Peacock, Drysdale, Oberauer, and Postle 2011, and Stokes 2015 for alternative accounts). These findings motivated the removal hypothesis of the retro-cue effect, which proposes that uncued items are simply removed from memory when a highly predictive retro-cue indicates the to-be-tested item (Souza & Oberauer, 2016). The removal process itself is closely related to - and indeed has been proposed as a component process of - a more general cognitive mechanism referred to as memory updating (Ecker, Oberauer, & Lewandowsky, 2014; Kessler & Meiran, 2006, 2008). Updating tasks typically require an observer to replace existing memory representations when new information becomes available. In effect, items at particular spatial, or temporal, locations are overwritten and the observer must keep track of only the most recently presented information. In practice, this requires an obsolete item to be removed (i.e., resources are deallocated from it), followed by encoding of a new item (i.e., resources are allocated to it). Ecker, Lewandowsky, and Oberauer (2014) demonstrated that a primary predictor of successful memory updating is the ability to efficiently remove outdated information. Intuitively, complete removal of obsolete items ought to facilitate the successful encoding of new information to the same contextual locations. Nevertheless, Ecker and colleagues argued that removal of outdated items is not a particularly easy task in contrast to studies that have claimed information can be wilfully discarded from memory (e.g., Williams et al., 2013; Williams & Woodman, 2012). Notably, most research on updating has used letter and digit stimuli within recognition-based paradigms. Accordingly, the precision of working memory has not been measured in these tasks, raising the possibility that items may have been degraded in fidelity rather than fully removed. Here we investigated how efficiently resources can be deallocated from items that are no longer required, using a modified analogue recall task (Prinzmetal, Amiri, Allen, & Edwards, 1998; Wilken & Ma, 2004). Our working hypothesis was that successful deallocation of resources from an irrelevant item would mean it no longer counted towards the “effective” set size determining precision for the remaining items. Specifically, successful updating was expected to result in an enhancement of memory precision compared to a condition where the amount of information that had to be stored in memory was equal to the total number of presented relevant and irrelevant objects. Conversely, a failure to fully withdraw resources was expected to be seen as a precision cost to other items compared to that ideal. In particular, following successful object removal, the performance ought to be comparable to the precision achieved with a set size equal to the number of relevant items only. Experiment 1 In our first experiment we examined how well participants could update the contents of VWM when presented with a single additional piece of information. In particular, we were interested in whether resources encoding a formerly task-relevant item could be completely reallocated to a new item. In this and subsequent experiments, all updated (i.e., replaced) features became obsolete and were never tested. Method Participants A total of 12 participants (7 females, 5 males; aged 21−45 years, M = 26.4, SD = 6.2) took part in the study having provided informed consent in accordance with the Declaration of Helsinki. Participants all had normal or corrected-to-normal visual acuity and reported having normal colour vision. Stimuli & Apparatus Stimuli were presented on an LCD monitor (45 cm × 28 cm) with a refresh rate of 60 Hz. Participants were positioned 60 cm from the screen with their head supported by a chin and forehead rest. Eye position was monitored online at 1000 Hz using an infra-red eye tracker (Eyelink 1000, SR Research). To establish fixation we presented a central white fixation point (.25° radius) against a grey background. Study arrays consisted of coloured discs (1° radius) that were randomly located at one of four equidistant points around the circumference of an imaginary circle (6° radius) centred on the fixation point, with a rotational offset chosen randomly on each trial. Disc colour was determined by randomly sampling from a colour wheel, defined by a circle in CIELAB space with constant luminance (L*= 50), centre a * = b * = 20, and radius of 60. Test arrays consisted of a circular white annulus (1° radius) acting as a memory cue to indicate the location of the item to be recalled, and a colour wheel (3° radius) centred on the fixation point. Procedure Following eye-tracker calibration, trials commenced with the presentation of a fixation point. Once a stable fixation had been established within 2° of the fixation point, the first study array was presented (1000 ms). This array always contained an initial set of three coloured discs. Participants were instructed to commit each of the three items to memory. A blank retention interval then followed (1000 ms), after which participants were shown one of four types of study arrays (1000 ms; see Figure 1). In the No Update condition the second array remained blank and only required participants to maintain the original three items in memory (Figure 1, orange outline). Subsequently, a target item was selected at random from the first sample array. The purpose of introducing this condition was to ensure that observers would encode first-array colours throughout the experiment. Participants each completed a total of 54 No Update trials. In the Repeat condition, one randomly chosen item was selected from the first array and presented again at the same location (Figure 1, purple outline). Although the second presentation of the item was physically identical to the first, we could not assume it coincided exactly with the colour in memory, so participants were instructed on Repeat trials to remember the most recently presented colour of the repeated item, while still maintaining memories of each non-repeated item. Participants completed a total of 72 Repeat trials, with each location cued equally often (i.e. 24 trials for the repeated item and 24 for each of the non-repeated items). In the Replace condition, a disc with a new randomly chosen colour, which we term the post-replacement stimulus, was presented at the same location as one of the original three items (Figure 1, green outline). Participants were again instructed to remember only the most recently presented colour and told that they would not be tested on the original colour, which we define as the pre-replacement stimulus. As a result, the Replace condition required the participant to update a single item in memory while continuing to maintain representations of the other two items from the first array. Participants completed a total of 72 Replace trials, with each location cued equally often (i.e. 24 trials for the replaced item and 24 for each of the non-replaced items). Finally, in the New condition, a disc with a randomly chosen colour was presented at a new array location that was unoccupied in the first array (Figure 1, blue outline). On these trials participants were told to remember the fourth colour in addition to the original three items. Participants completed a total of 72 New trials, with each item cued equally often (i.e. 18 trials for the item in the second array and 18 for each of the three items in the first array). In total, each participant completed 270 trials in a single one hour session. To avoid any uncertainty over trial types, we tested the potentially confusable Replace and Repeat conditions in two separate blocks, giving participants instructions applicable for each. One block randomly interleaved No Update (27 trials), Replace (72 trials), and New (72 trials); the other block interleaved No Update (27 trials) and Repeat (72 trials). Block order was counterbalanced across participants. Eye tracking was used to monitor fixation. If gaze position deviated by more than 2 dva before onset of the response cue, a message appeared on the screen, and the trial was aborted and restarted with newly randomized colours later in the same block. Analysis We measured recall error across each condition as the angular deviation between the reported and target colours on the colour wheel. The dispersion of recall errors, measured as circular standard deviation, was used to estimate how precisely items were retrieved from memory. We used Bayesian statistics for evaluating the evidence for and against our hypotheses, implemented in JASP (JASP Team, 2020) using the default Jeffreys-Zellner-Siow prior on effect sizes (Liang, Paulo, Molina, Clyde, & Berger, 2008). The Bayes factor compares the predictive adequacy of two competing hypotheses (e.g., alternative and null) and quantifies the change in belief that the data bring about for the hypotheses under consideration (Wagenmakers et al., 2018). For example, BF10 = 5 indicates that the data is five times more likely to occur under the alternative than the null hypothesis. Evidence for the null hypothesis is indicated by BF10 < 1, in which case the strength of evidence is indicated by 1/BF10. When presenting results from Bayesian ANOVAs we report the overall evidence for an effect. This is derived via Bayesian model averaging which averages over all candidate models that contain the effect of interest. We examined effects of the similarity between pre- and post-replacement items (in the Replace condition) by calculating recall variability as a function of the angular difference between the colours on the colour wheel. We did this by first pooling responses across observers (Figure 2A, dark green distribution, bottom panel) and then calculating the circular SD for 19 equally spaced colour distances, based on overlapping bins with widths of 50°. We additionally fit the three component mixture model to the response distributions from each condition (Bays, Catalao, & Husain, 2009; code available at https://bayslab.com/toolbox/). This model assumes a probabilistic mixture of responses distributed (a) around the target colour, (b) around each of the other, non-target colours presented on a trial, and (c) uniformly on the colour wheel. Owing to the small number of trials in individual conditions at the participant level, we fit the model to pooled participant data to obtain more reliable parameter estimates. The mixture model estimates are based on maximum likelihood principles, meaning that this method requires a sufficiently large number of observations to find the true parameter values with arbitrary precision (i.e., consistency property). In particular, simulations conducted using the mixture model have shown that the obtained number of trials at the participant level in Experiment 1 (i.e., <30 trials) is unlikely to provide reliable parameter estimates (https://bayslab.com/toolbox/), making it necessary to estimate parameters using pooled data. To be sure of detecting all possible intrusion (swap) errors, every colour presented on a trial was entered into the mixture model, either as the target or as a non-target. On Replace trials, if the replaced item was cued (i.e., indicated for recall) then the post-replacement colour was the target and the pre-replacement colour was entered as a non-target, while if one of the other items was cued both pre- and post-replacement colours were entered as non-targets. In the New condition, the cued item was the target and the other three colours were entered as non-targets, irrespective of the array in which they were presented. To distinguish between different kinds of intrusion error, we first used the trial-by-trial probability weights from the global mixture fit to classify as swap trials all trials where the posterior probability for the non-target component exceeded target and uniform components. Having identified these trials, we next inferred which non-target colour was most likely to have been reported in place of the target in each trial by examining the absolute distances between the reported colour and each non-target colour. Of particular importance, we identified a pre-replacement intrusion as any swap trial on which the pre-replacement colour was closest to the reported colour. Results The updating task of Experiment 1 was designed to yield clear quantitative predictions for the Replace condition. If the obsolete (pre-replacement) colour was efficiently removed from memory, then the allocation of resources to items should be the same as in the Repeat condition, with an effective set size of three. On the other hand, if the obsolete colour could not be removed and continued to occupy resources in the same way as the other items, the allocation should be the same as in the New condition, i.e. an effective set size of four. Note that, while both the No Update and Repeat conditions had an effective set size of three, the Repeat condition was designed to match Replace also in the delays between stimulus presentation and cue. Our main results are plotted in Figure 2, which shows the pooled response distribution for each condition (Figure 2A; top row first array items; bottom row second array items), along with the corresponding circular SD estimates for the second array probes only (Figure 2B). To address our central hypothesis we began by comparing recall precision between the second array probe conditions, i.e. when the new item (New condition), repeated item (Repeat condition) or post-replacement item (Replace condition) was indicated for report. As the experiment was carefully designed to contrast the precision of second-array probes, precision differences for first-array probes are consistent with multiple hypotheses, some of them unrelated to memory updating. Therefore, their analysis is provided in Supplementary Materials. For this initial analysis, we found that recall variability was greater for new items (dark blue bars) than for repeated items (dark purple bars; δ = -1.20, 95% CI :[-2.15, -0.34], BF10 = 10.07), but was similar in magnitude when compared to post-replacement items (dark green bars; δ = -.27, 95% CI :[-1.00, 0.38], BF10 =.42). However, no conclusive difference could be detected between post-replacement and repeated items (δ = -.53, 95% CI :[-1.34, 0.18], BF10 =.94). Because pre- and post-replacement colours were chosen independently at random, on some proportion of Replace trials the replacement colour could have been very similar to the colour it replaced. To examine the effect of this similarity we plotted recall variability as a function of the angular difference between colours of pre- and post-replacement items. The result is plotted in Figure 2C (filled circles). Group performance in the Repeat condition is shown for comparison (purple region). Values close to zero on the x -axis imply a high degree of colour similarity between the first- and second-array items presented at the replacement location. Perceptually, these instances will mimic a Repeat trial and, indeed, recall variability is very similar between the two conditions when colour similarity is high. However, we observed a substantial increase in variability when pre- and post-replacement items became more dissimilar. To investigate further, we plotted trial-by-trial responses for each observer as a function of colour similarity between pre- and post-replacement items (Figure 2D). This revealed a very clear explanation for why variability was greater for increasingly dissimilar colours. Though the majority of responses were clustered around the (correct) post-replacement colour, we observed a handful of errors that were displaced along the negative diagonal, consistent with observers erroneously reporting the pre-replacement colour. This would tend to inflate variability estimates when pooled with other trials. Mixture model fits further confirmed this interpretation. Our analysis identified a very small number of pre-replacement intrusions (only 10 in total; 3.47% of Replace trials on which the post-replacement item was probed), each coinciding with one of the data points falling along the negative diagonal (indicated by filled circles). We also noted a similar proportion of non-target reports on Replace trials where one of the non-replacement items was probed (3.47% of trials). Critically, when we removed the 10 trials identified as pre-replacement intrusions from analysis, we found post-replacement recall variability to be approximately invariant with colour distance (Figure 2C, unfilled circles). There was now strong evidence for a difference in precision between post-replacement (Figure 2B, lighter shaded bar) and new items (Figure 3; δ = -1.36, 95% CI :[-2.33, -0.31], BF10 = 19.5), and no consistent difference from recall of repeated items (δ = -0.04, 95% CI :[-0.73, 0.63], BF10 = 0.29). Discussion In the first experiment we successfully demonstrated that working memory resources can be efficiently reallocated from obsolete to relevant information. Critically, this was found only after discounting a very small proportion of trials on which observers failed to reallocate resources and reported the to-be-replaced colour. Although recall performance was overall worse on Replace than Repeat trials, an inspection of the trial-by-trial data revealed that this decrement was fully accounted for by a very small number of systematic failures where individuals mistakenly reported a pre-replacement stimulus feature. When influence of those trials was removed, precision estimates were aligned with performance in the Repeat condition where only three items were encoded into memory. The removal of intrusion errors should not be considered a mere data cleaning procedure. Instead, there are two main points to be taken regarding those errors. First, the identification of intrusion errors allowed us to uncover a specific way in which memory updating fails. Second, despite their comparative infrequency, these intrusions had a profound influence on our measure of recall precision. Following their removal, we were able to assess how efficient resource allocation was on the remaining majority of trials. Our results extend and clarify those of Kessler et al. (2015), who reported advantageous change detection accuracy for repeated relative to updated items. Prior to the removal of intrusion errors, the results of our initial analysis revealed a similar pattern of data to that observed by Kessler et al. (2015). This suggests that the accuracy cost for updated items observed in that study may correspond to occasional intrusions of a pre-replacement item, as we observed. The absence of difference between Replace and Repeat provides only partial evidence of successful resource reallocation. An equally important piece of evidence consists of demonstrating that the cost of updating is smaller than committing the additional item to memory. Such a cost would occur if the obsolete information could not be removed from memory and continued to occupy resources, affecting the precision of relevant objects. Although our initial analysis was ambiguous regarding this cost, following the removal of intrusion errors we found strong evidence that the precision cost of updating an existing item was smaller than storing an additional item. Together, comparisons of Replace with New and Replace with Repeat provided converging evidence that on the large majority (>95%) of trials, participants successfully reallocated resources from the obsolete item to its replacement. It remains unclear exactly why individuals occasionally failed to update their memories. One simple explanation is that these errors reflect momentary confusion about the task instructions: the subject may have temporarily forgotten which items they were supposed to remember, or mistaken a Replace trial for a New trial. Another possibility is that replacing only a subset of all encoded items is a cognitively more challenging operation than simply replacing all items. For example, Kessler and Meiran (2008) argued that “partial-set updating”, i.e. changes made to a subset of items in memory, requires a complex series of steps involving the decoupling and substitution of individual item features within each encoding context. In comparison, “whole-set updating” − where all items in memory are updated − is arguably a much simpler process. In this case the entire contents of memory can be discarded and completely new information encoded in its place. Because whole-set updating does not require removal to proceed in an item-wise fashion it is claimed to occur more quickly (Kessler & Meiran, 2008). It is conceivable, then, that the rare persistence of pre-replacement features in memory might be a consequence of individual updating. To test this hypothesis, the next exp eriment required observers to update the entire contents of memory on some trials. Experiment 2 To investigate whole-set updating of working memory, second arrays in Experiment 2 contained three items instead of just one (Figure 4). In other respects the design was very similar to Exp. 1. Method Participants A total of 12 new participants (12 females; aged 21–30 years, M = 25.1, SD = 3.4) took part in the study having provided informed consent in accordance with the Declaration of Helsinki. Participants all had normal or corrected-to-normal visual acuity and reported having normal colour vision. Procedure Study arrays again consisted of coloured discs (Figure 4). The location of each coloured disc was chosen at random from six equidistant locations positioned on an imaginary circle and rotated by a random offset from trial to trial. The No Update condition (54 trials per participant) was identical to Exp. 1. In the Repeat condition (72 trials), all three items in the first array were presented again in the second array. In the Replace condition (72 trials), three new coloured discs were presented at the same locations as the original three items. In this condition participants were instructed to remember the most recently presented colours and that the original items would not be tested. Finally, in the New condition (72 trials), three new coloured discs were presented at new array locations. On these trials participants needed to encode the new items while continuing to maintain the items in the first array. Targets were selected from either array with equal frequency. Participants competed a total of 270 trials in a single one hour session, with sessions blocked in the same way described for Experiment 1. Results For our second experiment, the predictions are essentially the same as Exp. 1: if observers can efficiently replace obsolete items, then only three items should be represented in memory in the Replace condition, making the effective set size the same as in the Repeat condition. If the obsolete items are not removed, the effective set size is six, matching the New condition. The results are plotted in Figure 5. Bayesian repeated measures t-tests showed strong evidence for better performance for repeated items than either post-replacement (δ = -1, 95% CI :[-1.77, -0.29], BF10 = 21.94), or new items (δ = -1.6, 95% CI :[-2.58, -0.69], BF10 = 422). When comparing performance between post-replacement and new items, the data moderately supported the null hypothesis of no difference (δ = -0.13, 95% CI :[-0.65, 0.38], BF10 = 0.33). Accordingly, preliminary analyses again suggested that observers were imperfectly removing pre-replacement features. We next considered the extent to which pre-replacement intrusions might have influenced this result. Despite asking observers to globally update their memory representations, our mixture analysis indicated that pre-replacement intrusions occurred at a somewhat higher rate (6.02% of trials in the Replace condition) to the previous experiment. In addition, in the Replace condition we also observed a number of within-display swaps (3.94%), i.e., reporting a non-target presented in the second array. Subsequent removal of pre-replacement intrusions produced some reduction in the variability estimate for post-replacement features (Figure 5, lighter green bar). Critically, the corrected estimate yielded moderate evidence for better recall of post-replacement than new items (δ = -0.67, 95% CI :[-1.33, -0.07], BF10 = 3.56) as well as weak evidence for worse recall compared to repeated items (δ = -0.59, 95% CI :[-1.23, -0.01], BF10 = 2.32) (Figure 6). Discussion The primary purpose of our second experiment was to examine the possibility that partial-set updating was responsible for updating failures in Exp. 1. In particular, ongoing maintenance of part of the memory content might disrupt a complex sequence of actions required for updating a single item, resulting in the occasional updating failures. To address this we instead asked observers to globally update all items in memory by presenting three entirely new items in the second stimulus array. Specifically, to avoid any deleterious effects of simultaneous memory maintenance on resource reallocation, we asked observers to forget old items and encode new items all at once. We nevertheless observed a similar rate of pre-replacement intrusions between the two experiments. Whole-set updating, then, had no appreciable effect upon how frequently pre-replacement intrusions occurred. Despite the low incidence of intrusions, they had a similarly deleterious effect on variability estimates. Indeed, when removed from the Replace condition data, the estimated variability was again lower than the New condition, consistent with successful deallocation of resources from obsolete items on the large majority of trials, although the fact that variability did not fall as low as the Repeat condition suggests a complete reallocation of resources from all three pre-replacement items may not have been achieved. Such partial resource reallocation is in contrast with findings from Exp. 1 where memory precision in Replace and Repeat conditions coincided. The comparable rates of pre-replacement intrusion in Experiments 1 and 2 suggest that whole-set updating did not benefit the removal of outdated information, as previously suggested Kessler and Meiran (2008). This, along with observed partial resource reallocation, raises the possibility that resources can only be deallocated item-by-item. Theoretical work by Ecker, Oberauer, and Lewandowsky (2014) showed that updating can be explained by a model that must first decouple item-context bindings in an item-wise fashion, regardless of number, before any items can be removed. It has further been argued that item-wise removal requires both time and effort (Fawcett & Taylor, 2008; Oberauer, 2001) to complete. Whereas it takes 50−100 ms to encode a single item into memory (Bays, Gorgoraptis, Wee, Marshall, & Husain, 2011; Vogel, Woodman, & Luck, 2006), removal of the same amount of information is a comparatively slower process, estimated at 500−600 ms (Ecker, Lewandowsky, & Oberauer, 2014). Earlier work using word lists has also provided estimates upward of two seconds for the complete removal of three items (Oberauer, 2001). If a similar amount of time is required for removal of three visual features from working memory, this could explain why we saw incomplete reallocation of resources in Exp. 2. We examined this possibility in our third experiment. Experiment 3 In this experiment we examined whether additional time would allow for more complete removal of obsolete items from memory. We used the same basic updating procedure as before, but introduced a temporal manipulation intended to provide more time to remove obsolete working memory contents following the presentation of the second stimulus array. If the removal of irrelevant items is a time-consuming process, as previously suggested (e.g., Oberauer, 2001), allowing more time should increase the frequency with which removal was successful and consequently reduce the frequency of intrusion errors. As the main focus of this experiment was on intrusion errors, we used only the Replace and New conditions. Method Participants A total of 12 new participants (7 females, 5 males; aged 19−39 years, M = 27, SD = 5.4) took part in the study having provided informed consent in accordance with the Declaration of Helsinki. Participants all had normal or corrected-to-normal visual acuity and reported having normal colour vision. We identified two participants, via the mixture model analysis, that had misunderstood the task. These individuals were found to have reported the pre-replacement instead of the post-replacement colour on almost all trials in the Replace condition. While this potentially indicates an inability to update WM contents, we believe a more likely explanation is that these observers misunderstood the task, since the remaining observers across all experiments showed very low or zero rate of intrusions. We therefore excluded these participants from further analysis. Keeping them in the sample would have led to the exclusion of the majority of their trials as intrusion errors and not meaningfully changed the outcome of the analysis. Procedure The experiment consisted only of Replace and New conditions, which were identical to those in Exp. 2 (Figure 4), except for a variable delay interval followed the offset of the second array of 1, 2, or 4 seconds. Condition and delay interval were randomly interleaved and participants completed two blocks of 144 trials in a single one-hour session. Results The main results from Experiment 3 are plotted in Figure 7. In contrast to our previous two experiments, we found strong evidence that observers recalled post-replacement items more precisely then new items, even before accounting for pre-replacement intrusions (2x3 Bayesian repeated measures ANOVA: BFindusion = 7.13 × 105). We also found strong evidence for an effect of delay (BFindusion = 49.4), though the effect appeared to be driven by poorer recall at the longest delay interval (1 second vs. 2 second: BF10 = 0.24; 2 second vs. 4 second: BF10 = 58.5). We found no evidence for an interaction between condition and delay BFindusion = 0.27). Our mixture analysis did not detect any pre-replacement intrusions in the shortest delay interval. However, intrusion rates in the Replace condition were comparable with previous experiments for both the two-second (5.63%) and four-second (6.88%) condition. The rates of other non-target reports in the same condition were also low but appeared to increase with delay interval, and we found 0%, 3.96%, and 6.67% of within-array swap errors for one-second, two-second, and four-second conditions, respectively. Removal of pre-replacement intrusions reduced estimated SD in both the 2 and 4 s delay condition (Figure 7B, lighter shaded bars) but the overall pattern of statistical results did not change (2x3 Bayesian repeated measures ANOVA: memory condition BFindusion = 1.31 × 108; delay BFindusion = 13.27; interaction BFindusion = 1.29). Discussion Our third experiment investigated whether observers would benefit from having more time to remove items from working memory. We varied the delay interval between the offset of the second array and the probe array to examine this possibility. We found that recall of replacement items was consistently better compared to the recall of new items across all delays, even when the influence of pre-replacement intrusions was left uncorrected. Nevertheless, these results largely discount the conjecture that observers would benefit from more time to remove information from memory. At the longest delay intervals, pre-replacement intrusions continued to occur at a similar low rate to previous experiments. Once these intrusions were removed, variability in the Replace condition was not consistently influenced by delay interval. General Discussion The present work explored whether observers could reallocate their visual working memory resources to accommodate new stimuli and remove obsolete information from memory. We presented one or more new stimuli at previously occupied locations to indicate to participants they should discard and replace information in memory corresponding to those locations. When comparing conditions that were matched in the timing and total number of items presented, the opportunity to replace items was expected to reduce the number of items maintained simultaneously in memory, resulting in more precise recall of the items remaining. Across three experiments, our results indicated that individuals were able to update their working memories efficiently on most trials, reallocating all or most of the resources dedicated to obsolete items to store new information. Importantly, we also observed resource reallocation occasionally failing. Indeed, an important caveat is that the successful updating was revealed only after discounting a very small number of pre-replacement failures, i.e., trials where a participant was cued to report a replacement colour but reported the original colour instead. Despite their infrequency, because responses on these trials were uniformly distributed with respect to the target, their inclusion had a disproportionate effect on variability estimates that tended to obscure the benefit of successful replacement that occurred on the large majority of trials. In order to evaluate the success of memory updating, we took advantage of the well-established set size effect on recall precision. Optimally efficient memory updating would result in an “effective set size” equal to the total number of relevant items after presentation of the second array. Conversely, a complete failure to discard obsolete objects would result in an effective set size equal to the sum of both relevant and irrelevant items. Therefore, a pattern of results indicating optimal memory updating would be signified by both a statistical equivalence in performance to a memory condition with the former set size and a statistical difference from a condition with the latter set size. While theoretically equally import ant, strong evidence as measured by the Bayes Factor for the former result could be harder to achieve in practice than for the latter result, given the asymmetry in how evidence for the null (i.e., absence of difference) and alternative (i.e., difference) accumulate (Keysers, Gazzola, & Wagenmakers, 2020; Stefan, Gronau, Schönbrodt, & Wagenmakers, 2019). Moreover, empirical patterns of recall precision might deviate from this desirable pattern, indicating only partial resource reallocation was achieved. While the results of Exp. 1 suggested complete memory removal is attainable, results of Exp. 2 were more ambiguous, suggesting updating may have been incomplete, either due to resources not being completely withdrawn from the obsolete object or not successfully allocated to the relevant object. We observed considerable consistency in intrusion rates across our experiments, despite attempting to manipulate factors that could interfere with updating of memory. In our second experiment we examined whether global replacement of all items in memory was more efficient than replacing just one (Kessler & Meiran, 2008), while our third experiment examined whether updating would benefit from more time to unbind obsolete items from their encoding context (Ecker, Lewandowsky, & Oberauer, 2014; Ecker, Oberauer, & Lewandowsky, 2014). The relative invariance of intrusion rates across these experiments, and the similarity in recall variability once intrusions were removed, suggests that neither of these manipulations had a significant effect on updating efficiency. The delay manipulation in Exp. 3 was intended to allow varying amounts of time for removal of obsolete information before memory was probed. The hypothesis that pre-replacement intrusions would become more infrequent with delay time was not supported, however one possible explanation is that the disengagement of memory resources from pre-replacement stimuli stopped once the second array disappeared, i.e. once there was no visible stimulus for them to be reallocated to. Although there would be no performance advantage to retaining pre-replacement stimuli on Replace trials (reporting them would be no better than guessing at random with respect to the instructed task) we cannot rule this account out based on our data. However, a recent study that tested whole-set updating of location information provides some evidence against this: Tabi et al. (2021) still observed intrusions of pre-replacement locations despite a much longer post-replacement stimulus duration (4 s) than in our study, suggesting prolonged exposure to replacement stimuli is not sufficient to achieve complete removal. Another possible approach would have been to cue items from the first array that were going to be replaced at varying intervals before the second array appeared (as in Ecker, Lewandowsky, & Oberauer, 2014), which would manipulate the time available for removal independently of encoding replacement information. A simple explanation for the intrusion errors observed in these experiments is that they reflect transient failures of vigilance or attention to the stimuli during presentation of the second array. If an observer did not store the new information in the second array they would report a pre-replacement colour in response to the probe. Another possibility is that participants occasionally mistook a Replace trial for a New trial, and so intentionally retained the pre-replacement colours from the first array in addition to the post-replacement colours. However, we think this is unlikely: the unique locations used for stimuli on a trial were all widely spaced, so we doubt the location of a pre-replacement item in the first array would be recalled so inaccurately as to mistake its replacement at the same location in the second array for a new item at a different location. Even if that mistake were to be made, to result in a pre-replacement intrusion the participant would further have to match the location of the probe to the (putatively) misremembered pre-replacement location rather than to the post-replacement location to which it actually corresponded. We view this particular combination of individually improbable events as a less parsimonious explanation than occasional lapses in encoding the second array. Whatever the cause of the rare intrusions, our results highlight the importance of identifying such contaminant responses, which can have a highly disproportionate influence on variability estimates. In the present case, failing to account for the possibility of pre-replacement intrusion would have led us to almost exactly opposite conclusions to the ones we have reached about the efficiency of updating. As we have emphasized previously (Taylor & Bays, 2018, 2020), mixture models can be a useful statistical tool to de-noise data even if mechanistic interpretations of the mixture components are uncertain. Although a large body of literature demonstrates that people can deliberately erase memory content, understanding the mechanisms by which this occurs has been challenging. Results of studies conducted by Kessler and Meiran (2006, 2008) suggested that object removal involves “dismantling” or “unbinding” the old representations. This idea was more formally conceptualized and implemented in the SOB-CS model of working memory complex span task (Ecker, Lewandowsky, & Oberauer, 2014; Oberauer, Farrell, Jarrold, Pasiecznik, & Greaves, 2012), but the logic applies to other WM tasks (Lewis-Peacock, Kessler, & Oberauer, 2018). Here, active information removal is accomplished by breaking the association between the memory content (e.g., colour) and its context (e.g., spatial position of coloured disk). In particular, removing an object proceeds by cueing the object with its location and then unlearning the association between the object and its location. In SOB-CS this unbinding process is implemented as Hebbian unlearning: unlike Hebbian learning, which forms item-context binding, unlearning simply removes previously formed associations. Previous studies have provided complementary neurophysiological evidence for the processes supporting memory updating, specifically the removal of items from WM and encoding of post-replacement items. Using multivariate decoding of EEG (LaRocque, Lewis-Peacock, Drysdale, Oberauer, & Postle, 2013) and fMRI activity patterns (Lewis-Peacock, Drysdale, Oberauer, & Postle, 2012) during WM maintenance, previous studies have found that presenting a cue indicating which item stored in memory is relevant for the task leads to an attenuation of decoder evidence for an uncued item relative to the cued item, consistent with removal of irrelevant information. More recently, this finding was extended by asking observers not only to remove memory content, but to replace it with new information (Kim, Smolker, Smith, Banich, & Lewis-Peacock, 2020). Similarly to previous studies, the item’s neural representation declined to baseline following the mid-trial instruction to replace that item with a new one. Furthermore the decline in decodability for the pre-replacement item was accompanied by an increase in classifier evidence for the post-replacement item. Together, these studies provide evidence for neural markers of WM content control mechanisms. Future research could aim to identify a neural signature of the occasional failures of resource reallocation that lead to intrusions of pre-replacement items. Research on WM updating has important implications for understanding the etiology and symptoms of psychiatric disorders. In particular, it has been shown that WM updating deficits characterize many psychiatric disorders including depression (Levens & Gotlib, 2010; Meiran, Diamond, Toder, & Nemets, 2011), post-traumatic stress disorder (Moores et al., 2008; Weber et al., 2005), schizophrenia (Galletly, MacFarlane, & Clark, 2007; van Raalten, Ramsey, Jansma, Jager, & Kahn, 2008), and autism (Lieder et al., 2019). In those conditions, a failure to update WM with new information can result in intrusive thoughts and perseverative, maladaptive behavior. Better understanding the mechanisms of WM updating could therefore aid in designing more focused interventions specifically aimed at treating cognitive impairments in those psychiatric populations. Taken together, our study provides evidence for efficient updating of working memory. Despite rare failures to update, our results clearly support the principle that observers can reallocate VWM resources from obsolete memoranda in order to maintain high precision representations of goal-relevant items. Future work could explore how these mechanisms operate in more naturalistic tasks and conditions requiring frequent updating of memory content. Supplementary Material Supplementary Material Declarations Funding This research was funded by the Wellcome Trust [Grant number 106926 to P.M.B.] For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. Availability of data and materials & Code availability Data and analysis code will be made publicly available upon acceptance of this manuscript. All data and code associated with this article can be found at https://doi.org/10.17863/CAM.88835. Figure 1 Procedure of Experiment 1. Note. After encoding an initial set of three coloured items, observers were subsequently shown one of four types of array: a No Update array (orange outline) contained no feature information; a Repeat array (purple outline) contained a single repeated item (here, the yellow disc); a Replace array (green outline) contained a new colour replacing the one previously shown at the same location (here, light blue replaces yellow); a New array contained a new colour at a previously unoccupied location (here, the pink disc). After a delay, a spatial cue was shown and the participant reported the corresponding colour in memory by selection from a colour wheel. Figure 2 Experiment 1 results. Note. a) Participant pooled error distributions. Lighter histograms reflect recall error for first array items, darker histograms for second array items. b) Corresponding circular standard deviation for second array probes only. Bars denote participant-averaged performance across each condition. The lighter green bar indicates group performance following removal of pre-replacement intrusions (as opposed to panel a)). Error bars indicate ± 1 SEM. c) Circular SD as a function of distance between features values presented at the replaced location. d) Trial-by-trial response errors for recall of replacement item feature, plotted as a function of distance between feature values presented at replacement location. Filled circles indicate all swap errors, i.e., intrusion errors and conventional non-target responses. e) Number of non-target reports in the Replace condition. Filled bars denote the number of pre-replacement intrusions. Unfilled bars are the number of conventional non-target reports. Figure 3 Difference in circular standard deviation for second array targets in Exp. 1. Note. Difference between the Replace and Repeat (purple) or New (blue) condition, before (left) and after (right) removing intrusion errors. Positive values indicate performance was better (lower variability) in the Replace condition. Error bars indicate ±1 SEM. Figure 4 Procedure of Experiment 2. Note. After encoding an initial set of three coloured items, observers were subsequently shown one of four types of array: a No Update array contained no feature information; a Repeat array contained three repeated items; a Replace array contained three new colours replacing those previously shown at the same locations; a New array contained three colours at previously unoccupied locations. After a delay, a spatial cue was shown and the participant reported the corresponding colour in memory by selection from a colour wheel. Figure 5 Experiment 2 results. Note. a) Participant pooled error distributions. Lighter histograms reflect recall error for first array items (No Update and New conditions only), darker histograms for second array items. b) Corresponding circular standard deviations for second array probes only. Bars denote participant-averaged performance in each condition. The lighter green bar indicates performance following removal of pre-replacement intrusions. c) Number of non-target reports. Filled bars denote pre-replacement intrusions, unfilled bars conventional non-target reports. Figure 6 Difference in circular standard deviation for second array targets in Exp. 2. Note. Difference between the Replace and Repeat (purple) or New (blue) condition, before (left) and after (right) removing intrusion errors. Positive values indicate performance was better (lower variability) in the Replace condition. Error bars indicate ±1 SEM. Figure 7 Experiment 3 results. Note. a) Participant pooled error distributions. Lighter histograms show recall error for first array items (New condition only), darker histograms for second array items. b) Corresponding circular standard deviations for second-array probes only. Bars denote participant-averaged performance across each condition. The lighter green bars indicate performance following removal of pre-replacement intrusions. Note that no lighter bar is presented for the 1s condition because no non-target responses were identified. c) Number of non-target reports. Filled bars denote pre-replacement intrusions, unfilled bars conventional non-target reports. Conflicts of interest: None. Ethics approval: This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of University of Cambridge. Consent to participate: Informed consent was obtained from all individual participants included in the study. Consent to publication: Not applicable. Authors’ contributions: P.M.B. conceived the idea; P.M.B., R.T., and D.A.-M. designed experiments; R.T. performed the research; R.T. and I.T. analysed data and wrote the initial draft; D.A.-M. provided feedback and I.T. and P.M.B. revised the manuscript. Bays PM Catalao RFG Husain M The precision of visual working memory is set by allocation of a shared resource Journal of vision 2009 9 1 11 10.1167/9.10.7 19761316 Bays PM Gorgoraptis N Wee N Marshall L Husain M Temporal dynamics of encoding, storage, and reallocation of visual working memory Journal of Vision 2011 11 10 6-6 10.1167/11.10.6 Bays PM Husain M Dynamic shifts of limited working memory resources in human vision Science (New York, NY) 2008 321 851 854 10.1126/science.1158023 Bays PM Taylor R A neural model of retrospective attention in visual working memory Cognitive Psychology 2018 100 43 52 10.1016/j.cogpsych.2017.12.001 29272732 Ecker UKH Lewandowsky S Oberauer K Removal of information from working memory: A specific updating process Journal of Memory and Language 2014 74 77 90 10.1016/j.jml.2013.09.003 Ecker UKH Oberauer K Lewandowsky S Working memory updating involves item-specific removal Journal of Memory and Language 2014 July 74 1 15 10.1016/j.jml.2014.03.006 Fawcett JM Taylor TL Forgetting is effortful: Evidence from reaction time probes in an item-method directed forgetting task Memory & Cognition 2008 36 1168 1181 10.3758/MC.36.6.1168 18927035 Fougnie D Suchow JW Alvarez GA Variability in the quality of visual working memory Nature Communications 2012 3 1229 Retrieved 2013-03-05, from http://www.nature.com/ncomms/journal/v3/n11/abs/ncomms2237.html 10.1038/ncomms2237 Galletly CA MacFarlane AC Clark CR Impaired updating of working memory in schizophrenia International Journal of Psychophysiology 2007 63 3 265 274 10.1016/j.ijpsycho.2006.11.004 17234290 Griffin IC Nobre AC Orienting Attention to Locations in Internal Representations Journal of Cognitive Neuroscience 2003 15 1176 1194 10.1162/089892903322598139 14709235 JASP Team JASP (Version 0.14.1)[Computer software] 2020 Retrieved from https://jasp-stats.org/ Kessler Y Meiran N All updateable objects in working memory are updated whenever any of them are modified: Evidence from the memory updating paradigm Journal of Experimental Psychology: Learning, Memory, and Cognition 2006 32 570 585 10.1037/0278-7393.32.3.570 16719667 Kessler Y Meiran N Two dissociable updating processes in working memory Journal of Experimental Psychology: Learning, Memory, and Cognition 2008 34 1339 1348 10.1037/a0013078 18980398 Kessler Y Rac-Lubashevsky R Lichtstein C Markus H Simchon A Moscovitch M Updating visual working memory in the change detection paradigm Journal of Vision 2015 15 9 18 10.1167/15.9.18 Keysers C Gazzola V Wagenmakers E-J Using Bayes factor hypothesis testing in neuroscience to establish evidence of absence Nature Neuroscience 2020 23 7 788 799 10.1038/s41593-020-0660-4 32601411 Kim H Smolker HR Smith LL Banich MT Lewis-Peacock JA Changes to information in working memory depend on distinct removal operations Nature Communications 2020 11 1 6239 10.1038/s41467-020-20085-4 Landman R Spekreijse H Lamme VAF Large capacity storage of integrated objects before change blindness Vision Research 2003 43 149 164 10.1016/S0042-6989(02)00402-9 12536137 LaRocque JJ Lewis-Peacock JA Drysdale AT Oberauer K Postle BR Decoding Attended Information in Short-term Memory: An EEG Study Journal of Cognitive Neuroscience 2013 25 1 127 142 10.1162/jocn_a_00305 23198894 Levens SM Gotlib IH Updating positive and negative stimuli in working memory in depression Journal of Experimental Psychology: General 2010 139 4 654 664 10.1037/a0020283 21038984 Lewis-Peacock JA Drysdale AT Oberauer K Postle BR Neural Evidence for a Distinction between Short-term Memory and the Focus of Attention Journal of Cognitive Neuroscience 2011 24 61 79 10.1162/jocn_a_00140 21955164 Lewis-Peacock JA Drysdale AT Oberauer K Postle BR Neural Evidence for a Distinction between Short-term Memory and the Focus of Attention Journal of Cognitive Neuroscience 2012 24 1 61 79 10.1162/jocn_a_00140 21955164 Lewis-Peacock JA Kessler Y Oberauer K The removal of information from working memory: The removal of information from working memory Annals of the New York Academy of Sciences 2018 1424 1 33 44 10.1111/nyas.13714 29741212 Liang F Paulo R Molina G Clyde MA Berger JO Mixtures of g Priors for Bayesian Variable Selection Journal of the American Statistical Association 2008 103 410 423 10.1198/016214507000001337 Lieder I Adam V Frenkel O Jaffe-Dax S Sahani M Ahissar M Perceptual bias reveals slow-updating in autism and fast-forgetting in dyslexia Nature Neuroscience 2019 22 2 256 264 10.1038/s41593-018-0308-9 30643299 Maxcey-Richard AM Hollingworth A The strategic retention of task-relevant objects in visual working memory Journal of experimental psychology Learning, memory, and cognition 2013 39 760 772 10.1037/a0029496 22845068 Meiran N Diamond GM Toder D Nemets B Cognitive rigidity in unipolar depression and obsessive compulsive disorder: Examination of task switching, Stroop, working memory updating and post-conflict adaptation Psychiatry Research 2011 185 1–2 149 156 10.1016/j.psychres.2010.04.044 20580836 Moores KA Clark CR McFarlane AC Brown GC Puce A Taylor DJ Abnormal recruitment of working memory updating networks during maintenance of trauma-neutral information in post-traumatic stress disorder Psychiatry Research: Neuroimaging 2008 163 2 156 170 10.1016/j.pscychresns.2007.08.011 Oberauer K Removing irrelevant information from working memory: A cognitive aging study with the modified Sternberg task Journal of Experimental Psychology: Learning, Memory, and Cognition 2001 27 948 957 10.1037/0278-7393.27.4.948 11486928 Oberauer K Removal of irrelevant information from working memory: sometimes fast, sometimes slow, and sometimes not at all Annals of the New York Academy of Sciences 2018 1424 239 255 10.1111/nyas.13603 29532484 Oberauer K Farrell S Jarrold C Pasiecznik K Greaves M Interference between maintenance and processing in working memory: The effect of item–distractor similarity in complex span Journal of Experimental Psychology: Learning, Memory, and Cognition 2012 38 3 665 685 10.1037/a0026337 22141748 Oberauer K Lin H-Y An interference model of visual working memory Psychological Review 2017 124 1 21 59 10.1037/rev0000044 27869455 Palmer J Attentional limits on the perception and memory of visual information Journal of Experimental Psychology: Human Perception and Performance 1990 16 2 332 350 10.1037/0096-1523.16.2.332 2142203 Pertzov Y Bays PM Joseph S Husain M Rapid Forgetting Prevented by Retrospective Attention Cues Journal of Experimental Psychology: Human Perception and Performance 2013 39 1224 1231 10.1037/a0030947 23244045 Prinzmetal W Amiri H Allen K Edwards T Phenomenology of attention: I. Color, location, orientation, and spatial frequency Journal of Experimental Psychology: Human Perception and Performance 1998 24 261 282 10.1037/0096-1523.24.1.261 Schneegans S Bays PM Neural Architecture for Feature Binding in Visual Working Memory The Journal of Neuroscience 2017 37 14 3913 3925 10.1523/JNEUROSCI.3493-16.2017 28270569 Schneegans S Bays PM Drift in Neural Population Activity Causes Working Memory to Deteriorate Over Time The Journal of Neuroscience 2018 May 38 21 4859 4869 10.1523/JNEUROSCI.3440-17.2018 29703786 Schneegans S Taylor R Bays PM Stochastic sampling provides a unifying account of visual working memory limits Proceedings of the National Academy of Sciences National Academy of Sciences Section: Biological Sciences 2020 10.1073/pnas.2004306117 Souza AS Oberauer K In search of the focus of attention in working memory: 13 years of the retro-cue effect Attention, Perception, & Psychophysics 2016 78 1839 1860 10.3758/s13414-016-1108-5 Souza AS Rerko L Lin H-Y Oberauer K Focused attention improves working memory: implications for flexible-resource and discrete-capacity models Attention, Perception, & Psychophysics 2014 76 2080 2102 10.3758/s13414-014-0687-2 Souza AS Rerko L Oberauer K Unloading and reloading working memory: attending to one item frees capacity Journal of experimental psychology Human perception and performance 2014 40 1237 1256 10.1037/a0036331 24730737 Stefan AM Gronau QF Schönbrodt FD Wagenmakers E-J A tutorial on Bayes Factor Design Analysis using an informed prior Behavior Research Methods 2019 51 3 1042 1058 10.3758/s13428-018-01189-8 30719688 Stokes MG ‘Activity-silent’ working memory in prefrontal cortex: a dynamic coding framework Trends in Cognitive Sciences 2015 19 394 405 10.1016/j.tics.2015.05.004 26051384 Tabi YA Husain M Manohar SG Recall cues interfere with retrieval from visuospatial working memory British Journal of Psychology 2019 10.1111/bjop.12374 Tabi YA Maio MR Fallon SJ Udale R Dickson S Idris MI Husain M Impact of processing demands at encoding, maintenance and retrieval in visual working memory Cognition 2021 214 104758 10.1016/j.cognition.2021.104758 33984741 Taylor R Bays PM Efficient coding in visual working memory accounts for stimulus-specific variations in recall The Journal of Neuroscience 2018 1018 18 10.1523/JNEUROSCI.1018-18.2018 Taylor R Bays PM Theory of neural coding predicts an upper bound on estimates of memory variability Psychological Review 2020 127 5 700 718 10.1037/rev0000189 32191074 van den Berg R Shin H Chou WC George R Ma WJ Variability in encoding precision accounts for visual short-term memory limitations Proceedings of the National Academy of Sciences 2012 109 8780 8785 10.1073/pnas.1117465109 van Raalten TR Ramsey NF Jansma JM Jager G Kahn RS Automatization and working memory capacity in schizophrenia Schizophrenia Research 2008 100 1–3 161 171 10.1016/j.schres.2007.10.035 18155446 Vogel E Woodman G Luck S The time course of consolidation in visual working memory J Exp Psychol Hum Percept Perform 2006 32 1436 1451 10.1037/0096-1523.32.6.1436 17154783 Wagenmakers E-J Marsman M Jamil T Ly A Verhagen J Love J Morey RD Bayesian inference for psychology. Part I: Theoretical advantages and practical ramifications Psychonomic Bulletin & Review 2018 25 1 35 57 10.3758/s13423-017-1343-3 28779455 Weber DL Clark CR McFarlane AC Moores KA Morris P Egan GF Abnormal frontal and parietal activity during working memory updating in post-traumatic stress disorder Psychiatry Research: Neuroimaging 2005 140 1 27 44 10.1016/j.pscychresns.2005.07.003 Wilken P Ma WJ A detection theory account of change detection Journal of Vision 2004 4 11 11 10.1167/4.12.11 Williams M Hong SW Kang M-S Carlisle NB Woodman GF The benefit of forgetting Psychonomic Bulletin & Review 2013 20 348 355 10.3758/s13423-012-0354-3 23208769 Williams M Woodman GF Directed forgetting and directed remembering in visual working memory Journal of Experimental Psychology: Learning, Memory, and Cognition 2012 38 1206 1220 10.1037/a0027389 22409182 Zhang W Luck SJ Discrete fixed-resolution representations in visual working memory Nature 2008 453 7192 233 235 nature06860 18385672

PMC007xxxxxx/PMC7877060.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101531846 37661 Cardiovasc Eng Technol Cardiovasc Eng Technol Cardiovascular engineering and technology 1869-408X 1869-4098 31098919 7877060 10.1007/s13239-019-00417-2 NIHMS1660378 Article Fabrication of low-cost patient-specific vascular models for particle image velocimetry Falk Katrina L. BS 1 Medero Rafael MS 2 Roldán-Alzate Alejandro PhD 123 1 Department of Biomedical Engineering, University of Wisconsin-Madison, USA 2 Department of Mechanical Engineering, University of Wisconsin-Madison, USA 3 Department of Radiology, University of Wisconsin-Madison, USA Author Statement All authors have contributed to the material presented in this manuscript and this material has not been submitted for publication elsewhere. * All authors use this below address* Alejandro Roldán-Alzate, Department of Radiology WIMR 2476, 1111 Highland Ave, Madison, WI 53705, fax. 608-263-0876 phone. 608-262-1780, roldan@wisc.edu 24 1 2021 9 2019 16 5 2019 11 2 2021 10 3 500507 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Purpose: Particle image velocimetry (PIV), an in-vitro experimentation technique that uses laser based optical measurements to analyze fluid velocity fields, has become increasingly popular to study flow dynamics in various vascular territories. However, it can be difficult and expensive to create patient-specific models for PIV due to the importance of refractive index matching of the model and the fluid. Here we implement and test a new lost-core casting technique to create low-cost, patient-specific models for PIV. Methods: The anonymized patient-specific anatomies were segmented and processed in Mimics/3Matic from 3D Digital Subtraction Angiographies to create patient-specific cores. The cores were 3D-Printed with poly-vinyl alcohol (PVA) and post-processed with a 80:20 water:glue mixture to smooth the surface. Two silicones, Sylgard 184 and Solaris, were used to surround the model and the PVA core was dissolved using warm water. Geometric accuracy was studied with circumferences and surface differences obtained from Computed Tomography scans. Results: Mean geometric differences in circumference along the inlet centerline and the mean surface difference along the aneurysm between the final Silicone Model and the desired STL Print geometry were statistically insignificant (0.6 mm, 95% CI [−1.4, 2.8] and 0.3 mm 95% CI [−0.1, 0.7], respectively). Particle illumination within each model was successful. The cost of one 10 cm x 10 cm x 5 cm model was $69. Conclusion: This technique was successful to implement and test a new, low-cost way of using a 3D-Printed PVA core to create patient-specific in-vitro models for PIV experimentation. Particle Image Velocimetry Patient-Specific Compliant Vascular Model 3D printing pmcIntroduction Particle image velocimetry (PIV) is an in-vitro experimentation technique that uses laser based optical measurements to analyze fluid velocity fields. This is done by optically tracking particle displacements throughout a fluid field in 2D and 3D with multiple high speed cameras. Particles of a variety of size are seeded into the liquid and are illuminated by the laser. In order to track these particles, the working fluid must have the same refractive index (RI) as the in-vitro model. In biomechanics, PIV is helpful to understand complex flow patterns in various vascular territories as well as to validate medical imaging techniques such as magnetic resonance image (MRI), Computed Tomography (CT), or Ultrasound. In order to do this, anatomically realistic in-vitro models are necessary, however, they can be difficult to manufacture, and prohibitively expensive as offered by commercial vendors. Yazdi et al., presented a review on the use of rigid and compliant in-vitro phantoms for hemodynamic analysis in the vascular system with PIV and emphasized the breadth of prototyping, manufacturing, and RI matching techniques currently used [15]. PIV analysis has previously been done with three-dimensional (3D) printed patient-specific rigid models; however, these materials tend to have a high RI relative to water, which makes fluid-material index matching difficult without the use of hazardous chemicals such as sodium iodide. Another limitation of rigid materials is the lack of distensibility, which poorly represents the in-vivo conditions [7]. On the other hand, compliant in-vitro patient-specific models provide more realistic mechanical properties and are most commonly made with Sylgard 184 (Dow Corning, Midland, MI) due to its low RI [15]. A lost-core casting technique has been used to create these compliant models, with cores made from wax, metal, 3D printable materials or even chocolate [1,5,12–14,17]. A variety of limitations such as manufacturing time and geometry complexion for the mold have been improved upon by the more recent use of additive manufacturing (3D Printing) to create the core. However, they still require chemical solutions that alter the properties of water to dissolve the core from the silicone [1,3,8,17]. Polyvinyl alcohol (PVA) is a water-soluble material used in fused deposition modeling (FDM) to print support structures. Although never used as the 3D printed core, PVA was previously used as a wall in a rigid phantom and to polish cores before casting in compliant phantoms [2,4,7,9]. By capitalizing on the purely water-soluble 3D printed material and the applicability of this method for many castable materials, we aim to implement and test a low-cost alternative of using a 3D-printed core with PVA to create patient-specific in-vitro vascular models for analysis with PIV. Methods There are 4 main steps to fabricate these models: anatomy segmentation, 3D-printing of the patient-specific core, silicone casting, and core removal. Gloves are worn during casting and the removal of the core for safety as well as to decrease finger prints on the silicone models which may disrupt PIV results. Anatomy Segmentation Three digital-subtraction angiography studies of intracranial aneurysms were retrospectively obtained after IRB approval. Each image dataset was imported into Mimics (Materialize, Leuven, Belgium) and segmented using global thresholding, region growing and manual selection, if needed, to create a 3D volume (Figure 1). Each volume was imported into 3-Matic (Materialize, Leuven, Belgium) for smoothing of the coarse walls (Laplacian (1st and 2nd Order) smoothing operations) and the addition of barbed connectors to create the patient-specific virtual core (Figure 1). Here the model was left true-to-size or scaled by 2. The surface of the patient-specific cores were exported in stereolithography (STL) format. 3D-Printed Patient-Specific Core All cores were printed out of PVA using an FDM printer (Ultimaker 3+ Extended, Cambridge, MA) with the parameters described in Table 1 and an oblique orientation with respect to the build plate to increase cross sectional area.. The extrusion head was equipped with PVA which produced a separation between the core and supports since they were deposited by two different continuous extrusions. Once finished, the supports were manually separated from the core to create the Unprocessed Core. To improve the surface quality and to hinder silicone from infusing into the model during casting, a thin layer of diluted polyvinyl acetate glue (80:20, glue:water) (Elmer’s Products, High Point, NC) was applied to the core to create the Processed Core. Silicone Casting A 5-sided acrylic box was created based on the specific dimensions of each core geometry. Acrylic pieces were placed and adhered together using hot melt adhesive to create a watertight box. The box was cleaned and the PVA core was secured to the walls at the appropriate height using the hot melt adhesive. Two silicone materials were used to cast the 3D printed models: Sylgard 184 and Solaris (Smooth-On, Macungie, PA). Properties of both silicones can be found in Table 2. Both were prepared according to instructions (Sylgard 184: 10A:1B ratio and Solaris: 1A:1B ratio) and were placed in a vacuum to remove bubbles before pouring over the Processed Core in the acrylic box. The silicone was left to cure for 48+ hours at room temperature. Core removal Once the silicone cured, the acrylic box was carefully disassembled by removing the hot melt adhesive leaving a silicone block (Figure 1). Using small tools, material was removed from the outlets by twisting and breaking off the material. Then warm water was inserted into the core with a 15% infill and continuously run through the PVA core to saturate the PVA. Once saturated, the PVA dissolved leaving a lumen inside of the silicone. Geometric Comparison The Unprocessed PVA Core, Processed PVA Core (with glue), and the final Silicone Model (made of Sylgard 184) were imaged with a Computed Tomography (CT; 0.15 × 0.15 × 0.315 mm resolution) scanner to analyze the geometric accuracy of each step compared to the ideal, STL Print of the core. The DICOM images obtained from the scan of each model was imported into Mimics for segmentation and analysis. Due to the clear attenuation difference between air and PVA or silicone, segmentation was automatic and confirmed by measuring an external connector. A centerline was generated from Mimics for each model and the circumferences along the inlet centerline were exported. Additionally, the surface of the aneurysm from the final Silicone Model was compared to surface of the aneurysm from the STL Print of the Core to observe any geometric differences. Statistical Analysis Qualitative and quantitative differences in the circumference along the centerline of the inlet of the model were used to compare each step in the model making process (Unprocessed PVA core, Processed PVA Core with glue, final Silicone Model) with the ideal, STL Print of the core. The average difference between the final Silicone Model, Processed Core, and Unprocessed Core were reported and statistical significance was defined with a confidence interval of 95%. Qualitative and quantitative comparisons of differences within the aneurysm between the final Silicone Model and the STL Print were also done. Results The fabrication of patient-specific models using a lost-core casting technique with a 3D printed PVA Core, was successful for intracranial aneurysms (both true-size and x2). The same technique was used with both casting materials, Sylgard 184 and Solaris, and the difference in mechanical properties, as stated in Table 2, was palpable. The CT scans of the Unprocessed PVA Core, Processed PVA Core (with glue), and the final Silicone Model (made of Sylgard 184) revealed centerlines throughout the whole anatomy. The circumference along the inlet centerline was chosen for analysis since the patient-specific inlet is not a complete circle in every location. Figure 2 illustrates the circumferences (mm) along the inlet centerline of each model. The STL Print of the core (red triangles) represents the ideal geometry and the Silicone Model represents the final compliant, in-vitro model for PIV. The mean difference in circumference along the inlet centerline between the Final Silicone and STL print was 0.6 mm, 95% CI [−1.4, 2.8]. The mean difference between the Unprocessed Core (without glue) and the STL Print was 2.1 mm, 95% CI [−2.4, 6.6] and the Unprocessed Core (without glue) and the Processed Core was 1.4 mm, 95% CI [−2.5, 5.3]. Absolute geometric differences along the aneurysm of the final Silicone Model and the STL Print range from 0–0.7 mm with a mean difference of 0.3 mm, 95% CI [−0.1, 0.7] (Figure 3). All mean differences in circumference along the inlet centerline and the aneurysm surface comparison were deemed insignificant, except for the difference between the Processed Core and the STL Print (3.5 mm, 95% CI [0.9, 6.2]). As illustrated in Figure 4, the patient-specific model of a real-size intracranial aneurysm made with both Sylgard 184 and Solaris were successfully index matched with a 58:42 glycerol:water mixture. Particles were imaged in the inlet (red box) of both models to show the ability to use these models for PIV. The overall time required to make the models varied based on the size of the anatomy and availability of resources. If segmentation, printing and casting could be done in 2 days, the final model would be completed in 4 days after curing. For instance, the intracranial aneurysm PVA core (Figure 1) required ~5 hours to print. It was also found that PVA rolls which were connected to the printer for a long period of time had failed prints more often than prints done with PVA rolls which were stored in a dry area. This was likely due to the saturation of the PVA rolls. The cost of 3D printed PVA core was less than $5 for each anatomy. As the size of the anatomy varied, so did the size of the entire model and therefore, the amount of silicone needed was the largest factor in the total cost of the model. For the 8 cm x 8 cm x 7 cm (Figure 1) and the 10 cm x 10 cm x 5 cm model (Figure 2) the total cost of the models were estimated to be around $62–69 (Sylgard 184 = ~ $0.128/mL). The Solaris silicone cost less per mL than the Sylgard 184, making those models even cheaper. Discussion This article presents a novel methodology of using 3D printed, patient-specific anatomy cores from PVA to create patient-specific in-vitro models for PIV analysis at low-cost. This technique has been successfully implemented and tested for geometric accuracy and PIV compatibility in patient-specific intracranial aneurysm models derived from a 3D-DSA. Sylgard 184 was chosen to compare for geometric accuracy because of its most extensive use for models in the PIV community 16. This circumference change along the inlet centerline varied between the processing steps. The greatest difference in circumference geometry occurred between the STL Print and the Processed Core, concluding that the addition of 80:20, water:glue made a significant difference to the ideal STL Print of the core. This difference, however, was minimized once the processed core was submerged into the silicone, likely being that the PVA core and water soluble glue absorbed some of the silicone to bring the final Silicone Model geometry back down to the desired STL Print size. Since the mean circumference difference along the centerline between the original, STL Print and the final, Silicone Model, was used to account for the non-circular inlet cross sections, the mean difference between the two in terms of diameter is ~0.1 mm. Although the model making process presented here focuses on two intracranial aneurysms, this technique was also successful in two other anatomical territories derived from MRI (a brain ventricle and a portal vein) (Figure 5). The portal vein silicone model was used for PIV analysis to compare flow patterns at different resolutions with MRI and computational fluid dynamic simulations[11]. Compared to other compliant in-vitro phantoms creation techniques, this method provides a novel way of printing the core anatomy with PVA and dissolving it out of a silicone block with only water. Additionally, the ability to use this technique with a more compliant silicone, such as Solaris, may allow users more structural variability for PIV experimentation. This process is limited by the size of the printer’s build volume, 197 × 215 × 300 mm. Therefore, anatomies larger than this build volume may need to be isolated into sections for this modeling technique. Also, it is important to note that different segmentation or post-processing techniques may produce a varied patient-specific anatomy6,10. The method explained here uses a 3D DSA where contrast has enhanced the vasculature of interest which aids in vessel segmentation and measurement checks with a physical external connector were included to aid in proper segmentation, however, geometric differences due to segmentation may remain. The Solaris silicone required more careful core extraction since the material was soft and more easily damageable by sharp extraction tools. Within the core and casting process, additional manufacturing artifacts such as debris or fingerprints were present but can be avoided by wearing gloves, improving the application of the polyvinyl acetate glue, and covering the silicone mold while curing. Conclusion Overall, this technique demonstrates a lost-core casting method of creating compliant, patient-specific in-vitro vascular models for analysis with PIV. The novelty of this technique includes the use of PVA to 3D print the core anatomy which is both a low-cost material and is easily extracted from the model using warm water. Also, the ability to apply this process with the Solaris silicone in addition to Sylgard 184, provides the opportunity for use of other silicone elastomers for in-vitro PIV experimentation. Acknowledgements Funding for this project was partially supported by a National Institutes of Health K12-DK100022 Grant (Dr. Roldán-Alzate). Figure 1: Model making process is illustrated with a patient-specific intracranial aneurysm. Projections from an anonymized 3D DSA acquisition were selected from an in-house database with IRB approval. Segmentation of the vascular anatomy was done with Mimics, model scaling (x2) and connectors were added with 3-Matic to create the STL Print of the core. The patient-specific vascular core was then 3D printed out of PVA using the Ultimaker 3+ Extended. The supports were carefully removed to create the Unprocessed Core. The 3D printed core is coated in a thin layer of glue to create the Processed Core. The Processed Core submerged in silicone. The final Silicone Model with the PVA core dissolved out with water. Figure 2: Comparison of the geometric circumference along the inlet of the centerline from points A-D in four different model making steps to create an intracranial aneurysm model: STL Print, Unprocessed Core, Processed Core, final Silicone Model. Circumferences were derived from CT images of the physical parts (resolution 0.15×0.15×0.315 mm). Centerlines were created in Mimics. Figure 3: Comparison of geometric difference across the surface of the aneurysm between the final Silicone Model and the ideal STL Print. Spatial differences (mm) are illustrated with the color bar where red represents the greatest difference nearing 0.7 mm. The mean difference across the aneurysm is 0.3 mm, 95% CI [−0.1, 0.7]. Geometry subtractions were done in Mimics. Figure 4: Distortion of checkerboard shown through the compliant intracranial aneurysm silicone models made of Sylgard 184 and Solaris. a) Sylgard 184 with water, b) Sylgard 184 with 58:42 glycerol:water mixture, c) Solaris with water and d) Solaris with 58:42 glycerol:water mixture. Red box highlights the area of the inlet where the particles were illuminated. Figure 5: Final in-vitro compliant silicone models (Sylgard 184) of a patient-specific portal vein and brain ventricle made with a 3D-printed PVA core. Table 1: 3D printer parameters for the Ultimaker 3+ Extended used to create the patient-specific PVA cores. Settings used to print PVA cores on the Ultimaker 3+ Extended Diameter Filament 2.85 mm Printing Temperature 215°C Nozzle Size 0.4 mm Build Plate Temperature 60°C Layer Height 0.1 mm Fan Cooling Speed 50% Infill Density 10% Support Pattern Triangles Printing Speed 35 mm/s Support Placement Touching Buildplate Table 2: Mechanical and physical properties of the sylgard 184 and Solaris silicones. Sylgard 184 Solaris Index of Refraction 1.41 1.41 Modulus (MPa) 1.32–1.84 (Johnston,NASA) 0.14–0.17 (Solaris Data sheet, NASA) Ultimate Tensile Strength (MPa) 5.13–7.07 (Johnston,Dow,NASA) 1.24 (Solaris Data sheet, NASA) Shore Hardness 43.8 15 Conflict of Interest K.L. Ruedinger, R. Medero and A. Roldán-Alzate declare that they have no conflicts of interest to report. References 1. Arcaute K , and Wicker RB . Patient-Specific Compliant Vessel Manufacturing Using Dip-Spin Coating of Rapid Prototyped Molds. J. Manuf. Sci. Eng 130 :051008, 2008. 2. Brunette J , Mongrain R , and Tardif JC . A Realistic Coronary Artery Phantom for Particle Image Velocimetry. J. Vis 7 :241–248, 2004. 3. Buchoux A , Valluri P , Smith S , Stokes AA , Hoskins PR , and Sboros V . Manufacturing of microcirculation phantoms using rapid prototyping technologies. Proc. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. EMBS 2015–Novem:5908–5911, 2015. 4. Büsen M , Kaufmann TAS , Neidlin M , Steinseifer U , and Sonntag SJ . In vitro flow investigations in the aortic arch during cardiopulmonary bypass with stereo-PIV. J. Biomech 48 :2005–2011, 2015.25907548 5. Cao P , Duhamel Y , Olympe G , Ramond B , and Langevin F . A new production method of elastic silicone carotid phantom based on MRI acquisition using rapid prototyping technique. Proc. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. EMBS 5331–5334, 2013.doi:10.1109/EMBC.2013.6610753 6. Despotović I , Goossens B , and Philips W . MRI Segmentation of the Human Brain: Challenges, Methods, and Applications. Comput. Math. Methods Med 2015 :1–23, 2015. 7. Geoghegan PH , Buchmann NA , Soria J , and Jermy MC . Time-resolved PIV measurements of the flow field in a stenosed, compliant arterial model. Exp. Fluids 54 :, 2013. 8. Hütter L , Geoghegan PH , Docherty PD , Lazarjan MS , and Clucas D . Fabrication of a compliant phantom of the human aortic arch for use in Particle Image Velocimetry ( PIV ) experimentation 2 :493–497, 2016. 9. Laumen M , Kaufmann T , Timms D , Schlanstein P , Jansen S , Gregory S , Wong KC , Schmitz-Rode T , and Steinseifer U . Flow analysis of ventricular assist device inflow and outflow cannula positioning using a naturally shaped ventricle and aortic branch. Artif. Organs 34 :798–806, 2010.20964698 10. Ruedinger KL , Rutkowski DR , Schafer S , Roldán-Alzate A , Oberstar EL , Strother C , and Ruedinger K . Impact of image reconstruction parameters when using 3D DSA reconstructions to measure intracranial aneurysms. J. Neurointerv. Surg 0 :1–5, 2017. 11. Rutkowski DR , Medero R , Garcia FJ , and Roldán-Alzate A . MRI-based modeling of spleno-mesenteric confluence flow. J. Biomech , 2019.doi:10.1016/j.jbiomech.2019.03.025 12. Sulaiman A , Boussel L , Taconnet F , Serfaty JM , Alsaid H , Attia C , Huet L , and Douek P . In vitro non-rigid life-size model of aortic arch aneurysm for endovascular prosthesis assessment. Eur. J. Cardio-thoracic Surg 33 :53–57, 2008. 13. Taylor TW , and Yamaguchi T . Three-dimensional simulation of blood flow in an abdominal aortic aneurysm using steady and unsteady computational methods. Asme Bioeng Div Publ Bed., Asme, New York, Ny(Usa), 1992, 22 :229–232, 1992. 14. Yagi T , Sato A , Shinke M , Takahashi S , Tobe Y , Takao H , Murayama Y , and Umezu M . Experimental insights into flow impingement in cerebral aneurysm by stereoscopic particle image velocimetry: transition from a laminar regime. J. R. Soc. Interface 10 :20121031, 2013.23427094 15. Yazdi SG , Geoghegan PH , Docherty PD , Jermy M , and Khanafer A . A Review of Arterial Phantom Fabrication Methods for Flow Measurement Using PIV Techniques , 2018.doi:10.1007/s10439-018-2085-8 16. Yazdi SG , Geoghegan PH , Docherty PD , Jermy M , and Khanafer A . A Review of Arterial Phantom Fabrication Methods for Flow Measurement Using PIV Techniques , 2018.doi:10.1007/s10439-018-2085-8 17. Yip R , Mongrain R , Ranga A , Brunette J , and Cartier R . Development of Anatomically Correct Mock-Ups of the Aorta for PIV Investigations. Can. Des. Eng. Netw. Conf 1–10 , 2004.

PMC008xxxxxx/PMC8149485.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9421525 8603 Gene Ther Gene Ther Gene therapy 0969-7128 1476-5462 33244179 8149485 10.1038/s41434-020-00206-w NIHMS1639137 Article Site-Specific Modifications to AAV8 Capsid Yields Enhanced Brain Transduction in the Neonatal MPS IIIB Mouse Gilkes Janine A (1) Judkins Benjamin L (1) Herrera Brontie N (1) Mandel Ronald J. (2) Boye Sanford L. (3) Boye Shannon E. (4) Srivastava Arun (4) Heldermon Coy D (1) (1) Department of Medicine, University of Florida College of Medicine, Gainesville, Florida, USA (2) Department of Neuroscience, University of Florida College of Medicine, Gainesville, Florida, USA (3) Department of Pediatrics, Powell Gene Therapy Center, University of Florida College of Medicine, Gainesville, Florida, USA (4) Department of Pediatrics, Division of Cellular and Molecular Therapy, University of Florida College of Medicine, Gainesville, Florida, USA Correspondence should be addressed to C.D.H. (coy.heldermon@medicine.ufl.edu), 1600 SW Archer Rd, Campus Box 100278, Gainesville, Florida 32608, USA, Tel: 1-352-273-7497, Fax: 1-352-273-5006 21 10 2020 26 11 2020 8 2021 23 8 2021 28 7-8 447455 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Mucopolysaccharidosis type IIIB (MPS IIIB) is an autosomal recessive lysosomal disease caused by defective production of the enzyme α-N-acetylglucosaminidase (NAGLU) (1). It is characterized by severe and complex central nervous system (CNS) degeneration. Effective therapies will likely target early onset disease and overcome the blood-brain barrier (2). Modifications of adeno-associated viral (AAV) vector capsids that enhance transduction efficiency have been described in the retina (3, 4). Herein, we describe for the first time, a transduction assessment of two intracranially administered Adeno-associated virus serotype 8 variants, in which specific surface-exposed tyrosine (Y) and threonine (T) residues were substituted with phenylalanine (F) and valine (V) residues, respectively. A double-mutant (Y444+733F) and a triple-mutant (Y444+733F+T494V) AAV8 were evaluated for their efficacy for the potential treatment of MPS IIIB in a neonatal setting. We evaluated biodistribution and transduction profiles of both variants compared to the unmodified parental AAV8, and assessed whether the method of vector administration would modulate their utility. Vectors were administered through four intracranial routes: six sites (IC6), thalamic (T), intracerebroventricular (ICV) and ventral tegmental area (VTA) into neonatal mice. Overall, we conclude that the IC6 method resulted in the widest biodistribution within the brain. Noteworthy, we demonstrate that GFP intensity was significantly more robust with AAV8 (double Y-F + T-V) compared to AAV8 (double Y-F). This provides proof of concept for the enhanced utility of IC6 administration of the capsid modified AAV8 (double Y-F + T-V) as a valid therapeutic approach for the treatment of MPS IIIB, with further implications for other monogenic diseases. Introduction One critical component of effective brain delivery is widespread delivery of therapeutic products throughout the CNS, as most of these disorders diffusely affect the CNS. Therefore, identifying a gene transfer injection method and vector that will diffuse through the brain more effectively is critical for any disorder that results from somatic mutations that are ubiquitous. The gene therapy results obtained in models of an array of autosomal recessive lysosomal disorders that predominantly affect the brain indicate that more widespread transduction and protein expression will result in better therapeutic effects in vivo. In addition, identifying the most efficient transduction regimen should also result in a safer therapeutic strategy. The requirement for efficient viral transduction is especially true of lysosomal disorders whose most devastating symptoms are the result of widespread brain pathology. Given that currently available recombinant adeno-associated viral vectors (rAAV) do not globally transduce the brain after peripheral injection, there are two basic parameters that can be modified to improve distribution of brain transduction: improving the efficiency of vector transduction via capsid alterations and/or choosing the most efficient route of injection. Recent studies aimed at improving our understanding of the cellular roadblocks affecting the efficiency of AAV transduction, revealed that the ubiquitin–proteasome pathway plays an essential role in AAV2 intracellular trafficking (4, 5). This is mediated at least in part by epidermal growth factor receptor protein tyrosine kinase (EGFR-PTK). Additionally a host cell protein, FK506-binding protein 52 (FKBP52) in its tyrosine-phosphorylated form, prevents viral second-strand DNA synthesis, resulting in inhibition of AAV-mediated transgene expression (6). Both the viral capsid and human FKBP52 protein can be phosphorylated by EGFR-PTK (7). This effect results in substantial numbers of ubiquitinated virions being recognized and targeted for proteosomal degradation on their way to the nucleus, and inefficient second strand synthesis, thus leading to inefficient nuclear transport. Phosphorylation prior to ubiquitination can occur at tyrosine, serine or threonine residues. Therefore, substitution of surface exposed tyrosine or threonine residues on AAV2 capsids was undertaken to allow the vectors to escape ubiquitination and subsequent degradation. Importantly, mutagenesis of highly conserved exposed tyrosine residues (Y444F, Y500F or Y730F) on AAV2 capsids enhanced transduction up to 10-fold in HeLa cells and 30-fold in mouse liver (8). Since then, single or combined tyrosine mutants of AAV2 have been successfully tested in vitro in fibroblasts and mesenchymal stem cells (9) and in vivo in murine hepatocytes (10), and the retina (3). Improved transduction of mouse skeletal muscle was also obtained with tyrosine mutants of AAV8 in the lungs (11) and in the skeletal muscle by AAV6 vectors (12). In our previous studies, we compared neonatal intracranial administration of AAV5, −8, −9 and –rh10 and concluded that AAV8 was superior to AAV5, −9 and rh10 in its ability to foster robust and widespread transduction within the MPS IIIB brain (13, 14). Consistent with previous studies, we also showed that AAV8 expressed a preference for neurons and astrocytes when injected in neonates. To improve upon the therapeutic capacity of AAV8, we capitalized on the improved transduction efficiency of capsid modified vectors in somatic organs. Use of AAV8 capsid tyrosine mutants in the brain of MPS IIIB animals has thus far not been assessed. In particular, based on its efficiency in a retinal model, we selected AAV8 (double Y-F) and AAV8 (double Y-F+T-V) for further analysis. To address the variable of the most efficient injection route to transduce the CNS, the aforementioned modified rAAV8 vectors were administered either via an intraparenchymal six site (IC6) route, an intracerebroventricular (ICV) injection, intra- thalamic (TH), and ventral tegmental area (VTA) methods. As the six site method, which we have tested previously is more invasive, requiring six burr holes, our goal is to identify an alternate less invasive method which would lead to the same level of global brain biodistribution. The four ventricular spaces in the brain are filled with CSF, which bathes the brain and spinal cord and protects these structures from injury. As this represents a promising route to achieve global brain coverage, several groups have attempted to exploit the ventricular system as a therapeutic approach. However, this approach displayed mixed results, often as a consequence of animal age and AAV serotype. Similarly, direct intra-parenchymal administration of rAAV has also been used to achieve widespread transduction. The inclusion of the thalamic and VTA injection sites are based on the wide spread projection pattern of these anatomical areas. The thalamus projects to the entire cerebral cortex while the VTA projects to the frontal cortex and basal ganglia. Since lysosomal enzymes can be functionally transported between cells via mannose-6-phosphate receptors, these small injection sites may affect large areas of the brain. To address these issues in the context of the MPS IIIB model, we performed a comprehensive analysis of two capsid modified AAV8 variants compared to parental AAV8 when administered via IC6, ICV, TH or VTA methods to identify the most efficient vector and optimal administration route. We observed a clearly superior transduction area and intensity with the IC6 AAV8 (double Y-F + T-V). We anticipate these results will contribute to clinical approaches by identifying the optimal gene delivery vector and method of delivery for treatment of MPS IIIB. Results Capsid Modification Results in Varying Effect on Brain Transduction We first sought to establish whether any differences existed between the different capsid modified AAV8 variants compared to the parental unmodified AAV8. We have previously shown that intracranial six site administration (IC6) of AAV8 lead to near global brain biodistribution within the CNS. We therefore initially assessed for changes in transduction efficiency as a consequence of capsid modification via IC6 administration. We qualitatively investigated four structural areas of functional significance within the brain for transduction. Surprisingly, we saw that the AAV8 (double Y-F) modified vector resulted in inefficient cellular transduction compared to unmodified AAV8. However, use of AAV8 (double Y-F+T-V) resulted in robust transduction of the cortex and hippocampus, with good transduction of the thalamus and cerebellum (Figure 1). To gauge the difference in GFP intensity as a consequence of capsid modification and mitigate the effects of GFP saturation in florescence signaling, we analyzed the cortical, hippocampal, thalamic and cerebellar regions of mid-sagittal brain sections of animals treated with each vector using a near-infrared dye. When data was normalized to the lowest expressing vector, AAV8 (double Y-F), we saw that use of AAV8 (double Y-F+T-V) resulted in the highest GFP intensity levels compared to unmodified AAV8 and AAV8 (double Y-F), (2612 a.u. vs. 1420 a.u. p<0.01 and 492 a.u. p<0.0001, respectively) (Figure 2). Interestingly, we also noted that unmodified AAV8 was superior to AAV8 (double Y-F) (p<0.05) These observations reflect a substantial difference in uptake kinetics and hints at some differences in trafficking kinetics. Taken together, we show that AAV8 (double Y-F+T-V) >AAV8> AAV8 (double Y-F), therefore, all subsequent experiments focused on the use of AAV8 (double Y-F+T-V). Global Brain Biodistribution is Achieved by Modulating rAAV Delivery Route As IC6 based vector administration represents a relatively invasive procedure, we next sought to establish whether comparable global brain biodistribution could be similarly achieved using alternate methods, namely intracerebroventricular (ICV), Thalamic (TH) and Ventral Tegmental Area (VTA), compared to IC6 with AAV8 (double Y-F+T-V) delivery (Figure 3). Each method was previously demonstrated to be a well-tolerated, effective method to achieve widespread brain biodistribution. Both ICV and TH methods represent bilateral vector administration in only two injection sites, whereas, the VTA method utilizes a single injection into the parenchyma. Three months after rAAV injection, four spatially distinct and relatively equidistant areas were selected for qualitative and quantitative assessment of GFP expression. Using the Allen Brain Atlas, we estimated the relative anatomical locations, in millimeters, medial to lateral, to be −4.2 (section 1), −3.72 (section 2), −2.72 (section 3) and −1.72 (section 4), respectively. Overall percentage area with GFP expression for all four sections was initially compared using MPS IIIB mice. Compared to the IC6 method, no other method resulted in global tissue penetration and biodistribution. Of the three other methods, ICV delivered rAAV showed relatively confined spread to the cortical area above the hippocampus, hippocampus, and moderate spread to the thalamus and cerebellum, although lesser penetration into other tissue sections was observed. The TH delivered similar overall GFP percentage area to ICV. VTA method was far lower by comparison. In the IC6 treated group, AAV8 (double Y-F+T-V) transduction is observed throughout the cortex, hippocampus, caudate putamen, inferior and superior colliculus and to a lesser degree, the thalamus and cerebellum. In the cortex, robust expression was observed throughout all layers associated with somatomotor and somatosensory cues, primarily layer V. Robust expression was also observed in all layers of the frontal cortex and in orbital areas of the pre-frontal cortex. Both sensory and motor related areas of the superior colliculus exhibited high GFP expression. In the hippocampus, widespread GFP expression was observed in all layers. In the thalamus, areas proximal to the hippocampus were transduced (Figure 3). Further, resulting vector transduction of the cerebellum was modest and resulted in GFP expression in what appeared based on morphologic assessment to be purkinje neurons, but not interneurons. Quantitatively, we see diminishing but substantial GFP expression in all tissue sections, overwhelmingly exhibited via the IC6 method, from approximately twenty-one percent total GFP positive area in tissue section one, down to approximately six percent in section four. It is important to note that overall, no other method of delivery achieves comparable levels of GFP biodistribution (IC6 vs ICV and TH p<0.05 for both and IC6 vs VTA, p<0.01 for all, Figure 4). Taken together, we conclude that IC6 rAAV administration is still the method of choice to foster global brain biodistribution. Systemic Transduction from CNS Delivery of rAAV is Observed As transport across the blood brain barrier of some rAAV vectors has previously been reported, we sought to determine whether this phenomenon would also be apparent in the MPS IIIB model. We therefore assessed various somatic organs including the heart, liver, muscle, kidney and spleen of AAV8 (double Y-F+T-V) treated MPS IIIB animals. We assessed for this phenomenon in IC6 and ICV treated animals as parenchymal transport may be less likely to occur than transport out of the ventricular spaces. Interestingly, we observed rAAV transduction into the heart and liver of treated animals with both methods, but did not observe muscle transduction (Figure 5). Compared to the IC6 method, ICV administration of AAV8 (double Y-F+T-V) resulted in a higher perceived degree of organ transduction. Discussion In the present study we investigated the impact of administration of AAV8 and capsid-mutant AAV8 vectors, AAV8 (double Y-F) and AAV8 (double Y-F+T-V), on the efficiency of brain transduction. To maximize therapeutic potential, we also investigated four brain administration methods: IC6, ICV, TH and VTA, for their effect on modulating global biodistribution. This is the first study that reveals the efficiency of capsid mutated AAV8 vectors in the central nervous system. The efficiency of rAAV transduction is dependent on multiple steps involving virus–host cell interactions, which include binding to cellular receptors, overcoming intracellular barriers that limit nuclear accumulation of the virus and the conversion of single-stranded viral genomes to double-stranded forms (15). As previously noted, the capsid is an essential element that influences both the extracellular events related to the recognition of specific receptors and intracellular processes affecting the trafficking and uncoating. Thus, the capsid plays an essential role in the cellular tropism, transduction kinetics, and intensity of efficiency of transgene expression (6, 16). Modulating these properties can improve both the effectiveness and safety of gene therapy. Tyrosine, serine or threonine phosphorylation serves as a signal for ubiquitination of intact AAV particles, leading to subsequent targeting for the proteasome-mediated vector degradation before reaching the nucleus. In this context, mutation of capsid tyrosine and threonine residues is predicted to allow the vectors to escape the proteasome degradation pathway and thus promote more vector genome delivery to the nucleus and more effective transgene expression. The increase in transduction efficiency gained from using capsid mutated vectors has been demonstrated in several disease models (3, 11, 12, 17, 18). We have previously shown that IC6 administration of AAV8 lead to global biodistribution within the brain of neonatal MPS IIIB mice (13). To investigate the potential benefit of using capsid modified vectors, we first selected available vectors based on AAV8, namely AAV8 (double Y-F) and AAV8 (double Y-F+T-V), for comparison. Our results showed a significant and robust difference in transduction efficiency among these different vectors (Figure 1 and Figure 2). Although, AAV8 (double Y-F+T-V) emerged superior to AAV8 and AAV8 (double Y-F), we were surprised to find that AAV8 (double Y-F) performed worse than AAV8. Since both AAV8 (double Y-F) and AAV8 (double Y-F+T-V) are mutated at the same residues, it can be concluded that the T-V substitution in AAV8 (double Y-F+T-V) plays an important role in modulating intracellular trafficking in neural cells. In comparing the effectiveness of the method of administration in fostering global brain biodistribution, we found that the IC6 method was far superior to the ICV, TH and VTA methods (Figure 3 and Figure 4). In subjective review of the images for the various injection methods, MPS IIIB mice appeared to have higher transduction than WT with the both the AAV8 (double Y-F+T-V) and with native AAV8, in agreement with our prior publications also showing higher transduction with AAV8 in MPS IIIB than WT (14, 19). We cannot rule out the possibility that some of this effect may be related to differing total volume of injection. The maximum injection volume for our Hamilton syringe is 4 ul and we were uncomfortable injecting more than this in any one site due to concerns for displacement of tissue and inducing further artifact by repeated needle placement at a site in order to reload the syringe and increase volume. We therefore adjusted concentrations of vector to fit the same total vector copy number/injection method in the volume we could comfortably inject with one syringe load for each site per method. Given the unique distribution profiles associated with each method, we postulate that a disease specific method of administration should be considered. For example, we saw that ICV method resulted in very high transduction of the hippocampus, and parts of the cortex and thalamus (Figure 3). This method of administration may be more effective in the treatment of Alzheimer’s, where severe pathology first affects parts of the cortex and the hippocampus. Similarly, we saw that a single VTA injection resulted in localized deposition in the mid-brain. This application may be better suited to the treatment of Parkinson’s disease which affects cells in a localized region, the substantia nigra. Further, consistent with previous findings, we also demonstrate brain to systemic spread of AAV8 (double Y-F+T-V) which was more pronounced when administered via ICV delivery (Figure 5). In conclusion, the present study is the first to analyze the transduction capacity of AAV8 capsid tyrosine/threonine modified vectors and their impact on brain transduction in the MPS IIIB model. As successful treatment of MPS IIIB will require global vector biodistribution and tissue penetration, we also sought to identify an optimal method of delivery which would maximize therapeutic potential. Taken together, our data revealed that IC6 administration of an AAV8 (double Y-F+T-V) vector results in enhanced biodistribution of transgene expression in the CNS. It is expected that when combined with a codon optimized NAGLU this will result in increased therapeutic benefit, although the modest transport of vector into systemic circulation may warrant use of immunomodulatory agents. Materials and Methods Mice The congenic C57BL/6 NAGLU-deficient mouse strain was a kind gift from Elizabeth Neufeld (UCLA) by way of Mark Sands (Washington University, St. Louis, MO), and was maintained and expanded by strict sibling mating (20). Wild type (+/+), heterozygous (−/+) (subsequently referred to as “Control”) and mutant (−/−). Genotyping was done on tissue of newborn mice (P2–3) by enzyme assay (21) or NAGLU exon 6 and neomycin insertion cassette PCR. All animal studies were performed in accordance with guidelines of the University of Florida Institutional Animal Care and Use Committee. AAV Constructs Recombinant AAV2 plasmids pseudotyped with capsid proteins from AAV8 were produced, purified, and titered at the University of Florida Powell Gene Therapy Center Vector Core Laboratory (Gainesville, FL) as previously described (22). Vector titer was determined by dot blot assays, diluted, aliquoted and stored at −80°C until use. Site-directed mutagenesis of surface-exposed tyrosine residues on AAV2 capsids has been described recently (6). Similar strategies were used to generate AAV serotype 8 vectors containing tyrosine to phenylalanine mutations (23). Vector preparations were produced using the plasmid co-transfection method. Each of the three single stranded rAAV vectors, AAV8, AAV8 (double Y-F) and AAV8 (double Y-F+T-V), express humanized green fluorescent protein (hGFP) driven by the hybrid cytomegalovirus enhancer/chicken beta-actin promoter and were kind gifts from Shannon Boye at the University of Florida. Treatments All treatments were performed in genotyped pups at 3–4 days of age and were well tolerated. For each comparison, 3–5 control and 3–5 MPS IIIB neonatal mice were injected by one of four methods; intracranial six site (IC6, N=5), Intracerebroventricular (ICV, N=3), Thalamic (TH, N=3) or Ventral Tegmental Area (VTA, N=3). Neonates were cryoanesthetized prior to and during treatment and were then placed on a warming pad after treatment, before being returned to their mothers. All treatments were well tolerated. Intracranial rAAV-GFP was administered using the following coordinates determined by ruler: IC6 – bilateral frontal (from bregma: 2 mm lateral and 1 mm posterior, 1.5 mm deep), bilateral temporal (from bregma: 3 mm lateral and 3 mm posterior, 2.5 mm deep) and bilateral cerebellar (from lambda: 1 mm lateral and posterior, 1.5 mm deep); TH – (bilateral injections from bregma: 4 mm lateral, 1 mm posterior and 3 mm deep); ICV – needle was placed perpendicular to the skull surface bilaterally, (0.25 mm lateral to the sagittal suture and 0.50–0.75 mm rostral to the neonatal coronary suture, 2 mm deep)(17); and VTA – (unilateral injection from bregma: 4.2 mm lateral and 4.mm deep; modified from Wolfe and colleagues (18). All injections were conducted by hand through the skull using a 32-gauge Hamilton syringe (Narishige Int., East Meadow, NY). All mice received a total of 1.4 × 1010 vector genomes in 4–12 microliter volumes over 1–3 minutes. Histological Procedures Animals were sacrificed three months after vector infusion. Mice were euthanized with 100 μl of Ketamine (120 mg/kg)/Xylazine (16 mg/kg) cocktail followed by thoracotomy. Transcardial perfusion with 1xPBS followed by fresh ice-cold 4% paraformaldehyde in 1xPBS solution followed. Brains were harvested, post fixed for 3 h in 4% paraformaldehyde at 4°C, followed by overnight incubation in 20% sucrose in 1xPBS at 4°C. One brain hemisphere was then embedded in O.C.T (Triangle Biomedical Sciences, Durham, NC) and rapidly frozen in a 2-methyl-butane/dry ice bath. Sagittal sections were cut to a thickness of 20 μm and stored in a cryoprotective solution at −80°C until use. Quantitation of GFP Quantitation of GFP positive area was conducted using the Scanscope FL instrument (Aperio Technologies, Vista, CA). Analysis was conducted using accompanying ImageScope software and the Positive Pixel Count FL v1 algorithm. Use of the tuning feature allowed for maximal capture of GFP based on pixel intensity. The minimum intensity was set between 0.2 and 0.22, and maximum intensity was set to 1. Regions of interest were demarcated using the Allen Reference Atlas (ARA) as a neuroanatomical reference. GFP positive area was determined using the average from three independent tests performed by three readers in a blinded manner. To reliably assess differences in GFP intensity as a consequence of AAV capsid modification, the Odyssey Infrared Imaging system (Li-Cor, Lincoln, NE), was utilized. Briefly, mid-sagittal brain sections of AAV8, AAV8 (double Y-F) or AAV8 (double Y-F+T-V) vector treated animals were incubated overnight with GFP antibody (1:2000, Abcam, Cambridge, MA; Cat. #: ab290) in 1xPBS/0.01% TBS-T/10% NDS/1% BSA buffer. To visualize GFP, donkey anti-rabbit 680 DR was used (1:5000, Li-Cor Biosciences, Lincoln, NE, Cat. # 926–68073). Sections are then mounted to slides and allowed to dry overnight followed by clearing with Xylene. Slide mounted sections were again air dried overnight and scanned using the Odyssey system the following day. Analysis was conducted using the Image Studio Lite version 4.0 software (Li-Cor Biosciences, Lincoln, NE), Statistical Analysis Sample sizes for groups were based on previously observed effect size in intensity of 4 and GFP percent area of 3 between AAV8 in control and MPS IIIB. In order to identify an effect size of 3.14 with alpha error of 0.05 and power of 0.8, a group size of 3 is required. GraphPad Prism 6 was used for statistical analysis. Two-tailed student’s t-test was used for unpaired data. Brown-Forsythe test was performed to confirm equal variance and Tukey’s test was used to correct for multiple comparisons for ANOVA analysis. For comparisons with unequal variance a log transform was used to normalize the data and reduce heterogeneity. Bar graphs are shown as mean ± SEM. Probability p < 0.05 was considered statistically significant. Animals were allocated to injection method by sequential selection of alternating method for each virus of animals randomly selected from each genotype. Allocation of areas of analysis for histology was carried out in a blinded fashion but the sample preparation and analysis itself was not blinded. From the intention-to-treat animals, those with evidence of vector extrusion during injection were excluded from analysis. Acknowledgements This work was supported by the Gatorade Trust through funds distributed by the University of Florida, Department of Medicine and by K085141-01 and NIH/NINDS R01NS102624 (CDH), R01EY024280 (SEB) and R01HL-097088, R01GM-119186, and R21 EB-015684 (AS). We would like to express appreciation to Andrew Kolarich for assistance with performance of neonatal intracranial injections. Figure 1. Assessment of transduction efficiency of AAV8, AAV8 (double Y-F) and AAV8 (double Y-F+T-V) when injected via the IC6 method. Three month old MPS IIIB mice were assessed for GFP expression and tissue penetration into the cortex, hippocampus, thalamus and cerebellum. Low magnification images of the entire mid-sagital section are shown below regional high magnification images. Images obtained using ScanScope FL. Cortex, thalamus and cerebellum (20x), scale bar = 100 μm; hippocampus (8x), scale bar = 300 μm. Figure 2. Use of AAV8 (double Y-F+T-V) results in superior GFP intensity levels. Brains of three month old MPS IIIB animals injected with AAV8, AAV8 (double Y-F) or AAV8 (double Y-F+T-V) via the IC6 method were collected and processed. Cortical, hippocampal, thalamic and cerebellar limits of sagittal tissue sections were delineated and differences in mean GFP intensity levels across the regions for each animal and vector injected were assessed using the Image Studio Lite software and quantitated using One-way ANOVA. Data were normalized to AAV8 (double Y-F) GFP intensity. *p<0.05; **p<0.01, $p<0.0001; n=4 Figure 3. The route of AAV administration impacts tissue penetration and biodistribution. AAV8 (double Y-F+T-V) was injected into neonatal MPS IIIB animals via the IC6, ICV, TH and VTA methods. Brains were extracted at three months, sectioned into four structurally unique, relatively equidistant sections, in millimeters from midline −4.2 (section 1), −3.72 (section 2), −2.72 (section 3) and −1.72 (section 4), respectively. and assessed for differences in tissue penetration and GFP biodistribution as a consequence of route of administration. Images obtained using ScanScope FL. Scale bar = 2mm. Figure 4. The IC6 method of administration results in the most widespread GFP biodistribution within the brain. Using three month old MPS IIIB animals, differences in GFP biodistribution as a consequence of method of administration was quantitatively assessed in each of the four tissue sections; as well as, cumulatively as indicated in the figure legend. Data were analyzed by two-way ANOVA. Data represented as mean ±SEM. *p<0.05; **p<0.01, #p<0.001, $p<0.0001; n=5 for IC6 and n=3 each for ICV, TH and VTA. Figure 5. CNS administration of AAV8 (double Y-F+T-V) results in somatic transduction of organs. Three months after IC6 and ICV routes of AAV administration, somatic organs were harvested and assessed for presence of GFP. Both injection methods result in preferential transduction of heart and liver, indicating vector entry to systemic circulation. 20X images obtained with ScanScope FL. Scale bar = 300μm. Conflict of Interest: CDH, RJM and AS declare stock ownership and co-founding of Lacerta Therapeutics which develops gene therapy for CNS diseases. AS also holds several issued US Patents on AAV vectors that have been licensed to various gene therapy companies, and is a co-founder of Nirvana Therapeutics. SEB and SLB declare stock ownership and co-founding of Atsena Therapeutics which develops gene therapies for ocular diseases. References 1. Di Natale P , Murino P , Pontarelli G , Salvatore D , Andria G . Sanfilippo B syndrome (MPS III B): altered residual alpha-N-acetylglucosaminidase activity in an unusual sibship. Clinica chimica acta; international journal of clinical chemistry. 1982;122 (2 ):135–43.6809360 2. Gilkes JA , Heldermon CD . Mucopolysaccharidosis III (Sanfilippo Syndrome)- disease presentation and experimental therapies. Pediatric endocrinology reviews : PER. 2014;12 Suppl 1 :133–40.25345095 3. Petrs-Silva H , Dinculescu A , Li Q , Min SH , Chiodo V , Pang JJ , High-efficiency transduction of the mouse retina by tyrosine-mutant AAV serotype vectors. Molecular therapy : the journal of the American Society of Gene Therapy. 2009;17 (3 ):463–71.19066593 4. Douar AM , Poulard K , Stockholm D , Danos O . Intracellular trafficking of adeno-associated virus vectors: routing to the late endosomal compartment and proteasome degradation. Journal of virology. 2001;75 (4 ):1824–33.11160681 5. Duan D , Yue Y , Yan Z , Yang J , Engelhardt JF . Endosomal processing limits gene transfer to polarized airway epithelia by adeno-associated virus. J Clin Invest. 2000;105 (11 ):1573–87.10841516 6. Zhong L , Li B , Jayandharan G , Mah CS , Govindasamy L , Agbandje-McKenna M , Tyrosine-phosphorylation of AAV2 vectors and its consequences on viral intracellular trafficking and transgene expression. Virology. 2008;381 (2 ):194–202.18834608 7. Qing K , Hansen J , Weigel-Kelley KA , Tan M , Zhou S , Srivastava A . Adeno-associated virus type 2-mediated gene transfer: role of cellular FKBP52 protein in transgene expression. Journal of virology. 2001;75 (19 ):8968–76.11533160 8. Zhong L , Li B , Mah CS , Govindasamy L , Agbandje-McKenna M , Cooper M , Next generation of adeno-associated virus 2 vectors: point mutations in tyrosines lead to high-efficiency transduction at lower doses. Proc Natl Acad Sci U S A. 2008;105 (22 ):7827–32.18511559 9. Li M , Jayandharan GR , Li B , Ling C , Ma W , Srivastava A , High-efficiency transduction of fibroblasts and mesenchymal stem cells by tyrosine-mutant AAV2 vectors for their potential use in cellular therapy. Hum Gene Ther. 2010;21 (11 ):1527–43.20507237 10. Markusic DM , Herzog RW , Aslanidi GV , Hoffman BE , Li B , Li M , High-efficiency transduction and correction of murine hemophilia B using AAV2 vectors devoid of multiple surface-exposed tyrosines. Molecular therapy : the journal of the American Society of Gene Therapy. 2010;18 (12 ):2048–56.20736929 11. Martini SV , da Silva AL , Ferreira D , Gomes K , Ornellas FM , Lopes-Pacheco M , Single tyrosine mutation in AAV8 vector capsid enhances gene lung delivery and does not alter lung morphofunction in mice. Cell Physiol Biochem. 2014;34 (3 ):681–90.25171090 12. Qiao C , Zhang W , Yuan Z , Shin JH , Li J , Jayandharan GR , Adeno-associated virus serotype 6 capsid tyrosine-to-phenylalanine mutations improve gene transfer to skeletal muscle. Hum Gene Ther. 2010;21 (10 ):1343–8.20497037 13. Gilkes JA , Bloom MD , Heldermon CD . Mucopolysaccharidosis IIIB confers enhanced neonatal intracranial transduction by AAV8 but not by 5, 9 or rh10. Gene therapy. 2015. 14. Gilkes JA , Bloom MD , Heldermon CD . Preferred transduction with AAV8 and AAV9 via thalamic administration in the MPS IIIB model: A comparison of four rAAV serotypes. Mol Genet Metab Rep. 2016;6 :48–54.27014573 15. Zolotukhin S , Potter M , Zolotukhin I , Sakai Y , Loiler S , Fraites TJ Jr. , Production and purification of serotype 1, 2, and 5 recombinant adeno-associated viral vectors. Methods. 2002;28 (2 ):158–67.12413414 16. Zolotukhin S , Byrne BJ , Mason E , Zolotukhin I , Potter M , Chesnut K , Recombinant adeno-associated virus purification using novel methods improves infectious titer and yield. Gene therapy. 1999;6 (6 ):973–85.10455399 17. Gholizadeh S , Tharmalingam S , Macaldaz ME , Hampson DR . Transduction of the central nervous system after intracerebroventricular injection of adeno-associated viral vectors in neonatal and juvenile mice. Human gene therapy methods. 2013;24 (4 ):205–13.23808551 18. Cearley CN , Wolfe JH . A single injection of an adeno-associated virus vector into nuclei with divergent connections results in widespread vector distribution in the brain and global correction of a neurogenetic disease. J Neurosci. 2007;27 (37 ):9928–40.17855607 19. Gilkes JA , Bloom MD , Heldermon CD . Mucopolysaccharidosis IIIB confers enhanced neonatal intracranial transduction by AAV8 but not by 5, 9 or rh10. Gene therapy. 2016;23 (3 ):263–71.26674264 20. Li HH , Yu WH , Rozengurt N , Zhao HZ , Lyons KM , Anagnostaras S , Mouse model of Sanfilippo syndrome type B produced by targeted disruption of the gene encoding alpha-N-acetylglucosaminidase. Proc Natl Acad Sci U S A. 1999;96 (25 ):14505–10.10588735 21. Marsh J , Fensom AH . 4-Methylumbelliferyl alpha-N-acetylglucosaminidase activity for diagnosis of Sanfilippo B disease. Clin Genet. 1985;27 (3 ):258–62.3921297 22. Sen D , Gadkari RA , Sudha G , Gabriel N , Kumar YS , Selot R , Targeted modifications in adeno-associated virus serotype 8 capsid improves its hepatic gene transfer efficiency in vivo. Human gene therapy methods. 2013;24 (2 ):104–16.23442071 23. Li C , Diprimio N , Bowles DE , Hirsch ML , Monahan PE , Asokan A , Single amino acid modification of adeno-associated virus capsid changes transduction and humoral immune profiles. Journal of virology. 2012;86 (15 ):7752–9.22593151

PMC008xxxxxx/PMC8595848.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0330500 7472 Sci Total Environ Sci Total Environ The Science of the total environment 0048-9697 1879-1026 34487902 8595848 10.1016/j.scitotenv.2021.150006 NIHMS1739876 Article Soil microbial community and abiotic soil properties influence Zn and Cd hyperaccumulation differently in Arabidopsis halleri Kushwaha Priyanka a Neilson Julia W. a Maier Raina M. a* Babst-Kostecka Alicja ab* a Department of Environmental Science, The University of Arizona, Tucson, AZ, 85721, USA b W. Szafer Institute of Botany, Polish Academy of Sciences, Department of Ecology, Lubicz 46, 31-512 Krakow, Poland * Both authors contributed equally to this work Corresponding author: Alicja Babst-Kostecka, ababstkostecka@arizona.edu CRediT authorship contribution statement: Priyanka Kushwaha: Investigation, Formal analysis, Writing - original draft. Julia W Neilson: Conceptualization, Formal analysis, Funding acquisition, Writing - review & editing. Raina M Maier: Conceptualization, Funding acquisition, Writing - review & editing. Alicja Babst-Kostecka: Conceptualization, Methodology, Investigation, Formal analysis, Data curation, Supervision, Funding acquisition, Writing - original draft. 16 9 2021 30 8 2021 10 1 2022 10 1 2023 803 150006150006 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Soil contamination with trace metal(loid) elements (TME) is a global concern. This has focused interest on TME-tolerant plants, some of which can hyperaccumulate extraordinary amounts of TME into above-ground tissues, for potential treatment of these soils. However, intra-species variability in TME hyperaccumulation is not yet sufficiently understood to fully harness this potential. Particularly, little is known about the rhizosphere microbial communities associated with hyperaccumulating plants and whether or not they facilitate TME uptake. The aim of this study is to characterize the diversity and structure of Arabidopsis halleri rhizosphere-influenced and background (i.e., non-Arabidopsis) soil microbial communities in four plant populations with contrasting Zn and Cd hyperaccumulation traits, two each from contaminated and uncontaminated sites. Microbial community properties were assessed along with geographic location, climate, abiotic soil properties, and plant parameters to explain variation in Zn and Cd hyperaccumulation. Site type (TME-contaminated vs. uncontaminated) and location explained 44% of bacterial/archaeal and 28% of fungal community variability. A linear discriminant effect size (LEfSe) analysis identified a greater number of taxa defining rhizosphere microbial communities than associated background soils. Further, in TME-contaminated soils, the number of rhizosphere-defining taxa was 6-fold greater than in the background soils. In contrast, the corresponding ratio for uncontaminated sites, was 3 and 1.6 for bacteria/archaea and fungi, respectively. The variables analyzed explained 71% and 76% of the variance in Zn and Cd hyperaccumulation, respectively; however, each hyperaccumulation pattern was associated with different variables. A. halleri rhizosphere fungal richness and diversity associated most strongly with Zn hyperaccumulation, whereas soil Cd and Zn bioavailability had the strongest associations with Cd hyperaccumulation. Our results indicate strong associations between A. halleri TME hyperaccumulation and rhizosphere microbial community properties, a finding that needs to be further explored to optimize phytoremediation technology that is based on hyperaccumulation. Graphical Abstract metal accumulation microbial diversity plant growth promoting bacteria pseudometallophyte soil metal contamination trace metal(loid) element pmc1. INTRODUCTION Anthropogenic activities such as mining and smelting significantly contribute to the local accumulation of harmful trace metal(loid) elements (TME) levels in the environment. There is a global legacy of contaminated mine tailings that are eroded and transported by wind and water into nearby ecosystems. Smelting additionally results in TME distribution by releasing metal-rich particles into the atmosphere, which are then deposited downwind into local habitats (Balabane et al., 1999). The sustainable management of contaminants associated with mining and smelting of metal ores is regarded as a worldwide challenge that the mining industry faces (Virgone et al., 2018). Establishing a lasting vegetation cover on contaminated sites is considered the best and most permanent way to minimize wind and water erosion and to stabilize mining impacted soils in situ (Mendez & Maier, 2008; Ali et al., 2013). Problematically, seed germination and plant growth are often inhibited under TME-rich conditions. Only a limited number of vascular plant species have evolved TME tolerance and are able to survive and reproduce in such environments (Baker, 1987). The majority of TME tolerant plants are labeled “excluders”, because their tolerance is based on minimizing metal(loid) uptake by the roots and limiting transport to shoots. Yet, certain species and populations are able to accumulate TME into above-ground tissues. Among these accumulators, a rare group called “hyperaccumulators” allocates extraordinarily large amounts of a given TME to their foliage, without showing any toxicity symptoms (Baker, 1981; Reeves & Baker, 2000). Importantly, the capacity for metal hyperaccumulation can differ among genotypes of the same plant species (Babst-Kostecka et al., 2018). Hyperaccumulators provide an opportunity to study plant adaptation to extreme environments and they also have potential to be applied in the development of phytoremediation technologies, especially where metal removal or recovery is desired. There is increasing evidence that TME hyperaccumulation depends on a complex set of interactions among soil properties including TME bioavailability, expression of detoxification genes and metal transporters, as well as the plant-associated microbiome (Thijs et al., 2017; Asad et al., 2019; Trivedi et al., 2020). Importantly, it is known that plants enrich the rhizosphere with organic root exudates that in turn selectively attract and stimulate the growth of certain microbial taxa (Honeker et al., 2019b; Trivedi et al., 2020). Accordingly, different plant species may have a different rhizosphere microbiome when grown in the same soil, but also plants of the same species may harbor similar microbial communities in different soils (Miethling et al., 2000). Individual plant genotypes further differentiate their rhizosphere communities to best support the physiology, fitness, and improvement of plant responses to environmental stress (Bressan et al., 2009; Micallef et al., 2009; Yeoh et al., 2017; Trivedi et al., 2020). Although several studies have reported on these plant-microbial interactions in hyperaccumulating plants, particularly little is known about factors that lead to TME hyperaccumulation in non-metalliferous sites, where the environmental impact of this phenomenon has remained largely neglected (Lodewyckx et al., 2002; Li et al., 2007; Lopez et al., 2017). The rhizosphere-associated microorganisms, including bacteria, archaea, and fungi, are thus very important for their host and can facilitate plant performance and ecosystem functions in many ways (Mendes et al., 2013). In addition, there is evidence that they can enhance the TME-accumulation capacity of plants, thereby increasing TME content in shoot tissues (Farinati et al., 2009; Farinati et al., 2011; Muehe et al., 2015). Therefore, gaining deeper insight in these interactions across diverse soil types (metalliferous and non-metalliferous) is an important step towards further discerning the hyperaccumulation processes (Rosatto et al., 2019). Arabidopsis halleri (L.) O’Kane and Al-Shehbaz is considered a model species for its ability to tolerate and hyperaccumulate excessive quantities of zinc (Zn) and cadmium (Cd; Roosens et al., 2008). It can tolerate high TME concentrations in contaminated soils (metallicolous [M] plant populations), but also thrives under natural conditions on uncontaminated soils (non-metallicolous [NM] plant populations). A broad quantitative variation in tolerance and hyperaccumulation of Zn and Cd has been observed in A. halleri across this range of environments (Bert et al., 2000; Bert et al., 2002; Talke et al., 2006; Meyer et al., 2010; Stein et al., 2017; Babst-Kostecka et al., 2018). For Zn and Cd, hyperaccumulation has been defined as over 3000 and 100 mg kg−1, respectively (van der Ent et al. 2013). Previous work has shown that while Zn and Cd tolerance levels are usually higher in M populations, NM populations on uncontaminated soils accumulate metals more efficiently than M populations on TME-contaminated soils. Thus, extremely high concentrations are found in A. halleri shoots, even when the element is present only at low concentration in the soil, indicating active TME foraging (Dietrich et al., 2019). We hypothesize that the associated soil microbiome in this case helps increase both plant TME accumulation capacity and plant fitness in A. halleri. In this study, we characterized A. halleri rhizosphere-influenced and background soil microbiomes in locations from TME-contaminated and uncontaminated field sites. We selected four plant accessions from Southern Poland, for which different TME hyperaccumulation capacities have recently been reported (Babst-Kostecka et al., 2018; Dietrich et al., 2021). These data were combined with multiple physicochemical soil variables and host-plant traits to test for plant-soil-microbe associations. Based on this setup, following questions are addressed: (1) Are differences in A. halleri Zn and Cd hyperaccumulation capacities associated with variation in rhizosphere-influenced microbial community structure, diversity, and richness? (2) Are there specific bacterial, archaeal, or fungal rhizosphere phylotypes that associate with different hyperaccumulation patterns, plant accessions, and/or TME soil contamination? (3) Do the microbial phylotypes differentiating A. halleri rhizosphere-influenced and background communities differ between soils from TME-contaminated vs. uncontaminated sites? 2. MATERIAL AND METHODS 2.1 Sampling and climatic data Four sites with naturally-occurring A. halleri were sampled in Southern Poland in late May 2018 (Table 1 and Fig. S1). Two were TME-contaminated sites, defined here as metalliferous (M), at low altitude in the Olkusz region (M_PL22 and M_PL27). Two were uncontaminated, defined here as non-metalliferous (NM); a sub-alpine location at the northern foothills of the Tatra Mts (NM_PL35) and a low altitude location in Niepołomice Forest (NM_PL14). The M sites differed in their history of industrial activity and source of TME contamination. Site M_PL22 was in the vicinity of the Zn smelter of the Bolesław Mine and Metallurgical Plant near Olkusz and site M_PL27 was near an open-pit mine that was closed in 1912 in Galman. Three replicates of both A. halleri rhizosphere-influenced and surrounding background soil were collected at each M and NM site. Rhizosphere-influenced (hereafter referred to as rhizosphere) samples were collected by first cutting away the aerial parts of the plant, followed by excavating soil at depth 0 to 15 cm to expose the plant roots. Roots were excised using sterile instruments, placed into a sterile plastic bag, and shaken to separate and collect ~ 50 g of soil that adhered to the roots for DNA analysis (Solís-Dominguez et al., 2012). An additional ~ 50 g of soil was sampled from the top 15 cm of the soil profile next to the A. halleri roots for geochemical analysis and placed into a plastic bag. To collect background samples at each site, three 5 × 5 m quadrats that were devoid of A. halleri vegetation were marked. For each quadrat, five ~ 50 g samples were taken at 10–15 cm depth devoid of visible plant roots (one from each corner and the center) and composited (Reimann et al., 2008). Subsamples for DNA analysis were taken as described by Kushwaha et al. (2021). All samples for DNA analysis were immediately placed on ice, transported back to the lab, and stored at −80°C until DNA extraction. Samples for geochemical analysis were stored at room temperature. After collecting the rhizosphere samples, roots and corresponding A. halleri shoots were sampled for elemental composition analysis. Shoots and roots were washed, dried at 80°C, and stored at room temperature. Note, that to avoid clonal replicates of A. halleri plants, the distance between samples was at least 5m. Monthly precipitation and air temperature data for the period 1997–2016 were obtained from meteorological databases at the Polish Institute of Meteorology and Water Management – National Research Institute (IMGW-NRI). Specifically, data were used from the meteorological station nearest to each of the four study sites (all within 15 km): Igołomia, Kościelisko-Kiry, Maczki, and Olewin meteorological stations for NM_PL14, NM_PL35, M_PL22, and M_PL27, respectively. Three bioclimatic variables were generated from the monthly temperature and precipitation data, averaged from 1997–2016: mean annual temperature (°C), annual precipitation (mm), and length of growing season (days). 2.2 Chemical analyses of plant material and geochemical analyses of soil For elemental composition analysis, 0.5 g of dry-ground plant material was mixed with 10 ml of HNO3 (69–70%) and HClO4 (70–72%) 4:1 v/v, left for 24h, and mineralized at 290°C (FOSS Tecator Digestor Auto). Total Zn, Cd, Pb, Cu, Fe K, Na, Mg, and Ca concentration was determined using flame or graphic furnace atomic absorption spectrometry (AAS; Varian AA280FS, AA280Z, Agilent Technologies, Santa Clara, USA). Phosphorus concentration was determined by the vanadium-molybdenum method of Barton (1948) using 0.5 g of dry-ground material dissolved in HClO4, mineralized at 290°C and mixed with vanadium-molybdenum mixture and water. Absorbance at 490 nm were read using a Hach-Lange DR3800 spectrophotometer. The Zn, Cd and Pb translocation factors (TF) were determined as the ratio between metal concentration in shoots and roots. Only rhizosphere soil samples were analyzed for pH, electrical conductivity (EC), total nitrogen (TN), total organic carbon (TOC), total carbon (TC), total inorganic carbon (TIC), NO2−-N, NO3−-N, NH4+-N, available P (P-Olsen), total Zn, Cd, Pb, Cu, Fe, Mg, Ca, K and Na; bioavailable Cd, Pb, Zn, and soil texture. Samples were sieved at 2 mm and dried. Soil pH (ISO 10390) and EC (PN-ISO 11265) were measured in 1:5 (w:v) water suspensions with a Hach HQ40D meter. Total N was determined using the Kjeldahl method; soil was digested in H2SO4 with Kjeltabs (K2SO4 + CuSO4•5H2O; Foss Tecator Digestor Auto) followed by distillation on a Foss Tecator Kjeltec 2300 Analyzer Unit. Total organic C, TC and TIC was determined with a dry combustion analyzer Leco RC-612 (ISO 10694). To analyze NO2−-N, NO3−-N, and NH4+-N concentrations, soil samples were shaken in water for 1 h (1:10, w:v), filtered through cellulose acetate membrane syringe filters (Huang & Schoenau, 1998) and anions in the extracts were determined with an ion chromatograph Dionex ICS-1100, while NH4+-N was determined with Dionex DX-100. P-Olsen was measured with an ion chromatograph (Dionex ICS-1100, Thermo Fisher Scientific) following soil extraction with 0.5 M NaHCO3. In order to determine total Zn, Cd, Pb, Cu, Fe, Mg, Ca, K and Na concentrations, ground samples were digested in hot concentrated HClO4 (FOSS Tecator Digestor Auto) at 288 – 292°C. Extracted elements were analyzed by flame or graphic furnace AAS as described above. Bioavailable fractions of Zn, Cd and Pb were assessed using Diffusive Gradients in Thin-films (DGT) method following the protocol provided by Dietrich et al. (2021). Soil texture was determined through a combination of sieving and sedimentation (ISO 11277). 2.3 Extraction of nucleic acids, amplicon sequencing and data processing DNA was extracted from 0.5 g soil using the FastDNA Spin Kit for Soil™ (MP Biomedicals, Solon OH, USA) as modified by Kushwaha et al. (2021). Soil samples were thawed on ice prior to extraction. Extracts were purified using DNeasy PowerClean Pro Cleanup kit (Qiagen, Hilden, Germany) to remove inhibitors and the DNA was quantified using a Qubit 2.0 Fluorometer and double stranded DNA (dsDNA) high sensitivity assay kit (Invitrogen, Carlsbad, California, USA). All DNA extraction steps were performed with negative control samples (blanks) containing only reagents. Bacterial/archaeal 16S rRNA gene primers 515F/806R and fungal internal transcribed spacer (ITS) primers ITS1f-ITS2 were used for paired-end amplicon sequencing of DNA extracts as described in Walters et al. (2016). The purified amplicons from all the samples were pooled in equimolar concentrations and sequenced with a paired-end read length of 2 × 150-bp on the Illumina MiSeq platform. The DNA library preparation and sequencing runs were conducted by the Microbiome Core at the Steele Children’s Research Center, University of Arizona. Raw reads were demultiplexed using the idemp tool (https://github.com/yhwu/idemp) and bioinformatics analysis was conducted using the DADA2 pipeline (Callahan et al., 2016). Demultiplexed reads were trimmed to the same length of 140 bases for the forward and reverse reads. The paired-end reads were merged using the default overlap of at least 12 bases and then grouped into amplicon sequence variants (ASVs). After removing poor quality and chimeric ASVs, a total of 1,336,123 and 3,353,095 sequence reads remained for the 16S rRNA and ITS genes respectively, with an average of 51,389 ± 23,065 (16S rRNA) and 128,965 ± 81,943 (ITS) reads per sample. Taxonomy identities were assigned to bacterial/archaeal and fungal ASVs using the SILVA (Quast et al., 2013) and UNITE ITS (Nilsson et al., 2018) databases, respectively. After removal of the contaminants from the ASV tables through comparisons of samples and the blanks, 11,665 and 6,166 ASVs were retained for analysis of bacterial/archaeal and fungal communities, respectively. Taxonomy tables were normalized using a cumulative-sum scaling approach prior to conducting statistical analysis (Paulson et al., 2013). The raw sequencing data for the 16S rRNA gene and ITS obtained in this study have been submitted to the NCBI BioProject number: PRJNA706064. 2.4 Statistical analyses Microbial richness (number of observed ASVs), Shannon diversity index, and community dissimilarity were determined using the vegan package (Oksanen et al., 2008). Bray-Curtis distance was used to calculate community dissimilarity and ordination plots were visualized using non-metric multidimensional scaling (NMDS). The differences in richness metrics and Shannon diversity index across site type (M vs. NM), location (NM_PL14, NM_PL35, M_PL22, and M_PL27), and sample type (rhizosphere vs. background soil) were tested using Wilcoxon and Kruskal-Wallis tests. The differences in microbial community compositions between these groups were examined using nested permutational multivariate analysis of variance (PERMANOVA; Anderson, 2001). The factor location was nested within site type, and sample was nested within location. Location and sample were considered as random factors. Further, a linear discriminant effect size (LEfSe) analysis (Segata et al., 2011) was performed to identify indicator microbial taxa most likely to explain the differences between i) M and NM sites and ii) rhizosphere vs. background soils from both M and NM sites (see http://huttenhower.sph.harvard.edu/galaxy/root). A logarithmic cutoff value of linear discriminant analysis (LDA) > 2.0 was applied for LEfSe analysis. Additionally, a similarity of percentages (SIMPER) analysis was performed to determine which ASVs contributed the most to the average dissimilarity in microbial community structure between rhizosphere and background soil samples (iterations=1000). SIMPER was conducted separately for samples from M and NM sites. Only ASVs with permutation p-values < 0.05 are reported. The LEfSe method identifies statistically significant microbial indicators across each group and it weights the uniqueness of the taxon rather than its overall abundance, whereas SIMPER analysis reports the specific ASVs that contribute most to the average dissimilarity between groups and weights the abundance of individual taxa. Non-parametric Kruskal-Wallis analysis was used to test for differences among the four sites with respect to abiotic and biotic soil properties, Zn and Cd shoot concentrations, and root-to-shoot translocation factors. Partial Least Square (PLS) regression was used to investigate the extent to which geography, climate, as well as biotic and abiotic soil and plant parameters explained variation in Zn and Cd hyperaccumulation. PLS is particularly well suited for situations where the number of predictors exceeds the number of observations (Carrascal et al., 2009), as is the case in our study. The PLS was implemented using function plsreg2 for multivariate cases from the plsdepot package in R (Sanchez & Sanchez, 2012). For each element, the analyses were run on two blocks of variables: a matrix of 95 predictors and a matrix of two responses (i.e., Zn shoot concentration and ZnTF for the Zn hyperaccumulation trait; and Cd shoot concentration and CdTF for the Cd hyperaccumulation trait). The predictors included: six geographic and climatic variables, nine elemental plant shoot concentrations, as well as 25 geochemical and 55 biological properties of corresponding rhizosphere samples. The biological properties included: i) microbial diversity metrics, e.g., richness, Shannon diversity index, and NMDS axes scores, ii) relative abundance of indicator microbial taxa likely to explain the differences between rhizosphere vs. background soil samples (rhizosphere taxa with LDA > 4.0 as per LEfSe analysis), and iii) relative abundance of microbial taxa that contributed to the average dissimilarity in microbial community structure (permutation p-values < 0.05 as per SIMPER) between rhizosphere and background soil samples from M and NM sites. A heatmap of the relative abundance of taxa identified as significantly associated with Zn and Cd hyperaccumulation was generated using ggplot2 in R. It is noted that geographic and climatic data were represented by a single value per population and that each value was replicated for all samples from the same sampling site. In the PLS analyses, the optimal number of components was determined by leave-one-out cross validation. Cumulative R2 (%) values are provided according to the retained number of components. These values reflect the explanatory power of the components for all dependent variables (cumulative R2Y) and for all explanatory variables (cumulative R2X). Predictors that contributed most to the underlying variation in metal hyperaccumulation were identified based on their “Variable Importance in Projection” (VIP) scores. Accordingly, variables with VIP scores greater than 0.8 were considered critical in a given PLS regression model (Farrés et al., 2015). Effect direction and intensity of each variable are specified by the sign and the absolute value of the corresponding standardized and scaled regression coefficients. All statistical analyses were performed using R 3.6.0 (R Core Team, Vienna, Austria) and XLSTAT. 3. RESULTS Geographic and climatic data for the sampling sites, elemental plant shoot concentrations, and geochemical properties of rhizosphere soil samples are provided in Supplementary Table S1. 3.1 General characterization of plant Zn and Cd hyperaccumulation All populations, independent of their edaphic origin, accumulated significantly more Zn (Fig. 1A, 1B) and Cd (Fig. 1E, 1F) in shoot than in root tissues. For Zn, the mean Znshoot concentration was on average twice as high in M than in NM plants (Table S1). Individual values ranged from 2013 mg kg−1 (NM_PL35) to 13,254 mg kg−1 (M_PL27), with only two NM_PL35 samples not reaching the Zn hyperaccumulation threshold (3000 mg kg−1; van der Ent et al. 2013). Interestingly, Znshoot concentrations in NM_PL14 plants did not differ from M populations (Fig. 1A), despite ~80-fold lower total and bioavailable Zn soil concentrations at the NM_PL14 site (Fig. 1D; Table 1). Further, bioavailable Zn soil concentrations were significantly higher at M sites when compared to NM sites (Fig. 1D). For Cd, M plants had on average nine-fold higher Cdshoot concentrations compared to NM plants. While all M plants hyperaccumulated Cd, none of the NM plants met the hyperaccumulation threshold for this metal (100 mg kg−1; van der Ent et al. 2013). All NM and M plants had translocation factors (ZnTF and CdTF) that exceeded one (Fig. 1C, 1G). Overall, Zn and Cd translocation factors were higher in M compared to NM populations; however, the difference was only significant for Cd translocation in NM vs. M_PL27. Similar to Zn soil concentrations, bioavailable Cd soil concentrations were significantly higher at M sites when compared to NM sites; in addition, the NM_PL35 had significantly higher Cd soil concentrations than NM_PL14 site (Fig. 1H). 3.2 Microbial community diversity and structure When rhizosphere and background soil samples were analyzed together, neither bacterial/archaeal nor fungal richness were significantly different across the four study sites (Fig. 2A, Table S2). In contrast, community ordination analysis showed that site type (M vs. NM) and location (NM_PL14, NM_PL35, M_PL22, M_PL27) explained 26% and 18%, respectively, of the variation in bacterial/archaeal community composition (p = 0.001), as well as 18% and 10% of the variation in fungal community composition (p = 0.001) (Fig. 2B). Overall, in this analysis 44% and 28% of the community variability for bacteria/archaea and fungi, respectively, could be explained by the combination of site type and location. Wilcoxon test revealed no significant variation for the richness and Shannon diversity index for microbial communities between rhizosphere and background soil (Table S2). 3.3 Key microbial taxa differentiating rhizosphere and background soil samples LEfSe and SIMPER analyses were conducted to identify defining taxa that differentiate the microbial communities of M and NM soils and their respective rhizosphere and background soil communities. LEfSe weights the uniqueness of a taxon within a community, whereas SIMPER weights the relative abundance. Results showed that M sites had four times as many unique bacterial/archaeal ASVs and almost twice the number of fungal ASVs than were found in NM sites (Table 2). As a result of this difference, the rhizosphere and background soil communities were analyzed separately for the M and NM sites. Overall, the analyses showed a higher number of microbial taxa defining the rhizosphere communities than the background soil communities (Table 2). Further, this difference was more pronounced in M sites for which the rhizosphere: background ratio of unique taxa was almost 6 for both bacteria/archaea and fungi. The corresponding ratio for NM sites, was 3 and 1.6 for bacteria/archaea and fungi, respectively (Table 2). The threshold on the logarithmic LDA score for discriminative features was set to 4, to evaluate the microbial taxa that were the most prominent in defining the differences between rhizosphere vs. background soils and M & NM sites. Accordingly, seven out of the 40 bacterial/archaeal taxa were identified as the top rhizosphere taxa at M sites, whereas rhizosphere taxa at NM sites had only one out of the 24 taxa with LDA > 4.0 (Figure S2; Table 2). The top bacterial/archaeal rhizosphere taxa at M sites included: Betaproteobacteriales, Burkholderiaceae, Gammaproteobacteria, Nitrosomonadaceae, Pseudonocardiaceae, Pseudonocardiales, and IS-44 (Fig. S2A; Table S3). In contrast, the top rhizosphere taxon at NM sites was only classified up to the bacterial kingdom level (Fig. S2B; Table S3). For M background soil community, the defining taxa belonged to the phyla Acidobacteria and Actinobacteria (LDA > 4.0). Interestingly, most of the defining NM background soil taxa belonged to Thaumarchaeota. At M sites, LEfSe identified 57 fungal taxa in rhizosphere soil that distinguished the rhizosphere community from the background soil community (Fig. S3A; Table S3). The top six fungal taxa that associated with the M rhizosphere community were Ascomycota, including four taxa that belonged to Dothiodemycetes order (LDA > 4.0; Fig. S3A). Although the number of fungal taxa in the M rhizosphere soil community was higher, the number of bacterial/archaeal and fungal taxa with LDA > 4.0 was the same. Ten fungal taxa were identified by LEFSe as related to the background soil at M sites. These fungal taxa included the phyla: Ascomycota, Basidiomycota, Mortierellomycota, and Glomeromycota (Fig S3A). In contrast, only 11 fungal taxa were identified in the NM rhizosphere community and Cantharellales was the only one fungal taxon with LDA > 4.0 (Fig. S3B). As with LEfSe, SIMPER analysis identified more taxa for M soils that contributed to the dissimilarity between rhizosphere and background soils than for the NM soils (Table S4). Twenty-three and 12 bacterial taxa differentiated the rhizosphere community from the background soil communities for M and NM sites, respectively (p < 0.05; Table S4). The genus Pseudomonas was the most significant taxa defining M rhizosphere samples, whereas the order Gaiellales was the most significant in NM samples. Regarding fungal communities, three taxa explained the differences between rhizosphere and background soil communities from M sites and two taxa from NM sites (p < 0.05; Table S4). The fungal species Russula depallens emerged as the most significant for M sites and its contribution towards the dissimilarity between rhizosphere and background soil communities was the highest among all taxa (1.4%). In NM soils, the fungal taxon from the Helotiales order had the highest contribution to the dissimilarity between rhizosphere and background soils. 3.4 Abiotic and biotic variables positively related to enhanced Zn and Cd hyperaccumulation Leave-one-out cross-validation indicated that the optimal number of partial least square (PLS) components for both regression analyses (i.e., performed on either Zn hyperaccumulation or Cd hyperaccumulation traits) was two. These two components together explained 71% and 76% of the variance in Zn and Cd hyperaccumulation, respectively (Fig. 3A; Fig. S4). Importantly, as our regression models were based on the same numbers of components, they are fully comparable. For Zn hyperaccumulation, all abiotic and biotic variables together (cumulative R2Y) and both Zn hyperaccumulation variables together (cumulative R2X from Znshoot and ZnTF) explained 50% and 47% of variance for the first PLS component and 21% and 11% for the second component, respectively. For Cd hyperaccumulation, cumulative R2Y and cumulative R2X (from Cdshoot and CdTF) explained 58% and 52% of variance for the first PLS component and 18% and 16% for the second component, respectively. To evaluate the importance of each predictor variable in our models, we used the VIP scores from the PLS output. Accordingly, 45 and 47 out of 95 variables were identified as relevant in the Zn and Cd hyperaccumulation PLS regression models, respectively (Table S5). The absolute values of the corresponding beta regression coefficients express the relative strength of these variables in explaining the variance in Zn and Cd hyperaccumulation (Fig. 3B). A striking result from the PLS analysis is that the three out of the four strongest variables that impacted Zn hyperaccumulation (Znshoot and ZnTF) were fungi-related including: fungal richness, diversity, and the taxa Humicola (Fig. 3B). In contrast, none of these fungi-related variables were important for Cd hyperaccumulation. The only important fungi-related variable for Cd was related to CdTF and in this case, the taxa Dothideomycetes was the strongest impacting variable. Interestingly, Dothideomycetes was also a strong variable (10th ranked) for ZnTF (Fig. 3B). Other important variables correlated with Zn uptake included plant shoot concentration of Mg and several abiotic variables including NH4+-N, TOC, TC, TN, and Cu. Bacteria were of less importance as variables that impacted high levels of Zn uptake. The results for Cd are in stark contrast to Zn. The strongest variables related to Cd hyperaccumulation were abiotic soil variables and bacteria. Abiotic variables of importance to enhanced Cdshoot concentrations included Cd and Zn bioavailability. Abiotic variables important for the CdTF included: Cu content, NH4+-N, TOC, and TC. Bacteria related to high levels of Cd uptake included Kineosporia and Lysinmonas for Cdshoot and Pseudomonas, Devosia, Flavobacterium, and Crossiella for CdTF. Though abiotic factors and bacteria were most important for Cd hyperaccumulation, the exception was that for CdTF the fungus Dothideomycetes was the most important factor. 3.5 Abiotic and biotic variables negatively related to Zn and Cd hyperaccumulation Six out of the seven strongest variables that negatively associated with Zn hyperaccumulation (Znshoot and ZnTF) were the same including: NO3—N, latitude, bacterial taxa Nitrospira, Phenylobacterium, Subgroup2; and fungal taxa Helotiales Incertae sedis (Fig. 3B). Interestingly, bioavailable concentrations of Cd, Pb, and Zn along with pH negatively related to translocation of both Zn and Cd. Seven bacteria taxa negatively impacted CdTF, whereas Cd hyperaccumulation was related to fungal NMDS axis score and Dothideomycetes. Notably, different nitrogen forms influenced Zn and Cd hyperaccumulation with NO3−-N and NH4+-N relating to Zn and Cd uptake, respectively. 3.6 Key microbial taxa that associated with Zn and Cd hyperaccumulation PLS analysis identified 29 microbial taxa (24 bacteria and 5 fungi) in rhizosphere as significant explanatory variables of Zn and Cd hyperaccumulation in A. halleri shoots (Fig. 3). Notably, some taxa were associated uniquely with Zn (7) or Cd (11) hyperaccumulation, whereas many were associated with both (11). The vast majority of these 29 taxa were present in the A. halleri rhizosphere at M sites, with 27 and 24 taxa linked to the M_PL27 and M_PL22 sites, respectively (Fig. 4). Only 13 of the taxa were found in the NM_PL35 site, but 21 taxa corresponded to the NM_PL14 site. Interestingly, there were 10 bacterial taxa that were present at both M sites and one NM site NM_PL14, but not at NM_PL35. The relative abundance of these taxa in NM_PL14 was either comparable or lower than in M sites. The bacterial taxa belonged to five phyla, including Acidobacteria, Actinobacteria, Bacteroidetes, Nitrospirae, and Proteobacteria and the five fungal taxa were from the Ascomycota phylum (Table S3). There were six bacterial genera (Actinomycetospora, Crossiella, Marmoricola, Bradyrhizobium, Rhodoplanes, and Pseudomonas) that had relative abundance > 0 in the M sites (M_PL22 and M_PL27) but were absent at NM sites (NM_PL14 and NM_PL35) (Fig. 4). In contrast, two bacterial genera, Nitrospira and Phelybacterium, were present exclusively at NM sites. 4. DISCUSSION 4.1 Arabidopsis halleri metal hyperaccumulation patterns in M and NM sites As shown recently, the four A. halleri populations (two each from M and NM sites) in this study are both genetically similar and in close geographic proximity (Babst-Kostecka et al., 2018). Here, we investigated the differences in Zn and Cd hyperaccumulation among these populations. All plants reached exceptionally high concentrations of both Zn and Cd in the shoots at M sites, whereas only Zn was hyperaccumulated at NM locations. These results are congruent with the findings of previous studies on A. halleri and underline the constitutive nature of Zn hyperaccumulation and the population-specific character of Cd hyperaccumulation in this species (Bert et al., 2000; Stein et al., 2017; Corso et al., 2018). The evolutionary dynamics of Zn hyperaccumulation by A. halleri are particularly interesting, as illustrated by a recent study that associated an increase in Zn hyperaccumulation in lowland NM population in Niepołomice Forest with the colonization of this location by plants from a former M population (Babst-Kostecka et al., 2018). Similarly, in this study the NM_PL14 population (Niepołomice Forest) accumulated remarkably high Zn concentrations, matching levels found at M locations even though the soil Zn concentration at the NM_PL14 site was ~80-fold lower compared to M sites. Interestingly, the Niepołomice Forest accessions have reduced neutral genetic variation compared to other NM and M populations from the same geographic region (Babst-Kostecka et al., 2018), which could negatively impact population fitness (Markert et al., 2010). It is thus surprising that these genetically less diverse A. halleri plants showed no visible toxicity symptoms even after extremely high Zn accumulation and reached the highest aboveground biomass of all investigated populations from Southern Poland in an earlier study (Dietrich et al., 2019). Taken together, these results suggest that soil TME concentration is not the main driver for A. halleri Zn hyperaccumulation and led us to consider other possible biotic and abiotic factors that could influence TME uptake. These are discussed in the following sections. 4.2 Microbial community dynamics in M and NM soils Elevated concentrations of TME are known to impact the soil microbial community. In this study, both M and NM sites were characterized by similar levels of microbial richness and diversity, but notable differences in microbial community composition were observed. Similarly, long-term exposure to historic metal-contaminated mine tailings has previously been reported to have weak or no impact on microbial alpha diversity in soils (Gans et al., 2005; Bamborough & Cummings, 2009). This was attributed to the local adaptation of microbial communities and the replacement of metal sensitive groups by more tolerant ones (Berg et al., 2012; Azarbad et al., 2015). In terms of microbial community structure, differences in the composition of both bacterial/archaeal and fungal communities in M and NM sites were observed both in the present and previous studies (Tipayno et al., 2018; Xu et al., 2019). As suggested above, these differences may be due to the replacement of metal-sensitive with metal-tolerant microbial populations. Not only did metal contamination (M vs. NM sites) affect the structure of bacterial/archaeal and fungal communities differently in this study but the contribution of site type in shaping the community composition was two-fold greater for bacteria/archaea than for fungi. A similar pattern was previously reported by Khan et al. (2010) and may be associated with differences in bacterial and fungal activities in M soils (Rajapaksha et al., 2004). Further, phospholipid fatty acid analysis – commonly used to quantify soil microbial responses to environmental stress – has shown positive correlation between soil available Cd and fungal indicators but for bacterial indicators the correlation is a negative one (Shentu et al., 2014). Of particular interest were the observed differences between A. halleri rhizosphere and background soil microbial communities from M and NM sites. In particular, the number of microbial taxa defining the rhizosphere was greater than for background soils. This difference was more pronounced in M than in NM sites. One possible explanation for these findings is that A. halleri recruits a more unique rhizosphere community from the background soil in M vs. NM sites. This is supported by Honeker et al. (2019), who reported a greater number of taxa enriched in Atriplex lentiformis rhizosphere in acidic pyritic mine tailings (24 rhizosphere taxa vs. 0 bulk taxa) compared to higher pH (5.2–7.8) substrates (15 rhizosphere taxa vs. 3 bulk taxa). Similar observations were derived from comparisons between A. lentiformis rhizosphere and bulk taxa in compost-amended and unamended pyritic tailings (Valentín-Vargas et al., 2018). Taken together, these findings indicate that plants growing under M conditions recruit a greater number of novel rhizosphere taxa than their counterparts in NM soils. Another possible explanation for a higher number of unique defining rhizosphere taxa in M sites may be the presence of specific metal tolerant or plant growth promoting bacteria (PGPB) that facilitate survival and help alleviate plant metal stress (Ma et al., 2015). Overall, PGPB can enhance plant growth by regulating plant stress responses through the production of siderophores, phytohormones, and enzymes (Penrose & Glick, 2001; Press et al., 2001; Patten & Glick, 2002). The present study identified several bacterial groups with the above-mentioned PGPB properties within the taxa associated with the A. halleri rhizosphere from M sites, including Burkholderiaceae and Sphingomonadaceae. For instance, the Pandoraea genus of Burkholderiaceae is known to produce the enzyme 1-aminocyclopropane-1-carboxylate (ACC) deaminase that alleviates stress by lowering ethylene concentrations in plants exposed to biotic and abiotic stress (Anandham et al., 2008). Genera of Sphingomonadaceae can produce phytohormones like gibberellins, abscisic acid, indole-3-acetic acid, and salicylic acid to promote plant growth (Yang et al., 2014). Taken together, the production of phytohormones and enzymes by rhizospheric PGPB are important adaptive strategies that are likely to facilitate the success of A. halleri survival and growth in M soils. The fungal taxa identified as most likely to explain differences between A. halleri rhizosphere and surrounding background soils were almost 6-fold higher in M than in NM sites (57 vs. 11 fungal taxa). Out of 57 unique rhizosphere fungal taxa associated with M sites, the existing literature characterizes very few with defined functional roles to support plants. Indeed, previous efforts that have characterized fungi in M soils and described fungal metal-tolerance mechanisms did not report on many of the specific taxa identified in our study (Miransari, 2010; Zarei et al., 2010; Miransari, 2011; Luo et al., 2014; Thijs et al., 2017). Exceptions include selected species of the genera Hormonema and Phialocephala which were reported to produce the auxin phytohormone indole-3-acetic acid and siderophores, respectively (Bartholdy et al., 2001; Soto et al., 2019). Given the number of novel fungal taxa that we found in the A. halleri rhizosphere from M sites, but not NM sites, the importance of fungi in Zn uptake (Fig. 3), and the unknown functional role of most of these taxa, further research is needed to unravel their potential roles in the rhizosphere of metal-hyperaccumulating plants. 4.3 Zn and Cd hyperaccumulation are shaped by different biotic and abiotic parameters We observed a large variability in the capacity for hyperaccumulation of Zn and Cd in M and NM populations of A. halleri from Southern Poland. While such behavior is well documented in the literature, the factors that drive this variation are not yet well-understood (Honjo & Kudoh, 2019). Recent studies have linked the concentration of Zn and Cd accumulated in plant shoots with soil element composition showing that soil metal concentrations explain only a small part of the variation in both hyperaccumulation traits (Stein et al., 2017; Frérot et al., 2018). To further investigate factors important for Zn and Cd hyperaccumulation this study measured 95 abiotic and biotic parameters. Results show that Zn hyperaccumulation was predominantly governed by biotic variables and similar factors were associated with both Zn shoot accumulation and the Zn translocation factor. Specifically, the strongest positive association of Zn hyperaccumulation was with fungal richness and diversity. This was followed by abiotic factors including ammonium N, total organic C, and total N. The strongest negative associations of Zn hyperaccumulation were with nitrate N and some bacterial taxa. In contrast with Zn, Cd hyperaccumulation was primarily explained by abiotic factors and different factors were more strongly associated with either Cd shoot accumulation or the Cd translocation factor. The strongest positive association of Cd in shoot tissues was with the bioavailability of Cd, Zn, and Pb. Similarly, concentrations of Cd and Pb in soil were previously reported to be strong drivers of the evolution of Cd hyperaccumulation in A. halleri (Frérot et al., 2018). Also strongly positively associated with Cd in shoot tissues were several bacteria/archaea taxa which are discussed further below. The variables with the strongest positive association with CdTF were a single fungal taxon, and three abiotic variables including soil Cu concentration, ammonium N, and total organic C. Recall that Cd was hyperaccumulated only at M sites. Several bacterial taxa that were abundant in the A. halleri rhizosphere at both M sites, but not at the NM sites, were strongly associated with Cd hyperaccumulation. These include: Actinomycetospora, Bradyrhizobium, Marmoricola, and Pseudomonas. Pseudomonas is a well-known PGPB and has previously been reported to reduce metal-induced stress in plants at M sites (Xiao et al., 2017; Honeker et al., 2019). Cadmium-resistant Pseudomonas sp. strains are capable of either leaching out Cd from Cd-complexed compounds by producing organic acids or biosorbing metals by releasing Cd-binding siderophores and peptides (Muehe et al., 2015). Some strains have also been shown to have a positive effect on the phytoextraction of Cd and Zn. Indeed, inoculation with Pseudomonas sp. strains increased Cd accumulation and stimulated the growth of roots and shoots in the Zn/Cd/Pb hyperaccumulator Brassica napus (Sheng & Xia, 2006; Dell’Amico et al., 2008; Dąbrowska et al., 2017), and increased growth and Cd and Zn content in Zn/Cd hyperaccumulator Sedum alfredii (Li et al., 2007). Also Bradyrhizobium, which is typically involved in atmospheric N fixation (Swanner and Templeton 2011), can produce siderophores and thus has the potential to increase metal solubility in the rhizosphere of metal hyperaccumulating plants (Asad et al., 2019). A second grouping of bacteria with PGPB activities includes taxa which were associated with both Zn and Cd hyperaccumulation: Rhodoplanes, Crossiella, and Solirubrobacteraceae, the latter two belonging to Actinobacteria phylum. While some Rhodoplanes are known for their N-fixation potential (Zhu et al., 2018), they are also capable of producing plant growth hormones (indole-3-acetic acid and 5-aminolevulinic) (Sun et al., 2015) that can alleviate metal-induced toxicity and stress in plants. Actinobacteria can enhance plant growth and yield through the fixation of atmospheric N, the solubilization of minerals such as P, K, and Zn, as well as the production of siderophores and plant growth hormones. Five bacterial/archaeal and fungal taxa were identified as negatively associated with hyperaccumulation of Zn and Cd in the shoots. These included taxa that were abundant solely in NM sites and belonged to the phylum Acidobacteria (order Subgroup2), genus Nitrospira and Phenylobacterium. Their restriction to NM soils suggests that these microorganisms might be sensitive to soil metal contamination. Indeed, previous studies showed that Acidobacteria, which are involved in various soil processes, are sensitive to M soils (Bell et al., 2015). In contrast, the well-known nitrite oxidizer Nitrospira has previously been shown to be adapted to M sites (Luo et al., 2018). Regarding Phenylobacterium, its functional relevance is currently unknown, however it was previously reported in the rhizosphere of A. halleri from an M soil (Muehe et al., 2015). Finally, we note a group of 10 bacterial taxa that were present at both M sites (M_PL27 and M_PL22) and at the lowland NM site (NM_PL14), but not at NM_PL35. Given the remarkably high Zn hyperaccumulation levels in plants from NM_PL14, M_PL27 and M_PL22 locations, these microbial taxa may play a role in the mechanism of Zn hyperaccumulation. In particular, Devosia, which was widely abundant in our study at M and NM_PL14 sites, is a well-known N-fixing bacteria (Laranjo et al., 2014). Devosia has previously been associated with Zn hyperaccumulator Thlaspi caerulescens (Lodewyckx et al., 2002) and Ni hyperaccumulator Alyssum murale (Lopez et al., 2017). It has been shown to occur in both bulk and rhizosphere soils at higher pH (5.2–7.8), but was present exclusively in the rhizosphere of acidic soils (Honeker et al., 2019). Accordingly, Devosia seems to be recruited by plants-root systems that encounter toxic conditions and is likely to be a relevant member of the rhizobacterial community associated with hyperaccumulating plants. Other interesting taxa observed were Lysinimonas and Galbitalea from the Microbacteriaceae family. Strains of these metal-resistant bacteria were isolated from the rhizosphere of Salix caprea grown at a M site. They were shown to significantly increase the extractability of Zn and Cd from contaminated soil and cause an increase of Zn and Cd concentration in S. caprea shoots (Kuffner et al., 2008; De Maria et al., 2011). Finally, members of Chitinophagaceae were found in both M sites and in NM_PL14. Recent research identified Chitinophagaceae in the rhizosphere of Zn/Cd hyperaccumulator Sedum alfredii and showed a positive correlation with Cd and Pb concentration in plant shoots and roots (Cao et al., 2020). Overall, the presence and ecological function of A. halleri rhizosphere taxa in both M sites and NM_PL14 (but not NM_PL35) suggests that selected members of microbial communities may affect Zn and Cd mobilization in soils and result in exceptionally efficient Zn uptake by A. halleri plants at both M and NM locations. These findings highlight the importance of investigating the role of rhizosphere microbial diversity in metal mobilization at M as well as NM sites. Yet, the diversity and abundance of the rhizosphere microbial community as characterized by DNA amplicon sequencing may vary from those of active microbial communities. As the latter may differently influence metal accumulation in A. halleri, future studies should evaluate both, DNA and RNA, and utilize RNA:DNA ratios as a proxy for microbial activity (Mei et al., 2016; Bowsher et al., 2019; Honeker et al., 2019). Similarly, assessing the bioavailable fractions of a larger suit of elements may identify additional relevant variables driving metal accumulation in plants. 5. CONCLUSIONS This study provides new insight into the association of soil microbial populations with Zn and Cd hyperaccumulation traits in A. halleri growing at M and NM sites. Metal contamination significantly altered the structure of soil bacterial/archaeal and fungal communities and influenced the number of unique taxa recruited by the A. halleri rhizosphere. Additionally, results show that Zn hyperaccumulation was predominantly governed by biotic variables, whereas variability in Cd hyperaccumulation was primarily explained by abiotic factors. We have identified a group of microbial taxa that consistently associated with Zn hyperaccumulation by A. halleri, regardless of soil metal contamination levels. These findings suggest that these taxa not only increase metal mobilization in the soil and hyperaccumulation by A. halleri, but also benefit overall plant performance. This can result in hyperaccumulation of Zn even in NM sites such as NM_PL14. The identification of taxa with potential to support metal hyperaccumulation is important from an applied perspective – it is a necessary step towards optimizing phytoextraction of metals from soil. We highlight the importance of selecting the most promising hyperaccumulating plant populations and the need to design highly specific microbial inocula with distinct functional properties to aid in the hyperaccumulation process. Future work should focus on: i) unraveling the functional role of the herein identified microorganisms for plant metal hyperaccumulation and ii) assessing the effective role of soil abiotic and biotic factors in the adaptive evolution of A. halleri at M and NM sites, e.g., through reciprocal transplant experiments. Supplementary Material 1 ACKNOWLEDGEMENTS This work was supported by the MINIATURA2 grant financed by the National Science Centre, Poland (2018/02/X/NZ8/00546), the POWROTY/REINTEGRATION programme of the Foundation for Polish Science cofinanced by the European Union under the European Regional Development Fund (POIR.04.04.00-00-1D79/16-01), and by the National Institute of Environmental and Health Sciences Superfund Research Program (Grant P42ES04940) at the University of Arizona. Figure 1. Uptake of Zn and Cd by A. halleri in the four study sites. Shoot and root concentrations of Zn (panels A and B, respectively) and Cd (panels E and F, respectively), root-to-shoot translocation factors (TF) for Zn (panel C) and Cd (panel G), and bioavailable (CDGT) fractions of Zn (panel D) and Cd (panel H) in rhizosphere soil are shown for each site. Each box represents the inter-quartile range of the data, with the median indicated by the horizontal line. The dotted lines in panels A and E indicate the thresholds for hyperaccumulation for Zn and Cd, respectively. The dotted lines in panels C and G indicate a TF = 1. NM = non-metalliferous and M = metalliferous soils. Different letters indicate statistically significant differences at p ≤ 0.05 (Kruskal-Wallis test). Figure 2. (A) Number of observed bacterial/archaeal and fungal amplicon sequence variants (ASVs) or phylotypes in the four study sites for rhizosphere and background soils. The boxes represent the inter-quartile range of the data, the median is indicated by the horizontal line. No comparisons between the means within the A panel were significant (Wilcoxon and Kruskal-Wallis tests). (B) Nonmetric multidimensional scaling (NMDS) ordination plots of microbial community structure in the four study sites. R2 represents the variation explained by site type (M vs. NM) and location (NM_PL14, NM_PL35, M_PL22, M_PL27) for the microbial community composition (p-value of 0.001 is represented as ***; PERMANOVA). NM = non-metalliferous and M = metalliferous soils. Figure 3. Biotic and abiotic drivers of Zn and Cd hyperaccumulation as determined by partial least square (PLS) regression analysis. (A) Correlations of the trait (black) and explanatory (colored) variables with the first two axes associated with the first two PLS components. The inner dashed circle denotes the correlation coefficient r = 0.75. The percentages of the variances in the matrix of responses and in the matrix of predictors explained by each variable are indicated on the respective axes. Note that variables with longer lines have greater loadings in the first or second component and are thus more influential in the model. Angles between the lines reflect the correlation between the variables; smaller angles indicate that variables are highly correlated. (B) Standardized regression coefficients (St. coeff.) for explanatory variables relevant for explanatory variables relevant for Zn and Cd hyperaccumulation. Note that only variables with Variable Importance in Projection > 0.8 are shown. TF, translocation factor; Temp, mean annual temperature; Precip, annual precipitation; CDGT, bioavailable fractions; EC, electrical conductivity; TN, total nitrogen; TC, total dissolved carbon; TOC, total organic carbon; TIC, total inorganic carbon. The letters in front of the microbial taxa names reflect the level of taxonomic hierarchy: c, class; o, order; f, family; g, genus. Figure 4. Heatmap showing the relative abundance of microbial taxa in A. halleri rhizosphere from contaminated (M_PL22, M_PL27) and uncontaminated (NM_PL14, NM_PL35) locations. The 29 taxa identified as significant drivers of Zn and Cd hyperaccumulation (as determined by partial least square regressions) were considered for the relative abundance comparison. The letters in front of the microbial taxa names reflect the level of taxonomic hierarchy: c, class; o, order; f, family; g, genus. The taxa are arranged alphabetically in the order of their phylum. Table 1. Geographic location, pH, bioavailable and total Zn and Cd concentrations in rhizosphere soil samples (mean ± SD, n=3) at the four study sites. Site Location Latitude [°N] Longitude [°E] Elevation (m) pH ZnCDGT1 (μg L−1) CdCDGT1 (μg L−1) Total Zn (mg kg−1) Total Cd (mg kg−1) NM_PL14 Niepołomice 50.108833 20.367467 188 5.8 ± 0.6 12 ± 5 0.08 ± 0.01 127 ± 35 0.31 ± 0.11 NM_PL35 Kościelisko 49.287056 19.879417 927 4.5 ± 0.3 35 ± 33 0.17 ± 0.06 61 ± 28 0.2 ± 0.1 M_PL22 Bukowno 50.282800 19.478717 339 7.3 ± 0.1 1052 ± 260 13 ± 5 6068 ± 4916 45.2 ± 37.2 M_PL27 Galman 50.198367 19.538817 471 6.1 ± 0.2 614 ± 335 8.6 ± 4.5 9401 ± 3160 102.4 ± 23.7 1 ZnCDGT and CdCDGT = bioavailable Zn and Cd NM = non-metalliferous sites; M = metalliferous sites. Table 2. Number of taxa identified by linear discriminant effect size (LEfSe) analysis that explain the differences between contaminated and uncontaminated site types and between rhizosphere and background soil communities within a given site type. Compared Sample Groups Number of taxa in Sample Group 1 vs. Sample Group 2; (ratio) Sample Group 1 Sample Group 2 Bacteria/Archaea Fungi M (contaminated sites) NM (uncontaminated sites) 20 vs. 5; (4) 32 vs. 19; (1.7) Rhizosphere soil (M sites) Background soil (M sites) 40 vs. 7; (5.7) 57 vs. 10; (5.7) Rhizosphere soil (NM sites) Background soil (NM sites) 24 vs. 8; (3) 11 vs. 7; (1.6) HIGHLIGHTS Key drivers of metal hyperaccumulation variability in plants are not fully understood A. halleri rhizosphere recruits more unique microbial taxa at contaminated sites Zn hyperaccumulation in A. halleri associates with rhizosphere microbial communities Cd hyperaccumulation in A. halleri is governed by abiotic soil parameters Locally optimized combinations of plants and microbes will enhance phytoremediation This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Declaration of interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. REFERENCES Ali H , Khan E , Sajad MA . 2013. Phytoremediation of heavy metals—concepts and applications. Chemosphere 91 (7 ): 869–881.23466085 Anandham R , Indira Gandhi P , Madhaiyan M , Sa T . 2008. Potential plant growth promoting traits and bioacidulation of rock phosphate by thiosulfate oxidizing bacteria isolated from crop plants. Journal of Basic Microbiology 48 (6 ): 439–447.18785656 Anderson MJ . 2001. A new method for non-parametric multivariate analysis of variance. Austral Ecology 26 (1 ): 32–46. Asad SA , Farooq M , Afzal A , West H . 2019. Integrated phytobial heavy metal remediation strategies for a sustainable clean environment-a review. Chemosphere 217 : 925–941.30586789 Azarbad H , Niklińska M , Laskowski R , van Straalen NM , van Gestel CAM , Zhou J , He Z , Wen C , Röling WFM . 2015. Microbial community composition and functions are resilient to metal pollution along two forest soil gradients. FEMS Microbiology Ecology 91 (1 ): 1–11. Babst-Kostecka A , Schat H , Saumitou-Laprade P , Grodzińska K , Bourceaux A , Pauwels M , Frérot H . 2018. Evolutionary dynamics of quantitative variation in an adaptive trait at the regional scale: The case of zinc hyperaccumulation in Arabidopsis halleri. Molecular Ecology 27 (16 ): 3257–3273. Baker AJM . 1981. Accumulators and Excluders - Strategies in the response of Plants to heavy metals. Journal of Plant Nutrition 3 (1–4 ): 643–654. Baker AJM . 1987. Metal tolerance. New Phytologist 106 (suppl. ): 93–111. Balabane M , Faivre D , van Oort F , Dahmani-Muller H . 1999. Mutual effects of soil organic matter dynamics and heavy metals fate in a metallophyte grassland. Environmental Pollution 105 (1 ): 45–54. Bamborough L , Cummings SP . 2009. The impact of increasing heavy metal stress on the diversity and structure of the bacterial and actinobacterial communities of metallophytic grassland soil. Biology and Fertility of Soils 45 (3 ): 273–280. Bartholdy BA , Berreck M , Haselwandter K . 2001. Hydroxamate siderophore synthesis by Phialocephala fortinii, a typical dark septate fungal root endophyte. BioMetals 14 (1 ): 33–42.11368273 Barton CJ . 1948. Photometric analysis of phosphate rock. Analytical Chemistry 20 (11 ): 1068–1073. Bell TH , Cloutier-Hurteau B , Al-Otaibi F , Turmel MC , Yergeau E , Courchesne F , St-Arnaud M . 2015. Early rhizosphere microbiome composition is related to the growth and Zn uptake of willows introduced to a former landfill. Environmental Microbiology 17 (8 ): 3025–3038.25970820 Berg J , Brandt KK , Al-Soud WA , Holm PE , Hansen LH , Sørensen SJ , Nybroe O . 2012. Selection for Cu-tolerant bacterial communities with altered composition, but unaltered richness, via long-term cu exposure. Applied and Environmental Microbiology 78 (20 ): 7438–7446.22904046 Bert V , Bonnin I , Saumitou-Laprade P , de Laguérie P , Petit D . 2002. Do Arabidopsis halleri from nonmetalicolous populations accumulate zinc and cadmium more effectively than those from metallicolous populations? New Phytologist 155 : 47–57. Bert V , Macnair MR , de Laguerie P , Saumitou-Laprade P , Petit D . 2000. Zinc tolerance and accumulation in metallicolous and non-metallicolous populations of Arabidopsis halleri (Brassicaceae). New Phytologist 146 : 225–233. Bowsher AW , Kearns PJ , Shade A . 2019. 16S rRNA/rRNA gene ratios and cell activity staining reveal consistent patterns of microbial activity in plant-associated soil. mSystems 4 (2 ): e00003–00019.30944883 Bressan M , Roncato MA , Bellvert F , Comte G , Haichar FEZ , Achouak W , Berge O . 2009. Exogenous glucosinolate produced by Arabidopsis thaliana has an impact on microbes in the rhizosphere and plant roots. ISME Journal 3 (11 ): 1243–1257. Callahan BJ , McMurdie PJ , Rosen MJ , Han AW , Johnson AJA , Holmes SP . 2016. DADA2: high-resolution sample inference from illumina amplicon data. Nat Methods 13 (7 ): 581–583.27214047 Cao X , Luo J , Wang X , Chen Z , Liu G , Khan MB , Kang KJ , Feng Y , He Z , Yang X . 2020. Responses of soil bacterial community and Cd phytoextraction to a Sedum alfredii-oilseed rape (Brassica napus L. and Brassica juncea L.) intercropping system. Science of the Total Environment 723 : 138152. Carrascal LM , Galván I , Gordo O . 2009. Partial least squares regression as an alternative to current regression methods used in ecology. Oikos 118 (5 ): 681–690. Corso M , Schvartzman MS , Guzzo F , Souard F , Malkowski E , Hanikenne M , Verbruggen N . 2018. Contrasting cadmium resistance strategies in two metallicolous populations of Arabidopsis halleri. New Phytologist 218 : 283–297. Dąbrowska G , Hrynkiewicz K , Trejgell A , Baum C . 2017. The effect of plant growth-promoting rhizobacteria on the phytoextraction of Cd and Zn by Brassica napus L. International journal of phytoremediation 19 (7 ): 597–604.27739900 De Maria S , Rivelli AR , Kuffner M , Sessitsch A , Wenzel WW , Gorfer M , Strauss J , Puschenreiter M . 2011. Interactions between accumulation of trace elements and macronutrients in Salix caprea after inoculation with rhizosphere microorganisms. Chemosphere 84 (9 ): 1256–1261.21612812 Dell’Amico E , Cavalca L , Andreoni V . 2008. Improvement of Brassica napus growth under cadmium stress by cadmium-resistant rhizobacteria. Soil Biology and Biochemistry 40 (1 ): 74–84. Dietrich CC , Bilnicki K , Korzeniak U , Briese C , Nagel KA , Babst-Kostecka A . 2019. Does slow and steady win the race? Root growth dynamics of Arabidopsis halleri ecotypes in soils with varying trace metal element contamination. Environmental and Experimental Botany 167 : 103862. Dietrich CC , Tandy S , Banaś A , Murawska K , Korzeniak U , Łopata B , Babst-Kostecka A . 2021. Phytoextraction efficiency of Arabidopsis halleri is driven by the plant and not by soil metal concentration. Chemosphere: 131437.34265706 Farinati S , DalCorso G , Bona E , Corbella M , Lampis S , Cecconi D , Polati R , Berta G , Vallini G , Furini A . 2009. Proteomic analysis of Arabidopsis halleri shoots in response to the heavy metals cadmium and zinc and rhizosphere microorganisms. Proteomics 9 (21 ): 4837–4850.19810031 Farinati S , DalCorso G , Panigati M , Furini A . 2011. Interaction between selected bacterial strains and Arabidopsis halleri modulates shoot proteome and cadmium and zinc accumulation. Journal of experimental botany 62 (10 ): 3433–3447.21357773 Farrés M , Platikanov S , Tsakovski S , Tauler R . 2015. Comparison of the variable importance in projection (VIP) and of the selectivity ratio (SR) methods for variable selection and interpretation. Journal of Chemometrics 29 (10 ): 528–536. Frérot H , Hautekèete N-C , Decombeix I , Bouchet M-H , Créach A , Saumitou-Laprade P , Piquot Y , Pauwels M . 2018. Habitat heterogeneity in the pseudometallophyte Arabidopsis halleri and its structuring effect on natural variation of zinc and cadmium hyperaccumulation. Plant and soil 423 : 157–174. Gans J , Wolinsky M , Dunbar J . 2005. Computational improvements reveal great bacterial diversity and high metal toxicity in soil. Science 309 (5739 ): 1387–1390.16123304 Honeker LK , Gullo CF , Neilson JW , Chorover J , Maier RM . 2019. Effect of Re-acidification on Buffalo Grass Rhizosphere and Bulk Microbial Communities During Phytostabilization of Metalliferous Mine Tailings. Frontiers in Microbiology 10 (1209 ). Honjo MN , Kudoh H . 2019. Arabidopsis halleri: a perennial model system for studying population differentiation and local adaptation. AoB Plants 11 (6 ): plz076.31832127 Huang W , Schoenau J . 1998. Fluxes of water-soluble nitrogen and phosphorus in the forest floor and surface mineral soil of a boreal aspen stand. Geoderma 81 (3–4 ): 251–264. Khan S , Hesham AEL , Qiao M , Rehman S , He JZ . 2010. Effects of Cd and Pb on soil microbial community structure and activities. Environmental Science and Pollution Research 17 (2 ): 288–296.19333640 Kuffner M , Puschenreiter M , Wieshammer G , Gorfer M , Sessitsch A . 2008. Rhizosphere bacteria affect growth and metal uptake of heavy metal accumulating willows. Plant and soil 304 (1 ): 35–44. Kushwaha P , Neilson JW , Barberán A , Chen Y , Fontana CG , Butterfield BJ , Maier RM . 2021. Arid Ecosystem Vegetation Canopy-Gap Dichotomy: Influence on Soil Microbial Composition and Nutrient Cycling Functional Potential. Applied and Environmental Microbiology 87 (5 ): 1–17. Laranjo M , Alexandre A , Oliveira S . 2014. Legume growth-promoting rhizobia: An overview on the Mesorhizobium genus. Microbiological Research 169 (1 ): 2–17.24157054 Li WC , Ye ZH , Wong MH . 2007. Effects of bacteria on enhanced metal uptake of the Cd/Zn-hyperaccumulating plant, Sedum alfredii. Journal of experimental botany 58 (15–16 ): 4173–4182.18039737 Lodewyckx C , Mergeay M , Vangronsveld J , Clijsters H , Van Der Lelie D . 2002. Isolation, characterization, and identification of bacteria associated with the zinc hyperaccumulator Thlaspi caerulescens subsp. calaminaria. International journal of phytoremediation 4 (2 ): 101–115.12655804 Lopez S , Piutti S , Vallance J , Morel J-L , Echevarria G , Benizri E . 2017. Nickel drives bacterial community diversity in the rhizosphere of the hyperaccumulator Alyssum murale. Soil Biology and Biochemistry 114 : 121–130. Luo Y , Wu Y , Wang H , Xing R , Zheng Z , Qiu J , Yang L . 2018. Bacterial community structure and diversity responses to the direct revegetation of an artisanal zinc smelting slag after 5 years. Environmental Science and Pollution Research 25 (15 ): 14773–14788.29541981 Luo ZB , Wu C , Zhang C , Li H , Lipka U , Polle A . 2014. The role of ectomycorrhizas in heavy metal stress tolerance of host plants. Environmental and Experimental Botany 108 : 47–62. Ma Y , Oliveira RS , Nai F , Rajkumar M , Luo Y , Rocha I , Freitas H . 2015. The hyperaccumulator Sedum plumbizincicola harbors metal-resistant endophytic bacteria that improve its phytoextraction capacity in multi-metal contaminated soil. Journal of Environmental Management 156 : 62–69.25796039 Markert JA , Champlin DM , Gutjahr-Gobell R , Grear JS , Kuhn A , McGreevy TJ , Roth A , Bagley MJ , Nacci DE . 2010. Population genetic diversity and fitness in multiple environments. BMC evolutionary biology 10 (1 ): 205.20609254 Mei R , Narihiro T , Nobu M , Liu WT . 2016. Effects of heat shocks on microbial community structure and microbial activity of a methanogenic enrichment degrading benzoate. Letters in applied microbiology 63 (5 ): 356–362.27490172 Mendes R , Garbeva P , Raaijmakers JM . 2013. The rhizosphere microbiome: significance of plant beneficial, plant pathogenic, and human pathogenic microorganisms. FEMS microbiology reviews 37 (5 ): 634–663.23790204 Mendez MO , Maier RM . 2008. Phytostabilization of mine tailings in arid and semiarid environments - An emerging remediation technology. Environmental Health Perspectives 116 (3 ): 278–283.18335091 Meyer C-L , Kostecka AA , Saumitou-Laprade P , Créach A , Castric V , Pauwels M , Frérot H . 2010. Variability of zinc tolerance among and within populations of the pseudometallophyte Arabidopsis halleri and possible role of directional selection. New Phytologist 185 (1 ): 130–142. Micallef SA , Shiaris MP , Colón-Carmona A . 2009. Influence of Arabidopsis thaliana accessions on rhizobacterial communities and natural variation in root exudates. Journal of experimental botany 60 (6 ): 1729–1742.19342429 Miethling R , Wieland G , Backhaus H , Tebbe CC . 2000. Variation of microbial rhizosphere communities in response to crop species, soil origin, and inoculation with Sinorhizobium meliloti L33. Microbial Ecology 40 (1 ): 43–56.10977876 Miransari M . 2010. Contribution of arbuscular mycorrhizal symbiosis to plant growth under different types of soil stress. Plant Biology 12 (4 ): 563–569.20636898 Miransari M . 2011. Hyperaccumulators, arbuscular mycorrhizal fungi and stress of heavy metals. Biotechnology Advances 29 (6 ): 645–653.21557996 Muehe EM , Weigold P , Adaktylou IJ , Planer-Friedrich B , Kraemer U , Kappler A , Behrens S . 2015. Rhizosphere microbial community composition affects cadmium and zinc uptake by the metal-hyperaccumulating plant Arabidopsis halleri. Applied and Environmental Microbiology 81 (6 ): 2173–2181.25595759 Nilsson RH , Larsson K-H , Taylor AFS , Bengtsson-Palme J , Jeppesen TS , Schigel D , Kennedy P , Picard K , Gï Ockner FO , Tedersoo L , 2018. The UNITE database for molecular identification of fungi: handling dark taxa and parallel taxonomic classifications. Nucleic Acid Res 47 (D1 ): D259–D264. Oksanen J , Kindt R , Legendre P , O’Hara B , Simpson GL , Solymos P , Stevens MHM , Wagner H . 2008. The vegan package. Community Ecology Package 10 : 631–637. Patten CL , Glick BR . 2002. Role of Pseudomonas putida indoleacetic acid in development of the host plant root system. Applied and Environmental Microbiology 68 (8 ): 3795–3801.12147474 Paulson JN , Pop M , Bravo HC . 2013. metagenomeSeq: Statistical analysis for sparse high-throughput sequencing. Bioconductor package 1 : 191–191. Penrose DM , Glick BR . 2001. Levels of ACC and related compounds in exudate and extracts of canola seeds treated with ACC deaminase-containing plant growth-promoting bacteria. Canadian Journal of Microbiology 47 (4 ): 368–372.11358177 Press CM , Loper JE , Kloepper JW . 2001. Role of iron in rhizobacteria-mediated induced systemic resistance of cucumber. Phytopathology 91 (6 ): 593–598.18943949 Quast C , Pruesse E , Yilmaz P , Gerken J , Schweer T , Yarza P , Rg Peplies J , Glö Ckner FO . 2013. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res 41 (D1 ): D590–D596.23193283 Rajapaksha RMCP , Tobor-Kapłon MA , Bååth E . 2004. Metal toxicity affects fungal and bacterial activities in soil differently. Applied and Environmental Microbiology 70 (5 ): 2966–2973.15128558 Reeves RD , Baker AJM 2000. Metal accumulating plants. In: Raskin I , Ensley BD eds. Phytoremediation of toxic metals: using plants to clean up the environment. New York: Wiley and Sons, 193–229. Reimann C , Albanese S , Batista M , Bel Lan A , Birke M , Cicchella D , Demetriades A , De Vivo B , De Vos W , Dinelli E . 2008. EuroGeoSurveys Geochemical mapping of agricultural and grazing land soil of Europe (GEMAS)-Field manual. NGU-rapport. Roosens NH , Willems G , Saumitou-Laprade P . 2008. Using Arabidopsis to explore zinc tolerance and hyperaccumulation. Trends in Plant Science 13 (5 ): 208–215.18407784 Rosatto S , Roccotiello E , Di Piazza S , Cecchi G , Greco G , Zotti M , Vezzulli L , Mariotti M . 2019. Rhizosphere response to nickel in a facultative hyperaccumulator. Chemosphere 232 : 243–243.31154185 Sanchez G , Sanchez MG . 2012. Package ‘plsdepot’. Partial Least Squares (PLS) Data Analysis Methods, v. 0.1 17 . Segata N , Izard J , Waldron L , Gevers D , Miropolsky L , Garrett WS , Huttenhower C . 2011. Metagenomic biomarker discovery and explanation. Genome Biology 12 (6 ): 1–18. Sheng X-F , Xia J-J . 2006. Improvement of rape (Brassica napus) plant growth and cadmium uptake by cadmium-resistant bacteria. Chemosphere 64 (6 ): 1036–1042.16516946 Shentu JL , He ZL , Zeng YY , He SY , Du ST , Shen DS . 2014. Microbial biomass and PLFA profile changes in rhizosphere of pakchoi (Brassica chinensis L.) as affected by external cadmium loading. Pedosphere 24 (4 ): 553–562. Solís-Dominguez FA , White SA , Hutter TB , Amistadi MK , Root RA , Chorover J , Maier RM . 2012. Response of key soil parameters during compost-assisted phytostabilization in extremely acidic tailings: effect of plant species. Environmental science & technology 46 (2 ): 1019–1027.22191663 Soto J , Ortiz J , Herrera H , Fuentes A , Almonacid L , Charles TC , Arriagada C . 2019. Enhanced arsenic tolerance in Triticum aestivum inoculated with arsenic-resistant and plant growth promoter microorganisms from a heavy metal-polluted soil. Microorganisms 7 (9 ): 1–14. Stein RJ , Höreth S , de Melo JRF , Syllwasschy L , Lee G , Garbin ML , Clemens S , Krämer U . 2017. Relationships between soil and leaf mineral composition are element-specific, environment-dependent and geographically structured in the emerging model Arabidopsis halleri. New Phytologist 213 (3 ): 1274–1286. Sun M , Xiao T , Ning Z , Xiao E , Sun W . 2015. Microbial community analysis in rice paddy soils irrigated by acid mine drainage contaminated water. Applied Microbiology and Biotechnology 99 (6 ): 2911–2922.25408313 Swanner E , Templeton A . 2011. Potential for nitrogen fixation and nitrification in the granite-hosted subsurface at Henderson Mine, CO. Frontiers in Microbiology, 2 , 254.22190904 Talke IN , Hanikenne M , Krämer U . 2006. Zinc-dependent Global Transcriptional Control, Transcriptional Deregulation, and Higher Gene Copy Numer for Genes in Metal Homeostasis of the Hyperaccumulator Arabidopsis halleri. Plant Physiology 142 : 148–167.16844841 Thijs S , Langill T , Vangronsveld J 2017. The bacterial and fungal microbiota of hyperaccumulator plants: Small organisms, large influence. Advances in Botanical Research: Elsevier, 43–86. Tipayno SC , Truu J , Samaddar S , Truu M , Preem JK , Oopkaup K , Espenberg M , Chatterjee P , Kang Y , Kim K , 2018. The bacterial community structure and functional profile in the heavy metal contaminated paddy soils, surrounding a nonferrous smelter in South Korea. Ecology and Evolution 8 (12 ): 6157–6168.29988438 Trivedi P , Leach JE , Tringe SG , Sa T , Singh BK . 2020. Plant–microbiome interactions: from community assembly to plant health. Nature reviews microbiology 18 (11 ): 607–621.32788714 Valentín-Vargas A , Neilson JW , Root RA , Chorover J , Maier RM . 2018. Treatment impacts on temporal microbial community dynamics during phytostabilization of acid-generating mine tailings in semiarid regions. Science of the Total Environment 618 : 357–368. van der Ent A , Baker AJM , Reeves RD , Pollard AJ , Schat H . 2013. Hyperaccumulators of metal and metalloid trace elements: Facts and fiction. Plant and soil 362 (1–2 ): 319–334. Virgone K , Ramirez-Andreotta M , Mainhagu J , Brusseau ML . 2018. Effective integrated frameworks for assessing mining sustainability. Environmental Geochemistry and Health 40 (6 ): 2635–2655.29808248 Walters W , Hyde ER , Berg-Lyons D , Ackermann G , Humphrey G , Parada A , Gilbert JA , Jansson JK , Gregory Caporaso J , Fuhrman JA , 2016. Improved bacterial 16S rRNA gene (V4 and V4–5) and fungal internal transcribed spacer marker gene primers for microbial community surveys. mSystems 1 (1 ): e00009–00015. Xiao X , Fan M , Wang E , Chen W , Wei G . 2017. Interactions of plant growth-promoting rhizobacteria and soil factors in two leguminous plants. Applied Microbiology and Biotechnology 101 (23–24 ): 8485–8497.29038972 Xu Y , Seshadri B , Bolan N , Sarkar B , Ok YS , Zhang W , Rumpel C , Sparks D , Farrell M , Hall T , 2019. Microbial functional diversity and carbon use feedback in soils as affected by heavy metals. Environment International 125 (February ): 478–488.30771648 Yang S , Zhang X , Cao Z , Zhao K , Wang S , Chen M , Hu X . 2014. Growth-promoting sphingomonas paucimobilis ZJSH1 associated with Dendrobium officinale through phytohormone production and nitrogen fixation. Microbial Biotechnology 7 (6 ): 611–620.25142808 Yeoh YK , Dennis PG , Paungfoo-Lonhienne C , Weber L , Brackin R , Ragan MA , Schmidt S , Hugenholtz P . 2017. Evolutionary conservation of a core root microbiome across plant phyla along a tropical soil chronosequence. Nature Communications 8 (1 ). Zarei M , Hempel S , Wubet T , Schäfer T , Savaghebi G , Jouzani GS , Nekouei MK , Buscot F . 2010. Molecular diversity of arbuscular mycorrhizal fungi in relation to soil chemical properties and heavy metal contamination. Environmental Pollution 158 (8 ): 2757–2765.20546984 Zhu B , Zhang X , Zhao C , Chen S , Yang S . 2018. Comparative genome analysis of marine purple sulfur bacterium Marichromatium gracile YL28 reveals the diverse nitrogen cycle mechanisms and habitat-specific traits. Scientific reports 8 (1 ): 1–11.29311619

PMC008xxxxxx/PMC8714704.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8411927 1552 J Biotechnol J Biotechnol Journal of biotechnology 0168-1656 1873-4863 34808251 8714704 10.1016/j.jbiotec.2021.11.004 NIHMS1759030 Article Foam fractionation of a recombinant biosurfactant apolipoprotein Lethcoe Kyle Fox Colin A. Ryan Robert O. 1 Department of Biochemistry and Molecular Biology, University of Nevada, Reno, Reno, NV 89557 Address correspondence to: Robert O. Ryan, robertryan@unr.edu, Robert O. Ryan, Biochemistry and Molecular Biology, University of Nevada, Reno, Mail Stop 0330, 1664 N. Virginia Street, Reno, NV 89557 25 11 2021 19 11 2021 10 1 2022 10 1 2023 343 2531 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Locusta migratoria apolipophorin III (apoLp-III) possesses the ability to exist as a water soluble amphipathic α-helix bundle and a lipid surface seeking apolipoprotein. The intrinsic ability of apoLp-III to transform phospholipid vesicles into reconstituted discoidal high-density lipoproteins (rHDL) has led to myriad applications. To improve the yield of recombinant apoLp-III, studies were performed in a bioreactor. Induction of apoLp-III expression generated a protein product that is secreted from E. coli into the culture medium. Interaction of apoLp-III with gas and liquid components in media produced large quantities of thick foam. A continuous foam fractionation process yielded a foamate containing apoLp-III as the sole major protein component. The yield of recombinant apoLp-III was ~0.2 g / liter bacterial culture. Mass spectrometry analysis verified the identity of the target protein and indicated no modifications or changes to apoLp-III occurred as a result of foam fractionation. The functional ability of apoLp-III to induce rHDL formation was evaluated by incubating foam fractionated apoLp-III with phosphatidylcholine vesicles. FPLC size exclusion chromatography revealed a single major population of particles in the size range of rHDL. The results described offer a novel approach to bioreactor-based apoLp-III production that takes advantage of its intrinsic biosurfactant properties. apolipoprotein reconstituted high density lipoprotein nanodisk bioreactor foam fractionation pmcIntroduction Locusta migratoria apolipophorin III (apoLp-III) is a 164 amino acid amphipathic exchangeable apolipoprotein. In solution, apoLp-III is organized as an elongated bundle of 5 amphipathic α-helices (Breiter et al, 1991). The five helices align adjacent to one another, each projecting their hydrophilic face toward the external aqueous environment while their opposing hydrophobic faces contact one another in the bundle interior. This organizational state provides a clear understanding of why apoLp-III is aqueous soluble as a monomeric protein while providing clues to explain its ability to reversibly associate with lipid surfaces (Wang et al, 2002). In vivo, apoLp-III functions in hormone-stimulated lipid mobilization associated with locust flight activity. During flight, release of adipokinetic hormone induces mobilization of fat body tissue triacylglycerol stores (Ryan and Van der Horst, 2000). Via a lipolytic process, stored triacylglycerol is converted to diacylglycerol (DAG), the transport form of neutral lipid in insects. DAG is transferred onto pre-existing high density lipophorin particles in hemolymph and, in so doing, the particle expands in size and decreases in density, forming low density lipophorin (LDLp). Concomitant with this, apoLp-III transitions from its aqueous soluble amphipathic α-helix bundle conformation to an elongated lipid bound state. ApoLp-III binding to lipophorin occurs as a function of particle DAG content, and serves to stabilize the LDLp particle structure. Subsequently, LDLp ferries its lipid cargo from fat body tissue to flight muscles, where DAG lipolysis and free fatty acid uptake occur. Delivery of fatty acids to flight muscle provides fuel molecules required for oxidative metabolism / ATP production that is essential for sustained flight. As the DAG content of LDLp declines, the particle shrinks and apoLp-III is released, whereupon it assumes its helix bundle conformation. Remarkably, this process, termed the “lipophorin shuttle” (Ryan, 1990), can be repeated multiple times, ensuring a continuous supply of fuel to flight muscle. Knowledge of apoLp-III function in flight muscle physiology has led to investigations of its surface properties and lipid binding activity (Kawooya et al, 1986; Soulages et al, 1995). ApoLp-III has been shown to be a surface-active protein that possesses the ability to induce conversion of phospholipid dispersions into a uniform population of nanoscale disk-shaped bilayer complexes (Wientzek et al, 1994). These stable particles have alternately been referred to as reconstituted high-density lipoprotein (rHDL) or, when formulated with an extraneous hydrophobic bioactive agent, nanodisks (ND) (Ryan, 2010). In these particles, apoLp-III functions as a scaffold that circumscribes the perimeter of a disk-shaped phospholipid bilayer, contacting otherwise aqueous exposed fatty acids at the edge of the disk. In this binding interaction, apoLp-III adopts its open, extended conformation (Narayanaswami and Ryan, 2000). rHDL have been studied extensively, and particles containing human apolipoprotein (apo) A-I as scaffold protein have been formulated for use in vivo as a therapy for atherosclerotic heart disease (Tsujita et al, 2018). At the same time, ND have been formulated with a wide range of hydrophobic bioactive compounds, transmembrane proteins and imaging agents (Ryan 2010; Fox et al, 2021). While new applications of ND technology continue to be developed at research scale (Borch and Hamann, 2009), high cost and labor-intensive downstream processing of the recombinant apolipoprotein represents a bottleneck that impedes industrial scale production of ND. Methods for recombinant apoLp-III expression in E. coli have been established using shaker flasks (Weers et al, 1998). In an effort to improve yield, increase production capacity and lower costs, recombinant apoLp-III expression was evaluated in a bioreactor. By employing a plasmid construct that encodes L. migratoria apoLp-III possessing an N-terminal E. coli pelB leader sequence extension (Hauser and Ryan, 2007), nascent apoLp-III is targeted to the periplasmic space where signal peptidase efficiently cleaves the pelB sequence. The resulting apoLp-III product, which contains no tags or extraneous sequence elements, escapes the bacteria and accumulates in the culture medium. As air and/or O2 gas is introduced into the bioreactor, following induction of apoLp-III expression, large quantities of foam are generated. When the foam was diverted into a collection vessel and collapsed to a liquid foamate, it was shown to be highly enriched in apoLp-III. Thus, as apoLp-III adsorbs to gaseous and liquid components in the media, copious amounts of foam are produced. This study demonstrates that members of the class of amphipathic exchangeable apolipoproteins may be produced with improved yield, lower cost, and reduced downstream processing via continuous foam fractionation. Materials and Methods Plasmid construct design / pelB leader sequence. A pET22b+ plasmid vector harboring the coding sequence of mature L. migratoria apoLp-III (isoform a; 164 amino acids; Van der Horst et al, 1991) was generated as described by Weers et al (1998). The construct encodes a chimera comprised of the E. coli pelB leader sequence fused to the coding sequence of apoLp-III. As previously reported (Ryan et al, 1995) with expression plasmids constructed in this manner, recombinant apoLp-III escapes from E. coli and accumulates in the culture medium as a mature apolipoprotein that possesses no extraneous sequence tags. The plasmid encoding L. migratoria apoLp-III was transformed into E. coli and maintained on streaked agar plates and as a frozen glycerol stock. Shaker flask expression. A saturated overnight culture of E. coli in NZCYM media containing 50 μg/ml carbenicillin / chloramphenicol was used to seed a 1 L culture at 37°C (1:50 dilution). When the culture optical density (OD) at 600 nm reached 0.6, isopropyl β-D-thiogalactopyranoside (IPTG) was added to a final concentration of 1 mM and the culture grown for another 5 h. Cells were then pelleted by centrifugation at 7800×g for 10 min at 4°C and the supernatant, containing recombinant apoLp-III, was concentrated to 100–150 ml at 4°C on a Sartorius Vivaflow 200 tangential flow concentrator fitted with a 10 kDa MWCO cassette. Concentrated sample was then dialyzed against 50 mM sodium phosphate, pH 7.0, 150 mM NaCl (PBS) for 48 h followed by sterile filtration (0.22 μm) and storage at −20°C. Bioreactor expression. A 5 L capacity Distek Model BIOne 1250 bioreactor was employed for expression studies. Following calibration of the dissolved O2 and pH meters, specified running parameters were set in the control module. For apoLp-III expression the bioreactor chamber contained 2 to 4 L NZCYM media or M9 minimal media, as specified. In either case, media was supplemented with 50 μg/ml carbenicillin / chloramphenicol. The bioreactor chamber was then seeded with an overnight culture of E. coli at a ratio of 50 mL per 1 L culture medium. After an initial growth period of 4 – 5 h, when the culture OD at 600 nm reached ~7.0, IPTG was added (1 mM final concentration). A series of parameters were independently examined in an effort to optimize cell growth, recombinant protein expression and/or foam production. These included aeration, temperature, impeller speed and dissolved oxygen proportional–integral–derivative cascade. In other experiments culture media, E. coli growth curve, induction time and IPTG concentration were examined. Culture pH was monitored using an EasyFerm Plus PHI VP 325 Pt100 probe (Hamilton). Fluctuations in pH were a common occurrence during bioreactor runs and, to maintain the pH at 6.8 ± 0.5, a solution of 2 M dibasic sodium phosphate, pH 9.0, was incrementally added. To achieve optimal cell growth, compressed air and/or O2 gas was delivered to the bioreactor chamber at 15 psi via the sparge unit. Air flow was monitored via the BIOne control tower in tandem with the dissolved O2 cascade setting. Gases were passed through two 0.22 μm sterile filters before entering the bioreactor chamber. Continuous foam fractionation. Soon after induction of apoLp-III expression, large quantities of foam were generated in the bioreactor chamber. Because the amount of foam produced exceeded the capacity of the bioreactor, it was diverted to an external collection chamber, driven by the pressure of continuous foam production. This foam diversion process required that all but one of the unused headplate ports be sealed to prevent unwanted escape of foam. The unsealed headplate port was fitted with flexible tubing into which accumulating foam flowed. The tubing fed into an external collection chamber that was placed on a bed of ice. Foam production continued for up to 5 h post induction. When foam production in the bioreactor chamber slowed or ceased, the foam-rich collection vessel was removed and stored a 4°C for 12 h to promote foam collapse into a liquid foamate. Foamate processing. The collapsed foamate was centrifuged at 9,000 × g for 30 min at 4 °C and the supernatant concentrated as described above. The concentrated sample (100 – 150 ml) was dialyzed against 4 L PBS for 48 h at 4 °C. Following dialysis, the concentrated foamate was chromatographed on a 2.5 cm × 100 cm column of Sephadex G50 equilibrated with PBS containing an additional 500 mM NaCL. Five mL aliquots of concentrated foamate were applied to the column with collection of 2.5 mL fractions. Elution fractions were analyzed by SDS-PAGE and those containing apoLp-III were pooled (fractions 25 – 58), concentrated and used in characterization studies. The protein content in samples of processed apoLp-III was determined using the bicinchoninic acid assay (Thermo Fisher Scientific). SDS-PAGE. Sodium dodecyl sulfate polyacrylamide Gel Electrophoresis (SDS-PAGE) was performed using 4–20% Bio-Rad Mini-PROTEAN TGX Precast Gels. Gels were electrophoresed at 150 V for 25–30 min. Following electrophoresis, gels were stained with a solution (90% methanol and 10% glacial acetic acid) containing Amido Black (1 g/L, Sigma) for 30 min. Gels were de-stained in a solution containing 40% deionized water, 40% methanol, 20% glacial acetic acid for 3 h. Gels were imaged using a Bio-Rad ChemiDoc instrument. rHDL formation. Dimyristoylphosphatidylcholine (DMPC) was purchased from Avanti Polar Lipids. Five mg aliquots were dissolved in 200 μL CHCl3:CH3OH (3:1 v/v) and dried under a stream of N2 gas, creating a thin film on the vessel wall. To formulate rHDL, 500 μL PBS was added to the dried lipid and the sample vortexed to disperse / hydrate the phospholipid, yielding an opaque lipid suspension. Two mg recombinant apoLp-III in 500 μL PBS was then added followed by bath sonication at 24 °C until the solution cleared (< 15 min). FPLC analysis. Processed foamate or apoLp-III rHDL (400 μl) were applied to a Superdex 200 Increase 10/300 GL column fitted to a GE AKTA Pure FPLC instrument. Samples were eluted with PBS at a flow rate of 0.75 ml/min. Absorbance was continuously monitored at 280 nm with collection of two mL fractions. Liquid chromatography / mass spectrometry analysis. Protein samples were quantified using Invitrogen EZQ Protein Quantitation Kit (R33200). Fifty μg of each sample was acetone precipitated, reduced / alkylated and digested with trypsin using an EasyPep Mini MS Sample Prep Kit (A40006; Thermo Scientific). Peptide mixtures were separated via liquid chromatography using an UltiMate 3000 RSLCnano system (Thermo Scientific, San Jose, CA). Column temperature was maintained at 50 °C and total gradient time was 175 min, with solvent B increasing from 2–90 % (Solvent A 0.1 % formic acid; Solvent B acetonitrile, 0.1 % formic acid). A self-packed Pico-Frit (New Objectives, Woburn, MA) matrix was used that included a 15 μm tip packed with 40 cm of 1.9 μm ReproSil-Pur 120 C18-AQ (Dr. Maisch GmbH, Germany). Mass spectral analysis was performed using an Orbitrap Fusion mass spectrometer (Thermo Scientific, San Jose, CA). The MS1 precursor selection range was from 400–1500 m/z at a resolution of 120K and an automatic gain control of 4.0 × 103 with a maximum injection time of 150 ms. Quadrupole isolation at 0.7 Th for MS2 analysis using CID fragmentation in the linear ion trap with a collision energy of 35 %. Data were analyzed using SEQUEST (Thermo Fisher Scientific, San Jose, CA, version v.27, rev. 11.) and Proteome Discoverer (Thermo Scientific, San Jose, CA. version 2.1). The mass of full length, bioreactor-derived apoLp-III was determined by electrospray ionization mass spectrometry. Direct infusion was performed on a 1 mg/mL solution of apoLp-III at a flow rate of 5 μL/min using an Orbitrap Eclipse mass spectrometer (Thermo Scientific, San Jose, CA) in intact mode. The MS1 precursor selection range was from 700–2000 m/z at a resolution of 240K. Deconvolution was performed using the Xcalibur Xtract algorithm (Thermo Fisher Scientific, San Jose, CA, version 3.0.63). Results Bioreactor production of recombinant apoLp-III. ApoLp-III is a well-studied member of the class of exchangeable apolipoproteins (Weers and Ryan, 2006). In addition to its role in lipoprotein metabolism, members of this protein class have been utilized in the formulation of aqueous soluble rHDL (Ryan, 2010). In the present study, the ability of recombinant apoLp-III to escape from E. coli when expressed with an N-terminal pelB leader sequence extension has been exploited. Following expression and trafficking to the periplasmic space, the pelB sequence is cleaved and mature apoLp-III transits to the extracellular space. Insofar as extracellular accumulation of recombinant apoLp-III should facilitate downstream processing of recombinant apoLp-III, expression experiments were conducted in a bioreactor. Initial studies confirmed that apoLp-III appears in the culture medium following induction. Confounding early expression runs, however, were large amounts of foam produced in the culture chamber. Further investigation revealed that foam production was contingent upon the air / O2 flow rate and induction of apoLp-III expression. When a foam sample was collected, condensed to a liquid (foamate) and analyzed by SDS-PAGE, apoLp-III was the sole major protein component. Thus, rather than trying to prevent foam formation, a vessel was connected to the bioreactor headplate via flexible tubing so that that foam generated in the chamber could be collected, as depicted in Figure 1. With this setup, as foam was produced and filled the bioreactor chamber, pressure exerted by continued foam production forced the foam to flow through the tubing and into the collection vessel. As the thick foam flowed out of the bioreactor chamber, it was observed to stack upon itself following deposition into the collection vessel, ultimately settling under its own weight. Bioreactor optimization studies. In subsequent experiments, different culture parameters were examined in an effort to optimize apoLp-III expression and foam production. Optimized bioreactor parameters, determined empirically, are presented in Table 1. To investigate if a correlation exists between foam production and apoLp-III content in the culture medium, aliquots of media were collected as a function of time following induction and analyzed by SDS-PAGE (Figure 2). The results indicate that, whereas the culture media did not contain apoLp-III pre-induction, by 1 h post induction, a band corresponding to apoLp-III appeared. The intensity of this band increased up to 2 h post induction but steadily declined at longer time points. The decline in apoLp-III content in the culture medium coincided with foam production, which was diverted into the collection vessel. When a sample of foam was condensed into foamate and analyzed by SDS-PAGE, a single major protein, apoLp-III, was present. Thus, the data indicate that, following induction of apoLp-III expression in the bioreactor, as it accumulates in the culture medium, it combines with gas and liquid present to create foam. Foam processing and apoLp-III isolation. Following foam collection, the vessel was stored at 4 °C for 12 h to induce foam collapse and conversion to a liquid foamate. From a 3 L bioreactor culture, ~1 L of apoLp-III enriched foamate was generated. Following centrifugation to remove cell debris and insoluble material, the foamate was concentrated and dialyzed. To remove minor contaminating proteins, concentrated foamate was subjected to preparative Sephadex G-50 gel filtration chromatography. The eluted fractions were collected and analyzed by SDS-PAGE (Figure 3). The data show that the bulk of contaminating proteins elute in the early fractions while apoLp-III was present in later eluting fractions. Fractions enriched in apoLp-III were pooled, concentrated and used for structure - function characterization studies. The yield of processed apoLp-III, determined by protein assay, ranged from 0.15 g/L to 0.2 g/L, and was dependent upon experimental parameters including culture media, culture temperature, induction O.D. and impeller speed. Characterization of apoLp-III. Given that exposure of apoLp-III to air / O2 gas occurred during foam fractionation, it is conceivable that the protein was modified or damaged by oxidation. To investigate this, an aliquot of processed apoLp-III was subjected to analytical FPLC gel filtration chromatography (Figure 4). The elution pattern obtained was indistinguishable from that observed for apoLp-III produced in a shaker flask (data not shown). Both proteins eluted as a single peak at 16.2 ml, corresponding to a protein of 17 kDa. FPLC elution fractions containing apoLp-III were collected for both shaker flask- and bioreactor-derived apoLp-III and each sample was subjected to trypsin digestion and mass spectrometry analysis. Table 2 depicts peptide fragments that were identified for each protein sample and revealed no differences in fragment mass between the two apoLp-III protein samples. Because the peptide fragments analyzed did not cover the entire sequence, electrospray ionization mass spectrometry was performed on a sample of foam fractionated apoLp-III. The experimentally-derived mono-isotopic mass for this sample was 17,580.46, in excellent agreement with its calculated molecular weight (17,580.33 Da) based on amino acid sequence. Taken together, mass spectral analysis of apoLp-III indicates that no modifications, including oxidation, occurred during foam fractionation. Biological activity of foam fractionated apoLp-III. Having shown that apoLp-III was not modified during foam fractionation and downstream processing, experiments were conducted to evaluate the functional properties of apoLp-III isolated from bioreactor foam. The known ability of apoLp-III to transform DMPC vesicles into a uniform population of rHDL particles (Wientzek et al, 1994) was evaluated. Two indicators of rHDL formation include 1) a change in sample appearance from turbid to clear following rHDL assembly and 2) formation of a homogeneous population of discrete particles, comprised of DMPC and apoLp-III. To examine this, an aqueous dispersion of 5 mg DMPC in buffer only appeared as an opaque, turbid lipid suspension following bath sonication (Figure 5, panel A). By contrast, when 2 mg apoLp-III was added to a 5 mg DMPC dispersion and the sample bath sonicated, the sample transitioned from opaque to clear. Thus, apoLp-III isolated from bioreactor foam retains its ability to transform DMPC vesicles into rHDL. To measure product particle size and homogeneity, FPLC gel filtration chromatography of the rHDL was performed. The chromatogram revealed a single major absorbance peak at ~8.5 ml (Figure 5, panel B) consistent with a homogenous population of particles with a molecular weight of ~200–300 kDa. When apoLp-III produced in a shaker flask was used to generate rHDL and analyzed by FPLC gel filtration chromatography, these particles also eluted at 8.5 ml. The finding that either expression method yielded rHDL with similar elution volumes indicates there was no detectable difference in rHDL particle size between these methods. Discussion Whereas chemical surfactants are generally synthesized from petrochemical and oleochemical sources (Desai et al, 1997), the term “biosurfactant” refers to a wide range of naturally occurring amphiphilic molecules produced by microorganisms. Biosurfactants are structurally diverse molecules, including glycolipids, phospholipids, lipopeptides and polymers that are produced by microorganisms including bacteria, yeast or fungi (Rahman and Gakpe, 2008). For example, rhamnolipids are glycolipid biosurfactants comprised of a rhamnose moiety joined to a 3-(3-hydroxyalkanoyloxy) alkanoic acid fatty acid tail. Of interest to the present study, Beuker et al (2016) reported heterologous rhamnolipid production from Pseudomonas putida in a bioreactor using integrated foam fractionation. On the other hand, surfactin is a cyclic lipopeptide discovered in Bacillus sp. This biosurfactant is organized as a heptapeptide attached to a β-hydroxy fatty acid chain forming a cyclic lactone ring structure. These compounds, and other biosurfactants, are secreted and can therefore can be processed by the technique of foam fractionation (Burghoff, 2012). Foam fractionation of biosurfactants represents a cost-effective, efficient processing strategy capable of concentrating and separating dilute products from culture media. The basic principle of this technology is adsorptive bubble separation, wherein air bubbles generated via gas flow generates foam. A certain amount of liquid is trapped between air bubbles (foam lamella) and, therefore, is lost within the foam. Gravity causes draining of entrapped liquid, which leads to foam collapse. The resulting foamate (collapsed foam) contains more concentrated surface- and nonsurface-active compounds than the initial liquid. To our knowledge, continuous foam fractionation has not been applied to recovery of recombinant proteins expressed in E. coli. Two key factors allowed for this method to be applied to recombinant apoLp-III in the current study. First, the plasmid construct employed encodes a 22 amino acid E. coli pelB leader sequence fused to the N-terminus of L. migratoria apoLp-III. The function of the pelB leader sequence is to direct nascent proteins to the periplasmic space. This is critical because, following entry to the periplasmic space, leader peptidase cleaves the pelB sequence, yielding intact mature apoLp-III. Subsequently, this protein escapes the bacteria and accumulates in the media. The second factor relates to the intrinsic biosurfactant properties of apoLp-III. In aqueous solution, apoLp-III is comprised of five elongated amphipathic α-helices organized as a five helix bundle. Each α-helix is amphipathic, possessing distinct hydrophilic and hydrophobic faces. The hydrophobic faces of apoLp-III α-helices contact one another in the interior of the helix bundle and these interactions stabilize the bundle conformation (Wang et al, 1998). Upon exposure to a lipid surface, however, the molecule undergoes a conformational change in which the protein adopts an extended open conformation that allows the hydrophobic faces of its α-helices to interact with the lipid surface (Weers and Ryan 2003). Whereas this property is essential to its role in lipoprotein metabolism, apoLp-III also displays general biosurfactant properties including adsorption at the air / water interface (Kawooya et al, 1986). Thus, when gas bubbles through a liquid solution containing apoLp-III, it can interact with the gas and liquid components to induce foam production. Importantly, foam produced during bioreactor-based expression of recombinant apoLp-III is a thick, dense foam that collects in the space above the culture medium. As the amount of foam accumulating in the confined headspace of the bioreactor increases, foam is forced out of the bioreactor via an open port into a tube that feeds into a collection vessel. This process is initiated by induction of the bacterial culture with IPTG and continues for several hours. Indeed, it is conceivable that adding more IPTG, providing additional nutrients, or modification of one or more other parameters, could lead to an increase in the duration of foam production and, thereby, increase apoLp-III concentration in the foamate. An important aspect of the findings reported in this study is the possibility that other members of the amphipathic exchangeable apolipoprotein class will display similar behavior. There are a dozen or more unique exchangeable apolipoproteins, many of which have been investigated in detail (Mahley et al, 1984). Collectively, members of the apolipoprotein class perform biological functions related to lipoprotein metabolism. Moreover, in the case of apolipoprotein A-I and apolipoprotein E, they are known to form rHDL for use in specific medical or biotechnological applications. The fact that the N-terminal 22 kDa fragment of human apoE3 (Wilson et al, 1991), when expressed as a pelB fusion protein, is also secreted into the culture media (Fisher et al, 1997), suggests this protein in is amenable to foam fractionation. An advantage of continuous foam fractionation for recombinant protein processing relates to potential cost savings when producing large quantities of apolipoprotein. Whereas it is conceivable that apoLp-III will find utility in biosurfactant applications, its role as a scaffold protein for a specialized category of rHDL, termed nanodisks (ND), is well established (Fischer et al, 2010, Wan 2011). These particles exist as nanoscale-size, aqueous soluble disk-shaped lipid bilayers that are circumscribed by apoLp-III, which functions as a scaffold protein. In this structural arrangement, apoLp-III contacts otherwise exposed fatty acyl chains at the edge of the disk via the hydrophobic face of its amphipathic α-helices. The particles display full aqueous solubility by presenting apoLp-III’s opposing, hydrophilic faces toward the aqueous environment. A wide range of applications of ND technology have been developed including their use as miniature membranes for characterization of transmembrane proteins in a native-like environment (Denisov and Sligar, 2017), as carrier vehicles for hydrophobic bioactive agents (Ryan, 2008) or contrast agents for molecular imaging (Kornmueller et al, 2019). Other potential large scale applications exist in agriculture. For example, ND technology may be used for treatment of plant fungal disease (Pérez-de-Luque et al, 2012) or as a water soluble delivery vehicle for hydrophobic pesticides. These applications require large scale, cost effective scaffold protein production methods to be feasible. The simplicity and relative ease of continuous foam fractionation for apolipoprotein processing at bioreactor production scale indicates this method offers an attractive strategy to increase yield, lower costs and improve the efficiency of downstream processing. Given the high cost of recombinant apolipoproteins from commercial sources, there is a growing need to develop new strategies to produce these unique proteins. Acknowledgements This work was supported by a grant from the National Institutes of Health (R37 HL64159), a Pre-Doctoral Fellowship from the American Heart Association to CAF and Service Award from the Nevada Proteomics Center. The Nevada Proteomics Center is supported in part by the Nevada INBRE, a grant from the National Institute of General Medical Sciences within the National Institutes of Health. The authors thank Dr. Patricia Ellison for access to the FPLC instrument and Dr. David Quilici for advice and assistance with mass spectrometry analysis. Abbreviations: apoLp-III apolipophorin III apo apolipoprotein DAG diacylglycerol LDLp low density lipophorin IPTG β-D-thiogalactopyranoside DMPC dimyristoylphosphatidylcholine PBS phosphate buffered saline OD optical density Figure 1: Schematic diagram of bioreactor foam fractionation process. Panel A) bioreactor chamber prior to the addition of IPTG. Panel B) bioreactor chamber following addition of 1mM IPTG to induce apoLp-III expression. Subsequently, a dense foam is produced causing an increase in internal bioreactor pressure. This pressure pushes accumulating foam out a collection tube and into in a separate collection vessel. Figure 2. SDS-PAGE of induced culture media and foamate. Aliquots of culture medium were collected at the time of induction and every hour thereafter for 5 h. During this period foam was collected and collapsed into foamate by centrifugation. Aliquots of each sample were applied to the lanes of a 4–20 % acrylamide gradient SDS-PAGE gel. Right side depicts the migration of molecular weight standards. The band for apoLp-III is prominent at 17 kDa. Figure 3. SDS-PAGE analysis of gel filtration column fractions. Processed apoLp-III foamate was subjected to Sephadex G-50 gel filtration chromatography with collection of 2.5 ml fractions. Aliquots of indicated fractions, spanning the elution profile, were applied to lanes of a 4–20% acrylamide gradient gel and subjected to SDS-PAGE. Figure 4. FPLC analysis of processed apoLp-III. A 400 μL aliquot of apoLp-III from Sephadex G-50 elution fraction 46 was analyzed by FPLC. The sample was applied to a Superdex 200 increase 10/300 column and eluted with PBS at 0.75 mL/min, with continuous monitoring of absorbance at 280 nm. Figure 5. rHDL formation and characterization. Panel A) Two 5 mg aliquots of DMPC were dispersed in PBS by vortexing. Subsequently, 200 μL PBS (Tube 1) or 200 μL PBS containing 2.0 mg apoLp-III (Tube 2) was added. Subsequently each sample was bath sonicated at 24 °C for ~15 min. Panel B) FPLC size exclusion chromatography of 200 μL of Tube 2 from panel A. The sample was applied to Superdex 200 increase 10/300 column and eluted with PBS at 0.75 mL/min, with continuous monitoring of absorbance at 280 nm. Table 1. Optimized bioreactor parameters for apoLp-III expression Parameter Value Seed Culture 5% vol/vol Culture pH 6.8 Culture temperature 37°C Induction OD600 ~7.0 Dissolved O2 50% Impeller speed 400–600 rpm ApoLp-III yield 0.2 g/L Table 2. Mass spectrometry analysis of apoLp-IIIa Bioreactor Sequence # spectra Modifications Avg mass (observed) Avg mass (calculated) Theoretical mass Start Residueb Stop Residue (K)IAEVTTSLK(Q) 10 none 481.2821 960.5496 960.5380 55 63 (K)IAEVTTSLKQEAEK(H) 2 none 773.9197 1545.8249 1545.8140 55 68 (K)EAAANLQNSIQSAVQKPAN(−) 1 none 977.5044 1952.9943 1952.9810 146 164 Shaker Flask Sequence # spectra Modifications Avg mass (observed) Avg mass (calculated) Theoretical mass Stop Start (K)IAEVTTSLK(Q) 29 none 481.2819 960.5492 960.5380 55 63 (K)IAEVTTSLKQEAEK(H) 5 none 773.9203 1545.8258 1545.8140 55 68 (K)EAAANLQNSIQSAVQKPAN(−) 4 none 977.5031 1952.9917 1952.9810 146 164 a Samples were digested with trypsin and analyzed by liquid chromatography / mass spectrometry as described in Materials and Methods. b Residue numbering corresponds to mature apoLp-III isoform a (Van der Horst et al, 1991). Highlights Apolipophorin III (apoLp-III) is a apolipoprotein with biosurfactant properties Recombinant apoLp-III with a pelB leader sequence is secreted from E. coli In a bioreactor, apoLp-III combines with gas and liquid in media to produce foam Foam fractionation can be used to streamline downstream processing of apoLp-III ApoLp-III has numerous biotechnology applications as a nanodisks scaffold This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Credit Author Statement Foam fractionation of a recombinant biosurfactant apolipoprotein (JBIOTEC-D-21–00771) Kyle Lethcoe: Methodology, data curation, figure preparation, manuscript editing Colin A. Fox: FPLC analyses, data interpretation, Table preparation, manuscript writing and editing Robert O. Ryan: Conceptualization, experimental design, data interpretation, manuscript writing, reviewing and editing Declaration of interests The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:, Robert O. Ryan reports administrative support was provided by University of Nevada Reno. No conflict to declare. References Beuker J , Steier A , Wittgens A , Rosenau F , Henkel M , Hausmann R , 2016. Integrated foam fractionation for heterologous rhamnolipid production with recombinant Pseudomonas putida in a bioreactor. AMB Express, 6 , 11.26860613 Borch J , Hamann T , 2009. The nanodisc: a novel tool for membrane protein studies. Biol. Chem 390 , 805–114.19453280 Breiter DR , Kanost MR , Benning MM , Wesenberg G , Law JH ,Wells MA , Rayment I , Holden HM , 1991. Molecular structure of an apolipoprotein determined at 2 5-Å resolution Biochemistry 30 ,603–608.1988048 Burghoff B (2012) Foam fractionation applications. J. Biotechnol 161 , 126–137.22484126 Denisov IG , Sligar SG , 2016 Nanodiscs for structural and functional studies of membrane proteins. Nat. Struct. Mol. Biol 23 , 481–486.27273631 Desai JD & Banat IM , 1997. Microbial production of surfactants and their commercial potential. Microbiol. Mol. Biol. Rev 61 , 47–64.9106364 Fischer NO , Blanchette CD , Segelke BW , Corzett M , Chromy BA , Kuhn EA , Bench G , Hoeprich PD , 2010. Isolation, characterization, and stability of discretely-sized nanolipoprotein particles assembled with apolipophorin-III. PLoS One. 5 , e11643.20657844 Fisher CA , Wang J , Francis GA , Sykes BD , Kay CM , Ryan RO , 1997. Bacterial overexpression, isotope enrichment, and NMR analysis of the N-terminal domain of human apolipoprotein E. Biochem. Cell Biol 75 , 45–53.9192073 Fox CA , Moschetti A , Ryan RO , 2021. Reconstituted HDL as a therapeutic delivery device. BBA Mol. Cell Biol. Lipids 1866 , 159025. Hauser PS , Ryan RO , 2007. Expressed protein ligation using an N-terminal cysteine containing fragment generated in vivo from a pelB fusion protein. Protein Expr. Purif 54 , 227–233.17493830 Kornmueller K , Vidakovic I , Prassl R , 2019. Artificial High Density Lipoprotein Nanoparticles in Cardiovascular Research. Molecules 24 , 2829. Kawooya JK , Meredith SC , Wells MA , Kézdy FJ , Law JH , 1986. Physical and surface properties of insect apolipophorin III. J. Biol. Chem 261 , 13588–13591.3759983 Mahley RW , Innerarity TL , Rall SC Jr , Weisgraber KH , 1984. Plasma lipoproteins: apolipoprotein structure and function. J. Lipid Res 25 , 1277–1294.6099394 Narayanaswami V , Ryan RO , 2000. The molecular basis of exchangeable apolipoprotein function. Biochim. Biophys. Acta 1483 , 15–36 10601693 Pérez-de-Luque A , Cifuentes Z , Beckstead JA , Sillero JC , Avila C , Rubio J , Ryan RO 2012. Effect of amphotericin B nanodisks on plant fungal diseases. Pest Manag. Sci 68 , 67–74.21710554 Rahman KSM , Gakpe E , 2008. Production, characterisation and applications of biosurfactants - Review, Biotechnology, 7 , 360–370. Ryan RO 1990. Dynamics of insect lipophorin metabolism. J. Lipid Res 31 , 1725–39.2079599 Ryan RO (2008) Nanodisks: hydrophobic drug delivery vehicle. Expert Opin. Drug Deliv 5 , 343–351.18318655 Ryan RO (2010) Nanobiotechnology applications of reconstituted high density lipoprotein. J. Nanobiotechnology 8 , 28.21122135 Ryan RO , Schieve D , Wientzek M , Narayanaswami V , Oikawa K , Kay CM , Agellon LB , 1995. Bacterial expression and site-directed mutagenesis of a functional recombinant apolipoprotein. J. Lipid Res 36 , 1066–10672.7658154 Ryan RO , Van der Horst DJ , 2000. Lipid transport biochemistry and role in energy production. Ann. Rev. Entomol 45 , 233–260 10761577 Soulages JL , Salamon Z , Wells MA , Tollin G , 1995. Low concentrations of diacylglycerol promote the binding of apolipophorin III to a phospholipid bilayer: a surface plasmon resonance spectroscopy study. Proc. Natl. Acad. Sci. USA 92 , 5650–5654 7777564 Tsujita M , Wolska A , Gutmann DAP , Remaley AT , 2018. Reconstituted discoidal high-density lipoproteins: bioinspired nanodiscs with many unexpected applications. Curr. Atheroscler. Rep 20 , 59.30397748 Van der Horst DJ , Van Doorn JM , Voshol H , Kanost MR , Ziegler R , Beenakkers AM , 1991. Different isoforms of an apoprotein (apolipophorin III) associate with lipoproteins in Locusta migratoria. Eur. J. Biochem 196 , 509–517.2007409 Wan CPL , Chiu MH , Wu X , Lee SK , Prenner EJ , Weers PMM , 2011. Apolipoprotein-induced conversion of phosphtidylcholine bilayer vesicles into nanodisks. Biochim. Biophys. Acta 1808 , 606–613.21111706 Wang J , Narayanaswami V , Sykes BD , Ryan RO , 1998. Interhelical contacts are required for the helix bundle fold of apolipophorin III and its ability to interact with lipoproteins. Protein Sci. 7 , 336–341 9521109 Wang J , Sykes BD , Ryan RO , 2002. Structural basis for the conformational adaptability of apolipophorin III, a helix-bundle exchangeable apolipoprotein. Proc. Natl. Acad. Sci. USA 99 , 1188–1193.11818551 Weers PM , Wang J , Van der Horst DJ , Kay CM , Sykes BD , Ryan RO , 1998. Recombinant locust apolipophorin III: characterization and NMR spectroscopy. Biochim. Biophys Acta 1393 , 99–107.9714761 Weers PMM , Ryan RO , 2003. Apolipophorin III: a lipid-triggered molecular switch. Insect Biochem. Mol. Biol 33 , 1249–126 14599497 Weers PM , Ryan RO 2006. Apolipophorin III: role model apolipoprotein. Insect Biochem. Mol. Biol 36 , 231–40.16551537 Wientzek M , Kay CM , Oikawa K , Ryan RO , 1994. Binding of insect apolipophorin III to dimyristoylphosphatidylcholine vesicles. Evidence for a conformational change. J. Biol. Chem 269 , 4605–461 8308032 Wilson C , Wardell MR , Weisgraber KH , Mahley RW , Agard DA , 1991. Three dimensional structure of the LDL receptor-binding domain of human apolipoprotein E. Science 252 , 1817–182 2063194

PMC008xxxxxx/PMC8752505.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9107782 8548 Curr Biol Curr Biol Current biology : CB 0960-9822 1879-0445 34735794 8752505 10.1016/j.cub.2021.10.030 NIHMS1750945 Article Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla Stucynski Joseph A. 1 Schott Amanda L. 1 Baik Justin 12 Chung Shinjae 1 Weber Franz 13* 1 Department of Neuroscience, Perelman School of Medicine, Chronobiology and Sleep Institute, University of Pennsylvania, 3400 Civic Center Blvd, Philadelphia, PA 19104, USA 2 Present address: Inscopix, 2462 Embarcadero Way, Palo Alto, CA 94303, USA 3 Lead Contact AUTHOR CONTRIBUTIONS J.A.S., S.C., and F.W. conceived and designed the study. J.A.S. performed all optogenetic, chemogenetic, and fiber photometry experiments and analyzed all sleep data. A.L.S. performed optogenetic pilot experiments. J.B. built the setup for optogenetic sleep recordings including the software to run experiments and the setup for fiber photometry experiments. J.A.S. and F.W. analyzed the data and wrote the manuscript. * Correspondence: Franz Weber, fweber@pennmedicine.upenn.edu 3 11 2021 10 1 2022 03 11 2021 10 1 2023 32 1 3750.e6 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY The two major stages of mammalian sleep – rapid eye movement sleep (REMs) and non-REM sleep (NREMs) – are characterized by distinct brain rhythms ranging from millisecond to minute-long (infraslow) oscillations. The mechanisms controlling transitions between sleep stages and how they are synchronized with infraslow rhythms remain poorly understood. Using opto- and chemogenetic manipulation in mice, we show that GABAergic neurons in the dorsomedial medulla (dmM) promote the initiation and maintenance of REMs, in part through their projections to the dorsal and median raphe nuclei. Fiber photometry revealed that their activity is strongly increased during REMs and fluctuates during NREMs in close synchrony with infraslow oscillations in the sleep spindle band of the electroencephalogram. The phase of this rhythm influenced the latency and probability with which dmM activation induced REMs. Thus, dmM inhibitory neurons strongly promote REMs, and their slow activity fluctuations may coordinate the timing of REMs episodes with infraslow brain rhythms. Sleep REM sleep brain state regulation infraslow oscillations pmcINTRODUCTION Mammalian sleep comprises two major stages: rapid eye movement sleep (REMs) and non-REM sleep (NREMs). Both stages are characterized by distinct oscillations in the electroencephalogram (EEG) that typically reside in the sub-second range, such as slow waves, sleep spindles, and hippocampal θ oscillations.1–3 In addition to these fast oscillations, the sleep architecture is also shaped by much slower processes such as the ultradian NREM-REM cycle or infraslow rhythms operating on a minute timescale.4–7 During NREMs, the EEG exhibits a distinct infraslow (~0.02 Hz) oscillation in the sleep spindle (σ) band, which plays a prominent role in timing transitions from NREMs to wakefulness.5 However, the mechanisms by which infraslow oscillations interact with sleep circuits to coordinate brain state switches, particularly from NREMs to REMs, remain largely unknown. Neural circuits involved in regulating transitions from NREMs to REMs have been identified in the hypothalamus and brainstem, including pons and medulla.2,3,8–10 Within the ventral medulla, GABAergic neurons have been shown to strongly promote REMs, partly through inhibition of REMs-suppressing (REM-off) neurons in the ventrolateral periaqueductal gray (vlPAG) within the midbrain.11 In addition to the ventral medulla, the dorsomedial medulla (dmM) has also been implicated in REMs control.12–19 Electrophysiological recordings in rodents revealed REMs-active neurons in both the nucleus prepositus hypoglossi (PH) and the dorsal paragigantocellular reticular nucleus (DPGi),15,17 and pharmacological inhibition of DPGi neurons expressing adrenergic α2-receptors specifically suppressed REMs.13 Inhibitory neurons in the PH and DPGi are thought to promote REMs by inhibiting REM-off neurons in the pons and midbrain, located within the locus coeruleus (LC), dorsal raphe (DRN), and vlPAG.12,14,15,18–20 DPGi GABAergic neurons, and neurons within the DPGi that project to the vlPAG or LC, express high levels of the immediate early gene c-Fos after a deprivation-induced REMs rebound, suggesting a strong activation during REMs.12,18,19 The slow time course of c-Fos expression, however, lacks the temporal precision to resolve activity changes associated with fast brain state transitions. Electrical stimulation of the PH resulted in an increased amount of REMs, which could be reversed by simultaneous pharmacological inhibition of the LC,16 but non-specific activation of axons-of-passage causing this effect cannot be ruled out. Although these studies suggest an important role of dmM inhibitory neurons in REMs regulation, their precise role in initiating and maintaining REMs and the underlying circuit mechanisms are still not fully understood. Furthermore, since the oscillation in the EEG σ power influences the timing of awakenings from NREMs,5 an interesting question is whether this infraslow rhythm is also involved in timing transitions from NREMs to REMs, possibly by influencing the activity of REMs-regulatory circuits such as the dmM inhibitory neurons. In this study, we found that optogenetic activation of GABAergic, GAD2-expressing neurons in the dmM strongly promotes the initiation and maintenance of REMs, while opto- and chemogenetic inhibition reduced the amount of REMs. The REMs-promoting effect was in part mediated dmM neurons projecting to the dorsal and median raphe nuclei (DR/MRN). Fiber photometry revealed that dmM neurons are activated during REMs. During NREMs, their activity closely followed the infraslow oscillation in the EEG σ power. In addition, we found that this oscillation modulates the latency and probability of optogenetically induced REMs episodes. Our findings delineate the role of dmM inhibitory neurons in REMs control and suggest a mechanism by which infraslow brain rhythms contribute to the timing of NREMs to REMs transitions. RESULTS Optogenetic activation of dmM GAD2 neurons promotes REMs. To probe the role of dmM GABAergic neurons in sleep-wake control, we injected Creinducible adeno-associated viruses (AAVs) expressing channelrhodopsin-2 fused with enhanced yellow fluorescent protein (AAV-EF1α-DIO-ChR2-eYFP) into the dmM of GAD2-Cre mice (Figure 1A). ChR2-eYFP was consistently expressed in the PH and DPGi across mice and to a lesser extent in neighboring areas including the medial vestibular nucleus (MV) and nucleus of the solitary tract (NST) (Figures 1A,B and S1). The optic fibers for laser stimulation were consistently placed on top of the PH (Figure S2A). While monitoring the animal’s brain state using EEG and electromyogram (EMG) recordings, laser stimulation (10 Hz, 120 s per trial) was applied randomly every 10 to 20 minutes (Figure 1C; Video S1; Methods). To quantify the laser-induced effect, we compared the percentage of each brain state during the 120 s laser interval with that during the preceding 120 s baseline interval (Methods). We found that optogenetic activation of dmM GAD2 neurons strongly increased the percentage of REMs (Figures 1D, S2B; mean difference [MD] = 26.04%, 95% confidence interval [CI] (18.64%, 33.46%), P < 0.0001; bootstrap, n = 9 mice) and the ratio of REMs to total sleep (NREMs + REMs) (Figure S2B). The percentage of NREMs was reduced by laser stimulation (MD = −34.99%, CI(−42.63%, −26.92%), P < 0.0001), while that of wakefulness was elevated (MD = 8.95%, CI(4.36%, 13.50%), P < 0.0001). In control mice expressing eYFP, laser stimulation had no significant effect (Figure S2B,C), and the laser-induced changes in brain states were significantly different between ChR2 and eYFP mice (Figure S2B bottom). Consistent with the strong REMs-promoting effect, the mean EEG spectrogram averaged across laser trials displayed a distinct increase in both the θ (6 - 9 Hz) and γ (55 - 90 Hz) power with a concomitant reduction in the δ (0.5 - 4.5 Hz) and σ (10 - 15 Hz) power during laser stimulation (Figures 1E, S2G; Methods). In eYFP mice, laser stimulation had no effect on the spectrogram (Figure S2F,G), and the laser-induced changes in each power band significantly differed between ChR2 and eYFP mice (Figure S2G bottom). Next, to disentangle whether the changes in the percentage of a specific brain state during laser stimulation are due to changes in the induction or maintenance of that state, we quantified how the laser affected the transition probability between each pair of brain states. For quantification, we compared the transition probabilities computed for consecutive 10 s epochs during the laser interval with those during the preceding 120 s baseline interval (Methods). Activation of dmM neurons caused a strong increase in the probability of NREM→REM transitions (Figures 1F, S2L; MD = 0.11, CI(0.10, 0.13), P < 0.0001, bootstrap, n = 9 mice). Consistent with this, the percentage of NREMs episodes that transitioned to REMs was increased from 2.29% (CI(0.58%, 4.35%)) during baseline without laser to 58.01% (CI(43.91%, 71.21%)) during laser stimulation (Figure S2D left). In control mice, there was no significant effect and the increase in the percentage significantly differed between ChR2 and eYFP mice. In addition, laser stimulation enhanced REM→REM transitions (MD = 0.094, CI(0.036, 0.166), P = 8.0e-4), and reduced REM→Wake transitions (MD = −0.096, CI(−0.17, −0.037), P = 8.0e-4), indicating that dmM GAD2 neuron activation also maintains REMs. In rare instances, laser stimulation also led to Wake→REM transitions (Figure S2K,L). Optogenetic activation also enhanced Wake→Wake transitions (Figures 1F, S2L; MD = 0.12, CI(0.082, 0.15), P < 0.0001) and reduced Wake→NREM transitions (MD = −0.12, CI(−0.16, −0.092), P < 0.0001). Consistent with this wake-maintaining effect, the mean duration of wake bouts originating during laser stimulation was longer than wake bouts during baseline without laser or wake bouts overlapping with laser in eYFP mice (Figure S2E). The probability of NREM→Wake transitions, however, was unchanged (MD = −0.0045, CI(−0.018, 0.0097), P = 0.53; bootstrap) and the percentage of NREMs episodes followed by wakefulness was also not altered by laser stimulation (Figure S2D right). This suggests that the increase in wakefulness during laser stimulation (Figure 1D) was not the result of a direct induction of wakefulness, but rather the result of the wake-maintaining effect and an indirect consequence of the laser-induced increase in REMs, since in rodents REMs is typically followed by wake episodes. Next, we tested to what extent the effect of laser stimulation on the EEG and EMG depended on the brain state of the animal. The power spectral density of the EEG during laser-induced REMs episodes was indistinguishable from that during spontaneous REMs, without significant changes in the δ, θ, or σ power (Figures 1G and S2I). In contrast, laser stimulation reduced the δ power during both NREMs and wakefulness and increased the NREMs θ power. We observed similar laser-induced changes in the δ and θ power during NREMs and wake when comparing ChR2 with eYFP mice and laser stimulation had no effect in eYFP mice (Figures S2H, S2I bottom). Finally, we found that optogenetic activation of dmM GAD2 neurons increased the EMG amplitude during wakefulness, but had no effect during REMs or NREMs (Figure S2J). Optogenetic activation of dmM GAD2 neurons enhances the intermediate stage. In rodents, REMs is preceded by a transition stage from NREMs to REMs (intermediate stage, IS), characterized by a gradual increase in θ power, a decay in δ power, and the presence of sleep spindles21–23. As stimulation of dmM GAD2 neurons during NREMs caused changes in the EEG resembling features of REMs – reduced δ and increased θ power – we rescored sleep to test whether activation of these neurons promotes IS (Methods). During IS the δ, θ, and σ power lay in between the values for REMs and NREMs (Figure S3A,B), did not differ between ChR2 and eYFP mice (Figure S3C,D), and were not changed by laser stimulation in ChR2 mice (Figure S3E right, S3F middle). Following 4-stage scoring to include IS, the differences in the δ and θ power between NREMs episodes with and without laser were reduced compared with those for 3-stage scoring, while the change in σ power was increased (Figure S3G). Optogenetic activation of dmM GAD2 neurons strongly increased the percentage of IS during laser stimulation (Figure S3H,I). The mean latency of laser-induced IS episodes (21.90 s, CI(17.39 s, 27.14 s)) was shorter than that for REMs episodes (44.44 s, CI(37.60 s, 51.95 s); MD = 22.54 s, CI(16.16 s, 28.92 s), P = 3.8e-5, T = 8.15, paired t-test, n = 9 mice). In eYFP mice, the percentage of IS was not significantly changed by laser stimulation, and the changes in IS significantly differed between ChR2 and eYFP mice (Figure S3I bottom). The average duration of laser-induced IS episodes was slightly longer than spontaneous episodes and episodes in control mice (Figure S3J). In addition to inducing IS, laser stimulation also increased the percentage of IS episodes that successfully transitioned to REMs during the laser interval from 62.12% in eYFP to 80.46% in ChR2 mice (Figure S3K). Closed-loop manipulation of dmM GAD2 neurons during REMs. To directly test the role of dmM GAd2 neurons in REMs maintenance, we applied an optogenetic closed-loop stimulation protocol (Figure 2A,B). The animal’s brain state was classified based on real-time analysis of the EEG/EMG recordings (Methods). As soon as the onset of a spontaneous REMs episode was detected, laser stimulation was initiated and stayed on until the episode ended. The laser was turned on randomly for only 50% of the detected REMs episodes. We found that ChR2-mediated activation of dmM GAD2 neurons prolonged REMs episodes (Figure 2C, Table S1). Stimulation in eYFP mice had no effect, and the laser-induced changes in REMs duration consequently differed between ChR2 and eYFP mice (Figure 2C). To test whether inactivation of dmM GAD2 neurons shortens the duration of REMs episodes, we expressed the blue light-activated bistable chloride channel SwiChR++ (AAV-EF1α-DIO-SwiChR++-eYFP) in these neurons (Methods).24 For closed-loop inhibition, we delivered 2 s step pulses (473 nm) at 30 s intervals throughout REMs for 50% of the detected episodes. REMs periods with laser were significantly shorter in SwiChR++ mice than episodes without laser in the same animals and episodes with laser in eYFP mice (Figure 2D, Table S1). Altogether, these results suggest that dmM GAD2 neuron activity contributes to the maintenance of REMs. As shown in multiple mammalian species, the interval between two successive REMs periods (inter-REM interval) is positively correlated with the preceding REMs duration.25–28 This correlation provides evidence for a homeostatic process regulating the ultradian timing of REMs episodes,26,29–31 in which homeostatic pressure for REMs building up during the inter-REM interval is dissipated during each REMs episode in proportion to its duration. In line with this notion, we found that extending the REMs duration through closed-loop activation resulted in a longer total duration of NREMs during the subsequent inter-REM interval, while the duration of wakefulness was not significantly altered by the laser (Figure 2E, Table S1). Sustained inhibition of dmM GAD2 neurons reduces the amount of REMs. To test how sustained inhibition of dmM GAD2 neurons affects the regulation of REMs, we optogenetically inactivated dmM GAD2 neurons for an extended period of time using SwiChR++ (Figure 3A,B). For analysis, we compared the percentage of each brain state during baseline recordings without laser and recordings with laser stimulation (1 s step pulses at 50 s intervals for 3 h), which were acquired on separate days (Figure 3C, Methods). SwiChR++-mediated inhibition reduced the percentage of REMs (Figure 3D, Table 1) as well as the ratio of REMs to total sleep (Figure 3E, Table 1). The reduction in REMs was caused by a decrease in the REMs frequency (Figure 3F, Table 1), but not the bout duration (Figure 3G). Optogenetic inhibition did not have any significant effect on the percentages of NREMs or wakefulness (Figure 3D). In addition to optogenetic inhibition, we also chemogenetically inactivated dmM GAD2 neurons by expressing hM4D(Gi) (AAV-hSyn-DIO-hM4D(Gi)-mCherry) in the dmM of GAD2-Cre mice (Figure S4A,B) and administering clozapine-N-oxide (CNO) for inhibition or saline for control (Methods). Compared with saline, CNO injection decreased the percentage of REMs in hM4D(Gi) mice due to a reduction in REMs episode frequency (Figure S4C,D,F) and had no significant effect in control mice expressing mCherry (Figure S4D). The effect of CNO on REMs was also significantly different between the mCherry- and hM4D(Gi)-expressing mice. There was a significant main effect of the drug on the amount of wakefulness, but the interaction between drug and virus was not significant. Despite an increase in NREMs in hM4D(Gi) mice, the ratio of REMs to total sleep was reduced (Figure S4E). The differences in how chemo- and optogenetic inhibition affected NREMs and wake may be caused by differences in the mechanisms by which the receptor hm4D(Gi) and ion channel SwiChR++ suppress neural activity.32 Additionally, we cannot exclude non-specific effects of CNO or its metabolite clozapine on NREMs or wakefulness,33 possibly reflected in the non-significant interaction between the drug and virus observed for the wake percentage, despite a significant main effect of the drug (Figure S4D right). But importantly, both methods consistently lowered the total duration of REMs and its ratio relative to total sleep, suggesting that the activity of dmM GAD2 neurons functionally contributes to regulating the amount of REMs. Activation of DR/MRN-projecting dmM GAD2 neurons specifically promotes REMs. We hypothesized that the REMs-promoting and wake-maintaining effects observed for dmM GAD2 neuron activation are mediated by different subpopulations within the dmM, and that the REMs-promoting subpopulation projects to areas containing REMs-suppressing (REM-off) neurons. To identify downstream areas receiving projections from dmM GAD2 neurons, we performed anterograde tracing using ChR2-eYFP. We observed strong projections to brainstem areas involved in brain state regulation including the LC, sublaterodorsal nucleus (SLD), the DR/MRN at the midbrain/pons boundary, and comparably sparse projections to the hypothalamus including the dorsomedial hypothalamus (DMH) (Figures 4A, S5A). A recent study has shown that tonic optogenetic activation of serotonergic neurons in the DRN strongly suppresses REMs, while maintaining NREMs.34 To specifically test the role of dmM GAD2 neurons projecting to the DR/MRN in REMs regulation, we injected AAVs with high retrograde efficiency (AAVrg-EF1α-DIO-ChR2-eYFP)35 into the DR/MRN (Figure 4B) and optogenetically stimulated the retrogradely labeled subpopulation of GAD2 neurons within the dmM. The DR/MRN-projecting dmM (dmM→DR/MRN) GAD2 neurons were localized adjacent to the 4th ventricle within the PH (Figure 4B,C). Optogenetic activation of the dmM→DR/MRN neurons (Figure 4D; 10 Hz, 120 s per trial) strongly increased the percentage of REMs during laser stimulation (Figures 4E, S5C top; MD = 18.87%, CI(12.58%, 24.65%), P < 0.0001; bootstrap, n = 9 mice) as well as the ratio of REMs to total sleep (Figure S5C top). The effect on the percentage of REMs was comparable to that found with stimulation of the whole dmM GAD2 population (Figure S5C bottom; Methods). Consistent with an enhanced probability of NREM→REM transitions (Figures 4G, S5I; MD = 0.069, CI(0.060, 0.077), P < 0.0001; bootstrap), the proportion of NREMs episodes that were followed by REMs was elevated during laser stimulation compared with that during baseline (Figure S5D left) and there was no significant effect of the stimulated dmM population on the percentage. In contrast, the proportion of NREMs episodes transitioning to wakefulness was not changed by laser stimulation and did not differ between the two dmM populations (Figure S5D right). The second factor contributing to the increase in REMs was an elevated REM→REM transition probability (Figures 4G, S5I; MD = 0.083, CI(0.043, 0.13), P < 0.0001). To directly probe the REMs-maintaining effect, we performed closed-loop stimulation of the dmM→DR/MRN GAD2 neurons and found that optogenetic activation extended the duration of REMs episodes (Figure S5K). The effect did not significantly differ between the two dmM populations (Figure S5K bottom). Similar to dmM GAD2 stimulation, closed-loop stimulation of dmM→DR/MRN neurons during REMs was associated with an increased total duration of NREMs during the subsequent inter-REM interval (Figure S5L left), while there was no effect on the total duration of wakefulness (Figure S5L right). Consistent with the REMs-promoting effect, optogenetic stimulation of the dmM→DR/MRN neurons also caused a distinct increase in the θ and γ power in the laser trial averaged EEG spectrogram with a concomitant reduction in the δ and σ range (Figures 4F, S5F). Contrary to stimulation of the whole dmM GAD2 population, activation of the dmM→DR/MRN GAD2 neurons did not alter the percentage of wakefulness (Figures 4E, S5C top; MD = −0.32%, CI(−4.07%, 3.38%), P = 0.87; bootstrap), and the change in the wake percentage consequently differed for stimulation of dmM and dmM→DR/MRN GAD2 neurons (Figure S5C bottom). Accordingly, in contrast to dmM GAD2 neuron stimulation, the Wake→Wake transition probability was not changed by dmM→DR/MRN GAD2 neuron activation (Figures 4G, S5J; MD = 0.013, CI(−0.020, 0.046), P = 0.42). Additionally, the duration of wake episodes initiated during laser stimulation was not changed, contrary to the increased wake bout duration found for ChR2-activation in dmM GAD2 mice (Figure S5E). These findings indicate that, in contrast to dmM GAD2 neurons, activation of the subpopulation of DR/MRN-projecting GAD2 neurons does not maintain wakefulness and thus specifically promotes REMs. Spectral analysis of the EEG revealed that during NREMs, optogenetic activation of dmM→DR/MRN GAD2 attenuated the δ power, while enhancing the θ power (Figures 4H, S5G). Consistent with these changes in the NREMs EEG, optogenetic activation increased the percentage of IS during the laser interval (Figure S5M,N); the magnitude of the effect was comparable to that for dmM GAD2 neuron stimulation (Figure S5N bottom). Thus, similar to dmM GAD2 neurons, the increase in REMs caused by the activation of dmM→DR/MRN GAD2 neurons was accompanied by an increase in IS. dmM GAD2 neurons are activated during REMs and synchronized with σ power during NREMs. To monitor the dynamics of dmM GAD2 neurons during spontaneous sleep, we injected Cre-dependent AAVs encoding the fluorescent calcium indicator GCaMP6s (AAV-Syn-Flex-GCaMP6s) into the dmM of GAD2-Cre mice and implanted an optic fiber for fiber photometry (Figures 5A,B, S6A; Video S2).36 The activity of dmM GAD2 neurons was significantly modulated by the brain state (Figure 5C, Table S2). The average activity during REMs was significantly higher than during NREMs (paired t-test with Bonferroni correction, MD = 11.03%, CI(5.29%, 16.76%), P = 0.013, T = 4.94), but did not differ from wake (P = 0.25). Individually, the activity significantly varied with the brain state in each mouse (one-way ANOVA, P < 0.05), and in 5 out of the 6 recorded animals, the dmM GAD2 neurons were most active during REMs (Tukey post-hoc test, P < 0.05); in the remaining mouse, the activity during both REMs and wake was higher than during NREMs. Analyzing the calcium changes at brain state transitions, we found that the activity of dmM GAD2 neurons during NREMs started increasing 35 s before the onset of REMs (Figure 5D; P < 0.05, paired t-test with Bonferroni correction, n = 62 transitions; Methods). Their activity remained elevated throughout REMs and began significantly decreasing 5 s before REMs ended (P < 0.05, n = 63 transitions). NREM→Wake transitions were associated with an initial increase in the activity (35 s before the transition, P < 0.05, n = 381 transitions), but then decayed before the actual brain state switch, consistent with our finding that optogenetic activation of dmM GAD2 neurons does not induce NREM→Wake transitions (Figure 1F). For Wake→NREM transitions, the activity was significantly reduced 15 s before the transition (n = 189 transitions). Interestingly, during NREMs the activity of the dmM GAD2 neurons rhythmically fluctuated on an infraslow timescale of tens of seconds (Figure 5B). A recent study has shown that the EEG displays a salient oscillation in the σ power on a similar timescale.5 We hypothesized that the dmM activity rhythm may follow this infraslow oscillation and indeed found that the σ power and calcium activity were strongly positively correlated during NREMs (Figures 5B, S6B,C). Compared with REMs and wake, the correlation between the two signals was strongest during NREMs (Figure S6B), and the power spectral densities of the two signals during NREMs peaked at comparable frequencies (Figure S6E). During NREMs, the cross correlation of the two signals was maximal at 0.75 s (CI(0.5 s, 0.96 s)), indicating that the dmM activity followed the σ power with a short time delay (Figure 5E). Compared with the δ and θ power, the correlation during NREMs was strongest for the σ power (Figure 5E, Table S2). In contrast, during wakefulness, the σ power and ΔF/F activity were anti-correlated (Figure S6B,C) and the cross-correlation accordingly exhibited a negative peak at 0.58 s (CI(0.0 s, 0.96 s); Figure S6D). Next, we analyzed how the ΔF/F activity evolved throughout single oscillation cycles of the σ power during NREMs. The ΔF/F activity exhibited a cosine-shaped activity modulation and closely matched the time course of the σ power (Figure 5F). The infraslow fluctuations in the dmM activity were particularly pronounced during consolidated NREMs (NREMs bouts with duration ≥ 120 s, only interrupted by microarousals (MAs), i.e. wake periods ≤ 10 s; Figure 5F, Table S2), but were also clearly present during shorter NREMs episodes (duration < 120 s, Figure S6F). Moreover, the correlation between the σ power and ΔF/F activity increased throughout single NREMs bouts (Figure S6G). The infraslow modulation likely contributes to the rise in activity before transitions to REMs and to the initial increase before the onset of wake (Figure 5D). To directly compare the activity preceding wakefulness with that before REMs on the infraslow timescale, we analyzed the time-normalized calcium signals starting from the last trough in σ power before the transition to wakefulness or REMs (Figure 5G). We found that the ΔF/F activity before the transition differed depending on whether the animal transitioned to wake or REMs (Table S2). In both cases, the activity was minimal at the preceding trough. But while the dmM activity displayed a cosine-like rise and decay in its activity before a wake onset, it remained elevated before a switch to REMs and finally reached its peak activity during REMs. In addition to the infraslow timescale, we analyzed whether the activity of dmM neurons is also modulated on the ultradian timescale (Figure S6H). We found that their activity during NREMs gradually increased throughout inter-REM intervals, while it decayed across wake episodes (Figure S6I). In accordance with the idea that longer REMs periods more strongly dissipate REMs pressure,26,28 the REMs duration was negatively correlated with the NREMs activity of dmM neurons during the following inter-REM interval (Figure S6J). In contrast, the wake activity was not correlated with the preceding REMs duration. The σ power oscillation modulates the delay and probability of optogenetically induced REMs. We showed that the infraslow oscillation in σ power strongly modulates the activity of dmM GAD2 neurons (Figure 5). Next, we tested whether this oscillation also plays a role in regulating the timing and probability of REMs episodes. In the optogenetic stimulation experiments of dmM and dmM→DR/MRN GAD2 neurons (Figures 1, 4), the delay between laser onset and the start of laser-induced REMs episodes considerably varied across trials (Figure 6A). We wondered whether the phase of the infraslow oscillation at laser onset may influence the REMs latency (Figure 6B). As the oscillation in σ power is particularly pronounced during consolidated NREMs (Figure 6C),5 we restricted the analysis to laser trials preceded by at least 120 s of NREMs (only interrupted by MAs). We found that the phase of the infraslow oscillation indeed modulated the delay between laser and REMs onset (Figure 6D, Table S3), while the dmM population had no effect on the delay. On average, the delay was shorter when the laser onset coincided with the falling phase of the σ power oscillation (0 to π rad.) than with its rising phase (−π to 0 rad.; Figure 6D, Table S3). In contrast to the delay, the probability with which dmM activation induced REMs was not significantly modulated by the phase (Figure 6E, Table S3), likely because the 120 s laser stimulation interval offered sufficient opportunity for the laser to eventually coincide with the optimal σ power range. We therefore shortened the laser stimulation interval to 50 s in a subsequent cohort of mice (Figure 6F). Optogenetic activation significantly increased the percentage of REMs during the laser interval (Figure S7A), and consolidated episodes of NREMs again exhibited a marked infraslow oscillation in the σ power (Figure S7B). The increase in REMs was less than for 120 s stimulation (Figure S7A bottom), allowing us to disentangle during which phases laser stimulation was most or least effective in inducing REMs. The probability that laser activation triggered REMs during the 50 s interval indeed depended on the phase of the σ power at laser onset (Figure 6H, Table S3). Stimulation was most likely to trigger REMs when the laser onset coincided with the falling phase of the infraslow rhythm (Table S3). As with 120 s stimulation, the latency of REMs periods was shortest during the falling phase (Figure 6G, Table S3). These findings demonstrate that optogenetic activation of dmM GAD2 neurons induces REMs with minimal latency and highest probability during the falling phase of the infraslow oscillation. For comparison, we also determined the preferred phases during which spontaneous transitions to REMs or IS occurred in eYFP control mice. The probability to observe a NREM→IS or IS→REM transition strongly depended on the phase (Figure S7C). NREM→IS transitions were more likely to occur during the falling phase of the infraslow oscillation, while the onset of REM preferentially coincided with its trough (−π to −π/2 rad. or π/2 to π rad.) (Figure S7C). Thus, the phase during which optogenetic activation of dmM GAD2 neurons most effectively triggered REMs coincided with the preferred phase of spontaneous NREM→IS transitions. DISCUSSION Combining opto- and chemogenetic manipulation with calcium imaging, we have identified an ascending medullary circuit in REMs regulation. Optogenetic activation of dmM GAD2 neurons greatly enhanced NREM→REM transitions and maintained REMs and wake episodes (Figures 1, 2), while opto- and chemogenetic inhibition reduced REMs (Figures 2, 3 and S4). Activation of dmM GAD2 neurons projecting to the DR/MRN enhanced REMs without affecting the duration of wake episodes, indicating that this subpopulation within the PH specifically promotes REMs (Figure 4). Calcium imaging using photometry demonstrated that the activity of dmM GAD2 neurons gradually increased throughout inter-REM and peaked during REMs (Figures 5 and S6). Combined with our findings from the opto- and chemogenetic experiments, this suggests that the endogenous activation of these neurons also promotes and maintains REMs during natural sleep. During NREMs their activity slowly fluctuated in close synchrony with the EEG σ power (Figures 5). The infraslow oscillation in σ power influenced the latency and probability with which optogenetic activation of dmM neurons triggered REMs (Figure 6), suggesting that it plays an important role in timing REMs episodes on the ultradian timescale. An ascending medullary circuit for REMs control. We found that activation of dmM GAD2 neurons projecting to the DR/MRN is sufficient to trigger REMs. A recent study showed that tonic optogenetic activation of serotonergic neurons in the DRN suppresses REMs, while maintaining NREMs.34 Photometry imaging, consistent with previous electrophysiological recordings,37 demonstrated that serotonergic neurons are least active during REMs. The ascending projections of the dmM GAD2 neurons to the DR may inhibit the serotonergic DRN neurons during REMs, and the suppression of these REM-off neurons may contribute to the induction and maintenance of REMs. In addition to the DRN and MRN, the LC also receives strong axonal projections from inhibitory neurons in PH and DPGi.19,20 Pharmacological inhibition of the LC suppressed the extension of REMs episodes elicited by electrical stimulation of the PH, suggesting that this projection contributes to REMs maintenance.16,38 As previous studies also demonstrated the importance of the DPGi in regulating REMs,13,15,19 it would be interesting to similarly test the specific role of DPGi inhibitory neurons using projection targets preferentially innervated by these neurons or by identifying specific molecular markers for these neurons.39,40 Stimulation of dmM GAD2 neurons strongly promoted θ oscillations during NREMs indicating that these neurons directly or indirectly interact with circuits involved in generating hippocampal θ oscillations. Pharmacological inhibition41–43 and lesions of the MRN44 are known to elicit θ oscillations, an effect likely mediated by serotonergic neurons.43 Postsynaptic inhibition of the MRN by the dmM may therefore contribute to the generation of θ oscillations during both IS and REMs. Current dynamical systems models posit that mutual inhibition between REM-on and REM-off neurons stabilizes the brain state in either NREMs or REMs.3,23,45,46 The slow increase in the activity of REM-on neurons during inter-REM, as observed in this study for dmM GAD2 neurons, or the slow decay of REM-off neurons in the vlPAG,47 may reflect the accumulation of REM pressure and may control when a NREM→REM transition occurs. Activation of REM-on neurons in a mutual inhibition network, either through the build-up of REMs pressure or optogenetic excitation, is expected to inhibit REM-off neurons, thus destabilizing the NREMs state, which eventually forces a transition to REMs. IS, with features of both NREMs and REMs, may reflect this unstable, transitional network regime.23,48 Continuous activation of dmM neurons throughout REMs in turn should stabilize it, resulting in longer REMs bouts as observed for closed-loop stimulation of dmM GAD2 neurons. The finding that closed-loop activation resulted in a longer NREMs duration during the following inter-REM interval, but did not extend the duration of wake bouts supports the idea that REMs pressure specifically accumulates during NREMs and not wakefulness.26 This may further explain why longer REMs episodes are negatively correlated with the activity of dmM neurons during the following NREMs, but not wakefulness. Infraslow oscillations and brain state regulation. The infraslow oscillation in σ power has been previously shown to modulate the arousability of mice during NREMs. When presented with acoustic stimuli, mice tended to sleep through when the stimulus coincided with the rising phase, but to wake up during the falling phase.5 Our findings suggest that the infraslow oscillation also plays an important role in the induction of REMs. During the falling phase, NREMs appears to be less stable and transitions out of NREMs are therefore more likely; awakenings can be triggered by sensory stimuli, while the endogenous activation of REMs-promoting circuits, such as the dmM GAD2 neurons, may result in the emergence of IS, followed by a transition to REMs. In contrast, during the rising phase, NREMs is less likely to be disrupted by stimuli or spontaneous state transitions. These protected periods of NREMs may be ideal for sleep-dependent processes such as memory consolidation or synaptic remodeling that require uninterrupted NREMs.4,49 Consistent with the strong contribution of sleep spindles to the σ power,50 multiunit activity in the thalamus, which is critical for spindle generation, is modulated on the infraslow timescale.51 In addition, the calcium activity in apical dendrites of cortical neurons is also tightly correlated with σ power oscillations,52 indicating that this oscillation influences the spontaneous activity in thalamocortical circuits. A recent study showed that enhancing the spindle rate through optogenetic activation of the thalamic reticular nucleus facilitates transitions to REMs.53 As the calcium activity of the dmM GAD2 neurons lags behind the cortical σ power (Figure 5), it is possible that dmM neurons receive direct or indirect inputs from the thalamocortical system, potentially driving their oscillatory activity during NREMs. We speculate that frontal cortical areas involved in spindle generation may relay neural activity related to sleep spindles through their extensive subcortical projections to REMs-regulatory neurons in the hypothalamus or brainstem.54,55 Thus, the dmM GAD2 neurons may be part of a brain-wide network extending well beyond the brainstem that coordinates NREM→REM transitions with infraslow fluctuations in neural activity. STAR METHODS RESOURCE AVAILABILITY Lead contact For information regarding materials used, experimental methods, and data analysis, direct all inquiries to the Lead Contact, Franz Weber (fweber@pennmedicine.upenn.edu). Materials availability This study did not generate any unique materials or reagents. Data and code availability All sleep recordings with opto- or chemogenetic manipulation and fiber photometry recordings have been deposited at Zenodo (https://zenodo.org) and are available as of the date of publication. All original code has also been deposited at Zenodo and is publicly available as of the date of publication. DOIs for data sets and code are listed in the key resources table. Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact upon request. EXPERIMENTAL MODEL AND SUBJECT DETAILS Mice All experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Pennsylvania and conducted in accordance with the National Institutes of Health Office of Laboratory Animal Welfare Policy. Experiments were performed in male or female GAD2-IRES-Cre mice (Jackson Laboratory stock no. 010802). Animals were housed on a 12-h dark/12-h light cycle (lights on between 7 am and 7 pm) and were aged 6-12 weeks at the time of surgery. All mice were group-housed with ad libitum access to food and water. METHOD DETAILS Surgical procedures. All surgeries were performed following the IACUC guidelines for rodent survival surgery. Prior to surgery, mice were given meloxicam subcutaneously (5 mg/kg). Mice were anesthetized using isoflurane (1 - 4%) and positioned in a stereotaxic frame. Animals were placed on a heating pad to maintain the body temperature throughout the procedure. Following asepsis, the skin was incised to gain access to the skull. For virus injection or implantation, a small burr hole was drilled above the dorsomedial medulla (anteroposterior (AP) −6.4 to −6.7 mm, mediolateral (ML) 0 mm). For optogenetic activation of dmM GAD2 neurons, 0.1 - 0.3 μl of AAV1-EF1α-DIO-hChR2-eYFP-WPRE-hGH (University of Pennsylvania Vector Core, RRID:Addgene_20298) was injected into the target area of GAD2-Cre mice using Nanoject II (Drummond Scientific) via a glass micropipette (dorsoventral (DV) −3.6 mm). For optogenetic inhibition experiments, 0.1 - 0.2 μl of AAV1-EF1α-DIO-SwiChR++-eYFP (University of North Carolina Vector Core, RRID:Addgene_55631) were injected (DV - 3.6 mm). For controls, we injected 0.1 - 0.3 μl of AAV2-EF1α-DIO-eYFP (University of Pennsylvania Vector Core, RRID:Addgene_27056) into the same area. To optogenetically stimulate dmM GAD2 neurons projecting to the DR/MRN, an additional burr hole was drilled (AP −4.8 mm, ML 0.0 to −0.25 mm) and 0.1 - 0.25 μl of AAVrg-EF1α-DIO-hChR2-eYFP-WPRE-HGHpA (Addgene, RRID:Addgene_20298) were injected into the DR/MRN (DV −3.2 to −3.4 mm). To localize the injection site, the micropipette was coated with red DiI (Invitrogen). After virus injection, an optic fiber (0.2 mm diameter) was inserted into the dmM (DV −3.4 to −3.5 mm). For fiber photometry experiments, 0.1 - 0.3 μl of AAV1-Syn-Flex-GCaMP6s-WPRE-SV40 (University of Pennsylvania Vector Core, RRID:Addgene_100845) were injected into the dmM and the optic fiber (0.4 mm diameter) was placed on top of the injection site (DV −3.3 to −3.5 mm). For chemogenetic inhibition experiments, 0.1 - 0.2 μl of AAV8-hSyn-DIO-hM4D(Gi)-mCherry (University of North Carolina Vector Core, RRID:Addgene_44362) were injected (DV −3.6 mm, ML 0 mm). Control mice were instead injected with AAV8-hSyn-DIO-mCherry (University of North Carolina Vector Core, RRID:Addgene_50459). EEG signals were recorded using stainless steel wires attached to two screws, one on top of the hippocampus and one on top of the prefrontal cortex. The reference screw was inserted on top of the left cerebellum. For EMG recordings, two stainless steel wires were inserted into the neck muscles. All electrodes, screws, connectors, and optic fibers were secured to the skull using dental cement. After injection and implantation were finished, bupivacaine (2 mg/kg) was administered at the incision site. For fiber photometry, opto-, and chemogenetic experiments, we excluded animals where no virus expression could be detected or where the virus expression extended below the dmM, as were mice in which the optic fiber tip was located below the virus expression site. Histology. Mice were deeply anesthetized and transcardially perfused with 0.1 M phosphate-buffered saline (PBS) followed by 4% paraformaldehyde in PBS. After removal, brains remained overnight in fixative and were then stored in 30% sucrose by volume in PBS solution for at least one night. After embedding and freezing, brains were sliced into 30 or 40 μm sections using a cryostat and mounted onto glass slides. For immunohistochemistry, brain sections were washed in PBS, permeabilized using PBST (0.3% Triton X-100 in PBS) for 30 minutes and then incubated in blocking solution (5% normal donkey serum in PBST, Jackson ImmunoResearch Laboratories, Inc., 017-000-001) for 1 hour. To stain eYFP-expressing axon fibers, brain sections were subsequently incubated with a chicken anti-GFP primary antibody (Aves Lab, GFP8794984, 1:1,000) diluted in the blocking solution for one night at 4°C. The next day, brain sections were washed in PBS and incubated for 2 hours with a species-specific secondary antibody conjugated with green Alexa fluorophore (Jackson ImmunoResearch Laboratories, Inc., 703-545-155, 1:500; donkey anti-chicken) diluted in PBS. The slices were washed with PBS followed by counterstaining with Hoechst solution (Thermo Scientific) and coverslipped with Fluoromount-G (Southern Biotechnic). Fluorescence images were taken using a fluorescence microscope. Virus expression analysis. To generate heatmaps of the virus expression across mice (Figures 1B, 3B, 4C, and S4B), coronal reference images were downloaded from Allen Mouse Brain Atlas for the corresponding AP coordinates (© 2015 Allen Institute for Brain Science). For a given AP reference section, the corresponding histology section from each mouse was overlaid and fitted to the reference. Regions in which eYFP- or mCherry-labeled cell bodies were present were then outlined manually. Custom Python programs then detected these outlines and determined for each location on the reference picture the number of mice with overlapping virus expression, which was encoded using different green color intensities. Polysomnographic recordings. Sleep recordings were performed in the animal’s home cage or in a cage to which the mouse had been habituated for 3 days, which was placed within a sound-attenuating box. All recordings were performed during the light phase between 8 am and 5 pm. For opto- and chemogenetic studies, EEG and EMG signals were recorded using an RHD2132 amplifier (Intan Technologies, sampling rate 1 kHz) connected to the RHD USB Interface Board (Intan Technologies). For fiber photometry, we used a TDT RZ5P amplifier (sampling rate 1.5 kHz). EEG and EMG signals were referenced to a common ground screw, placed on top of the cerebellum. Videos were recorded using a camera placed above the mouse cage. During the recordings, EEG and EMG electrodes were connected to flexible recording cables using a small connector. To determine the brain state of the animal, we first computed the EEG and EMG spectrogram with consecutive fast Fourier transforms (FFTs) calculated for sliding, half-overlapping 5 s windows, resulting in 2.5 s time resolution. Next, we computed the time-dependent δ, θ, σ, and high γ power by integrating frequencies in the range 0.5 to 4 Hz, 5 to 12 Hz, 12 to 20 Hz, and 100 to 150 Hz, respectively. We also calculated the ratio of the θ and δ power (θ/δ) and the EMG power in the range 50 to 500 Hz. For each power band, we used its temporal mean to separate it into a low and high part (except for the EMG and θ/δ ratio, where we used the mean plus one standard deviation as threshold). REMs was defined by high θ/δ ratio, low EMG, and low δ power. A state was set as NREMs if δ power was high, the θ/δ ratio was low, and EMG power was low. In addition, states with low EMG power, low δ, but high σ power were scored as NREMs. Wake encompassed states with low δ power and high EMG power and each state with high gamma power (if not otherwise classified as REMs). Finally, we manually rescored the automatic classification to correct any errors using a graphical user interface, visualizing the raw EEG, EMG signals, EEG spectrogram, EMG amplitude, and hypnogram. The software for automatic brain state classification and manual rescoring was programmed in Python. IS was scored manually based on visual inspection of the raw EEGs, the EEG spectrograms (binned in 2.5 s) and EMG. IS was identified by a gradual increase in θ power in the hippocampal EEG channel and a reduction in δ power, coinciding with the presence of sleep spindles in the prefrontal EEG channel. In the case that IS was followed by REMs, the offset of IS was marked by the disappearance of δ power and dominant θ activity in both EEG channels together with the absence of sleep spindles. In the case of a wake transition, the offset of IS coincided with a reduction in the EEG amplitude and an increase in EMG activity. Optogenetic manipulation. We performed optogenetic experiments 3 to 6 weeks after surgery. Animals were habituated for at least two days to the recording setup. After habituation, sleep recordings for optogenetics were performed during the light cycle (between 8 am and 6 pm) and lasted 6.5 hours on average. For optogenetic experiments, mice were tethered to an optic fiber patch cable in addition to the cable used for EEG/EMG recordings. For optogenetic open-loop stimulation, we repeatedly presented 10 Hz pulse trains (10 ms up, 90 ms down) lasting for 120 s generated by a blue 473 nm laser (4 - 6 mW, Laserglow). The inter-stimulation interval was randomly chosen from a uniform distribution ranging from 10 to 20 min. TTL pulses to trigger the laser were controlled using a Raspberry Pi, which in turn was controlled by a custom-programmed user interface programmed in Python. For optogenetic closed-loop stimulation, the program determined whether the animal was in REMs or not based on real-time spectral analysis of the EEG and EMG signals. The onset of REMs was defined as the time point where the EEG θ/δ ratio exceeded a hard threshold (mean + one std of θ/δ), which was calculated using previous recordings from the same animal. REMs lasted until the θ/δ ratio dropped below a soft threshold (mean of θ/δ) or if the EMG amplitude passed an offline calculated threshold. To compare REMs episodes with and without laser stimulation within the same recording session, the laser was turned on only for a randomly selected 50% of REMs episodes. For closed-loop activation, 10 Hz pulses (473 nm, 4 - 6 mW) were continuously presented throughout each REMs episode (that was selected for closed-loop stimulation). For SwiChR++-mediated closed-loop inhibition, we delivered 2 s step pulses (473 nm, 1 - 2 mW) every 30 s throughout REMs. For SwiChR++ mediated sustained inhibition of dmM GAD2 neurons, we continuously delivered 1 s step pulses at 50 s intervals for a total of 3 h. Baseline recordings without laser and recordings with laser stimulation were recorded on separate days. For each animal we collected two or three baseline and laser recordings each. Chemogenetic manipulation. On recording days, saline or CNO (5 mg/kg, clozapine N-oxide dihydrochloride, Tocris Bioscience) was injected intraperitoneally (i.p.) into GAD2-Cre mice expressing hM4D(Gi) or mCherry in the dmM. Injections occurred at 10:30 am. Each recording session began 30 min after injection and each animal contributed 2 saline and 2 CNO recording sessions to the data set. The order of injection days (saline vs CNO) was randomly assigned. For statistical analysis, we compared the impact of CNO or saline in hM4D(Gi) or mCherry-expressing mice on the brain state using mixed ANOVA with virus (mCherry vs hM4D(Gi)) and drug (saline vs CNO) as within and between factors, followed by pairwise t-tests with Bonferroni correction. Fiber photometry. Calcium imaging using fiber photometry was performed in mice freely moving in their home cages, placed within a sound-attenuating chamber. The implanted optic fiber was connected to a flexible patch cable. In addition, a flexible cable was connected to the implanted EEG/EMG electrodes via a mini-connector. For calcium imaging, a first LED generated the excitation wavelength of 465 nm and a second LED emitted 405 nm light, which served as control for bleaching and motion artifacts, as the emission signal from the 405 nm illumination is independent of the intracellular calcium concentration. The 465 and 405 nm signals were modulated at two different frequencies (210 and 330 Hz). Both lights were passed through dichroic mirrors before entering a patch cable attached to the optic fiber. Fluorescence signals emitted by GCaMP6s were collected by the optic fiber and passed via the patch cable through a dichroic mirror and GFP emission filter before entering a photoreceiver. Photoreceiver signals were relayed to a TDT RZ5P amplifier and demodulated into two signals using TDT’s Synapse software, corresponding to the 465 and 405 nm excitation wavelengths. For further analysis, we used custom-written Python scripts. First, both signals were low-pass filtered at 2 Hz using a 4th order digital Butterworth filter. Next, using linear regression, we fitted the 405 nm to the 465 nm signal. Finally, the linear fit was subtracted from the 465 nm signal (to correct for photo-bleaching or motion artifacts) and the difference was divided by the linear fit yielding the ΔF/F signal. To determine the brain state, EEG and EMG signals were recorded together with fluorescence signals using the RZ5P amplifier. All recordings were performed during the light phase between 8 am and 4 pm and lasted for 2 hours. We excluded fiber photometry recordings which contained sudden shifts in the baseline (likely due to a loose connection between optic fiber and patch cord), or with less than two REMs episodes (as we could not perform inter-REM interval analyses with these recordings). QUANTIFICATION AND STATISTICAL ANALYSIS Analysis of laser-induced brain state changes. To test whether optogenetic activation of dmM GAD2 neurons induced significant changes in the brain state, we compared for each state the mean percentages during the 120 s baseline interval before laser onset (−120 s to 0 s in Figure 1D, 4E) with the mean percentages during the 120 s laser interval (0 s to 120 s). P-values were calculated using bootstrapping: We resampled the complete data set comprising m trials from n mice, by randomly selecting with replacement m laser stimulation trials from the mice. During each of the 10,000 bootstrap iterations, we computed the difference between the mean percentages for the baseline and laser interval (across mice), resulting in a sampling distribution of the paired mean difference for each brain state. Using this distribution we then determined the equal-tail bootstrap p-value.5 The p-values were not corrected for multiple comparisons. The 95% confidence interval (CI) ranged from the 2.5th to the 97.5th percentile of the sampling distribution. To compare the effect of laser stimulation between different experimental groups of mice (ChR2 vs eYFP stimulation, or dmM vs dmM→DR/MRN stimulation), we determined for each experimental group the changes in the mean percentage (Δ percentage) of each state between the 120 s baseline and 120 s laser interval. Using bootstrapping, we then calculated the sampling distribution of the mean difference between Δ percentage in ChR2 and eYFP mice, which was used to determine the confidence interval and equal-tail bootstrap p-value. Analysis of ΔF/F activity relative to brain state changes. To calculate the activity of dmM GAD2 neurons relative to brain state transitions from state X to Y (Figure 5D), we first aligned the ΔF/F signals for all X→Y transitions from all mice relative to the time point of the transition (t = 0 s). Next, we ensured that for each NREM→Y transition, the preceding NREMs episode lasted for at least 100 s (only interrupted by MAs, wake episodes ≤ 10 s). In the case of REM→Wake and Wake→NREM transitions, the preceding REMs or Wake episode was at least 30 s long. To determine the time point at which the activity started increasing or decreasing, we used the first 10 s of state X as baseline. For all n X→Y transitions we averaged the z-scored activity during the baseline bin, resulting in n baseline values. We then subsequently compared this vector with the n time-averaged ΔF/F values computed for consecutive, non-overlapping 10 s bins using paired t-tests (the time point for a given 10 s bin was set to its midpoint). To account for multiple comparisons, we divided the significance level (α = 0.05) by the number of comparisons (Bonferroni correction). Transition analysis. For the transition analysis in Figures 1F and 4G, hypnograms were downsampled to 10 s epochs. All laser stimulation trials were first aligned to the laser onset (t = 0 s). To test whether laser stimulation induced a change in the probability of transitions from state X to Y (P(X→Y)), we calculated Markov transition probabilities for consecutive 10 s time bins for the 120 s interval preceding laser stimulation and the 120 s laser interval. More precisely, for each brain state pair (X, Y) we calculated the transition probability P(X→Y) by dividing the number of cases where state X in time bin ti was followed by state Y in bin ti+1 by the total number of bins in state X. The p-values were calculated using bootstrapping. (We used a non-parametric test for statistical testing as the distribution of the probability of transitions occurring with only low likelihood (e.g. NREM→REM transitions) is typically not normal.) For each of the 10,000 bootstrap iterations, we resampled the complete data set (m trials from n mice) by randomly selecting with replacement m laser stimulation trials from the n mice and then computing for each state pair (X, Y) the transition probabilities during the 120 s baseline and laser interval. By subtracting the laser from the baseline probability, we obtained a sampling distribution of the paired mean difference for each state pair (Figures S2L, S5I), which was used to calculate the confidence interval (ranging from the 2.5th to 97.5th percentile) and the equal-tail bootstrap p-value.56 In case all 10,000 sampled mean differences were consistently larger or smaller than 0, we set the p-value as P < 0.0001. The gray bars and black lines in Figures 1F and 4G depict the Markov transition probabilities and 95% CIs for 30 s intervals computed for consecutive 10 s time bins, whereas the red lines represent the Markov transition probabilities calculated for the 120 s baseline intervals. Spectral density and power estimation. The power spectral density of the EEG was computed using Welch’s method with Hanning window for consecutive 5 s, half overlapping intervals. To calculate the power within a given frequency band, we approximated the corresponding area under the spectral density curve using a midpoint Riemann sum. To compute the EMG amplitude, we also first calculated the power spectral density of the EMG and then integrated frequencies in the range 5 - 100 Hz. To test in Figures 1G and 4H whether laser stimulation during a specific brain state changed the spectral density (or EMG amplitude), we determined for each mouse the δ, θ, or σ power (or EMG amplitude) for that state with and without laser, which served as input for a paired t-test. To determine the time dependent power bands in Figures 1E, S2F, and 4F, we first computed the normalized EEG spectrogram using half-overlapping 5 s windows, normalized each frequency component in the spectrogram by its temporal mean, and then calculated for each frequency range and time point the corresponding Riemann sum. For better representation, we plotted the laser trials averaged spectrogram using a logarithmic frequency axis. To test whether laser stimulation significantly changed the normalized EEG power within a given frequency band, we computed for each animal the mean power during the 120 s laser interval and the preceding 120 s baseline interval, and tested whether these values were significantly different across animals using a paired t-test. For the EEG power spectral density analysis in Figure 4H and the EMG amplitude analysis in Figure S5H we excluded one recording due to artifacts in the EEG or EMG signal, respectively. These recordings were still included in the remaining analyses, because the artifacts did not confound sleep state scoring. Analysis of infraslow σ power oscillations. To calculate the power spectral density of the EEG σ power (Figures 6C, S6E, and S7B), we first calculated for each recording the EEG power spectrogram by computing the FFT for consecutive sliding, half-overlapping 5 s windows. Next, we normalized the spectrogram by dividing each frequency component by its mean power and calculated the normalized σ power by averaging across the spectral density values in the σ range (10 to 15 Hz). As the infraslow rhythm is most pronounced in consolidated NREMs bouts,5 we only considered NREMs bouts that lasted at least 120 s, possibly interrupted by MAs (wake periods ≤ 10s). We then calculated the power spectral density using Welch’s method with Hann window for each consolidated NREMs bout and averaged for each animal across the resulting densities. To determine the instantaneous phase of the infraslow rhythm, we smoothed the σ power using a 10 s box filter and band-pass filtered it in the range of 0.01 – 0.03 Hz using a 4th order digital Butterworth filter. Finally, we computed the phase angle by applying the Hilbert transform to the band-pass filtered σ power signal (see Figures 5B and 6B). Based on the phase, we could then isolate the beginning and end of single infraslow cycles to average the ΔF/F activity during single cycles (Figure 5F,G) or to determine the phase at the onset of laser stimulation in the optogenetic open-loop experiments (Figure 6). To calculate the ΔF/F activity during single cycles, we first downsampled the calcium signals to the same temporal resolution (2.5 s) as the σ power, and then normalized the cycle durations for averaging. To determine the ΔF/F activity during the last σ power cycle before NREM→Wake or NREM→REM transitions, we only included transitions preceded by at least 120 s of NREMs (possibly interrupted by MAs) and excluded trials without any distinct local minimum (< −1 rad.) corresponding to a trough in the instantaneous phase before the transition. To compute Pearson’s r between the ΔF/F signals and the normalized σ power (or other power bands) (Figures S6B), we correlated for each mouse all states scored as REMs, NREMs, or wake with the corresponding 2.5 s bins in the calcium signal. To calculate the cross correlation between the ΔF/F signal and the σ power (or other power bands) (Figure 5B,E and S6D), we first calculated for all NREMs bouts with duration ≥ 120 s (possibly interrupted by MAs) the σ power (s) from the EEG spectrogram, computed using consecutive 2.5 s windows with 80% overlap to increase the temporal resolution. We again normalized the spectrogram by dividing each frequency component by its mean power. Using the same overlapping binning, we downsampled the ΔF/F signal (d) to prevent any time lags resulting from differences in downsampling and then calculated the cross correlation of both signals. The cross correlation was normalized by dividing it by the product of the standard deviation of s and d and the number of data points in s. For each mouse we finally obtained the mean cross correlation by averaging across all NREMs bouts. Statistical tests. Statistical analyses were performed using the python modules scipy.stats (scipy.org) and pingouin (https://pingouin-stats.org).57 We did not predetermine sample sizes, but cohorts were similarly sized as in other relevant sleep studies.58,59 All statistical tests were two-sided. The significance of changes in brain state percentages or transition probabilities between brain states induced by laser stimulation were tested using bootstrapping. Otherwise, data were compared using t-tests or using ANOVA followed by multiple comparisons tests (pairwise t-tests with Bonferroni correction for mixed or one-way repeated-measures ANOVA and Tukey’s post hoc test for one-way ANOVA). Using the Shapiro-Wilk test we verified that the data were normally distributed. For repeated-measures and mixed ANOVA, Mauchly’s test was applied to check the sphericity of the data. Results in text and figures are represented as mean with 95% confidence intervals. For all tests, a (corrected) p-value < 0.05 was considered significant. The statistical results for the main figures are presented in the text, in Table 1, and in Tables S1–3. The test results related to supplemental figures are included in the supplemental figure legends. Supplementary Material Supplemental Material Video S1 Video S1. Optogenetic open-loop stimulation of dmM GAD2 neurons, related to Figure 1. Video S2 Video S2. Calcium activity of dmM GAD2 neurons during NREM→REM transition, related to Figure 5. ACKNOWLEDGEMENTS This work was supported by the National Institute of Health (NIH)/National Heart, Lung, and Blood Institute (NHLBI), R01HL149133, a NARSAD Young Investigator grant (#27799) by the Brain & Behavior Research Foundation, and by a grant from the Margaret Q. Landenberger Foundation to F.W. We thank J. Hong for help with sleep recordings and J. Smith and X. Li for help with histology. Figure 1. Optogenetic activation of dmM GAD2 neurons promotes REMs. (A) Schematic of optogenetic experiment. Left, coronal diagram of mouse brain indicating injection site and optic fiber placement. Gray rectangle, optic fiber. Right, fluorescence image of dmM in a GAD2-Cre mouse injected with AAV-DIO-ChR2-eYFP (green). Blue, Hoechst stain. Scale bar, 1 mm. (B) Outline of areas with ChR2-eYFP expressing cell bodies along the rostrocaudal axis within three coronal brain sections. The green color code indicates in how many mice the virus expression overlapped at the corresponding location (n = 9 mice). The coronal brain schemes were adapted from Allen Mouse Brain Atlas (© 2015 Allen Institute for Brain Science) PH, nucleus prepositus hypoglossi; DPGi, dorsal paragigantocellular nucleus; mlf, medial longitudinal fasciculus; MV, medial vestibular nucleus; NTS, nucleus of the solitary tract; XII, hypoglossal nucleus. (C) Example experiment. Shown are EEG spectrogram, EMG amplitude, and color-coded brain states. The blue patches indicate 120 s laser stimulation intervals (473 nm, 10 Hz). Two EEG and EMG raw traces are shown at an expanded timescale for the selected time points (dashed lines; scale bars, 1 s and 0.5 mV). PSD, power spectral density. (D) Percentage of REMs, NREMs, and wake before, during, and after laser stimulation (n = 9 mice). Shadings, 95% confidence intervals (CIs). (E) Impact of laser stimulation on the EEG spectrogram and different power bands. Top, laser trial averaged EEG spectrogram with logarithmic frequency axis. Each frequency component of the spectrogram was normalized by its mean power across the recording. Bottom, time course of δ (0.5 - 4.5 Hz), θ (6 - 9 Hz), σ (10 - 15 Hz), and γ power (55 - 90 Hz) before, during, and after laser stimulation. Shadings, 95% CIs. (F) Effect of laser stimulation of dmM GAD2 neurons on transition probabilities between brain states. Transition probabilities were calculated for hypnograms binned in 10 s epochs. Each bar represents the transition probability calculated for a 30 s interval. The red line depicts the baseline transition probability computed for the 120 s interval preceding laser stimulation. Error bars, 95% CIs. Bootstrap, ***P < 0.001. (G) Power spectral density of EEG during REMs, NREMs, and wake with and without (w/o) laser stimulation. Paired t-tests, **P < 0.01; ***P < 0.001. See also Figures S1, S2, S3 and Video S1. Figure 2. Closed-loop manipulation of dmM GAD2 neurons during REMs. (A) Schematic of closed-loop stimulation protocol. The brain state was continuously monitored; once a REMs episode was detected, laser stimulation was initiated and maintained throughout REMs for 50% of the detected episodes. (B) Example experiment. Shown are EEG spectrogram, EMG amplitude, and color-coded brain states. The blue patches indicate laser stimulation intervals (473 nm, 10 Hz). The arrows indicate a preceding REMs episode (REMpre) and the subsequent inter-REM interval. PSD, power spectral density. (C) Duration of REMs episodes (REMpre) with and without laser in ChR2 (n = 8) and eYFP mice (n = 10). † and ‡ indicate a significant main effect of laser and significant interaction between laser and virus, respectively, in mixed ANOVA; * indicates significance in pairwise t-tests with Bonferroni correction. **P < 0.01; ***,†††,‡‡‡P < 0.001. Bars, mean duration; error bars, 95% CIs; lines, individual mice. (D) Duration of REMs episodes with and without laser (473 nm, 2 s step pulse every 30 s) in SwiChR++ (n = 8) and eYFP mice (n = 5). Mixed ANOVA followed by pairwise t-tests with Bonferroni correction. *,†,‡P < 0.05. (E) Effects of laser stimulation during REMpre on the total NREMs and wake duration during the following inter-REM interval in ChR2 (n = 10) and eYFP mice (n = 8). Mixed ANOVA followed by pairwise t-tests with Bonferroni correction. *P < 0.05; ††,‡‡P < 0.01; ***P < 0.001. See also Table S1. Figure 3. Sustained inhibition of dmM GAD2 neurons reduces the amount of REMs. (A) Schematic of optogenetic inhibition experiment. Left, coronal diagram of mouse brain. Right, fluorescence image of dmM in a GAD2-Cre mouse injected with AAV2-DIO-SwiChR++-eYFP (green). Blue, Hoechst stain. Scale bar, 1 mm. (B) Heatmap outlining areas with cell bodies expressing SwiChR++-eYFP within three consecutive sections along the rostrocaudal axis, n = 7 mice. (C) Example hypnograms from two recordings (3 h) with and without laser. (D) Percent of time spent in REMs, NREMs, and wake during the 3 h laser and baseline recordings in mice expressing SwiChR++ (n = 7) and eYFP (n = 7). † and ‡ indicate a significant main effect of laser and significant interaction between laser and virus, respectively, in mixed ANOVA; * indicates significance in pairwise t-tests with Bonferroni correction. Bars, mean across mice; error bars, 95% CIs. **,‡‡P < 0.01; †††P < 0.001. (E) Effects of laser stimulation on the ratio of REMs to total sleep (REMs + NREMs) in SwiChR++ and eYFP mice. Mixed ANOVA followed by pairwise t-tests with Bonferroni correction. **P < 0.01; ***,†††,‡‡‡P < 0.001. (F) Effects of laser stimulation on REMs frequency. Mixed ANOVA followed by pairwise t-tests with Bonferroni correction. **,††,‡‡P < 0.01. (G) Effects of laser stimulation on REMs bout duration. See also Table 1 and Figure S4. Figure 4. Activation of DR/MRN-projecting dmM GAD2 neurons specifically promotes REMs. (A) Left, schematic sagittal brain section depicting injection of AAV-DIO-ChR2-eYFP into the dmM of a GAD2-Cre mouse. Right, fluorescence image of injection site showing expression of ChR2-eYFP in the dmM (bottom) and image showing axonal projections of dmM GAD2 neurons in midbrain and pons (top). Blue, Hoechst stain. Scale bars, 1 mm. vlPAG, ventrolateral periaqueductal gray; Aq, aqueduct; DRN, dorsal raphe nucleus; MRN, median raphe nucleus. (B) Schematic depicting AAVrg-DIO-ChR2-eYFP injection into the DR/MRN of a GAD2-Cre mouse and implantation of an optic fiber into the dmM for stimulation of retrogradely labeled neurons. Top, expression of ChR2-eYFP at the injection area. Bottom, retrogradely labelled neurons in the dmM. Magenta, pseudo-colored red DiI labeling the injection needle tract. Scale bars, 1 mm. (C) Expression of ChR2-eYFP in the dmM of GAD2-Cre mice injected with AAVrg-DIO-ChR2-eYFP into the DR/MRN. The green color code indicates in how many mice the virus expression overlapped at the corresponding location (n = 9 mice). (D) Example experiment. Shown are EEG spectrogram, EMG amplitude, and brain states. Two EEG, EMG raw traces during NREMs and REMs are represented on an expanded timescale for the selected time points (dashed lines; scale bars, 1 s and 0.5 mV). PSD, power spectral density. (E) Percentage of REMs, NREMs, and wake before, during, and after laser stimulation of dmM→DR/MRN GAD2 neurons (n = 9 mice). Shadings, 95% CIs. (F) Impact of laser stimulation on the EEG spectrogram and different power bands. Top, laser trial averaged EEG spectrogram with logarithmic frequency axis. Each frequency of the spectrogram was normalized by its mean power across the recording. Bottom, δ, θ, σ, and γ power before, during, and after laser stimulation. Shadings, 95% CIs. (G) Effect of dmM→DR/MRN GAD2 neuron stimulation on transition probabilities between brain states. Red line, baseline transition probabilities. Error bars, 95% CIs. Bootstrap, ***P < 0.001. (H) Power spectral density of EEG during REMs, NREMs, and wake with and without laser stimulation. Paired t-test; **P < 0.01, ***P < 0.001. See also Figure S5. Figure 5. dmM GAD2 neurons are activated during REMs and synchronized with σ power during NREMs. (A) Top, schematic of calcium imaging using fiber photometry. AAV-FLEX-GCaMP6s was injected into the dmM of GAD2-cre mice. Gray rectangle, optic fiber. Bottom, fluorescence image showing expression of GCaMP6s (green) in the dmM. Blue, Hoechst stain. Scale bar, 0.5 mm. (B) Top, example fiber photometry recording. Shown are EEG spectrogram, EMG amplitude, normalized σ power (10 - 15 Hz), color-coded brain states, and ΔF/F signal. PSD, power spectral density. Bottom left, the gray box indicates an interval for which the σ power (top) and ΔF/F signal (bottom) are shown at an expanded timescale. The filtered σ power (dashed line) was used to determine the phase of the σ power oscillation (middle). Bottom right, cross correlation between the calcium signal and σ power during NREMs for the entire example recording. (C) Average ΔF/F activity during REMs, wake, and NREMs. † indicates a significant effect of brain state in one-way repeated-measures ANOVA; * indicates significance in pairwise t-tests with Bonferroni correction; †,*P < 0.05. Bars, mean across mice; lines, individual mice; error bars, 95% CIs; n = 6 mice. (D) Calcium activity (ΔF/F, z-scored) at brain state transitions. For each transition from state X to Y, the mouse was in state X for at least 100 s (NREM→REM and NREM→Wake) or 30 s (REM→Wake and Wake→NREM). Shadings, 95% CIs. (E) Cross correlation between calcium activity and δ, θ, or σ power during NREMs. Shadings, 95% CIs. (F) Average σ power and calcium activity during a single cycle of the σ power oscillation. The analysis only includes NREMs episodes with duration ≥ 120 s (including MAs ≤ 10 s). Each σ power cycle was normalized in time, ranging from −π to π rad. Shadings, 95% CIs. (G) Calcium activity preceding NREM→REM or NREM→Wake transitions (wake also includes MAs). The interval from the preceding trough in σ power until the transition and the duration of the following REMs or wake episode were normalized in time. Shadings, 95% CIs. See also Figure S6, Video S2, and Table S2. Figure 6. The σ power oscillation modulates the delay and probability of optogenetically induced REMs. (A) Distribution of delay times between onset of laser stimulation and REMs episodes for 120 s open-loop stimulation of dmM and dmM→DR/MRN GAD2 neurons (data from Figures 1 and 4; n = 9 mice for each data set). Only trials preceded by consolidated NREMs (duration ≥ 120 s, possibly interrupted by MAs) were included in the analysis. (B) Example of a laser stimulation trial. Shown are the EEG spectrogram, brain states and σ power including its phase. The filtered σ power (dashed line) was used to determine the phase of the σ power oscillation at laser onset. (C) Power spectral densities of the σ power during consolidated NREMs dmM and dmM→DR/MRN mice. For each animal, the spectral density was normalized by its mean power. Shadings, 95% CIs. (D) Delay between laser and REMs onset depending on the phase of the σ power oscillation at laser onset for 120 s stimulation of dmM and dmM→DR/MRN GAD2 neurons. Phase values were divided into four equally sized bins. For all laser trials within a given bin, we calculated the average delay time between laser and REMs onset across mice. † indicates a significant main effect of the phase in mixed ANOVA. ††P < 0.01. Shadings, 95% CIs. (E) Probability of REMs induction during the 120 s stimulation interval in dependence of the σ power phase. For each mouse and phase bin, we divided the number of successful trials (where REMs was triggered during laser) by the sum of successful and unsuccessful trials for this bin to obtain the probability. Shadings, 95% CIs. (F) Percentage of REMs, NREMs, and wake before, during, and after 50 s laser stimulation of dmM GAD2 neurons (n = 9 mice). Shadings, 95% CIs. (G) Delay of REMs onset for 50 s stimulation of dmM GAD2 neurons. The delay was calculated only for REMs episodes initiated during the stimulation interval. † indicates a significant effect of the phase in one-way repeated-measures ANOVA. †††P < 0.001. (H) Probability of REMs induction for 50 s stimulation of dmM GAD2 neurons. A trial was counted as successful if REMs was triggered during the 50 s stimulation interval. ††P < 0.01. See also Figure S7, Table S3. Table 1. Statistical results for SwiChR++-mediated inhibition of dmM GAD2 neurons. Mixed ANOVA with virus (SwiChR++ vs eYFP) and laser as between and within factors, followed by pairwise t-tests with Bonferroni correction; n = 7 SwiChR++ and n = 7 eYFP mice. DOF, degrees of freedom; CI – 95% CI. Figure 3D – Effect of optogenetic inhibition on %REM ANOVA F DOF P ηp2 Laser x virus 17.33 (1,12) 0.0013 0.59 Main effect of laser 26.28 (1,12) 2.5e-4 0.69 t-tests MD CI P T SwiChR++: Laser – W/o laser −3.12% CI(−4.27%,−1.97%) 0.0011 6.62 Laser: SwiChR++ – eYFP −2.36% CI(−3.65%,−1.07%) 0.0036 −3.99   Figure 3E - Effect of optogenetic inhibition on REM/(REM + NREM) ANOVA F DOF P ηp2 Laser x virus 28.84 (1,12) 1,68e-4 0.71 Main effect of laser 65.05 (1,12) 3.0e-6 0.84 t-tests MD CI P T SwiChR++: Laser – W/o laser −4.41% CI(−5.0%, −3.0%) 1.1 e-4 10.17 Laser: SwiChR++ – eYFP −3.05% CI(−5.0%, −1.0%) 0.0044 −3.89   Figure 3F - Effect of optogenetic inhibition on REM frequency ANOVA F DOF P ηp2 laser x virus 13.80 (1,12) 0.0030 0.53 main effect of laser 15.80 (1,12) 0.0018 0.57 t-tests MD CI P T SwiChR++: Laser – W/o laser −1.64 h−1 CI(−2.37 h−1, −0.92 h−1) 0.0029 5.54 Laser: SwiChR++ – eYFP −1.25 h−1 CI(−1.94 h−1, −0.57 h−1) 0.0035 −4.00 KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Normal Donkey Serum Jackson ImmunoResearch Laboratories, Inc. Cat# 017-000-001, RRID:AB_2337254 Chicken polyclonal anti-GFP Aves Lab Cat# GFP1202, RRID:AB_2734732 Donkey anti-chicken, Alexa Fluor 488 Jackson ImmunoResearch Laboratories, Inc. Cat# 703-545-155, RRID:AB_2340375 Bacterial and virus strains AAV1-EF1α-DIO-hChR2-eYFP-WPRE-hGH University of Pennsylvania Vector Core Cat# 20298-AAV1, RRID:Addgene_20298 AAV2-EF1α-DIO-SwiChR++-eYFP University of North Carolina (UNC) Vector Core Cat# 55631, RRID:Addgene_55631 AAV2-EF1α-DIO-eYFP University of Pennsylvania Vector Core Cat# 27056-AAV1, RRID:Addgene_27056 AAVrg-EF1α-DIO-hChR2-eYFP-WPRE-HGHpA Addgene Cat# 20298-AAVrg, RRID:Addgene_20298 AAV1-Syn-Flex-GCaMP6s-WPRE-SV40 University of Pennsylvania Vector Core Cat# 100845-AAV1, RRID:Addgene_100845 AAV8-hSyn-DIO-hM4D(Gi)-mCherry UNC Vector Core Cat# 44362-AAV8, RRID:Addgene_44362 AAV8-hSyn-DIO-mCherry UNC Vector Core Cat# 50459-AAV8, RRID:Addgene_50459 Chemicals, peptides, and recombinant proteins Clozapine-N-Oxide dihydrochloride (50mg) Tocris Bioscience Cat# 4936 Hoechst stain Thermo Scientific Cat# 33342 Fluoromount-G slide mounting medium Southern Biotechnic Cat# OB10001 Invitrogen Vybrant Di-I cell labeling solution Invitrogen Cat# V22885 Deposited data All code generated for Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5527140 Fig 1 ChR2 open loop, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5519364 Fig 1 eYFP open loop, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5522796 Fig 2 ChR2 closed loop, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5507513 Fig 2 eYFP closed loop activation controls, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5519498 Fig 2 Swichr++ closed loop, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5519599 Fig 2 eYFP closed loop inactivation controls, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5523065 Fig 3 Swichr++ sustained inactivation, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5519726 Fig 3 eYFP sustained inactivation controls, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5519744 Fig S4 hM4D(Gi) chemogenetic inhibition, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5519678 Fig S4 mCherry controls, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5522825 Fig 4 AAVrg open loop, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5528959 Fig S5 AAVrg-ChR2 closed loop, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5519556 Fig 5 photometry, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5519450 Fig 6 ChR2 50 s open loop, in Regulation of REM Sleep by Inhibitory Neurons in the Dorsomedial Medulla This paper Zenodo: https://doi.org/10.5281/zenodo.5518204 Software and algorithms Python Programming Language https://www.anaconda.com/ RRID:SCR_008394 Synapse Tucker-Davis Technologies, https://www.tdt.com/component/synapse-software/ RRID:SCR_006495 RHD USB Interface Board Software http://intantech.com/downloads.html?tabSelect=Software RRID:SCR_019278 SciPy http://www.scipy.org/ RRID:SCR_008058 Pingouin 57, https://pingouin-stats.org https://doi.org/10.21105/joss.01026 Other Allen Mouse Brain Atlas http://mouse.brain-map.org/static/atlas RRID:SCR_002978 Nanoject II Drummond Scientific Cat# 3-000-204 Raspberry Pi 3 Model B Raspberry Pi, https://www.raspberrypi.org/ N/A RHD2000 USB Interface Board Intan Technologies Cat# C3100 RHD2132 amplifier Intan Technologies, https://intantech.com/products_RHD2000.html Cat# C3334 RZ5P amplifier Tucker-Davis Technologies, https://www.tdt.com/component/rz5p/ Cat# RZ5P 473 nm lasers Laserglow Cat# LRS-0473-PFM-00050-05 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. DECLARATION OF INTERESTS The authors declare no competing financial interests. REFERENCES 1. Adamantidis AR , Gutierrez Herrera C , and Gent TC (2019). Oscillating circuitries in the sleeping brain. Nat. Rev. Neurosci 20 , 746–762.31616106 2. Scammell TE , Arrigoni E , and Lipton JO (2017). Neural circuitry of wakefulness and sleep. Neuron 93 , 747–765.28231463 3. Saper CB , Fuller PM , Pedersen NP , Lu J , and Scammell TE (2010). Sleep state switching. Neuron 68 , 1023–1042.21172606 4. Fernandez LMJ , and Lüthi A (2019). Sleep spindles: mechanisms and functions. Physiol. Rev 100 , 805–868.31804897 5. Lecci S , Fernandez LMJ , Weber FD , Cardis R , Chatton J-Y , Born J , and Lüthi A (2017). Coordinated infraslow neural and cardiac oscillations mark fragility and offline periods in mammalian sleep. Sci. Adv 3 , e1602026.28246641 6. Terzano MG , Mancia D , Salati MR , Costani G , Decembrino A , and Parrino L (1985). The cyclic alternating pattern as a physiologic component of normal NREM sleep. Sleep 8 , 137–145.4012156 7. Watson BO (2018). Cognitive and physiologic impacts of the infraslow oscillation. Front. Syst. Neurosci 12 , 44.30386218 8. Luppi P-H , Clement O , Sapin E , Peyron C , Gervasoni D , Léger L , and Fort P (2012). Brainstem mechanisms of paradoxical (REM) sleep generation. Pflüg. Arch. Eur. J. Physiol 463 , 43–52. 9. Park S-H , and Weber F (2020). Neural and homeostatic regulation of REM sleep. Front. Psychol 11 , 1662.32793050 10. Peever J , and Fuller PM (2017). The Biology of REM Sleep. Curr. Biol 27 , R1237–R1248.29161567 11. Weber F , Chung S , Beier KT , Xu M , Luo L , and Dan Y (2015). Control of REM sleep by ventral medulla GABAergic neurons. Nature 526 , 435–438.26444238 12. Clément O , Sapin E , Libourel P-A , Arthaud S , Brischoux F , Fort P , and Luppi P-H (2012). The lateral hypothalamic area controls paradoxical (REM) sleep by means of descending projections to brainstem GABAergic neurons. J Neurosci 32 , 16763–16774.23175830 13. Clément O , Garcia SV , Libourel P-A , Arthaud S , Fort P , and Luppi P-H (2014). The Inhibition of the Dorsal Paragigantocellular Reticular Nucleus Induces Waking and the Activation of All Adrenergic and Noradrenergic Neurons: A Combined Pharmacological and Functional Neuroanatomical Study. PLOS ONE 9 , e96851.24811249 14. Gervasoni D , Peyron C , Rampon C , Barbagli B , Chouvet G , Urbain N , Fort P , and Luppi P-H (2000). Role and Origin of the GABAergic Innervation of Dorsal Raphe Serotonergic Neurons. J. Neurosci 20 , 4217–4225.10818157 15. Goutagny R , Luppi P-H , Salvert D , Lapray D , Gervasoni D , and Fort P (2008). Role of the dorsal paragigantocellular reticular nucleus in paradoxical (rapid eye movement) sleep generation: a combined electrophysiological and anatomical study in the rat. Neuroscience 152 , 849–857.18308473 16. Kaur S , Saxena R , and Mallick BN (2001). GABAergic neurons in prepositus hypoglossi regulate REM sleep by its action on locus coeruleus in freely moving rats. Synapse 42 , 141–150.11746711 17. Sakai K (2018). Behavioural state-specific neurons in the mouse medulla involved in sleep-wake switching. Eur. J. Neurosci 47 , 1482–1503.29791042 18. Sapin E , Lapray D , Bérod A , Goutagny R , Léger L , Ravassard P , Clément O , Hanriot L , Fort P , and Luppi P-H (2009). Localization of the brainstem GABAergic neurons controlling paradoxical (REM) sleep. PLoS ONE 4 , e4272.19169414 19. Verret L , Fort P , Gervasoni D , Léger L , and Luppi P-H (2006). Localization of the neurons active during paradoxical (REM) sleep and projecting to the locus coeruleus noradrenergic neurons in the rat. J. Comp. Neurol 495 , 573–586.16498678 20. Ennis M , and Aston-Jones G (1989). Potent inhibitory input to locus coeruleus from the nucleus prepositus hypoglossi. Brain Res. Bull 22 , 793–803.2475220 21. Gottesmann C (1996). The transition from slow-wave sleep to paradoxical sleep: evolving facts and concepts of the neurophysiological processes underlying the intermediate stage of sleep. Neurosci. Biobehav. Rev 20 , 367–387.8880730 22. Durán E , Oyanedel CN , Niethard N , Inostroza M , and Born J (2018). Sleep stage dynamics in neocortex and hippocampus. Sleep 41 . 23. Grace KP , and Horner RL (2020). A Focal Inactivation and Computational Study of Ventrolateral Periaqueductal Gray and Deep Mesencephalic Reticular Nucleus Involvement in Sleep State Switching and Bistability. eNeuro 7 , ENEURO.0451-19.2020. 24. Berndt A , Lee SY , Wietek J , Ramakrishnan C , Steinberg EE , Rashid AJ , Kim H , Park S , Santoro A , Frankland PW , (2016). Structural foundations of optogenetics: Determinants of channelrhodopsin ion selectivity. Proc. Natl. Acad. Sci 113 , 822–829.26699459 25. Barbato G , and Wehr TA (1998). Homeostatic regulation of REM sleep in humans during extended sleep. Sleep 21 , 267–276.9595605 26. Benington JH , and Heller HC (1994). REM-sleep timing is controlled homeostatically by accumulation of REM-sleep propensity in non-REM sleep. Am. J. Physiol.-Regul. Integr. Comp. Physiol 266 , R1992–R2000. 27. Ursin R (1970). Sleep stage relations within the sleep cycles of the cat. Brain Res. 20 , 91–97.4315521 28. Vivaldi EA , Ocampo A , Wyneken U , Roncagliolo M , and Zapata AM (1994). Short-term homeostasis of active sleep and the architecture of sleep in the rat. J. Neurophysiol 72 , 1745–1755.7823099 29. Le Bon O (2020). Relationships between REM and NREM in the NREM-REM sleep cycle: a review on competing concepts. Sleep Med. 70 , 6–16.32179430 30. Ocampo-Garcés A , Bassi A , Brunetti E , Estrada J , and Vivaldi EA (2020). REM sleep-dependent short-term and long-term hourglass processes in the ultradian organization and recovery of REM sleep in the rat. Sleep 43 , 1–13. 31. Ocampo-Garcés A , Molina E , RodrÍguez A , and Vivaldi EA (2000). Homeostasis of REM sleep after total and selective sleep deprivation in the rat. J. Neurophysiol 84 , 2699–2702.11068012 32. Wiegert JS , Mahn M , Prigge M , Printz Y , and Yizhar O (2017). Silencing Neurons: Tools, Applications, and Experimental Constraints. Neuron 95 , 504–529.28772120 33. Gomez JL , Bonaventura J , Lesniak W , Mathews WB , Sysa-Shah P , Rodriguez LA , Ellis RJ , Richie CT , Harvey BK , Dannals RF , (2017). Chemogenetics revealed: DREADD occupancy and activation via converted clozapine. Science 357 , 503–507.28774929 34. Oikonomou G , Altermatt M , Zhang R-W , Coughlin GM , Montz C , Gradinaru V , and Prober DA (2019). The serotonergic raphe promote sleep in zebrafish and mice. Neuron 103 , 686–701.31248729 35. Tervo DGR , Hwang B-Y , Viswanathan S , Gaj T , Lavzin M , Ritola KD , Lindo S , Michael S , Kuleshova E , Ojala D , (2016). A designer AAV variant permits efficient retrograde access to projection neurons. Neuron 92 , 372–382.27720486 36. Lerner TN , Shilyansky C , Davidson TJ , Evans KE , Beier KT , Zalocusky KA , Crow AK , Malenka RC , Luo L , Tomer R , (2015). Intact-brain analyses reveal distinct information carried by SNc dopamine subcircuits. Cell 162 , 635–647.26232229 37. McGinty DJ , and Harper RM (1976). Dorsal raphe neurons: depression of firing during sleep in cats. Brain Res. 101 , 569–575.1244990 38. Kaur S , Saxena R , and Mallick BN (1997). GABA in locus coeruleus regulates spontaneous rapid eye movement sleep by acting on GABA_A receptors in freely moving rats. Neurosci. Lett 223 , 105–108.9089684 39. Girard F , von Siebenthal M , Davis FP , and Celio MR (2020). Gene expression analysis in the mouse brainstem identifies Cart and Nesfatin as neuropeptides coexpressed in the Calbindin-positive neurons of the Nucleus papilio. Sleep 43 , 1–14. 40. Gutierrez Herrera C , Girard F , Bilella A , Gent TC , Roccaro-Waldmeyer DM , Adamantidis A , and Celio MR (2019). Neurons in the Nucleus papilio contribute to the control of eye movements during REM sleep. Nat. Commun 10 , 1–11.30602773 41. Kinney GG , Kocsis B , and Vertes RP (1994). Injections of excitatory amino acid antagonists into the median raphe nucleus produce hippocampal theta rhythm in the urethane-anesthetized rat. Brain Res. 654 , 96–104.7982102 42. Kinney GG , Kocsis B , and Vertes RP (1995). Injections of muscimol into the median raphe nucleus produce hippocampal theta rhythm in the urethane anesthetized rat. Psychopharmacology (Berl.) 120 , 244–248.8524970 43. Vertes RP , Kinney GG , Kocsis B , and Fortin WJ (1994). Pharmacological suppression of the median raphe nucleus with serotonin1a agonists, 8-OH-DPAT and buspirone, produces hippocampal theta rhythm in the rat. Neuroscience 60 , 441–451.8072690 44. Maru E , Takahashi LK , and Iwahara S (1979). Effects of median raphe nucleus lesions on hippocampal EEG in the freely moving rat. Brain Res. 163 , 223–234.218681 45. Dunmyre JR , Mashour GA , and Booth V (2014). Coupled flip-flop model for REM sleep regulation in the rat. PloS One 9 , e94481.24722577 46. Diniz Behn C , Ananthasubramaniam A , and Booth V (2013). Contrasting Existence and Robustness of REM/Non-REM Cycling in Physiologically Based Models of REM Sleep Regulatory Networks. SIAM J. Appl. Dyn. Syst 12 , 279–314. 47. Weber F , Hoang Do JP , Chung S , Beier KT , Bikov M , Saffari Doost M , and Dan Y (2018). Regulation of REM and Non-REM sleep by periaqueductal GABAergic neurons. Nat. Commun 9 , 1–13.29317637 48. Grace KP , Vanstone LE , and Horner RL (2014). Endogenous Cholinergic Input to the Pontine REM Sleep Generator Is Not Required for REM Sleep to Occur. J. Neurosci 34 , 14198–14209.25339734 49. Rolls A , Colas D , Adamantidis A , Carter M , Lanre-Amos T , Heller HC , and de Lecea L (2011). Optogenetic disruption of sleep continuity impairs memory consolidation. Proc. Natl. Acad. Sci. U. S. A 108 , 13305–13310.21788501 50. Fernandez LM , Vantomme G , Osorio-Forero A , Cardis R , Béard E , and Lüthi A (2018). Thalamic reticular control of local sleep in mouse sensory cortex. eLife 7 , e39111.30583750 51. Csernai M , Borbély S , Kocsis K , Burka D , Fekete Z , Balogh V , Káli S , Emri Z , and Barthó P (2019). Dynamics of sleep oscillations is coupled to brain temperature on multiple scales. J. Physiol 597 , 4069–4086.31197831 52. Seibt J , Richard CJ , Sigl-Glöckner J , Takahashi N , Kaplan DI , Doron G , de Limoges D , Bocklisch C , and Larkum ME (2017). Cortical dendritic activity correlates with spindle-rich oscillations during sleep in rodents. Nat. Commun 8 , 1–13.28232747 53. Bandarabadi M , Herrera CG , Gent TC , Bassetti C , Schindler K , and Adamantidis AR (2020). A role for spindles in the onset of rapid eye movement sleep. Nat. Commun 11 , 1–12.31911652 54. Siapas AG , and Wilson MA (1998). Coordinated Interactions between Hippocampal Ripples and Cortical Spindles during Slow-Wave Sleep. Neuron 21 , 1123–1128.9856467 55. Sesack SR , Deutch AY , Roth RH , and Bunney BS (1989). Topographical organization of the efferent projections of the medial prefrontal cortex in the rat: An anterograde tract-tracing study with Phaseolus vulgaris leucoagglutinin. J. Comp. Neurol 290 , 213–242.2592611 56. MacKinnon JG (2009). Bootstrap Hypothesis Testing. In Handbook of Computational Econometrics (John Wiley & Sons, Ltd), pp. 183–213. 57. Vallat R (2018). Pingouin: statistics in Python. Journal of Open Source Software, 3 (31 ), 1026. 58. Eban-Rothschild A , Rothschild G , Giardino WJ , Jones JR , and Lecea L. de (2016). VTA dopaminergic neurons regulate ethologically relevant sleep–wake behaviors. Nat. Neurosci 19 , 1356–1366.27595385 59. Yu X , Li W , Ma Y , Tossell K , Harris JJ , Harding EC , Ba W , Miracca G , Wang D , Li L , (2019). GABA and glutamate neurons in the VTA regulate sleep and wakefulness. Nat. Neurosci 22 , 106–119.30559475

PMC008xxxxxx/PMC8752517.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101120028 22411 Dev Cell Dev Cell Developmental cell 1534-5807 1878-1551 34942115 8752517 10.1016/j.devcel.2021.12.004 NIHMS1763937 Article chinmo-mutant spermatogonial stem cells cause mitotic drive by evicting non-mutant neighbors from the niche Tseng Chen-Yuan 1 Burel Michael 1 Cammer Michael 2 Harsh Sneh 1 Flaherty Maria Sol 1 Baumgartner Stefan 45 Bach Erika A. 136 1 Department of Biochemistry and Molecular Pharmacology 2 DART Microscopy Laboratory 3 Helen L. and Martin S. Kimmel Center for Stem Cell Biology, NYU Grossman School of Medicine, New York, NY 10016, USA 4 Department of Experimental Medical Sciences, Lund University, 22184 Lund, Sweden 5 Department of Biology, University of Konstanz, 78467 Konstanz, Germany 6 Lead contact AUTHOR CONTRIBUTIONS Conceptualization, MSF, MB, CYT, EAB; Investigation, CYT, MB, MSF, MC, SH, SB; Supplied Key Reagents, SB; Manuscript Writing, CYT, MB, EAB; Funding, CYT, MB, EAB; Supervision, EAB Correspondence: erika.bach@nyu.edu 13 12 2021 10 1 2022 22 12 2021 10 1 2023 57 1 8094.e7 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY Niches maintain a finite pool of stem cells via restricted space and short-range signals. Stem cells compete for limited niche resources, but the mechanisms regulating competition are poorly understood. Using the Drosophila testis model, we show that germline stem cells (GSCs) lacking the transcription factor Chinmo gain a competitive advantage for niche access. Surprisingly, chinmo−/− GSCs rely on a new mechanism of competition in which they secrete the extracellular matrix protein Perlecan to selectively evict non-mutant GSCs and then upregulate Perlecan-binding proteins to remain in the altered niche. Over time, the GSC pool can be entirely replaced with chinmo−/− cells. As a consequence, the mutant chinmo allele acts as a gene drive element: the majority of offspring inherit the allele despite the heterozygous genotype of the parent. Our results suggest the influence of GSC competition may extend beyond individual stem cell niche dynamics to population-level allelic drift and evolution. Graphical Abstract eTOC Blurb How stem cells compete for niche access remains largely unknown. Tseng et al. show that spermatogonial stem cells (SSCs) lacking the transcription factor Chinmo remodel the niche to their own benefit, causing the expulsion of non-mutant SSCs. This leads to a homozygous germline and biased inheritance of the chinmo-mutant allele. Drosophila testis germline stem cell niche competition chinmo Perlecan Laminin Dystroglycan biased inheritance aging pmcINTRODUCTION Stem cells reside in discrete microenvironments called niches that maintain a supply of undifferentiated stem cells via molecular signals. Because these signals are short-range and niche space is often limited, stem cells compete with one another for niche occupancy. In stem cell competition, “winners” remain in the niche and retain stem cell identity while “losers” exit and differentiate (Simons and Clevers, 2011). Under normal conditions, stem cells compete through a stochastic process: any given stem cell has an equal probability of remaining in the niche or being lost and replaced by its neighbor. In this homeostatic process, known as neutral competition, no one stem cell has a long-term competitive advantage over another (Klein and Simons, 2011). Competition can be biased if one stem cell gains an advantage over the others. Indeed, one advantaged stem cell and its progeny can hijack a niche to compose the entire stem cell pool, resulting in tissue monoclonality (Vermeulen et al., 2013, Amoyel et al., 2014, Issigonis et al., 2009, Snippert et al., 2014). However, the factors and mechanisms regulating stem cell competition – and germline stem cell competition in particular - remain largely mysterious. The Drosophila testis is an ideal model system to study stem cell competition due to its well-characterized niche, genetic tractability, and conserved stem cell maintenance pathways (Fig. 1A and (Greenspan et al., 2015)). The testis is a coiled blind-ended tube, enveloped by a muscle sheath, responsible for spermatogenesis throughout adult male life (Hardy et al., 1979, Fuller, 1998). The niche includes a group of post-mitotic cells, which is anchored to the testis apex by integrins (Tanentzapf et al., 2007). The niche secretes self-renewal signals that support two distinct stem cell populations. GSCs divide continuously throughout life and ultimately produce sperm. Approximately 8 GSCs adhere to the niche through the adhesion molecule E-Cadherin (E-Cad) (Yamashita et al., 2003). The niche also supports somatic cyst stem cells (CySCs), which co-differentiate with GSCs into post-mitotic cysts that enclose developing spermatogonia (Hardy et al., 1979, Fabrizio et al., 2003). Stem cells in the Drosophila testis, as those in other tissues and organisms, compete for limited niche space and resources. CySCs have documented competitive behavior, and several mechanisms triggering biased competition in favor of single CySCs have been uncovered (Amoyel et al., 2016, Amoyel et al., 2014, Issigonis et al., 2009, Singh et al., 2010, Stine et al., 2014). Significantly less is known, however, about competition in GSCs. Regular GSC loss and replacement by neighbors has been observed (Wallenfang et al., 2006, Sheng and Matunis, 2011, Salzmann et al., 2013). While this behavior is suggestive of neutral competition, no underlying mechanisms have been elucidated. Biased competition has also never been observed in male GSCs, though single female GSCs harboring specific genetic mutations can outcompete their neighbors in the Drosophila ovary (Jin et al., 2008, Rhiner et al., 2009). Here we report biased competition in male GSCs driven by mutations in the transcriptional repressor Chinmo (Zhu et al., 2006). Chinmo is a known regulator of somatic stem cell activity in the Drosophila ventral nerve cord and testis (Flaherty et al., 2010, Narbonne-Reveau et al., 2016). In the testis, Chinmo is required for the maintenance of male sexual identity. CySCs devoid of Chinmo undergo sex-reversal and transdifferentiate into feminized somatic stem cells (Grmai et al., 2018, Ma et al., 2014, Ma et al., 2016). Despite transcriptomic analyses of somatic stem cells lacking Chinmo, direct Chinmo target genes have not yet been identified (Grmai et al., 2021, Narbonne-Reveau et al., 2016). Here, we show that loss of chinmo in single GSCs confers a competitive advantage for niche access so that, over time, the GSC pool becomes “fixed” with chinmo−/− cells. Surprisingly, chinmo−/− GSCs do not rely on increased proliferation or canonical adhesion systems to gain their advantage. Instead, chinmo−/− GSCs ectopically secrete the extracellular matrix protein (ECM) Perlecan (Pcan) that creates a de novo ECM around the niche within the testis lumen. This resculpted niche selectively evicts non-mutant GSCs from the niche but retains chinmo−/− GSCs, which upregulate ECM-binding proteins. As a result this competition, the majority of adult offspring inherit the mutant chinmo allele despite the heterozygous genotype of the parent. These results provide the experimental evidence that GSC competition underlie “mitotic” drive, a proposed but not proven mechanism by which GSCs with a competition advantage transmit greater than the expected 50% Mendelian ratio (Otto and Hastings, 1998). Thus, GSC competition may not only be important for understanding the fundamental principles of stem cell dynamics but also have long-term implications for genetic drift and evolution (Hastings, 1989). RESULTS chinmo−/− GSCs evict non-mutant neighbor GSCs from the niche Chinmo is expressed in male GSCs (Fig. S1A), but its role in these cells is unknown. To investigate this, we used mitotic recombination to induce single GFP-negative control and chinmo homozygous mutant (chinmo−/−) clones in adult flies and monitored their niche retention and lineages over time (Fig. 1B). We used a null allele (chinmo1), a hypomorphic allele (chinmok13009) and validated RNAi-depletion (chinmo-i) to generate chinmo-deficient GSCs clones (Fig. S1A–F). Control GSC clones and chinmo-deficient GSC clones were induced at the same frequency as assessed at 2 days post clone induction (dpci) (Fig. 1C,D). Both types of clones contained single, Vasa-positive cells at the niche that had dot fusomes characteristic of GSCs (Fig. 1G,H). Surprisingly, chinmo−/− GSCs composed a majority of the GSC pool by 28 dpci (Fig. 1F) compared to control clones (Fig. 1E). Prior work has shown that mutant CySCs with a competitive advantage prevail over both wild-type (WT) CySCs and WT GSCs for niche access. In contrast, chinmo−/− GSCs only outcompeted non-mutant GSCs while leaving the CySC population unaffected. We observed comparable numbers of CySCs expressing the stem cell marker Zfh1 in testes harboring either chinmo−/− or control GSC clones (Fig. 1-I, Table S1 #16). Thus, chinmo−/− GSCs provide a new model system for studying GSC-GSC specific stem cell competition. We hypothesized that single GSCs lacking chinmo gain a competitive advantage over their non-mutant neighbors for niche access, so we quantified the percent of clones composing the GSC pool, which we termed “clone occupancy,” at 2, 7, 14, 21, and 28 dpci (Fig. 1B). At 2 dpci, both control and chinmo−/− GSC clones occupied the niche in equal proportions (Fig. 1M, Table S1 #37–46). By 28 dpci, however, chinmo−/− GSCs occupied a majority of the GSC pool (78.52 ± 3.77%, Table S1 #16), while control clones occupied significantly less (45.97 ± 5.00%, P < 2 × 10−5, Table S1 #11) (Fig. 1J). chinmo−/− GSCs occupied a majority of the GSC pool at 7, 14, and 21 dpci (Fig. 1J, Table S1 #13–15). Similar trends were observed at 14 dpci for the other chinmo alleles (Table S1 #76,80). Indeed, the probability of recovering testes with “fixed” niches - where a GSC clone composed the entire GSC pool - was significantly higher when GSC clones lacked chinmo (12.50% vs. 54.00% in control and chinmo−/− clones, respectively, at 28 dpci, P < 3 × 10−7) (Fig. 1K, Table S1 #17–26). These data indicate that loss of chinmo in single GSCs biases competition for niche access in their favor. To determine whether chinmo−/− GSC clones expanded their niche presence by over-proliferation and/or by losing non-mutant neighbor GSCs, we quantified the average number of GSC clones and the average number of non-mutant neighbor GSCs over time in testes containing either control or chinmo−/− clones. Control and chinmo−/− clones themselves were present at nearly identical levels over time (3.68 ± 0.45 vs. 3.46 ± 0.25, control and chinmo−/− clones, respectively, by 28 dpci, P = 0.47) (Fig. 1M, Table S1 #37–46). By contrast, neighbor GSCs were precipitously lost from the niche in the presence of chinmo−/− but not control clones (4.38 ± 0.46 vs. 1.00 ± 0.18 neighbors of control or chinmo−/− clones, respectively, by 28 dpci, P = 7 × 10−7) (Fig. 1N, Table S1 # 47–56). As a result, niches harboring chinmo−/− clones experienced a net loss in total GSCs (8.05 ± 0.29 vs. 4.46 ± 0.22 in control and chinmo−/− clones, respectively, at 28 dpci, P = 4 × 10−21) (Fig. 1L, Table S1, #27–36). We obtained similar results with other chinmo alleles (Table S1, #98,102). These results suggest that chinmo−/− GSCs gain their competitive advantage not by increasing their numbers but rather by imperiling niche access of neighbor GSCs (Fig. 1O). chinmo−/− GSCs ectopically secrete Pcan into the niche milieu The competitive advantages of chinmo−/− GSCs could result from behaviors observed in “winning” cells in the Drosophila ovary or epithelia, including increased proliferation, increased E-Cad, or induced death of neighbors (Jin et al., 2008, Amoyel and Bach, 2014). However, there were no substantial differences in these parameters between chinmo−/− GSC clones and control GSC clones or non-mutant neighbors (Fig. S2A–C). Additionally, chinmo−/− GSC clones still divided asymmetrically as evidenced by the mother centrosome being located at the GSC-niche interface and the daughter centrosome 180° away (Fig. S2D–F). We considered the possibility that chinmo−/− GSCs had increased self-renewal signaling compared to neighbors. GSCs require BMP/Mad and JAK/STAT pathways for self-renewal and niche adhesion, respectively (Leatherman and Dinardo, 2010). However, there are no differences in phosphoMad or phospho-STAT between chinmo−/− clones and their non-mutant neighbors (Fig. S2G–H). We observed similar robust E-Cad levels at the GSC-niche interface in chinmo−/− clones and in their non-mutant neighbor GSCs (Fig. S2I), suggesting decreased E-Cad levels were not responsible for neighbor loss. Electron micrograph of the stem cell niche at 28 dpci revealed a ring of proteinaceous material encircling the niche in testes with chinmo−/− GSCs but not those with control GSC clones (Fig. 2A,B). This material resembled and was contiguous with the ECM composing the muscle basal lamina (BL). We termed this structure the “moat” because it resembled the ditch around a castle. A similar “gap” between GSCs and the niche can be observed at 28 dpci in single confocal images of testes with chinmo−/− GSC clones (Fig. S3B) but not with control GSC clones (Fig. S3A). The majority of chinmo−/− GSCs were displaced significantly farther from the niche by 28 dpci than control clones (Fig. S3C, compare blue to gray bar). Moats were observed in a majority of testes with chinmo−/− clones but only rarely seen in testes with control clones (Fig. S3D). Based on these results, we speculate that chinmo−/− GSCs secrete ectopic ECM into the niche milieu. We examined expression of Pcan, a conserved secreted proteoglycan that cross-links the ECM, allows for cell-ECM adhesion, and facilities ligand diffusion (Broadie et al., 2011). In testes with control GSC clones, Pcan is found in the muscle BL and not within the testis lumen at 21 dpci (Fig. 2C,G). However, Pcan localized to the moat when chinmo−/− GSCs were present and this was observed as early as 7 dpci (Fig. 2D,G, Fig. S3E,F). We also examined expression of Laminin (Lan), a major component of BL. In testes with control GSC clones, Lan localized to the muscle BL and not the testis lumen (Fig. 2E,G). However, Lan localized to the moat when chinmo−/− GSCs were present as early as 7 dpci (Fig. 2F,G, Fig. S3G,H). Further, 3D reconstructions of confocal z-stacks revealed that Lan surrounded the niche in testes with chinmo−/− clones but not those with control clones (Fig. 3H,I, Movie S1). Other ECM proteins were expressed at low levels in testes and were not localized to the moat (Fig. S3M–T), indicating that the moat is specifically composed of Pcan and Lan. Pcan is necessary and sufficient for perturbing neighbor GSC niche occupancy If ectopic Pcan underlies the competitive advantage of chinmo−/− GSCs, these should be true: (1) Lan accumulation in the moat should depend on Pcan secretion by chinmo−/− GSCs; (2) ectopic Pcan secretion should be sufficient to induce GSC loss in an otherwise WT testis; and (3) chinmo−/− GSCs should require Pcan for their competitive properties. To test the first condition, we generated positive-marked chinmo−/− GSC clones and chinmo−/− GSC clones depleted for Pcan. In testes with chinmo−/− GSC clones, Pcan is found in the moat (Fig. 2J,N) but this was significantly reduced when these clones were depleted for Pcan (Fig. 2K,N). Lan in the moat also declined significantly in chinmo−/− clones depleted for Pcan (Fig. 2L,M,O). These results indicate that the moat is caused by ectopic Pcan secretion by chinmo−/− GSCs and suggest that Lan is recruited to the moat from the muscle BL. In support of this, Lan in the moat was significantly reduced when Lan was depleted from muscle (Fig. S3IK). HCR-FISH revealed that Pcan transcripts were significantly increased in chinmo−/− GSCs compared to control GSCs (Fig. S4A–F), suggesting that Pcan is Chinmo target in GSCs. To test the second condition - whether ectopic Pcan can impede WT GSCs - we used the GSC driver nanos (nos) to mis-express Pcan in otherwise WT GSCs. We observed ectopic Pcan and Lan surrounding the niche in nos>Pcan testes, similar to the chinmo−/− GSC phenotype (Fig. 3A–D). As expected, GSC-secreted Pcan phenocopied the loss in total GSC number (Fig. 3E, Table S1 #57,58) observed in testes containing chinmo−/− GSC clones at same time point (i.e., 21 days) (Fig. 1L). Thus, ectopic Pcan secreted by GSCs is sufficient to alter the niche architecture and disadvantage WT GSC niche access. To test the third condition - whether Pcan is necessary for chinmo−/− GSCs to expel non-mutant neighbor GSCs - we used the MARCM technique to deplete Pcan using two validated RNAi lines (Fig. S1K–N) in control or chinmo−/− GSC clones. Depletion of Pcan from chinmo−/− GSC clones robustly blocked the increase in chinmo−/− clone occupancy (48.21 ± 3.49% vs. 33.52 ± 2.50% in chinmo−/− and chinmo−/−, Pcan-RNAi clones, respectively, at 14 dpci, P < 0.0018) (Fig. 3F, Table S1 #68, 70). By contrast, clone occupancy of control clones was unaffected by Pcan depletion (Fig. 3F, Table S1 #60, 62). This reduction in competitive behavior is not because Pcan-depleted chinmo−/− GSCs were lost from the niche (Fig. 3I, Table S1 #134, 136), but rather because non-mutant neighbor GSCs were no longer preferentially evicted (3.37 ± 0.27 vs. 4.67 ± 0.23 WT neighbors of chinmo−/− and chinmo−/−, Pcan-RNAi clones, respectively, at 14 dpci, P < 0.0006) (Fig. 3H, Table S1, #112, 114). Consistent with the partial rescue of neighbor GSCs, the average total number of GSCs was also significantly increased by depleting Pcan in chinmo−/− GSCs (6.28 ± 0.26 vs. 6.97 ± 0.19 GSCs in testes with chinmo−/− and chinmo−/−, Pcan-RNAi clones, respectively, at 14 dpci, P < 0.044) (Fig. 3G, Table S1 #90, 92). A similar autonomous requirement for Pcan was observed in chinmok13009 clones (Table S1, #76, 78, 98, 100, 120, 122). By contrast, Lan depletion from chinmo−/− GSC clones using a validated RNAi line (Fig. S1H–J) did not significantly alter clone occupancy, the average number of GSCs, or the average number of neighbors (Table S1 #74, 96, 118). Thus, removal of Pcan but not Lan from chinmo−/− GSCs greatly impairs their competitive phenotype. Consistent with the model that Lan is recruited to the moat, depleting Lan from the muscle significantly reduced Lan in the moat and the clone occupancy of chinmo−/− GSCs (Fig. S3K,L, Table S1 #147, 148). We conclude that chinmo−/− GSCs non-autonomously compromise neighbor GSC access by secreting Pcan into the niche (Fig. 3J). chinmo−/− GSCs remain in the altered niche by upregulating Dystroglycan (Dg) We reasoned that chinmo−/− GSCs remain in the niche because they upregulate ECM-binding proteins. Dg is a transmembrane protein that interacts with Pcan via its extracellular domain and the actin cytoskeleton via its intercellular domain (Schneider et al., 2006). Dg is significantly increased at the GSC-niche interface in chinmo−/− GSCs but not in control GSC clones (Fig. 4A–C). Furthermore, HCR-FISH revealed that Dg transcripts were significantly increased in chinmo−/− GSCs compared to control GSCs (Fig. S4G–I), suggesting that Dg may be regulated by Chinmo in GSCs. To test this, we depleted Dg using a validated RNAi line (Fig. S1O–Q) from control and chinmo−/− clones. Depleting Dg from chinmo−/− clones significantly decreased their clone occupancy (Fig. 4D, Table S1 #154, 156), restored the average total number of GSCs (Fig. 4E, Table S1 #166, 168) and rescued non-mutant neighbors (Fig. 4F, Table S1 #178, 180). Dg knockdown in chinmo−/− clones – but not control clones - caused a slight but significant decrease in clone recovery at 14 dpci (Fig. 4G, Table S1 #190, 192). Similar results were observed with the chinmok13009 allele (Table S1 #158, 160, 170, 172, 182, 184, 194, 196). We conclude that Dg expression is an important mechanism used by chinmo−/− GSCs to remain in the remodeled niche (Fig. 4H). chinmo−/− GSCs remain in the altered niche via βPS integrin We next examined integrins, which mediate cell-ECM interactions, in GSC competition. The integrin-associated protein Talin is significantly increased at the GSC-niche interface in chinmo−/− GSC clones compared to non-mutant GSC neighbors (Fig. S5B,C). By contrast, there is no significant change in Talin levels between control GSC clones compared to their neighbor GSCs (Fig. S5A,C). Only βPS - and not other integrin subunits - was increased at the GSC-niche interface in chinmo-deficient GSCs (Fig. S5D–P). HCR-FISH revealed that βPS transcripts were significantly increased in chinmo-depleted GSCs compared to control GSCs (Fig. S4J–L), suggesting that βPS may be regulated by Chinmo in GSCs. To determine whether Talin or βPS was required for chinmo−/− clones to remain in the altered niche, we depleted either factor using validated RNAi lines (Fig. S1R–Y) in control or chinmo−/− clones and assessed parameters of competition. Knocking down Talin in chinmo−/− clones significantly decreased their clone occupancy (Fig. S6A, Table S1 #202, 204), restored the average total number of GSCs (Fig. S6B, Table S1 #210, 212) and rescued non-mutant neighbors (Fig. S6C, Table S1 #218, 220). Talin depletion from chinmo−/− clones - but not from control clones - reduced their niche residence (Fig. S6D, Table S1 #226, 228). Depleting βPS with either RNAi line in chinmo−/− clones significantly decreased clone occupancy (Fig. S6E, Table S1 #236, 238, 240), restored the average total number of GSCs (Fig. S6F, Table S1 #248, 250, 252) and rescued non-mutant neighbors (Fig. S6G, Table S1 #260, 262, 264). Control GSC clones depleted for βPS using RNA-i #1 could not be recovered at 2 dcpi (Fig. S6H, Table S1 #267), and control clone recovery was greatly reduced using the second βPS RNAi line (Table S1 #269). These data suggest that βPS is required for GSC maintenance. We could not test this hypothesis using mosaic analysis because mutation of the X-linked βPS gene is male lethal. However, in support of this model, there were significantly fewer GSCs at the niche when βPS or Talin were depleted from GSCs (Fig. S1Z). Importantly, chinmo−/− GSC clones depleted for βPS could remain in the altered niche (Fig. S6H, Table S1 #273, 274). These results indicate that increased integrin expression is used by chinmo−/− GSCs to remain in the resculpted niche (Fig. S6I). They also demonstrate that Dg and βPS integrin are non-redundant mechanisms employed by competitive GSCs for niche occupancy. Downregulation of ECM binding proteins in the chinmo−/− clone alters Pcan and Lan levels in the moat. In testes with chinmo−/− GSCs depleted for βPS or Dg, there is significantly less Pcan and Lan in the moat (Fig. 2N,O). These results suggest that (1) there is a feedback loop from ECM binding receptors promoting Pcan secretion or (2) ECM binding proteins in GSCs stabilize the moat. Providing neighbors with more Dg rescues them from competition We next tested whether non-mutant neighbor GSCs supplied with more Dg would adhere to the moat and be rescued from competition (Fig. 5A). To test this, we mis-expressed Dg in all GSCs using nos and then generated control or chinmo−/− GSC clones. We compared the number of non-mutant neighbor GSCs at 14 dpci when competition is robustly occurring (Fig. 5B). The moat generated by chinmo−/− GSCs (Fig. 5C,F) was still observed when Dg was mis-expressed in all GSCs (Fig. 5D–F). Despite the presence of the moat, non-mutant neighbor GSCs remained in the altered niche when they had increased Dg (Fig. 5D, yellow arrowheads, Fig. 5G, last bar, Table S2 #6, 8). Furthermore, the number of GSCs/testis is rescued when neighbors have increased Dg (Fig. 5H, Table S2 #14, 16). GSC competition causes biased inheritance Since testes with chinmo−/− GSCs frequently have a homozygous mutant germline, the chinmo−/− allele should be inherited at a super-Mendelian frequency (i.e., greater > 50%). To assess this, we generated males which had a chinmo+ or chinmo− allele in trans to a ubi-GFP-labeled sister chromosome that could be scored in the next generation (Fig. 6A). We generated chinmo+/+ GSC clones or chinmo−/− GSC clones and aged the males until 21 dpci, at which time we mated single males with two virgin WT females for two days. At 23 dpci, we isolated the testes of the mated males and assessed germline clonality. Our model predicts that if competitive chinmo−/− GSCs cause mitotic drive, the chinmo− chromosome should be inherited by more than 50% of the F1 progeny while the GFP-positive sister chromosome should be passed on to less than 50%. By contrast, since chinmo+/+ GSCs should not have a competitive advantage, the chinmo+ chromosome should be inherited by 50% of the F1 progeny and the GFP-positive sister chromosome should be inherited by the other 50% (Fig. 6A). As expected, testes with chinmo+/+ clones contained both GFP-negative GSC clones and GFP-positive non-mutant neighbor GSCs (Fig. 6B). As predicted, testes with chinmo−/− GSC clones were frequently monoclonal or “fixed”, containing only GFP-negative chinmo−/− GSCs (Fig. 6C). For the allele inheritance assay, we followed the sister chromosome by inheritance of the GFP-positive transgene it harbors (Fig. 6D, green part of each bar) and the chinmo allele (either chinmo+ or chinmo−) by inheritance of the GFP-negative chromosome (Fig. 6D, black part of each bar). As predicted, 50% of the offspring of males with chinmo+/+ GSC clones inherited the chinmo+ allele and the other 50% inherited the GFP-positive sister chromosome (Fig. 6D, first bar, Table S3 #1, 2). Consistent with our model, 65% of offspring of males with chinmo1-mutant GSCs clones inherited the chinmo1 mutant allele and 35% inherited the GFP-positive sister chromosome (Fig. 6D, second bar, Table S3 #3, 4). We see a similar trend of biased inheritance with the chinmok13009 allele, (Fig. 6D, last bar, Table S3 #5, 6). These results indicate that GSC competition can cause biased inheritance. Declining Chinmo levels in GSCs cause physiological aging of the testis niche To assess whether the competition phenotypes could also be observed in physiological processes like aging and whether they also depended on reduced Chinmo expression, we examined testes from 2- (“young”) and 42-day-old (“aged”) males. Chinmo protein is expressed at a moderate level in GSCs in young testes (Fig. 7A,C) but is significantly decreased in GSCs in aged testes (Fig. 7B,C). A time course revealed that Chinmo is progressively and significantly decreased in GSCs during adulthood (Fig. 7C). The average total number of GSCs significantly declines during adulthood, and by 42 days there are on average 6.3 GSCs (Fig. 7D, Table S1 #283). This is similar to the decline in average total number of GSCs when chinmo−/− clones are induced in young males (Fig. 1L). A moat composed of Pcan and Lan is present in the lumen of aged but not young testes (Fig. 7E–K). Most testes from aged males had a moat, a significant increase compared to young males (Fig. 7I), that was phenotypically indistinguishable from the moat caused by competition (Fig. 2D,F). The moat is a valid age-related phenotype because it was observed two distinct genotypes, OregonR and nos>lacZ (see below). Additionally, GSCs in aged WT testes have increased Dg at the GSC-niche interface (Fig. 7M,N), similar to what we observe in chinmo−/− GSCs in younger males (Fig. 4B). Increased Dg was not observed in GSCs in young WT testes (Fig. 7L,N). To determine whether these aging phenotypes could be reserved by artificially maintaining Chinmo levels in GSCs throughout adulthood, we over-expressed Chinmo using nos (nos>chinmo) (Fig. 7O). As the control, we over-expressed a neutral protein lacZ using the same promoter (nos>lacZ). We found that all age-related phenotypes (the moat, Dg upregulation, and reduced GSC number) were significantly suppressed by maintaining moderate levels of Chinmo in GSCs throughout adulthood (Fig. 7O–X, Table S1 #284, 285). These data demonstrate that aging of the Drosophila testis stem cell niche is regulated by declining levels of Chinmo in GSCs. They also indicate that chinmo−/− GSCs in young males coopt an aging mechanism to remodel the niche to benefit themselves and to disadvantage non-mutant neighbors. Chinmo represses Pcan expression in somatic stem cells To test the possibility that Chinmo represses Pcan in other Drosophila stem cells to maintain tissue homeostasis, we examined the role of Pcan in male-to-female sex transformation of CySCs induced by loss of Chinmo. CySCs lacking chinmo transdifferentiate into ovarian, epithelial cells that express Fas3 and that generate an ectopic BL composed of Pcan (Fig. S7A–C”’, arrowheads, E). chinmo-mutant CySCs depleted for Pcan had significantly reduced Pcan deposition in the testis lumen (Fig. S7D–E), suggesting that the chinmo-deficient stem cells were the source of Pcan and that Chinmo normally represses Pcan in CySCs. Depleting Pcan from chinmo-deficient CySCs significantly reduced Fas3 expression, indicating that sex transformation was at least partially blocked (Fig. S7E). Thus, Chinmo represses Pcan expression in at least two distinct stem cell populations to maintain tissue function. DISCUSSION This work reveals an unexpected model of GSC competition that results in biased inheritance (Fig. 6E). These results provide mechanistic demonstration of the postulated “mitotic drive” by which germline stem cells with a competition advantage transmit competitive alleles at greater than the expected 50% Mendelian ratio (Otto and Hastings, 1998). Studies in plants, yeast, flies and mice have shown that selfish genetic elements can cause gene drive through various molecular mechanisms. These include “meiotic drivers” that coopt meiotic divisions or that kill viable gametes which do not inherit the selfish element (Bravo Nunez et al., 2018). “Mitotic drive” occurs earlier in the germline lineage, at the stem cell level, and is an understudied area. Prior work in the Drosophila ovary has shown that dedifferentiation-defective female GSCs “win” by upregulating E-Cad at the GSC-niche interface and gradually pushing WT GSCs out of the niche (Jin et al., 2008). Since differentiation-defective GSCs do not differentiate into gametes, this study could not examine biased allele inheritance. Female GSCs with 4x gene dose of Myc were reported to be “winners” but biased inheritance was not assessed (Rhiner et al., 2009). Given the competitive advantage of GSCs lacking chinmo, why has evolution not selected for male GSCs with no chinmo expression? Since chinmo is an essential gene required for development (Zhu et al., 2006), loss of chinmo in GSCs might cause reduced chinmo expression in other tissues, likely reducing organismal fitness. Furthermore, chinmo-dependent competition is a progressive phenotype requiring at least two weeks of adulthood. If males with mosaic chinmo−/− clones in the germline were mated as young adults, both the chinmo+ and chinmo− allele would be passed on to offspring, and this would be sufficient to maintain chinmo+ allele in the population. Although the GSC pool in testes with chinmo−/− cells is often monoclonal, why is the chinmo− allele is not passed on to 100% of offspring. We maintain males as virgins until we dissect their testes or until mating. This means that both chinmo− GFP-negative and ubi-GFP-positive spermatids are stored in the seminal vesicle throughout the male’s lifetime and the single round of mating in our experiments allows for the transmission of both types of sperm. We identified three phenotypes – competition, aging and transdifferentiation – that are dependent on ectopic Pcan expression in chinmo-deficient stem cells. This remarkable finding raises the important question of what factors regulate Chinmo expression in GSCs during adulthood and what genes are direct targets of Chinmo. We identified chinmo as a JAK/STAT target gene (Flaherty et al., 2010), and but STAT-deficient GSCs still express Chinmo protein (not shown), indicating that as-yet unidentified factors regulate Chinmo in GSCs and perhaps in other stem cells. Additionally, future molecular work will be needed to determine whether Pcan, Dg and βPS are direct Chinmo target genes in germline and somatic stem cells. Our work raises the possibility that other mutant stem cells can “cheat” by resculpting their microenvironment and then ensuring their own retention in this remodeled milieu. Paternal age effect disorders (PAEs) encompass a broad spectrum of spontaneous congenital disorders and are thought to arise from rare selfish GSCs have that are positively selected and clonally expand (Goriely and Wilkie, 2012). While the current model of PAE postulates increased proliferation of mutant GSCs as the competitive mechanism, other selfish cellular behaviors such of the ones we have discovered could also be functioning in the mammalian testis. Cancer stem cells could utilize the mechanisms described in our study to colonize a tissue. Leukemic stem cells induce progressive remodeling of the bone marrow niche, and this altered niche favors the mutant stem cells while impairing normal hematopoietic stem cell residence and contributes to bone marrow fibrosis (Schepers et al., 2013, Sperling et al., 2017). In sum, our work raises the possibility that selfish stem cells across species cheat using the mechanisms we have discovered in competitive GSCs in the Drosophila testis. Limitations of the Study Although our study found significant increases in Pcan, Dg and βPS transcripts in GSCs deficient for Chinmo, our HCR FISH analyses of relative intensity are not strictly quantitative. As such, we cannot rule out that Chinmo affects other aspects of mRNA regulation, such as splicing. Because we have as-yet not obtained the transcriptome of chinmo−/− GSCs at sufficient resolution and have not been able to perform Chinmo ChIP-seq in GSCs, we cannot conclude that Chinmo directly represses these genes. It will be important in the future to determine Chinmo occupancy on chromatin in GSCs and in other stem cells. Surprisingly, the niche spaces vacated by non-mutant neighbor GSCs are occupied by CySCs and not by chinmo−/− GSCs. We do not understand why the CySCs predominate, and future studies employing ex vivo live-imaging will be important to gain insights into this process. STAR METHODS RESOURCE AVAILABILITY Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Erika Bach (erika.bach@nyu.edu). Materials availability This study did not generate new unique reagents. All Drosophila stocks generated in this study are available from the Lead Contact without restriction. Data and code availability All data reported in this paper will be shared by the Lead Contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact upon request. EXPERIMENTAL MODEL AND SUBJECT DETAILS Fly Lines and Maintenance Drosophila melanogaster strains used in this study are listed in the Key Resources Table. Drosophila were reared on food made with these ingredients: 1800mL molasses (LabScientific, Catalog no. FLY-8008–16), 266 g agar (Mooragar, Catalog no. 41004), 1800 g cornmeal (LabScientific, Catalog no. FLY-8010–20), 744g Yeast (LabScientific, Catalog no. FLY-8040–20F), 47 L water, 56 g Tegosept (Sigma no. H3647–1KG), 560mL reagent alcohol (Fisher no. A962P4), and 190mL propionic acid (Fisher no. A258500). Flies were raised at 25°C except nos-Gal4 and mef2-Gal4 crosses, which were maintained at 18°C until eclosion, and the adult flies were transfer to 29°C. We used the following Drosophila stocks: nos-Gal4-VP16 (Van Doren et al., 1998); tj-Gal4 (Kyoto #104055); UAS-lacZ (BDSC #3956); UAS-Pcan-i #1 (BDSC #33642) and #2 (BDSC #29440); Pcan-i on II (VDRC #24549); UAS-control-i (BDSC #61501); UAS-chinmo-i (BDSC #33638); UAS-LanB1-i (BDSC #42616); UAS-βPS-i #1(BDSC #33642) and #2 (BDSC #27735); UAS-talin-i (BDSC #32999); UAS-Dg-i (BDSC #34895); mef2-Gal4 (BDSC #50742); UAS-Dg (Deng et al., 2003); βPS-GFP (Klapholz et al., 2015); UAS-PcanRG (Cho et al., 2012); chinmo1 (Zhu et al., 2006); chinmok13009; (Kyoto #111100); tub-Gal80ts (McGuire et al., 2004); UAS-5’UTR-chinmo-3’UTR (Zhu et al., 2006). For a list of full genotypes by figure, see Table S4. We used only adult Drosophila males in this study. METHODS DETAILS Drosophila Genetics and Clonal Analysis Negatively-marked GSC clones were generated using the FLP/FRT technique after a single 1 hour heat shock at 37°C in 2-day old adult males (Xu and Rubin, 1993). Males were returned to 25°C until dissection at 2, 7, 14, 21, 23, 28 dpci. Positively marked clones were generated by the MARCM technique after a single 1 hour heat shock at 37°C in 2-day old adult males (Lee and Luo, 1999). Males were returned to 25°C until dissection at 2, 14 dpci. Lineage-wide mis-expression or depletion was achieved using the Gal4/UAS system (Brand and Perrimon, 1993). A Gal80ts transgene was used with nos-Gal4 to deplete Dg or Talin from adult GSCs in RNA-i efficiency experiments in Fig. S1O,P,T,U,W and X (McGuire et al., 2004). tj-Gal4 was used to deplete Chinmo from CySCs for 5 days of adulthood in Fig. S7A–D. Immunofluorescence Dissections and staining were carried out as previously described (Flaherty et al., 2010). Briefly, testes were dissected in 1x phosphate buffered saline (PBS), fixed for 15 minutes in 4% paraformaldehyde (PFA) in 1xPBS, washed for 1 hour at 25°C in 1xPBS with 0.5% Triton X-100, and blocked in PBTB (1xPBS 0.2% Triton X-100 and 1% bovine serum albumin) for 1 hour at 25°C. Primary antibodies were incubated overnight at 4°C except for pSTAT which was incubated overnight at 25°C. They were washed two times for 30 minutes in PBTB and incubated for 2 hours in secondary antibody in PBTB at 25°C and then washed two times for 30 minute in 1xPBS with 0.2% Triton X-100. They were mounted in Vectashield or Vectashield with DAPI (Vector Laboratories). For 5-ethynyl-20-deoxyuridine (EdU, Life Technologies) labeling, samples were incubated for 30 minutes before fixation in Ringer’s medium containing 10 μM EdU. Testes were fixed and processed normally for antibody labeling and then treated per manufacturer’s instructions. Confocal images were captured using Zeiss LSM 510 and LSM 700 microscopes with a 63x objective. Z-stacks for 2D and 3D images were captured on a Nikon W1 spinning disk confocal microscope with lasers at 405, 488, 561, 640 nm, narrow pass filters for emission, a SR HP Plan Apo 100X 1.35 Silicon Oil λS DIC lens, and an Andor 888 Live EMCCD camera. Electron Microscopy Prior to fixation, we dissected 10 testes of each genotype and visualized the extent of GFP-positivity (indicating non-mutant germ cells) under a GFP-dissecting microscope. Germ cells are the most numerous cell type in the testis, and it is relatively straightforward to gain a rough assessment of germline clonality in mosaic testes using GFP. Most of the testes with control clones contained approximately equal levels of GFP-negative and GFP-positive germ cells. Most of the testes with chinmo−/− clones contained few GFP-positive germ cells, indicating that the clones had outcompeted the non-mutant neighbors. All of the samples were processed for TEM, and we are presenting in Fig. 2A,B representative examples of each genotype. Drosophila testes were fixed with 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) with 1 mM CaCl2 for 2 hours. After fixation, they were treated with 1% osmium tetroxide for 1 hour, followed by block staining with 1% uranyl acetate aqueous solution overnight at 4°C. The samples were rinsed in water, dehydrated in graded series of ethanol, infiltrated with propylene oxide/EMbed 812 mixtures and embedded in EMbed 812 resin (Electron Microscopy Sciences, PA USA). 70 nm ultra-thin sections were cut and mounted on 200 mesh copper grids and stained with uranyl acetate and lead citrate. Imaging was performed by Talos120C transmission electron microscope (Thermo Fisher Scientific, Hillsboro, OR) and recorded using Gatan (4k × 4k) OneView Camera with software Digital Micrograph (Gatan Inc., Pleasanton, CA). 3D Image Rendering – ImageJ 1.53 and Imaris 9.7 were used to generate 2D and 3D images in Fig. 2H,I, Fig. 7J,K and Movie S1. Inheritance Assay We induced control, chinmo1 or chinmok13009 GSC clones and aged the males for 21 days. Each male was mated singly to two OregonR (WT) virgins for 2 days. At 23 dpci, we dissected the testes of each mated male to determine the germline clonality. We scored the percentage of adult F1 offspring from the mated male for the inheritance of the chinmo+ allele (from control GSC clones) or the chinmo− allele (from chinmo1 or chinmok13009 GSC clones) by the lack of GFP expression under a Zeiss Stemi 11 GFP-dissecting scope. We scored inheritance in F1 offspring of the sister chromosome, which carries a ubi-GFP transgene, by the expression of GFP. Hybridization Chain Reaction (HCR) Fluorescent in situ Hybridization (HCR-FISH) We purchased from Molecular Instruments, Inc. the HCR probe set against Pcan, Dg and βPS mRNAs, the HCR amplifier, and the hybridization, wash, and amplification buffers. The protocol for immunostaining with HCR-FISH was adapted from (Zimmerman et al., 2013, Choi et al., 2018). Briefly, testes were fixed in 4% PFA in 0.1% Triton X-100 in 1xPBS-DEPC for 30 minutes at 25°C, washed with 0.5% Triton X-100 in 1xPBS-DEPC two times for 30 minutes at 25°C. Samples were blocked in 0.1% Triton X-100 in 1xPBS-DEPC with 50 μg/mL heparin and 250 μg/mL yeast tRNA (buffer hereafter called “PBTH”), and then they were then incubated with primary antibodies overnight at 4°C. The next day, the samples were washed twice in PBTH for 30 minutes. Samples were then incubated with fluorescently-labeled secondary antibodies in PBTH for 2 hours at 25°C. Samples were washed in PBTH twice for 30 minutes at 25°C. Samples were then dehydrated and rehydrated with a series of ethanol washes (25%, 50%, 75%, 100%) in 1xPBS-DEPC for 10 minutes at 25°C. Samples were treated for 7 minutes with 50 μg/mL Proteinase K, which was then inactivated by washing with 0.2% glycine twice in 1xPBS-DEPC for 5 minutes at 25°C. After Proteinase K treatment, the samples were fixed again in 4% PFA in 1xPBS-DEPC for 30 minutes at 25°C. The re-fixed samples were pre-hybridized in hybridization buffer provided by Molecular Instruments Inc. for 10 minutes at 25°C and then incubated with HCR probes overnight (12 – 16 hr) at 37°C. Samples were then washed 6 times 10 minutes at 37°C with wash buffer provided by Molecular Instruments Inc. and then twice for 5 min in 5xSCC at 25°C. Samples were incubated in amplification buffer provided by Molecular Instruments Inc. for 5 minutes at 25°C. The secondary reagents called “Hairpin h1 DNA” and “Hairpin h2 DNA” were prepared by heating each for 90 seconds at 95°C and cooling them at 25°C in a dark drawer for 30 minutes. Hairpin h1 DNA and Hairpin h2 DNA were mixed together at a 1:1 ratio and then added to the samples, which were then incubated in the dark environment overnight (16 hr) at 25°C. Samples were washed 6 times for 5 minutes with 5xSSC at 25°C and mounted in Vectashield plus DAPI for confocal analysis. QUANTIFICATION AND STATISTICAL ANALYSIS Quantification of Pcan/Lan in the moat We captured confocal z-stacks (at 1 μM intervals) encompassing all niche cells (typically 16–20 slices) in testes with GSC clones and stained then with Pcan or Lan antibodies. The niche was counted as “with Pan/Lan” if the ECM protein was present next to at least one niche cell facing the testis lumen in any of the slices. Quantification of βPS-GFP/Dg/Talin/E-Cad expression at the GSC-niche interface We captured confocal z-stacks (at 1 μM intervals) encompassing all niche cells (typically 16–20 slices). We measured fluorescence intensity by ImageJ in single z slices at the area of maximal contact of a GSC with the niche. The background signal was measured in the nucleus of a gonialblast then subtracted from each GSC-niche measurement. In the mis-expression/knockdown experiments performed using the Gal4/UAS technique, each data point represents the intensity of βPS-GFP/Dg/Talin in one GSC. In the clonal analyses, the fluorescence intensity of Dg/Talin/E-Cad at the GSC-niche interface of the clone was normalized to that of a non-mutant neighbor GSC. In the clonal analyses, each data point represents one GSC. Quantification of Chinmo protein and Pcan, Dg, βPS mRNAs in GSCs We captured confocal z-stacks (at 1 μM intervals) encompassing all niche cells (typically 16–20 slices). We measured Chinmo (or HCR-probe) fluorescence intensity by ImageJ in a single z slice taken at the maximal width of the GSC. The background signal was measured in the nucleus of a muscle cell then subtracted from each measurement. Each data point represents one GSC. Quantification of Pcan/Lan expression in the muscle basal lamina and in the testis lumen We captured confocal z-stacks (at 1 μM intervals) encompassing the entire width of testes including the basal lamina and all niche cells (typically 20–25 slices). Measurements were performed using ImageJ on a single z section taken at the position where the niche attaches to the basal lamina. The background signal was measured in the nucleus of a niche cell then subtracted from each measurement. Each data point represents Pcan/Lan intensity in one testis. Quantification of mis-oriented centrosomes in GSC clones Centrosomes were labeled with a γ-tubulin antibody. GSCs with mis-oriented centrosomes were defined as having neither the mother nor the daughter centrosome located next to the niche as described in (Cheng et al., 2008). To measure the percentage of GSCs with mis-oriented centrosomes, we scanned each testis with GSC clones on a laser scanning confocal microscope. Z-sections were taken at 1 μM intervals (typically 16–20 slices in total/testis). We calculated the number of GFP-negative GSC clones and GFP-positive non-mutant neighbor GSCs with mis-oriented centrosomes (Fig. S2F). Quantification of the GSC-niche distance To measure the distance from GSCs to the niche (Fig. S3C), we used ImageJ to analyze images of testes captured on a laser scanning confocal microscope. We measured the distance from the GSC plasma membrane to the edge of the closest niche cell for each GSC in testes with control or chinmo−/− GSCs clones at 28 dpci. Quantification of the number of testes with a moat Testis with control or chinmo1 GSC clones were examined at 2, 7, 14, 21, 28 dpci for a gap (corresponding to the moat) between the GSCs and niche cells (Fig. S3D). Statistical Analysis Statistical analyses were performed using two-tailed Student’s t-tests except in Fig. 1K, which was performed using two-way Anova, and in Figs. 2G, 2N, 2O, 5F, 6D, 7I, S2F, S3K, and S7E, which were performed using χ2 tests; and Fig. S3D, which was performed by a Mann-Whitney test. Data were analyzed by GraphPad Prism and Microsoft Excel. Statistical significance was assumed by P < 0.05. Individual P values are indicated. Data are represented by the mean and standard error of mean (SEM), except Fig. S3D in which data are represented by the mean and standard deviation (SD). Supplementary Material 1 Table S1 (excel file): CySC counts, clone occupancy, GSC number, number of neighbors GSC, and number of GSC clones in the analyzed genotypes (control and chinmo GSC clones), related to Figs. 1, 3, 4 and 7. 2 Table S4 (excel file): genotypes analyzed, related to Figs. 1–7 3 4 Movie S1 (mp4 file), related to Fig. 2 mp4 movie of Imaris-generated views of a testis with control GSC clones (left) and with chinmo-mutant GSC clones (right). Niche cells are shown by their blue nuclei and Lan (red) is highly expressed in the muscle basal lamina. The time point is 14 dpci. See also Fig. 2H,I. (Left) Niche cells are visible as a ball. Some niche cells adhere to Lan present in the muscle basal lamina, thereby anchoring the niche to the apical tip of the testis. (Right) Niche cells are visible as a ball. Some niche cells adhere to and are partially obscured by a distended “hat” of Lan (upper arrow labeled “moat”) that is contiguous with Lan in the muscle basal lamina. In addition, there is ectopic Lan in the testis lumen. Some of this ectopic Lan is present as three “claws” on distal niche cells (lower arrow labeled “moat”). An “tendril” of ectopic Lan extends from these three “claws” to the right side towards the muscle basal lamina. ACKNOWLEDGEMENTS We thank NYU Langone Health DART Microscopy Laboratory, A. Liang, C. Petzold and K. Dancel-Manning for consultation and assistance with TEM work. The Microscopy Laboratory is partially supported by Laura and Isaac Perlmutter Cancer Center Support Grant NIH/NCI P30CA016087. We thank K. White, D. Godt, P. Rangan, N. Sokol, T. Volk, G. Morata, E. Laufer, B. Hudson, M. Ringuette, S. Hayashi, M. Crozatier, Y. Nakanishi for antibodies, and T. Lee, A. Kolodkin, N. Brown, Bloomington Stock Center (BDSC) and Kyoto Stock Center for fly stocks. The BDSC is supported by a grant from the Office of the Director of the NIH (P40OD018537). We are grateful to FlyBase, which is supported by a grant from the National Human Genome Research Institute at NIH (U41 HG000739). Work is the Bach lab is supported by grants from the NIH (R03-HD090422; R01-GM085075). CYT was supported by a New York State Department of Health/NYSTEM institutional training grant (#C322560GG). Figure 1: chinmo−/− GSCs dominate the niche by evicting non-mutant GSCs. (A) The adult Drosophila testis. A GSC produces a gonialblast (Gb), which undergoes transita-mplifying divisions to produce spermatogonia that differentiate into sperm. CySCs divide to produce cyst cells, two of which envelope a Gb and its descendants. (B) The clone occupancy assay. In a WT testis, 8 GSCs surround the niche. GSC clones (either control or chinmo−/−) lack GFP. The other GSCs (labeled “neighbors”) are not mutant and express GFP. Clone occupancy is measured by dividing the number of GSC clones by the total number of GSCs in that testis. (C-F) Confocal images of testes with control (C,E, arrows) or chinmo−/− (D,F, arrows) GSC clones at 2 (C,D) or 28 (E,F) dpci. Clones lack GFP. Vasa (red) marks germ cells. Tj (blue) marks the nuclei of CySCs and early cyst cells. Arrowheads mark non-mutant GSC neighbors. (G-H) Confocal images of testes with control (G, arrow) or chinmo−/− (H, arrow) clones at 21 dpci stained with αSpectrin (red) to mark the fusome. Clones lack GFP. Vasa (blue). (I) Graph showing the number of Zfh1-positive CySCs in testes with control (gray), chinmo1 (blue) and chinmok13009 GSC clones (purple) at 2 and 28 dpci. (J) Graph showing clone occupancy of control (gray) and chinmo−/− (blue) GSC clones at 2, 7, 14, 21, 28 dpci. (K) Graph showing percent monoclonal testes within the GSC lineage at 2, 7, 12, 21 and 28 dpci when control (gray) or chinmo−/− (blue) GSC clones are present. (L) Graph showing the average total number of GSCs in testes with control (gray) or chinmo−/− GSC clones (blue) at 2, 7, 14, 21 and 28 dpci. (M) Graph showing the average number of control (dashed line) or chinmo−/− (solid line) GSC clones at 2, 7, 14, 21 and 28 dpci. (N) Graph showing non-mutant GSC neighbors in testes with control (dashed line) or chinmo−/− (solid line) GSC clones at 2, 7, 14, 21 and 28 dpci. (O) Model. Control GSC clones (gray cells, upper panel) and chinmo−/− GSC clones (gray cells, lower panel) are induced at the same low frequency. Over time, both types of clones expanded to a similar extent (gray cells). chinmo−/− clones cause the loss of non-mutant neighbor GSCs (green cells, lower panel) and this does not occur to the non-mutant neighbors (green cells, upper panel) of control GSC clones. In C-F, G,H, an asterisk marks the niche. Scale bar = 10 μM In I,J,L,M,N, error bars represent SEM. n.s. = not significant; * P ≤ 0.05; *** P ≤ 0.001; **** P ≤ 0.0001 as assessed by Student’s t-test (I,J,LM,N) or by χ2 test (K). See also Table S1, Fig. S1. Figure 2: chinmo−/− GSCs create a “moat” around the niche by secreting Pcan (A,B) TEMs of testes with control (A) or chinmo1 (B) GSC clones at 28 dpci. The micrographs are pseudocolored to show ECM-like material (light blue) in the muscle basal lamina (“BL”, purple arrowhead in A,B) or in the testis lumen (yellow arrowhead, B). Yellow arrowheads indicate GSC-niche interface. Magnification 5,600x. Scale bar is 2 μM. (C-F) Pcan (C,D, red) and Lan (E,F, red) in testes with control (C,E arrows) chinmo−/− GSC (D,F, arrow) clones at 21 dpci. Arrowhead (D’,F’) indicates ectopic ECM. (G) Graph quantifying the percentage testes with (gray portion of bar) or without (white portion of bar) ECM proteins Pcan or Lan surrounding the niche when control, chinmo1 or chinmok13009 clones are present. (H,I) Imaris-generated view of a testis with control (H) or chinmo−/− (I) GSC clones (not shown) at 14 dpci. Niche cells (blue) are visible as a ball, and some niche cells adhere to Lan (red) present in the muscle (H,I). In I, some niche cells are partially obscured by a distended “hat” of Lan (red, upper arrow labeled “moat”) that is contiguous with Lan in the muscle. In I, there is ectopic Lan in the testis lumen, which form three “claws” on distal niche cells (lower arrow labeled “moat”). A “tendril” of ectopic Lan extends to the right. In H’,I’, the testis is rotated 90°. All niche cells are visible in H’, but in I’ several niche cells are covered by Lan (arrows). (J-M) Expression of Pcan (red in J,K) and Lan (red in L,M) around the niche when chinmo−/− GSC clones (J,L) or chinmo−/− clones depleted for Pcan (K,M) are present. Arrowheads indicate ectopic ECM. (N-O) Graph quantifying the percentage of testes with (gray portion of bar) or without (white portion of bar) Pcan (N) or Lan (O) in the moat when chinmo−/− clones (first bars), chinmo−/− clones depleted for Pcan (second bars), for Dg (third bars) or for βPS (fourth bars) are present. In A-F, clones lack GFP. In H-M, clones express GFP. In C-F, J-M, Vasa is blue and an asterisk marks the niche. Scale bar = 10 μM * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001 as assessed by χ2 test. See also Figs. S1, S2, S4, S7. Figure 3: chinmo−/− GSCs require Pcan to evict non-mutant neighbors (A,D) Confocal images of nos>lacZ (A,C) or nos>Pcan (B,D) testes in which Pcan was mis-expressed in all GSCs for 21 days. Arrowheads (B,B’,D) indicate ectopic ECM. Pcan (green, A,B); Lan (green, C,D); Vasa (red) and DNA (ToPro, blue A,B); Fas3 (blue, C,D). (E) Graph showing average number of GSCs in nos>lacZ (gray) or nos>Pcan (yellow) testes. (F-I) Box and whisker plots showing clone occupancy (F), average total number of GSCs (G), average number of non-mutant GSC neighbors (H); average number of clones (I) in testis with control GSC clones (dark gray bars), with control GSC clones depleted for Pcan (light gray bars), with chinmo−/− GSC clones (dark blue bars) or chinmo−/− GSC clones depleted for Pcan (light blue bars) at 2 and 14 dpci. (J) Model: chinmo−/− GSC clones (gray cells) secrete Pcan (dark blue symbol), which seeds the moat. Lan (light blue symbol) is recruited from the muscle BL. Boxed area at right illustrates a chinmo−/− GSC clone in contact with the moat. By contrast, non-mutant neighbor GSCs (green stem cells in middle cartoon) are lost from the niche. The smaller, light green cells are niche cells. Scale bar = 10 μM In F-I, error bars represent SEM. n.s. = not significant; * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001; **** P ≤ 0.0001, as assessed by Student’s t-test. See also Table S1, Figs. S1, S3, S7. Figure 4: chinmo−/− GSCs require Dg to remain in the altered niche (A,B) Confocal images of Dg (red) in testes harboring control (A, arrow) or chinmok13009-mutant GSC clones (B, arrows) at 14 dpci. Orange arrowhead (A’), magenta arrowhead (A’) and blue arrowheads (B’) indicate Dg at the GSC-niche in a non-mutant GSC neighbor, a control clone and a chinmo−/− GSC, respectively. Clones lack GFP. Neighbors (yellow arrowhead) express GFP. Vasa (blue). (C) Graph of relative Dg expression at the GSC-niche interface in control, chinmo1 or chinmok13009 GSC clones relative to that of neighbor GSCs in the same testis. (D-G) Box and whisker plots showing clone occupancy (D), average total number of GSCs (E), average number of non-mutant GSC neighbors (F), and average numbers of GSC clones (G) in testes with control GSC clones (dark gray bars), with control GSC clones depleted for Dg (light gray bars), with chinmo−/− GSC clones (dark blue bars) or chinmo−/− GSC clones depleted for Dg (light blue bars) at 2 and 14 dpci. (H) Model: chinmo−/− GSC clones (gray cells) have increased Dg (orange symbol) at the GSC-niche interface, allowing them to remain in the resculpted niche. Boxed area at right illustrates a chinmo−/− GSC clone remaining in contact with the moat through increased localized Dg expression. Non-mutant neighbor GSCs do not have increased Dg at the GSC-niche interface and cannot remain long-term in the niche. Scale bar = 10 μM In D-G, error bars represent SEM. * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001; **** P ≤ 0.0001, as assessed by Student’s t-test. See also Table S1, Figs. S1, S4, S5, S6. Figure 5: Neighbors are rescued from competition when provide with ectopic Dg (A) Model: If non-mutant neighbor GSCs (green) are provided with increased Dg (dark orange), they should be able to remain in the niche despite the induction of chinmo−/− GSC clones (gray) and the formation of the moat (Pan, Lan, blue symbols). Yellow symbols represent integrins. (B) In the “neighbor rescue” assay, chinmo−/− clones (abbreviated chinmomut) were generated in a background where all GSCs express UAS-Dg driven by the GSC driver nos-Gal4. We scored the number of non-mutant neighbors at 2 and 14 dpci (see G), average number of GSCs (see H) and presence a moat (see C,D,F). (C-E) Pcan (red) in testes containing chinmok13009 GSC clones in nos>lacZ (C) or nos>Dg (D). The chinmo clone is not visible in D (“focal plan #1”) but is visible in E (arrow, “focal plane #2”). Clones lack GFP. Pcan (red). Vasa (blue). (F) Graph showing percentage of testes with a “moat” when chinmok13009 GSC clones are generated in a nos>lacZ or nos>Dg background. (G, H) Box and whisker plots showing average number of non-mutant GSC neighbors (G) or average number of GSCs (H) in testis with control GSC clones in nos>lacZ (dark gray bars), control GSC clones in nos>Dg (light gray bars), chinmok13009 GSC clones in nos>lacZ (dark blue bars), or chinmok13009 GSC clones in nos>Dg (light blue bars) at 2 and 14 dpci. Scale bar = 10 μM In G,H, error bars represent SEM. n.s. = not significant; ** P ≤ 0.01; *** P ≤ 0.001 as assessed by Student’s t-test (G,H) and by χ2 test (F). See also Table S2. Figure 6: GSC competition causes biased inheritance (A) Schematic of the “inheritance assay” - see STAR Methods for details. Box shows expected outcomes. (B,C) Confocal images of a testis with a control (B,B’, arrow) or a chinmok13009 GSC clone (C,C’, arrows) at 23 dpci. Non-mutant neighbor GSCs are marked by an arrowhead. The only GFP-positive cells in C are somatic support cells. Clones lack GFP. Vasa is red. (D) Graph showing inheritance of the chinmo chromosome (chinmo+, chinmo1 or chinmok13009 allele) (in black) or the ubi-GFP chromosome (in green). (E) Model. chinmo−/− GSC clones (gray) secrete Pcan (dark blue), which causes the moat. Lan (light blue) is recruited to the moat from the muscle BL. chinmo−/− GSC clones increase Dg (orange) and βPS integrin (yellow) at the GSC-niche interface, allowing them to remain in the resculpted niche but non-mutant neighbor GSCs (green cells) do not and differentiate. Scale bar = 10 μM *** P ≤ 0.001 as assessed by χ2 test (D). See also Table S3. Figure 7: Age-related phenotypes of the testis stem cell niche are caused by declining Chinmo levels in GSCs (A-D) Confocal images of young (2-day-old) (A) or aged (42-day-old) (B) WT testes. Chinmo (green). Arrowheads indicate GSCs in A,B. (C,D) Graphs showing Chinmo levels in GSCs (C) and average number of GSCs/testis (D) during aging. (E-H) Confocal image of Pcan (green in E,F) and Lan (green in G,H) in young (E,G) and aged testes (F,H). (I) Graph showing percentage of testis with Pcan (orange) or Lan (yellow) surrounding the niche. (J,K) Imaris-generated views of young (J) and aged (K) testes. Niche cells (blue) are visible as a ball (J,K). In a young testis, some niche cells interact with Lan (red) present in muscle BL (J). In an aged testis, some niche cells are partially obscured by ectopic Lan (red) in the testis lumen (K). (L-N) Confocal image of young (L) and aged (M) testis stained with Dg (green). (N) Graph showing relative Dg levels at the GSC-niche interface in young or old testes. (O) Graph showing Chinmo intensity in GSCs in nos>lacZ (gray) or nos>chinmo (orange). (P-T) Confocal images of aged nos>lacZ (P,R) and aged nos>chinmo (Q,S) testes stained for Pcan (green in P,Q) or Lan (green in R,S). (T) Graph showing ECM intensity in aged nos>lacZ (gray) or aged nos>chinmo (orange) testes. (U-X) Confocal images of aged nos>lacZ (U) or aged nos>chinmo (V) testes stained with Dg (green). Arrowheads indicate Dg at GSC-niche interface. (W) Graph of Dg intensity at the GSC-niche interface in aged nos>lacZ (gray) or in aged nos>chinmo (orange) testes. (X) Graph showing number of GSCs in aged nos>lacZ (gray) or in aged nos>chinmo (orange) testes. In A,B,E-H,L,M,P-S,U,V, Vasa is red and Fas3 is blue. Scale bar = 10 μM In C,D,N,O,T,W,X, error bars represent SEM. * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001; **** P ≤ 0.0001 as assessed by Student’s t-test (C,D,N,O,T,W,X) and by χ2 test (I). See also Table S1. Key Resource Table REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Rabbit polyclonal anti-GFP (1:1000) Invitrogen Cat# A6455 Goat polyclonal anti-Vasa (1:200) Santa Cruz Cat# sc26877; RPID: AB_793877 Rabbit polyclonal anti-Zfh1 (1:200) K. White (University of Chicago, USA) N/A Guinea pig polyclonal anti-Traffic jam (Tj) (1:1000) D. Godt (University of Toronto, Canada) N/A Rabbit polyclonal anti-Pcan (1:1000) Baumgartner lab N/A Chicken polyclonal anti-Vasa (1:200) P. Rangan (SUNY, Albany, USA) N/A Rat anti-Chinmo (1:200) N. Sokol (Indiana University, USA) N/A Chicken polyclonal anti-GFP (1:1000) Abcam Cat# ab13970 RRID:AB_300798 Guinea pig anti-Lan (1:1000) T. Volk (Weizmann Institute, Israel) N/A Rabbit polyclonal anti-Dg (1:500) Baumgartner lab N/A Mouse monoclonal anti-Talin carboxy terminus 534 amino acids (1:20) Developmental Studies Hybridoma Bank (DSHB) Cat# Talin E16B, RRID:AB_10683995 Mouse monoclonal anti-Talin carboxy terminus 534 amino acids (1:20) DSHB Cat# Talin A22A, RRID:AB_10660289 Mouse monoclonal anti-β-galactosidase (1:50) DSHB Cat# 40-1a RRID:AB_528100 Mouse monoclonal anti-βPS integrin (1:20) DSHB Cat# cf.6g11 RRID:AB_528310 Guinea pig anti-Myc (1:50) G. Morata (CSIC-UAM, Spain) N/A Rabbit anti-cleaved Dcp-1 (1:500) Cell Signaling Cat# 9578 RRID:AB_2721060 Mouse monoclonal anti-γ-tubulin (1:1000) Sigma-Aldrich Cat# T6557 RRID:AB_477584 Rabbit polyclonal anti-phospho-Mad (1:1250) E. Laufer (Columbia University, USA) N/A Rabbit polyclonal anti-pSTAT (1:50) Bach lab N/A Rat monoclonal anti-E-Cad (1:5) DSHB Cat# DCAD2 RRID:AB_528120 Rabbit polyclonal anti-ColIV (1:500) B. Hudson (Vanderbilt University, USA) N/A Rabbit polyclonal anti-Testican (1:500) Baumgartner lab N/A Rabbit polyclonal antiNidogen (1:400) Baumgartner lab N/A Rabbit polyclonal anti-Sparc (1:200) M. Ringuette (University of Toronto, Canada) N/A Mouse monoclonal anti-αPS1 integrin (1:100) DSHB Cat# dk.1a4 RRID:AB_528303 Mouse monoclonal anti-αPS2 integrin (1:100) DSHB Cat# cf.2c7 RRID:AB_528304 Rabbit polyclonal anti-αPS3 integrin (1:100) S. Hayashi (RIKEN Center for Developmental Biology, Japan) N/A Rabbit polyclonal anti-αPS4 integrin (1:100) M. Crozatier (Université de Toulouse, France) N/A Mouse monoclonal anti-βν integrin (1:200) Y. Nakanishi (Kanazawa University, Japan) N/A Mouse monoclonal anti-Drosophila α-Spectrin (1:20) DSHB Cat# 3A9 RRID:AB_528473 Cy3-AffiniPure Donkey Anti-Mouse IgG (1:400) Jackson ImmunoResearch Labs Cat# 715-165-150 RRID:AB_2340813 Alexa Fluor 488-AffiniPure Donkey Anti-Rabbit IgG (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 711-545-152 RRID:AB_2313584 Cy3-AffiniPure Donkey Anti-Rabbit IgG (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 711-165-152 RRID:AB_2307443 Cy5-AffiniPure Donkey Anti-Rabbit IgG (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 711-175-152 RRID:AB_2340607 Alexa Fluor 488-AffiniPure Donkey Anti-Rat IgG (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 712-545-150 RRID:AB_2340683 Cy3-AffiniPure Donkey Anti-Rat IgG (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 712-165-150, RRID:AB_2340666 Cy5-AffiniPure Donkey Anti-Rat IgG (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 712-175-150, RRID:AB_2340671 Alexa Fluor 488 AffiniPure Donkey Anti-Chicken IgY (IgG) (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 703-545-155, RRID:AB_2340375 Cy3-AffiniPure Donkey Anti-Chicken IgY (IgG) (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 703-165-155, RRID:AB_2340363 Cy5-AffiniPure Donkey Anti-Chicken IgY (IgG) (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 703-175-155, RRID:AB_2340365 Cy3-AffiniPure Donkey Anti-Guinea Pig IgG (1:400) Jackson ImmunoResearch Labs Cat# 706-165-148, RRID:AB_2340460 Cy5-AffiniPure Donkey Anti-Guinea Pig IgG (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 706-175-148, RRID:AB_2340462 Cy3-AffiniPure Donkey Anti-Goat IgG (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 705-165-003, RRID:AB_2340411 Alexa Fluor 647 AffiniPure Donkey Anti-Goat IgG (H+L) (1:400) Jackson ImmunoResearch Labs Cat# 705-605-003, RRID:AB_2340436 Chemicals, peptides, and recombinant proteins VECTASHIELD Mounting Medium with DAPI Vector Laboratories Cat# H-1200, RRID:AB_2336790 VECTASHIELD Mounting Vector Laboratories Cat# H-1000, RRID:AB_2336789 Paraformaldehyde, 16% w/v aq. soln., methanol free (PFA) Thermo Fisher Scientific Cat# 43368-9L Heparin Sigma-Aldrich Cat# H4784 tRNA Roche Cat# 10109495001 Protector RNase Inhibitor Roche Cat# 3335399001 20 × Saline Sodium Citrate (SSC) Thermo Fisher Cat# 15557-044 Proteinase K Thermo Fisher Scientific Cat# EO0491 Glycine Thermo Fisher Scientific Cat# BP381-500 Diethyl pyrocarbonate (DEPC) MilliporeSigma Cat# D5758 Click-iT EdU Imaging Kits Thermo Fisher Scientific Cat# C10340 Molasses Labscientific Cat# FLY-8008-16 Agar Mooragar Cat# 41004 Cornmeal LabScientific Cat# FLY-8010-20 Yeast LabScientific Cat# FLY-8040-20F Tegosept Sigma Cat# H3647-1KG Reagent alcohol Fisher Cat# A962P4 Propionic acid Fisher Cat# A258500 Experimental Models: Organisms/Strains D. melanogaster; y, w, hs-flp122/Y; ubi-GFP, FRT40A Bach lab N/A D. melanogaster; w; chinmo1,UAS-mCD8-GFP, FRT40A/CyO T. Lee (Janelia Research Camp, USA) N/A D. melanogaster; w; P[ry[+t7.2]=neoFRT]40A; Bach lab N/A D. melanogaster; y, w, hs-flp122, tub-Gal4, UAS-nls-GFP/Y; tub-Gal80, FRT40A/CyO Bach lab N/A D. melanogaster; y1 v1; P[TRiP.GL01153]attP2 (Pcan-i #1) BDSC RRID:BDSC_42783 D. melanogaster; y1 v1; P[TRiP.JF03376]attP2 (Pcan-i #2) BDSC RRID:BDSC_29440 D. melanogaster; y, w; P[w+, nos-GAL4 VP16] on II Bach lab N/A D. melanogaster; w1118; P[w[+mC]=UAS-lacZ.NZ]J312 Insertion on III BDSC RRID:BDSC_3956 D. melanogaster; w; UAS-PcanRG Insertion on III A. Kolodkin (Johns Hopkins School of Medicine, USA) N/A D. melanogaster; y1 sc* v1 sev21;P[TRiP.HMS01240]attP 2(Dg-i) BDSC RRID:BDSC_34895 D. melanogaster; w; P[lacW]chinmok13009, FRT40A/CyO Kyoto Stock Center RRID:DGGR_111100 D. melanogaster; y, w, βPS-GFP N. Brown (University of Cambridge, UK) N/A D. melanogaster; y1 v1; P[TRiP.HMS00036]attP2/TM3, Sb1 (chinmo-i) BDSC RRID:BDSC_33638 D. melanogaster; y1 v1; P[TRiP.HMS00043]attP2 (βPS-i #1) BDSC RRID:BDSC_33642 D. melanogaster; y1 v1; P[TRiP.JF02819]attP2 (βPS-i #2) BDSC RRID:BDSC_27735 D. melanogaster; w; UAS-Dg Insertion on III W. Deng (Tulane University, USA) N/A D. melanogaster; y1 sc* v1 sev21; P[TRiP.HMS00799]attP2 (talin-i) BDSC RRID:BDSC_32999 D. melanogaster; y1 sc* v1 sev21;P[TRiP.HMS02451]attP 2(Lan-i) BDSC RRID:BDSC_42616 D.melanogaster; w*; P[Mef2-Gal4.247]3 BDSC RRID:BDSC_50742 D. melanogaster; y1 sc* v1 sev21; P[TRiP.HMS00799]attP2 (talin-i) BDSC BDSC: 32999 FlyBase: FBst0032999 D. melanogaster; w; +; UAS-5’UTR-chinmo-3’UTR T. Lee (Janelia Research Camp, USA) N/A D. melanogaster; w*; P[UAS-mCherry.scramble.sponge]att P40; P[UAS mCherry.scramble.sponge]att P2 (control-i) BDSC RRID:BDSC_61507 D. melanogaster; w1118; P[GD7744]v24549 (Pcan-i on II) VDRC RRID:SCR_24549 D. melanogaster; w*; P[GawB]NP1624/CyO; (tj-Gal4) Bach lab RRID:DGGR_104055 D. melanogaster; w 1118 ; P[w+mC=UAS-Dcr-2.D]10 (UAS-Dcr-2) Bach lab RRID:BDSC_24651 D.melanogaster; w*; P[tubP-Gal80ts]2/TM2 BDSC RRID:BDSC_7071 Oligonucleotide HCR probe of Pcan Molecular Instruments, Inc. LOT: PRJ993 HCR probe of Dg Molecular Instruments, Inc. LOT: PRJ994 HCR probe of βPS Molecular Instruments, Inc. LOT: PRJ995 Software and algorithms ImageJ/Fiji Fiji http://fiji.sc/ Photoshop/Illustrator Adobe https://www.adobe.com/products/ Prism GraphPad https://www.graphpad.com ZEN Zeiss https://www.zeiss.com/microscopy/us/products/microscopesoftware/zen.html Excel Microsoft https://products.office.com/en-us/excel Imaris Oxford Instruments https://imaris.oxinst.com/ Highlights chinmo-mutant GSCs secrete ECM proteins to remodel the niche, evicting WT GSCs. chinmo-mutant GSCs remain in the altered niche by increasing ECM-binding proteins. Inheritance of the chinmo-mutant allele is biased and occurs in >50% of F1 progeny. Aged testes have a remodeled niche caused by declining levels of Chinmo in GSCs. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. DECLARATION OF INTERESTS The authors declare no interests. REFERENCES AMOYEL M , ANDERSON J , SUISSE A , GLASNER J & BACH EA 2016. Socs36E Controls Niche Competition by Repressing MAPK Signaling in the Drosophila Testis. PLoS Genet, 12 , e1005815.26807580 AMOYEL M & BACH EA 2014. Cell competition: how to eliminate your neighbours. Development, 141 , 988–1000.24550108 AMOYEL M , SIMONS BD & BACH EA 2014. Neutral competition of stem cells is skewed by proliferative changes downstream of Hh and Hpo. EMBO J, 33 , 2295–313.25092766 BRAND AH & PERRIMON N 1993. Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development, 118 , 401–15.8223268 BRAVO NUNEZ MA , NUCKOLLS NL & ZANDERS SE 2018. Genetic Villains: Killer Meiotic Drivers. Trends Genet, 34 , 424–433.29499907 BROADIE K , BAUMGARTNER S & PROKOP A 2011. Extracellular matrix and its receptors in Drosophila neural development. Dev Neurobiol, 71 , 1102–30.21688401 CHENG J , TURKEL N , HEMATI N , FULLER MT , HUNT AJ & YAMASHITA YM 2008. Centrosome misorientation reduces stem cell division during ageing. Nature, 456 , 599–604.18923395 CHO JY , CHAK K , ANDREONE BJ , WOOLEY JR & KOLODKIN AL 2012. The extracellular matrix proteoglycan perlecan facilitates transmembrane semaphorin-mediated repulsive guidance. Genes Dev, 26 , 2222–35.23028146 CHOI HMT , SCHWARZKOPF M , FORNACE ME , ACHARYA A , ARTAVANIS G , STEGMAIER J , CUNHA A & PIERCE NA 2018. Third-generation in situ hybridization chain reaction: multiplexed, quantitative, sensitive, versatile, robust. Development, 145 . DENG WM , SCHNEIDER M , FROCK R , CASTILLEJO-LOPEZ C , GAMAN EA , BAUMGARTNER S & RUOHOLA-BAKER H 2003. Dystroglycan is required for polarizing the epithelial cells and the oocyte in Drosophila. Development, 130 , 173–84.12441301 FABRIZIO JJ , BOYLE M & DINARDO S 2003. A somatic role for eyes absent (eya) and sine oculis (so) in Drosophila spermatocyte development. Dev Biol, 258 , 117–28.12781687 FLAHERTY MS , SALIS P , EVANS CJ , EKAS LA , MAROUF A , ZAVADIL J , BANERJEE U & BACH EA 2010. chinmo is a functional effector of the JAK/STAT pathway that regulates eye development, tumor formation, and stem cell self-renewal in Drosophila. Dev Cell, 18 , 556–68.20412771 FULLER MT 1998. Genetic control of cell proliferation and differentiation in Drosophila spermatogenesis. Semin Cell Dev Biol, 9 , 433–44.9813190 GORIELY A & WILKIE AO 2012. Paternal age effect mutations and selfish spermatogonial selection: causes and consequences for human disease. Am J Hum Genet, 90 , 175–200.22325359 GREENSPAN LJ , DE CUEVAS M & MATUNIS E 2015. Genetics of gonadal stem cell renewal. Annu Rev Cell Dev Biol, 31 , 291–315.26355592 GRMAI L , HARSH S , LU S , KORMAN A , DEB IB & BACH EA 2021. Transcriptomic analysis of feminizing somatic stem cells in the Drosophila testis reveals putative downstream effectors of the transcription factor Chinmo. G3-Genes Genomes Genetics, 11 . GRMAI L , HUDRY B , MIGUEL-ALIAGA I & BACH EA 2018. Chinmo prevents transformer alternative splicing to maintain male sex identity. PLoS Genet, 14 , e1007203.29389999 HARDY RW , TOKUYASU KT , LINDSLEY DL & GARAVITO M 1979. The germinal proliferation center in the testis of Drosophila melanogaster. J Ultrastruct Res, 69 , 180–90.114676 HASTINGS IM 1989. Potential germline competition in animals and its evolutionary implications. Genetics, 123 , 191–7.2806882 ISSIGONIS M , TULINA N , DE CUEVAS M , BRAWLEY C , SANDLER L & MATUNIS E 2009. JAK-STAT signal inhibition regulates competition in the Drosophila testis stem cell niche. Science, 326 , 153–6.19797664 JIN Z , KIRILLY D , WENG C , KAWASE E , SONG X , SMITH S , SCHWARTZ J & XIE T 2008. Differentiation-defective stem cells outcompete normal stem cells for niche occupancy in the Drosophila ovary. Cell Stem Cell, 2 , 39–49.18371420 KLAPHOLZ B , HERBERT SL , WELLMANN J , JOHNSON R , PARSONS M & BROWN NH 2015. Alternative mechanisms for talin to mediate integrin function. Curr Biol, 25 , 847–57.25754646 KLEIN AM & SIMONS BD 2011. Universal patterns of stem cell fate in cycling adult tissues. Development, 138 , 3103–11.21750026 LEATHERMAN JL & DINARDO S 2010. Germline self-renewal requires cyst stem cells and stat regulates niche adhesion in Drosophila testes. Nat Cell Biol, 12 , 806–11.20622868 LEE T & LUO L 1999. Mosaic analysis with a repressible cell marker for studies of gene function in neuronal morphogenesis. Neuron, 22 , 451–61.10197526 MA Q , DE CUEVAS M & MATUNIS EL 2016. Chinmo is sufficient to induce male fate in somatic cells of the adult Drosophila ovary. Development, 143 , 754–63.26811385 MA Q , WAWERSIK M & MATUNIS EL 2014. The Jak-STAT Target Chinmo Prevents Sex Transformation of Adult Stem Cells in the Drosophila Testis Niche. Dev Cell, 31 , 474–86.25453558 MCGUIRE SE , MAO Z & DAVIS RL 2004. Spatiotemporal gene expression targeting with the TARGET and gene-switch systems in Drosophila. Sci STKE, 2004 , pl6. NARBONNE-REVEAU K , LANET E , DILLARD C , FOPPOLO S , CHEN CH , PARRINELLO H , RIALLE S , SOKOL NS & MAURANGE C 2016. Neural stem cell-encoded temporal patterning delineates an early window of malignant susceptibility in Drosophila. Elife, 5 . OTTO SP & HASTINGS IM 1998. Mutation and selection within the individual. Genetica, 102–103 , 507–24. RHINER C , DIAZ B , PORTELA M , POYATOS JF , FERNANDEZ-RUIZ I , LOPEZ-GAY JM , GERLITZ O & MORENO E 2009. Persistent competition among stem cells and their daughters in the Drosophila ovary germline niche. Development, 136 , 995–1006.19211674 SALZMANN V , INABA M , CHENG J & YAMASHITA YM 2013. Lineage tracing quantification reveals symmetric stem cell division in Drosophila male germline stem cells. Cell Mol Bioeng, 6 , 441–448.24465278 SCHEPERS K , PIETRAS EM , REYNAUD D , FLACH J , BINNEWIES M , GARG T , WAGERS AJ , HSIAO EC & PASSEGUE E 2013. Myeloproliferative neoplasia remodels the endosteal bone marrow niche into a self-reinforcing leukemic niche. Cell Stem Cell, 13 , 285–99.23850243 SCHNEIDER M , KHALIL AA , POULTON J , CASTILLEJO-LOPEZ C , EGGER-ADAM D , WODARZ A , DENG WM & BAUMGARTNER S 2006. Perlecan and Dystroglycan act at the basal side of the Drosophila follicular epithelium to maintain epithelial organization. Development, 133 , 3805–15.16943280 SHENG XR & MATUNIS E 2011. Live imaging of the Drosophila spermatogonial stem cell niche reveals novel mechanisms regulating germline stem cell output. Development, 138 , 3367–76.21752931 SIMONS BD & CLEVERS H 2011. Strategies for homeostatic stem cell self-renewal in adult tissues. Cell, 145 , 851–62.21663791 SINGH SR , ZHENG Z , WANG H , OH SW , CHEN X & HOU SX 2010. Competitiveness for the niche and mutual dependence of the germline and somatic stem cells in the Drosophila testis are regulated by the JAK/STAT signaling. J Cell Physiol, 223 , 500–10.20143337 SNIPPERT HJ , SCHEPERS AG , VAN ES JH , SIMONS BD & CLEVERS H 2014. Biased competition between Lgr5 intestinal stem cells driven by oncogenic mutation induces clonal expansion. EMBO Rep, 15 , 62–9.24355609 SPERLING AS , GIBSON CJ & EBERT BL 2017. The genetics of myelodysplastic syndrome: from clonal haematopoiesis to secondary leukaemia. Nat Rev Cancer, 17 , 5–19.27834397 STINE RR , GREENSPAN LJ , RAMACHANDRAN KV & MATUNIS EL 2014. Coordinate regulation of stem cell competition by Slit-Robo and JAK-STAT signaling in the Drosophila testis. PLoS Genet, 10 , e1004713.25375180 TANENTZAPF G , DEVENPORT D , GODT D & BROWN NH 2007. Integrin-dependent anchoring of a stem-cell niche. Nat Cell Biol, 9 , 1413–8.17982446 VAN DOREN M , WILLIAMSON AL & LEHMANN R 1998. Regulation of zygotic gene expression in Drosophila primordial germ cells. Curr Biol, 8 , 243–6.9501989 VERMEULEN L , MORRISSEY E , VAN DER HEIJDEN M , NICHOLSON AM , SOTTORIVA A , BUCZACKI S , KEMP R , TAVARE S & WINTON DJ 2013. Defining stem cell dynamics in models of intestinal tumor initiation. Science, 342 , 995–8.24264992 WALLENFANG MR , NAYAK R & DINARDO S 2006. Dynamics of the male germline stem cell population during aging of Drosophila melanogaster. Aging Cell, 5 , 297–304.16800845 XU T & RUBIN GM 1993. Analysis of genetic mosaics in developing and adult Drosophila tissues. Development, 117 , 1223–37.8404527 YAMASHITA YM , JONES DL & FULLER MT 2003. Orientation of asymmetric stem cell division by the APC tumor suppressor and centrosome. Science, 301 , 1547–50.12970569 ZHU S , LIN S , KAO CF , AWASAKI T , CHIANG AS & LEE T 2006. Gradients of the Drosophila Chinmo BTB-zinc finger protein govern neuronal temporal identity. Cell, 127 , 409–22.17055440 ZIMMERMAN SG , PETERS NC , ALTARAS AE & BERG CA 2013. Optimized RNA ISH, RNA FISH and protein-RNA double labeling (IF/FISH) in Drosophila ovaries. Nat Protoc, 8 , 2158–79.24113787

PMC008xxxxxx/PMC8779329.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101130617 29778 Cancer Cell Cancer Cell Cancer cell 1535-6108 1878-3686 34822775 8779329 10.1016/j.ccell.2021.11.002 NIHMS1760104 Article The allergy mediator histamine confers resistance to immunotherapy in cancer patients via activation of the macrophage histamine receptor H1 Li Hongzhong 11113 Xiao Yi 113 Li Qin 112 Yao Jun 1 Yuan Xiangliang 1 Zhang Yuan 1 Yin Xuedong 1 Saito Yohei 1 Fan Huihui 2 Li Ping 1 Kuo Wen-Ling 1 Halpin Angela 3 Gibbons Don L. 4 Yagita Hideo 5 Zhao Zhongming 26 Pang Da 7 Ren Guosheng 8 Yee Cassian 9 Lee J. Jack 10 Yu Dihua 114* 1 Department of Molecular and Cellular Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA. 2 Center for Precision Health, School of Biomedical Informatics, The University of Texas Health Science Center at Houston, Houston, TX 77030, USA. 3 Clinical Informatics, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA. 4 Department of Thoracic/Head and Neck Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA. 5 Department of Immunology, Juntendo University School of Medicine, Bunkyo-ku, Tokyo, Japan. 6 Human Genetics Center, School of Public Health, The University of Texas Health Science Center at Houston, Houston, TX 77030, USA 7 Department of Breast Surgery, Harbin Medical University Cancer Hospital, Harbin, Heilongjiang, China. 8 Department of Endocrine and Breast Surgery, The First Affiliated Hospital of Chongqing Medical University, Yuzhong, Chongqing, China. 9 Department of Melanoma Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA. 10 Department of Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA. 11 Current address: Chongqing Key Laboratory of Molecular Oncology and Epigenetics, The First Affiliated Hospital of Chongqing Medical University, Chongqing, China. 12 Current address: Department of Oncology, Beijing Friendship Hospital, Capital Medical University, Beijing, China. 13 These authors contributed equally to this work. 14 Lead contact. * Correspondence to: dyu@mdanderson.org. AUTHOR CONTRIBUTIONS H.L., Y.X., and D.Y. developed the original hypothesis and designed experiments. H.L., Y.X., J.Y., X.Y., Y.Z., X.Y., Y.S., W.X., P.L., and D.Y. performed experiments and/or analyzed data. D.L.G., H.Y., D.P. and G.R. provided critical mouse strains, cell line, reagents and/or samples. H.F., Z.Z., and J.Y. analyzed scRNA-seq data. Y.X., A.H., C.Y., J.L. and D.Y. analyzed the MDACC clinical data using the Epic SlicerDicer software. H.L. and Q.L. collected and analyzed clinical data of cancer patients receiving anti-PD1 treatment in the basket trail. H.L., Y.X., and D.Y. wrote and edited the manuscript. D.Y. supervised the study. 11 12 2021 10 1 2022 24 11 2021 10 1 2023 40 1 3652.e9 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY Reinvigoration of anti-tumor immunity remains an unmet challenge. Our retrospective analyses revealed that cancer patients who took antihistamines during immunotherapy treatment had significantly improved survival. We uncovered that histamine and histamine receptor H1 (HRH1) are frequently increased in the tumor microenvironment and induce T cell dysfunction. Mechanistically, HRH1-activated macrophages polarize toward an M2-like immunosuppressive phenotype with increased expression of the immune checkpoint VISTA, rendering T cells dysfunctional. HRH1 knockout or antihistamine treatment reverted macrophages immunosuppression, revitalized T cell cytotoxic function, and restored immunotherapy response. Allergy, via the histamine-HRH1 axis, facilitated tumor growth and induced immunotherapy resistance in mice and humans. Importantly, cancer patients with low plasma histamine levels had more than tripled objective response rate to anti-PD-1 treatment compared to patients with high plasma histamine. Altogether, pre-existing allergy or high histamine levels in cancer patients can dampen immunotherapy responses and warrant prospectively exploring antihistamines as adjuvant agents for combinatorial immunotherapy. Graphical Abstract eTOC blurb Li et al. investigate how cancer cells evade immune attack and resist immunotherapies. Cancer cell-derived or allergy-released histamine binds to HRH1 on tumor-associated macrophages that suppress CD8+ T cell function, accelerate tumor growth, and confer immunotherapy resistance. H1-antihistamines counteract histamine-mediated immunosuppression, reinforce anti-tumor immunity, and significantly enhance immunotherapy response. pmcINTRODUCTION T cell-mediated anti-tumor immunity plays a central role in host’s defense against cancer. However, cancer cells can co-evolve with the tumor immune microenvironment and develop different strategies to evade T cell immune destruction. Tumor infiltrating T cells often manifest impaired effector function (i.e., dysfunction) and fail to eliminate cancer cells owing to various T cell inhibitory signals, e.g., cytotoxic lymphocyte antigen-4 (CTLA-4) and programmed cell death protein 1 (PD-1)/programmed death ligand 1 (PD-L1) (Jiang et al., 2018). Anti-CTLA-4 and anti-PD-1/PD-L1 antibodies, as immune checkpoint blockade (ICB) therapies, have yielded significant clinical benefits and durable responses in a subset of cancer patients. Yet, many cancer patients cannot benefit from these treatments, and it is highly challenging to reach immunotherapy’s full potential (Ribas and Wolchok, 2018; Sharma et al., 2017; Verma et al., 2019). To gain insights on what may impact on cancer patients’ immunotherapy response, we retrospectively analyzed patients who took different common pharmaceutical drugs during ICB treatments. Surprisingly, we found that antihistamines are associated with significantly improved clinical outcome and antihistamines may achieve this via reinforcing anti-tumor immunity, raising an interesting question: how antihistamines, which block histamine binding to histamine-receptors, influence anti-tumor immunity? Histamine, a metabolite of histidine, is best known for its release from mast cells as a response to allergic reactions or tissue damage. Histamine exerts its effects primarily by binding to G protein-coupled receptors, designated histamine receptors H1 through H4 (HRH1/2/3/4). Among them, HRH1 is the major one involved in allergic response. During allergic reactions, mast cell-released histamines activate HRH1, which triggers contraction of smooth muscles and increases capillary permeability, resulting in classic allergy symptoms. HRH1 antagonists, mostly over-the-counter (OTC) drugs, are widely used to relieve allergy symptoms and to prevent nausea and vomiting in cancer treatment (Simons, 2004; Thurmond et al., 2008). Elevated levels of histamine have been detected in cancer patients’ blood and cancerous tissues although histamine has not been suggested in cancer etiology (Haak-Frendscho et al., 2000; Moriarty et al., 1988; Sieja et al., 2005; von Mach-Szczypinski et al., 2009). In addition, cancer cells frequently up-regulate a histamine-synthesizing enzyme, L-histidine decarboxylase (HDC), leading to increased histamine in cancer patients (Haak-Frendscho et al., 2000; Massari et al., 2018). Generally, the roles of histamine and histamine receptors in cancer development are unclear. Previous studies focus on HRHs’ expressions on various cancer cells, which lead to controversial reports that HRHs either promote or inhibit cancer growth (Faustino-Rocha et al., 2017). Moreover, allergic reaction releases lots of histamine and affects tens of millions of people every year, yet the potential impacts of allergy on cancer and cancer therapies have not been investigated. RESULTS Patients receiving antihistamines have better survival with ICB therapies To assess the impact of taking other medications on therapeutic response to immunotherapy in cancer patients, we retrospectively evaluated the clinical outcomes of melanoma patients who took another medicine among forty charted common drugs (Table S1) while receiving immunotherapies (anti-PD-1/PD-L1) at The University of Texas MD Anderson Cancer Center. Our data showed that taking antibiotics (e.g., ampicillin) was associated with an increased death rate in immunotherapy-treated patients, consistent with a previous report (Elkrief et al., 2019); whereas taking aspirin was correlated with a reduced death rate in immunotherapy-treated patients, as found in mouse models (Figures 1A and S1A) (Zelenay et al., 2015). Among the forty common drugs examined, only HRH1-specific antihistamines (H1-antihistamines or second-generation antihistamines) significantly correlated with better survival of patients besides aspirin (Figure 1A). Clearly, melanoma patients who took H1-antihistamines during anti-PD-1/PD-L1 treatments had a highly significantly reduced death rate than did age-, sex-, or stage-matched patients without taking the H1-antihistamines (Figures 1B, 1C, S1B and Table S2). Among lung cancer patients receiving anti-PD-1/PD-L1 treatments, those taking H1-antihistamines also showed a statistically significant reduction of death rate than those without taking the H1-antihistamines (Figure 1B, and Table S2). Kaplan-Meier survival analysis also indicated significantly improved overall survival in melanoma and lung cancer patients who took H1-antibhistamins during anti-PD-1/PD-L1 treatment compared to control groups that didn’t (Figure 1D). Additionally, breast and colon cancer patients taking H1-antihistamines while receiving anti-PD-1/PD-L1 therapies also showed trends of reduced death rate (Figure S1C and Table S2), although no statistical significance due to relatively smaller numbers of patients taking the H1-antihistamines. These clinical data indicated that H1-antihistamines may augment immunotherapy. Notably, H1-antihistamines had minimal effect on the survival of chemotherapy-treated patients (Figures 1E and S1D), suggesting that H1-antihistamines may not target tumor cells directly. HRH1 correlates with T cell dysfunction in human cancers The above clinical findings suggested that H1-antihistamines may enhance anti-tumor immunity. H1-antihistamines specifically block histamine binding to HRH1. Therefore, we examined whether HRH1 high-expressing tumors were associated with suppressed anti-tumor immunity in cancer patients. We adapted Tumor Immune Dysfunction and Exclusion computational framework (Jiang et al., 2018) to evaluate the impacts of HRH1 on T cell infiltration and T cell dysfunction in patient samples from The Cancer Genome Atlas (TCGA). Evidently, HRH1 expression yielded high tumor immune dysfunction scores in 9 of 12 TCGA cancer types analyzed, a higher proportion than CD274 (PD-L1) and SERPINB9 (Figure 1F), two genes well-known for inducing T cell dysfunction (Jiang et al., 2018). Notably, HRH1 expression was not associated with cytotoxic T lymphocyte (CTL) infiltration (Figure S1E), suggesting that HRH1 high expression may primarily induce T cell dysfunction. In contrast, other histamine-receptors (HRH/2/3/4) had much lower effect on T cell dysfunction compared with HRH1 (Figure 1F). The strong association between HRH1 expression and T cell dysfunction prompted us to examine whether high HRH1 expression correlates with poor clinical outcome in cancer patients, especially patients with CTL infiltrated (CTL+) tumors (Figure S1F). Indeed, high HRH1 expression was significantly associated with poor survival in patients with CTL+ triple negative breast cancer (TNBC) and lung adenocarcinoma as well as a strong trend of poor survival in melanoma patients (Figures 1G and S1G-S1I). Notably, HRH1 is among the top 20 genes that are strongly associated with poor survival in CTL+ TNBC patients (hazard ratio>2, Figure S1J). Given that HRH1 is associated with T cell dysfunction in cancer patients and patients receiving H1-antihistamines along with ICB had better survival, we further explored whether high HRH1 expression is associated with immunotherapy resistance. Among melanoma patients treated with the anti-PD-1 drug pembrolizumab (GSE78220) (Hugo et al., 2016), non-responders had higher HRH1 mRNA expression in pre-treatment tumors than responders (Figure 1H, left). Anti-PD-1-treated patients with high HRH1-expressing tumors had devastatingly short overall survival compared with patients with low HRH1-expressing tumors (Figure 1H, right). Histamine and HRH1 are up-regulated in tumor microenvironment When deciphering how HRH1 induces T cell dysfunction, we surprisingly found that HRH1 expression was barely detectable in most of the tested human and mouse tumor cell lines (Figures S2A and S2B). Instead, using two deconvolution algorithms, i.e., Tumor Immune Estimation Resource (TIMER) and CIBERSORT (Li et al., 2017; Newman et al., 2015), we found that HRH1 expression was negatively correlated with tumor purity but positively correlated with tumor-associated macrophage (TAM) in the tumor microenvironment (TME) (Figures S2C and S2D), particularly in immunosuppressive M2-like macrophages among various cell types in human TME (Figure 2A). Furthermore, HRH1 was expressed mainly on M2-polarized (IL-4-treated) macrophages and TAMs in the TME, instead of naïve macrophages, M1-polarized (IFN-γ-treated) macrophages, or resident macrophages from normal mammary glands, in both humans (Figures 2B-2D and S2E top) and mice (Figures 2E-2G, S2E bottom, and S2F). In addition to HRH1 up-regulation on IL-4-induced M2-like macrophages (Orecchioni et al., 2019) (Figures 2B, 2E, and S2E), tumor-derived TGF-β also induced HRH1 expression on macrophages (Figure S2G). Additionally, we detected significantly increased levels of HRH1 ligand histamine in the blood of TNBC or colon cancer patients compared with that of healthy subjects (Figure 2H). Intriguingly, high histamine levels in TNBC patients’ blood was significantly correlated with low density of tumor-infiltrating GZMB+ cells (cytotoxic CD8+ T cell or NK cells) (Figure 2I), suggesting a potential link between histamine levels and immune cytotoxic cell infiltration. Increased histamine levels were also detected in tumor tissues (Figure S2H) and blood of tumor-bearing mice (Figure S2I) compared with corresponding normal tissues and blood from tumor-free mice, respectively, consistent with other reports (Moriarty et al., 1988; Sieja et al., 2005; von Mach-Szczypinski et al., 2009). Additionally, significantly increased histamine levels were detected in the tumor-conditioned-medium (TCM) derived from all the exanimated murine tumor cell lines and human breast cancer cell lines compared to normal culture medium and control medium from normal human breast epithelial cell line MCF-12A (Figure S2J), suggesting that cancer cells may be a major source of increased histamine detected in tumor-bearing mice and cancer patients (Figures 2H and S2H). Consistently, increased expression of HDC, the histamine-synthesizing enzyme, was also detected in patients’ breast cancer cells (Figure S2K). These data indicate that both histamine and HRH1 are up-regulated in the immunosuppressive TME. Inhibition of HRH1 on macrophages restores T cell anti-tumor immunity To investigate the specific function of the histamine-HRH1 axis in macrophages, we generated bone marrow-derived macrophages (BMDMs) from wild-type (WT) and HRH1-knockout (HRH1−/−) mice and treated them with TCM. Alternatively, we added an H1-antihistamine (fexofenadine, abbreviated as FEXO) to TCM-treated WT BMDMs. The expression ratio of major histocompatibility complex class II (MHCII, an M1 marker) versus CD206 (an M2 marker) were used to evaluate M1-M2 polarization status, which generally denotes anti-tumor versus pro-tumor activities of TAMs, although it does not fully reflect TAM’s complexity in the TME (Guerriero, 2018). Both HRH1−/− and FEXO treatment polarized macrophages toward an M1-like phenotype characterized by increased MHCII and decreased CD206 (Figure 3A). TAMs isolated from EO771 mammary tumors in HRH1−/− mice also had up-regulated M1-like pro-inflammatory molecules (Il1b, Il6, Il12b, and Nos2) and attenuated M2-like marker Arg1 compared with that in WT mice (Figure S3A). To examine the impact of macrophage HRH1 blockade on T cell activation, HRH1−/− or FEXO-treated macrophages were cultured with WT splenic T cells. HRH1−/− or FEXO treatment abrogated TAM-mediated T cell suppression, as signified by enhanced T cell proliferation, up-regulated cytotoxic and cytolytic effector molecules, including interferon gamma (IFN-γ) and perforin-1 (PRF1), and increased ovalbumin (OVA)-specific OT-I cell-mediated killing of OVA-transduced EO771 tumor cells (Figures 3B, and S3B-S3D). Notably, FEXO treatment of HRH1−/− macrophages didn’t increase MHCII:CD206 ratio (Figure 3A) or T cell activation/killing compared with vehicle-treated HRH1−/− macrophages (Figures 3B, S3C, and S3D), indicating that FEXO’s effects on macrophages are mediated by HRH1. Conversely, histamine (10 μM) treated mouse macrophages had increased M2-like marker CD206 and reduced M1-like marker MHCII compared to vehicle-treated ones (Figure S3E). Importantly, histamine-treated macrophages significantly suppressed T cell activation compared to vehicle-treated macrophages (Figures S3F and S3G). Next, we investigated the impact of histamine-HRH1 axis on TAMs and T cell immunity using two syngeneic tumor models in vivo. The EO771 mammary tumor cells or B16-GM (denotes B16-GM-CSF tumors with high tumor-infiltrating TAMs) melanoma cells (De Henau et al., 2016) were inoculated orthotopically into HRH1−/− mice and WT C57BL/6 mice. In separate experiments, WT mice were transplanted with these two cell lines and treated with vehicle or FEXO. Enhanced MHCII:CD206 ratio in the TAMs, increased numbers of IFN-γ+ and PRF1+ CD8+ T cells, and reduced tumor growth were found in HRH1−/− mice and FEXO treated mice compared to WT mice and vehicle treated mice, respectively (Figures 3C-3E, PRF1 data not shown). Furthermore, enzyme-linked immune absorbent spot (ELISPOT) assay revealed that tumor reactive T cells were increased in tumors from HRH1−/− mice compared to that from WT mice (Figure S3H). Similar changes were also detected in TNBC (4T1) or lung carcinoma tumors (LLC) in FEXO-treated mice compared with vehicle-treated mice (Figures S3I-S3K). The inhibition of B16-GM tumor growths in both HRH1−/− mice and FEXO-treated WT mice were blocked by depleting CD8+ T cells with anti-CD8 antibodies (Figures 3F and S3L), indicating that the enhanced anti-tumor activities by HRH1 blockade depend on CD8+ T cells. Although HRH1 was also expressed on endothelial cells (Lu et al., 2010), FEXO treatment did not show significant impact on angiogenesis in EO771 tumors (Figure S3M). To test whether HRH1 expressed on CD8+ T cells contributes to their biological functions, we compared HRH1−/− versus WT T cells and FEXO-treated versus vehicle-treated T cells, respectively, and found that they had similar proliferation rates and activities (Figure S3N), suggesting that increased T cell activation by blocking HRH1 in mice was unlikely resulted from direct inhibition of HRH1 on CD8+ T cells. To explore whether loss of HRH1 expression on non-immune cells in the TME may also contribute to tumor suppression in HRH1−/− mice, we generated chimeric mice by transplanting WT bone marrows into HRH1−/− mice (WT BM in HRH1−/−) or HRH1−/− bone marrows into WT mice (HRH1−/− BM in WT) following lethal irradiation. One month later, mice were orthotopically inoculated with EO771 tumor cells. HRH1−/− BM in WT mice showed decreased tumor growth along with increased MHCII:CD206 ratio in TAMs and increased IFN-γ+ CD8+ T cell infiltration compared to that of control WT mice reconstituted with WT bone marrow cells (WT BM in WT) (Figure S4A). Interestingly, WT BM in HRH1−/− mice seemed to have a minor tumor reduction although statistically insignificant (Figure S4A). To further determine the critical function of HRH1 on macrophages, we co-implanted WT or HRH1−/− macrophages with various types of cancer cells into recipient mice. Co-implantation of HRH1−/− BMDMs with B16-GM melanoma cells into WT mice significantly increased the activity of tumor-infiltrating CD8+ T cells and reduced tumor growth, which phenocopied the B16-GM tumors implanted in the HRH1−/− mice; whereas co-implanting WT BMDMs with B16-GM cells into HRH1−/− hosts enhanced tumor growth and suppressed CD8+ T cell activity, similar to the B16-GM tumors in WT mice (Figures 3G and 3H). Similarly, co-implanting HRH1−/− BMDMs with EO771 mammary tumor and LLC lung tumor cells into WT mice also significantly enhanced anti-tumor immunity and reduced tumor growths (Figure S4B, data not shown). Together, these data demonstrated that HRH1 activation on macrophages promotes CD8+ T cell activity suppression and tumor growth. To evaluate the general impact of HRH1 blockade on the tumor immune microenvironment, we profiled CD45+ immune cells isolated from EO771 tumors grown in WT and HRH1−/− mice using mass cytometry (CyTOF), which revealed 12 distinct subsets, or clusters, of cells (Figures S4C-S4E). EO771 tumors from HRH1−/− mice had significantly fewer M2-like macrophages (cluster 7), whereas cytotoxic immune cells, including CD8+ T cells (cluster 2) were increased in tumors from HRH1−/− mice (Figures 3I and S4F), suggesting enhanced anti-tumor immunity. Moreover, the MHCII:CD206 ratio of TAMs was increased, along with granzyme B (GZMB)+ CD8+ T cells, in HRH1−/− mice compared with WT mice (Figure S4G). HRH1 blockade also enhanced anti-tumor immunity in lung metastatic sites of two spontaneous lung metastasis models B16-GM and 4T1, as shown by increased M1-like polarization of resident alveolar macrophages (Misharin et al., 2013), increased cytotoxic CD8+ T cells, and reduced lung metastases (Figures S5A-S5C). B16-GM lung metastases in FEXO-treated mice were also inhibited compared to vehicle-treated mice after B16-GM primary melanomas were surgically removed upon growing to 100 mm3 of tumor size (Figure S5D), indicating that antihistamines enhanced anti-metastasis immune response. HRH1 activation promotes VISTA membrane localization To explore how HRH1 on macrophages suppresses T cell activities, EO771 TCM-treated WT or HRH1−/− macrophages were co-cultured with WT CD8+ T cells in direct contact or separately in a trans-well. Modulation of IFN-γ+PRF1+ CD8+ T cell by macrophage HRH1 was largely dependent on direct cell-cell contact (Figures 4A and S6A). Since macrophages and dendritic cells can regulate T cell function via engagement of co-stimulatory or inhibitory receptors on T cells (Guerriero, 2019; Ostuni et al., 2015), we investigated whether HRH1 on macrophages induces T cell dysfunction via regulating co-stimulatory or inhibitory receptors on T cells. Among the 13 ligands with co-stimulatory or inhibitory activities screened on EO771 or B16-GM TCM-treated macrophages, VISTA and TIM-3, known inhibitory molecules (Lines et al., 2014; Ocana-Guzman et al., 2016), were the most down-regulated molecules on HRH1−/− macrophages compared with WT macrophages (Figures 4B and S6B). Functionally, when WT macrophages in TCM were pre-treated with VISTA-blocking antibody and co-cultured with T cells, IFN-γ and PRF1+ CD8+ T cell levels and tumor cell killing activity increased to similar levels of the T cells co-cultured with HRH1−/− macrophages; a TIM-3-blocking antibody had a lesser effect (Figures 4C, S6C, and S6D), suggesting that VISTA is a major HRH1 downstream mediator of T cell dysfunction. In EO771 tumors from WT mice, HRH1 expression on TAMs was strongly correlated with VISTA expression (Figure S6E). Reduced VISTA expression was detected on TAMs from EO771 and B16-GM tumors in HRH1−/− mice or FEXO-treated WT mice compared to respective controls (Figure 4D). Decreased VISTA on TAMs was also observed in FEXO-treated other tumor models (e.g., 4T1 and LLC) in WT Balb/c or B6 mice compared to that of vehicle-treated ones (Figure S6F). Similarly, VISTA expression on alveolar macrophages from lung metastases of HRH1−/− mice bearing B16-GM tumor or of FEXO-treated WT mice bearing 4T1 tumor was also down-regulated compared with their controls (Figure S6G). Blocking HRH1 significantly reduced VISTA membrane expression on macrophages (Figures 4D and 4E) but VISTA mRNA and total protein expression didn’t change significantly (Figures 4E and S6H). Cell fractionation analysis confirmed that HRH1 blockade decreased cell membrane VISTA protein expression (Figure 4E). Calcium (Ca2+) facilitates protein trafficking to the plasma membrane and HRH1 activation induces Ca2+ release from the endoplasmic reticulum (Micaroni, 2010, 2012; Parsons and Ganellin, 2006). Thus, we explored whether HRH1 activation may foster VISTA membrane trafficking via releasing Ca2+. Indeed, after histamine or TCM treatment that induced activation of HRH1, intracellular free Ca2+ levels were higher in WT macrophages than that in HRH1−/− macrophages and FEXO-treated WT macrophages (Figure S6I). Blocking Ca2+ flux by BAPTA-AM, an intracellular calcium chelator, reduced HRH1-mediated membrane VISTA expression on WT TAMs, while the Ca2+ flux agonist ionomycin increased cell-surface VISTA expression on HRH1−/− TAMs (Figures 4F and S6I). These data suggest that HRH1-modulated Ca2+ release is critical for VISTA membrane localization. HRH1 activation reshapes the transcriptomic landscape of macrophages To gain deep insight into HRH1 downstream signaling that may contribute to the immunosuppressive phenotype of macrophages, we profiled the global transcriptome of TCM-treated WT and HRH1−/− macrophages by RNA sequencing. Compared to WT macrophages, HRH1−/− macrophages showed higher expression of genes associated with M1 polarization (e.g., CXCL10 and CD40), but lower expressions of many genes associated with M2-like phenotype (e.g., C1QB, C1QC, and MRC1) (Figure 5A). Gene set enrichment analysis (GSEA) identified key canonical pathways specifically up- or down-regulated in HRH1−/− macrophages compared to WT macrophages (Figure 5B). For example, TCM-treated HRH1−/− macrophages showed significantly upregulated TNF-α signaling, LPS and IFN-γ signaling (Figure 5B). Various pro-inflammatory cytokines and chemokines (e.g, Il6, Il1α, Cxcl10 and Cxcl11) are also significantly higher in HRH1−/− macrophages than in WT macrophages (Figure S6J). These upregulated signaling pathways and molecules in HRH1−/− macrophages are tightly associated with M1 polarization of macrophages and anti-tumor immune reactivity of macrophages (DeNardo and Ruffell, 2019), conceivably, they may contribute to the increased anti-tumor activities of HRH1−/− macrophages. On the other hand, reduced M2-polarized macrophage signature (Gerrick et al., 2018) was detected in HRH1−/− macrophages (Figure 5B), consistent with their M1-like polarization phenotype. Intriguingly, cholesterol biosynthesis and targets of sterol regulatory element binding transcription factors (SREBF1/2) are also among the inhibited signaling pathways in HRH1−/− macrophages compared to WT macrophages (Figure 5B). To further explore the broad impact of HRH1 blockade on macrophage phenotype and the landscape of tumor immune microenvironment in vivo, the CD45+ immune cells were isolated from EO771 tumors growing in WT versus HRH1−/− mice for single-cell RNA sequencing (scRNA-seq) analyses. Major immune cell types were predicted using built-in annotated human PBMC datasets as a reference and the automatic cell annotations were further calibrated by examining the most highly expressed marker genes between clusters (Figure 5C). The data showed that HRH1−/− primarily impacted on TAMs and T cells among CD45+ immune cells. Cell composition analysis showed reduced M2-like macrophages and slightly increased M1-like macrophages in tumors from HRH1−/− mice compared to that in WT mice (Figure S6K). We also calculated M1 and M2 gene signature-based scores at single-cell level in TAMs. Overall, macrophages isolated from EO771 tumors in HRH1−/− mice showed significantly higher M1-gene signature scores, but much lower M2-gene signature scores compared to macrophages from WT mice (Figure 5D). Since our above studies suggested correlation between HRH1 expression and T cell dysfunction (Figures 1F and S1F-S1H), we further evaluated exhausted CD8+ T cell gene signature score at single-cell level in CD8+ T cells isolated from the tumors. Indeed, CD8+ T cells isolated from EO771 tumors in HRH1−/− mice showed much lower exhausted CD8+ T cell gene signature scores compared to those from the WT mice (Figure 5E), indicating reduced T cell dysfunction by HRH1 blockade. To validate the above findings in cancer patients, we further analyzed correlations between HRH1 and human M1- and M2- macrophage markers at single-cell level in TAMs collected from melanoma patients (GSE115978) (Jerby-Arnon et al., 2018). We found that HRH1 strongly and positively correlated with well-known human M2-macrophage markers (Martinez et al., 2006), e.g. CD163, CD209, C1QB/C, at the single-cell level (Figures 5F and S6M). On the other hand, HRH1 negatively correlated with human M1-macrophage markers, including IRF1 and IDO1, both of which are downstream of IFN-γ signaling (Figure 5G). Taken together, blocking HRH1 reshaped the transcriptomic landscape of immune cells, among which reduced M2-like macrophage signatures and enhanced cytotoxic T cell functions mostly contribute to the alleviation of immunosuppression in TME. HRH1 inhibition enhances therapeutic responses to ICB HRH1 activation promotes VISTA membrane localization (Figure 4E) and patients with high HRH1-expressing tumors showed poor responses to anti-PD-1 immunotherapy (Figure 1G). VISTA and PD-1/PD-L1 suppress T cell activity non-redundantly, and up-regulation of VISTA has been linked with ICB resistance in cancer patients (Blando et al., 2019; Gao et al., 2017; Liu et al., 2015; Nowak et al., 2017). Therefore, we further investigated whether HRH1 high expression, via induction of VISTA membrane expression, would confer immunotherapy resistance, and HRH1 blockade could enhance response to immunotherapy. We found that among EO771 tumors in mice having heterogeneous responses to anti-PD-1 treatment, non-responding tumors had higher HRH1 and VISTA expression on TAMs than did partially responding tumors (Figure S7A). To test if inhibition of HRH1 would enhance anti-tumor activity of PD-L1 blockade, EO771 tumors in WT or PD-L1−/− mice were treated by vehicle or FEXO. FEXO-treated PD-L1−/− mice showed the most effective tumor inhibition and dramatically prolonged survival (Figure 6A) with 50% of mice remaining tumor free, whereas only 10% of vehicle-treated PD-L1−/− mice were tumor free. Similarly, growth of anti-PD-1-treated EO771 tumor in HRH1−/− mice was effectively inhibited accompanied by lower VISTA membrane expression on TAMs and more IFN-γ+ CD8+ T cells compared to that of anti-PD-1-treated WT mice (Figures 6B and S7B). Next, we examined whether inhibiting HRH1 activation with antihistamine could also enhance therapeutic efficacy of anti-CTLA-4 immune checkpoint inhibitor. FEXO or anti-CTLA-4 treatment both delayed tumor growth of CT26 murine colorectal carcinoma model in mice, and combinatorial treatment of FEXO plus anti-CTLA-4 more effectively inhibited CT26 tumor growth (Figure 6C, left). Remarkably, FEXO plus anti-CTLA-4 combinatorial treatment resulted in complete tumor remission and tumor-free survival in 40% of the mice while none of the mice in other groups survived by day 41 post-injection (Figure 6C, right). Furthermore, in ICB-resistant B16-GM melanoma model, FEXO treatment combined with ICB (anti-PD-1 plus anti-CTLA-4) achieved the highest therapeutic response and drastically inhibited both primary tumor growth and lung metastasis compared with FEXO or ICB alone (Figures 6D, 6E and S7C). Complete tumor remission was observed in 50% of the FEXO+ICB combination treatment group but in none of the other groups (Figure 6E), along with significantly down-regulated VISTA expression on TAMs and enhanced T cell function at both primary and metastatic tumor sites (Figures 6F and S7D). Next, the FEXO+ICB-treated B16-GM tumor-free mice were re-inoculated with B16-GM cells or EO771 cells. B16-GM cells, not EO771 cells, were rejected, indicating a persistent T cell memory for B16-GM tumor cells (Figure S7E). Together, HRH1 knockout or antihistamines combined with ICB greatly improved the therapeutic response of multiple tumor models, echoing our clinical findings that cancer patients who took H1-antihistamines during ICB treatments had better overall survival (Figure 1B). VISTA-blocking antibodies are currently being tested in clinical trials for anti-tumor efficacy (Nowak et al., 2017). We tested whether FEXO, a low-cost OTC drug, may have similar effects as anti-VISTA antibodies. FEXO monotherapy showed similar anti-tumor activity to that of VISTA antibodies in the B16-GM model (Figure 6D). When combined with ICB therapy, FEXO and the anti-VISTA antibody also had similar efficacy in controlling primary tumor growth (Figure 6D). Amazingly, FEXO+ICB was more effective than anti-VISTA+ICB for prolonging survival of mice because 50% of FEXO+ICB-treated mice had tumor-free survival but none of the anti-VISTA+ICB-treated mice survived (Figure 6E). FEXO+ICB was also more potent than anti-VISTA+ICB in promoting macrophage M1-like polarization and inhibiting lung metastasis (Figures 6G and S7F). Allergies induce immunotherapy resistance which is mitigated by HRH1 blockade The above studies indicate that histamine-HRH1 up-regulation in TME induces T cell dysfunction and immunotherapy resistance. Since allergic reactions release lots of histamine, we questioned whether allergy similarly impacts on anti-tumor immunity and immunotherapy response. To address the question, an OVA-induced allergic airway disease model (Nials and Uddin, 2008), in which BALB/c mice had two rounds of allergen (OVA) sensitization, was transplanted with tumor cells, followed by 1 week of airway OVA allergen exposure and treatment with FEXO, ICB, or FEXO+ICB (Figure 7A). To study allergy’s impact on tumor immunity and immunotherapy, we used two murine tumor models, EMT6 (mammary tumor) and CT26 (colon cancer), both of which were derived from BALB/c background that is susceptible to OVA-induced allergy (Kumar et al., 2008). High levels of histamine in plasma and tumor tissues were detected in mice after exposure to OVA (Figures 7B and S7G), indicating allergic reaction to OVA. Compared with the sham control group, mice with an OVA-induced allergic response had significantly accelerated EMT6 tumor growth, which was largely blocked by FEXO treatment (Figure 7C). Similarly, tumor growth of CT26 colon cancer cells in OVA-allergic BALB/c mice was significantly increased than that in sham control mice, and was largely blocked by FEXO treatment (Figure S7H). Both EMT6 or CT26 tumors in allergic mice also had increased VISTA expression on TAMs, a decreased MHCII:CD206 ratio, and reduced IFN-γ+ CD8+ T cells, all of which could be partially reversed by FEXO treatment (Figures 7D, S7I, and S7J). EMT6 and CT26 tumors are relatively sensitive to ICB treatment, as seen in sham control mice (Figures 7E and 7F) and as previously reported (Khononov et al., 2021; Mosely et al., 2017). However, both EMT6 and CT26 tumors became completely resistant to ICB therapy in allergic mice (Figures 7E and 7F). Remarkably, FEXO treatment largely restored sensitivity of EMT6 and CT26 tumors to ICB therapy in allergic mice (Figures 7E and 7F). The data indicate that allergic reaction promotes cancer immune evasion and immunotherapy resistance via the histamine-HRH1 axis, and this immune evasion could be mostly blocked by antihistamines. Next, to examine the clinical impact of allergic response on immunotherapy efficacy in cancer patients, we retrospectively analyzed survival data of melanoma and lung cancer patients who reported allergic reactions before receiving anti-PD-1/PD-L1 treatment versus those who did not (Table S3). Indeed, cancer patients who experienced allergies had significantly worse outcomes compared with those patients who had no allergy (Figure 7G, melanoma patients: 51% versus 41% deceased; lung cancer patients: 64% versus 58% deceased). Last, we directly examined whether plasma histamine levels are associated with patients’ response to immunotherapies. We measured pre-treatment histamine levels in plasma collected from a cohort of cancer patients (n=70) enrolled in a basket trial of anti-PD-1 treatment, which included lung cancer, breast cancer, and colon cancer patients. We found markedly lower levels of histamine in the blood of patients with complete response (CR) or partial response (PR) compared to that of patients with progressive disease (PD) (Figure 7H). Patients with stable disease (SD) had plasma histamine levels lower than PD patients but higher than CR/PR patients (Figure 7H). Next, we separated cancer patients into three groups based on their plasma histamine levels, patients with low levels of histamine (<0.3 ng/ml, which is the average level of histamine among healthy subjects, see Figure 2H), medium levels of histamine (0.3-0.6 ng/ml), and high levels of histamine (>0.6 ng/ml) (Tables S4-S6). Patients with low levels of plasma histamine had more than tripled overall response rate (ORR) and doubled disease control rate (DCR) (ORR: 55.6% vs 16%; DCR: 88.3% vs 44%) compared to patients with high levels of plasma histamine (Figures 7I and 7J). Notably, there were no significant differences in age, gender, and tumor stage among the three groups (Figure 7K and Tables S4-S6). These data from immunotherapy-treated cancer patients support the clinical relevance of our experimental findings from mouse tumor models with OVA-induced allergic airway disease (Figures 7E and 7F), suggesting that histamine release either from allergy response or by cancer cells attenuates response to immunotherapies, which can be mostly rescued by antihistamines. DISCUSSION In this study, we found that melanoma and lung cancer patients taking H1-antihistamines during immunotherapy treatment exhibited improved clinical outcomes with statistical significance. Similar trends were also observed in ICB-treated breast and colon cancer patients although didn’t achieve statistical significance which was likely due to the smaller patient numbers enrolled in the ICB treatment at the time compared to melanoma and lung cancer patients. These clinical data suggest that H1-antihistamines augment T cell-mediated anti-tumor immunity. There were previous controversial reports on histamine modulation of myeloid-derived suppressive cells (MDSCs). Using different mouse and tumor models, some studies modulated histamine production and suggested that histamine reduced MDSCs and suppressed tumor growth (Grauers Wiktorin et al., 2019; Yang et al., 2011), while others found that histamine from mast cells increased MDSC proliferation and survival and promoted B16 melanoma metastasis (Martin et al., 2014). It is possible that different histamine concentrations and histamine receptors were inducing distinct effect on MDSCs. Antihistamines, when combined with chemotherapies, were reported to have either inhibitory or promoting effects on certain types of cancers, but antihistamines had not been tested for combination with any other therapies, especially not immunotherapy, for cancer treatment (Fritz et al., 2020). Our studies suggest that the histamine-HRH1 axis could serve as a potential biomarker of T cell dysfunction and immunotherapy response, as well as promising therapeutic targets for enhancing immunotherapy response. The strong correlation between low levels of plasma histamine and better response to ICB treatment in cancer patients infers that patients who have high levels of histamine in plasma thus respond poorly to immunotherapies may particularly benefit from antihistamine treatment. Based on our data, we consider that the low-cost OTC H1-antihistamines can be used as an adjuvant therapy in combination with immunotherapy to more effectively treat cancer patients. HRH1 is also expressed in non-immune cells, including endothelial cells. Additionally, histamine may disintegrate endothelial barrier and induce vascular hyperpermeability (Ashina et al., 2015; Kugelmann et al., 2018). Our chimeric mice and macrophage co-implantation experiments indicated that HRH1 loss on TAMs is the major contributor of the enhanced immunity in HRH1−/− mice. However, HRH1−/− mice reconstituted with WT bone marrow (WT in HRH1−/−) also exhibited a slight tumor inhibition, suggesting that HRH1 loss in non-immune cells may also partly contribute to the tumor inhibition in HRH1−/− mice. Indeed, we found that compared to EO771 tumors in WT mice, tumors in HRH1−/− mice had reduced CD31+ blood vessel although they showed no significant difference in vascular permeability (data not shown). However, similar CD31+ blood vessel density was detected in FEXO-treated and vehicle-treated EO771 tumors. It is possible that HRH1−/− in endothelial cells induced impaired angiogenesis but FEXO treatment only temporally blocks binding of histamine to HRH1 without significant effects on endothelial cells as that in HRH1−/− mice. Some phenotypes observed in HRH1−/− mice reconstituted with WT bone marrow may also be associated with reduced angiogenesis in HRH1−/− mice. A major downstream effector of the histamine-HRH1 axis is VISTA, which has been implicated in ICB resistance in patients (Blando et al., 2019; Gao et al., 2017; Liu et al., 2015). Knockout of HRH1 gene or antihistamine treatment reduced membrane VISTA on TAMs and boosted T cell anti-tumor immunity, similar to anti-VISTA antibody. Recently, VISTA was identified as an acidic pH-selective ligand for the co-inhibitory receptor P-selectin glycoprotein ligand-1 (PSGL-1) on T cells, thus suppresses T cell function (Johnston et al., 2019). It was suggested that acidic pH, which is frequently found in TME, is required for VISTA engaging with PSGL-1 and suppressing T cell immunity. Interestingly, we found that tumors growing in WT mice was more acidic compared to that in HRH1−/− mice (data not shown), which may favor binding of VISTA with PSGL-1 on T cells. Remarkably, when combined with ICB, antihistamines elicited a strong anti-tumor response superior to that of anti-VISTA antibody combined with ICB, suggesting that antihistamines also regulate other downstream effectors of immune stimulation/suppression, in addition to VISTA. A most interesting finding from our studies is the potential impact of allergic reaction and histamine on anti-tumor immunity and immunotherapy response. Currently, studies regarding the relationship between allergy and cancer are controversial from epidemiological findings (Rittmeyer and Lorentz, 2012; Turner et al., 2006). Some studies suggested that allergies may reduce the risk of cancer by either increased immune surveillance after the immune hyperresponsiveness that may exert a protective effect against the development of cancer, or the physical effects of allergy symptoms that may inhibit cancer via removing potential carcinogens. In contrast, others suggested that the Th2 response and inflammation induced by allergy may facilitate development of cancer. The relationship between allergy and cancer was unclear since the potential impact of allergy on cancer hasn’t been experimentally investigated so far. Here, our experimental data from both mammary tumor and colon cancer models in mice clearly demonstrated that allergy fueled tumor growth and triggered resistance to immunotherapy through histamine-HRH1-mediated suppression of anti-tumor immunity, underscoring the previously unrecognized tumor-prone activity of allergy. Finally, our clinical data from ICB-treated cancer patients indicates that pre-existing allergy with high plasma histamine impairs cancer patients’ anti-tumor immune response and leads to their poor responses to immunotherapy. Our clinical studies have a limitation that the numbers of patients with pre-existing allergies who received antihistamine treatment before ICB therapy were not recorded. Nevertheless, our finding that plasma histamine levels of, and uptake of antihistamines by, cancer patients are associated with their response to immunotherapies strongly supports using antihistamines to treat cancer patients who have allergy with high levels of plasma histamine. OTC H1-antihistamines can restore T cell function suppressed by cancer cell-secreted and/or allergy-released histamine and improve the efficacy of immunotherapies such as ICB therapies. Our findings necessitate further clinical studies to prospectively test the effect of H1-antihistamines as adjuvant therapies for enhancing immunotherapy responses in cancer patients. STAR METHODS RESOURCE AVAILABILITY Lead Contact Further information and request for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Dihua Yu (dyu@mdanderson.org). Materials Availability The plasmids, antibodies, stable cell lines and mouse strains generated in this study have not been deposited to any repositories yet, however, these materials would be available upon request. Data and Code Availability RNA-seq data generated during this study have been deposited in the Gene Expression Omnibus (GEO) database under accession numbers GSE161484. Published datasets used in this study are available through GEO: GSE78220 (Hugo et al., 2016), GSE115978 (Jerby-Arnon et al., 2018) or cBioPortal database (http://www.cbioportal.org). Single-cell sequencing data has been submitted to Sequence Read Archive (SRA) and is available in SRA Run Selector with accession PRJNA756466. EXPERIMENTAL MODEL AND SUBJECT DETAILS Mice C57BL/6, BALB/c, and C57BL/6-background HRH1-knockout mice were purchased from The Jackson Laboratory. C57BL/6-background PD-L1-knockout mice were obtained and maintained as previously described (Chen et al., 2014). All mouse protocols and experiments were performed in accordance with National Institutes of Health guidelines and were approved by the MD Anderson Institutional Animal Care and Use Committee. All mice used in our experiments were between 6-8 weeks of age, and were housed under standard housing conditions at the MDACC animal facilities. Both male and female C57BL/6 or BALB/c mice were used for lung cancer model (LLC model), melanoma model (B16-GM model) and colon cancer model (CT26), and female BALB/c or C57BL/6 mice were used for breast cancer models (4T1, EMT6 and EO771 models). Animal numbers of each group were calculated by power analysis and animals are grouped randomly for each experiment. Cell lines Human mammary tumor cell lines including HCC1806, HS578T, BT-20, MDA-MB-435, MDA-MB- 436, MDA-MB-231 and BT-549, murine mammary tumor cell lines including 4T1, EMT6 and EO771, murine lung carcinoma cell line LLC1 (or LLC), colon cancer cell line CT26, L929 cell line and 293T cell line were obtained from American Type Culture Collection (ATCC) and cultured in endotoxin-free DMEM/F-12 medium supplemented with 10% fetal bovine serum (FBS) (HyClone). Murine melanoma cell line B16-BL6 obtained from ATCC and B16-GM described previously (De Henau et al., 2016) were cultured in endotoxin-free RPMI1640 medium supplemented with 10% FBS (HyClone). All the cells are not among commonly misidentified cell lines, and were tested for mycoplasma contamination annually using a Mycoplasma Detection Kit (Biotool #B3903). In order to prevent potential contamination, all the media were supplemented with Penicillin-Streptomycin (15-140-122, Thermo Fisher Scientific) according to the manufacturer’s instructions. Human Samples The untreated breast cancer samples including tumor tissues and blood plasma from TNBC patients who had undergone a mastectomy before therapy, normal breast tissues from women undergoing cosmetic breast surgery, and blood plasma from colon cancer patients and healthy subjects were collected at the First Affiliated Hospital of Chongqing Medical University. The blood plasma from patients with advanced lung (n=48), colon (n=12) and breast (n=10) cancers were collected pre-anti-PD1 treatment (patients were treated with Camrelizumab between Dec 24, 2019 and Feb 27, 2021) at the Beijing Friendship Hospital of Capital Medical. Best percentage change in the sum of the diameters for the selected target lesion is defined by Response Evaluation Criteria in Solid Tumors (RECIST version 1.1) on minimum 2 computed tomographic scans before treatment and 1 computed tomographic scan during treatment. The use of pathological specimens, as well as the review of all pertinent patient records, were approved by the Research Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (project approval number 1005367 2017-012) and the Research Ethics Committee of Beijing Friendship Hospital of Capital Medical University (project approval number 2017-P2-141-01), and written informed consent were obtained from each participant and/or their legal representative, as appropriate. The information about the samples used is summarized in Table S7. METHOD DETAILS Survival analysis using The Cancer Genome Atlas (TCGA) data TCGA breast cancer gene expression data and clinical data were downloaded from TCGA data portal (https://tcga-data.nci.nih.gov/tcga/dataAccessMatrix.htm). Triple-negative breast cancer (TNBC) cases were collected based on ESR1, PGR, and ERBB2 expression (RNAseq V2 RSEM) by using K-means 2 separation to define positive and negative groups and then combining triple-negative samples. CD8+ cytotoxic T lymphocyte (CTL) groups in TNBC samples were based on K-means 3 separation using a 15-gene signature (CD3E, CD8A, CCL2, CCL3, CCL4, CXCL9, CXCL10, GZMA, GZMK, HLA-DOA, HLA-DOB, HLA-DMB, ICOS, IRF1 and PRF1; Figure S1A). Survival analysis was performed for 77 CTL-high TNBC cases with survival data using the coxph function of R software using a resampling procedure (Yao et al., 2012). Candidate prognostic genes were collected from 100-round runs with a coxph P value of <0.02 in >75 rounds. Volcano plots were drawn using hazard ratios and P values from the coxph analysis on 16,975 genes after removal of low expressers with candidate genes associated with poor prognosis and favorable prognosis marked in both CTL-high and CTL-low groups. To test the association between HRH1 gene expression level and patient survival, Kaplan-Meier survival analysis was performed using the similar program described in the Human Protein Atlas (www.proteinatlas.org/humanpathology) (Uhlen et al., 2017). Briefly, survival analysis was performed using R package survival and survmine. Patients were stratified into gene expression high and low groups using Kmeans two separation of gene expression values (consensus from 20 iterations), which is an unbiased separation based on gene expression dispersion. Epic SlicerDicer software analysis Our retrospective study was conducted on melanoma patients encountered during 2016 to 2017 at The University of Texas MD Anderson Cancer Center, which draws a diverse range of local, regional, national, and international patients. The study population consisted of all melanoma patients with of International Classification of Diseases (ICD) code C43 (10th version). Anonymized aggregate-level data were collected using the SlicerDicer function within MD Anderson Epic electronic medical records. Institutional Review Board approval was therefore waived. Using the Epic SlicerDicer, we identified 9922 patients with a visit diagnosis, billing diagnosis, or active problem list with malignant skin of melanoma (ICD-10-CM:C43.*). EPIC SlicerDicer was used to further identify 878 of these patients who received anti-PD-1 antibodies (pembrolizumab, nivolumab, or cemiplimab) or anti-PD-L1 antibodies (durvalumab, atezolizumab, or avelumab) treatment during the study period. The comparison group was 1986 patients with the same diagnosis who received chemotherapies but no immunotherapeutic treatment. To analyze the impacts of 40 different common pharmaceutical drugs (Supplementary Table 1) on patients’ outcomes, the 878 patients receiving anti-PD-1/PD-L1 treatment were further divided into 40 patient subgroups based on the medications they took during immunotherapeutic treatment. The estimated patient deaths of each subgroup were calculated based on the patient number of each subgroup and the average patient death rate (39%, based on deceased patients out of total patients analyzed). The real patient deaths of each subgroup were compared to the estimated patient deaths in order to determine the overall impact of the medication on clinical outcome (Figure 1A). For further analysis of the H1-antihistamines, the 878 patients (received immunotherapy treatment) and 1986 patient (receiving chemotherapy treatment) were further subdivided into patients subgroups with or without uptake of H1-antihistamines that selectively target HRH1 (including fexofenadine, loratadine, desloratadine, cetirizine, levocetirizine, and azelastine) at the same time with anti-PD-1/PD-L1 antibody treatment or chemotherapies. Patient information, including age (<30 years, 30 to <50 years, 50-70 years, >70 years), sex (male or female), disease stage (0, I, II, III, IV, other, or unknown), survival (alive or dead), was extracted. Because many patients’ direct responses to therapy were not available in Epic SlicerDicer, the overall survival (up to June 15, 2021) was used as a surrogate indicator of patients’ therapeutic response. The Fisher exact test was used to identify any nonrandom association between antihistamine uptake and patients’ overall survival status. A similar analysis was performed in breast cancer patients (ICD code: C50) encountered at MD Anderson during 2016-2018, lung cancer patients (ICD code: C34) encountered at MD Anderson during 2016-2018, and colon cancer patients (ICD code: C18) encountered at MD Anderson during 2016-2018. To perform patient overall survival analysis, patient death date and the date receiving anti-PD-1 or PD-L1 treatment were pulled out from Epic. The survival time was calculated as the days between the two dates. To investigate the potential impact of melanoma patients’ allergy status on their response to ICB therapies, melanoma patients who received anti-PD-1/PD-L1 immunotherapy were divided into two groups based on their allergy status. Patients were considered to have had an allergic response if they had a specified diagnosis of allergy status to other drugs, medicaments, or biological substances (ICD code: Z88) documented as a visit diagnosis, billing diagnosis, or active problem list. To avoid conflating these allergic responses induced by anti-PD-1/PD-L1 antibodies, which may interfere with the treatment efficacy, we reserved the allergic response category for patients who reported an allergic response within 10 days before receiving anti-PD-1/PD-L1 treatment. RNA sequencing and data analysis WT and HRH1−/− BMDM were cultured with TCM or DMEM medium for two days and then total RNA was purified using Trizol (Invitrogen). After removing DNA by DNase, RNA was further purified using RNeasy MinElute Cleanup Kit (Qiagen). RNA samples were sent to UT health sequencing core for library construction and sequencing. After Raw fastq files from RNA sequencing were mapped to mm10 genome using STAR2, read counts for genes were prepared by htseq from which TPM were calculated. Principle component (PCA) analysis was done using R software. GSEA analysis was done using Java Web Start downloaded from https://www.gsea-msigdb.org/gsea/downloads.jsp. Differential expressed genes between the wild-type and HRH1-KO samples were obtained by using R package DESeq2 with filtering parameters of fold change above 3, adjusted P<0.01, and average log2(TPM) in the high expression group above 0. Volcano plots were drawn using log2 (fold change) and −log10 (adjusted p), ceiled at 5 and 100 respectively. We obtained a list of M1 and M2 macrophage signatures genes from previous publication (Gerrick et al., 2018) and generated two customized terms (M1_UP_SHORT and M2_UP_SHORT) using only genes with fold changes above 5. These terms were appended to the gmt file in mySigDB and used for GSEA analysis. For graphing top GSEA terms, we calculated a p value score using the average of −log10 (NOM p-val) and −log10 (FDR q-val) listed in the GSEA report where p values less than 1e-9 were set to 1e-9. Single-cell transcriptomic profiling 0.2x106 EO771 tumor cells were transplanted into mammary fat pad of wide type or HRH1−/− mice. Three weeks later, tumors were harvested and 0.1 gram of tumor tissue of each tumor were collected. Three tumor tissues from the same group were randomly combined into one mixed sample and proceeded to digestion using 2mg/ml collagenase A. Single-cell suspension was generated following the method described below in “Isolation of tumor-infiltrating cells”. 106 live CD45+ immune cells were sorted by flowcytometry and submitted to UTHealth Cancer Genomics Core (CGC) for sequencing. Cells were labeled with multiplexing oligoes (#1000261, 10x Genomics, Pleasanton, CA) by following the cell multiplexing oligo labeling protocol (CG000391). The labeled samples were then pooled together. The single cell capture and library construction were performed by following the 10x Genomics Chromium Next GEM Single Cell 3′ Reagent Kits v3.1 protocol (CG000388). Briefly, the pooled samples were loaded onto Chromium Next GEM Chip G (PN-1000120, 10xGenomics, Pleasanton, CA) with partitioning oil and barcoded singe cell gel beads. The barcoded and full-length cDNA is produced after incubation of the gel beads-in-emulsion (GEMs) and amplified via PCR for library construction. The library preparation was performed by following the protocol of Chromium Single Cell 3′ GEM, Library & Gel Bead Kit v3 (PN-1000121, 10xGenomics, Pleasanton, CA). The quality of the final libraries was examined using Agilent High Sensitive DNA Kit (#5067-4626) by Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, USA), and the library concentrations were determined by qPCR using Collibri Library Quantification kit (#A38524500, Thermo Fisher Scientific) on a QuantStudio3 (ThermoFisher Scientific). The libraries were pooled evenly and underwent for the paired-end sequencing on an Illumina NextSeq 550 System (Illumina, Inc.) using High Output Kit v2.5 (#20024907, Illumina, Inc.). Single-cell gene expression processing and analysis CellRanger (10x Genomics; v6.0.1) subcommand multi was used to process the raw sequencing reads and generate count matrix per sample (Zheng et al., 2017). Reads were aligned to the mouse genome assembly (version mm10) which is pre-built by 10x Genomics. Raw count matrices were then merged and analyzed using the Seurat R package (v4.0.1) (Hao et al., 2021). Cell matrices were initially filtered by removing cells with barcodes lower than 20 UMIs, with lower than 200 expressed genes, or with more than 10% of reads mapping to the mouse mitochondrial genes. To avoid low-quality cells, empty droplets or multiplets, we further filtered cells based on the number of unique genes detected in each cell, which is capped in the range from 2.5th to 97.5th percentile. Counts for the remaining cells were normalized against library size and regressed for the unwanted cycling bias among proliferating cells, using S and G2M phase scores calculated by the CellCycleScoring function in Seurat package. Scaled and centered read counts were used as gene expression for further analysis. Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP) was applied to visualize inferred cell clusters (cite: arXiv:1802.03426v3) based on the top 30 principal components. Automatic immune cell annotation was performed using scPred package (v 1.9.0)(Alquicira-Hernandez et al., 2019). The built-in annotated human PBMC datasets were combined and used as a reference to predict the major immune cell types in our in-house mouse single-cell dataset. Manual inspections were carried out to calibrate the automatic cell annotations by examining the most highly expressed marker genes between clusters, as well as literature- and database-derived cell markers. Murine M1- and M2-like cell markers were extracted from prior publication (Jablonski et al., 2015), while exhausted CD8+ T cell markers were downloaded from CellMarker database (Zhang et al., 2019). To examine the phenotype shift between mouse wildtype and knockout groups, we calculated the M1-like and M2-like macrophages and the exhausted CD8+ T cell expression-based scores at single-cell level. Specifically, we overlapped cell type-specific markers with the top 50 highly expressed genes between clusters, and then took the average expression value as the score for each set of cell type-specific markers. All the statistical analyses were conducted using R software (v4.0.4). To examine correlation of HRH1 gene expression with M1/M2 macrophage markers (Martinez et al., 2006) in human cancer, we obtained scRNA-seq data from GSE115978 (Jerby-Arnon et al., 2018) which included macrophages and other immune cells isolated from melanoma patients. To deal with missing data points in scRNA-seq, we focused on macrophage cells with normalized TPM above 0.5 for HRH1 and other macrophage marker genes of interest. Pearson correlation and scatter plots were done using log2 (1+TPM/10) values as describe in previous publication (Jerby-Arnon et al., 2018). Generation of stable cells using lentiviral infection Mouse TGF-β1-targeting shRNAs (shTGFB1-1: TRCN0000065993; shTGFB1-2: TRCN0000065994) were purchased from Sigma-Aldrich. For lentiviral production, the lentiviral expression vector was co-transfected with the third-generation lentivirus packing vectors into 293T cells using LipoD293 DNA in vitro Transfection Reagent (SignaGen Laboratories). Then, 48-72 hours after transfection, cancer cell lines were stably infected with viral particles. Generation of naïve bone marrow-derived macrophages (BMDMs) Bone marrow cells were collected from femurs obtained from 8- to 10-week-old C57BL/6 or BALB/c mice. After red blood cell lysis, bone marrow cells were seeded at a density of 5×106 cells/150×15 mm Petri dish and cultured at 37°C in complete Dulbecco modified Eagle medium (DMEM) containing 20% L929 cell-conditioned medium, providing macrophage colony-stimulating factor. Macrophages were ready for use on day 7 following a fresh medium change on day 4. Isolation of murine peritoneal macrophages 2 ml of 3% Brewer thioglycollate medium per mouse was injected into the peritoneal cavity to trigger an inflammatory response. Allow inflammatory response to proceed for 4 days, and then collect peritoneal exudate macrophages. After red blood cell lysis, peritoneal macrophages were cultured at 37°C in complete DMEM medium, and ready for use. In vitro co-culture with T cells Spleens from wild-type C57BL/6 or BALB/c mice were harvested and filtered through a 40-μm cell strainer to generate a single-cell suspension. After red blood cell lysis, splenocytes were counted and plated in complete Roswell Park Memorial Institute 1640 medium supplemented with 50 μM β-mercaptoethanol and 10 mM HEPES onto 12-well plates coated with 2.5 μg/ml anti-CD3 (clone 145-2C11, BioLegend) and 3 μg/ml anti-CD28 (clone 37N, BioLegend) antibodies. Spleen T cells were activated for 48 hours before co-culture with macrophages. To educate the macrophages, we co-cultured naïve BMDMs with tumor cells at a ratio of 1:1 or with tumor cell-derived conditioned medium for 48 hours. Macrophages were seeded with activated T cells at a ratio of 5:1. After co-culture for another 24 or 48 hours, T cells were collected for flow cytometry analysis. In vitro T cell killing assays Ovalbumin (OVA)-expressing EO771 cells alone or co-cultured with macrophages at a ratio of 1:2 were plated into 12-well plates. The following day, OT-I T cells pre-activated by OVA257-264 were added to the plates at the indicated ratio. After co-culture for another 48 hours, the tumor cells were harvested and counted to determine T cell killing ability. Macrophage polarization and stimulation To polarize macrophages toward an M1-like phenotype, we stimulated PBMC-derived macrophages or BMDMs with IFN-γ (20 ng/ml, PeproTech) for 48 hours. To induce an M2-like phenotype, we treated THP-1-derived macrophages or BMDMs with IL-4 (20 ng/ml, PeproTech) for 48 hours. To stimulate macrophages with histamine, we cultured peritoneal macrophages with histamine (10 μM) for 48 hours. To stimulate macrophages with tumor cell-derived conditioned medium (TCM), we cultured BMDMs with complete medium containing 50% TCM (volume). Tumor inductions and treatment experiments For 4T1 and EO771 models, 2×105 tumor cells were orthotopically injected into female BALB/c or C57BL/6 mice, respectively. For LLC and B16-GM models, 2×105 tumor cells were subcutaneously injected into the back of C57BL/6 mice. For the CT26 model, 2×105 CT26 cells were subcutaneously injected into the back of BALB/c mice. The induction of allergic airway disease has been described previously. Briefly, BALB/c mice were sensitized using OVA (Sigma-Aldrich) at a dose of 0.01 mg/mouse in 0.2 ml incomplete Freund's adjuvant intraperitoneally on days 0 and 12. Control mice received the same volume of phosphate-buffered saline in incomplete Freund's adjuvant. All groups of mice were challenged daily with 5% OVA (aerosolized for 20 minutes) via the airways between days 19 and 24. EMT6 and CT26 models were constructed by orthotopic injection of 2×105 tumor cells on day 18 after the first OVA sensitization. Treatments were given as single agents or in combination, with the following regimen for each drug. HRH1 antagonist fexofenadine hydrochloride was administered by oral gavage once per day at 30 mg/kg. Treatments were initiated on day 5 after tumor inoculation for the entire duration of the experiment. Anti-PD-1 antibody (clone RMP1-14; the hybridoma RMP1-14 for αPD-1 production was provided by Dr. Hideo Yagita, 10 mg/kg) and anti-CTLA-4 antibody (clone 9D9, Bio X Cell, 5 mg/kg) were injected intraperitoneally on days 7, 10, 13, 16, and 19 after tumor inoculation. Anti-VISTA antibody (clone 13F3, Bio X Cell, 300 μg/mouse) was injected subcutaneously every day for 10 days starting on day 7, followed by continuous injection every 2 days for the rest of the duration of the experiment (Le Mercier et al., 2014). For in vivo CD8+ T cell depletion, mice were treated with 200 μg of anti-CD8 antibody every 3-4 days starting at 3 days before B16-GM tumor inoculation. For in vivo macrophage adoptive transfer experiments, B16-GM tumor cells mixed with primary wild-type or HRH1−/− BMDMs at a ratio of 1:5 were injected subcutaneously into HRH1−/− or wild-type host mice, respectively. For the re-challenge study, mice with complete responses were re-challenged with 2×105 B16-GM or EO771 tumor cells (on day 150 after tumor implant). Tumor size was measured by calipers every second or third day when tumors were palpable, and the volume was calculated using the formula V = as (width×width×length)/2. Bone marrow transplantation Bone marrow transplantation followed previous publication (Khononov et al., 2021). In brief, on day 1, recipient mice (8 weeks old) received 1,000 rads total body irradiation (137Cesium Gammacell source), and 4 hours later, they were transplanted with 5×106 bone marrow cells collected from femurs of the donor mice (5 weeks old) via tail vein injection. In a successful graft, the immunological reconstitution is expected complete within 4-5 weeks. 30 days after bone marrow transplantation, the recipient mice were ready for following experiments. Isolation of tumor-infiltrating cells Mouse tumor samples were chopped with scissors and then subjected to enzymatic digestion with 2 mg/ml collagenase A (Roche) in DMEM for 1 hour at 37°C. Next, tissues were filtered through 70-μm filters (BD Biosciences) to achieve single-cell suspensions. After treatment with red blood cell lysis buffer for 5 minutes at room temperature, all samples were washed and re-suspended in flow cytometry buffer (phosphate-buffered saline/0.5% albumin/2 mM EDTA) or DMEM depending on further use. Flow cytometry staining and analysis Live single cells were sub-gated by staining with Fixable Viability Dye eFluor 450 (eBioscience) for 15 minutes at 4°C. For blocking of Fc receptors, cells were then pre-incubated with purified anti-CD16/32 antibody (clone 93, BioLegend) for 10 minutes on ice before immunostaining. After one wash with flow cytometry buffer, cells were incubated with appropriate dilutions of various combinations of the following antibodies. Primary antibodies to cell surface markers directed against CD45 (30-F11), CD3 (145-2C11), CD8a (53-6.7), CD11b (M1/70), Gr-1 (RB6-8C5), F4/80 (BM8), CD206 (C068C2), I-A/I-E (M5/114.15.2), VISTA (MH5A), Tim-3 (B8.2C12), CD11c (N418), CD24 (M1/69), GITRL (YGL 386), ICOSL (HK5.3), 4-1BBL (TKS-1), CD276 (RTAA15), OX40L (RM134L), Galectin-9 (RG9-35), PD-L1 (10F.9G2), and PD-L2 (TY25) were from BioLegend; against HRH1 (480054), from R&D Systems; against HRH1 (AHR-001), from Alomone Labs; and against Siglec-F (E50-2440), from BD Biosciences. For intracellular staining, cells were fixed, permeabilized using Foxp3/Transcription Factor Staining Buffer Set (eBioscience), and then stained with fluorochrome-conjugated antibodies to Ki-67 (16A8) and PRF1 (S16009A) from BioLegend. For cytokine staining, cells were first stimulated with Cell Stimulation Cocktail (eBioscience) at 37°C for 4 hours, and then stained with anti-IFN-γ (XMG1.2) from BioLegend. The stained cells were acquired by a BD FACSCanto II Flow Cytometer using BD FACSDiva software (BD Biosciences), and data generated were processed using FlowJo software. Mass cytometry and data analysis Mouse tumor tissues were digested as described above. Then, for CyTOF analysis, cells were incubated with 25 μM cisplatin for 1 minute (viability staining) and subsequently stained with a metal-labeled monoclonal antibody cocktail against cell surface molecules. After treatment with the Fixation/Permeabilization Buffer (eBioscience), cells were further incubated with monoclonal antibody cocktails against intracellular proteins. Antibodies used in the mass cytometry analysis were purchased from Fluidigm. The samples were analyzed using the CyTOF 2 instrument (Fluidigm) in the Flow Cytometry and Cellular Imaging Core Facility at MD Anderson. All CyTOF files were normalized and manually gated in Cytobank software. Data were transformed using the cytofAsinh function before they were applied to the downstream analysis. Phenograph clustering analysis in the R cytofkit package was performed on pooled samples to automatically identify underlying immune subsets. Heat-maps were generated on the basis of the mean value for each marker in clusters. Cell frequency in each cluster was calculated as the assigned cell events divided by the total CD45+ cell events in the same sample. Purification of myeloid cells or macrophages from tumors Single-cell suspensions of mouse tumors were generated as described in the previous section. Single cells were stained with CD11b microbeads (Miltenyi Biotec) according to the manufacturer’s instructions to enrich the myeloid fractions. Cells were then stained with Fixable Viability Dye eFluor 450 (eBioscience) to exclude dead cells, and anti-Gr-1- phycoerythrin (PE) (clone RB6-8C5) and anti-F4/80- fluorescein isothiocyanate (FITC) (clone BM8) for flow sorting on a FACSAria II Cell Sorter (BD Biosciences). qRT-PCR cDNA was prepared using 1 μg of RNA with the iScript cDNA Synthesis Kit (Bio-Rad). SYBR green-based qRT-PCR was performed using mouse primers to Il1b, Il6, Il10, Il12b, Nos2, Arg1, Tgfb1, Vista, Hrh1, and 18s (Integrated DNA Technologies). mRNA levels were normalized to 18s (ΔCt=Ctgene of interest−Ct18s) and presented as relative mRNA expression (ΔΔCt=2−(ΔCtsample-ΔCtcontrol)) or fold change. Western blotting Western blotting was done as previously described (Zhang et al., 2015). The following primary antibodies were used: HRH1 (LS-C331459, LifeSpan BioSciences), VISTA (54979, Cell Signaling Technology) and CD11b (NB110-40766, Novus Biologicals). Enzyme-linked immunosorbent assay (ELISA) The levels of histamine in cell culture supernatant, serum/plasma, and tissues were detected by Histamine ELISA kits (ENZ-KIT140-0001, Enzo; ab213975, Abcam) according to the manufacturer’s instructions. IFN-γ ELISPOT assay The IFN-γ ELISPOT assay was done following the manufacturer’s protocol. Briefly, the CD45+ leukocytes were sorted from tumors by flow cytometry. The leukocytes were counted and seeded at 5x105 cells/well into pre-coated PVDF plates (ImmunoSpot® Kits, Cellular Technology Limited), stimulated with anti-CD3 antibody and IL-2 overnight, and secreted IFN-γ was quantified following standard protocol. Assay plates were scanned and analyzed using an automated ELISPOT reader system. Immunohistochemistry (IHC) and immunofluorescence (IF) staining Standard IHC and IF staining was performed as described previously (Zhang et al., 2020). The primary antibodies used for IHC staining include anti-GZMB (ab255598, Abcam), anti-HDC (ab37291, Abcam), and anti-CD31 (77699, Cell Signaling); used for IF staining include anti-CD68 (ab955, Abcam) and anti-HRH1 (ab75236, Abcam). DyLight 488- or DyLight 594-conjugated secondary antibodies against rabbit or mouse IgG were purchased from Thermo Fisher Scientific. Intracellular calcium measurement Intracellular Ca2+ was determined using the Fluo-Forte calcium assay kit (Enzo Life Sciences) according to the manufacturer’s instructions. Fluorescence was measured using a BD FACSCanto II Flow Cytometer. Quantification and Statistical analysis Prism 8.0 software (GraphPad) was used for statistical analysis. Analysis for significance was performed by one-way or two-way ANOVA when more than two groups were compared and by parametric or nonparametric Student t-test when only two groups were compared. Fisher exact test was used when percentages of cancer patients from different groups were compared. Chi-square test was performed to determine whether there was any significant difference of gender, age, and tumor stages between the patient groups that took antihistamines and that didn’t take antihistamines. P<0.05 was considered statistically significant (*P<0.05, **P<0.01, ***P<0.001, ****P<0.0001). Survival was evaluated using the Kaplan-Meier method and analyzed by the Mantel-Cox log-rank test. All experiments were performed at least twice, and n refers to biological replicates. Supplementary Material Supplementary Figures and Tables Table S7 Human samples used in the study, related to STAR Methods. ACKNOWLEDGMENTS We thank Dr. Jedd D. Wolchok for providing B16-GM cells; Dr. Mien-Chie Hung for providing EMT6 cells; Drs. Anil K. Sood, Chunru Lin, Lin Zhang, and Xiang H.-F. Zhang for reading the manuscript; Dr. Akosua A. Badu-Nkansah and Ms. Sarah Bronson for manuscript editing; and members of the Yu laboratory for helpful discussions. Dr. Weiya Xia helped to read and scored some IHC slides. Funding: This work was supported by the National Institutes of Health (NIH) grants R01CA112567 (D.Y.), R01CA184836 (D.Y.), R01CA208213 (D.Y.), the METAVivor grants 56675 and 58284 (D.Y.), and NIH Cancer Center Support Grant P30CA016672 to MD Anderson Cancer Center (Functional Genomics Core, Flow Cytometry and Cellular Imaging facility, Research Histology Core Laboratory, and Research Animal Support Facility - Houston). The sequencing data was generated by the UTHealth Cancer Genomics Core supported by the Cancer Prevention & Research Institute of Texas (CPRIT) grant RP180734. D.Y. is the Hubert L. & Olive Stringer Distinguished Chair in Basic Science at MDACC. Figure 1. Uptake of antihistamines correlated with better survival in ICB-treated patients and HRH1 expression is associated with T cell dysfunction (A) Scatter plot of the real numbers of deceased melanoma patients who took various commonly-used medicines (40 different drugs as listed in Table S1) during ICB treatment versus their estimated deaths (at 39% death rate based on 336 deceased patients out of total 865 ICB-treated patients at MDACC). Each dot represents a group of patients who took one type of medicine along with ICB. (B) Percentages of deceased cancer patients taking H1-antihistamines during anti-PD-1/PD-L1 treatment versus those did not (Fisher exact test). (C) Percentages of deceased melanoma patients who took H1-antihistamines during anti-PD-1/PD-L1 treatment or chemotherapy compared with sex- or stage-matched melanoma patients who did not. (D) Kaplan-Meier overall survival analysis of cancer patients taking H1-antihistamines during anti-PD-1/PD-L1 treatment versus those who did not. (E) Percentages of deceased cancer patients taking H1-antihistamines during chemotherapy treatment versus those did not (Fisher exact test). (F) T cell dysfunction scores of HRH1-4 in indicated cancer types assessed by TIDE. T cell dysfunction score is defined as the z score of d/standard error (s.e.) following previous publication (Jiang et al., 2018). (G) Kaplan-Meier overall survival analysis for CTL+ TNBC patients based on HRH1 level detected in tumors. The numbers at risk are the stratified HRH1 level high and low patient numbers of those who remained alive and uncensored after a certain time period. (H) HRH1 mRNA expression in pre-treatment tumors of responder (n=15) versus non-responder (n=13) melanoma patients (t-test), and comparison of overall survival of melanoma patients who had high HRH1 vs. low HRH1 expressed in the tumors before anti-PD-1 treatment (GSE78220). Mean±SEM, *P<0.05. See also Figure S1, Tables S1 and S2. Figure 2. Activated histamine-HRH1 axis in tumor microenvironment (A) Relative HRH1 mRNA levels in immune cell subsets assessed by CIBERSORT. (B) Flow cytometry analysis of HRH1 expression on human peripheral blood monocytes (PBMC) - derived macrophages (n=3, one-way ANOVA). (C) Representative images and quantification of HRH1+ macrophages in human breast tissues (n=9) and TNBC tumors (n=32, t-test). Blue, DAPI; red, CD68; green, HRH1. Scale bar, 25 μm. (D) Percentage of CD68− or CD68+ cells (macrophages) in total HRH1+ cells in human TNBC tissues (n=32, t-test). (E) Flow cytometry analysis of HRH1 expression on mouse bone marrow-derived (BMDM) naïve macrophages (M0) and polarized macrophages (M1 and M2-like) (n=3, one-way ANOVA). (F) Mean florescent intensity (MFI) of HRH1 in indicated cell subsets of mouse mammary tumors (4T1 and EO771). Macrophages, CD45+CD11b+Grl−F4/80+; neutrophils, CD45+CD11b+Grl+; lymphocytes, CD45+CD11b−. (G) Flow cytometry analysis of HRH1 expression on resident macrophages (nMφ) from mammary fat pad (MFP) of BALB/c mice and TAMs (4T1 and EMT6 tumors) (n=5, one-way ANOVA). (H) Histamine levels in blood plasma from healthy subjects (n=20), patients with TNBC (n=50), colon cancer (n=28) detected by ELISA (one-way ANOVA). (I) Pearson correlation analysis of the relationship between serum histamine level and GZMB+ cell density (%) in cancer tissues from TNBC patients (n=50). Mean ± SEM, **P<0.01, ****P<0.0001. See also Figure S2. Figure 3. Inhibiting HRH1 on macrophages enhances T cell anti-tumor immunity (A) Flow cytometry analysis of M1-like (MHC II+) versus M2-like (CD206+) populations in bone marrow-derived macrophages (BMDMs) that were generated from wild-type (WT) or HRH1−/− mice and treated with vehicle or fexofenadine (FEXO) (10 μM) in the presence of EO771 tumor cell-conditioned medium (TCM) for 48 hours. (B) Analysis of IFN-γ+CD8+ T cells in splenocytes co-cultured with vehicle- or FEXO (10 μM)-treated WT or HRH1−/− BMDMs (TCM-educated). (C and D) Relative MHC II:CD206 MFI ratios of tumor-associated macrophages (TAMs) (C) and percentage of IFN-γ+CD8+ T cells (D) in EO771 tumors (left) or B16-GM tumors (right) growing in WT versus HRH1−/− mice, and vehicle-treated versus FEXO-treated WT mice (n=5-6, t-test). (E) EO771 (left) and B16-GM (right) tumor growth in WT versus HRH1−/− mice, and vehicle-treated versus FEXO-treated WT mice (n= 5-8 mice/group, two-way ANOVA). (F) B16-GM tumor growth with indicated treatment. CD8+ T cells were depleted by anti-CD8 antibodies (n=6-7 mice/group, two-way ANOVA). (G) Growth of B16-GM tumor cells co-implanted with WT or HRH1−/− BMDMs in HRH1−/− or WT recipient mice, respectively (n=6-9 mice/group, two-way ANOVA). (H) Percentages of IFN-γ+ and PRF1+ CD8+ T cells in primary tumors from WT and HRH1−/− mice transplanted with B16-GM tumor cells alone, or both tumor cells and BMDMs (HRH1−/− or WT respectively) (n=6, one-way ANOVA). (I) t-distributed stochastic neighbor embedding (tSNE) plot of tumor-infiltrating leukocytes overlaid with color-coded clusters in EO771 tumors from WT or HRH1−/− mice. Dotted ellipses highlight clusters with significant differences between two groups. Mean ± SEM, *P<0.05, **P<0.01, ***P<0.001. See also Figures S3, S4 and S5. Figure 4. Histamine-HRH1 activation promotes VISTA membrane localization (A) Percentages of IFN-γ+ CD8+ T cells co-cultured with EO771 TCM-treated WT or HRH1−/− BMDMs in direct contact or separately in transwells (n=6, t-test). (B) Heat-map depicting relative expression of co-stimulatory/inhibitory molecules on naïve or TCM-treated WT or HRH1−/− BMDMs measured by flow cytometry. The mean fluorescence intensity (MFI) of each molecule was normalized to the MFI of the naïve WT group. (C) Percentages of IFN-γ+ CD8+ T cells co-cultured with EO771 TCM-treated WT or HRH1−/− BMDMs pretreated with IgG, anti-TIM-3 (10 μg/ml) and/or anti-VISTA (10 μg/ml) antibodies (n=3-4, one-way ANOVA). (D) Flow cytometry analysis of VISTA+ TAMs (CD45+ CD11b+ F4/80+) from EO771 and B16-GM tumors growing in WT versus HRH1−/− mice, and vehicle-treated versus FEXO-treated WT mice (n=5-6, t-test). (E) Western blot analysis of total VISTA and membrane VISTA expression on naïve or TCM-treated WT or HRH1−/− BMDMs. β-actin and CD11b were used as loading controls. (F) Percentages of VISTA+ BMDMs after treatment with 10 μM BAPTA-AM (an intracellular calcium chelator) or 1 μg/ml ionomycin-Ca2+ (n=3, t-test). All in vitro experiments were performed at least twice. Mean ± SEM, *P<0.05, **P<0.01, ***P<0.001, NS: not significant. See also Figure S6. Figure 5. HRH1 knockout reshapes the transcriptomic landscape of macrophages (A) Volcano plots of log2 fold change (FC) and log10 adjusted P -value of differentially expressed genes between TCM-treated WT and HRH1−/− macrophages. Red dots: genes up-regulated in WT macrophages; Blue dots: genes up-regulated in HRH1−/− macrophages. (B) The pathway enrichment map of differentially expressed genes between WT and HRH1−/− macrophages. (C) Dimensionality reduction by UMAP using all the single cells reveals the fine structure of CD45+ immune cells in mice. Manually inspected automatic annotation of cell identities are labeled in colors as their corresponding cell populations. (D) Violin plot showing the expression-based score of M1-like (top) and M2-like macrophage (bottom), with their mean scores and P values labeled. The horizontal lines represent 25th percentile, median and 75th percentile of the scores. Significance levels are computed using the nonparametric Games-Howell Post-Hoc Test. (E) Violin plot showing the expression-based score of exhausted CD8 T cells with their mean scores and P values labeled. Significance levels are computed using the nonparametric Games-Howell Post-Hoc Test. (F) Scatter-plot results from the Pearson's correlation analysis of HRH1 and M2-like macrophage markers CD163 (left) and CD209 (right) at single-cell level in TAMs of human melanomas (GSE115978). (G) Scatter-plot results from the Pearson's correlation analysis of HRH1 and M1-like macrophage markers IDO1 (left) and IRF1 (right) at single-cell level in TAMs of human melanomas (GSE115978). See also Figure S6. Figure 6. HRH1 inhibition enhances ICB therapeutic efficacy (A) EO771 tumor growth and survival analysis of vehicle- or FEXO-treated WT and PD-L1−/− mice. n=15-23 mice/group for tumor volume analysis, n=9 mice/group for survival analysis, two-way ANOVA for tumor volume comparison, log-rank test for survival comparison. (B) Flow cytometry analysis of VISTA+ TAMs and IFN-γ+ CD8+ T cells in EO771 tumors collected from WT and HRH1−/− mice receiving IgG or anti-PD-1 antibody treatment (n=6, one-way ANOVA). (C) Primary tumor growth (left) and survival analysis (right) of CT26 tumor-bearing WT mice treated with FEXO alone, anti-CTLA-4 alone, or FEXO+anti-CTLA-4. n=9-10 mice/group for tumor volume and survival analysis. (D and E) Tumor growth (D) and survival analysis (E) of B16-GM tumor-bearing WT mice with the indicated treatment (n=6-10 mice/group). (F) Flow cytometry analysis of VISTA+ TAMs and IFN-γ+ CD8+ T cells in primary tumor tissues from B16-GM-bearing mice treated with indicated regimens (n=5, one-way ANOVA). (G) Flow cytometry analysis of MHCII:CD206 ratios of TAMs and IFN-γ+ CD8+ T cells in B16-GM primary tumor tissues from mice treated with indicated regimens (n=6, one-way ANOVA). Mean ± SEM, two-way ANOVA for tumor volume comparison, log-rank test for survival comparison. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. See also Figure S7. Figure 7. HRH1 blockade rescues allergy-induced immunotherapy resistance (A) Experimental schematics of EMT6 tumor model with concurrent allergy. (B) Serum histamine levels detected by ELISA in age-matched healthy mice, allergic mice, and tumor-bearing mice (10 days after EMT6 tumor cell inoculation) with or without induced allergy (n=6, t-test). (C) EMT6 tumor growth in sham control group, allergy group, and allergy plus FEXO treatment group (n=6 mice per group). (D) Flow cytometry analysis of VISTA+ TAMs and IFN-γ+ CD8+ T cells in EMT6 tumor tissues from sham control group, allergy induction group, and allergy plus FEXO treatment group (n=5, one-way ANOVA). (E) EMT6 tumor growth in mice with or without concurrent allergic disease, treated with vehicle, ICB, or ICB+FEXO (n=6 mice per group). (F) CT26 tumor growth in mice with or without concurrent allergic disease, treated with indicated therapies (n=7 mice per group). (G) Comparison of deceased patient percentages according to patient allergy status before receiving anti-PD-1/PD-L1 treatment in melanoma and lung cancer patients. (H) Comparison of plasma histamine levels in pre-treatment blood collected from cancer patient groups with different responses to anti-PD-1 treatment (one-way ANOVA). CR: complete response (100% remission), PR: partial response (≥30% remission), SD: stable disease (<30% remission to <20% increase of tumor size), PD: progressive disease (≥20% increase). (I) A waterfall plot depicting the responses to anti-PD-1 treatment in cancer patients with low levels (<0.3 ng/ml), medium levels (>0.3 ng/ml to <0.6 ng/ml), and high levels (>0.6 ng/ml) of plasma histamine. (J) Assessment of the objective response rate (ORR) and disease control rate (DCR) among cancer patients with different plasma histamine levels (low, medium, and high) (Fisher exact test). (K) The distributions of age, sex, and tumor stage among anti-PD-1 treated lung cancer patients (n=48) with indicated histamine level (Fisher exact test). Mean ± SEM, two-way ANOVA for tumor volume comparison. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, NS, no significant. See also Figure S7 and Tables S3-S6. KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Anti-mouse CD45 PE/Cyanine5 BioLegend Cat# 103110; RRID: AB_312975 Anti-mouse CD3e APC BioLegend Cat# 100312; RRID: AB_312677 Anti-mouse CD8a PE/Cyanine7 BioLegend Cat# 100722; RRID: AB_312761 Anti-mouse/human CD11b BV 510™ BioLegend Cat# 101263; RRID: AB_2629529 Anti-mouse Ly-6G/Ly-6C (Gr-1) PE BioLegend Cat# 108408; RRID: AB_313373 Anti-mouse F4/80 FITC BioLegend Cat# 123108; RRID: AB_893502 Anti-mouse CD206 PE BioLegend Cat# 141706; RRID: AB_10895754 Anti-mouse I-A/I-E Alexa Fluor® 488 BioLegend Cat# 107616; RRID: AB_493523 Anti-mouse VISTA PE/Cyanine7 BioLegend Cat# 143714; RRID: AB_2632662 Anti-mouse CD366 (Tim-3) PE BioLegend Cat# 134004; RRID: AB_1626177 Anti-mouse CD11c APC BioLegend Cat# 117310; RRID: AB_313779 Anti-mouse CD24 Alexa Fluor® 647 BioLegend Cat# 101818; RRID: AB_493484 Anti-mouse GITR Ligand PE BioLegend Cat# 120306; RRID: AB_2207248 Anti-mouse CD275 (ICOS Ligand) PE BioLegend Cat# 107405; RRID: AB_2248797 Anti-mouse 4-1BB Ligand PE BioLegend Cat# 107105; RRID: AB_2256408 Anti-mouse CD276 (B7-H3) PE BioLegend Cat# 124508; RRID: AB_1279206 Anti-mouse CD252 Alexa Fluor® 647 BioLegend Cat# 108810; RRID: AB_2207379 Anti-mouse Galectin-9 APC BioLegend Cat# 136110; RRID: AB_2561658 Anti-mouse Ki-67 FITC BioLegend Cat# 652410; RRID: AB_2562141 Anti-mouse PRF1 PE BioLegend Cat# 154306; RRID: AB_2721639 Anti-mouse IFN-γ PE BioLegend Cat# 505810; RRID: AB_315404 Anti-mouse CD274 PE/Cyanine7 BioLegend Cat# 124314; RRID: AB_10643573 Anti-mouse CD3ε antibody BioLegend Cat# 100372; RRID: AB_2800556 Anti-mouse CD28 antibody BioLegend Cat# 102132; RRID: AB_2810333 Anti-mouse CD16/32 antibody BioLegend Cat# 101302; RRID: AB_312801 Anti-human HRH1 Alexa Fluor® 647 R&D Systems Cat# FAB4726R Anti-mouse Siglec-F PerCP-Cy™5.5 BD Biosciences Cat# 565526; RRID: AB_2739281 HRH1 antibody Alomone Labs Cat# AHR-001; RRID: AB_2039915 HRH1 antibody LifeSpan BioSciences Cat# LS-C331459 HRH1 antibody Abcam Cat# ab75236; RRID: AB_2092479 CD68 antibody Abcam Cat# ab955; RRID: AB_307338 VISTA antibody Cell Signaling Technology Cat# 54979; RRID: AB_2799474 CD11b/c antibody Novus Biologicals Cat# NB110-40766; RRID: AB_714950 β-actin Santa Cruz Biotechnology Cat # SC47778; RRID: AB_2714189 Anti-mouse IgG, DyLight 594 Thermo Fisher Scientific Cat # 35510; RRID: AB_1185569 Anti-Rabbit IgG, DyLight 488 Thermo Fisher Scientific Cat # 35552; RRID: AB_844398 Anti-mouse CTLA-4 (CD152) Bio X Cell Cat # BE0164; RRID: AB_10949609 Anti-mouse VISTA Bio X Cell Cat # BE0310; RRID: AB_2736990 Anti-mouse CD8α Bio X Cell Cat # BE0117; RRID: AB_10950145 CD45, Label: 89Y Fluidigm Cat # 3089005B; RRID: AB_2651152 FoxP3, Label: 165Ho Fluidigm Cat # 3165024A; RRID: AB_2687843 Granzyme B, Label: 173Yb Fluidigm Cat # 3173006B; RRID: AB_2811095 TIM-3, Label: 162Dy Fluidigm Cat # 3162029B; RRID: AB_2687841 CD357, Label: 143Nd Fluidigm Cat # 3143019B CD80, Label: 171Yb Fluidigm Cat # 3171008B CD86, Label: 172Yb Fluidigm Cat # 3172016B CD40, Label: 161Dy Fluidigm Cat # 3161020B CD278, Label: 176Yb Fluidigm Cat # 3176014B CD39, Label: 142Nd Fluidigm Cat # 3142005B CD11b, Label: 148Nd Fluidigm Cat # 3148003B; RRID: AB_2814738 CD11c, Label: 209Bi Fluidigm Cat # 3209005B; RRID: AB_2811244 Ly-6C, Label: 150Nd Fluidigm Cat # 3150010B Ly-6G, Label: 141Pr Fluidigm Cat # 3141008B; RRID: AB_2814678 CD38, Label: 175Lu Fluidigm Cat # 3175014B I-A/I-E, Label: 174Yb Fluidigm Cat # 3174003B CD206, Label: 169Tm Fluidigm Cat # 3169021B; RRID: AB_2832249 CD274, Label: 153Eu Fluidigm Cat # 3153016B; RRID: AB_2687837 CD4, Label: 115In BioLegend Cat # 100506; RRID: AB_312709 CD3e, Label: 152Sm BioLegend Cat # 100302; RRID: AB_312667 CD8a, Label: 146Nd BioLegend Cat # 100702; RRID: AB_312741 NK1.1, Label: 170Er BioLegend Cat # 108702; RRID: AB_313389 CD19, Label: 149Sm BioLegend Cat # 115502; RRID: AB_313637 T-bet, Label: 154Sm BioLegend Cat # 644825; RRID: AB_2563788 IRF4, Label: 151Eu BioLegend Cat # 646402; RRID: AB_2280462 CD152, Label: 163Dyv BioLegend Cat # 106202; RRID: AB_313247 CD69, Label: 156Gd BioLegend Cat # 104533; RRID: AB_2563760 CD14, Label: 158Gd BioLegend Cat # 123302; RRID: AB_940592 F4/80, Label: 159Tb BioLegend Cat # 123102; RRID: AB_893506 Ly-6G/C, Label: 139La BioLegend Cat # 108402; RRID: AB_313367 CD103, Label: 147Sm BioLegend Cat # 121401; RRID: AB_535944 GATA3, Label: 145Nd Thermo Fisher Scientific Cat # 14-9966-82; RRID: AB_1210519 Ki67, Label: 168Er BD Biosciences Cat # 556003; RRID: AB_396287 CCR7, Label: 155Gd Thermo Fisher Scientific Cat # 16-1971-85; RRID: AB_494123 Bacterial and Virus Strains DH5α Thermo Scientific Cat# 18265017 Stbl3 Thermo Scientific Cat# C737303 Biological Samples Human samples (Normal breast tissues, breast and colon cancer tissues) The First Affiliated Hospital of Chongqing Medical University, China. Project approval number 1005367 2017-012 The blood plasma from patients with advanced lung (n=48), colon (n=12) and breast (n=10) cancers collected pre-anti-PD1 treatment (patients were treated with Camrelizumab between Dec 24, 2019 and Feb 27, 2021) Beijing Friendship Hospital of Capital Medical Project approval number 2017-P2-141-01 Chemicals, Peptides, and Recombinant Proteins Fexofenadine HCl Sigma-Aldrich Cat # PHR1685; CAS: 153439-40-8 Ovalbumin Sigma-Aldrich Cat # A5503; CAS: 9006-59-1 Puromycin dihydrochloride Sigma-Aldrich Cat# P8833; CAS: 58-58-2 Collagenase A Roche Cat # 11088793001 16% Formaldehyde, Methanol-free Pierce Cat # 28906 Ionomycin, Calcium Salt Cell Signaling Technology Cat # 9995; CAS: 56092-82-1 BAPTA-AM Selleckchem Cat # S7534; CAS: 126150-97-8 Recombinant Murine IFN-γ PeproTech Cat # 315-05 Recombinant Murine IL-4 PeproTech Cat # 214-14 Fixable Viability Dye eFluor™ 450 eBioscience Cat # 65-0863-14 CD11b MicroBeads Miltenyi Biotec Cat # 130-049-601 Cell-ID™ Cisplatin Fluidigm Cat # 201064 Critical Commercial Assays Histamine ELISA kits Enzo Life Sciences Cat # ENZ-KIT140-0001 iScript™ cDNA Synthesis Kit Bio-rad Cat # 1708891 Intracellular Fixation & Permeabilization Buffer Set eBioscience Cat # 88-8824-00 Maxima SYBR Green qPCR Master Mix Thermo Fisher Scientific Cat# K0253 QIAGEN Plasmid Maxi Kit QIAGEN Cat# 12163 ImmunoSpot® Kits Cellular Technology Limited Mouse IFN-γ Single-Color ELISPOT Deposited Data RNA-seq with mouse BMDM (Raw and analyzed data) This paper GEO: GSE161484 RNA-seq with human melanomas Hugo et al 2016 GEO: GSE78220 Single-cell RNA-seq of human melanoma Livnat Jerby-Arnon et al 2018 GEO: GSE115978 Single-cell RNA-seq with CD45+ immune cells isolated from EO771 tumors This paper SRA Run Selector Accession: PRJNA756466   Experimental Models: Cell Lines 293T ATCC Cat# ACS-4500; RRID:CVCL_4V93 4T1 ATCC Cat# CRL-2539; RRID:CVCL_0125 B16/BL6 ATCC Cat# CRL-6475; RRID:CVCL_0159 BT20 ATCC Cat# HTB-19; RRID:CVCL_0178 BT549 ATCC Cat# HTB-122; RRID:CVCL_1092 CT26 ATCC Cat# CRL-2638; RRID:CVCL_7256 EMT6 ATCC Cat# CRL-2755; RRID:CVCL_1923 EO771 ATCC Cat# CRL-3461; RRID:CVCL_GR23 HCC1806 ATCC Cat# CRL-2335; RRID:CVCL_1258 HS578T ATCC Cat# HTB-126; RRID:CVCL_0332 L929 ATCC Cat# CCL-1; RRID:CVCL_0462 LLC1 ATCC Cat# CRL-1642; RRID:CVCL_4358 MDA-MB-231 ATCC Cat# HTB-26; RRID:CVCL_0062 MDA-MB-435 ATCC Cat# HTB-129; RRID:CVCL_0622 MDA-MB-436 ATCC Cat# HTB-130; RRID:CVCL_0623 THP-1 ATCC Cat# TIB-202; RRID:CVCL_0006 B16-GMCSF N/A Experimental Models: Organisms/Strains Mouse: C57BL/6 The Jackson Laboratory N/A Mouse: BALB/c The Jackson Laboratory N/A Mouse: B6.129P2-Hrh1tm1Wtn/BrenJ The Jackson Laboratory Stock No: 029346 Mouse: PD-L1−/− Dr. Don L. Gibbons Chen et al. (2014) Oligonucleotides See Table S2 for a detailed primer list This paper N/A Recombinant DNA pLKO.1-shTGFB1-1 Sigma-Aldrich TRCN0000065993 pLKO.1-shTGFB1-2 Sigma-Aldrich TRCN0000065997 Software and Algorithms FlowJo BD Biosciences https://www.flowjo.com/solutions/flowjo/downloads GraphPad Prism GraphPad https://www.graphpad.com/scientific-software/prism/ ImageJ NIH https://imagej.nih.gov/ij/ R Studio N/A https://www.rstudio.com Cytofkit Jinmiao Chen Lab https://github.com/JinmiaoChenLab/cytofkit2 Highlights Histamine binding of HRH1 on macrophage induces an immunosuppressive phenotype; H1-antihistamine treatment enhances immunotherapy response; Allergic reaction promotes immune evasion and resistance to immunotherapy; High histamine and HRH1 levels correlate with poor immunotherapy response in patients. COMPETING INTERESTS The authors declare no competing interests. INCLUSION AND DIVERSITY While citing references scientifically relevant for this work, we also actively worked to promote gender balance in our reference list. The author list of this paper includes contributors from the location where the research was conducted who participated in the data collection, design, analysis, and/or interpretation of the work. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. REFERENCES Alquicira-Hernandez J , Sathe A , Ji HP , Nguyen Q , and Powell JE (2019). scPred: accurate supervised method for cell-type classification from single-cell RNA-seq data. Genome Biol 20 , 264.31829268 Ashina K , Tsubosaka Y , Nakamura T , Omori K , Kobayashi K , Hori M , Ozaki H , and Murata T (2015). Histamine Induces Vascular Hyperpermeability by Increasing Blood Flow and Endothelial Barrier Disruption In Vivo. PLoS One 10 , e0132367.26158531 Blando J , Sharma A , Higa MG , Zhao H , Vence L , Yadav SS , Kim J , Sepulveda AM , Sharp M , Maitra A , (2019). Comparison of immune infiltrates in melanoma and pancreatic cancer highlights VISTA as a potential target in pancreatic cancer. Proc Natl Acad Sci U S A 116 , 1692–1697.30635425 Chen L , Gibbons DL , Goswami S , Cortez MA , Ahn YH , Byers LA , Zhang X , Yi X , Dwyer D , Lin W , (2014). Metastasis is regulated via microRNA-200/ZEB1 axis control of tumour cell PD-L1 expression and intratumoral immunosuppression. Nat Commun 5 , 5241.25348003 De Henau O , Rausch M , Winkler D , Campesato LF , Liu C , Cymerman DH , Budhu S , Ghosh A , Pink M , Tchaicha J , (2016). Overcoming resistance to checkpoint blockade therapy by targeting PI3Kgamma in myeloid cells. Nature 539 , 443–447.27828943 DeNardo DG , and Ruffell B (2019). Macrophages as regulators of tumour immunity and immunotherapy. Nat Rev Immunol 19 , 369–382.30718830 Elkrief A , Derosa L , Kroemer G , Zitvogel L , and Routy B (2019). The negative impact of antibiotics on outcomes in cancer patients treated with immunotherapy: a new independent prognostic factor? Ann Oncol 30 , 1572–1579.31268133 Faustino-Rocha AI , Ferreira R , Gama A , Oliveira PA , and Ginja M (2017). Antihistamines as promising drugs in cancer therapy. Life Sci 172 , 27–41.27986539 Fritz I , Wagner P , Bottai M , Eriksson H , Ingvar C , Krakowski I , Nielsen K , and Olsson H (2020). Desloratadine and loratadine use associated with improved melanoma survival. Allergy. Gao J , Ward JF , Pettaway CA , Shi LZ , Subudhi SK , Vence LM , Zhao H , Chen J , Chen H , Efstathiou E , (2017). VISTA is an inhibitory immune checkpoint that is increased after ipilimumab therapy in patients with prostate cancer. Nat Med 23 , 551–555.28346412 Gerrick KY , Gerrick ER , Gupta A , Wheelan SJ , Yegnasubramanian S , and Jaffee EM (2018). Transcriptional profiling identifies novel regulators of macrophage polarization. PLoS One 13 , e0208602.30532146 Grauers Wiktorin H , Nilsson MS , Kiffin R , Sander FE , Lenox B , Rydstrom A , Hellstrand K , and Martner A (2019). Histamine targets myeloid-derived suppressor cells and improves the anti-tumor efficacy of PD-1/PD-L1 checkpoint blockade. Cancer Immunol Immunother 68 , 163–174.30315349 Guerriero JL (2018). Macrophages: The Road Less Traveled, Changing Anticancer Therapy. Trends Mol Med 24 , 472–489.29655673 Guerriero JL (2019). Macrophages: Their Untold Story in T Cell Activation and Function. Int Rev Cell Mol Biol 342 , 73–93.30635094 Haak-Frendscho M , Darvas Z , Hegyesi H , Karpati S , Hoffman RL , Laszlo V , Bencsath M , Szalai C , Furesz J , Timar J , (2000). Histidine decarboxylase expression in human melanoma. J Invest Dermatol 115 , 345–352.10951267 Hao Y , Hao S , Andersen-Nissen E , Mauck WM 3rd , Zheng S , Butler A , Lee MJ , Wilk AJ , Darby C , Zager M , (2021). Integrated analysis of multimodal single-cell data. Cell 184 , 3573–3587 e3529.34062119 Hugo W , Zaretsky JM , Sun L , Song C , Moreno BH , Hu-Lieskovan S , Berent-Maoz B , Pang J , Chmielowski B , Cherry G , (2016). Genomic and Transcriptomic Features of Response to Anti-PD-1 Therapy in Metastatic Melanoma. Cell 165 , 35–44.26997480 Jablonski KA , Amici SA , Webb LM , Ruiz-Rosado Jde D , Popovich PG , Partida-Sanchez S , and Guerau-de-Arellano M (2015). Novel Markers to Delineate Murine M1 and M2 Macrophages. PLoS One 10 , e0145342.26699615 Jerby-Arnon L , Shah P , Cuoco MS , Rodman C , Su MJ , Melms JC , Leeson R , Kanodia A , Mei S , Lin JR , (2018). A Cancer Cell Program Promotes T Cell Exclusion and Resistance to Checkpoint Blockade. Cell 175 , 984–997 e924.30388455 Jiang P , Gu S , Pan D , Fu J , Sahu A , Hu X , Li Z , Traugh N , Bu X , Li B , (2018). Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med 24 , 1550–1558.30127393 Johnston RJ , Su LJ , Pinckney J , Critton D , Boyer E , Krishnakumar A , Corbett M , Rankin AL , Dibella R , Campbell L , (2019). VISTA is an acidic pH-selective ligand for PSGL-1. Nature 574 , 565–570.31645726 Khononov I , Jacob E , Fremder E , Dahan N , Harel M , Raviv Z , Krastev B , and Shaked Y (2021). Host response to immune checkpoint inhibitors contributes to tumor aggressiveness. J Immunother Cancer 9 . Kugelmann D , Rotkopf LT , Radeva MY , Garcia-Ponce A , Walter E , and Waschke J (2018). Histamine causes endothelial barrier disruption via Ca(2+)-mediated RhoA activation and tension at adherens junctions. Sci Rep 8 , 13229.30185878 Kumar RK , Herbert C , and Foster PS (2008). The "classical" ovalbumin challenge model of asthma in mice. Curr Drug Targets 9 , 485–494.18537587 Le Mercier I , Chen W , Lines JL , Day M , Li J , Sergent P , Noelle RJ , and Wang L (2014). VISTA Regulates the Development of Protective Antitumor Immunity. Cancer Res 74 , 1933–1944.24691994 Li T , Fan J , Wang B , Traugh N , Chen Q , Liu JS , Li B , and Liu XS (2017). TIMER: A Web Server for Comprehensive Analysis of Tumor-Infiltrating Immune Cells. Cancer Res 77 , e108–e110.29092952 Lines JL , Pantazi E , Mak J , Sempere LF , Wang L , O'Connell S , Ceeraz S , Suriawinata AA , Yan S , Ernstoff MS , (2014). VISTA is an immune checkpoint molecule for human T cells. Cancer Res 74 , 1924–1932.24691993 Liu J , Yuan Y , Chen W , Putra J , Suriawinata AA , Schenk AD , Miller HE , Guleria I , Barth RJ , Huang YH , (2015). Immune-checkpoint proteins VISTA and PD-1 nonredundantly regulate murine T-cell responses. Proc Natl Acad Sci U S A 112 , 6682–6687.25964334 Lu C , Diehl SA , Noubade R , Ledoux J , Nelson MT , Spach K , Zachary JF , Blankenhorn EP , and Teuscher C (2010). Endothelial histamine H1 receptor signaling reduces blood-brain barrier permeability and susceptibility to autoimmune encephalomyelitis. Proc Natl Acad Sci U S A 107 , 18967–18972.20956310 Martin RK , Saleem SJ , Folgosa L , Zellner HB , Damle SR , Nguyen GK , Ryan JJ , Bear HD , Irani AM , and Conrad DH (2014). Mast cell histamine promotes the immunoregulatory activity of myeloid-derived suppressor cells. J Leukoc Biol 96 , 151–159.24610880 Martinez FO , Gordon S , Locati M , and Mantovani A (2006). Transcriptional profiling of the human monocyte-to-macrophage differentiation and polarization: new molecules and patterns of gene expression. J Immunol 177 , 7303–7311.17082649 Massari NA , Nicoud MB , and Medina VA (2018). Histamine receptors and cancer pharmacology: an update. Br J Pharmacol. Micaroni M (2010). The role of calcium in intracellular trafficking. Curr Mol Med 10 , 763–773.20937019 Micaroni M (2012). Calcium around the Golgi apparatus: implications for intracellular membrane trafficking. Adv Exp Med Biol 740 , 439–460.22453953 Misharin AV , Morales-Nebreda L , Mutlu GM , Budinger GR , and Perlman H (2013). Flow cytometric analysis of macrophages and dendritic cell subsets in the mouse lung. Am J Respir Cell Mol Biol 49 , 503–510.23672262 Moriarty CM , Stucky JL , Hamburger KW , Patil KD , Foley JF , and Koefoot RR (1988). Blood histamine and solid malignant tumors. J Cancer Res Clin Oncol 114 , 588–592.3204106 Mosely SI , Prime JE , Sainson RC , Koopmann JO , Wang DY , Greenawalt DM , Ahdesmaki MJ , Leyland R , Mullins S , Pacelli L , (2017). Rational Selection of Syngeneic Preclinical Tumor Models for Immunotherapeutic Drug Discovery. Cancer Immunol Res 5 , 29–41.27923825 Newman AM , Liu CL , Green MR , Gentles AJ , Feng W , Xu Y , Hoang CD , Diehn M , and Alizadeh AA (2015). Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 12 , 453–457.25822800 Nials AT , and Uddin S (2008). Mouse models of allergic asthma: acute and chronic allergen challenge. Dis Model Mech 1 , 213–220.19093027 Nowak EC , Lines JL , Varn FS , Deng J , Sarde A , Mabaera R , Kuta A , Le Mercier I , Cheng C , and Noelle RJ (2017). Immunoregulatory functions of VISTA. Immunol Rev 276 , 66–79.28258694 Ocana-Guzman R , Torre-Bouscoulet L , and Sada-Ovalle I (2016). TIM-3 Regulates Distinct Functions in Macrophages. Front Immunol 7 , 229.27379093 Orecchioni M , Ghosheh Y , Pramod AB , and Ley K (2019). Macrophage Polarization: Different Gene Signatures in M1(LPS+) vs. Classically and M2(LPS−) vs. Alternatively Activated Macrophages. Front Immunol 10 , 1084.31178859 Ostuni R , Kratochvill F , Murray PJ , and Natoli G (2015). Macrophages and cancer: from mechanisms to therapeutic implications. Trends Immunol 36 , 229–239.25770924 Parsons ME , and Ganellin CR (2006). Histamine and its receptors. Br J Pharmacol 147 Suppl 1 , S127–135.16402096 Ribas A , and Wolchok JD (2018). Cancer immunotherapy using checkpoint blockade. Science 359 , 1350–1355.29567705 Rittmeyer D , and Lorentz A (2012). Relationship between allergy and cancer: an overview. Int Arch Allergy Immunol 159 , 216–225.22722389 Sharma P , Hu-Lieskovan S , Wargo JA , and Ribas A (2017). Primary, Adaptive, and Acquired Resistance to Cancer Immunotherapy. Cell 168 , 707–723.28187290 Sieja K , Stanosz S , von Mach-Szczypinski J , Olewniczak S , and Stanosz M (2005). Concentration of histamine in serum and tissues of the primary ductal breast cancers in women. Breast 14 , 236–241.15927833 Simons FE (2004). Advances in H1-antihistamines. N Engl J Med 351 , 2203–2217.15548781 Thurmond RL , Gelfand EW , and Dunford PJ (2008). The role of histamine H1 and H4 receptors in allergic inflammation: the search for new antihistamines. Nat Rev Drug Discov 7 , 41–53.18172439 Turner MC , Chen Y , Krewski D , and Ghadirian P (2006). An overview of the association between allergy and cancer. Int J Cancer 118 , 3124–3132.16395696 Uhlen M , Zhang C , Lee S , Sjostedt E , Fagerberg L , Bidkhori G , Benfeitas R , Arif M , Liu Z , Edfors F , (2017). A pathology atlas of the human cancer transcriptome. Science 357 . Verma V , Shrimali RK , Ahmad S , Dai W , Wang H , Lu S , Nandre R , Gaur P , Lopez J , Sade-Feldman M , (2019). PD-1 blockade in subprimed CD8 cells induces dysfunctional PD-1(+)CD38(hi) cells and anti-PD-1 resistance. Nat Immunol 20 , 1231–1243.31358999 von Mach-Szczypinski J , Stanosz S , Sieja K , and Stanosz M (2009). Metabolism of histamine in tissues of primary ductal breast cancer. Metabolism 58 , 867–870.19375125 Yang XD , Ai W , Asfaha S , Bhagat G , Friedman RA , Jin G , Park H , Shykind B , Diacovo TG , Falus A , (2011). Histamine deficiency promotes inflammation-associated carcinogenesis through reduced myeloid maturation and accumulation of CD11b+Ly6G+ immature myeloid cells. Nat Med 17 , 87–95.21170045 Yao J , Zhao Q , Yuan Y , Zhang L , Liu X , Yung WK , and Weinstein JN (2012). Identification of common prognostic gene expression signatures with biological meanings from microarray gene expression datasets. PLoS One 7 , e45894.23029298 Zelenay S , van der Veen AG , Bottcher JP , Snelgrove KJ , Rogers N , Acton SE , Chakravarty P , Girotti MR , Marais R , Quezada SA , (2015). Cyclooxygenase-Dependent Tumor Growth through Evasion of Immunity. Cell 162 , 1257–1270.26343581 Zhang L , Yao J , Wei Y , Zhou Z , Li P , Qu J , Badu-Nkansah A , Yuan X , Huang YW , Fukumura K , (2020). Blocking immunosuppressive neutrophils deters pY696-EZH2-driven brain metastases. Sci Transl Med 12 . Zhang L , Zhang S , Yao J , Lowery FJ , Zhang Q , Huang WC , Li P , Li M , Wang X , Zhang C , (2015). Microenvironment-induced PTEN loss by exosomal microRNA primes brain metastasis outgrowth. Nature 527 , 100–104.26479035 Zhang X , Lan Y , Xu J , Quan F , Zhao E , Deng C , Luo T , Xu L , Liao G , Yan M , (2019). CellMarker: a manually curated resource of cell markers in human and mouse. Nucleic Acids Res 47 , D721–D728.30289549 Zheng GX , Terry JM , Belgrader P , Ryvkin P , Bent ZW , Wilson R , Ziraldo SB , Wheeler TD , McDermott GP , Zhu J , (2017). Massively parallel digital transcriptional profiling of single cells. Nat Commun 8 , 14049.28091601

PMC009xxxxxx/PMC9117566.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9100813 25889 Account Res Account Res Accountability in research 0898-9621 1545-5815 34743618 9117566 10.1080/08989621.2021.2002693 NIHMS1766989 Article Reducing Tensions and Expediting Manuscript Submission Via an Authorship Agreement for Early-Career Researchers: A Pilot Study Norman Marie K. PhD http://orcid.org/0000-0003-1307-0115 Institute for Clinical Research Education, University of Pittsburgh, School of Medicine Pittsburgh, PA, USA; Clinical and Translational Science Institute, University of Pittsburgh Proulx Chelsea N. MPH http://orcid.org/0000-0001-9269-2355 Institute for Clinical Research Education, University of Pittsburgh, School of Medicine Pittsburgh, PA, USA; Clinical and Translational Science Institute, University of Pittsburgh Rubio Doris M. PhD http://orcid.org/0000-0001-9540-6174 Institute for Clinical Research Education, University of Pittsburgh, School of Medicine Pittsburgh, PA, USA; Clinical and Translational Science Institute, University of Pittsburgh Mayowski Colleen A. EdD, MLIS http://orcid.org/0000-0003-3752-6455 *Institute for Clinical Research Education, University of Pittsburgh, School of Medicine Pittsburgh, PA, USA; Clinical and Translational Science Institute, University of Pittsburgh * Corresponding author: Twitter: @PittICRE mayowski@pitt.edu 12 2 2022 12 2023 19 11 2021 01 12 2024 30 7 379392 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Authorship can be a source of tension on research teams, in academic/industry collaborations, and between mentors/mentees. Authorship misconduct is prevalent among biomedical researchers, and disputes about authorship can generate tensions that have the potential to disrupt professional relationships and damage careers. Early-career researchers may experience particular challenges navigating authorship both because of inexperience and power differentials; in effect, they lack the language and confidence to have these conversations and may feel unwilling to challenge the status quo. The authors implemented an Authorship Agreement for use when collaborating on a manuscript and hypothesized that using this agreement would reduce authorship tensions and speed time to manuscript submission by helping early-career investigators manage authorship conversations more effectively. The authors surveyed trainees (n=65) on the prevalence of authorship-related tensions and compared the results from the first survey in 2017 to the final survey administered in 2020. The decrease in tensions around meeting deadlines was significant (z=2.59, p=0.010). The authors believe the effect of an Authorship Agreement on authorship-related tensions has not previously been investigated. This work extends what is known about the prevalence of commonly cited authorship tensions, and provides evidence of the effectiveness of steps that can be taken to alleviate them. pmcIntroduction The publication of peer-review papers is centrally important to academic career success and thus a high-stakes enterprise (Cutas and Shaw 2015; Moffatt 2018; Seeman and House 2015; Smith and Master 2017; Resnik and Master 2011; Smith, et al. 2019a; Tang 2018). As a result, authorship can be a source of considerable personal and professional tension on research teams, in academic/industry collaborations, and in the mentor/mentee relationship (Borenstein and Shamoo 2015; Cutas and Shaw 2015; Smith and Williams-Jones 2012). Tensions are known to occur around instances of ghost authorship (Wislar et al. 2011) and guest authorship (Moffatt 2011). Authorship order is a high-stakes issue as well, with conventions varying across disciplines (Borenstein and Shamoo 2015; Seeman and House 2015; Smith and Master 2017). Conflicts can also arise on authoring teams over issues of accountability, with tensions potentially arising over missed deadlines or inequitable contributions. Whether these are simple disagreements expressed by both parties, a feeling of unfairness repressed or expressed by an individual, or a screaming match, these tensions are consequential and have been described as detrimental to individual careers and to research integrity as a whole (Cutas and Shaw 2015; Maggio et al. 2018; Smith, et al. 2019b; Martinson et al. 2006; Youtie and Bozeman 2014). While documented across a variety of disciplines, authorship tensions are known to be especially prevalent in STEM fields and among biomedical researchers (Bozeman and Youtie 2016; Cutas and Shaw 2015; Uijtdehaage, Mavis, and Durning 2018). Although conflicts and tensions around authorship can and do occur among researchers at any career stage, early-career researchers are most vulnerable, in many cases due to real or perceived power differentials with their more senior colleagues, and the concern that these conversations might offend, convey mistrust, or seem less than collegial (Roberts 2017; Cutas and Shaw 2015; Bozeman and Youtie 2016; Kwok 2005; Youtie and Bozeman 2016; Borenstein and Shamoo 2015; Seeman and House 2015; Primack, Cigliano, and Parsons 2014). When documenting these tensions, researchers have proposed solutions ranging from dismantling authorship in favor of contributorship (Rennie, Yank, and Emanuel 1997), using checklists and guidelines to decide whether a contribution rises to the level of authorship (Bennett and Gadlin 2012; Smith 1997; COPE 2021; ICMJE 2021; WAME 2016; Primack, Cigliano, and Parsons 2014), and implementing detailed grids (Phillippi, Likis, and Tilden 2018) to determine authorship order in qualitative and quantitative research papers. All of these worthwhile suggestions contribute to transparency and fairness in authorship decisions. Additionally, when discussing these solutions, researchers often emphasize the importance of early and frequent communication (Youtie and Bozeman 2016, 2014; Bozeman and Youtie 2016; Heard 2016) between the people authoring the manuscript, and recommend documenting the results of these conversations in writing (Youtie and Bozeman 2016; Roberts 2017; Bozeman and Youtie 2016; Bennett, Gadlin, and Marchand 2018; Primack, Cigliano, and Parsons 2014; Gadlin and Jessar 2002; Bennett and Gadlin 2012). Such communication strategies are widely viewed as essential for preventing tensions that can arise around authorship. However, there appears to be a disconnect between recommended practices regarding frequent communication and documentation and “real-life” behavior. Indeed, the pervasiveness of authorship misconduct and abuses reported in the literature suggests that such recommendations are not commonly followed—perhaps because initiating these conversations is potentially uncomfortable. At the Institute for Clinical Research Education (ICRE) in the University of Pittsburgh, where we train the next generation of clinical and translational scientists, (“ICRE Home Page” 2021) we wanted to reduce the anxiety around these conversations for early-career investigators and facilitate more productive, early, and consistent dialogue about authorship. We designed and implemented an Authorship Agreement for ICRE trainees and their senior colleagues to use when collaborating on a manuscript, hypothesizing that using this agreement would reduce commonly reported authorship tensions and speed time to manuscript submission. In this paper, we describe how we created the agreement, tested it, refined it, and assessed its effects on tensions on authoring teams and time to manuscript submission. Materials and Methods Our goal was to develop both an agreement and a process for using it that would help early-career researchers comfortably initiate authorship conversations with their mentor or research team and thus reduce the probability of tensions (Roberts 2017; Bennett and Taylor 2003). We wanted the agreement to satisfy six criteria. (See Suggested Call-out Box 1. Six Criteria for the Agreement) First, it needed to be simple, so as not to impose an undue burden on authorship teams. Second, it had to be flexible, and speak to authorship teams of varied sizes and compositions. Third, it had to enumerate accepted authorship criteria, so that it educated teams and provided a common vocabulary for discussing authorship while also providing guidelines to intended authors. (Although none of the widely known authorship criteria fit perfectly, after reviewing COPE (2021), ICMJE (2021), and WAME (2016), we chose the ICMJE criteria, as many of the journals our trainees submit to have adopted these guidelines. Obviously, if journal guidelines contradict the ICMJE criteria, teams would negotiate that themselves.) Fourth, it had to include fields for designating both authorship credit and authorship order. Fifth, we wanted a field for the target journal as well as an intended submission date to promote collaborative goal-setting. Finally, it had to be a living, editable document, in recognition that circumstances change (Norman, Mayowski, and Fine 2020; Primack, Cigliano, and Parsons 2014) and authorship decisions may as well (Youtie and Bozeman 2016). We began by surveying the literature to determine whether such an agreement existed. We found checklists (COPE 2021), scorecards (Gaffey 2021), grids (Phillippi, Likis, and Tilden 2018), and even a “prenuptial agreement” (Gadlin and Jessar 2002; Bennett, Gadlin, and Marchand 2018), but we were unable to find an actual agreement form that met our needs. We then created a prototype, soliciting input from target users (ICRE pre- and post-doc trainees and junior faculty) during a regularly scheduled training seminar time, collecting trainee feedback and refinements on the agreement. We finalized the revised agreement (Figure 1. Authorship Agreement Form) and made it available to our trainees in October 2017, explaining its purpose and asking them to begin using it. To further encourage mentors and trainees to complete an Authorship Agreement form when writing together, ICRE leadership published a statement of support in the ICRE program handbooks (Authorship Agreement Statement, Figure 2.) In this way, we avoided putting the burden of implementing a new process on the very people it was supposed to free from an uncomfortable task. In practice, this statement serves as a recommendation, and adherence is not actively monitored. To measure the effectiveness of the Authorship Agreement, we created a survey designed to gather data on the prevalence and nature of authorship-related tensions experienced by ICRE trainees, how long it took to write and submit a manuscript, and trainees’ perception of whether that process moved in a timely way. To develop the questions on our survey, we drew on the common areas of tension cited in the literature, as well as our goals for the authorship agreement. Face validity was established by vetting the survey with experienced ICRE researchers and mentors. Study Population The trainees surveyed represented a diverse, interdisciplinary group of early-career clinical and translational science researchers from a variety of biomedical research disciplines and educational backgrounds; for example, critical care medicine, human genetics, bioengineering, nursing, immunology, pharmacy, and social work. During the 2017–2020 study period, the ICRE trained a total of 65 TL1 and KL2 scholars. The TL1 grant mechanism is designed to provide predoctoral and postdoctoral clinical and translational research training (supported by an NIH Clinical and Translational Science Award [CTSA]). The KL2 supports newly trained clinicians for activities related to the development of a successful clinical and translational research career. While the number surveyed remained constant, the individuals comprising the study population changed during the study period due to normal, expected events such as trainees transitioning out and new trainees joining the programs. At each timepoint during this study, the ICRE was training 10 TL1 pre-docs, 10 TL1 post-docs, and 9 KL2 scholars. Therefore, we surveyed 29 early-career researchers at each timepoint. (Table 1). Survey The survey (Figure 3. Survey) was administered via an emailed link in September 2017 prior to introducing the Authorship Agreement form. Following the initial administration, we surveyed ICRE KL2 junior faculty and TL1 pre- and post-doc trainees twice per year. Question 1 was answered on a 5-point Likert scale with possible responses of Never; Rarely; Sometimes; Often; Very Often. Question 2 allowed for an open-ended text response. Question 3 was answered on a 5-point Likert scale with possible responses of Significantly Delayed; Somewhat Delayed; On Time; Somewhat Ahead of Schedule; Significantly Ahead of Schedule. To answer our research question, we compared responses to questions 1, 2, and 3 collected from the baseline survey in September 2017 with responses collected at the study’s endpoint, June 2020. Response rates were high: 26 trainees (87%) responded to the 2017 survey, and 22 (73%) responded in 2020. Participation was optional and responses were deidentified prior to analysis. Follow-up surveys included additional questions designed to explore the experience of using the form and are beyond the scope of this manuscript. This study was approved by the University of Pittsburgh IRB, Protocol# STUDY20050167. Analysis Stata Version 14.2 (College Station, TX: StataCorp LLC) was used for all analyses. We compared the prevalence of reported authorship tensions experienced by ICRE trainees who responded to our survey in September 2017—just before the Authorship Agreement form was introduced—and June 2020, the endpoint of the study. For Question 1, “Think about the last manuscript you submitted (it does not need to be accepted). To what extent were there tensions among co-authors about: a) Who should receive authorship credit? b) Authorship order? c) A target journal? d) Roles and responsibilities? e) Meeting deadlines?” we dichotomized the Likert scale for the sub-questions due to small cell count, with responses of “Sometimes,” “Often,” or “Very Often,” indicating experienced tension. As we did with Question 1, we dichotomized responses to Question 3, “To what extent do you think the manuscript was submitted in a timely manner?” to represent delayed vs timely, with delayed represented by “Significantly Delayed” and “Somewhat Delayed.” All dichotomized responses were analyzed using a two-sample z-test of proportions between the two survey time-points. Responses to Question 2, “Please estimate how long it took for the manuscript to go from first draft to submission” were converted to days; for example, if a participant responded “6 weeks,” it was converted to 42 days. The difference in mean time to submission between the two time-points was tested using a two-sample t-test. Significance for all tests were determined by p-value<0.05. Results Responses to Question 1 and 3 are shown in Table 2. The decrease in tensions around Meeting Deadlines reached statistical significance, decreasing from 69.3% to 31.8% (z=−2.59, p=0.010). We observed a decrease in questions assessing tensions surrounding Authorship Credit, Authorship Order, Target Journal, and Timeliness, but these did not reach statistical significance. Mean days to paper submission (Question 2) decreased from 159 in 2017 to 116 in 2020—a decrease of 43 days from first draft to manuscript submission; however, this difference was not significant (t=1.28, p=0.208). Discussion We hypothesized that using an Authorship Agreement form would reduce authorship tensions for early-career researchers and lessen the time from first draft to manuscript submission. We found that providing a thoughtfully designed Authorship Agreement form to early-career researchers and their mentors, and then supporting its use, accompanied a statistically significant decline in tensions around meeting deadlines. While we observed decreases in other authorship tensions as well as time from first draft to manuscript submission, these did not rise to the level of statistical significance. The effect of an Authorship Agreement on authorship-related tensions experienced by early-career researchers, although predicted in the literature, (Bozeman and Youtie 2016; Roberts 2017; Youtie and Bozeman 2016; Bennett and Gadlin 2012; Primack, Cigliano, and Parsons 2014; Bennett, Gadlin, and Marchand 2018; Gadlin and Jessar 2002) has not to our knowledge been previously investigated. We credit the decline in tensions around meeting deadlines to the implementation of an Authorship Agreement, while acknowledging other concurrent ICRE efforts described in this article (for example, the Authorship Agreement Statement shown in Figure 2) may have influenced this finding. While our results do not allow us to make sweeping claims about the effectiveness of the Authorship Agreement, they do lead to some interesting insights that may be of value to others working to reduce authorship tensions and make authorship ethics a topic of open discussion at their institutions. For example, we now believe that simply providing the agreement form and asking early-career researchers to use it is not sufficient for some to overcome their reluctance to engage their senior colleagues in authorship conversations. While perhaps this should not have been surprising to us—it is, after all, a reluctance commonly cited in the literature (Seeman and House 2015; Youtie and Bozeman 2016; Primack, Cigliano, and Parsons 2014)—it made us realize that we had to actively avoid putting the burden of implementing a new process on the very people it was supposed to free from an uncomfortable task. Our response to this realization was the Authorship Agreement Statement that supported the use of the form. (Figure 2.) We believe a policy, or at the very least, an Authorship Agreement Statement such as the one now in place at the ICRE, is an advisable step, so that mentors and other senior colleagues expect to be approached with the form. (See Suggested Call-out Box 2: Summary of Recommendations for Leaders.) Though the prevalence of policies aimed at minimizing authorial misconduct has been explored, (Rasmussen et al. 2020) to our knowledge, the impact of a policy that strongly encourages documentation of authorship conversations has not been previously reported in the literature. We anticipate interviewing those who have submitted completed Authorship Agreements, to examine more closely how the policy and using the form affected authorship tensions. Though beyond the scope of the current study, these are topics we will explore in future work. Although we did not set out to determine how widespread the use of the Authorship Agreement would become, we have reason to believe it is gaining momentum, based on the number of completed agreements received, and requests for the form that author CAM is receiving from other departments at the University of Pittsburgh as well as other institutions. Interestingly, within the ICRE, we believe that authorship tensions may be declining even for those who have not tried using the form. We base this speculation on the work of other researchers who have found that trainees may derive benefits simply by training and working in an environment that promotes use of an Authorship Agreement, where leadership has embraced its use (Buljan, Barać, and Marušić 2018), and where authorship ethics is a topic of open discussion (Norman, Mayowski, and Fine 2020). This claim is supported by others who have shown that when leadership works to create an ethical research environment, and researchers believe they are working in an ethical environment, they respond by behaving ethically (Buljan, Barać, and Marušić 2018; Seeman and House 2015; Yeo-Teh and Tang 2020; NAS 2017; Martinson et al. 2006). This study does have limitations. It was conducted at one US research-intensive university, in a small translational science institute. This would not allow for randomization without a strong likelihood of contamination between the treatment (Authorship Agreement) and usual care (no Authorship Agreement) groups. Sample sizes were small for our surveys, which may have contributed to a lack of power to detect significance for several of our tests; however, our sample was limited by the size of our population (29 scholars and trainees at each time-point). A more complex statistical model that stratified or controlled for duration of program would have been ideal, but we were not sufficiently powered for those techniques. It is possible that other educational opportunities available to our trainees, for example, Responsible Conduct of Research workshops offered by the university’s Clinical and Translational Science Institute, or other programming (Norman, Mayowski, and Fine 2020) may have influenced their level of tension related to academic authorship. And, as previously mentioned, we believe the introduction of the Authorship Agreement Statement in and of itself may have contributed to an atmosphere of support for ethical authorship behavior, which could decrease tensions. Nevertheless, our work contributes to what is known about the prevalence of commonly cited authorship tensions experienced by early-career researchers, and provides evidence of the effectiveness of steps that can be taken to alleviate them. In conclusion, as we look ahead, we expect to see authorship-related tensions continue to decrease and eventually plateau as the research environment for those in ICRE training programs completes the shift to a place where these authorship conversations are safe to have and expected to happen (Norman, Mayowski, and Fine 2020). We currently are engaged in creating an online version of the Authorship Agreement form with extended capabilities which will allow us to more closely monitor usage, make the form easier to update and share, and introduce an archiving function, which will facilitate easy record-keeping and documentation of these conversations. Even given the very different nature of research across different disciplines (Youtie and Bozeman 2014; Marušić, Bošnjak, and Jerončić 2011; Moffatt 2018), our Authorship Agreement form and its supporting statement can likely be generalized across disciplines to help enable authorship conversations without fear of repercussion, contribute to openness, transparency, and fairness in research authorship (Cutas and Shaw 2015), and shorten the time to manuscript submission. Acknowledgments The authors would like to thank Megan Miller, M.Ed., for her contributions to Table 1. The authors would also like to thank Reviewer 3 for their help in describing how tensions between researchers might be expressed. Funding This study was funded by NCATS/NIH Grant UL1 TR001857-04. Figure 1. Authorship Agreement. Figure 2. Leadership Statement of Support for Completing the Authorship Agreement Figure 3. Survey. Question 1 was answered on a 5-point Likert scale with possible responses of Never; Rarely; Sometimes; Often; Very Often. Question 3 was answered on a 5-point Likert scale with possible responses of Significantly Delayed; Somewhat Delayed; On Time; Somewhat Ahead of Schedule; Significantly Ahead of Schedule. Table 1. Descriptive Demographic Characteristics of 2017-2020 ICRE Pre- and Post-Doc TL1 Trainees and KL2/PCORI K12 Trainees, n=65 Characteristics n (%) Program TL1 Pre-Doc TL1 Post-Doc KL2/PCORI K12 Total 65 (100%) 19 (29%) 22 (34%) 24 (37%) Female 39 (60%) 10 (53%) 14 (64%) 15 (63%) Race/Ethnicity  Asian 9 (14%) 3 (16%) 3 (14%) 3 (13%)  Black 6 (9%) - 2 (9%) 4 (17%)  White 42 (66%) 14 (74%) 15 (68%) 13 (54%)  Other* 8 (12%) 2 (10%) 2 (9%) 4 (17%) * includes Multiracial, Hispanic/Latinx, and those who chose not to respond Table 2. Proportion of participants reporting authorship tensions on 2017 and 2020 surveys. Proportion Reporting Tensions Source of Tension 2017 (n=26) 2020 (n=22) z p-value  Authorship Credit 9 (34.6) 7 (31.8) −0.20 0.838  Authorship Order 10 (38.4) 4 (18.2) −1.54 0.124  Target Journal 8 (30.8) 4 (18.2) −1.00 0.316 Roles/Responsibilities 7 (26.9) 7 (31.8) 0.37 0.710  Meeting Deadlines 18 (69.2) 7 (31.8) −2.59 * 0.010 *  Timely Manner 18 (69.2) 15 (68.2) −0.08 0.938 * Denotes p<0.05. Call-out Box 1: Six Criteria for the Agreement 1 Offers a simple design, so as not to impose an undue burden on authorship teams 2 Provides flexibility so as to be applicable to a wide range of authorship teams 3 Includes accepted authorship criteria to educate teams and build a shared vocabulary 4 Includes fields for designating both authorship credit and authorship order 5 Solicits target journal and intended submission date to promote collaborative goal-setting 6 Functions as a living, editable document, in recognition that circumstances chang Call-out Box 2: Summary of Recommendations for Leaders 1 Actively acknowledge the power dynamics at play in authorship conversation 2 Adopt an Authorship Agreement form such as the one described her 3 Publish a policy or statement in support of using the for 4 Disseminate the form to both mentors and mentee 5 Ensure authorship tensions and authorship ethics are topics of open discussio Declaration of interest statement The authors have no conflicts of interest to declare. References Bennett Dianne M , and Taylor David McD . 2003. “Unethical practices in authorship of scientific papers.” Emergency Medicine Australasia 15 (3 ):263–70. Bennett L Michelle , and Gadlin Howard . 2012. “Collaboration and team science: from theory to practice.” Journal of Investigative Medicine 60 (5 ):768–75.22525233 Bennett L Michelle , Gadlin Howard , and Marchand Christophe . 2018. Collaboration Team Science: Field Guide: US Department of Health & Human Services, National Institutes of Health…. Borenstein Jason , and Shamoo Adil E . 2015. “Rethinking authorship in the era of collaborative research.” Accountability in Research 22 (5 ):267–83.25928178 Bozeman Barry , and Youtie Jan . 2016. “Trouble in Paradise: Problems in Academic Research Co-authoring.” Science and Engineering Ethics 22 (6 ):1717–43. doi: 10.1007/s11948-015-9722-5.26573303 Buljan Ivan , Barać Lana , and Marušić Ana . 2018. “How researchers perceive research misconduct in biomedicine and how they would prevent it: A qualitative study in a small scientific community.” Accountability in Research 25 (4 ):220–38.29637796 COPE (Committee on Publication Ethics). “How to handle authorship disputes.” 2018. COPE, Accessed October 20, 2021. https://publicationethics.org/resources/guidelines-new/how-handle-authorship-disputesa-guide-new-researchers. Cutas Daniela , and Shaw David . 2015. “Writers blocked: On the wrongs of research co-authorship and some possible strategies for improvement.” Science and Engineering Ethics 21 (5 ):1315–29.25348894 Gadlin Howard , and Jessar Kevin . 2002. “Preempting discord: prenuptial agreements for scientists.” The NIH Catalyst 10 (3 ). Gaffey A . 2021. “Determining and Negotiating Authorship.” American Psychological Association, Accessed February 26. https://www.apa.org/science/about/psa/2015/06/determining-authorship. Heard Stephen B . 2016. The scientist’s guide to writing: How to write more easily and effectively throughout your scientific career: Princeton University Press. ICMJE (International Committee of Medical Journal Editors). 2021. “Recommendations for the Conduct, Reporting, Editing and Publication of Scholarly Work in Medical Journals.” Accessed October 20, 2021. http://www.icmje.org/recommendations/browse/roles-and-responsibilities/defining-the-role-of-authors-and-contributors.html ICRE (Institute for Clinical Research Education).”ICRE Home Page.” University of Pittsburgh, Institute for Clinical Research Education, Accessed February 25 2021. https://www.icre.pitt.edu/. Kwok Lance S . 2005. “The White Bull effect: abusive coauthorship and publication parasitism.” Journal of Medical Ethics 31 (9 ):554–6.16131560 Maggio Lauren A. , R. Anthony Jr Katherine Artino Picho , and W. Erik Driessen . 2018. “Are You Sure You Want to Do That? Fostering the Responsible Conduct of Medical Education Research.” Academic Medicine 93 (4 ):544–9.28678102 Martinson Brian C , Anderson Melissa S , Crain A Lauren , and De Vries Raymond . 2006. “Scientists’ perceptions of organizational justice and self-reported misbehaviors.” Journal of Empirical Research on Human Research Ethics 1 (1 ):51–66. Marušić Ana , Bošnjak Lana , and Jerončić Ana . 2011. “A systematic review of research on the meaning, ethics and practices of authorship across scholarly disciplines.” Plos One 6 (9 ):e23477.21931600 Moffatt Barton . 2011. “Responsible authorship: why researchers must forgo honorary authorship.” Accountability in Research 18 (2 ):76–90.21390872 Moffatt Barton . 2018. “Scientific authorship, pluralism, and practice.” Accountability in Research 25 (4 ):199–211.29400074 National Academies of Sciences, Engineering, and Medicine. 2017. Fostering integrity in research: National Academies Press. Norman Marie K , Mayowski Colleen A , and Fine Michael J . 2020. “Authorship stories panel discussion: Fostering ethical authorship by cultivating a growth mindset.” Accountability in Research:1–10. Phillippi Julia C , Likis Frances E , and Tilden Ellen L . 2018. “Authorship grids: Practical tools to facilitate collaboration and ethical publication.” Research in Nursing & Health 41 (2 ):195–208.29603766 Primack R , Cigliano J , and Parsons C . 2014. “Editorial: Co-authors gone bad—how to avoid publishing conflicts five examples of conflicts among co-authors.” Biological Conservation 176 :277–80. Rasmussen Lisa M , Williams Courtney E , Hausfeld Mary M , Banks George C , and Davis Bailey C . 2020. “Authorship Policies at US Doctoral Universities: A Review and Recommendations for Future Policies.” Science and Engineering Ethics:1–21.31123979 Rennie D , Yank V , and Emanuel L . 1997. “When authorship fails: A proposal to make contributors accountable.” JAMA 278 (7 ):579–85. doi: 10.1001/jama.1997.03550070071041.9268280 Resnik David B. , and Master Zubin . 2011. “Criteria for Authorship in Bioethics.” The American Journal of Bioethics 11 (10 ):17–21. doi: 10.1080/15265161.2011.603795. Roberts Laura Weiss . 2017. “Addressing Authorship Issues Prospectively: A Heuristic Approach.” Academic Medicine 92 (2 ):143–6. doi: 10.1097/acm.0000000000001285.27355782 Seeman Jeffrey I , and House Mark C . 2015. “Authorship issues and conflict in the US academic chemical community.” Accountability in Research 22 (6 ):346–83.26155731 Smith Elise , and Master Zubin . 2017. “Best practice to order authors in multi/interdisciplinary health sciences research publications.” Accountability in Research 24 (4 ):243–67.28128975 Smith Elise , and Williams-Jones Bryn . 2012. “Authorship and responsibility in health sciences research: a review of procedures for fairly allocating authorship in multi-author studies.” Science and Engineering Ethics 18 (2 ):199–212.21312000 Smith Elise , Williams-Jones Bryn , Master Zubin , Larivière Vincent , Sugimoto Cassidy R , Paul-Hus Adèle , Shi Min , Diller Elena , Caudle Katie , and Resnik David B . 2019a. “Researchers’ perceptions of ethical authorship distribution in collaborative research teams.” Science and Engineering Ethics 26 , no. 4 (2020): 1995–2022.31165383 Smith Elise , Williams-Jones Bryn , Master Zubin , Larivière Vincent , Sugimoto Cassidy R , Paul-Hus Adèle , Shi Min , and Resnik David B . 2019b. “Misconduct and misbehavior related to authorship disagreements in collaborative science.” Science and engineering ethics 26 , no. 4 (2020): 1967–1993.31161378 Smith Richard . 1997. “Misconduct in research: editors respond: the committee on publication ethics (COPE) is formed.” In.: British Medical Journal Publishing Group. StataCorp. 2016. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP. Accessed October 20, 2021. https://www.stata.com/support/faqs/resources/history-of-stata/ Tang Bor Luen . 2018. “Letter to the editor: On plurality and authorship in science.” Accountability in Research 25 (4 ):254–8.29587536 Uijtdehaage Sebastian , Mavis Brian , and Durning Steven J. . 2018. “Whose Paper Is It Anyway? Authorship Criteria According to Established Scholars in Health Professions Education.” Academic Medicine 93 (8 ):1171–5. doi: 10.1097/acm.0000000000002144.29384749 WAME (World Association of Medical Editors). “Recommendations on Publication Ethics Policies for Medical Journals.” 2016. Accessed February 14, 2021. https://wame.org/recommendations-on-publication-ethics-policies-for-medical-journals. Wislar Joseph S , Flanagin Annette , Fontanarosa Phil B , and DeAngelis Catherine D . 2011. “Honorary and ghost authorship in high impact biomedical journals: a cross sectional survey.” BMJ 343 :d6128.22028479 Yeo-Teh Nicole Shu Ling , and Tang Bor Luen . 2020. “Research ethics courses as a vaccination against a toxic research environment or culture.” Research Ethics:1747016120926686. Youtie Jan , and Bozeman Barry . 2014. “Social dynamics of research collaboration: Norms, practices, and ethical issues in determining co-authorship rights.” Scientometrics 101 (2 ):953–62. Youtie Jan , and Bozeman Barry . 2016. “Dueling Co-Authors: How Collaborators Create and Sometimes Solve Contributorship Conflicts.” Minerva 54 (4 ):375–97. doi: 10.1007/s11024-016-9303-z.

PMC009xxxxxx/PMC9246243.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0374660 4131 Hum Factors Hum Factors Human factors 0018-7208 1547-8181 34435529 9246243 10.1177/00187208211039102 NIHMS1816424 Article Head scanning behavior predicts hazard detection safety before entering an intersection Savage Steven W. 1* Zhang Lily 1 Swan Garrett 1 Bowers Alex R. 1 1 Schepens Eye Research Institute of Massachusetts Eye and Ear, Department of Ophthalmology, Harvard Medical School, Boston, MA * Corresponding author: Steven William Savage, Ph.D.; CPsychol., Steven.W.Savage@gmail.com 25 6 2022 8 2023 26 8 2021 01 8 2024 65 5 942955 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Objective: We conducted a driving simulator study to investigate scanning and hazard detection before entering an intersection. Background: Insufficient scanning has been suggested as a factor contributing to intersection crashes. However, little is known about the relative importance of the head and eye movement components of that scanning in peripheral hazard detection. Methods: Eleven older (mean 67 years) and 18 younger (mean 27 years) current drivers drove in a simulator while their head and eye movements were tracked. They completed two city drives (42 intersections per drive) with motorcycle hazards appearing at 16 four-way intersections per drive. Results: Older subjects missed more hazards (10.2% vs 5.2%). Failing to make a scan with a substantial head movement was the primary reason for missed hazards. When hazards were detected, older drivers had longer RTs (2.6s vs. 2.3s), but drove more slowly, thus safe response rates did not differ between the two groups (older 83%; younger 82%). Safe responses were associated with larger (28.8° vs. 20.6°) and more numerous (9.4 vs. 6.6) gaze scans. Scans containing a head movement were stronger predictors of safe responses than scans containing only eye movements. Conclusion: Our results highlight the importance of making large scans with a substantial head movement before entering an intersection. Eye-only scans played little role in detection and safe responses to peripheral hazards. Application: Driver training programs should address the importance of making large scans with a substantial head movement before entering an intersection. Précis: Larger and more numerous gaze scans before entering four-way intersections were associated with safer detections of peripheral motorcycle hazards approaching along the cross street. Scans containing a significant head component (larger scans) were a stronger predictor of detection likelihood and response safety than scans containing only eye movements (smaller scans). Ageing Simulated Driving Head and Eye Movements Hazard Detection pmc1 Introduction Early detection of hazards is crucial for executing a safe response when driving. This is particularly important at intersections where a large field of view must be scanned with eye and head movements in order to check for peripheral hazards approaching from both sides of the driver’s vehicle (Garber & Srinivasan, 1991; Ryan, Legge, & Rosman, 1998). This gaze scanning typically includes two types of scans: head+eye scans (scans containing both a substantial head and an eye movement component) and eye-only scans (scans containing no meaningful head component) (Savage, Zhang, Swan, & Bowers, 2020). Although insufficient scanning has often been suggested as a reason for increased crash risk (Bao & Boyle, 2009; Romoser & Fisher, 2009; Romoser, Pollatsek, Fisher, & Williams, 2013), relatively few studies have directly investigated the relationship between scanning and the detection of hazards at intersections (Bowers, Bronstad, Spano, Goldstein, & Peli, 2019; Yamani, Samuel, Roman Gerardino, & Fisher, 2016). Using the safe environment of a driving simulator, Yamani et al (2016) found that older drivers who made fewer scans toward potential hazards after entering the intersection were more likely to crash than older drivers who made more scans. More recently, Bowers et al (2019) examined the relationship between head scanning and the detection of pedestrians at intersections, again in a driving simulator. Head scan magnitudes of older drivers were 15° smaller when the target pedestrian was not detected as compared to when it was detected. Thus the results of these two prior driving simulator studies suggest that both scan numbers and scan magnitudes are important in safe detection of hazards. However, each study had limitations. Yamani et al (2016) did not analyze scan magnitudes and measured only eye position within the scene without recording head position. Bowers et al (2019) tracked only head movements (and not eye movements) and the target was a stationary pedestrian positioned 90° to the left or right of the intersection that presented no immediate danger. In the current study we addressed these limitations by recording head (head-in-world) and eye (eye-in-head) movements, characterizing scanning in terms of scan numbers and magnitudes, and using moving motorcycles as ecologically-valid hazards at intersections. Furthermore, the current study extended prior research by also evaluating the relative importance of head+eye scans and eye-only scans to hazard detection likelihood and response safety at intersections. From the age of 70 crash rates at intersections begin to increase (Bryer, 2000). An analysis of police accident reports suggested that over 50% of older drivers’ crashes at intersections were related to inadequate scanning behavior (Hakamies-Blomqvist, 1993). Consistent with these data, we previously reported that older drivers scanned less extensively on approach to intersections in a driving simulator (Savage et al., 2020). We did not, however, address the relationship between scanning behavior and detection of hazards. Three questions remain unanswered: 1) Does scanning behavior at an intersection predict detection likelihood and response safety for hazards at that same intersection?; 2) Does the reduced scanning behavior of older subjects mean that they are more likely to fail to detect hazards or have unsafe responses?; and 3) Is the contribution of the head or the eye components of scanning more important to detection and response safety? 2 Methods 2.2 Subjects We analyzed data for 29 subjects, including 11 older subjects (≥ 60 years) and 18 younger subjects (20 to 40 years). Characteristics of study subjects are shown in Table 1. They were the same subjects as those reported in Savage et al. (2020). They were all current drivers who met the vision requirements for driving in Massachusetts, had at least two years of driving experience and no adverse ocular history (self-reported) of eye disease that might affect visual acuity or visual fields. Both older and younger subjects had visual acuity (Test Chart 2000 PRO, Hertfordshire, UK) and contrast sensitivity (MARS chart, The Mars Perceptrix Corporation, Chappaqua, NY) within the typical normal range (Elliott, Yang, & Whitaker, 1995; Haymes, Roberts, Cruess, Nicolela, LeBlanc, Ramsey, Chauhan, & Artes, 2006). There was no statistical difference in the number of miles driven per year between the two age groups (W= 76.5; p= .12; Table 1). The study followed the tenets of the Declaration of Helsinki and was approved by the institutional review board at the Schepens Eye Research Institute. Informed consent was obtained from each subject. 2.3 Materials 2.3.1 Driving simulator and head and eye tracker The driving simulator (LE-1500, FAAC Corp, Ann Arbor, MI) consisted of five 42-inch LCD monitors yielding approximately 225° horizontal field of view. The central screen provided the view through the windshield, while the screens to the left and right provided the view through the side windows. The dashboard, including a speedometer, was displayed at the bottom of the central monitor. The controls and dashboard resembled a fully automatic transmission Ford Crown-Victoria along with a motion base seat with 3 degrees of freedom (Figure 1). Driving simulator data were collected at 30 Hz, including the position of the subject’s vehicle, its speed and heading, as well as information about other scripted vehicles in the virtual world. Head and eye movement data were recorded using a remote, digital 6-camera tracking system at 60 Hz (SmartEye Pro Version 6.1, Goteborg, Sweden, 2015 - Figure 1 yellow circles). The system tracked head and eye movements across approximately 180° (90° to the left and right), enabling us to capture the large scans that drivers made before entering an intersection. Matlab software (Mathworks, R2015a) was developed to post-process the merged simulator and head and eye data to generate plots and output variables used in the statistical analysis. 2.3.2 Driving scenarios A route with 42 intersections was selected. Events along the route were scripted using Scenario Toolbox software (version 3.9.4. 25873, FAAC Incorporated). The route was set in a light industrial virtual world including both city and residential areas with the roads on a grid system. The posted speed limit was 35 mph. The route included oncoming and cross traffic. Subjects drove the route twice, once guided by GPS instructions (“turn right/left at next intersection”) and once following a lead car since the method used to guide drivers through the virtual world could affect scanning behaviors and also detection performance. In this paper we report data for 16 of the 42 intersections. These 16 intersections had motorcycle hazards approaching along the cross-street from either the left or right side. They were all 4-way intersections with a range of traffic control devices, cross traffic, and intersection maneuvers (see Table 2 for details). They were selected to provide a variety of intersections as might be encountered in real world driving. In addition to the 16 motorcycles at intersections, there were seven motorcycle distractors, which did not appear at intersections, to avoid anticipation of motorcycles appearing only at intersections. Data for an additional 15 of the 42 intersections without any hazards were reported previously (Savage et al. 2020). All 16 intersection motorcycle events had the same design (Figure 2), which simulated a motorcyclist who was speeding through an intersection and failed to perceive the subject’s vehicle. The motorcyclist did not obey any control devices and did not follow normal priority rules. When the subject’s vehicle was at 30 meters from the intersection, the motorcycle was triggered to appear 60 meters away from the intersection along the cross street to the left (n = 8) or right (n = 8). The motorcycle then traveled at a constant speed of 45mph (or 20.1 m/s), well in excess of the 35 mph speed limit, until it entered the intersection, about 3 s after being triggered. These parameters were determined empirically based on measurements in a pilot study of subjects’ driving behaviors (speed, deceleration) when approaching intersections used for the experimental drives. They were selected to create a scenario in which the motorcycle represented a realistic hazard for a range of subjects’ intersection approach behaviors and deceleration profiles. However, to avoid any potential psychological stress caused by a collision between the motorcycle and the subject’s vehicle, the motorcycle was programmed to disappear just before entering the potential collision zone. Subjects were able to drive at whatever speed they were comfortable with up to a maximum of 35 mph (speed was capped at 35 mph or 15.7 m/s). If the subject’s vehicle was travelling at this maximum speed their vehicle would reach the intersection 1.8 s after triggering the motorcycle (or 1.2 s before the motorcycle). However, because subjects typically slowed down on approach to intersections, it took longer than this minimum time to reach the intersection and the motorcycle always represented a potential hazard. 2.4 Procedure The driving simulator session began with two acclimation drives. The first took place on a rural highway without other vehicles. The second was in the same city as the experimental drives but without other vehicles. For these two acclimation drives, subjects were given as much time (10-20 minutes) as they needed in order to become comfortable maneuvering the simulated vehicle. After the acclimation drives, we adjusted the six cameras of the SmartEye tracking system and calibrated the subject’s head position. Next, subjects completed a practice drive which included all the elements (motorcycle hazards, oncoming and cross traffic) of the experimental drives. During the practice drive the SmartEye system automatically built a profile of the subject and tracked their facial features. Eye position was calibrated after the practice drive by means of a five point calibration procedure on the center screen. After calibration, subjects completed two experimental drives (GPS and Lead Car; see Figure 3 for example screenshots) in counterbalanced order. Eye and head data were collected and analyzed for these two experimental drives. Each drive typically lasted between 10 and 15 minutes depending on the subjects’ driving speed. Subjects were instructed to drive as they normally would, follow all rules of the road and to press the horn as soon as they saw a motorcycle. Occasionally subjects forgot to press the horn and responded verbally to indicate detection of a motorcycle. Any such events were noted by the experimenter. If a subject stepped out of the simulator for a break, re-calibration of the SmartEye tracker was performed before the second experimental drive. In the GPS drive, auditory navigation instructions were delivered when the subject’s vehicle was approximately 70 meters from an intersection. In the Lead Car drive, subjects followed a car that was scripted to drive at 35 mph. The Lead Car made periodic stops to ensure that subjects did not lose sight of it. Subjects were instructed to drive as they normally would and follow all normal traffic rules. In the Lead Car drive they were instructed to follow the Lead Car at a safe distance, as they would when following a friend’s car in an unfamiliar city. Subjects were not, however, given any specific instructions about how to scan. 2.5 Quantifying detection performance and response safety Events where the subject responded verbally rather than by pressing the horn (1.9% of all motorcycle intersection events) were manually entered as seen and included in analyses of miss rates. However, they were not included in analyses of reaction times and response safety since there was no horn press response. 2.5.1 Miss rates Miss rates were the number of intersection motorcycles that were missed (i.e., events for which there was no horn-press response) expressed as a percentage of the total number of motorcycle intersection events. 2.5.2 Reaction times when a hazard was detected Reaction time was computed as the time between when the motorcycle first became visible and the horn-press response. 2.5.3 Safety of responses when hazard was detected Two metrics were used to quantify the safety of responses, each capturing a separate aspect of safety. For both metrics, the hypothetical collision point was used, given that the motorcycle disappeared prior to entering the collision zone. The first metric was the post-encroachment time (PET), which is defined as the time difference between the first road user leaving the collision zone and the second road user entering it (Allen, Shin, & Cooper, 1978). It represents the safety margin or the extent to which the two road users miss each other. The lower the PET, the more critical the situation. A PET of less than 1 second is considered a dangerous situation in real-world traffic conflicts (Hupfer, 1997; Kraay, van der Horst, & Oppe, 2013; Zangenehpour, Strauss, Miranda-Moreno, & Saunier, 2016; Johnson, Laureshyn & De Ceunynck, 2018), and was therefore classified as unsafe in the current study. The second metric was the deceleration-to-safety time (DST; (Hupfer, 1997)). This metric describes the minimum deceleration rate (in m/s2) needed to avoid a collision; i.e., to stop the subject’s car from entering the collision zone before the motorcycle leaves the collision zone. The higher the DST, the more critical the event. In real-world situations, a deceleration rate greater than 4 m/s2 is considered potentially dangerous (Hupfer, 1997), and was therefore classified as unsafe in the current study. Decelerating at a rate of 4 m/s2 could result in a driver either becoming a hazard to themselves (if the road conditions were wet or slippery) or to the driver in the vehicle behind them. The required deceleration was calculated using the following equation: Deceleration=2(Scar−Vcartmc)tcar2 Scar is the distance in meters between the location of the subject’s vehicle at the horn-press and the location where their vehicle will enter the collision zone. Vcar is the velocity of the subject’s vehicle in meters per second at the time of the horn-press. tcar is the time in seconds it would take the subject’s vehicle to travel from its current location (given their speed at the time of the horn press) to the point at which it enters the collision zone. tmc is the time in seconds it would take for the motorcycle hazard to exit the collision zone, thereby no longer being a hazard. In summary, when hazards were detected, the response was classified as unsafe if: 1) the subject’s vehicle entered the intersection before the motorcycle (and the motorcycle might have crashed into them); 2) the PET was less than 1 second; or 3) the DST was greater than 4 m/s2, or any combination of these situations. 2.6 Quantifying gaze scanning behavior Gaze scans were analyzed from 100 m before the intersection until the point at which the subject’s car first entered the intersection (0 m), defined as crossing the white line at controlled intersections or where the white line would have been at intersections without a control device. Thus, we included all scans made on approach to the intersection as well as any scans that were made when the subject’s vehicle was stationary and the subject was scanning for hazards before entering the intersection. We did not evaluate scans made after entering the intersection. We define a gaze scan as the whole series of lateral movements that take the eyes the furthest to the right or to the left from straight ahead (0°). Gaze scans could be composed of a single saccadic eye movement or multiple eye movements, either with or without an associated head movement (see Figure 4). Gaze scans were automatically detected using a custom algorithm; see Swan et al. (2021) for full details. In brief, the algorithm first marked saccades (velocity above 30 °/sec, greater than 1° eccentricity, and longer than 0.33s duration) and then merged sequential saccades that were headed in the same direction (i.e., towards the left or right), on the same side of the straight ahead gaze position (i.e., on the left or right side), and were within 400ms of each other. Merging saccades was necessary given that many gaze scans, especially large gaze scans, were composed of multiple saccades (see Figure 4 – right panel, G2; and Figure S1 in the supplement). Each gaze scan had a corresponding start and end eccentricity, with the difference being the scan magnitude. For each gaze scan, there was a corresponding head movement that was estimated by setting the start of the head movement as the start of the gaze scan and the end of the head movement as the local maximum of head eccentricity around the end of the marked gaze scan (Savage et al., 2020). Only gaze scans above 4° (more than four times the manufacturer’s accuracy under ideal conditions) headed away from straight ahead (0°) were used in analyses. Gaze scans were classified into two major categories (Savage et al., 2020): 1) scans which comprised predominately eye movement only (“eye-only” scans); and 2) scans which contained both a substantial head and eye movement component (“head+eye” scans). The classification was based on the magnitude of the head scan component of each gaze scan (Table 4), as implemented in prior driving simulator studies (Bowers et al., 2014; Bowers et al., 2019). The term “all-gaze” scans is used when data are pooled across eye-only and head+eye scans. 2.8 Statistical Analyses Prior to conducting statistical analyses we excluded single intersections on a per-subject basis where there was excessive noise in the gaze data for that intersection (removed ~7% of all intersections). Ultimately 856 intersections and 7538 scans were included in analyses. For the analyses of continuous numerical data, Linear Mixed Models (LMMs) were constructed in RStudio of the R statistical programming environment (Version 3.6.2 – R CoreTeam, 2019). Categorical outcome variables were analyzed by means of Generalized Linear Mixed Models (GLMMs). Our first set of analyses evaluated the effects of age (older vs. younger) on the three motorcycle detection measures (miss rates, reaction times and safe response rates). We created models in which we entered age (younger vs. older) as a fixed factor. Guidance type (GPS vs. Lead Car) was not included as a factor because preliminary analyses found no significant main effects of guidance type and no significant interactions with age (see supplementary materials). For all of our models we entered the unique event (intersection) number as a random factor to account for any variance contributed by the individual intersections as well as a random effects structure for subject to account for the variability contributed by individual differences. The second set of analyses evaluated whether scanning behavior was predictive of the safety of responses when hazards were detected using the following metrics: the magnitude and number of all-gaze scans, head+eye scans, and eye-only scans. We created a GLMM in which we entered the safety of responses (safe or unsafe) as the outcome variable and the magnitude or number of scans as the predictor variable. Age group (older vs. younger) was also added as a fixed factor to these models. Magnitudes of eye-only scans and all-gaze scans were normalized with a log2 transform prior to entering them into our models. All other continuous numerical outcome variables were roughly normally distributed. Outliers greater than 3 standard deviations from the mean (in transformed units) were removed. When reporting the data for normalized variables, we transformed them back to their raw unit format for ease of understanding. P-values for main effects were estimated by means of the lmerTest package (Kuznetsova, Brockhoff, & Christensen, 2017). P-values for any interactions between age and safety were calculated by model comparisons. We compared the simplest form of each model (with all interactions removed) with the same model plus the interaction of interest. The interaction model and baseline model were then compared using an analysis of variance (ANOVA), with the resulting p-value derived from our χ2 statistic representing the significance of the interactions of interest. 3 Results The main results are reported below. Additional analyses addressing detection types (fixational or peripheral), effects of guidance (lead car vs. GPS) and control device (stop sign vs. no sign) are provided in the supplementary materials. 3.1 Detection performance 3.1.1 Missed hazards Older subjects had significantly higher miss rates than younger subjects (10.2% vs. 5.2%; χ2(1)= 6.84; p= .009). 3.1.2. Detected hazards When hazards were detected, RTs were slower for older than younger subjects (means 2.63 s vs. 2.25 s; β= .−.33, SE= .14; t= −2.37; p= .027), and for unsafe than safe responses (2.7 s vs. 2.3 s; β= .18, SE= .08; t= 2.28; p= .023). There was no interaction between age and safety for the RT data, χ2(1,6)= 2.49; p= .11. In contrast, rates of safe responses did not differ between the two age groups (older 83% vs. younger 82%; β= .−.09, SE= .63; z= .15; p= .88). Analysis of speed at the time of horn press response revealed that both older and younger subjects drove slower when responses were considered safe (mean= 13.93 mph) as opposed to unsafe (mean= 25.47 mph), β= 12.6, SE= 1.6; t= 8.03; p< .001, and older subjects drove significantly slower (mean= 13.7 mph) than younger subjects (mean= 17.1 mph), β= 4.4, SE= 1.66; t= 2.65; p= .013. We also found a significant interaction between age and safety χ2(1,6)= 5.46; p= .02. This interaction came about as the difference between safe and unsafe responses was larger for younger (difference: 12 mph) than older drivers (10.5 mph). 3.2. Scanning behavior and missed hazards When the hazard was missed, subjects failed to make a head+eye scan toward the motorcycle in the majority (93%) of missed events but failed to make an eye-only scan in that direction for only a minority of missed events (15%). In other words, when a hazard was missed subjects frequently made eye-only scans toward the hazard but rarely made a head+eye scan. When expressed as a proportion of all events, rates of failing to make a head+eye scan and failing to detect the hazard were higher for older than younger subjects (9% vs. 5%; χ2(1) = 5.78; p= .016). 3.3. Scanning behavior and safety of responses for detected hazards 3.3.1 Number of scans Subjects made significantly more head+eye scans when the response was safe (3.8) as compared to unsafe (1.8), β= −.84, SE= .14; z= −6.07; p< .0001 (Figure 5, left panel), significantly more eye-only scans when the response was safe (5.6) as compared to unsafe (4.8), β= −.17, SE= .08; z= −2.25; p= .024 (Figure 5, middle panel), and significantly more all-gaze scans when their responses were safe (9.4) as compared to unsafe (6.6), β= −.36, SE= .062; z= −5.91; p< .0001 (Figure 5, right panel). We found no main effects of age for numbers of head+eye scans, β= −.016, SE= .15; z= .11; p= .91, eye-only scans, β= −.2, SE= .14; z= −1.43; p= .15, and all-gaze scans β= −.09, SE= .08; z= −1.5; p= .25. However, there was a significant interaction between safety and age for head+eye scans, χ2(1,6)= 4.08; p= .043. The difference in the number of scans between older and younger subjects was slightly greater when detections were unsafe (difference= .59 scans fewer for older subjects) than when they were safe (difference= .37 scans). For eye-only and all-gaze scans numbers, there were no significant interactions between age and safety (χ2(1,6)= .02; p= .89 , and χ2(1,6)= .71; p= .4, respectively). 3.3.2 Scan magnitudes The magnitude of head+eye scans was significantly larger when subjects’ responses were safe (53.7°) as compared to unsafe (46.7°), β= −10.03, SE= 2.91; t= −3.45 p< .001 (Figure 6, left panel). However, eye-only scan magnitudes did not differ between safe (10.7°) and unsafe responses (10.2°), β= .01, SE= .078 t= .13; p= .89 (Figure 6, middle panel). The magnitude of all-gaze scans was also larger when the response was safe (28.8°) than when it was unsafe (20.6°), β= −.28, SE= .06; t= −4.57; p< .001 (Figure 6, right panel). Older subjects made significantly smaller eye-only scans than younger subjects (9.6° vs. 11.3°); β= .023, SE= .08; t= 2.93; p= .007, and significantly smaller all-gaze scans (23.8° vs. 30°), β= −.15, SE= .07; t= −2.12; p= .04. For head+eye scan magnitudes, the main effect of age did not reach statistical significance β= 4.16, SE= 2.18; t= 1.91 p= .07. We found no interactions between age and safety for head+eye scan magnitudes, χ2(1,6)= 2.54; p= .11, eye-only scan magnitudes χ2(1,6)= .88; p= .35, and all-gaze scan magnitudes, χ2(1,6)= .49; p= .48. 4 Discussion 4.1 The effects of age on detection performance Both older and younger subjects detected the majority of motorcycles. However, older subjects missed a significantly higher percentage of motorcycles than younger subjects (10% vs. 5%). In the real world, any failure to detect a motorcyclist speeding through an intersection could result in the most serious of traffic conflicts, a collision. Thus the finding of a higher rate of missed hazards in the older group is concerning and is consistent with reports of higher rates of at-fault collisions in older drivers at intersections (Garber & Srinivasan, 1991; Pai, 2011; Preusser, Williams, Ferguson, Ulmer, & Weinstein, 1998). When motorcycles were detected, older subjects had longer reaction times than their younger counterparts, but the rate of safe responses did not differ between the two age groups because older subjects drove more slowly. Our calculations of response safety took account of the speed and distance of both road users at the time of the horn press. As such, driving more slowly increased the time window in which subjects were able to make a safe response. Older drivers have previously been reported to have overall slower driving speeds in simulated (Doroudgar, Chuang, Perry, Thomas, Bohnert, & Canedo, 2017; Zhang, Bowers, & Savage, 2020) and on-road driving (Porter, & Whitton, 2002; Horberry et al., 2004), as well as slower speeds when approaching intersections in simulated (Caird, Chisholm, Edwards, & Creaser, 2007) and on-road driving (Liu, 2007). However, even though safe responses rates of older drivers did not differ from those of younger drivers when motorcycles were detected, they still missed more motorcycles, which is the most critical kind of unsafe event with the greatest potential for adverse consequences in on-road driving. 4.2 The relationship between scanning behavior and missed hazards In the majority of events when a motorcycle was missed, there was no head+eye scan toward the motorcycle (on average 93% of missed detections), but there was usually at least one eye-only scan in that direction. These results clearly highlight the importance of making at least one large scan with a substantial head movement component in each direction before entering the intersection. Rates of failing to make a head+eye scan before entering the intersection and failing to detect the hazard were higher for older than younger subjects, especially at intersections without signage (see supplement S4.1), suggesting one reason for higher collision rates of older drivers in real-world driving. 4.3 The relationship between scanning behavior and response safety for detected hazards The number and magnitude of all-gaze scans were significant predictors of response safety when hazards were detected. Subjects made significantly more all-gaze scans (9.4 vs. 6.6) and significantly larger all-gaze scans (28.8° vs 20.6°) when responses were safe as compared to unsafe. Consistent with our prior study of scanning behaviors at intersections without motorcycle hazards (Savage et al., 2020), we found that older subjects made significantly smaller all-gaze scans than younger subjects but did not differ from younger subjects for the number of all-gaze scans. As a next step we wanted to determine whether head+eye or eye-only scans were more important for safe responses when the hazard was detected. Both older and younger subjects made more head+eye scans when responses were safe (3.8) as compared to unsafe (1.8), and the size of head+eye scans was greater when responses were safe (53.7°) as compared to unsafe (46.7°). Subjects also made more eye-only scans when responses were safe (5.6) as compared to unsafe (4.8). However, the magnitude of eye-only scans did not differ between safe (10.7°) and unsafe (10.2°) responses. The difference in eye-only scan numbers between safe and unsafe responses was less than 1 scan per intersection. Conversely, the differences between safe and unsafe responses for head+eye scan magnitudes (9.1°) and numbers (2.3 scans) were much greater and have both statistical and ecological significance. Thus our results strongly suggest that head+eye scans were more crucial than eye-only scans to response safety when motorcycles were detected. 4.4 Limitations Previous research has demonstrated that the risk of being involved in collisions at intersections starts to increase from the age of 70 years (Bryer, 2000). However, the average age of our older subjects was only 67 years. Thus we might have found greater age-related differences in safe response rates if we had included more subjects over 70 years of age. Nevertheless, the findings of higher miss rates and smaller all-gaze scans in older subjects are in the expected direction. In the current study we instructed our subjects to press the horn as soon as they detected a motorcycle. This task is conceptually quite different from a typical hazard perception task (Savage, Potter, & Tatler, 2013). Because the target in our current study was always the same, subjects most likely would have simply pressed the horn as soon as they saw a motorcycle, without needing to make a judgment about the level of hazard posed by the approaching motorcycle. In contrast to this, in a hazard perception task, detecting a hazard relies on knowledge of the rules of the road as well as contextual information which needs to be recalled and processed before a hazard judgment can be made. Therefore, the processing of potential hazards is most likely more cognitively demanding than was the case for our motorcycle detection task. The instructions to press the horn as soon as a motorcycle was detected may have resulted in more active visual scanning behaviors than would be the case in real-world driving. Moreover, subjects experienced a relatively high prevalence of motorcycles in each drive. Both these factors may have increased the likelihood and speed of detection (Beanland, Lenné & Underwood, 2014). Thus scanning deficits and detection deficits might be greater in on-road driving than in our driving simulator study. On the other hand, feelings of simulator discomfort (which were higher in the older than the younger group (Savage et al., 2020)) may have resulted in less active scanning than in real-world driving. Intersections in the real world contain many different types of hazards, which we did not test in our current experiment. Our paradigm focused on detection of a peripheral hazard which appeared at a relatively large eccentricity (about 60°). The results suggest that large head+eye scans are crucial for detection of this kind of hazard. However, other hazards may appear at smaller eccentricities e.g., an oncoming car which suddenly makes a turn without signaling. Eye-only scans may be important for detection of such hazards. The next step will be to apply our methodology to investigating the relationship between scanning and responses to hazards for different types of ecologically-valid hazards at intersections. 4.4 Conclusions and practical implications Older subjects failed to detect significantly more motorcycles than younger subjects, primarily because they did not make a head+eye scan in the direction of the hazard, which may place them at increased risk for collisions in real-world driving. When a motorcycle was detected, older subjects were slower to respond to the motorcycle but were not less safe as they drove slower than their younger counterparts. Our approach of analyzing head+eye and eye-only scans separately provides new insights into the relative importance of head and eye scanning in detection of peripheral hazards before entering four-way intersections as well as response safety when detected. In particular, failure to make a large head+eye scan in the direction of the motorcycle was the primary reason for missed detections, while making fewer head+eye scans and smaller head+eye scans were strongly associated with unsafe responses when motorcycles were detected. In contrast, eye-only scans played little role in the detection of peripheral hazards, though they may be important for detection of other kinds of hazards. It is well established that older persons have age-related reductions in neck flexibility that reduce maximum head rotation extent (Isler et al. 1997, Dukic and Broberg 2012, Chen et al. 2015). Reduced neck flexibility likely contributed to the finding that older subjects were more likely to fail to make a head+eye scan and made smaller head+eye scans than younger subjects. We previously reported that older subjects made more eye-only scans than younger subjects (Savage et al., 2020), which might be an attempt to compensate for making fewer head+eye scans. However, even if older drivers make more eye-only scans, the results of the current study suggest that eye-only scans are not sufficient for detection of peripheral hazards. Prior studies have demonstrated the effectiveness of training older drivers to make a scan to check for hazards after entering an intersection (Romoser & Fisher, 2009; Romoser, 2013). Our findings suggest that training programs for older drivers should also address the need to make at least one large head+eye scan in each direction before entering four-way intersections. Given limitations in neck flexibility, shoulder movements may be needed. The results also highlight that younger drivers may need training in the importance of head+eye scanning before entering an intersection. Designing and evaluating such training programs are topics for future studies. Supplementary Material Supplementary materials with figures Acknowledgements The authors would like to acknowledge Sarah Sheldon for assistance in driving simulator scenario development, and Sarah Sheldon and Dora Pepo for their help with recruiting and testing subjects. Funding Information: Funded in part by NIH grants R01-EY025677, S10-RR028122, and P30- EY003790. Steven W. Savage is a postdoctoral fellow at Schepens Eye Research Institute of Massachusetts Eye and Ear, Department of Ophthalmology, Harvard Medical School in Boston, MA, USA. He received his PhD in psychology from the University of Dundee, Scotland in 2015. Lily Zhang is a research assistant at Schepens Eye Research Institute of Massachusetts Eye and Ear in Boston, MA, USA. She received her MSc in mechanical engineering from the University of Minnesota in 2009. Garrett Swan is a postdoctoral fellow at Schepens Eye Research Institute of Massachusetts Eye and Ear, Department of Ophthalmology, Harvard Medical School in Boston, MA, USA. He received his PhD in cognitive psychology from the Pennsylvania State University in 2017. Alex R. Bowers, PhD is an associate scientist at Schepens Eye Research Institute of Massachusetts Eye and Ear and an associate professor in the Department of Ophthalmology, Harvard Medical School in Boston, MA, USA. She received her PhD in vision rehabilitation, from Glasgow Caledonian University, Scotland in 1998. Figure 1. FAAC driving simulator with SmartEye 6-camera head and eye tracking system (yellow circles) Figure 2. Schematic representation of motorcycle intersections (not drawn to scale). The red square represents the collision zone between the subject’s vehicle and the scripted motorcycle hazard. Figure 3. Examples of the scene on the center screen for each of the guidance methods within the light industrial city, taken from the point of view of the subject’s vehicle. Left panel shows the view when guided by GPS instructions. The right panel shows the view when following a lead car. Screenshots were taken from the center screen of the driving simulator Figure 4. Examples of lateral gaze (blue) and head (red) movements on approach to an intersection. The left panel serves as a reference, with the car’s travel path being left to right (i.e. increasing in time). Movements towards the left have a negative eccentricity and movements towards the right have a positive eccentricity. The right panel includes the gaze scans (green) detected using our custom algorithm. Gaze scan G1 is an eye-only scan comprising 1 eye saccade, G2 and G3 are eye-only with 2 eye saccades, G4, G5, and G6 are eye-only with one eye saccade, G7 is eye-only with 2 eye saccades, G8 is eye-only with one eye saccade, G9 is a head+eye scan comprised of a large lateral head rotation and eye saccade, and G10 is eye-only with one eye saccade. Figure 5: Average number of scans per intersection when motorcycles were detected for head+eye scans (left), eye-only scans (middle) and all-gaze scans (right). Data are split by safe and unsafe responses for older and younger subjects. Error bars represent the SEM Figure 6: Average scan magnitudes when motorcycles were detected for head+eye scans (left), eye-only scans (middle) and all-gaze scans (right). Data are split by safe and unsafe responses for older and younger subjects. Error bars represent the SEM Table 1. Characteristics of study subjects Factor Older (N=11) Younger (N=18) Age [years], mean (SD) 67.5 (6.7) 26.5 (5.9) Male [N] (%) 7 (64) 9 (50) Binocular visual acuity [logMAR1], mean (SD) 0.00 (0.07) 20/20 −0.07 (0.05) 20/17 Binocular contrast sensitivity [log], mean (SD) 1.70 (0.09) 1.78 (0.07) Annual mileage [miles], median 2860 1170 Driving experience [years], median 49 6 1 logMAR – Logarithm of the minimum angle of resolution Table 2. Inventory of all 4-Way Intersections with Motorcycle Hazards INT # Traffic control device on the subjects’ approach Other Traffic Intersection maneuver MC Side 1 Stop Car left Straight Right 2 Stop Van left / Car ahead Straight Right 3 Stop None Straight Right 4 Stop Car right Straight Left 5 None None Straight Left 6 None Car right Straight Left 7 None None Straight Left 8 Stop None Straight Right 9 Yield None Right Left 10 None Bus left / Bus ahead Straight Right 11 Traffic Light1 None Right Left 12 Traffic Light1 Bus left Right Left 13 None None Straight Right 14 None None left Left 15 Yield None Left Right 16 Stop Van left Straight Right 1. The traffic light turned to red when the subject’s vehicle was about 20 m from the stop line Table 4. Classification of gaze scans based on the size of the head movement component and the subject’s distance to the intersection. Distance to Intersection [m] Size of Head movement1 [°] Classification of Scan 100 – 50 ≥ 4 head+eye < 4 eye-only 50 – 20 ≥ 6 head+eye < 6 eye-only 20 – 0 ≥ 10 head+eye < 10 eye-only 1 Thresholds were the same as those used to define a head scan in prior driving simulator studies (Bowers et al., 2014; Bowers et al., 2019) 5. Key Points Older drivers missed more hazards than younger drivers Failing to make a large scan was the main reason for failing to detect hazards When motorcycles were detected, older drivers’ responses were not less safe because they drove more slowly Safe responses were associated with larger and more numerous gaze scans in both younger and older drivers Gaze scans with substantial head component were stronger predictor of response safety than eye only scans Disclosures: none 6. References Allen B , Shin T , & Cooper P (1978). Analysis of traffic conflicts and collisions. Transportation Research Record, 667 , 67–74. Bao S , & Boyle LN (2009). Age-related differences in visual scanning at median-divided highway intersections in rural areas. Accident Analysis & Prevention, 41 (1 ), 146–152.19114149 Beanland V , Lenne MG , & Underwood G (2014). Safety in numbers: Target prevalence affects the detection of vehicles during simulated driving. Attention Perception & Psychophysics, 76 (3 ), 805–813. Bowers AR , Ananyev E , Mandel AJ , Goldstein RB , & Peli E (2014). Driving with hemianopia IV: Head scanning and detection at intersections in a simulator. Investigative Ophthalmology & Visual Science, 55 (3 ), 1540–1548.24474265 Bowers AR , Bronstad PM , Spano LP , Goldstein RB , & Peli E (2019). The effects of age and central field loss on head scanning and detection at intersections. Translational Vision Science & Technology, 8 (5 ), 14. Bryer T (2000). Characteristics of motor vehicle crashes related to aging. . In Schaie KW & Pietrucha M (Eds.), Mobility and Transportation in the Elderly (pp. 157–206). New York: Springer Publishing Company, Inc. Caird JK , Chisholm SL , Edwards CJ , & Creaser JI (2007). The effect of yellow light onset time on older and younger drivers' perception response time (PRT) and intersection behavior. Transportation Research Part F-Traffic Psychology and Behaviour, 10 (5 ), 383–396. Chen KB , Xu X , Lin J-H , & Radwin RG (2015). Evaluation of older driver head functional range of motion using portable immersive virtual reality. Experimental Gerontology, 70 , 150–156.26315289 Doroudgar S , Chuang HM , Perry PJ , Thomas K , Bohnert K , & Canedo J (2017). Driving performance comparing older versus younger drivers. Traffic Injury Prevention, 18 (1 ), 41–46.27326512 Dukic T , & Broberg T (2012). Older drivers’ visual search behaviour at intersections. Transportation Research Part F: Traffic Psychology and Behaviour, 15 (4 ), 462–470. XElliott DB , Yang KCH , & Whitaker D (1995). Visual acuity changes throughout adulthood in normal, healthy eyes: seeing beyond 6/6. Optometry and Vision Science, 72 (3 ), 186–191.7609941 Garber NJ , & Srinivasan R (1991). Characteristics of accidents involving elderly drivers at intersections. Transportation Research Record (1325 ), 8–16. Hakamies-Blomqvist LE (1993). Fatal accidents of older drivers. Accident Analysis & Prevention, 25 (1 ), 19–27.8420531 Haymes SA , Roberts KF , Cruess AF , Nicolela MT , LeBlanc RP , Ramsey MS , Chauhan BC , Artes PH (2006). The letter contrast sensitivity test: clinical evaluation of a new design. Investigative Ophthalmology and Visual Science, 47 (6 ), 2739–45.16723494 Horberry T , Hartley L , Gobetti K , Walker F , Johnson B , Gersbach S , & Ludlow J (2004). Speed choice by drivers: The issue of driving too slowly. Ergonomics, 47 (14 ), 1561–1570.15697070 Hupfer C (1997). Deceleration to Safety Time (DST) - a Useful Figure to Evaluate Traffic Safety. Paper presented at the 1997 ICTCT Conference Proceedings of Seminar 3, Lund, Sweden. Isler RB , Parsonson BS , & Hansson GJ (1997). Age related effects of restricted head movements on the useful field of view of drivers. Accident Analysis & Prevention, 29 (6 ), 793–801.9370015 Johnsson C , Laureshyn A , & Tim De Ceunynck T (2018) In search of surrogate safety indicators for vulnerable road users: a review of surrogate safety indicators. Transport Reviews, 38 (6 ), 765–785 Kraay JH , van der Horst ARA , & Oppe S (2013). Manual conflict observation technique DOCTOR: Dutch Objective Conflict Technique for Operation and Research (No. 2013-1). Voorburg. Kuznetsova A , Brockhoff PB , & Christensen RHB (2017). lmerTest Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software, 82 (13 ), 26. Liu BS (2007). Association of intersection approach speed with driver characteristics, vehicle type and traffic conditions comparing urban and suburban areas. Accident Analysis and Prevention, 39 (2 ), 216–223. doi:10.1016/j.aap.2006.07.005 16950148 Pai C-W (2011). Motorcycle right-of-way accidents—A literature review. Accident Analysis & Prevention, 43 (3 ), 971–982.21376890 Porter MM , & Whitton MJ (2002). Assessment of driving with the global positioning system and video technology in young, middle-aged, and older drivers. Journals of Gerontology Series A-Biological Sciences and Medical Sciences, 57 (9 ), M578–M582.12196494 Preusser DF , Williams AF , Ferguson SA , Ulmer RG , & Weinstein HB (1998). Fatal crash risk for older drivers at intersections. Accident Analysis & Prevention, 30 (2 ), 151–159.9450119 R Core Team (2019). R: A language and environment for statistical computing. Vienna, Austria.: R Foundation for Statistical Computing. Retrieved from https://www.R-project.org/ Romoser MRE (2013). The long-term effects of active training strategies on improving older drivers' scanning in intersections: A two-year follow-up to Romoser and Fisher (2009). Human Factors, 55 (2 ), 278–284 23691824 Romoser MRE , & Fisher DL (2009). The effect of active versus passive training strategies on improving older drivers’ scanning in intersections. Human Factors, 51 (5 ), 652–668.20196291 Romoser MRE , Pollatsek A , Fisher DL , & Williams CC (2013). Comparing the glance patterns of older versus younger experienced drivers: Scanning for hazards while approaching and entering the intersection. Transportation Research Part F: Traffic Psychology and Behaviour, 16 , 104–116.23148130 Ryan GA , Legge N , & Rosman D (1998). Age related changes in drivers' crash risk and crash type. Accident Analysis & Prevention, 30 (3 ), 379–387.9663297 Savage SW , Potter DD , & Tatler BW (2013). Does preoccupation impair hazard perception? A simultaneous EEG and eye tracking study. Transportation Research Part F: Traffic Psychology and Behaviour, 17 , 52–62. Savage SW , Zhang L , Swan G , & Bowers AR (2020). The effects of age on the contributions of head and eye movements to scanning behavior at intersections. Transportation Research Part F: Traffic Psychology and Behaviour 73 , 128–142 32742187 Swan G , Goldstein RB , Savage SW , Zhang L , Ahmadi A , & Bowers AR (2021). Automatic processing of gaze movements to quantify gaze scanning behaviors in a driving simulator. Behaviour Research Methods 53 (2 ), 487–506 Yamani Y , Samuel S , Roman Gerardino L , & Fisher DL (2016). Extending analysis of older drivers’ scanning patterns at intersections. Transportation Research Record, 2602 (1 ), 10–15. Zhang T , Bowers AR , & Savage SW (2020). The effects of age, distraction, and simulated central vision impairment on hazard detection in a driving simulator. Optometry and Vision Science; 97 (4 ), 239–248.32304533 Zangenehpour S , Strauss J , Miranda-Moreno L , & Saunier N (2016). Are signalized intersections with cycle tracks safer? A case-control study based on automated surrogate safety analysis using video data. Accident Analysis and Prevention 86 , 161–172.26562673

PMC009xxxxxx/PMC9308671.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8809320 1600 Neuron Neuron Neuron 0896-6273 1097-4199 35397211 9308671 10.1016/j.neuron.2022.03.020 NIHMS1797223 Article Prelimbic cortex drives discrimination of non-aversion via amygdala somatostatin interneurons Stujenske Joseph M. 118* O’Neill Pia-Kelsey 23418 Fernandes-Henriques Carolina 56 Nahmoud Itzick 7 Goldburg Samantha R. 8 Singh Ashna 1 Diaz Laritza 9 Labkovich Margarita 10 Hardin William 11 Bolkan Scott S. 12 Reardon Thomas R. 13 Spellman Timothy J. 14 Salzman C. Daniel 2341516 Gordon Joshua A. 17 Liston Conor 1 Likhtik Ekaterina 5619* 1 Department of Psychiatry, Weill Cornell Medical College, New York, NY 10065, USA 2 Department of Psychiatry, Columbia University, New York, NY 10027, USA 3 The Mortimer B. Zuckerman Mind, Brain, and Behavior Institute, Columbia University, New York, NY 10027, USA 4 Department of Neuroscience, Columbia University, New York, NY 10027, USA 5 Biology Program, The Graduate Center, City University of New York, New York, NY 10016, USA 6 Department of Biological Sciences, Hunter College, City University of New York, New York, NY 10065, USA 7 Wayne State University School of Medicine, Detroit, MI 48201, USA 8 Department of Ophthalmology, Northwell Health, Great Neck, NY 11021, USA 9 Charles E. Schmidt College of Medicine, Boca Raton, FL 33431, USA 10 Icahn School of Medicine at Mount Sinai, New York, NY 10029, USA 11 Amazon Games, San Diego, CA 92127, USA 12 Princeton Neuroscience Institute, Princeton, New Jersey 08540, USA 13 Meta Reality Labs, Redmond, WA 98052, USA 14 Department of Neuroscience, UConn School of Medicine, Farmington, CT 06032, USA 15 Kavli Institute for Brain Sciences, Columbia University, New York, NY 10027, USA 16 New York State Psychiatric Institute, 1051 Riverside Drive, New York, NY 10032, USA 17 National Institute of Mental Health, Bethesda, MD 20892, USA 18 The authors contributed equally 19 Lead contact AUTHOR CONTRIBUTIONS Conceptualization, J.M.S., J.A.G., and E.L.; methodology, J.M.S., T.J.S., and E.L.; software, J.M.S.; formal analysis and data curation, J.M.S., P.-K.O., I.N., A.S., M.L., and E.L.; writing, J.M.S. and E.L.; visualization, J.M.S., P.-K.O., and E.L.; project administration, J.M.S. and E.L.; investigation, J.M.S., P.-K.O., C.F.-H., I.N., S.R.G., A.S., L.D., W.H., S.S.B., and E.L.; resources, T.R.R. and T.J.S.; supervision, J.M.S., C.D.S., J.A.G., C.L., and E.L. * Correspondence: jmstujenske@gmail.com (J.M.S.), elikhtik@genectr.hunter.cuny.edu (E.L.) 25 4 2022 20 7 2022 08 4 2022 20 7 2023 110 14 22582267.e11 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY The amygdala and prelimbic cortex (PL) communicate during fear discrimination retrieval, but how they co-ordinate discrimination of a non-threatening stimulus is unknown. Here, we show that somatostatin (SOM) interneurons in the basolateral amygdala (BLA) become active specifically during learned non-threatening cues and desynchronize cell firing by blocking phase reset of theta oscillations during the safe cue. Furthermore, we show that SOM activation and desynchronization of the BLA is PL-dependent and promotes discrimination of non-threat. Thus, fear discrimination engages PL-dependent coordination of BLA SOM responses to non-threatening stimuli. In brief Overgeneralization of fear to non-aversive stimuli is observed in several psychiatric disorders. Using a mouse model, Stujenske et al. identify the dorsomedial prefrontal cortex acting via somatostatin interneurons in the basolateral amygdala as the mechanism that desynchronizes sensory activation of the amygdala during non-aversive stimuli, thereby suppressing generalized fear responses. Graphical Abstract pmcINTRODUCTION Generalized fear is a cardinal symptom in disorders of trauma and anxiety and is associated with a hyperactive amygdala and a hypoactive medial prefrontal cortex (mPFC) (Ghosh and Chattarji, 2015; Hayes et al., 2012; Milad et al., 2009). Neural signatures of fear discrimination are present in the basolateral amygdala (BLA) (Grewe et al., 2017; Grosso et al., 2018; Sangha et al., 2013), and mice with dysfunctional BLA inhibitory circuits generalize fear (Bergado-Acosta et al., 2008; Shaban et al., 2006). During classical fear conditioning, the BLA parvalbumin-positive (PV) and somatostatin-positive (SOM) interneurons (INs) play opposing roles (Krabbe et al., 2019; Wolff et al., 2014), promoting and impeding fear acquisition, respectively. However, the role of BLA INs during discrimination of non-threatening cues is unknown. Furthermore, the question remains whether discrimination of non-threat is an actively maintained or a passive process. Recent work showed that during conditioning, BLA principal neurons (PNs) remap to encode fear-conditioned stimuli, whereas neutral stimuli remain passively maintained (Grewe et al., 2017). However, it was also shown that discrimination learning recruits specific PNs and INs that prevent generalization (Grosso et al., 2018), suggesting that generalization is actively prevented. Given that BLA SOMs are posited to dampen defensive freezing and gate BLA PN potentiation (Ito et al., 2020), we sought to investigate how BLA SOMs participate in the retrieval of learned non-threat and interact with the mPFC, which is critical for fear discrimination (Likhtik et al., 2014). RESULTS BLA SOM INs selectively upregulate during the CS We trained mice (n = 18) on a 3-day auditory differential fear conditioning (DFC) paradigm (Likhtik et al., 2014) (Figure 1A).The CS+ was paired with a shock, and the CS− was explicitly un-paired (tones of 2 or 8 kHz, 50 ms pips at 1 Hz for 30 s). After con-ditioning, mice were exposed to either CS+ (n = 9) or CS− (n = 9) in a new context to quantify BLA SOMs and PV c-fos co-labeling (Figures 1A–1B). SOM INs were significantly more active during CS− than CS+ retrieval (Figures 1C and 1D). PV INs exhibited a smaller, non-significant difference (Figures 1C and 1D). Total c-fos expression also did not differ (Figure S1A), consistent with our prior electrophysiology findings (Likhtik et al., 2014; Stujenske et al., 2014). Fiber photometry in SOM-cre (n = 4) and PV-cre (n = 8) mice revealed that BLA SOM INs exhibited significantly increased activity during the CS−, whereas there was a weaker, non-significant trend toward a difference in PV INs (Figures 1E, S1B, and S1C). Individual SOM IN responses during DFC were further characterized with two-photon calcium imaging (Figures 1F, 1G, and S1D–S1F), first during stimulus habituation and then after DFC, during retrieval (n = 378 total cells from seven mice). SOM INs weakly responded to both stimuli during habituation (Figure 1H; n = 322). During retrieval, only the CS− induced a strong response (Figures 1H–1J; n = 306). The next day, all mice discriminated while freely moving (Figure 1K). SOM INs responded preferentially to the CS− with short latency (Figures 1L and S1G). 54% of SOM INs (n = 306) were significantly activated by the CS− (with 20% also active to the CS+), compared with 22% of SOM INs activated by the CS+ (with 91% also responsive to the CS−; Figure 1M). We found that the CS− response was largely homogeneous (81% of variance explained by a single principal component; Figures S1H–S1K). The computational relevance of the SOM IN response was confirmed using a support vector machine (SVM) linear classifier, which demonstrated high decoding accuracy of stimulus identity, both in individual mice and across mice (Figures 1N and S1I–S1M). BLA SOM INs mediate CS− discrimination by reducing BLA theta synchrony Next, we tested BLA SOM and PV function during discrimination retrieval. SOM− Cre and PV-Cre mice were injected with cre-dependent viruses expressing the inhibitory opsin, eArch3.0, or eYFP, and optic fibers were implanted above the BLA (Figures 2A–2C and S2). During DFC retrieval, light was paired with half of the CS trials (Figures 2B, 2C, and S2A). SOM inhibition selectively enhanced freezing to the CS− in eArch mice (Figure 2D; n = 8) but not in controls (Figure 2A; n = 11). Inhibition decreased the discrimination score (see STAR Methods) by 62.1% ± 11.3% (Figure 2E). PV IN inhibition did not have a significant effect on discrimination (Figure 2E; n = 8), although there was variability between mice (Figure S2A). Given that SOM INs mediate distal dendritic inhibition of BLA PNs (Muller et al., 2007), we tested whether SOM INs gate incoming sensory input, which reliably induces fear-related synchronous changes in the BLA-mPFC circuit (Ito et al., 2020; Krabbe et al., 2019). We therefore recorded local field potentials (LFPs) in the BLA and prelimbic cortex (PL) with and without BLA SOM silencing. As previously reported (Likhtik et al., 2014), during retrieval, the CS+ evoked higher power and phase reset of the ongoing theta oscillations in BLA than the CS− (Figures 2F–2I and S2; n = 13), indicating synchronized population neuronal activation during the CS+. In SOM-Arch mice (n = 6) but not in controls (n = 7), SOM inhibition induced a theta phase reset to the CS− with similar temporal dynamics to the CS+, without a significant effect during CS+ (Figures 2H–2J and S2). Notably, the effect of silencing on the theta reset was highly specific, as light did not alter BLA or PL theta power (Figure 2K), BLA-PL theta coherence, or the PL-to-BLA direction of information transfer during the CS− (Figure S2), which followed expected patterns of activity (Likhtik et al., 2014). We then recorded BLA single units during fear retrieval in another set of SOM-Arch mice (n = 179 single units from n = 5 mice; Figure 2L). Putative SOM INs were optogenetically identified (n = 11), and we found that CS− pips evoked a time-locked response, while CS+ pips did not (Figure 2M; CS− versus CS+ average Z score difference of 2.90 ± 0.85, p = 0.0004, paired t test). Putative PV INs (n = 14), defined based on previously described electrophysiological characteristics (Wolff et al., 2014), exhibited time-locked firing after both CS+ and CS− pips (Figure 2M), with no difference in response magnitude (CS− versus CS+ Z score difference of −0.67 ± 0.82, p = 0.42, paired t test) but a non-significant trend for the response dynamics (time × CS interaction, p = 0.06; Figure 2L). Thus, PV INs are engaged in the discrimination process but likely in a more nuanced way than SOM INs. SOM IN silencing did not change the pip-evoked firing pattern of putative PV INs (Figure 2M) but induced a significantly higher firing rate (FR) in over 50% (Figures 2N and S2). Putative PNs showed mixed changes in firing when SOM INs were silenced during the CSs (Figure 2N), such that there was no significant net change in population firing. These findings suggest that when SOM INs are suppressed, average PN firing is stabilized–likely by increased PV IN activity. Lastly, we studied theta-frequency modulation of firing in putative PNs (Figure 2O; STAR Methods). We reasoned that if individual cells exhibit theta resets, this would be visible as a frequency component in the average pip-evoked firing. During light off, the average pip-evoked firing of PNs showed a stronger theta (6–10 Hz) component during the CS+ than CS− (Figure 2O and 2P), but when SOM INs were inhibited during the CS−, there was a drastic increase in the theta component, analogous to LFP theta reset (Figures 2H and 2I). Putative PV INs exhibited a similar pattern, with a non-significant trend toward the same effect of light as PNs (Figures 2O and 2P). These data suggest that BLA SOM INs selectively gate incoming sensory input to PNs during the CS−, specifically preventing theta reset. PL upregulates BLA SOM activity during the CS Given that the PL partakes in discrimination learning (Lee and Choi, 2012; Meyer and Bucci, 2014) and PL-BLA communication contributes to DFC retrieval (Burgos-Robles et al., 2017; Klavir et al., 2013; Likhtik et al., 2014; Stujenske et al., 2014; Taub et al., 2018), we hypothesized that the PL modulates BLA SOM IN activity during retrieval. We first verified that the PL is necessary for discrimination. Mice trained on DFC (n = 10) were infused in the PL with saline or the GABA-A receptor agonist muscimol on two days of retrieval (counterbalanced). Muscimol, but not saline, led to dramatic fear generalization to the CS−, without an effect on CS+ freezing (Figures 3A and 3B). In the same mice as in Figure 1, we imaged SOM INs during DFC retrieval before and after muscimol or saline infusion in the PL (Figure 3C). PL silencing selectively impaired activation of BLA SOM INs during the CS− in all mice (Figures 3D, 3E, and S3, n = 117 neurons from n = 3 mice), whereas saline had no effect (Figure S3, n = 127 cells from n = 3 mice). Again, PL silencing did not significantly alter SOM IN CS+ responses (Figures 3D, 3E, and S3). The CS− activated 58% of SOM INs pre-muscimol but only 19% post-muscimol (Figures 3F1 and S3), whereas saline had no effect (Figure 3F2). An SVM classifier confirmed that cue encoding was stable post-saline but changed dramatically post-muscimol (Figure 3G). Using rabies tracing, we confirmed that at least one pathway coordinating SOM IN activity consists of monosynaptic connections from mPFC (Figure S3E). PL inputs to the BLA drive CS− discrimination To investigate the functional relevance of the mPFC-driven activity in the BLA, we utilized 129S6/SvEvTac mice, which are pre-disposed to fear generalization (Figure 4A). These mice exhibit a bimodal distribution in their ability to discriminate, and they were classified as discriminators or generalizers (Figure 4B; see STAR Methods). Generalizing mice exhibit impaired safety-related directional mPFC (PL and IL)-to-BLA synchrony and inappropriately high theta resetting during safe stimuli (Likhtik et al., 2014) (summarized in Figure 4C). To confirm that mPFC input to the BLA was necessary for discrimination, we expressed the inhibitory opsin eArch3.0 or eYFP in the mPFC and placed optic fibers with microelectrodes in the BLA (Figure 4D). In mice that discriminated the two stimuli well during light-off trials, there was a lower BLA theta reset during the CS− than the CS+ (n = 9 LFPs from n = 7 mice; Figures 4E, 4F, and S4). Inhibition of mPFC input to the BLA resulted in increased theta reset in the BLA, specifically during the CS− (Figures 4E and 4F), mimicking effects of SOM IN silencing (Figure S4). Behaviorally, discrimination was impaired by mPFC terminal silencing (Figure 4G; n = 8 Arch, n = 5 eYFP), although this was not due to a specific effect on the CS− (Figure S4). Last, we tested whether enhancing PL input to the BLA was sufficient to improve discrimination. We expressed either ChR2 or mCherry in the PL and trained 129S6/SvEvTac mice on DFC (Figure 4H). We then stimulated terminals in the BLA with 6 Flz oscillatory illumination on half of CS+ and CS− cues during retrieval (Figure 4K), simulating endogenous mPFC theta-frequency activity (Likhtik et al., 2014; Stujenske et al., 2014). In ChR2-expressing mice that generalized during light-off trials (n = 8), stimulation of PL terminals in the BLA selectively decreased CS− freezing, resulting in a significant increase in the discrimination score (Figures 4I–4K and S4), whereas controls exhibited no change (n = 6). DISCUSSION Our data support the view that SOM INs filter the synchronous drive of BLA pyramidal cells (Ito et al., 2020) by virtue of mediating dendritic inhibition at synapses in close proximity to excitatory inputs (Muller et al., 2007). We highlight the active role of the PL in preventing generalized fear via PL inputs to BLA, driving SOM IN-mediated suppression of fear to non-aversive cues (Figure S4I). Whereas previous work has shown the role of the PL in supporting fear expression (Cummings and Clem, 2020; Likhtik et al., 2014; Meyer and Bucci, 2014; Sierra-Mercado et al., 2011; Vieira et al., 2015), here we demonstrate a role for PL fear suppression during discrimination of non-threat, similar to the role of the anterior cingulate in non-human primates during reversal learning (Klavir et al., 2013). Nevertheless, PL input to the BLA drives excitation of PN cells via AMPA receptor-mediated currents (Arruda-Carvalho and Clem, 2014; Arruda-Carvalho et al., 2017), supporting fear expression (Arruda-Carvalho and Clem, 2014; Likhtik et al., 2005). Thus, our data suggest that the PL serves opposing functions under different conditions. PL input can mediate direct feed-forward activation of both BLA PNs and INs and makes many indirect connections with both cell types (Likhtik et al., 2005; Marek et al., 2018; Rosenkranz and Grace, 2001). In vitro, inhibiting BLA SOM IN creates windows for enhanced plasticity of PL-to-BLA excitatory input (Ito et al., 2020). Thus, PL- dependent SOM IN activation may serve an autoregulatory function. It is likely that mPFC-dependent activation of BLA SOM INs is a critical circuit motif regulating opposing changes in defensive behavior. BLA PV INs engage in upregulation and downregulation of fear during acquisition and extinction, respectively (Davis et al., 2017; Trouche et al., 2013; Wolff et al., 2014). Interestingly, we show that during retrieval, CS+ and CS− both activate putative BLA PV but only the CS+ induces time-locked theta-frequency firing (Figures 2K and 2M), consistent with their known role in CS+ processing. This suggests a complex, timing-dependent role for BLA PV INs in CS processing, which could explain why constant optogenetic inhibition of these cells did not affect discrimination. Given their known connectivity (Wolff et al., 2014), PV IN silencing would also likely disinhibit SOM INs, potentially masking their role in controlling PNs. Disturbance of mPFC-to-BLA functional connectivity, with heightened theta phase resetting to the CS−, is associated with generalization. We demonstrate that the mPFC-to-BLA projection is necessary and sufficient for discrimination. Thus, it is likely that the presence of PL-dependent BLASOM IN activation underlies individual variation in the discrimination of mice. These findings are highly relevant to understanding the circuit basis of overgeneralization in individuals with a variety of psychiatric disorders. STAR★METHODS RESOURCE AVAILABILITY Lead contact Further information and requests for data or other materials should be directed to the lead contact, Ekaterina Likhtik (elikhtik@genectr.hunter.cuny.edu). Materials availability The study did not generate any unique reagents. Data and code availability All data reported in this paper will be shared by the lead contact upon request. Original code utilized in this paper has been deposited at Zenodo and github and is publicly available as of the date of publication (see the key resources table). Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. EXPERIMENTAL MODEL AND SUBJECT DETAILS All experiments were performed on three to five-month-old male mice. All mice were C57/B6J mice (The Jackson Laboratory), except for mPFC terminal activation and inactivation experiments, which were in 129S6/SvEvTac (Taconic Biosciences). Wildtype mice were acquired directly from The Jackson Laboratory. PV-IRES-cre (RRID:IMSR_JAX:017320) and SOM-IRES-cre (RRID:IMSR_JAX:018973) were maintained on a C57/B6J background and mice used for experiments were heterozygous, produced from a homozygous to wildtype cross. C57/B6J mice were group housed throughout all completed experiments. 129S6/SvEvTac were singly housed, with enrichment (nestlets), starting after surgical procedures. Littermates were randomly assigned to experimental and control groups. When group housed, they were maintained in the same cage. Sample sizes reported include only mice with verified, accurate placements and/or viral expression in the relevant brain regions; excluded mice are noted in each applicable section of the methods. All procedures were conducted in accordance with National Institutes of Health regulations and approved by the Columbia University and New York State Psychiatric Institute Institutional Animal Care and Use Committees, the Hunter College Institutional Animal Care and Use Committee, and the Weill Cornell Medical College Institutional Animal Care and Use Committee. METHOD DETAILS Differential fear conditioning Mice were first habituated to handling, optical fiber and/or electronic interface board connection, and, where applicable, fiber optic light illumination in a small rectangular box (30 × 20 cm) in the dark for twenty minutes. 465 nm (activation experiments) or 532 nm (inactivation experiments) light was directed through their fiber optic for two minutes, followed by two minutes of no light, for twenty minutes. 532nm light was generated by DPSS lasers (3% power stability, OEM lasers) and 465nm light was generated by fiber-coupled LED modules (Plexon). On a subsequent day, mice were habituated in dim light (30 lux) to the fear conditioning chamber (Med Associates, 15.24 × 13.34 × 12.7 cm) with a grid floor of stainless-steel bars for shock delivery, and exposed to two separate tones: 2kHz and 8 kHz pure tones (5 times each, pseudorandom order, presented as thirty 50 ms pips, amplitude modulated with 25ms linear increase followed by 25ms linear decrease, once per second). Mice were then trained as previously described (Likhtik et al., 2014; Stujenske et al., 2014) over the next three days to associate one tone (CS+) with a co-terminating shock (0.4 mA, 1s), whereas the other tone (CS−) was explicitly unpaired. This shock intensity was specifically chosen to avoid ceiling effects in evaluation of freezing to the CS+ and CS−. We also tested 0.6 mA and 0.9 mA shock intensities, which induced CS+ freezing of approximately 90%. Thus, we utilized a higher shock intensity for our c-fos experiments to assure a robust response in our mice (see below), as a ceiling effect was not a problem in that experiment. CS identity was counterbalanced between mice, with variable inter-trial-intervals (ITI) of 60,80,100, 120 seconds. On the fourth day, mice that would receive optogenetic manipulation were again habituated to optogenetic light by five 30 second light exposures, alternating with 30 seconds of no light. The mice were then placed in a new context (a wooden enclosure, 60 Lux) for testing how well they had learned the differential associations with the two stimuli. They were habituated to the new chamber for 3 minutes prior to the first tone presentation. Stereotactic viral injections Mice were anesthetized with isoflurane (0.5–1.5%) and head-fixed in a surgical stereotactic apparatus (Kopf). Viral injections were made with a pressure injector (Quintessential Stereotactic Injector, Stoetling) connected to a 10 μL Hamilton Syringe terminating in a glass pipette with fine tapered end (~30 gauge). Temperature was monitored and maintained at 36.6°C with a feedback-regulated heating pad. The skull was leveled using bregma and lambda landmarks and craniotomies were made using anterior-posterior (AP) coordinates from bregma, medo-lateral (ML) coordinates from the midline and dorso-ventral coordinates (DV) from brain surface. For PL terminal excitation experiments, 3-4 month-old 129S6/SvEvTac mice were injected with 150 nL of adeno-associated virus (AAV) expressing either channelrhodopsin (AAV5-CamKIIa-hChR2-H134R-mCherry-WPRE-pA) or a control virus (AAV5-CamKIIa-H134R-mCherry-WPRE-pA; UNC Vector Core) at two locations bilaterally to drive expression in the PL: ML ±0.3mm; AP +1.75mm; DV −1.1 and −1.2 mm (relative to brain surface). All injections were made by first lowering to 0.05 mm deeper than the intended target, and then raising the needle to create a “pocket,” minimizing backflow up the pipette. The mice were then implanted with bilateral ferrule-coupled fiber optic implants (200 μm core, 0.39 NA) above the amygdala basal nuclei (ML: +/−3.5, AP: −1.6, DV: −3.8). To allow for viral expression at the PL axon terminals in the BLA, behavioral testing was performed 6-7 weeks after the initial viral injection. Mice in this experiment were all singly housed, with bedding for enrichment, to be consistent with our previous behavioral approach (Likhtik et al., 2014), yielding discriminating and generalizing mice. For mPFC terminal inactivation experiments, the preparation was the same, except 500 nL of either AAV5-CamKIIa-eArchaerhodopsin-eYFP or control virus (AAV5-CamKIIa-eYFP; UNC Vector Core) was injected at each ML ±0.3mm; AP +1.75mm; DV −1.6 and −2.0 mm (relative to brain surface). Behavioral testing was performed 12 weeks after the viral injection. For silencing genetically defined interneuron populations, 3-5 month old male C57/B6 mice expressing cre-recombinase in either SOM+ cells (SOM-IRES-cre) (Taniguchi et al., 2011) or PV+ cells (PV-IRES-cre) (Hippenmeyer et al., 2005) were utilized. Mice were stereotactically injected in the BLA (ML 3.3, AP −1.7, DV, 4.3) with 0.5 mL of an AAV that expressed the gene for Arch in a cre-dependent manner (AAV5-Ef1a-DIO-eArch3.0-eYFP) or a control virus (AAV5-Ef1a-DIO-eYFP; UNC Vector Core). Mice were group housed, with cagemates divided equally into Arch and control groups. During the same surgery, bilateral ferrule-coupled fiber optic implants (200 umcore, 0.39 NA) were inserted over the basolateral nuclei (ML +/−3.3, AP −1.7, DV −3.9). A subset of mice had tungsten electrodes affixed to the fiber optic directed to the BLA (electrode tip at DV 4.3) and electrodes implanted in the PL (ML −0.3 mm; AP +1.8 mm; DV −1.6). Behavioral testing began 4 weeks post-surgery to allow for expression of virally-delivered proteins. For imaging SOM+ interneurons in the basolateral nuclei, a midline incision was made in the scalp, and 1 μL of AAV5-hsyn-FLEX-GCaMP6s (Penn Vector Core, Addgene) was injected unilaterally in the left BLA (ML: −3.3, AP: −1.55, DV: −4.9 to −4.5) of SOM-cre mice. This volume of injection was used to assure robust expression and spread, and as a result, expression in the central amygdala could not be avoided and was tolerated as these cells were far from the subsequent field of imaging. The midline incision was subsequently closed with VetBond Adhesive (3M). For fiber photometry recordings, 500 nL of AAV2/5-hsyn-FLEX-GCaMP6s (Addgene) was injected into the left BLA(ML: 3.3, AP: −1.55, DV: −4.7) of SOM-cre or PV-cre mice, and they were implanted with a unilateral optic fiber (400 um core, 0.37 NA) at the same coordinate. Note that there are differences in the PL and BLA coordinates between 129S6/SvEvTac and C57/B6 background mice due to slight anatomical differences. In all cases, the experimenters were blinded to viral identity at the time of viral injection (when relevant), and to mouse identity until after the experiments were complete. For mice with optical fibers, surgical screws were placed in the bone over the frontal cortex and cerebellum as anchors for implanted components, and the skull was covered with dental cement (Dentsply or Parkell). Perioperatively, mice were given analgesics (Carpofen, 5 mg/kg, s.c. or Meloxicam 2 mg/kg, s.c.) and monitored for comfort and weight gain. Fiber photometry recordings Fluorescence signals were recorded using the Neurophotometrics FP3002, which uses a high-resolution camera (FLIR) to capture fluorescence signal exiting the tip an optic fiber. Data acquisition consisted of alternating pulses of fiber-coupled ultraviolet (405 nm) and blue (470 nm) LEDs, acquired at an effective rate of 20 Hz for each channel, capturing an isosbestic (control) trace and a calcium-dependent trace. The calcium-dependent trace was corrected by the control channel as follows (function detrend_photometry.m). The isosbestic channel was fit with the sum of two exponentials to capture signal decay due to bleaching (Matlab curve fitting toolbox), yielding its overall trend and a detrended isosbestic trace. We then determined the best linear fit of the isosbestic trend to a running baseline of the calcium-dependent trace (determined by applying a median filter of length 5s followed by a minmax filter of 100s width). Light artifacts were removed by subtracting a zero-meaned and rescaled isosbestic trace. The corrected signal was then detrended by dividing by the isosbestic trend and multiplying by the trend’s initial value. We confirmed that light artifacts were small relative to the true signal and that similar results were obtained if the isosbestic trace was not utilized. 2-3 weeks after viral injection and fiber optic implantation, mice were connected to the recording equipment to confirm presence of fluorescence signals. Two SOM-cre mice were excluded from this experiment, one because of lack of signal and the other because of incorrect fiber placement. No PV-cre mice were excluded. Two wildtype mice were injected with the same cre-dependent virus to confirm that there was a lack of fluorescence signal in the absence of the calcium indicator. After confirming expression, mice were habituated to the tones and trained on three days of DFC, as described above, with a few changes. The mice were fear conditioned in a rat-size chamber (Med Associates, 29.53 × 24.84 × 18.67 cm), with one wall transparent for camera recording of the mouse behavior. The chamber was within an enclosure covered on all walls with gray sound-dampening material, with the exception of a small hole where the camera was placed and a hole in the ceiling, around which a white LED ring was placed for illumination and through which the optic fiber was able to pass. Retrieval was performed in the same chamber, except the grid floor was covered with a white plastic sheet, a curved white insert was placed in the back of the chamber to change the context, and a black and white checkerboard pattern was placed on the wall of the enclosure visible to the mouse. Mice were re-exposed to the tones in the absence of any shock, while the fiber photometry signal was captured. The fiber photometry signal was then analyzed by calculating the percentage change for the signal during each 30-second tone relative to the preceding 30-seconds and averaging over 3 repetitions (which we called ΔF/F). For plotting, the traces were smoothed with a boxcar filter (width of 2s). 2-photon Imaging One week after viral injections, mice were anesthetized with isoflurane, and a 0.85mm microendoscope probe (Grintech GmBH) was stereotactically implanted in the BLA positioned at the same coordinates as the BLA injection. Brain tissue was aspirated to a depth of ~2mm below skull surface to reduce intracranial pressure from the implant. We lowered the implant to above the amygdala basal nuceli (DV −4.75 mm, relative to skull surface at bregma) and fixed it in place using a small amount of VetBond adhesive (3M) followed by a layer of dental cement (Dentsply). A custom designed metal head post was attached with dental cement caudal to the microendoscope to be used for head fixation during imaging. 26-gauge dual barrel cannulae (PlasticsOne) were also implanted over the PL (AP: +1.65 ML: ±0.3 DV:−1.1, allowing for infusion at DV −1.6 after insertion of longer internal syringe) for bilateral drug delivery. The cannulae were secured to the skull using dental cement and small screws for stability. We then applied Kwik-cast (World Precision Instruments) to the GRIN lens to protect it from damage between imaging sessions. Mice recovered for two weeks and were habituated to the recording context and head fixation. After habituation we checked the level of GCaMP6s expression using the two-photon microscope. If expression was sufficiently bright we commenced experiments. All images were acquired using a multi-photon microscopy system (Prairie Technologies) with a 10x objective (0.3 NA, Nikon) and an ultra-fast pulsed laser beam (920-nm wavelength, Coherent). We continuously acquired GCaMP fluorescence with a photomultiplier tube (GaAsP PMT) operated by PrairieView software. Mice were habituated to the optical instrumentation for five days. In the first two days, mice were allowed to walk freely on a Styrofoam ball. On subsequent days, mice were securely head fixed on the ball setup with a pair of clamps designed to match the implanted metal head post. Each mouse received two imaging sessions, one prior to fear conditioning (Habituation) and one the day following three days of discriminative fear conditioning (Retrieval). Fear discrimination training occurred as in other experiments. On each day, about 1 hour after fear conditioning, mice also received a head-fixed habituation session on the Styrofoam ball for 15 minutes. During habituation and testing imaging sessions, mice were head fixed and time series were collected at 128 × 256 pixels covering a 500 × 500um field of view, cropped to include only areas with active cells, yielding a recording rate between 6.8 and 8.8 Hz. Calcium imaging data was synchronized with tone presentation using an Arduino and custom-written Python software. Three CS− tones were followed by three CS+ tones, with variable inter-trial-intervals (ITIs) of 90, 110, 130, or 150 seconds. Then either muscimol or saline was infused bilaterally into the PL(1 μgm/μl, 0.5 μl per hemisphere, infused at a rate of 0.1 μl/min). Twenty minutes later, the mice were retested on another set of three CS− and CS+ trials (same variable ITI schedule). Movements of the styrofoam ball were simultaneously recorded with a rotary encoder (US Digital MA3-A10-125-B), which was sensitive enough to detect movements smaller than a single step of a mouse. The starts of bouts of movement were marked by deflections of the ball on the rotary encoder and stops were marked when no deflections were detected for at least 1 second. The amount of movement was calculated as the sum of the lengths of all movement bouts. The day after the final imaging session, mice were placed in a behavioral chamber for retrieval in a freely moving condition (retrieval occurred in a different context from acquisition; for half the mice this was a wooden enclosure, for the other half this was a plastic enclosure of the same dimensions). Mice were habituated to the chamber for three minutes and then presented with three CS− tones and three CS+ tones, as during imaging. Subsequently, mice were infused with DiI in the PL to determine the affected area of infusion. Twenty minutes later, mice were deeply anesthetized and perfused to determine cannula and GRIN lens placement as well as level of viral expression. 1 mouse was included in habituation-recall analysis (Figure 1) but not in the infusion experiment due to localization of infusion cannulae over the IL, rather than the PL. Three mice were excluded from analysis due to GRIN lens misplacement (two placements over the lateral nuclei, one placement in endopiriform cortex). Two mice were excluded due to lack of GCaMP signal. Calcium imaging time-series were motion corrected using a planar hidden Markov model (SIMA) (Kaifosh et al., 2014) and in the event of poor performance of SIMA, NoRMCorre (Pnevmatikakis and Giovannucci, 2017) was used as backup. Regions of interest (ROIs) were manually drawn over motion corrected time-series in ImageJ (NIH) with the aid of maximum, average, and standard deviation projections, as well as the full image stack. Data were then imported to MATLAB for further analysis. All calcium signals were investigated for the presence of calcium transients, and any signals without transients in any time epoch were excluded. First, calcium signals were denoised using an L0-constrained non-negative deconvolution with 1.8 second time constant implemented using deconvolution_standalone.m in the Suite2p analysis package (Pachitariu M., 2017) to reduce variability between mice prior to combining data, though findings were similar without denoising. To standardize across mice for the purpose of analyses at different time points, data was interpolated to generate a 10Hz signal using nearest neighbor interpolation (the main results were tested and confirmed without interpolation). While mice did not move for a majority of the time on testing day, there were occasionally sustained bouts of running, which induced substantial variability in the SOM IN activity, likely due to a relationship between their activity and movement. As we were specifically interested in the activity of these interneurons as they related to stimulus processing rather than movement, any tones in which mice exhibited more than 2 seconds of movement were excluded (with this criterion, this amounted to at most 1 of each tone excluded per mouse per recording session). Tone-evoked intracellular calcium signals were normalized to the pre-tone calcium signal by subtracting the average pre-tone calcium signal in the preceding 15 seconds and dividing by the average activity in the preceding 50 seconds of each tone presentation (to get a more robust baseline and account for any time-variant changes in fluorescence over the long recording session), excluding any periods of movement, leading to a ΔF/F output. Tone-evoked responses were quantified as the average area under the curve of each recorded cell for the 30s that the tone was on. Acquired data from habituation and recall sessions were aligned using non-rigid image registration (Matlab function, image_registration, from publicly available matlab suite “B-spline Grid, Image and Point based Registration”). Some cells were only recorded during habituation or recall, while the majority (66%) of cells were detected on both days. Rather than excluding from analysis any cells that were not recorded on both days, we accounted for a mixture of paired and unpaired data in the construction of our rm-ANOVA test and post-hoc testing when analyzing changes from habituation to fear recall. The main results were confirmed by analyzing the subset of cells that were present on both days, with the same pattern of changes to the CS+ and CS− following fear conditioning. For the majority of analyses presented in the main figures, cells were combined together across mice, but we separately confirmed that the main effects were reproduced in each individual mouse by analyzing each population of recorded cells separately (Figures 1H–1M and S1). Analysis of significant tone responses To determine which cells were significantly responsive to the CS− and CS+, a shift predictor was generated by repeating the same procedure for generating ΔF/F values, but using only time periods in which no tones were being presented and randomly defining three time points as tone onsets, defining the subsequent 30s as a pseudo-tone presentation and the prior 50s as a pseudo-pretone for normalization. The three results were averaged. This was repeated 2000 times to generate an expected distribution for every time point before and after tone onset. This method was chosen rather than a z-score approach, as the use of a defined period for ΔF/F normalization introduced a slight bias for the value to increase as time moves away for the zeroing period (into either the past or future), which was accounted for by this randomization procedure. ΔF/F responses to the CS+ and CS− were averaged across tones for the 30s of tone presentation and compared to the distribution generated by averaging the 2000 iterations of the shift predictor. A cell was defined as significantly increased during a tone if its ΔF/F value exceeded the 97.5th percentile of the shift predictor or significantly decreased during a tone if the value was below the 2.5th percentile. As the percentage of cells with a significant decrease was at chance (Figures 1L and 1M), the possibility of inactivation responses to the tone was subsequently excluded and we focused exclusively on activated cells. To the contrary, during fear conditioning, it was previously found that BLA SOM IN predominantly exhibit a small decrease in FR in response to CSs when recorded electrophysiologically (Wolff et al., 2014). Consistent with this, we did see a trend for such an effect in our fiber photometry data, which is pooled data across many cells, though it was highly variable between mice. Our calcium imaging technique is likely more reliable at detecting cells with increased activity, as GCaMP6 indicators have been previously demonstrated to be unreliable at detecting drops in FR (Hara-Kuge et al., 2018), particularly in cells with low baseline firing and subtle drops in FR (Ali and Kwan, 2020). The ~20% of SOM IN that we found to be significantly activated to the CS+ (Figure 1m) is similar to the ~15% of cells that were found to be activated by CSs during fear conditioning electrophysiologically (Wolff et al., 2014). For the analysis of significance at each time point (Figure 1L), a cell was considered significantly activated by a tone for each time point if its value at a time point exceeded the 97.5th percentile of the shift predictor distribution for that time point. This same procedure was used to evaluate when a unit was significantly more active to the CS+ vs the CS−, except that in the absence of enough CS data to generate a full distribution, we used the shift predictor 95% confidence interval as a proxy for the expected spread of data. A unit was considered to be significantly more active to the CS+ compared to the CS−, or vice versa, at a time point (or for the average of the 30s of tone presentation) if their values differed by greater than the 95% confidence interval of the shift predictor. Investigating patterns of SOM IN responses To investigate the diversity of patterns of SOM IN response to tone presentation, we utilized a principal component decomposition of the tone-evoked response (Figure S1). PCA1 accounted for the great majority of the variance of the response to the CS− and to a lesser extent for the CS+, owing to the fact that CS+ evoked responses were weaker. PCA of only units with significant responses to the CS+ yielded similar results as for the CS−. The first PCA component was highly correlated with the average DF/F tone responses, so we investigated whether other PCA components may be more revealing of diverse SOM responses. In investigating PCA2-5 (each of which accounted for >1% of the variance, at least for the CS−), we found patterns of activity that reflected some diversity at the level of individual cells. We saw that the PCA2 component looked like a faster latency response, so we reasoned that this may be a reasonable proxy for response latency. We thereby classified cells as type 1 if they had a negative PCA2 component (i.e., they looked dissimilar to PCA2) or as type 2 if they had a positive PCA2 component. We investigated the latency of response in type 1 and type 2 cells by using the time at which 95% of the maximum ΔF/F value was reached. Our findings indicated that overall, despite some minor heterogeneity, there was a largely homogenous response of SOM BLA cells to the CS+ and CS−, and for this reason, we treated the cells as a roughly homogenous population for our analysis rather than investigating differential responses in different cell populations. Tone identity decoding by SOM+ interneurons Tone-evoked calcium signals from all cells in each group of mice was calculated, as described. For most analyses, cells from the mice were combined into one cell population, but the decoding properties of cells from individual mice was also performed to confirm consistent results (Figure S1). Tone-evoked calcium responses were divided into 30 one second blocks, providing for 60-90 data points per tone per cell (depending on whether tones were excluded due to movement in an individual mouse). We iteratively subsampled our population of cells 1000 times for each of a variety of cell population size (2-150 cells). A support vector machine (Matlab function, fitcsvm) was trained for binary classification of CS+ vs CS− on a subset of the data from standardized inputs (centered and divided by standard deviation for each cell) with uniform prior (indicating equal likelihood of CS+ or CS−) and linear kernel function and then tested on the left-out data (Matlab function, predict). For the n=1 datapoint, we established the binary threshold that best separated an individual cell’s response to the CS+ and CS− on training data and used that to classify test data. For investigating the coding properties of cells in the recall session before any infusion, all 30 one second samples from an individual tone were left out at a time, while the SVM was trained on the samples from remaining tones for each subsampled population. The accuracy of the decoding of each CS across the entire 30s, for each iteration, was calculated as the average number of correctly identified tones (defined as >50% of the 30 samples being defined correctly as CS+ or CS−; Figure S1). Alternatively, for the analysis of decoding accuracy over time, the accuracy of the decoding at each time point after tone onset (Figure 1N) was merely calculated as the percentage of tones correctly identified for that individual time sample. For both analyses, results were then averaged across all 1000 subsamples at each population size and presented as the mean and 95% confidence interval (calculated by the lower 2.5th and upper 2.5th percentile). For the time analysis, a shuffle was generated by repeating the above procedure but randomly labeling tones as CS+ and CS− in five unique simulations (none of which were the two permutations which would correspond to the correct binary classification). For investigating the change in coding properties of cells before and after saline or muscimol infusion (Figure 3G), the SVM was trained on the samples from all tones before infusion and then tested on samples from each tone before and after infusion. Accuracy of decoding was defined separately for the before and after infusion sessions, as above. This approach likely over-estimates the decoding reliability of session 1, given that no session 1 data was excluded, but the decoding reliability within this session was already separately analyzed with cross-validation (see above; Figure S1), and this analysis was interrogating session 1 to session 2 changes between saline and muscimol mice, rather than determining the unbiased session 1 decoding accuracy. Fear discrimination training and optogenetics All optogenetic manipulations were performed only during fear retrieval in the absence of shock. For IN silencing experiments, C57B6 mice were exposed to six CS+ and six CS− tones (pseudorandom order), half with 532 nm (10 mW) optical illumination of the BLA and half with no illumination (n=3 trials each). For PFC and PL terminal inhibition and activation experiments, the structure was the same except, 129S6/SvEvTac mice were used and exposed to ten CS+ and ten CS− tones, half paired with 465nm (4-5mW, for ChR2) or 532nm (10mW, for Arch) illumination of the BLA (n=5 trials each). All SOM-cre and PV-cre mice included for optogenetic experiments had expression in the basolateral amygdala, with a fiber optic placed over the BLA, and no expression of opsins in the central amygdala, which has a strong SOM presence (Li et al., 2013). One SOM-cre mouse was excluded from behavioral analysis due to beingageneralizer (CS+, CS− difference < 10%) prior to manipulation. All mice utilized for PFC terminal manipulation experiments exhibited expression within the injected part of the prefrontal cortex, had visualizable prefrontal terminal fields in the BLA, and had fiber optics placed over the BLA. Two mice were excluded due to lack of expression or incorrect fiber optic placements. For the PL terminal activation experiments, we were interested in identifying a cohort of fear generalizing mice, and we therefore used 129S6/SvEvTac mice, a group that has a higher proportion of generalizers, instead of C57B6 mice, which has a higher proportion of discriminators (see Animals [above], Defining Generalization… [below], Figure 4A), due to the overall larger proportion of these mice that generalizer fear. We have previously found that on average, 129S6/SvEvTac mice extinguish fear to the CS+ and CS− slower than C57B6 mice, and therefore to decrease statistical variability from mouse to mouse, more CS+ (5 Light on and 5 Light Off), and CS− (5 Light On and 5 Light Off) presentations were used in the terminal manipulation experiments. For the experiments where we tested the role of BLA SOM or PVIN in discrimination retrieval, we used C57B6 mice, which extinguish faster, and thus we used fewer CS+ (3 Light on and 3 Light Off), and CS− (3 Light On and 3 Light Off) presentations. For Arch terminal inhibition experiments, despite our having run 10 CS+ and 10 CS− trials, we nevertheless only used the first 3 of each to calculate discrimination changes between Light On and Light Off trials, making them equivalent with the discrimination changes that we calculated in Figure 2 for the C57B6 mice. This is done for two reasons. First, we wanted to compare the same number of inactivation trials between mouse strains. Second, the mPFC terminal inactivation experiment was performed in Discriminators, which we found to extinguish faster as a group than when we average all the SvEv129s together, because generalizing mice extinguish slower. Thus, behaviorally, during tones 13-20, mice were no longer reacting differently to the CS+ and CS−, which was the difference that we intended to manipulate. In the terminal activation experiments in generalizing mice (Figures 4H–4K), we used all 5 trials of each type to calculate discrimination Light On vs Light Off as mice continued to freeze and generalize throughout. In all experiments, illumination turned on five seconds before tone presentation, continued for the entire 30 sec tone presentation period, and turned off five seconds after the tone ended. This was done so as to prevent light artifact at the onset of tones, and prevent any effects from possible rebound excitation at light offset, rather than the end of the CS. Any light-induced changes were controlled for by light-only controls. During input excitation experiments, the 465nm light was presented as a 6Hz sine wave,delivered by one PlexBright Compact LED Module per side (Plexon), mounted on a Plexon magnetic commutator, and controlled using a PlexBright LD-1 Single Channel LED driver. For inhibition experiments, 532nm coherent light was generated continuously by a laser (OEM lasers, 3% power stability) and divided bilaterally using a Doric optical commutator (FRJ_1x2i_FC-2FC). All light was delivered through 200 μm diameter patch cords terminated in ceramic ferrules (Thorlabs). The ITIs remained the same during retrieval as during training. With our electrophysiological recording technique, we found that the sinusoidal fiber optic illumination induced a larger artifact in our LFP, even in non-opsin expressing mice, that prevented proper data collection. For this reason, electrophysiology was not performed during this experiment. For behavioral analysis, freezing to the two tones was quantified offline by an observer blinded to the identity of the tones and the mice. Freezing was also quantified for the 30 seconds prior to each tone onset and averaged to calculate ITI freezing. Mice were considered to be freezing at times when there was complete immobility with the exception of breathing or a periodic, stereotyped shuttering movement (time-locked to the pips). We defined a priori that we would investigate the freezing of mice to the CS+ and CS−, with and without light, as well as discrimination between CS+ and CS−. To control for multiple comparisons, we first used repeated measure ANOVA to analyze the effect of CS (CS+ vs CS−), light (on vs off), and CS x light interaction for each experimental group. If a significant effect was found for CS x light interaction, we used Bonferroni-corrected paired t-tests for post-hoc testing for all four conditions. Defining generalization in 129S6/SvEvTac mice Mice were defined as generalizers if the absolute difference in freezing between CS+ and CS− was within ±10%, as in our previous studies (Likhtik et al., 2014; Stujenske et al., 2014). We also quantified the discrimination strength of each mouse by the discrimination score (Figures 4C and S2A4). For the input activation experiments, we used only mice that were deemed generalizers on the Light Off trials, in order to test whether we can improve their discrimination. The proportion of generalizer mice were not significantly different between the two groups (mCherry, 6 of 14 total mice (42.9%), ChR2, 8 of 15 total mice were generalizers (53.3%), z-score of proportions = −0.56, p=0.58). To confirm the reliability of our previously defined generalization threshold in 129S6/SvEvTac mice, we fit a bimodal Gaussian distribution to the combined behavioral data based on least squares minimization (Matlab function fit, simulated using multiple random starting points to find best fit), assuming the same standard deviation for both generalizer and discriminator distributions and different means, allowing for different numbers of mice in each distribution. For the purpose of defining these distributions most robustly, data from mice implanted with microdrives but not injected with virus was combined with data from mice that received microdrives and viral injections. Based on best fit Gaussians, 129S6/SvEvTac mice that generalized were defined with 91% sensitivity and 95% specificity based on our previously established cutoff. The possibility of incorrect classification was accounted for by using control groups for all experiments. We confirmed that when we attempted to fit a bimodal distribution to data from C57B6 mice, the best fit was with a near 0 amplitude for one of the two Gaussian distributions, supporting the existence of a unimodal distribution in this data set, which shows higher discrimination for the C57B6 population of mice (Figure 4A). Microdrive construction and implantation Custom microdrives were constructed using interface boards (EIB-8, EIB-16, or EIB-32, Neuralynx, Bozeman, MT), or our custom-designed boards (Likhtik Lab), and implanted under isoflurane anesthesia. Briefly, craniotomies were made over the BLAand PL (coordinates defined above) with subsequent implantation of either 76.2 μm Formvar-coated tungsten electrodes (California Fine Wire, Grover Beach, CA) for LFP recordings or a stereotrode bundle (10-12 per mouse), constructed with 25 μm Formvar-coated tungsten micro wire (California Fine Wire, Grover Beach, CA), fastened to a cannula attached to the interface board. As explained above, electrodes directed to the BLA were affixed to fiber optics with super glue, with approximately 400μm distance between the end of the fiber optics and the electrode tips. In 129S6/SvEvTac mice, placements were made bilaterally; In C57/B6J mice, they were placed in the left hemisphere only. Surgical screws affixed to bone over the cerebellum and frontal pole, and in contact with cerebrospinal fluid, served as ground and reference, respectively. Behavior and electrophysiological activity were compared for equal light on and light off periods during fear recall after fear conditioning. Physiology data acquisition and analysis LFP signals were recorded using Digital Lynx hardware (Neuralynx) or Cerebrus Neural Signal Processor (Blackrock Neurotech), high-pass filtered (0-1000 Hz), acquired at 2 kHz, and referenced against the frontal screw. Sounds were generated by a PC, and the sound output was split between a speaker and a DC input into the electrophysiology hardware. The time of pip presentations was determined from this digital recording, to account for audio latency. Data was imported into Matlab (Natick, MA) for analysis. Three SOM-cre mice (two eArch, one control) were excluded from LFP analyses due to poor signal quality that prevented analysis, likely due to damage to the board or electrode. No 129S6/SvEvTac mice were excluded from LFP analyses during the mPFC terminal silencing experiment. Multitaper spectral power was calculated with a 250 ms moving window, 245ms overlap, a time-bandwidth product of 1.5 with 2 tapers, and 2048 FFTs. These parameters were chosen as they allowed for the use of multiple tapers and frequency resolution of +/− 3Hz, allowing the theta band to be distinguished from delta or beta. The effect of light on pip-evoked power was quantified as the average power in the 6-10 Hz frequency range in the 250ms window, starting at the onset of the pips. This value was normalized for each mouse by dividing by the average power in the 6-10Hz range during the pre-tone period (the aggregate of 30s before each tone, amounting to a total of 360 seconds). For plotting of spectrograms, each frequency band was normalized by its respective pre-tone average, and results were averaged across mice. We found that, as previously shown (reviewed in (Buzsaki and Mizuseki, 2014)), power follows a log normal distribution, and therefore log(power) was used for parametric statistical testing. To calculate the phase of ongoing theta oscillations, a bandpass filter was applied using a zero-phase-delay FIR filter with Hamming window (filter0, provided by K. Harris and G. Buzsáki, New York University, USA) and the Hilbert transform of the bandpass-filtered signal. Pip-evoked theta band activity was extracted with a 6-10 Hz bandpass. The first pip was excluded from the analysis given that it was the only pip that did not have expected timing for the mice. For SOM-cre mice, the strength of theta phase reset was calculated by quantifying the consistency of phases at each time 1-200 ms post-pip for each of the remaining 87 pips using the mean resultant length (MRL), which is a measure of the degree of unimodality of circularly distributed variables, ranging from 0 to 1. The lower bound of MRL value expected by change is dependent on sample size, and we calculated the MRL value that was expected by chance for 87 samples as .095, which was used for statistical testing of the significance of CS+ and CS− phase resetting. As expected for a bounded variable, we found that MRL exhibited a right-tailed distribution which was approximately normal after log transform, so log(phase consistency) was used for all parametric statistical testing. For 129 mice with virus expressed in mPFC, theta phase reset was calculated the same way as for SOM-cre mice, with two exceptions. First, due to increased variability in the strength of phase resetting in these mice compared to C57s, pips from two further CS presentations of each type were included for the analysis to increase robustness. Second, the phase reset was found to extend for a longer period of time in these mice, so phase consistency was averaged from 0-300 ms. For plotting of the data from 129 mice, the MRL value was normalized by dividing by the average of the CS+ and CS− light off conditions (since the log value was used, this is equivalent to subtracting by the log average), but this did not affect the statistical results of the performed ANOVA, as the repeated measures design is invariant to shifts in the mean across conditions. This normalization was performed for illustrative purposes due to the enhanced variability in the baseline MRL within 129 mice compared to C57 mice. Phase coherence between PL and BLA was calculated using a multi-taper approach, except equalizing extracted amplitude between the two signals, and calculating the coherence exclusively accounted for by changes in phase (250 ms moving window, 245ms overlap, a time-bandwidth product of 1.5 with 2 tapers, and 2048 FFTs). Interestingly, in contrast to our prior findings of overall coherence, phase coherence was increased in a consistent way that was not time locked to the pip, so average coherence values in the 6-10Hz range were averaged across the entire 30s period of tone presentation. Coherence values were normalized by subtracting average 6-10Hz coherence in the pre-tone period. Normalized coherence was normally distributed. Prefrontal cortex and basolateral amygdala theta lead probabilities were determined by iteratively performing PL-BLA power-power correlations, as previously described (Adhikari et al., 2010; Likhtik et al., 2014), using the average multitaper spectral power in the 6-10Hz range during tone presentation. This lead analysis consisted in analyzing the cross-correlation between the PL and BLA power in 1 second windows, stepping by 5 ms. The lag at which the cross-correlation was maximized was calculated in each time window, with possible values between −1s and 1s. The majority of time windows were maximized with lag = 0, suggesting no significant lead of either signal, but a substantial number of windows demonstrated either BLA lead or PL lead (negative or positive lag, respectively). PL lead probability was determined as the number of time windows which demonstrated a PL lead divided by the number of time windows with PL or BLA lead. As we expected no significant preponderance of PL or BLA lead during the CS+, while we expected a PL lead preponderance during the CS−, we defined a priori that we would test each distribution of lead probabilities using a one sample t-test vs 50%. As we did not anticipate a consistent change between CS+ and CS− in each mouse given our previous findings (Likhtik et al., 2014; Stujenske et al., 2014), we did not plan between condition testing and therefore did not perform it. As we intended to present the results of each separate t-test as our output statistic, no multiple comparison correction was applied for this analysis. For single unit recordings in SOM-eArch mice, a bundle of 16 stereotrodes (pairs of tightly-wound microwire), attached to an optic fiber terminating 300 microns above the microwire tips, was implanted in the BLA of 7 SOM-cre mice along with viral injection (as above). 2 mice were excluded from analysis due to poor signal quality that prevented further analysis. Unfiltered activity was acquired at 30 kHz with Blackrock hardware and post-processed using Kilosort2 with bandpass filtering between 300 and 6000 Hz and then manually curated using Phy2 (see Key Resources). Given that the same units can be seen across some stereotrodes, data was clustered by inputting stereotrode wires as a linear probe, such that 16 channels were considered per each spike. Wires within a stereotrode were adjacent, but otherwise ordering was random. All 32 wires could not be considered simultaneously due to a lack of support for this functionality by Kilosort2. Noise artifacts were identified and removed from clusters by using k-means clustering and visual identification. Only well-isolated single units were utilized. Cell responses were z-scored using the responses in non-overlapping 100ms bins in the 500ms before pip onset, averaged across pips. We averaged the standard deviation calculated across the four tone conditions (CS+ off, CS− off, CS+ on, CS− on) because we were interested in normalizing the size of responses between units and not within units. This improved interpretability of within-type comparisons, as it assured that differences in z-scored firing between conditions were not simply due to changes in spike variability. All aggregate tone-induced FRs changes reported are the average z-scored FR 0-400ms post-pip onset, relative to either the pre-tone mean (difference calculations reported in the text) or the pre-pip mean (as in Figure 2). For light off vs light on comparisons in the absence of tone, differences were compared between the 25 seconds before light onset and the 5 seconds after. Putative somatostatin interneurons were identified as cells that exhibited decreases in z-scored FR during light on that were below −2 and had a significantly prolonged latency of spiking on at least half of light presentations. This was determined by calculating the distribution of interspike intervals (ISI) in the intertrial interval and then questioning whether the first spike after light on exhibited an ISI that exceeded the 99th percentile. This second criterion helped to distinguish Arch-expressing SOM IN from cells that were inhibited by the light due to polysynaptic effects. Putative PV IN were identified by utilizing baseline FR and spike half-width, based on previously published data for optogenetically identified BLA PV IN (Wolff et al., 2014). Spike waveforms were isolated by taking the spike-triggered average recording in each channel, and the waveform from the recording channel with the highest spike amplitude was selected. Half-width was calculated as the width of the spike at half of the spike amplitude. In comparing our distribution of spike half-width vs FR to that previously reported (Wolff et al., 2014), we found a very similar distribution, with the exception of lacking one cluster of cells: a portion of cells with long spike half-width and high FR, some of which were identified as PV+. Interestingly, we did observe this cluster when we liberalized our single unit criterion to include some multiunits, suggesting that this cluster may be due to multi-units comprised of synchronous cell pairs; alternatively, it is possible that we did not identify this cluster due to differences in how the data was acquired and filtered. Of note, we did observe that our recorded distribution was shifted by approximately 50 μs in half-width, which is likely due to a methodological differences. Our spike waveforms are from a non-filtered recording, as opposed to threshold-crossing spike acquisition which uses bandpass filtering, artificially widening spike shape. To define PV IN, we selected units that either had a half-width <240 μs or <280 μs with a spontaneous FR of >5 Hz. Putative principal neurons were defined as non-putative SOM IN with a spike half-width >260 μs and spontaneous FR <3 Hz. As previously found using optogenetic cell identification (Wolff et al., 2014) we found that optogenetically identified SOM IN had spontaneous FRs and spike half-widths that were very similar to putative pyramidal cells. SOM IN had a FR of 1.35 +/− 1.23 Hz (mean +/− standard deviation) and spike half-width of 314.1 +/− 33.4 μs. Putative PN had a FR of 1.01 +/− .85 Hz and spike half-width of 325.4 +/− 44.8 μs. Non-SOM, non-PV IN had a FR of 2.1 +/− 2.5473 Hz and spike half-width of 325.5 +/− 46.1 μs. Individual single unit pip-evoked power spectra were calculated from pip-aligned spike times using the Chronux package (function coherencysegpt). Histology After behavioral testing, all mice were deeply anesthetized. In mice with implanted electrodes, electrothermal lesions were made to visualize electrode tip placements. Mice were trans-cardially perfused with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS). Brains were extracted and post-fixed in 4% PFA for at least three days. After cryoprotection in 30% sucrose in PBS, 40-micron histological sections were prepared on a cryostat to localize viral expression and ferrule location. A subset of sections was utilized for immunohistochemistry to validate expression patterns of the cre-dependent viruses. Mice were excluded from behavioral or electrophysiological analyses based on two criteria: 1. expression in an improper location, especially the central amygdala, which is strongly SOM+(Li et al., 2013); 2. electrode placements outside the BLA or ferrule locations that would not properly illuminate the BLA. Quantifying IN activity immunohistochemically 12-week-old, male, wild type C57/B6 mice (Jackson Laboratory) were used in this experiment, no mice were excluded from analysis. The activity of PV and SOM populations was probed using expression of the immediate-early gene c-fos, which is upregulated 1-3 hours after activation of neurons (Kovacs, 1998). All mice were trained on the differential fear conditioning paradigm, as described above, with three differences to encourage robust freezing and discrimination: 1. Mice were exposed to 6 trials of the CS+ and 6 trials of the CS−, interspersed, during training, 2. The mice were pre-exposed to context B for 20 minutes on the day of habituation, 3. On day 4, mice were exposed to either the CS+ or CS− for six trials. Animals were trained with 0.6 mA shocks for the CS+. Freezing during the tone presentations was scored offline by a researcher that was blind to group. Ninety minutes after the last tone presentation, mice were deeply anesthetized with a mixture of ketamine (100mg/kg), and xylazine (7mg/kg), and trans-cardially perfused (4% paraformaldehyde). The brains were then removed, fixed overnight, and then cryoprotected (30% sucrose in PBS). Slices were subsequently washed in PBS, blocked with 10% Normal Donkey Serum and 1% Triton, and incubated with primary antibodies for c-fos (Rabbit, Abcam, 1:2000), PV (Guinea Pig, Synaptic Systems, 1:1000), and SOM (Goat, Santa Cruz, 1:500) overnight at 4°C. The next day, slices were washed in PBS, incubated with secondary antibodies (Donkey, Jackson Immunoresearch) for 1 hour at room temperature and again washed with PBS before being mounted onto slides and coverslipped (ProLong® Gold liquid mount). Slides were imaged on a confocal microscope (Nikon A1) at 20X and positive cells counted, blinded with respect to group. C-fos co-labeling with either PV or SOM was quantified as the percentage of total PV or total SOM IN also labeling with c-fos. Muscimol infusions Male 3-5 mo C57/B6 mice were anesthetized with isofurane and stereotactically implanted with 26-gauge dual barrel cannulae (PlasticsOne) over the PL bilaterally (AP: +1.65, ML: ±0.3, DV: −1.1, allowing for infusion at DV −1.6), as already described. One week after implantation, mice were trained on the differential fear conditioning paradigm as previously described. On day four, the mice were exposed to 3 CS− and 3 CS+ tones to confirm that they were appropriately discriminating. Mice were then briefly sedated with isofurane to allow for infusion with either saline or muscimol in the PL, as already described. 30 minutes later, mice were re-exposed to 3 CS− and 3 CS+ tones and freezing to these tones was scored post-hoc. This procedure was then repeated on day 5, but infusing the opposite substance to the day prior. Mice were counterbalanced for which order they received saline and muscimol. There was no discernible change in pre-infusion discrimination between day 4 and 5. Paired analyses were performed to compare freezing of mice to the CS+ and CS− after saline and muscimol infusions. The main effect is presented as only the freezing rates after infusion, such that the effect of handling and isofurane were controlled, while the order of infusion was controlled by equally counterbalancing mice receiving saline (n=5) or muscimol (n=5) first. It is notable that no extinction was observed from day 4 to day 5 in either group. This was possibly due to the amnesiac effects of isoflurane, which was shown to affect the consolidation of learning when administered after Pavlovian fear conditioning (Dutton et al., 2002). After testing, the mice were deeply anesthetized, infused with DiI, and their brains fixed to make histological sections. Only mice with cannulae terminating in the PL were included in the analysis (n=7 mice for retrieval data in Figure 1, whereas only n=6 mice for infusion mice due to one mouse with a misplaced infusion). Monosynaptic rabies tracing Viral tracing was done using viral constructs for rabies tracing experiments that dramatically increase trans-synaptic efficacy (Reardon et al., 2016). Viruses (0.4 nL of an AAV helper virus expressing GFP and glycoprotein in a cre-dependent manner) was stereotaxically injected in the BLA (ML: −3.3, AP: −1.7, DV: −4.3) of SOM-cre mice, such that helper virus only expressed in cre expressing cells. Two weeks later, the same mice were injected with 1 μL of Glycoprotein (G)-deleted rabies expressing mCherry. Thus, starter cells were labeled with both GFP and mCherry, while presynaptically connected cells were labeled only with mCherry. GFP-expressing cells not expressing mCherry were not traced. We observed a very sparse population of starter cells (at most 2 / coronal BLA slice), but this was sufficient to label presynaptic targets to conclusively demonstrate the existence of monosynaptic prefrontal inputs to amygdala SOM interneurons. To estimate the amount of connectivity, we counted the number of starter cells in coronal slices of the BLA, each collected every 200 microns of A-P distance. We then counted the number of presynaptic cells in medial prefrontal cortex subregions by the same methodology. From these counts, we estimated the total number of cells each the brain structure based on it’s A-P extent, and then we divided to estimate the number of upstream cells per starter cell. QUANTIFICATION AND STATISTICAL ANALYSIS Statistical details are found in supplementary tables associated with each figure and supplementary figure legends, including sample sizes, test statistics, and p-values. As most experiments consisted of CS+ and CS− responses as repeated measures in the presence or absence of manipulation (e.g. light or drug infusion), two-way repeated measures ANOVA was used throughout with explanatory variables of CS, manipulation (e.g. light on vs off, pre vs post infusion), and interaction term. Statistical testing was performed in Matlab using the built-in function anovan, with the repeated measure (mouse identity) identified as a hidden random variable. In experiments with only one explanatory variable (e.g. group), one-way ANOVA was utilized (built-in function, anova1), including hidden variables to account for any paired data collection when appropriate (as in Figure 1C; using the Matlab Statistics and Machine Learning toolbox functions fitrm and ranova). Post-hoc testing was only performed if an interaction term was found to be significant, to interpret the nature of the interaction. For one-way anova, the multcompare function was used in Matlab to perform Bonferroni post hoc test, and for two-way repeated measures anova, Bonferroni-corrected paired t-tests were performed (six comparisons). For post-hoc testing with mixed paired and unpaired data (e.g., comparing calcium imaging between habituation and retrieval), the optimal pooled t-test, T0pool, was used as an alternative (Guo and Yuan, 2017). For comparing paired proportions, McNemar’s test was used, with Bonferroni correction when appropriate. For comparing unpaired proportions, a chi-square test was performed. For the evaluation of effects in individual mice, we used Bonferroni-corrected paired t-tests. For analysis of theta power directionality, we used uncorrected paired t-tests as explained above. For calcium imaging, we used uncorrected paired t-tests to compare calcium responses to the shift predictor (as in 1i, “vs shuffle”), while for LFP data, we used uncorrected one-sample t-tests compared to empirically known chance values (as in Figure S2). Uncorrected paired t-tests were also used to test for significant changes in discrimination score within individual groups of mice. All statistical tests and p-values were two-sided throughout this study. Correlation strength was calculated using Pearson’s R, where applicable. Error is reported in figures as mean ± standard error of the mean (± SEM), unless otherwise noted. One notable exception is that decoding data were presented as mean ± 95% confidence interval, as stated above. Supplementary Material 2 ACKNOWLEDGMENTS We thank Phebe Warren and Estelle Hofgaertner for technical assistance. Funding – J.M.S., P.-K.O., C.D.S., J.A.G., C.L., and E.L. J.M.S. is supported by NIMH K08MH127383. P.-K.O. is supported by NIMH K01MH123783. C.D.S. is supported by 5R01MH082017, 1R21MH116348, and the Simons Foundation. E.L. is supported by NIMH R01MH118441, BBRF, and PSC-CUNY. This article was partially prepared while J.A.G. was employed at Columbia University. The opinions expressed in this article are the author’s own and do not reflect the view of the National Institutes of Health, the Department of Health and Human Services, or the United States government. Figure 1. BLA SOM interneurons are selectively upregulated during CS− retrieval (see Table S1 for statistics) (A) Summary of 3-day differential fear conditioning paradigm. (B) On day 4, mice were randomized to undergo CS+ (n = 9) or CS— (n = 9) retrieval. Mice froze significantly more to the CS+ (top, red) than the CS− (bottom, blue). (C) PV and SOM co-labeling with c-fos revealed increased SOM IN activity during CS−. (D) PV INs (blue, full arrow, left) and SOM INs (red, arrowhead, right) co-labeled with c-fos. Scale bars, 20 μm. (E) Fiber photometry of SOM (n = 4) and PV INs (n = 8) during retrieval. (E1) Mean time course of CS response. (E2) Average CS response of PV and SOM INs. (E3) Average CS freezing. (F) Schematic of calcium imaging experiments. Note that the mice were also exposed to another three CS− and three CS+ trials on day 4 in a subsequent experiment (Figure 3). (G) Left, maximum projection of GCaMP6s fluorescence. Right, SOM INs exhibit reliable CS− responses. (H) Average SOM IN activity during habituation and retrieval. (I) Magnitude of SOM IN responses to the CSs; “vs shuffle” compares to a shift predictor. (J) Per mouse change in average response from habituation to recall. (K) The day after imaging, all mice behaviorally discriminated, despite some partial extinction. (L) Left, percent of SOM INs with a significant CS− > CS+ response (blue) or CS+ > CS− (red). Right, 42% of cells exhibited a greater CS− response while only 2% of cells exhibited a greater CS+ response (chance: 2.5%). (M) SOM INs with significantly increased (left) or decreased (right) activity during the CSs. Purple reflects response to both CSs. (N) SVMs were trained for subsamples of 150 neurons from the full dataset to decode CS+ versus CS−. The probability of correct tone identification is plotted as the mean ±95% CI. Time delay to peak performance reflects the time constant of the fluorophore. *p < 0.05, **p < 0.01, ***p < 0.001. n.s., non-significant. Unless specified, data plotted as mean ± SEM. Figure 2. BLA SOM mediate discrimination by desynchronizing inputs to the amygdala during CS− (see Table S2 for statistics) (A) Top, Cre-dependent adeno-associated viruses (AAVs) expressed eArch3.0 or eYFP in the BLA of SOM-cre mice. A subset of mice had electrodes implanted for recording LFPs and single cells. Bottom, example LFP recordings (CS+ pips, red lines). (B) During fear retrieval, half of CS presentations were paired with light. (C) Histology of SOM-cre mouse expressing eYFP. Scale bars, 200 μm. (D) Freezing of eArch group (n = 7) during CSs in the presence or absence of light in the BLA. (E) Change in discrimination score with light in SOM-Arch (purple, n = 7), control (n = 8 SOM-cre [gray], n = 3 PV-cre [black]), and PV-Arch (orange, n = 8) mice. (F) Heat map depicts phase consistency at different frequencies of BLA LFP, averaged over all SOM-cre mice (n = 6, eArch; n = 7, eYFP) during CSs (light off). Average bandpass signals are overlaid (2-Hz bandwidth, centered at frequency on y axis). Consistency at 1–3 Hz may reflect delta oscillations or pip-evoked activity. Note that temporal smoothing from the Hamming window artificially blurs phase reset prior to the pip. Also see Figure S2. (G) Average spectrogram for mice as in (F). (H) CS pip-evoked theta-filtered (6–10 Hz) phase-only LFPs (black traces) for individual CS presentations are overlain from a SOM-Arch mouse during light off and on. (I) Average theta phase consistency of Arch-expressing mice (n = 6) during CSs with and without light. Insets, schematic illustrating the effect: during the CS− (blue arrow), SOM INs are activated and block PN theta reset via dendritic inhibition. When SOM INs are less active (CS+ or light on), reset occurs. Sine waves are scaled to the magnitude of the average resets. (J) Pip-evoked theta reset strength of SOM-YFP and SOM-Arch mice during CSs, with or without light. Asterisks indicate the main effect of CS or significant Bonferroni post hoc tests. (K) As in (J), for theta power, which was not changed by light. (L) Top, plot of single unit spike half-width versus FR for putative PN (green), PV (orange), SOM (purple), and non-classified (gray) neurons (see STAR Methods), with group average ± standard deviation, n = 114 PN, 14 PV, 11 SOM, 40 non-classified from n = 5 mice. Bottom, average spike waveform for each neuronal group, normalized to the spike trough. (M) Peri-pip FR changes for putative SOM (top) and PV (bottom) INs during light on and off. Dashed lines, pip. (N) Top, light-induced FR changes during CSs for PN and PV. Bottom, percent of putative PN and PV that exhibited significant changes in firing with light during CSs. (O) Power spectra of average pip-evoked single unit responses (FR normalized as fold change from 150–500 Hz) for putative PN (top, n = 95 with sufficient spikes) and PV neurons (bottom, n = 14), with and without light. (P) Average of theta power in PN (top) and PV (bottom). *p < 0.05, **p < 0.01, ***p < 0.001. Unless specified, data plotted as mean ± SEM. Figure 3. BLA SOM activity during the CS− depends on the prelimbic cortex (see Table S3 for statistics). (A) Histology of Dil infusion through cannulae in the PL. (B) Top, saline or muscimol was infused into the PL on subsequent retrieval days (order randomized between mice). Bottom, freezing and discrimination during both days. (C) Experimental design for PL silencing during imaging of SOM INs. These mice are a subset of the mice shown in Figure 1. (D) Average CS− evoked calcium signals across all cells in mice infused with muscimol in the PL (n = 117 cells from n = 3 mice) pre- and post-infusion. Box summarizes post hoc comparisons (*p < 0.05; **p < 0.01; ***p < 0.001; also see Figure S3). (E) Heat map of individual cell responses, corresponding to (D), sorted by the average stimulus-evoked response (across all conditions). (F) Percent of cells significantly upregulated during CSs pre- and post-muscimol (F1, n = 117 cells, n = 3 mice) or saline (F2, n = 127 cells, n = 3 mice). Some cells were active to a CS both pre- and post-infusion (“stable”) or only in one imaging epoch (“changed”). (G) Accuracy decoding tone identity by SVMs trained on pre-infusion data, pre- (gray) and post- (green) muscimol (G1), or saline (G2). Plotted as mean ± 95% CI for models of different subsample sizes. Figure 4. Silencing mPFC inputs to the BLA permit synchronous cell-firing and theta reset during the CS− (see Table S4 for statistics) (A) Freezing of 129S6/SvEvTac mice (n = 67) and C57/B6J mice (n = 26) during fear retrieval. (B) Difference between CS+ and CS− freezing for 129S6/SvEvTac mice exhibits a bimodal distribution. (C) Summary of previous findings in the 129S6/SvEvTac model of fear generalization (Likhtik et al., 2014). (D) CamKII-eArch3.0 or CamKII-eYFP was expressed in the mPFC of 129S6 mice; optical fibers and electrodes were implanted in the BLA. (E) All pip-evoked theta (6–10 Hz) phase-only LFPs (black traces) for CS presentations from an Arch mouse with and without light. (F) Average pip-evoked theta reset in the BLA of Arch mice with and without light in the BLA. (G) Silencing mPFC-to-BLA terminals significantly decreased discrimination. (H) Mice were injected with viruses expressing CamKII-ChR2 or CamKII-mCherry in the PL for activation of terminals in the BLA. (I) Expression of ChR2 virus in PL bodies (top) and BLA terminals (bottom). Scale bars, 500 μm. (J) Stimulating PL inputs to the BLA in ChR2-expressing generalizers decreased CS− evoked freezing. (K) Activating PL-to-BLA terminals significantly increased discrimination. *p < 0.05, **p < 0.01, ***p < 0.001. Unless specified, data plotted as mean ± SEM. Key resources table REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Anti-c-fos (rabbit) Abcam Cat#ab190289; RRID:AB_2737414 Anti-parvalbumin (guinea pig) Synaptic Systems Cat#195 004; RRID:AB_2156476 Anti-SOM (goat) Santa Cruz Cat#sc-7819; RRID:AB_2302603   Bacterial and virus strains AAV5-CamKIIa-hChR2-H134R-mCherry-WPRE-pA UNC Vector Core N/A AAV5-CamKIIa-H134R-mCherry-WPRE-pA UNC Vector Core N/A AAV5-CamKIIa-eArchaerhodopsin-eYFP UNC Vector Core N/A AAV5-CamKIIa-eYFP UNC Vector Core N/A AAV5-Ef1a-DIO-eArch3.0-eYFP UNC Vector Core N/A AAV5-Ef1a-DIO-eYFP UNC Vector Core N/A AAV5-hsyn-FLEX-GCaMP6s Penn Vector Core, Addgene Cat#(Addgene):100845-AAV5; Plasmid RRID:Addgene_100845 AAV1-Ef1a-FLEX-H2B-GFP-P2A-N2c[G] Laboratory of Attila Losonczy, Reardon et al., 2016 Plasmid RRID:Addgene_73476 AAV1-FLEX-TVA-mCherry Laboratory of Attila Losonczy, Reardon et al., 2016 Plasmid RRID:Addgene_38044 RabV CVS-N2cΔG-dsRed[EnvA] Laboratory of Attila Losonczy, Reardon et al., 2016 Plasmid RRID:Addgene_73460   Chemicals, peptides, and recombinant proteins Muscimol Tocris Cat#0289   Experimental models: Organisms/strains PV-IRES-cre transgenic mice Jackson Laboratory RRID:IMSR_JAX:017320 SOM-IRES-cre transgenic mice Jackson Laboratory RRID:IMSR_JAX:018973 Wildtype C57/B6J mice Jackson Laboratory RRID:IMSR_JAX:000664 Wildtype 129S6/SvEvTac Taconic Biosciences RRID:IMSR_TAC:129sve   Software and algorithms Kilosort2 Github, retrieved on Sept 18, 2020 https://github.com/MouseLand/Kilosort/releases/tag/v2.0 Chronux2.12v03 http://chronux.org/, retrieved on October 27, 2021 http://chronux.org/ Circular Statistics Toolbox (Matlab) Mathworks https://www.mathworks.com/matlabcentral/fileexchange/10676-circular-statistics-toolbox-directional-statistics B-spline Grid, Image and Point based Registration Toolbox (Matlab) Mathworks https://www.mathworks.com/matlabcentral/fileexchange/20057-b-spline-grid-image-and-point-based-registration deconvolution_standalone.m (Matlab) Github, retrieved on April 26, 2019 https://github.com/cortex-lab/Suite2P Phy2 Github, retrieved on Sept 18, 2020 https://github.com/cortex-lab/phy ImageJ Research Services Branch, National Institute of Mental Health https://imagej.nih.gov/ij/ Matlab 2021a Mathworks https://www.mathworks.com/products/matlab.html Spectral, single unit, fiber photometry analyses This paper DOI: 10.5281/zenodo.6347608 https://github.com/jmstujenske/PL_drives_nonaversion_via_SOM_IN   Other 532nm DPSS laser (3% power stability) OEM Laser Systems GR-532-00100-CWM-SD-03-LED-0 Optical commutator Doric FRJ_1x2i_FC-2FC PlexBright Dual LED Commutator Plexon COM-Dual LED PlexBright Compact LED Modules (465nm) Plexon LED_Blue_Compact_LC_magnetic PlexBright One Channel LED Driver Plexon 1ch_LED_Driver Fiber Photometry System Neurophotometrics FP3002 Digital Acquisition System Neuralynx Digital Lynx 4SX Digital Acquisition System Blackrock Neurotech Cerebrus Neural Signal Processor Custom GRIN lens GrinTech GmBH NEM-085-50-00-920-S-1.0p Two-photon microscope Prairie Technologies Ultima IV Model Tunable mode-locked laser Coherent Chameleon Vision model Tungsten Wire with Formvar coating (25 μm) California Fine Wire Material #100211, Coating NEMA MW 15, custom order Tungsten Wire with Formvar coating (76.2 μm) California Fine Wire Material #100211, Coating NEMA MW 15, custom order Electronic interface boards (8, 16, or 32 channel) Neuralynx EIB-8, EIB 16, EIB 32 narrow Electronic interface boards (32 channel) Gold Phoenix Custom Design, http://likhtiklab.com/tools Stainless steel binding screws (000-120 thread, 1/16” long) Antrin Miniature Specialties Part#B002SG89KW Highlights BLA SOM interneurons are selectively activated by learned non-aversive stimuli SOM interneurons filter out inputs that synchronize BLA, supporting discrimination Prelimbic cortex is necessary for BLA SOM interneuron activation and discrimination Prelimbic input to the BLA drives discrimination SUPPLEMENTAL INFORMATION Supplemental information can be found online at https://doi.Org/10.1016/j.neuron.2022.03.020. DECLARATION OF INTERESTS The authors declare no competing interests. REFERENCES Adhikari A , Sigurdsson T , Topiwala MA , and Gordon JA (2010). Crosscorrelation of instantaneous amplitudes of field potential oscillations: a straightforward method to estimate the directionality and lag between brain areas. J. Neurosci. Methods 191 , 191–200.20600317 Ali F , and Kwan AC (2020). Interpreting in vivo calcium signals from neuronal cell bodies, axons, and dendrites: a review. Neurophotonics 7 , 011402.31372367 Arruda-Carvalho M , and Clem RL (2014). Pathway-selective adjustment of prefrontal-amygdala transmission during fear encoding. J. Neurosci 34 , 15601–15609.25411488 Arruda-Carvalho M , Wu WC , Cummings KA , and Clem RL (2017). Optogenetic examination of prefrontal-amygdala synaptic development. J. Neurosci 37 , 2976–2985.28193691 Bergado-Acosta JR , Sangha S , Narayanan RT , Obata K , Pape HC , and Stork O (2008). Critical role of the 65-kDa isoform of glutamic acid decarboxylase in consolidation and generalization of Pavlovian fear memory. Learn. Mem 15 , 163–171.18323571 Burgos-Robies A , Kimchi EY , Izadmehr EM , Porzenheim MJ , Ramos-Guasp WA , Nieh EH , Felix-Ortiz AC , Namburi P , Leppla CA , Presbrey KN , (2017). Amygdala inputs to prefrontal cortex guide behavior amid conflicting cues of reward and punishment. Nat. Neurosci 20 , 824–835.28436980 Buzsáki G , and Mizuseki K (2014). The log-dynamic brain: how skewed distributions affect network operations. Nat. Rev. Neurosci 15 , 264–278.24569488 Cummings KA , and Clem RL (2020). Prefrontal somatostatin interneurons encode fear memory. Nat. Neurosci 23 , 61–74.31844314 Davis P , Zaki Y , Maguire J , and Reijmers LG (2017). Cellular and oscillatory substrates of fear extinction learning. Nat. Neurosci 20 , 1624–1633.28967909 Dutton RC , Maurer AJ , Sonner JM , Fanseiow MS , Laster MJ , and Eger EI 2nd (2002). Isoflurane causes anterograde but not retrograde amnesia for Pavlovian fear conditioning. Anesth. 96 , 1223–1229. Ghosh S , and Chattarji S (2015). Neuronal encoding of the switch from specific to generalized fear. Nat. Neurosci 18 , 112–120.25436666 Grewe BF , Gründemann J , Kitch LJ , Lecoq JA , Parker JG , Marshall JD , Larkin MC , Jercog PE , Grenier F , Li JZ , (2017). Neural ensemble dynamics underlying a long-term associative memory. Nature 543 , 670–675.28329757 Grosso A , Santoni G , Manassero E , Renna A , and Sacchetti B (2018). A neuronal basis for fear discrimination in the lateral amygdala. Nat. Commun 9 , 1214.29572443 Guo B , and Yuan Y (2017). A comparative review of methods for comparing means using partially paired data. Stat. Methods Med. Res 26 , 1323–1340.25834090 Hara-Kuge S , Nishihara T , Matsuda T , Kitazono T , Teramoto T , Nagai T , and Ishihara T (2018). An improved inverse-type Ca2+ indicator can detect putative neuronal inhibition in Caenorhabditis elegans by increasing signal intensity upon Ca2+ decrease. PLoS One 13 , e0194707.29694380 Hayes JP , Hayes SM , and Mikedis AM (2012). Quantitative meta-analysis of neural activity in posttraumatic stress disorder. Biol. Mood Anxiety Disord 2 , 9.22738125 Hippenmeyer S , Vrieseling E , Sigrist M , Portmann T , Laengle C , Ladle DR , and Arber S (2005). A developmental switch in the response of DRG neurons to ETS transcription factor signaling. PLoS Biol. 3 , e159.15836427 Ito W , Fusco B , and Morozov A (2020). Disinhibition-assisted long-term potentiation in the prefrontal-amygdala pathway via suppression of somatostatin-expressing interneurons. Neurophotonics 7 , 015007.32090134 Kaifosh P , Zaremba JD , Danielson NB , and Losonczy A (2014). SIMA: Python software for analysis of dynamic fluorescence imaging data. Front. Neuroinform 8 , 80.25295002 Klavir O , Genud-Gabai R , and Paz R (2013). Functional connectivity between amygdala and cingulate cortex for adaptive aversive learning. Neuron 80 , 1290–1300.24314732 Kovács KJ (1998). c-Fos as a transcription factor: a stressful (re)view from a functional map. Neurochem. Int 33 , 287–297.9840219 Krabbe S , Paradiso E , d’Aquin S , Bitterman Y , Courtin J , Xu C , Yonehara K , Markovic M , Müller C , Eichlisberger T , (2019). Adaptive disinhibitory gating by VIP interneurons permits associative learning. Nat. Neurosci 22 , 1834–1843.31636447 Lee YK , and Choi JS (2012). Inactivation of the medial prefrontal cortex interferes with the expression but not the acquisition of differential fear conditioning in rats. Exp. Neurobiol 21 , 23–29.22438676 Li H , Penzo MA , Taniguchi H , Kopec CD , Huang ZJ , and Li B (2013). Experience-dependent modification of a central amygdala fear circuit. Nat. Neurosci 16 , 332–339.23354330 Likhtik E , Pelletier JG , Paz R , and Paré D (2005). Prefrontal control of the amygdala. J. Neurosci 25 , 7429–7437.16093394 Likhtik E , Stujenske JM , Topiwala MA , Harris AZ , and Gordon JA (2014). Prefrontal entrainment of amygdala activity signals safety in learned fear and innate anxiety. Nat. Neurosci 17 , 106–113.24241397 Marek R , Xu L , Sullivan RKP , and Sah P (2018). Excitatory connections between the prelimbic and infralimbic medial prefrontal cortex show a role for the prelimbic cortex in fear extinction. Nat. Neurosci 21 , 654–658.29686260 Meyer HC , and Bucci DJ (2014). The contribution of medial prefrontal cortical regions to conditioned inhibition. Behav. Neurosci 128 , 644–653.25285456 Pachitariu M , Stringer C , Dipoppa M , Schröder S , Rossi LF , Dalgleish H , Carandini M , and Harris KD (2017). Suite2p: beyond 10,000 neurons with standard two-photon microscopy. Preprint at BioArxiv. 10.1101/061507. Pnevmatikakis EA , and Giovannucci A (2017). NoRMCorre: an online algorithm for piecewise rigid motion correction of calcium imaging data. J. Neurosci. Methods 291 , 83–94.28782629 Reardon TR , Murray AJ , Turi GF , Wirbiich C , Croce KR , Schneii MJ , Jessell TM , and Losonczy A (2016). Rabies virus CVS-N2c(ΔG) strain enhances retrograde synaptic transfer and neuronal viability. Neuron 89 , 711–724.26804990 Rosenkranz JA , and Grace AA (2001). Dopamine attenuates prefrontal cortical suppression of sensory inputs to the basolateral amygdala of rats. J. Neurosci 21 , 4090–4103.11356897 Sangha S , Chadick JZ , and Janak PH (2013). Safety encoding in the basal amygdala. J. Neurosci 33 , 3744–3751.23447586 Shaban H , Humeau Y , Herry C , Cassasus G , Shigemoto R , Ciocchi S , Barbieri S , van der Putten H , Kaupmann K , Bettler B , and Lüthi A (2006). Generalization of amygdala LTP and conditioned fear in the absence of presynaptic inhibition. Nat. Neurosci 9 , 1028–1035.16819521 Sierra-Mercado D , Padiiia-Coreano N , and Quirk GJ (2011). Dissociable roles of prelimbic and infralimbic cortices, ventral hippocampus, and basolateral amygdala in the expression and extinction of conditioned fear. Neuropsychopharmacology 36 , 529–538.20962768 Stujenske JM , Likhtik E , Topiwala MA , and Gordon JA (2014). Fear and safety engage competing patterns of theta-gamma coupling in the basolateral amygdala. Neuron 83 , 919–933.25144877 Taniguchi H , He M , Wu P , Kim S , Paik R , Sugino K , Kvitsiani D , Fu Y , Lu J , Lin Y , (2011).A resource of Cre driver lines for genetic targeting of GABAergic neurons in cerebral cortex. Neuron 71 , 995–1013.21943598 Taub AH , Perets R , Kahana E , and Paz R (2018). Oscillations synchronize amygdala-to-prefrontal primate circuits during aversive learning. Neuron 97 , 291–298.e3.29290553 Vieira PA , Corches A , Lovelace JW , Westbrook KB , Mendoza M , and Korzus E (2015). Prefrontal NMDA receptors expressed in excitatory neurons control fear discrimination and fear extinction. Neurobiol. Learn. Mem 119 , 52–62.25615540 Wolff SB , Gründemann J , Tovote P , Krabbe S , Jacobson GA , Müller C , Herry C , Ehrlich I , Friedrich RW , Letzkus JJ , and Lüthi A (2014). Amygdala interneuron subtypes control fear learning through disinhibition. Nature 509 , 453–458.24814341

PMC009xxxxxx/PMC9308684.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8809320 1600 Neuron Neuron Neuron 0896-6273 1097-4199 35584693 9308684 10.1016/j.neuron.2022.04.031 NIHMS1809322 Article Global and subtype-specific modulation of cortical inhibitory neurons regulated by acetylcholine during motor learning Ren Chi 1 Peng Kailong 13 Yang Ruize 14 Liu Weikang 1 Liu Chang 25 Komiyama Takaki 16* 1 Neurobiology Section, Center for Neural Circuits and Behavior, Department of Neurosciences, and Halıcıoğlu Data Science Institute, University of California San Diego, La Jolla, CA 92093, USA 2 Department of Computer Science and Engineering, University of California San Diego, La Jolla, CA 92093, USA 3 Present address: Interdepartmental Neuroscience Program, Yale University, New Haven, CT 06520, USA 4 Present address: Interdepartmental Neuroscience Program, Northwestern University, Chicago, IL 60611, USA 5 Present address: Department of Bioengineering, University of California Berkeley, Berkeley, CA 94720, USA 6 Lead contact Author Contributions T.K. and C.R. conceived the project. C.R. performed behavioral training, imaging, and manipulation experiments and analyzed the data with inputs from T.K. and assistance from K.P., R.Y., W.L., and C.L.. C.L., C.R., K.P., and R.Y. performed immunostaining. C.R. and T.K. wrote the paper. * Correspondence: Takaki Komiyama, tkomiyama@ucsd.edu 29 5 2022 20 7 2022 17 5 2022 20 7 2023 110 14 23342350.e8 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Summary Inhibitory neurons (INs) consist of distinct subtypes with unique functions. Previous studies on INs mainly focused on single brain regions, and thus it remains unclear whether the modulation of IN subtypes occurs globally across multiple regions. Here, we monitored the activity of different cortical IN subtypes at both macroscale and microscale in mice learning a lever-press task. Learning evoked a global modulation of IN subtypes throughout the cortex. The initial learning phase involved strong activation of vasoactive intestinal peptide-expressing INs (VIP-INs) and weak activation of somatostatin-expressing INs (SOM-INs). Inactivating VIP-INs increased SOM-IN activity and impaired initial learning. Concurrently, cortical cholinergic inputs from the basal forebrain were initially more active but became less engaged over learning. Manipulation of the cholinergic system impaired motor learning and differentially altered activity of IN subtypes. These results reveal that motor learning involves a global and subtype-specific modulation on cortical INs regulated by the cholinergic system. eTOC blurb With longitudinal calcium imaging at both macroscale and microscale in behaving mice, Ren et al. uncover a global and subtype-specific modulation of cortical inhibitory neuron activity regulated by the basal forebrain cholinergic system during motor learning. inhibitory neuron cholinergic system motor learning wide-field calcium imaging two-photon calcium imaging pmcIntroduction Cortical inhibitory neurons (INs) consist of heterogeneous populations, and each IN subtype carries out unique functions based on the morphological features, anatomical connectivity, and physiological properties (Hattori et al., 2017; Kepecs and Fishell, 2014; Tremblay et al., 2016). According to the expression of molecular markers, most cortical INs can be divided into three major subtypes: the vasoactive intestinal peptide- (VIP-), somatostatin- (SOM-), and parvalbumin- (PV-) expressing INs (Rudy et al., 2011). Even though each of these three subtypes includes heterogeneous subgroups (Tasic et al., 2018), they nevertheless represent largely distinct functional groups. PV-INs mainly target the perisomatic regions of excitatory neurons and inhibit their outputs, whereas SOM-INs typically synapse onto the distal dendrites and exert focal inhibition of synaptic inputs to excitatory neurons. In contrast, VIP-INs mostly inhibit other IN subtypes, mediating disinhibition of excitatory neurons (Letzkus et al., 2015; Markram et al., 2004; Naka and Adesnik, 2016). Such regulations by distinct IN subtypes are critical for shaping excitatory circuits in various brain functions during development and adulthood (Arriaga and Han, 2019; Bicks et al., 2020; Fu et al., 2014, 2015; Kim et al., 2016a; Krabbe et al., 2019; Kuchibhotla et al., 2016; Levelt and Ḧubener, 2012; Makino and Komiyama, 2015; Pi et al., 2013). During motor learning, neural activity becomes more refined and reproducible both within the primary motor cortex (M1) and across spatially distributed cortical regions (Makino et al., 2017; Peters et al., 2014). Motor learning reorganizes local excitatory circuits in M1 layer 2/3 through the plasticity of dendritic spines at their distal dendrites (Chen et al., 2015b; Peters et al., 2014; Xu et al., 2009). It has been suggested that such learning-related spine reorganization is regulated by dendritic disinhibition from SOM-INs (Chen et al., 2015b). Other studies have also demonstrated that the proper functioning of SOM- and PV-INs is important to maintain learned motor skills (Adler et al., 2019; Cichon and Gan, 2015; Donato et al., 2013; Vallentin et al., 2016). Although accumulating evidence has suggested that different IN subtypes play unique roles in motor learning, these studies mainly focused on M1. Given that motor learning accompanies a reorganization of the cortex-wide activity in excitatory neurons (Makino et al., 2017), it is of particular interest to investigate whether learning also modulates IN subtypes globally throughout the cortex. Furthermore, it remains elusive how individual IN subtypes are modulated from the initial learning phase to the expert stage. Additionally, the underlying mechanisms driving these subtype-specific modulations remain unclear. To address these questions, we systematically monitored the activity of different cortical IN subtypes at both macroscale and microscale in mice learning a lever-press task. We found that different IN subtypes exhibited distinct learning-related changes in their activity during movements, not only in M1 but throughout the cortex. The IN subtype modulation was mediated by a strong engagement of cortical cholinergic inputs from the basal forebrain during the initial learning phase. Our results reveal a global, subtype-specific modulation of cortical INs regulated by the cholinergic system during motor learning. Results Global and subtype-specific modulation of IN activity in the dorsal cortex To investigate the learning-induced dynamics of different IN subtypes throughout the cortex, we used motor learning as a platform and trained mice to learn a lever-press task (Peters et al., 2014) over weeks (n = 40 mice from session 1 to 21, n = 17 mice for session 22, one session per day, Figure 1A; STAR Methods). Using the same lever-press task, we have recently shown that motor learning evokes a reorganization of the cortex-wide macroscopic activity pattern in cortical excitatory neurons (Makino et al., 2017). In this task, water-restricted mice learned to use their left forelimb to press a lever beyond the set threshold during an auditory cue to receive a water reward. Mice showed a gradual improvement in performance over learning, indicated by the increased fraction of rewarded trials (p < 0.0001, mixed-effects model, Figure 1B). Furthermore, mice developed more reproducible lever-press movements with training (Figure 1C), as shown by the increased correlation of trial-by-trial lever trajectories both within (p < 0.0001, mixed-effects model) and across sessions (p < 0.0001, mixed-effects model, Figures 1D and 1E; see Figure S1 for other behavioral parameters and more example lever trajectories). To monitor the cortex-wide activity of different IN subtypes, we performed longitudinal wide-field calcium imaging in task-performing mice with GCaMP6f expressed in VIP-, SOM-, or PV-INs broadly across the cortex (Figures 2A and S2A; STAR Methods). This was achieved by crossing VIP-Cre, SOM-Cre (Taniguchi et al., 2011), or PV-Cre mice (Hippenmeyer et al., 2005) with Cre-dependent GCaMP6f reporter transgenic mice (Madisen et al., 2015). Similar to the global activation of cortical excitatory neurons during movements (Makino et al., 2017; Musall et al., 2019), all three IN subtypes were activated during movements throughout the dorsal cortex during task performance (Figures 2B and S2B; Movie S1). The amplitudes of activity varied across regions with the largest amplitudes in regions that are most dorsal (e.g., S1HL). This pattern was similar to observations in wide-field calcium imaging of cortical excitatory neurons during the same lever-press task in our previous work (Makino et al., 2017) and also during the imaging of spontaneous activity (not shown). Therefore, amplitude differences across cortical regions may be of technical origin and so we limited our analysis to activity changes within each region. Additional control experiments using transgenic mice with GFP expressed in VIP-INs confirmed that that the majority of fluorescence changes we report are indeed calcium signals (Figures S2C–S2F; STAR Methods). To investigate the dynamics of IN subtypes during the learning of this motor task, we focused on the activity surrounding rewarded movements (STAR Methods). VIP- and SOM-INs demonstrated different changes in their activity during movement epochs. The activity of VIP-INs was initially strong and decreased in most of the cortical regions, while the activity of SOM-INs was initially weak and increased globally. In contrast, the activity of PV-INs was more stable in the dorsal cortex throughout learning, only showing a marginal increase in a minority of cortical regions (n (VIP) = 11 mice, n (SOM) = 12 mice, n (PV) = 11 mice, Figure 2C). We further characterized the changes in activity amplitude and duration by quantifying the peak and the full width at half maximum (FWHM) of trial-based IN activity in individual cortical regions respectively (STAR Methods). VIP-INs decreased the activity amplitude with learning while the activity duration remained stable. SOM-INs increased both the activity amplitude and duration (Figure S3). Therefore, motor learning evokes a global and subtype-specific modulation of IN activity in the dorsal cortex. The initial learning phase was accompanied by a strong global activation of VIP-INs and weak global activation of SOM-INs. We also note that, even though VIP- and SOM-INs changed their activity rather globally, there was also some heterogeneity across cortical regions. For example, the learning-related changes of VIP- and SOM-IN activity were strongly anti-correlated in M1, while VIP-IN activity remained relatively stable in RSC (Figure 2C). To further evaluate the contributions of different behavioral events to neural activity during movement epochs, we constructed a generalized linear model (Musall et al., 2019; Pinto and Dan, 2015) to predict the activity of IN subtypes in each cortical region using a set of behavioral events as predictors (Figure 3A; STAR Methods). This model was able to closely reproduce the activity profile of three IN subtypes (Figures 3B–3D), predicting 47.98% ± 1.99%, 60.80% ± 1.40%, 59.92% ± 1.40% of variance in VIP-, SOM-, and PV-IN activity respectively (n (VIP) = 11 mice, n (SOM) = 12 mice, n (PV) = 11 mice, mean ± SEM). To determine the contributions of each behavioral event during movement epochs, we calculated the activity predicted by individual behavioral events by examining individual terms of the full model, and compared the predicted activity with the true imaged activity. Overall, the lever-press movements had the highest contribution to the activity, followed by licking, reward, and then the auditory cue (Figure 3E). Furthermore, the learning-related decreases of VIP-IN activity were driven by decreases in the activity related to lever-press movements, and increases of SOM-IN activity were attributed to lever-press movements and licking (Figure 3F). These results suggest that the activity and learning-related modulations that we report in this study are most strongly related to lever-press movements. Modulation of individual VIP- and SOM-INs in M1 during motor learning Wide-field calcium imaging monitors spatially integrated activity of neurons expressing the activity indicator. We further characterized the learning-induced dynamics of individual VIP- and SOM-INs, as these two subtypes showed more prominent modulations during learning in the wide-field experiments described above. The activity of layer 2/3 VIP- and SOM-INs in the right M1 was monitored longitudinally using two-photon calcium imaging in task-performing mice (Figure 4A; STAR Methods). We injected adeno-associated viruses (AAV) encoding Cre-dependent GCaMP6f (AAV-Syn-FLEX-GCaMP6f) (Chen et al., 2013) into the forelimb area of the right M1 in VIP-Cre or SOM-Cre transgenic mice (Taniguchi et al., 2011). The forelimb area of the right M1 recorded in two-photon calcium imaging largely overlapped with the r-M1 module in wide-field calcium imaging. Individual VIP- and SOM-INs exhibited diverse activity profiles during movement epochs, displaying activated and suppressed responses during movements (870 VIP-INs from 7 mice, 533 SOM-INs from 10 mice, Figures 4B and 4G; STAR Methods). Consistent with the observations from wide-field calcium imaging, VIP- and SOM-INs displayed subtype-specific changes in their activity during learning, showing a decrease in the averaged population activity of VIP-INs (p < 0.0001, mixed-effects model, Figures 4C and 4D) but an increase of SOM-IN activity (p = 0.0007, mixed-effects model, Figures 4H and 4I). Consistently, at the single-cell level, the fraction of VIP-INs that significantly decreased their activity throughout learning was larger than those that increased (Early v.s. Naive: p < 0.0001, Middle v.s. Naive: p < 0.0001, Late v.s. Naive: p < 0.0001, mixed-effects model, corrected for multiple comparisons by false discovery rate, Figures 4E, 4F, and S4A). In contrast, the activity increased in a higher fraction of SOM-INs than the fraction of decreased SOM-INs (Early v.s. Naive: p = 0.0011, Middle v.s. Naive: p = 0.0011, Late v.s. Naive: p = 0.0011, mixed-effects model, corrected for multiple comparisons by false discovery rate, Figures 4J, 4K, and S4D). Given the diversity of activity of individual INs during movements (e.g., movement-activated vs. suppressed), we further investigated how the changes in the activity level during learning related to the types of activity during movements of individual neurons. In VIP-INs, the majority of neurons that decreased their activity were movement-activated at the naive stage but became less activated or non-modulated over learning (Figure S4C). In SOM-INs, neurons that increased their activity were mainly non-modulated at the naive stage but became movement-activated with learning (Figure S4F). Repeating the same analyses on the neurons from randomly selected 50% of animals consistently generated similar results (Figures S4B, S4C, S4E, and S4F), indicating that the phenomena we observed were not driven by a small number of outlier animals. Taken together, these results further demonstrate the subtype-specific modulation during motor learning at the cellular resolution. The activation of the majority of VIP-INs is strong at the beginning of learning and gradually decreases, while the activity of many SOM-INs is initially weak and increases during learning. Inactivation of VIP-INs increases the SOM-IN activity during movements and impairs motor learning in naive animals The strong activation of VIP-INs at the naive stage may allow learning-related plasticity of excitatory neurons by releasing them from SOM-IN inhibition (Chen et al., 2015b). To investigate the function of VIP-INs in the initial learning phase, we inactivated VIP-INs with hM4Di and CNO. CNO administration (10 mg/kg body weight) effectively inactivated VIP-INs during spontaneous lever-press movements (Figures S5A–S5C; STAR Methods). To further characterize the effective window of hM4Di under repeated CNO administrations during training, we expressed GCaMP6f in VIP-INs, with a subset of them also co-expressing hM4Di (Figure S5D; STAR Methods). This allowed us to compare the activity of hM4Di-expressing VIP-INs to non-hM4Di-expressing VIP-INs within the same animal under CNO administrations. The inactivation was only effective in the first two CNO administrations (one administration per day, 10 mg/kg body weight, Figure S5E), possibly reflecting receptor desensitization and downregulation following repeated dosing with CNO (Roth, 2016). Therefore we restricted our experiments to the first two days of training, which correspond to the naive stage of learning. We inactivated VIP-INs while monitoring the activity of SOM-INs in the right M1 with two-photon imaging during task performance in naive animals. This was achieved by using double transgenic mice in which the Cre recombinase was expressed in VIP-INs and the Flp recombinase was expressed in SOM-INs (VIP-Cre::SOM-Flp (He et al., 2016; Taniguchi et al., 2011)). We co-injected AAVs expressing Cre-dependent hM4Di (AAV-hSyn-DIO-hM4Di-mCherry) and Flp-dependent GCaMP6f (AAV-EF1a-fDIO-GCaMP6f) into the right M1 (hM4Di group, Figure 5A; STAR Methods). A cohort of animals expressing only mCherry in VIP-INs served as a control group (mCherry group, Figure 5A). With CNO administration, the SOM-INs were more strongly activated in the hM4Di group compared to the mCherry group (p = 0.0186, mixed-effects model, 619 neurons from 11 animals in the mCherry group, 559 neurons from 11 animals in the hM4Di group, Figures 5B–5E). The elevated activation of SOM-INs was still observed when we examined the neurons from randomly selected 50% of animals in each group (Figure S5F), indicating the consistency of the observation. Furthermore, the increased activation of SOM-INs during movements was accompanied by impairments in motor learning. Naive animals in the hM4Di group showed a lower fraction of rewarded trials and a longer time interval from movement onset to reward (mixed-effects model, 11 animals in each group, Figure 5F). These impairments in task performance were unlikely due to reduced motivation or deficits in movement generation (Figure S5G). In addition, animals in the hM4Di group also performed worse at the beginning of the session after CNO manipulation sessions compared to the control group (Figure 5G), suggesting a deficit from poor learning in prior CNO manipulation sessions. These results suggest that the strong activation of VIP-INs at the naive stage of learning suppresses SOM-INs, creating a window for learning-related plasticity in excitatory neurons and allowing the acquisition of new motor skills. Strong activation of basal forebrain cholinergic projections to the cortex in the initial learning phase What is the mechanism underlying the globally-orchestrated, subtype-specific dynamics of INs during motor learning? We hypothesized that the basal forebrain cholinergic system could mediate these changes based on the following reasons. In a subset of animals, we performed pupillometry recordings and observed a large pupil dilation during movements in the initial learning phase that decreased with learning (Figures S6A and S6B; STAR Methods). As the pupil diameter correlates with noradrenergic and cholinergic activity in the cortex (Reimer et al., 2016), the changes in pupil dilation suggest the involvement of neuromodulators, including acetylcholine (ACh), during learning. Previous literature has also supported the role of ACh in mediating learning-related modulations of IN activity. First, the basal forebrain cholinergic system is widely implicated in learning (Crouse et al., 2020; Guo et al., 2019; Lin et al., 2015). Second, cortical INs express ACh receptors (AChRs) and are responsive to cholinergic inputs (Dasgupta et al., 2018; Fu et al., 2014; Gasselin et al., 2021; Kuchibhotla et al., 2016; Yao et al., 2021). Third, basal forebrain cholinergic neurons project broadly to the cortex (Do et al., 2016; Kim et al., 2016b; Li et al., 2018; Woolf, 1991), which may drive the global changes in IN activity during learning. To monitor the dynamics of basal forebrain cholinergic inputs to the cortex during motor learning, we injected AAV-hSyn-FLEX-axon-GCaMP6s (Broussard et al., 2018) into the basal forebrain of ChAT-Cre transgenic mice (Rossi et al., 2011) and recorded the activity of cholinergic axons in the cortex with two-photon calcium imaging during learning (Figures 6A, 6B, and S6C–S6E; STAR Methods). We imaged the right M1, S1HL, and PPC in separate sessions to cover a range of areas across the dorsal cortex. In all three cortical regions, the activation of cholinergic axons during movements was most prominent at the naive stage, which then was gradually attenuated with learning (p (M1) = 0.0016, p (S1HL) = 0.0059, p (PPC) = 0.0190, mixed-effects model, n (M1) = 7 mice, n (S1HL) = 6 mice, n (PPC) = 6 mice, Figure 6C), suggesting a global learning-related modulation in which the cholinergic projections are highly recruited across distributed cortical regions in naive but less engaged in expert mice. This observation is consistent with the decreased pupil dilation during learning (Figure S6B) and shows a similar trend as the decreased VIP-IN activity with learning (Figures 2 and 4). Basal forebrain cholinergic inputs to the cortex are necessary for efficient motor learning Having established that the cholinergic projections are particularly active during the initial learning phase, we next asked whether the cholinergic system contributed to the learning of new motor skills. To test this idea, we used two different methods to impair the cholinergic system and examined the effect on behavior. First, we ablated basal forebrain cholinergic neurons bilaterally by injecting AAV-EF1a-FLEX-taCaspase3 (Yang et al., 2013) into the basal forebrain of ChAT-Cre mice (STAR Methods). A separate cohort of ChAT-Cre animals receiving saline injections served as a control group (Figure 6D). Four weeks after injections, histological analyses revealed that this approach achieved a nearly complete ablation of cholinergic neurons restricted to basal forebrain nuclei (Figures S7A–S7F). Compared to the control group, mice in the ablation group showed deficits in learning the lever-press task. The ablation animals were slower in achieving high success rates, and their movements did not reach the same level of reproducibility as control animals (n (Ablation) = 15 mice, n (Control) = 13 mice, Figures 6E, 6F; see Figure S7G for other behavioral parameters). Chronic ablation of cholinergic neurons could induce compensatory mechanisms that limit behavioral effects. Additionally, because basal forebrain cholinergic neurons also send projections to subcortical regions, the behavioral effects above could originate from cholinergic functions in subcortical regions. To address these issues, we also performed a cortex-wide and acute ontogenetic inactivation targeting basal forebrain cholinergic projections to the cortex using the inhibitory opsin eOPN3 (Mahn et al., 2021). Activating eOPN3 with green light (~532 nm, ~2 mW/mm2 for 15 s) reduced GCaMP6f responses by ~50% in cholinergic axons (Figures S7H–S7J; STAR Methods). This is a larger degree of suppression than the results reported in the original paper (Mahn et al., Neuron, 2021), suggesting an effective (but not complete) suppression of the cholinergic activity with eOPN3. To achieve a cortex-wide inactivation of the basal forebrain cholinergic projections, we bilaterally injected AAV-hSyn-SIO-eOPN3-mScarlet in the basal forebrain of ChAT-Cre animals (eOPN3 group, Figure 6G; STAR Methods). A separate cohort of ChAT-Cre animals received injections of AAV-hSyn-DIO-mCherry and served as a control group (mCherry group). Green light was delivered to the entire dorsal cortex through the skull (~532 nm, ~4 mW/mm2 at the skull surface and ~2 mW/mm2 at the brain surface, for 15 s every 5 min spanning entire sessions, Figures S7I and S7J; STAR Methods) in both groups during training. Compared to the control group, the eOPN3 group showed a lower fraction of rewarded trials at the early stage of learning (Figure 6H, n (eOPN3) = 9 mice, n (mCherry) = 8 mice). Furthermore, the lever-press movements of eOPN3 group were less reproducible (Figure 6I, see Figure S7K for other behavioral parameters). Given the incomplete inactivation by eOPN3, the results are lower-bound estimates of the true effects of cholinergic signaling in learning. These results suggest that cholinergic projections to the cortex play an important role in the acquisition of new motor skills. Then, could the behavior be improved by artificially elevating cortical cholinergic activity? To test this idea, we bilaterally injected AChR agonists (3 μM nicotine and 0.5 mM carbamoylcholine chloride) into M1 in naive animals and trained them with the lever-press task (STAR Methods). This elevation of cholinergic signaling in M1 improved task performance and led to a significant increase in the fraction of rewarded trials (Figure 6J, n (Ago.) = 5 mice, n (Sal.) = 5 mice). Other measures of task performance were not significantly altered. Thus, even though the observed effect is mild, this result implies that cholinergic activity in normal mice during the initial stage of learning is not saturated. Rather, it is possible to facilitate initial learning by artificial activation of cholinergic signaling. Manipulation of cholinergic signaling alters the activity during movements of VIP- and SOM-INs in motor learning The learning-related dynamics observed in IN subtypes and cholinergic inputs to the cortex are consistent with a simple circuit model that modulates the gain and plasticity of visual cortex neurons during locomotion (Fu et al., 2014, 2015). Our results suggest that this mechanism is engaged globally across cortex during motor learning (Figure 7A). At the initial stage of learning, the cholinergic system is highly active during movements, driving a strong activation of VIP-INs, which in turn inhibit SOM-INs. With learning, the cholinergic system becomes less engaged. The decreased excitation from cholinergic inputs attenuates VIP-IN activity, allowing a stronger activation of SOM-INs during movements. This model could be a mechanism behind our previous observation that decreased inhibition from SOM-INs to excitatory neurons during learning enhances excitatory neuron plasticity and permits motor learning (Chen et al., 2015b). To examine whether the cholinergic system modulates the activity of IN subtypes during motor learning, we pharmacologically manipulated cholinergic signaling in the right M1, while imaging the activity of VIP- and SOM-INs in both naive and expert animals. According to the model above, artificially suppressing the cholinergic system in naive mice should decrease and increase the activity of VIP-INs and SOM-INs during movements, respectively. In contrast, boosting cholinergic signaling in expert mice should elevate and suppress the activity of VIP-INs and SOM-INs, respectively. To test the first prediction, we injected a cocktail of AChR antagonists (3 mM mecamylamine and 100 μM atropine) into the right M1 in one training session during the first two days of training (STAR Methods). Antagonist sessions and control sessions were alternated across animals to balance the potential learning effects on neural activity in both groups. At the naive stage, blocking cholinergic signaling with AChR antagonists significantly decreased the activity of VIP-INs (p = 0.0024, mixed-effects model, 775 VIP-INs from 10 mice, Figures 7B–7F) and increased the activity of SOM-INs (p = 0.0062, mixed-effects model, 1128 SOM-INs from 11 mice, Figures 7G–7K), consistent with our prediction. Similar to the changes during learning, the majority of VIP-INs with decreased activity were movement-activated in the control session and became non-modulated with the application of AChR antagonists (Figure S8B). In SOM-INs, most neurons showing increased activity were previously non-modulated during movements but became activated with AChR antagonists (Figure S8D). Concurrent with the alteration of IN activity, blocking cholinergic signaling also mildly impaired learning at the naive stage, resulting in a lower fraction of rewarded trials (Figure S8I). We further examined the roles of different AChR subtypes in mediating the modulation of VIP-IN activity in naive mice by applying nicotinic AChR (nAChR) antagonist (3 mM mecamylamine) and muscarinic AChR (mAChR) antagonist (100 μM atropine) in separate groups of animals. Applying nAChR antagonist alone significantly reduced the activity of VIP-INs during movements (Figures 7L–7P), which largely recapitulated the effects of injecting a cocktail of nAChR and mAChR antagonists (Figures 7B–7F). In contrast, mAChR antagonist did not robustly reduce the VIP-IN activity (Figures 7Q–7U), suggesting that cholinergic signaling acts on VIP-INs mainly through nAChRs. To test the second prediction, we activated cholinergic signaling in the right M1 with AChR agonists (3 μM nicotine and 0.5 mM carbamoylcholine chloride) in expert mice. This manipulation altered the activity of INs during movements in the direction opposite to the antagonist experiments in naive animals above. Applying AChR agonists significantly increased the activity of VIP-INs (p = 0.0063, mixed-effects model, 995 VIP-INs from 11 mice, Figures 8A–8E) and decreased the activity of SOM-INs during movements (p = 0.0166, mixed-effects model, 762 SOM-INs from 10 mice, Figures 8F–8J). The majority of VIP-INs showing increased activity were non-modulated in expert mice but became activated during movements with AChR agonists (Figure S8F), and the majority of SOM-INs with decreased activity were movement-activated but became non-modulated with cholinergic activation (Figure S8H). Repeating the same analyses on the neurons pooled from randomly selected 50% of animals generated similar results (Figures S8A–S8H), indicating that the results are consistent across animals. We noticed that applying AChR agonists slightly increased the movement stereotypy in expert animals (Figure S8J). This result is similar to an observation in songbirds (Jaffe and Brainard, 2020) and suggests that ACh may reduce the variability of the activity of the motor cortex during well-learned movements. Taken together, these results support our model (Figure 7A) and demonstrate that the movement-related dynamics of cortical IN subtypes are modulated by cholinergic inputs during motor learning. Discussion Previous studies have revealed distinct roles of IN subtypes in M1 during motor learning (Adler et al., 2019; Chen et al., 2015b; Cichon and Gan, 2015; Donato et al., 2013). Here, we extended these studies by systematically monitoring the activity from IN subtypes at both macroscale and microscale throughout motor learning. Equipped with the large-scale activity monitoring with wide-field calcium imaging, we found that motor learning induced subtype-specific changes in IN activity throughout the cortex. The initial phase of learning accompanied high activity of VIP-INs and low activity of SOM-INs during movements. As various task events temporally overlapped, such as the auditory cues, lever-press movements, and rewards, we constructed a generalized linear model to dissect the contributions of individual behavioral events to neural dynamics. Although this is a common approach utilized in many recent studies, a more complex task design with a better temporal separation between task events could further clarify the relationships between behavioral events and dynamics of IN subtypes during learning. VIP-INs have been demonstrated to be critical in state-dependent modulation and associative learning by suppressing other IN subtypes, mediating disinhibition on excitatory neurons (Donato et al., 2013; Fu et al., 2014, 2015; Gasselin et al., 2021; Krabbe et al., 2019; Letzkus et al., 2015; Pi et al., 2013; Zhang et al., 2014). Our results suggest that a similar disinhibition mechanism is also adopted in motor learning and this occurs globally throughout the cortex. Due to the limited effective window of DREADD, we only assessed the function of VIP-INs in the initial learning phase. Other manipulation methods that can achieve more extended periods (e.g., optogenetics) will be beneficial to study the function of VIP-INs in developing behavioral features requiring long-term training (e.g., movement reproducibility). The selective disinhibition through SOM-INs could arise from a preferential connection from VIP-INs to SOM-INs in cortical circuits (Lee et al., 2013; Pfeffer et al., 2013; Pi et al., 2013). This aligns with the previous observation that learning-induced spine plasticity is largely restricted to the apical dendrites of M1 excitatory neurons, the dendritic compartment inhibited by SOM-INs (Chen et al., 2015b). However, we note that the modulation of IN activity happened immediately at the beginning of learning, while the structural plasticity occurred more gradually over days (Chen et al., 2015b). The dissociation of timing between the activity modulation and structural plasticity suggests that multiple disinhibitory mechanisms contribute to a sequential regulation of plasticity. The rapid activity modulation of INs may play a permissive role in broadly opening the plasticity window, while the slower structural plasticity of INs may provide more targeted signals by relying on the interactions with individual excitatory neurons (Bloodgood et al., 2013; Xue et al., 2014) and neighboring astrocytes (Allen and Eroglu, 2017). It should be noted that our two-photon calcium imaging experiments focused on layer 2/3, and so the learning-related modulation of INs in deeper layers remains unclear. It has been shown that SOM-INs are heterogeneously modulated in different cortical layers (Munoz et al., 2017), suggesting that a layer-specific modulation may also occur during learning. We noticed that wide-field signals of SOM-INs continued to increase throughout learning but their two-photon signals appeared to plateau earlier (Figure 2C compared to Figure 4I). As wide-field signals also contain the activity of layer 1 axons from SOM-INs in deep layers (Jiang et al., 2015; Nigro et al., 2018), the prolonged changes of wide-field signals could implicate a more gradual modulation of SOM-INs in deep layers. Future studies are required to directly examine the learning-related modulation of INs in different cortical layers. Concurrent with the changes in cortical INs, the basal forebrain cholinergic projections to the cortex were particularly active in the initial learning phase. Such learning-related modulation on the basal forebrain cholinergic system may reflect a global state change when animals are engaged in a new task. The basal forebrain cholinergic neurons are rapidly recruited by reinforcement signals (Hangya et al., 2015). Therefore, the global modulation of cortical INs may not be unique to motor learning but generally occur in various forms of learning. The strength of such global modulation may depend on the exact learning paradigms and cognitive demands of the tasks. Furthermore, recent studies have pointed out the heterogeneity of the projection targets of individual cholinergic neurons (Kim et al., 2016b; Li et al., 2018) and the cortex-wide cholinergic signaling in spontaneous activity (Lohani et al., 2020). Directly characterizing the cortex-wide cholinergic activity with simultaneous large-scale recordings during task engagement will be of particular interest in future studies. To determine the contributions of cholinergic inputs to motor learning, we used two different methods to impair the cholinergic system. We note that both manipulation methods have their own caveats: chronic ablation could induce compensatory mechanisms, and optogenetic inactivation with eOPN3 is likely incomplete. Therefore, the results probably underestimate the true effects of cholinergic signaling in learning. Nevertheless, we observed robust impairments in motor learning from different manipulation methods applied to the basal forebrain cholinergic system, suggesting basal forebrain cholinergic inputs to the cortex play an important role in learning new motor skills. Furthermore, the deficits in developing movement reproducibility caused by the lack of cholinergic inputs recapitulate the behavioral effects of M1 lesion before training in the same lever-press task (Peters et al., 2014). Manipulation of cholinergic signals also affected IN activity both in the initial and expert learning phases, suggesting the involvement of cholinergic signaling in modulating IN activity during motor learning. While the current study focused on the net effect of cholinergic inputs on IN activity during motor learning, ACh can act directly on nearly every type of cortical neurons through both mAChRs and nAChRs (Colangelo et al., 2019; Rudy and Munoz, 2014; Yao et al., 2021). Therefore, ACh likely acts at multiple loci in cortex during learning, including both direct effects on INs and indirect modulation through networks. For example, cholinergic inputs can excite VIP-INs through at least 3 mechanisms: via direct depolarization through AChRs on VIP-INs (Arroyo et al., 2012; Chen et al., 2015a), by enhancing excitatory presynaptic inputs onto VIP-INs (Lee et al., 2013; Williams and Holtmaat, 2019; Zhang et al., 2014), and by suppressing other IN subtypes that inhibit VIP-INs (Karnani et al., 2016; Zhang et al., 2014). Furthermore, the cholinergic system can also modulate excitatory neurons (Colangelo et al., 2019) and thalamocortical inputs (Kruglikov and Rudy, 2008). Besides the strong VIP-IN activation, the high activity of cholinergic inputs in the initial learning phase may also suppress the recurrent cortical activity and enhance thalamocortical inputs (Hasselmo and Sarter, 2011). All these ACh-mediated modulations may work together to facilitate the integration of task-relevant information in the cortex (Ballinger et al., 2016; Minces et al., 2017). Thus, it is likely that the cholinergic system contributes to motor learning not only through the modulation of VIP-INs but through the regulation of various circuit components. A more comprehensive understanding of how cholinergic signaling modulates the circuit dynamics could be aided by computational models of cortical circuits with detailed characterization of the expression pattern of AChRs in each circuit component (Ramaswamy et al., 2018; Saunders et al., 2018; Zeisel et al., 2018). In addition to the cholinergic system, a mixture of different inputs, such as other neuromodulators (Hattori et al., 2017) and long-range inputs from other brain regions (Ährlund-Richter et al., 2019; Makino and Komiyama, 2015; Wall et al., 2016; Williams and Holtmaat, 2019; Zhang et al., 2014), can drive the learning-related modulation in cortical INs. A detailed anatomical and functional characterization of inputs to IN subtypes will help complete the picture of underlying mechanisms. Neuromodulatory control of IN subtypes likely subserves a signal to open the temporal window of plasticity for excitatory circuits. The global nature of this signaling makes it unlikely for this mechanism to provide the spatial specificity with regards to which synapses in which brain regions become plastic. However, we note that motor learning evokes a reorganization of cortex-wide macroscale activity patterns of excitatory neurons (Makino et al., 2017). Therefore, motor learning probably involves spatially distributed synaptic reorganizations, and the global signaling described in this study may be best suited to allow temporal coordination of plasticity across spatial scales. Furthermore, we acknowledge that the behavioral effects of various manipulations in this study are relatively mild, slowing learning rather than blocking it completely. We believe that the mechanisms we describe here modulate the degree of learning rather than serving as a binary gating switch. Taken together, our results reveal a global and subtype-specific modulation on cortical INs regulated through the cholinergic system during motor learning, and provide insights into how different IN subtypes contribute to learning-induced reorganization in excitatory circuits. STAR Methods RESOURCE AVAILABILITY Lead Contact Further information and requests for resources, reagents, and data should be directed to and will be fulfilled by the lead contact Takaki Komiyama (tkomiyama@ucsd.edu). Materials Availability This study did not generate new unique reagents. Code and Data Availability Analyzed data have been deposited at Mendeley Data (DOI: 10.17632/tcnk38zkyz.1) and are publicly available as of the date of publication. The DOI is also listed in the key resources table. The raw data are too large to be deposited in a public repository, but will be shared by the lead contact upon request. The custom MATLAB codes have been deposited at https://github.com/CRen2333/InhibitoryNeuron_LeverPress.git and the link is also listed in the key resources table. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. EXPERIMENTAL MODEL AND SUBJECT DETAILS Mouse lines All procedures were performed following protocols approved by the UCSD Institutional Animal Care and Use Committee and guidelines of the National Institute of Health. Mice were acquired from Jackson laboratories, including PV-Cre [JAX:017320], SOM-Cre [JAX:013044], VIP-Cre [JAX:010908], SOM-Flp [JAX:028579], Ai95 [JAX:024105], and ChAT-Cre [JAX:006410], and used to generate double transgenic mice. Mice were group-housed in disposable plastic cages with standard bedding in a room with a reversed light cycle (12 h-12 h). Experiments were performed during the dark period. Both male and female healthy adult mice (6 weeks or older) were used. Mice had no prior history of experimental procedures that could affect the results. For manipulation experiments, littermates were randomly assigned to experimental and control groups. METHOD DETAILS Surgeries and virus injections For two-photon calcium imaging Surgical procedures were performed as previously described (Peters et al., 2014). Adult mice (6 weeks or older, male and female) were anesthetized with 1%-2% isoflurane and injected with Baytril (10 mg/kg), dexamethasone (2 mg/kg), and buprenorphine (0.1 mg/kg) subcutaneously at the beginning of surgery to prevent infection, inflammation, and discomfort. A custom-built head-plate was glued and cemented to the skull. For imaging cholinergic projections in the right S1HL and PPC, a large hexagonal craniotomy (~6 × 5.5 mm) was performed to encompass both cortical regions. For other two-photon imaging experiments, craniotomy (~3 mm in diameter) was performed over the right caudal forelimb area (0.3 mm anterior and 1.5 mm lateral from the bregma). For injections within the right M1, virus solutions were injected around the center of the caudal forelimb area at 5 sites (~250 μm deep, ~500 μm apart). Pipettes were left in the brain for 4-5 min after each injection to avoid backflow. For imaging cortical IN subtypes, virus solutions of AAV-Syn-FLEX-GCaMP6f were injected with 30-40 nL at each site over ~3 min in respective Cre lines. For confirming the effectiveness of hM4Di in VIP-INs, virus solutions of AAV-hSyn-DIO-hM4Di-mCherry and AAV-Syn-FLEX-GCaMP6f at a titer ratio of 1:1 were injected with 200 nL at each site over ~10 min in VIP-Cre mice. For characterizing the effectiveness window of hM4Di under repeated dosing of CNO, virus solutions of AAV-hSyn-DIO-hM4Di-mCherry and AAV-Syn-FLEX-GCaMP6f at a titer ratio of 1:5 were injected with 200 nL at each site over ~10 min in VIP-Cre mice. This mixture ratio was to achieve co-expression of hM4Di only in a subset of GCaMP6F-expressing VIP-INs. For the manipulation of VIP-IN activity with simultaneous imaging of SOM-INs, virus solutions of AAV-hSyn-DIO-hM4Di-mCherry (AAV-hSyn-DIO-mCherry in control animals) and AAV-EF1a-fDIO-GCaMP6f were injected with 200 nL at each site over ~10 min in VIP-Cre::SOM-Flp mice. A cohort of animals that were littermates of the hM4Di group received injections of virus solutions of AAV-hSyn-DIO-mCherry and AAV-EF1a-fDIO-GCaMP6f and served as a control group. The surgical procedures were identical between two groups. For imaging cholinergic projections to the cortex, virus solutions of AAV-hSyn-FLEX-axon-GCaMP6s were injected in the basal forebrain with ~1 μL virus solutions per hemisphere over ~10-15 min and pipettes were left in the brain for 10 min after each injection to minimize backflow. For imaging cholinergic projections to the right M1, virus solutions were injected through angled injections in ChAT-Cre mice to avoid inserting the pipette directly through the right M1 and infecting local ChAT+ neurons by backflow. Pipettes were inserted through small craniotomies (~0.5 mm) around 2.0 mm posterior and 1.7 mm lateral from the bregma at 18° relative to the vertical plane, and the pipette tips targeted the region around 0.3 mm posterior and 1.7 mm lateral from the bregma and 4.75 mm deep from the brain surface. For imaging cholinergic projections to the right S1HL and PPC, pipettes were inserted vertically through small craniotomies (~0.5 mm) around 0.3 mm posterior and 1.7 mm lateral from the bregma to 4.75 mm deep from the brain surface. For confirming the effectiveness of eOPN3, ~1.8 μL virus solutions of 50:50 mixtures of AAV-hSyn-SIO-eOPN3-mScarlet and AAV-hSyn-FLEX-axon-GCaMP6s were injected in the right basal forebrain through angled injections, as described above. After virus injections, a glass window was implanted over the craniotomy. The edges between the window and the skull were filled with Vetbond (3M). The window was further secured with cyanoacrylate glue and dental acrylic. For imaging IN subtypes, experiments were performed ~3-4 weeks after surgery. For imaging cholinergic axons, experiments were performed ~5-8 weeks after surgery. For test the effectiveness of eOPN3, experiments were performed ~7 weeks after surgery. For wide-field calcium imaging Adult mice (6 weeks or older, male and female) were anesthetized with 1%-2% isoflurane and injected with Baytril (10 mg/kg), dexamethasone (2 mg/kg), and buprenorphine (0.1 mg/kg) subcutaneously at the beginning of surgery. A custom-built head-bar was glued and cemented to the skull (~1 mm posterior to lambda). To improve the signal-to-noise ratio limited by the relatively low density of each IN subtype, we replaced most of the dorsal skull with a curved transparent glass window (Kim et al., 2016c). A large craniotomy was performed to remove most of the dorsal skull. A curved glass window consisting of a hexagonal glass plug (~8 × 7 mm) and a base (~10 × 8.5 mm) was implanted over the craniotomy. The edges between the window and the skull were filled with Vetbond. The window was secured with cyanoacrylate glue and dental acrylic. A custom-designed 3D-printed hexagonal crown (~10 mm × 10 mm) was glued to the circumference to protect the window and minimize the entry of the excitation light to the eyes during imaging. Experiments were performed ~4 weeks after surgery. For ablation of basal forebrain cholinergic neurons Adult ChAT-Cre mice (6 weeks or older, male and female) were anesthetized with 1%-2% isoflurane and injected with Baytril (10 mg/kg), dexamethasone (2 mg/kg), and buprenorphine (0.1 mg/kg) subcutaneously at the beginning of surgery. Virus solutions of AAV-EF1a-FLEX-taCaspase3 were injected bilaterally to the basal forebrain through angled injections, as described above. ~1 μL virus solutions were injected over ~10-15 min on each side. After injections, the craniotomies were sealed with Vetbond. A custom-built head-bar was glued and cemented to the skull (~1 mm posterior to lambda), and the exposed skull was covered by cyanoacrylate glue and dental acrylic. Experiments were performed ~4-5 weeks after surgery. Saline was similarly injected in control animals that were littermates of the ablation group and the surgical procedures were identical to the ablation group. For inactivation of basal forebrain cholinergic projections to the cortex with eOPN3 Surgical preparations were identical to the preparations for ablation of basal forebrain cholinergic neurons. Virus solutions of AAV-hSyn-SIO-eOPN3-mScarlet were injected bilaterally to the basal forebrain through angled injections in adult ChAT-Cre mice (6 weeks or older, male and female). Virus solutions of AAV-hSyn-DIO-mCherry were similarly injected in control animals that were littermates of the ablation group. ~1 μL virus solutions were injected over ~10-15 min on each side. Experiments were performed ~7 weeks after surgery. ~2 days before the training started, a ring of black heat shrink tube (~5 mm in height, ~2 cm in diameter) was glued to the circumference of the skull to minimize the entry of the green light (~532 nm) to the eyes during optogenetic inactivation. Experiments were performed ~7 weeks after surgery. Behavior Water restriction started 2 weeks before the behavioral training at 1 mL per day. After water restriction, mice were trained to perform the lever-press task 1 session per day for ~3 weeks under a microscope (21 days for wide-field calcium imaging and 22 days for two-photon calcium imaging). The hardware and software used for behavioral training have been previously described (Peters et al., 2014). In brief, a 6 kHz tone marked a cue period (up to 10 s), during which a successful lever press was rewarded with water (~10 μL per trial) paired with a 500 ms, 12 kHz tone, and followed by an inter-trial interval (ITI, variable duration of 8-12 s). Licking was monitored using an infrared lickometer (Island Motion Co.). A successful lever-press movement was defined as crossing two thresholds (~1.5 mm and ~ 3.375 mm below the resting position) within 200 ms. Failure to press the lever passing the two thresholds during the cue period triggered a loud white-noise sound and the start of an ITI. Lever presses during ITIs were neither rewarded nor punished. In two-photon calcium imaging, each session consisted of around 100 trials. In wide-field calcium imaging, each session consisted of 80–150 trials and was terminated when mice reached 80 successful trials or performed 150 trials, whichever came first. Sessions were binned into 4 stages in analyses (naive: sessions 1-2; early: 3-8; middle: 9-16; late: 17-22 for animals under two-photon imaging and 17-21 for animals under wide-field imaging). To examine the effects of ablating basal forebrain cholinergic neurons, both ablation and control groups performed 100 trials per session for 21 days. For chemogenetic inactivation experiments, Clozapine-N-Oxide (CNO, Enzo Life Sciences) was dissolved in deionized water to a 2.5 mg/mL concentration and injected intraperitoneally at a 10 mg/kg body weight dose 30 min before behavioral training. For the following manipulation experiments, including chemogenetic inactivation of VIP-INs, ablation of basal forebrain cholinergic neurons, optogenetic inactivation of cholinergic projections to the cortex, and pharmacological activating cholinergic signaling in naive mice, the training procedures were identical between the manipulation and control groups, and littermates belonging to two groups were trained in parallel on the same days. Lever-press movement analysis Lever traces were processed and movement bouts were identified as previously described (Peters et al., 2014). In brief, lever traces were downsampled from 10 kHz to 1 kHz and filtered with a low-pass Butterworth filter (4-pole 10 Hz). Movement bouts were detected by a velocity threshold (4.9 mm/s) using the filtered lever traces, and the onset and offset of movement bouts were refined by the lever position leaving or entering the resting period, respectively. Cued trials were defined as the trials with an at least 100-ms movement-quiescent period preceding the cue onset. Only rewarded movements in the cued trials were included in further analyses. Lever trajectories for these movements were collected from 0-2 s relative to the movement onset, which was approximately the duration of rewarded movements for all animals (2.19 ± 0.18 s, mean ± SEM). The similarity of lever trajectories across trials was computed by Pearson correlation (Figures 1D, 1E, 5F, 6F, 6I, 6J, S8I, and S8J). Licking analysis Licking bouts were identified as previously described (Komiyama et al., 2010). In brief, licking bouts were defined as no less than three continuous licks with inter-lick intervals < 300 ms. The licking bout onsets were defined as the start of individual licking bouts (Figure 3). Imaging data acquisition For two-photon calcium imaging For cortical INs, imaging was conducted with a commercial two-photon microscope (MOM, Sutter Instrument, retrofitted with a resonant galvanometer-based scanning system from Thorlabs), 16 × objective (Nikon), and 925 nm excitation light (Ti:Sapphire laser, Newport) controlled by Scanlmage (Vidrio Technologies). Images were recorded at ~28 Hz continuously, alternating between 2 depths within layer 2/3. At each depth, images were acquired at ~14 Hz with a field of view of ~590 × 635 μm with 512 × 512 pixels. Imaging was alternated between two locations in the right M1 each day such that each field was imaged every other day. For basal forebrain cholinergic projections to the right M1, images were acquired using a commercial two-photon microscope (B-Scope, Thorlabs) at ~30 Hz with a field of view of ~204 × 220 μm with 512 × 512 pixels in layer 1. For basal forebrain cholinergic projections to the right S1HL and PPC, the two fields of view were alternated between imaging sessions within the same animal. To minimize photobleaching and phototoxicity, at the early and middle stages of learning, axonal imaging was only performed in the middle two sessions within each stage. Frame times were recorded and synchronized with behavioral recordings by the Ephus software. Slow drifts in the field of view were manually corrected using reference images during imaging. For wide-field calcium imaging Wide-field calcium imaging was performed using a commercial fluorescence microscope (Axio Zoom.V16, Zeiss, objective lens (1 ×, 0.25 NA)) and a CMOS camera (ORCA-Flash4.0 V2, Hamamatsu) as previously described (Makino et al., 2017). The light source for wide-field calcium imaging was HXP 200 C (Zeiss). The filter set (000000-1021-600, Zeiss) for imaging GCaMP signals consisted of a bandpass filter for the excitation light (485 ± 17 nm), a beamsplitter (500 nm), and a tunable bandpass filter centered at 520 nm for the emission light. Images were acquired using HCImage Live (Hamamatsu) at 29.98 Hz, 512 × 512 pixels (field of view: ~8.5 × 8.5 mm, binning: 4, 16 bit) every other session of behavioral training. Each session consisted of several 5-min blocks, and every other block was imaged to minimize photobleaching. Imaging and behavioral data were acquired simultaneously and aligned off-line based on a synchronization signal. Two-photon imaging analysis ROI identification and fluorescence analysis Images were first aligned frame-by-frame using a custom MATLAB program to correct lateral movements and distortions (Hattori and Komiyama, 2021; Mitani and Komiyama, 2018). For cortical INs, ROIs were manually drawn and aligned across sessions using a custom MATLAB program (Peters et al., 2014) by visual inspection. ROIs showing a filled nucleus by GCaMP6f were excluded from all analyses. A ring-shaped ‘background ROI’ containing neuropil signals was created from the border of each neuronal ROI to a width of 6 pixels. For cholinergic axons, ROIs were first identified with Suite2P (Pachitariu et al., 2016) and then selected with visual inspection. ROIs clearly belonging to the same axon were manually combined. For each field of view, three background ROIs were manually drawn in the dark area without obvious fluorescent transients. Their averaged fluorescence time series was used as the background fluorescent trace for all axonal ROIs in the same field of view. Fluorescence analysis was processed as described (Peters et al., 2014). Briefly, pixels within each ROI were averaged to create a fluorescence time series, and the background fluorescent trace was subtracted. To estimate the time-varying baseline (f) of a fluorescence trace, the raw fluorescence trace was smoothed with a 4-min moving average window, and f was estimated based on the inactive portions of the trace. Classification of neurons showing significant changes in their activity during movements The mean activity from 0 s to 2 s after movement onset in each trial, after subtraction of the baseline (the averaged activity between −357 ms and −214 ms relative to the movement onset) was used as activity during movements for comparison (Figures 4, 5B–5E, 7, 8, S4, S5B, S5C, S5E, and S8). To determine whether a given neuron significantly changed its activity during movements with learning, the trials from the early, middle, or late stage were pooled with trials from the naive stage, forming an activity dataset. The trial identity in the dataset was shuffled 1,000 times and generated a null distribution of the activity change relative to the naive stage. If the actual activity change fell in the left (< 2.5%) or the right (> 97.5%) tail of the null distribution, the neuron was considered to decrease or increase its activity during movements, respectively (Figures 4F, 4K, S4A, S4B, S4D, and S4F). The same analyses were performed to determine whether the activity was significantly changed by pharmacological manipulation of cholinergic signaling in the right M1 (Figures 7F, 7K, 7P, 7U, 8E, 8J, S8A, S8C, S8E, and S8G). Classification of movement-modulated neurons At each learning stage, movement-modulated neurons were identified by comparing the activity during the baseline period and the activity at each frame after the movement onset by bootstrap (1,000 times). The frame was classified as movement-modulated (activated or suppressed) if its p-value was less than 0.05 (two-tailed). For a given neuron, if more than 25% of the frames after movement onset were either activated or suppressed, this neuron was classified as movement-activated or movement-suppressed, respectively (Figures S4C, S4F, S8B, S8D, S8F, and S8H). Ca event detection in cholinergic activity The Δf/f trace was first smoothed (loess, 3 s) and the first derivative (velocity) of the smoothed Δf/f trace was calculated. The inactive portion was defined as the periods when the velocity was within the standard deviation of the whole velocity trace. Events were defined if the velocity trace crossed the standard deviation of the inactive portion of the velocity trace. This method detected sharp rises in Δf/f. Events with an active period less than 167 ms were excluded from further analyses. For each event, the onset time was estimated as the time when the velocity exceeded the velocity criterion (Figure S7I and S7J). Wide-field imaging analysis Fluorescence analysis was processed as described (Makino et al., 2017). Briefly, images were first downsampled from 512 × 512 to 128 × 128 pixels. To obtain Δf/f time series for each pixel, time-varying baseline fluorescence (f) was estimated for a given time point as the 10th percentile value over 30 s around it. For the beginning and end of each imaging block, the following and preceding 15 s window was used to determine the baseline, respectively. Images across sessions from the same animals were aligned to the first session using a semi-automated method written in MATLAB. To remove hemodynamic contamination and motion artifacts, Δf/f time series for each pixel during the peri-movement epoch (−0.5 s to 2 s relative to the movement onset) were concatenated across all trials from all sessions for each animal to perform principal component analysis followed by independent component analysis (PCA-ICA) with 80 components retained (94.71% ± 0.44% of the total variance, mean ± SEM). The ICA algorithm adopted in the current study was JADER (Cardoso, 1999). Independent components corresponding to hemodynamic signals and motion artifacts were visually identified and excluded (Figure S2C), and Δf/f time series were reconstructed using the remaining components. ROI masks of cortical regions were determined by overlaying the scaled common cortical modules on the average image from the first session in each animal (Figure S2A). Common cortical modules were previously identified based on the cortex-wide activity of excitatory neurons (Makino et al., 2017). The coordinates forthe centers of these 16 modules are (in mm): M2: ± 1.2 ML, +2.3 AP; left and right S1/M1FL: ± 3.2 ML, +1.3 AP; left and right aS1BC: ± 3.8 ML, −0.4 AP; left and right M1: ± 2.6 ML, −0.4 AP; left and right S1HL: ± 1.8 ML, −1.0 AP; left and right pS1BC: ± 3.3 ML, −1.6 AP; PPC: ± 2.1 ML, −1.8 AP; aRSC: ± 0.8 ML, −1.8 AP; pRSC: ± 0.9 ML, −3.1 AP; left and right visual cortex: ± 3.0 ML, −3.3 AP. Pixels overlapping with main blood vessels or included in more than one module were excluded from ROI masks. For each cortical region, Δf/f time series was computed as the mean of the pixel values within its ROI mask. For the activity during movement epochs, the averaged activity between −367 ms and −233 ms relative to the movement onset (baseline) was subtracted from each trial, then the mean activity from 0 s to 2 s after movement onset in each trial was used for comparison between stages (Figure 2C). The activity amplitude and duration of each cortical region were analyzed using the activity from 0 s to 2 s after movement onset after subtracting the baseline period in each trial. The amplitude was measured by the maximum Δf/f value (peak) in each trial and the mean of all peak values was taken at each learning stage within each animal. The duration was measured by the full width at half maximum (FWHM) in each trial and the median of all FWHM values was taken at each learning stage within each animal. In a small fraction of trials (3.45% ± 0.47%, 3.23% ± 0.45%, 3.45% ± 0.47% for VIP-, SOM- and PV-INs, n (VIP) = 11 mice, n (SOM) = 12 mice, n (PV) = 11 mice, mean ± SEM), the activity did not return to half maximum within 2 s after movement onset, and in these cases the duration was defined as the interval between the time that the value went above half maximum to 2 s after movement onset. GFP control experiments for wide-field signals To confirm that the signals detected with our approach were calcium signals from GCaMP6f rather than artifacts, we performed control experiments using transgenic mice with GFP expressed in VIP-INs and imaging during the lever-press task (Figures S2C–S2F). We chose VIP-INs as they are the least abundant among the three IN subtypes and therefore likely to be most susceptible to artifacts with wide-field calcium imaging. The experiments and data analysis were identical to GCaMP6f-expressing animals. The wide-field signals from both GFP- and GCaMP6f-expressing mice were passed to PCA-ICA analysis to exclude hemodynamic components and then reconstructed. Using reconstructed signals, we first calculated the mean fluorescence changes (Δf/f) from 0 s to 2 s after movement onset of each cortical region in both groups, after subtracting the baseline period. Then, the values across cortical regions were averaged for each mouse to generate a single value for the dorsal cortex, which was further normalized to the mean value across GCaMP6f-expressing mice. For the 3 GFP-expressing mice, the relative fluorescence changes were −0.15%, 1.56%, and 5.55%, respectively (mean ± SD: 2.32% ± 2.92%). Given the heterogeneity across cortical regions, we also compared individual cortical regions between GFP- and GCaMP6f-expression mice separately, using the mean fluorescence changes during movements calculated as mentioned above. For each cortical region, the values were first averaged across GFP-expressing mice, and then normalized to the mean value across GCaMP6f-expressing mice. Using this comparison method, the relative fluorescence changes of individual cortical regions ranged 1.37% ± 6.23% (mean ± SD, M2: −1.91%, I-M1/S1FL: 10.10%, r-M1/S1FL: −5.09%, I-M1: 6.52%, r-M1: 6.82%, I-aS1BC: 1.61%, r-aS1BC: −10.16%, I-S1HL: 6.33%, r-S1HL: 10.24%, PPC: 2.08%, I-pS1BC: −2.97%, r-pS1BC: −8.24%, aRSC: 5.24%, pRSC: 1.65%, I-Visual: 3.74%, r-Visual: 4.07%). These results demonstrate that the large majority of fluorescence changes we report are indeed calcium signals. Generalized linear model To estimate the contributions of different behavioral events to neural activity, we constructed a generalized linear model (Musall et al., 2019; Pinto and Dan, 2015) to predict the activity of each inhibitory neuron subtype in 16 cortical regions using a set of behavioral variables as predictors. All behavioral variables were downsampled to 30 Hz to match the imaging frame rate. The predictors included both analog and binary predictors related to lever-press movements, licking, auditory cue, and reward. Predictors of each binarized behavioral variable consisted of a binary event trace containing pulses at the occurrences of the relevant event, and its time-shifted copies, each shifted in time by one frame relative to the original binary trace. For binary motor events, including the onsets of lever-press movements and licking bouts, the time-shifted copies spanned the frames from −0.5 s to +2 s relative to each event. For sensory stimuli, including the auditory cue and reward delivery sound, the time-shifted copies spanned all frames from stimulus onset until 0.5 s after each event. Analog behavioral variables, including the lever speed and licking rate, were not time-shifted. The full model had a total of 186 predictors and the general bias term. We used ridge regression (Karabatsos, 2018; Musall et al., 2019) to prevent overfitting and assessed the model performance using tenfold cross-validated correlations between the predicted and true imaged activity. The contributions of each behavioral event were assessed by calculating the predicted activity during movements as mentioned in the previous section using individual behavioral events and their coefficients in full models, and comparing the predicted activity with the true activity (Figure 3). Pupillometry recordings The right eye was monitored using a commercial camera (DMK 23U618, The Imaging Source) mounted with a CCTV lens (35mm f/1.7, Fujian) through 4-5 extension rings (5mm, RioRand). Images were acquired using IC Capture 2.4 (The Imaging Source) at 15 Hz, 640 × 480 pixels. The light source for the camera was an IR LED illuminator (IRINB04L, JCHENG). Frame times were recorded and synchronized with behavioral recordings by the Ephus software. To prevent complete pupil dilation in darkness, a blue led array lamp (LIU470A, Thorlabs) softened with several layers of Kimwipes paper (Kimberly-Clark Professional) was placed ~60 cm away in front of animals to provide ambient light in the task. The space between the craniotomy window and the microscope objective was enclosed with a blackout material (Thorlabs) to protect calcium imaging from contamination from the IR and ambient light. After the end of each session, the blue led array lamp was blocked with a black cap, and the fully dilated pupil in darkness was imaged for ~30 s as the baseline. The pupil diameter was fitted off-line with custom codes written in MATLAB. Before fitting images from the entire session, every 100th frame was collected and parameters for pupil fitting were manually tuned on this subset. Each frame was denoised with 2D-meidan filtering (3 × 3 neighborhood around) and segmented using Otsu’s multilevel threshold method. The darkest cluster containing more than 5000 pixels was used to generate a binary image for further refinement. The hole corresponding to the corneal reflection of the IR illuminator was filled first, and then the largest connected region in the binary image was retained as the mask for pupil edge detection. Edges were detected using Canny method. The pupil center was roughly estimated using the mean coordinates of the mask. Edges corresponding to the eyelids and small reflection spots were eliminated based on the distance and angle relative to the pupil center. The remaining edges were visually inspected, and those not aligned with the true pupil boundary were cropped out by ROIs drawn manually. The cleaned edges were fitted with an ellipse with Hough transform and the length of the long axis of the ellipse was taken as pupil diameter (Figure S6A). The parameters tuned on this subset of frames, including the number of pixel clusters, ROIs, and fitting parameters in Hough transform, were used to fit the entire session. For baseline recording, parameters were directly tuned on the entire image set (~450 frames for 30 s). ~40 images with good fitting were manually selected, and their averaged pupil diameter was used as baseline diameter. Pupil diameter during task performance was normalized to the baseline diameter within each session to control for individual variances in pupil size and distance to the camera. For pupil diameter during the peri-movement epoch, the averaged pupil diameter between −500 ms and 0 ms relative to movement onset at the naive stage was subtracted to control for the individual variances in the sensitivity to the ambient light (Figure S6B). Optogenetic inactivation of cholinergic projections to the cortex with eOPN3 Optogenctic inactivation was performed seven weeks after injections throughout training. The position of the laser (GL532T3, Shanghai Laser & Optics Century) was adjusted to ensure the green light covered the entire dorsal skull. In every session, the green light (~532 nm) was delivered through the skull to the entire dorsal cortex. Given that the activity already partially recovered during 10-12 min post light delivery (Figures S7I and S7J), the green light was delivered for 15 s every 5 min spanning the entire training session to achieve a continuous inactivation. The skull transmission was measured by placing moisturized fresh skull samples between the light source and power meter. The measured skull transmission was ~50%−55%. Therefore, to achieve a light intensity of ~2 mW/mm2 at the brain surface, we adjusted the light intensity to ~4 mW/mm2 at the skull surface. Pharmacological manipulation of cholinergic inputs in the right M1 Imaging fields of view were first determined under the two-photon microscope before AChR drug injections. Animals were then placed under light anesthesia with 0.5%-0.8% isoflurane, and a small craniotomy was opened ~1mm away from the center of the predetermined field of view. This was to avoid heating the field of view during drilling through the glass window. We usually drilled along the window edge to avoid window cracks. Solutions of AChR drugs were injected with a beveled pipette (~20 μm tip in diameter) through the small craniotomy at the depths of 250 μm and 500 μm. Drugs were fully dissolved in saline, and ~250 nL drug solutions were injected over 5-8 min at each depth. For naive animals, a mixture of mecamylamine (3 mM, Tocris) and atropine (100 μM, Sigma-Aldrich) was injected in one of the first two sessions. The session without injections served as controls. For expert animals, a mixture of nicotine (3 μM, Sigma-Aldrich) and carbamoylcholine chloride (0.5 mM, Sigma-Aldrich) was injected in session 20, and session 19 was used as controls. Pipettes were left in the brain for ~5 min after each injection. The small craniotomy was sealed with Vetbond and cyanoacrylate glue. Following injections, animals recovered in their home cage for 40-60 min before behavioral testing and imaging. For enhancing cholinergic signaling in naive animals, a mixture of nicotine (3 μM, Sigma-Aldrich) and carbamoylcholine chloride (0.5 mM, Sigma-Aldrich) was injected bilaterally in M1 (0.3 mm anterior and 1.5 mm lateral from the bregma) in the first two training sessions. Drug solutions were injected through small craniotomies at the depths of 250 μm and 500 μm with ~250 nL at each depth. The small craniotomy was sealed with Vetbond and cyanoacrylate glue after injections. Immunohistochemistry Mice were anesthetized (ketamine/xylazine, 150 mg/kg, 12 mg/kg body weight) and perfused transcardially with 4% paraformaldehyde. Brains were then cryoprotected in a 30% sucrose solution until brains sank. 50-60 mm coronal sections were cut with a microtome (Microm HM 430, Thermo Scientific) and blocked in a solution consisting of 10% donkey serum, 1% BSA, and 0.3% Triton X-100 in 1 × PBS for 1 h at room temperature. They were then incubated overnight at 4°C with primary antibodies (1:1000 Chicken anti-GFP, Aves Labs; 1:300 Goat anti-ChAT, Millipore; Rabbit anti-mCherry, Abcam) diluted in the blocking solution. After washing, sections were incubated in Alexa Fluor-conjugated secondary antibodies (1:1000 anti-chicken 488, Jackson Immuno Research; 1:1000 anti-goat 594, Jackson Immuno Research ; 1:1000 anti-rabbit 594, Invitrogen) for 2 h at room temperature. Slices were mounted with a CC mounting medium (Sigma-Aldrich) and imaged using a fluorescence microscope (Zen and ApoTome.2, Zeiss). QUANTIFICATION AND STATISTICAL ANALYSIS Experimenters were not blind to the experimental conditions. Statistical significance was defined by alpha pre-set to 0.05. Error bars indicate standard errors of the mean (SEM) unless noted otherwise. All the statistical details are described in the figure legends and each test was selected based on data distributions using histograms. Sample sizes were predetermined without any statistical methods but based on those generally employed in the field. Two-tailed tests were used unless noted otherwise. Multiple comparisons were corrected by false discovery rate. Formulas used in the mixed-effects model are listed below. For the behavior performance throughout training (Figures 1B, 1E, S1A–S1C): y∼1+session+(1∣animal)+(session−1∣animal) where (1|animal) and (session – 1|animal) indicate a random effect constant and a random effect slope term for each animal, session is a discrete variable representing the training session, animal is a categorical variable representing the animal identity, and y is behavioral measurement. The coefficient of session was tested against 0. For the comparison of activity and pupil diameter during movements throughout learning (Figures 2C, 4D, 4E, 4I, 4J, 6C, and S6B): y∼1+stage+(1∣animal)+(stage−1∣aniaml) where (1|animal) and (stage – 1|aniamal) indicate a random effect constant and a random effect slope term for each animal, stage is a discrete variable representing the learning stage, and y is the activity or pupil diameter during movements. The coefficient of stage was tested against 0. For the comparison of activity during movements between the hM4Di and mCherry groups (Figures 5D and 5E): y∼1+group+(1∣animal) where (1|animal) indicates a random effect constant, group is a categorical variable indicating which group the animal belonged to, and y is the activity during movements. The coefficient of group was tested against 0. For the comparison of behavior performance between the hM4Di and mCherry groups (Figures 5F, 5G, and S5D), or between the ablation and control groups (Figures 6E, 6F, and S7G), or between the eOPN3 and mCherry groups (Figures 6H, 6I, and S7K), or between the agonist and saline groups in naive mice (Figure 6J): y∼1+group+(1∣session) where (1|session) indicates a random effect constant, group is a categorical variable indicating which group the animal belonged to, and y is behavioral measurement. The coefficient of group was tested against 0. For the comparison of activity during movements and behavior performance between pharmacological manipulation and control sessions (Figures 7D, 7E, 7I, 7J, 7N, 7O, 7S, 7T, 8C, 8D, 8H, 8I, S8I, and S8J): y∼1+drug+(1∣animal)+(drug−1∣animal) where (1|animal) and (drug – 1|aniamal) indicate a random effect constant and a random effect slope term for each animal, drug is a categorical variable indicating whether the session was the manipulation or control session, and y is the activity during movements or behavioral measurement. The coefficient of drug was tested against 0. Supplementary Material 2 Movie S1 (Related to Figure 2). Single-trial cortex-wide activity of IN subtypes from example trials. Time 0 ms corresponds to the onset of lever-press movements. Note that besides the global activation after movement onset, the wide-field calcium signals also resolved spatially restricted activation on a moment-by-moment basis. 3 Acknowledgments We thank members of the Komiyama lab, especially E. Gjoni and R. Hattori for comments and discussions; A. Kim, K. O’Neil, O. Arroyo, Q. Chen, Y. Magaña, S. Jilani, L. Hall, and E. Hall for technical assistance; A. Mitani, and R. Hattori for motion correction algorithm; C. Matéo and T. Sato for help with pupillometry recordings; and E. Wang, X. Ren, and S. Liu for sharing viral vectors. This work was supported by grants from NIH (R01 NS091010, R01 EY025349, R01 DC014690, and P30 EY022589), NSF (1940202), and David & Lucile Packard Foundation to T.K. Figure 1. Motor learning task (A) Experimental setup and task structure. ITI: inter-trial interval. (B) Fraction of rewarded trials increases with learning (p < 0.0001, mixed-effects model, n = 40 mice from session 1 to 21, n = 17 mice for session 22, mean ± SEM). Gray dashed lines indicate the naive (session 1-2), early (session 3-8), middle (session 9-16), and late (session 17-22) stages. (C) Example lever trajectories in rewarded trials from one mouse. Grey lines represent trajectories from randomly selected 10 trials and black lines represent their average. Red dashed line indicates the movement onset. (D) Correlation matrix of the lever trajectories in individual trials. Each square represents the median of trial-by-trial correlation coefficients of rewarded lever-press movements within or across sessions, averaged across animals. (E) Trial-by-trial correlations of the lever trajectories within each session and across adjacent sessions increase with training (p < 0.0001 for both comparisons, mixed-effects model, mean ± SEM). These comparisons correspond to the diagonals indicated by the black and gray arrows in (D). See also Figure S1. Figure 2. Cortex-wide, subtype-specific modulation of IN activity during motor learning revealed by cell type-specific wide-field calcium imaging (A) Left, experimental setup. Right: an example field of view of wide-field calcium imaging.Scale bar, 1 mm. (B) Activity of VIP-, SOM- and PV-INs aligned to the movement onset from example cortical regions, averaged across animals (n (VIP) = 11 mice, n (SOM) = 12 mice, n (PV) = 11 mice, mean ± SEM). S1HL: primary somatosensory cortex hindlimb area. (C) Activity of VIP-, SOM- and PV-INs during movements in individual cortical regions at each learning stage (indicated by colors), measured by mean Δf/f between 0 and +2 s relative to the movement onset (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, mixed-effects model, corrected for multiple comparisons by false discovery rate, mean ± SEM). See also Figures S2, S3, and Movie S1. Figure 3. Contributions of various behavioral events to cortex-wide IN activity characterized by a generalized linear model (A) Predictors in the generalized linear model. Binary behavioral events are time-shifted to account for potential lags between the events and neural activity. (B-D) The true imaged activity and the activity predicted by the full model or individual behavioral events of each IN subtype in r-M1 as examples, aligned to the movement onset and averaged across animals (n (VIP) = 11 mice, n (SOM) = 11 mice, n (PV) = 11 mice, mean ± SEM). Note that the full model closely fit the activity profile. (E) True or predicted activity during movements averaged across animals (mean ± SEM). N: naive, E: early, M: middle, L: late. (F) Contributions of behavioral events to learning-related changes of cortex-wide IN activity during movements, compared to the naive stage (mean ± SEM). Color code in (C)-(F) is the same as in (B). Figure 4. Subtype-specific modulation of the activity during movements of individual VIP- and SOM-INs in M1 during motor learning. (A) Example fields of view of two-photon calcium imaging of VIP- and SOM-INs in the right M1. Scale bar, 100 μm. (B) Activity of VIP-INs at each learning stage aligned to the movement onset (870 VIP-INs from 7 mice). Each row represents the activity averaged across trials of individual neurons, sorted according to their activity level during movements within each stage. (C) Activity of VIP-INs averaged across movements and neurons at each learning stage aligned to the movement onset (mean ± SEM). (D) Activity during movements averaged across VIP-INs at each learning stage (p < 0.0001, mixed-effects model, mean ± SEM). (E) Distribution of changes in the activity level during movements across VIP-INs at the early, middle, and late stages compared to the naive stage (Early v.s. Naive: p < 0.0001, Middle v.s. Naive: p < 0.0001, Late v.s. Naive: p < 0.0001, mixed-effects model, corrected for multiple comparisons by false discovery rate). (F) The fraction of VIP-INs with significant changes in the activity level during movements at the late stage compared to the naive stage. (G)-(K) Same as (B)-(F) but for SOM-INs (533 SOM-INs from 10 mice). For (I), p = 0.0007, mixed-effects model. For (J), Early v.s. Naive: p = 0.0011, Middle v.s. Naive: p = 0.0011, Late v.s. Naive: p = 0.0011, mixed-effects model, corrected for multiple comparisons by false discovery rate. The activity of SOM-INs during movements is weak at the naive stage and increases with further training. See also Figure S4. Figure 5. Inactivation of VIP-INs in naive animals increases SOM-IN activity and impairs motor learning (A) Schematic of injections to selectively express hM4Di (or mCherry in controls) in VIP-INs and GCaMP6f in SOM-INs in the right M1. Scale bar, 100 μm. (B) Activity of SOM-INs at the naive stage aligned to the movement onset (619 neurons from 11 animals in the mCherry group, 559 neurons from 11 animals in the hM4Di group). Each row represents the activity averaged across trials of individual neurons, sorted according to their activity level during movements. (C) Averaged activity of SOM-INs in mCherry and hM4Di groups aligned to the movement onset (mean ± SEM). (D) Activity during movements averaged across SOM-INs in mCherry and hM4Di groups (p = 0.0186, mixed-effects model, mean ± SEM). (E) Distribution of the activity level of SOM-INs during movements in mCherry and hM4Di groups. Note the rightward shift of the hM4Di group compared to the mCherry group. (F) Behavior performance of naive animals in mCherry and hM4Di groups in CNO sessions (mixed-effects model, n (hM4Di) = n (mCherry) = 11 mice, 2 sessions per animal, mean ± SEM). Circles represent behavioral measurements of individual sessions. (G) Behavioral performance in the first 10 trials in the first session after the CNO sessions (mixed-effects model, 11 animals in each group). Circles represent behavioral measurements of individual animals. See also Figure S5. Figure 6. The basal forebrain cholinergic system is involved in motor learning (A) Schematic of injections to selectively express axon-GCaMP6s in basal forebrain cholinergic neurons. (B) Example fields of view of two-photon calcium imaging of basal forebrain cholinergic axons in the right M1, S1HL, and PPC. Scale bar, 20 μm. (C) Traces show the activity of cholinergic axons at each learning stage aligned to the movement onset (n (Ml) = 7 mice, n (SIHL) = 6 mice, n (PPC) = 6 mice, mean ± SEM). Line plots show the averaged activity of cholinergic axons during movements at each learning stage (p (M1) = 0.0016, p (S1HL) = 0.0059, p (PPC) = 0.0190, mixed-effects model, mean ± SEM). Gray lines represent individual animals. (D) Schematic of injections to selectively ablate basal forebrain cholinergic neurons bilaterally through Caspase3 expression. (E) Behavior performance of ablation and control groups (**p < 0.01, ***p < 0.001, ****p < 0.0001, mixed-effects model, n (Ablation) = 15 mice, n (Control) = 13 mice, mean ± SEM). Gray dashed lines indicate the naive (session 1-2), early (session 3-8), middle (session 9-16), and late (session 17-22) stages. (F) Trial-by-trial correlations of the lever trajectories within (left) and across (right) sessions (*p < 0.05, **p < 0.01, ***p < 0.001, mixed-effects model, mean ± SEM). (G) Schematic of injections to selectively express eOPN3 or mCherry in basal forebrain cholinergic neurons bilaterally. (H) Behavior performance of eOPN3 and mCherry groups (*p < 0.05, mixed-effects model, n (eOPN3) = 9 mice, n (mCherry) = 8 mice, mean ± SEM). (I) Trial-by-trial correlations of the lever trajectories within (left) and across (right) sessions (*p < 0.05, **p < 0.01, ***p < 0.001, mixed-effects model, mean ± SEM). (J) Naive mice receiving bilateral injections of AChR agonists (3 μM nicotine and 0.5 mM carbamoylcholine chloride) into M1 showed a higher fraction of rewarded trials than mice receiving saline injections (mixed-effects model, n (Ago.) = 5 mice, n (Sal.) = 5 mice, 2 sessions per animal, mean ± SEM). Circles represent behavioral measurements of individual sessions. Sal.: saline, Ago.: agonists. See also Figures S6 and S7. Figure 7. Pharmacological inactivation of cholinergic signaling decreases VIP- and increases SOM-IN activity during movements in naive mice (A) Schematic of the model for learning-related modulation in cortical circuits. (B) Activity of VIP-INs in the right M1 at the naive stage aligned to the movement onset, with or without local applications of AChR antagonists (775 VIP-INs from 10 mice). Each row represents the activity averaged across trials of individual neurons, sorted according to their activity level during movements within each session. Ctrl.: control sessions, Ant.: manipulation sessions with AChR antagonists. (C) Activity of VIP-INs averaged across movements and neurons in control and manipulation sessions aligned to the movement onset (mean ± SEM). (D) Activity during movements averaged across VIP-INs in control and manipulation sessions (p = 0.0024, mixed-effects model, mean ± SEM). (E) Distribution of changes in the activity level of VIP-INs during movements in the manipulation session compared to the control session. (F) Fraction of VIP-INs with significant changes in the activity level during movements in the manipulation session compared to the control session at the naive stage. (G)-(K) Same as (B)-(F) but for SOM-INs (1128 SOM-INs from 11 mice). For (I), AChR antagonists increased the activity of SOM-INs during movements in naive animals, p = 0.0062, mixed-effects model, mean ± SEM. (L)-(P) Same as (B)-(F) but in naive animals with local applications of only nAChR antagonist in the right M1 (896 VIP-INs from 6 mice). For (N), nAChR antagonist significantly reduced the activity of VIP-INs in naive animals (p = 0.0031, mixed-effects model, mean ± SEM). Ctrl.: control sessions, nAnt.: manipulation sessions with nAChR antagonist. (Q)-(U) Same as (B)-(F) but with mAChR antagonist (1066 VIP-INs from 7 mice). For (S), mAChR antagonist did not significantly decrease the mean activity of VIP-INs in naive animals (p = 0.2302, mixed-effects model, mean ± SEM). See also Figure S8. Figure 8. Pharmacological activation of cholinergic signaling increases VIP- and decreases SOM-IN activity during movements in expert mice (A) Activity of VIP-INs in the right M1 at the expert stage aligned to the movement onset, with or without local applications of AChR agonist (995 VIP-INs from 11 mice). Each row represents the activity averaged across trials of individual neurons, sorted according to their activity level during movements within each session. Ctrl.: control sessions, Ago.: manipulation sessions with AChR agonists. (B) Activity of VIP-INs averaged across movements and neurons in control and manipulation sessions aligned to the movement onset (mean ± SEM). (C) Activity during movements averaged across VIP-INs in control and manipulation sessions (p = 0.0063, mixed-effects model, mean ± SEM). (D) Distribution of changes in the activity level of VIP-INs during movements in the manipulation session compared to the control session. (E) Fraction of VIP-INs with significant changes in the activity level during movements in the manipulation session compared to the control session at the naive stage.. (F)-(J) Same as (A)-(E) but for SOM-INs (762 SOM-INs from 10 mice). For (H), AChR agonists decreased the activity of SOM-INs in expert animals (p = 0.0166, mixed-effects model, mean ± SEM). See also Figure S8. Key resources table REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Chicken anti-GFP Aves Labs Cat#GFP-1020; RRID: AB_10000240 Goat anti-ChAT Millipore Cat#AB144P; RRID: AB_2079751 Rabbit anti-mCherry Abcam Cat#ab167453; RRID: AB_2571870 Donkey anti-chicken IgG (H+L), Alexa Fluor 488 Jackson Immuno Research Cat#703-545-155; RRID: AB_2340375 Donkey anti-goat IgG (H+L), Alexa Fluor 594 Jackson Immuno Research Cat#705-585-003; RRID: AB_2340432 Donkey anti-rabbit IgG (H+L), Alexa Fluor 594 Invitrogen Cat#A-21207; RRID: AB_141637 Bacterial and virus strains AAV1-Syn-FLEX-GCaMP6f Chen et al., 2013 Addgene viral prep # 100833-AAV1; RRID: Addgene_100833 AAVDJ-hSyn-DIO-hM4Di-mCherry Krashes et al., 2011, Byungkook Lim Addgene plasmid # 44362; RRID: Addgene_44362 AAV1-hSyn-DIO-mCherry Krashes et al., 2011 Addgene viral prep # 50459-AAV1; RRID: Addgene_50459 AAVDJ-EF1a-fDIO-GCaMP6f Sciolino et al., 2019, Byungkook Lim Addgene plasmid # 128315; RRID: Addgene_128315 AAV5-hSyn-FLEX-axon-GCaMP6s Broussard et al., 2018 Addgene viral prep # 112010-AAV5; RRID: Addgene_112010 AAV1-EF1a-FLEX-taCaspase3 Yang et al., 2013, Upenn Vector Core Addgene plasmid # 45580; RRID: Addgene_45580 AAVDJ-hSyn-SIO-eOPN3-mScarlet Mahn et al., 2021, Byungkook Lim Addgene plasmid # 125713; RRID: Addgene_125713 Chemicals, peptides, and recombinant proteins Clozapine N-oxide (CNO) Enzo Life Sciences Cat#BML-NS105-0025; Mecamylamine Tocris Cat#2843 Atropine Sigma-Aldrich Cat#A0132 Nicotine Sigma-Aldrich Cat#SML1236 Carbamoylcholine chloride Sigma-Aldrich Cat#C4382 Deposited data Analyzed data This paper DOI: 10.17632/tcnk38zkyz.1 Experimental models: Organisms/strains Mouse: PV-Cre: B6.129P2-Pvalbtm1(cre)Arbr/JSOM-Cre [JAX:013044], The Jackson Laboratory RRID: IMSR_JAX:017320 Mouse: SOM-Cre: Ssttm2.1(cre)Zjh/J The Jackson Laboratory RRID: IMSR_JAX:013044 Mouse: VIP-Cre, Viptm1(cre)Zjh/J The Jackson Laboratory RRID: IMSR_JAX:010908 Mouse: SOM-Flp: Ssttm3.1(flpo)Zjh/J The Jackson Laboratory RRID: IMSR_JAX:028579 Mouse: Ai95: B6;129S-Gt(ROSA)26Sortm95.1(CAG-GCaMP6f)Hze/J The Jackson Laboratory RRID: IMSR_JAX:024105 Mouse: ChAT-Cre: B6;129S6-Chattm2(cre)Lowl/J The Jackson Laboratory RRID: IMSR_JAX:006410 Software and algorithms MATLAB MathWorks RRID: SCR_001622 Fiji Schindelin et al., 2012 RRID: SCR_002285 LabView National Instruments RRID: SCR_014325 Bpod Sanworks https://sites.google.com/site/bpoddocumentation/home?authuser=0 Arduino IDE Arduino https://www.arduino.cc/en/software HCImage Live Hamamatsu RRID: SCR_015041 ScanImage Vidrio Technologies RRID: SCR_014307 IC Capture The Imaging Source RRID: SCR_016047 Zen Zeiss RRID: SCR_013672 JADER Cardoso, 1999 https://www.mathworks.com/matlabcentral/mlc-downloads/downloads/submissions/67527/versions/3/previews/jadeR.m/index.html ridgeMML Karabatsos, 2018; Musall et al., 2019 http://churchlandlab.labsites.cshl.edu/code Custom MATLAB code This paper https://github.com/CRen2333/InhibitoryNeuron_LeverPress.git Other Axio Zoom. V16 Zeiss https://www.zeiss.com/microscopy/int/products/stereo-zoom-microscopes/axio-zoom-v16.html Illuminator HXP 200C Zeiss 435716-0000-000 ORCA-Flash4.0 V2 Hamamatsu https://www.hamamatsu.com/us/en/product/cameras/cmos-cameras.html Movable objective microscope (MOM) Sutter Instrument RRID: SCR_018860 B-Scope Thorlabs https://www.thorlabs.com/thorproduct.cfm?partnumber=B-SCOPE Ti:Sapphire laser (Mai Tai HP) Newport MTEV HP1040S Monochrome industrial camera The Imaging Source DMK 23U618 TV lens (35mm f/1.7) Fujian https://www.amazon.com/Fujian-Mount-Camera-Adapter-bundle/dp/B075RZT7P8 Green laser 532 nm Shanghai Laser & Optics Century GL532T3 Arduino Arduino https://www.arduino.cc/ Highlights Global and subtype-specific modulation of IN activity during motor learning VIP-INs (SOM-INs) show strong (weak) activation in the initial learning phase Global modulation of the cholinergic inputs across cortical regions during learning Perturbation of the cholinergic system alters IN activity and impairs learning This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Declaration of Interests The authors declare no competing interests. References Adler A , Zhao R , Shin ME , Yasuda R , and Gan WB (2019). Somatostatin-expressing interneurons enable and maintain learning-dependent sequential activation of pyramidal neurons. Neuron 102 , 202–216.e7 doi: 10.1016/j.neuron.2019.01.036.30792151 Ährlund-Richter S , Xuan Y , van Lunteren JA , Kim H , Ortiz C , Pollak Dorocic I , Meletis K , and Carlén M (2019). A whole-brain atlas of monosynaptic input targeting four different cell types in the medial prefrontal cortex of the mouse. Nat. Neurosci 22 , 657–668 doi: 10.1038/s41593-019-0354-y.30886408 Allen NJ , and Eroglu C (2017). Cell biology of astrocyte-synapse interactions. Neuron 96 , 697–708 doi: 10.1016/j.neuron.2017.09.056.29096081 Arriaga M , and Han EB (2019). Structured inhibitory activity dynamics in new virtual environments. Elife 8 , e47611 doi: 10.7554/eLife.47611.31591960 Arroyo S , Bennett C , Aziz D , Brown SP , and Hestrin S (2012). Prolonged disynaptic inhibition in the cortex mediated by slow, non-α7 nicotinic excitation of a specific subset of cortical interneurons. J. Neurosci 32 , 3859–3864 doi: 10.1523/JNEUROSCI.0115-12.2012.22423106 Ballinger EC , Ananth M , Talmage DA , and Role LW (2016). Basal forebrain cholinergic circuits and signaling in cognition and cognitive decline. Neuron 91 , 1199–1218 doi: 10.1016/j.neuron.2016.09.006.27657448 Bicks LK , Yamamuro K , Flanigan ME , Kim JM , Kato D , Lucas EK , Koike H , Peng MS , Brady DM , Chandrasekaran S , (2020). Prefrontal parvalbumin interneurons require juvenile social experience to establish adult social behavior. Nat. Commun 11 , 1003 doi: 10.1038/s41467-020-14740-z.32081848 Bloodgood BL , Sharma N , Browne HA , Trepman AZ , and Greenberg ME (2013). The activity-dependent transcription factor NPAS4 regulates domain-specific inhibition. Nature 503 , 121–125 doi: 10.1038/nature12743.24201284 Broussard GJ , Liang Y , Fridman M , Unger EK , Meng G , Xiao X , Ji N , Petreanu L , and Tian L (2018). In vivo measurement of afferent activity with axon-specific calcium imaging. Nat. Neurosci 21 , 1272–1280 doi: 10.1038/s41593-018-0211-4.30127424 Cardoso JF (1999). High-order contrasts for independent component analysis. Neural Comput. 11 , 157–192 doi: 10.1162/089976699300016863.9950728 Chen N , Sugihara H , and Sur M (2015a). An acetylcholine-activated microcircuit drives temporal dynamics of cortical activity. Nat. Neurosci 18 , 892–902 doi: 10.1038/nn.4002.25915477 Chen SX , Kim AN , Peters AJ , and Komiyama T (2015b). Subtype-specific plasticity of inhibitory circuits in motor cortex during motor learning. Nat. Neurosci 18 , 1109–1115 doi: 10.1038/nn.4049.26098758 Chen T-W , Wardill TJ , Sun Y , Pulver SR , Renninger SL , Baohan A , Schreiter ER , Kerr R. a , Orger MB , Jayaraman V , (2013). Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature 499 , 295–300 doi: 10.1038/nature12354.23868258 Cichon J , and Gan W-B (2015). Branch-specific dendritic Ca2+ spikes cause persistent synaptic plasticity. Nature 520 , 180–185 doi: 10.1038/naturel4251.25822789 Colangelo C , Shichkova P , Keller D , Markram H , and Ramaswamy S (2019). Cellular, synaptic and network effects of acetylcholine in the neocortex. Front. Neural Circuits 13 , 24 doi: 10.3389/fncir.2019.00024.31031601 Crouse RB , Kim K , Batchelor HM , Girardi EM , Kamaletdinova R , Chan J , Rajebhosale P , Pittenger ST , Role LW , Talmage DA , (2020). Acetylcholine is released in the basolateral amygdala in response to predictors of reward and enhances the learning of cue-reward contingency. Elife 9 , e57335 doi: 10.7554/eLife.57335.32945260 Dasgupta R , Seibt F , and Beierlein M (2018). Synaptic release of acetylcholine rapidly suppresses cortical activity by recruiting muscarinic receptors in layer 4. J. Neurosci 38 , 5338–5350 doi:10.1523/JNEUROSCI.0566-18.2018.29739869 Do JP , Xu M , Lee SH , Chang WC , Zhang S , Chung S , Yung TJ , Fan JL , Miyamichi K , Luo L , (2016). Cell type-specific long-range connections of basal forebrain circuit. Elife 5 , e13214 doi: 10.7554/eLife.13214.27642784 Donato F , Rompani SB , and Caroni P (2013). Parvalbumin-expressing basket-cell network plasticity induced by experience regulates adult learning. Nature 504 , 272–276 doi: 10.1038/nature12866.24336286 Fu Y , Tucciarone JM , Espinosa JS , Sheng N , Darcy DP , Nicoll RA , Huang ZJ , and Stryker MP (2014). A cortical circuit for gain control by behavioral state. Cell 156 , 1139–1152 doi: 10.1016/j.cellL.2014.01.050.24630718 Fu Y , Kaneko M , Tang Y , Alvarez-Buylla A , and Stryker MP (2015). A cortical disinhibitory circuit for enhancing adult plasticity. Elife 4 , e05558 doi: 10.7554/eLife.05558.25626167 Gasselin C , Hohl B , Vernet A , Crochet S , and Petersen CCH (2021). Cell-type-specific nicotinic input disinhibits mouse barrel cortex during active sensing. Neuron 109 , 778–787.e3. doi: 10.1016/j.neuron.2020.12.018.33472037 Guo W , Robert B , and Polley DB (2019). The cholinergic basal forebrain links auditory stimuli with delayed reinforcement to support learning. Neuron 103 , 1164–1177.e6 doi:10.1016/j.neuron.2019.06.024.31351757 Hangya B , Ranade SP , Lorenc M , and Kepecs A (2015). Central cholinergic neurons are rapidly recruited by reinforcement feedback. Cell 162 , 1155–1168 doi: 10.1016/j.cell.2015.07.057.26317475 Hasselmo ME , and Sarter M (2011). Modes and models of forebrain cholinergic neuromodulation of cognition. Neuropsychopharmacology 36 , 52–73 doi: 10.1038/npp.2010.104.20668433 Hattori R , and Komiyama T (2021). PatchWarp: Corrections of non-uniform image distortions in two-photon calcium imaging data by patchwork affine transformations. bioRxiv 468164 doi: 10.1101/2021.11.10.468164. Date: 2021.11.13 Hattori R , Kuchibhotla KV , Froemke RC , and Komiyama T (2017). Functions and dysfunctions of neocortical inhibitory neuron subtypes. Nat. Neurosci 20 , 1199–1208 doi: 10.1038/nn.4619.28849791 He M , Tucciarone J , Lee S , Osten P , Rudy B , Huang Correspondence J , Nigro MJ , Kim Y , Levine JM , Kelly SM , (2016). Strategies and tools for combinatorial targeting of GABAergic neurons in mouse cerebral cortex. Neuron 91 , 1–16 doi: 10.1016/j.neuron.2016.08.021.27387643 Hippenmeyer S , Vrieseling E , Sigrist M , Portmann T , Laengle C , Ladle DR , and Arber S (2005). A developmental switch in the response of DRG neurons to ETS transcription factor signaling. PLoS Biol. 3 , 0878–0890 doi: 10.1371/journal.pbio.0030159. Jaffe PI , and Brainard MS (2020). Acetylcholine acts on songbird premotor circuitry to invigorate vocal output. Elife 9 , e53288 doi: 10.7554/eLife.53288.32425158 Jiang X , Shen S , Cadwell CR , Berens P , Sinz F , Ecker AS , Patel S , and Tolias AS (2015). Principles of connectivity among morphologically defined cell types in adult neocortex. Science 350 , aac9462 doi: 10.1126/science.aac9462.26612957 Karabatsos G (2018). Marginal maximum likelihood estimation methods for the tuning parameters of ridge, power ridge, and generalized ridge regression. Commun. Stat. Simul. Comput 47 , 1632–1651 doi: 10.1080/03610918.2017.1321119. Karnani MMM , Jackson J , Ayzenshtat I , Tucciarone J , Manoocheri K , Snider WGG , and Yuste R (2016). Cooperative subnetworks of molecularly similar interneurons in mouse neocortex. Neuron 90 , 86–100 doi: 10.1016/j.neuron.2016.02.037.27021171 Kepecs A , and Fishell G (2014). Interneuron cell types are fit to function. Nature 505 , 318–326 doi: 10.1038/nature12983.24429630 Kim D , Jeong H , Lee J , Ghim JW , Her ES , Lee SH , and Jung MW (2016a). Distinct roles of parvalbumin- and somatostatin-expressing interneurons in working memory. Neuron 92 , 902–915 doi: 10.1016/j.neuron.2016.09.023.27746132 Kim J , Jung A , Jeong D , Choi I , Kim K , Shin S , Kim SJ , and Lee XS (2016b). Selectivity of neuromodulatory projections from the basal forebrain and locus ceruleus to primary sensory cortices. J. Neurosci 36 , 5314–5327 doi: 10.1523/JNEUROSCI.4333-15.2016.27170128 Kim TH , Zhang Y , Lecoq J , Jung JC , Li J , Zeng H , Niell CM , and Schnitzer MJ (2016c). Long-term optical access to an estimated one million neurons in the live mouse cortex. Cell Rep. 17 , 3385–3394 doi: 10.1016/j.celrep.2016.12.004.28009304 Komiyama T , Sato TR , O’Connor DH , Zhang Y-X , Huber D , Hooks BM , Gabitto M , and Svoboda K (2010). Learning-related fine-scale specificity imaged in motor cortex circuits of behaving mice. Nature 464 , 1182–1186 doi: 10.1038/nature08897.20376005 Krabbe S , Paradiso E , Aquin S , Bitterman Y , Courtin J , Xu C , Yonehara K , Markovic M , Müller C , Eichlisberger T , (2019). Adaptive disinhibitory gating by VIP interneurons permits permits associative learning. Nat. Neurosci 22 , 1834–1843 doi: 10.1038/s41593-019-0508-y.31636447 Krashes MJ , Koda S , Ye CP , Rogan SC , Adams AC , Cusher DS , Maratos-Flier E , Roth BL , and Lowell BB (2011). Rapid, reversible activation of AgRP neurons drives feeding behavior in mice. J. Clin. Inves 121 , 1424–1428 doi: 10.1172/JCI46229. Kruglikov I , and Rudy B (2008). Perisomatic GABA release and thalamocortical integration onto neocortical excitatory cells are regulated by neuromodulators. Neuron 58 , 911–924 doi: 10.1016/j.neuron.2008.04.024.18579081 Kuchibhotla KV , Gill JV , Lindsay GW , Papadoyannis ES , Field RE , Sten TAH , Miller KD , and Froemke RC (2016). Parallel processing by cortical inhibition enables context-dependent behavior. Nat. Neurosci 20 , 62–71 doi: 10.1038/nn.4436.27798631 Lee S , Kruglikov I , Huang ZJ , Fishell G , and Rudy B (2013). A disinhibitory circuit mediates motor integration in the somatosensory cortex. Nat. Neurosci 16 , 1662–1670 doi: 10.1038/nn.3544.24097044 Letzkus JJ , Wolff SBE , and Lu A (2015). Disinhibition, a circuit mechanism for associative learning and memory. Neuron 88 , 264–276 doi: 10.1016/j.neuron.2015.09.024.26494276 Levelt CN , and Ḧubener M (2012). Critical-period plasticity in the visual cortex. Annu. Rev. Neurosci 35 , 309–330 doi: 10.1146/annurev-neuro-061010-113813.22462544 Li X , Yu B , Sun Q , Zhang Y , Ren M , Zhang X , Li A , Yuan J , Madisen L , Luo Q , (2018). Generation of a whole-brain atlas for the cholinergic system and mesoscopic projectome analysis of basal forebrain cholinergic neurons. Proc. Natl. Acad. Sci 115 , 415–420 doi: 10.1073/pnas.1703601115.29259118 Lin S-C , Brown RE , Hussain Shuler MG , Petersen CCH , and Kepecs A (2015). Optogenetic dissection of the basal forebrain neuromodulatory control of cortical activation, plasticity, and cognition. J. Neurosci 35 , 13896–13903 doi: 10.1523/JNEUROSCI.2590-15.2015.26468190 Lohani S , Moberly AH , Benisty H , Landa B , Jing M , Li Y , Higley MJ , and Cardin JA (2020). Dual color mesoscopic imaging reveals spatiotemporally heterogeneous coordination of cholinergic and neocortical activity. bioRxiv 418632 doi: 10.1101/2020.12.09.418632. Date: 2021.09.21 Madisen L , Garner AR , Shimaoka D , Chuong AS , Klapoetke NC , Li L , van der Bourg A , Niino Y , Egolf L , Monetti C , (2015). Transgenic mice for intersectional targeting of neural sensors and effectors with high specificity and performance. Neuron 85 , 942–958 doi: 10.1016/j.neuron.2015.02.022.25741722 Mahn M , Saraf-Sinik I , Patil P , Pulin M , Bitton E , Karalis N , Bruentgens F , Palgi S , Gat A , Dine J , (2021). Efficient optogenetic silencing of neurotransmitter release with a mosquito rhodopsin. Neuron 109 , 1621–1635.e8 doi: 10.1016/j.neuron.2021.03.013.33979634 Makino H , and Komiyama T (2015). Learning enhances the relative impact of top-down processing in the visual cortex. Nat. Neurosci 18 , 1116–1122 doi: 10.1038/nn.4061.26167904 Makino H , Ren C , Liu H , Kim AN , Kondapaneni N , Liu X , Kuzum D , and Komiyama T (2017). Transformation of cortex-wide emergent properties during motor learning. Neuron 94 , 880–890.e8 doi: 10.1016/j.neuron.2017.04.015.28521138 Markram H , Toledo-Rodriguez M , Wang Y , Gupta A , Silberberg G , and Wu C (2004). Interneurons of the neocortical inhibitory system. Nat. Rev. Neurosci 5 , 793–807 doi: 10.1038/nrn1519.15378039 Minces V , Pinto L , Dan Y , and Chiba AA (2017). Cholinergic shaping of neural correlations. Proc. Natl. Acad. Sci. U. S. A 114 , 5725–5730 doi: 10.1073/pnas.1621493114.28507133 Mitani A , and Komiyama T (2018). Real-time processing of two-photon calcium imaging data including lateral motion artifact correction. Front. Neuroinform 12 , 98 doi: 10.3389/fninf.2018.00098.30618703 Munoz W , Tremblay R , Levenstein D , and Rudy B (2017). Layer-specific modulation of neocortical dendritic inhibition during active wakefulness. Science 959 , 954–959 doi: 10.1126/science.aag2599. Musall S , Kaufman MT , Juavinett AL , Gluf S , and Churchland AK (2019). Single-trial neural dynamics are dominated by richly varied movements. Nat. Neurosci 22 , 1677–1686 doi: 10.1038/s41593-019-0502-4.31551604 Naka A , and Adesnik H (2016). Inhibitory circuits in cortical layer 5. Front. Neural Circuits 10 , 35 doi: 10.3389/fncir.2016.00035.27199675 Nigro MJ , Hashikawa-Yamasaki Y , and Rudy B (2018). Diversity and connectivity of layer 5 somatostatin-expressing interneurons in the mouse barrel cortex. J. Neurosci 38 , 1622–1633 doi: 10.1523/JNEUROSCI.2415-17.2017.29326172 Pachitariu M , Stringer C , Dipoppa M , Schröder S , Rossi LF , Dalgleish H , Carandini M , and Harris K (2016). Suite2p: beyond 10,000 neurons with standard two-photon microscopy. bioRxiv 061507 doi: 10.1101/061507. Date: 2017..07.20 Peters AJ , Chen SX , and Komiyama T (2014). Emergence of reproducible spatiotemporal activity during motor learning. Nature 510 , 263–267 doi: 10.1038/nature13235.24805237 Pfeffer CK , Xue M , Fie M , Huang ZJ , and Scanziani M (2013). Inhibition of inhibition in visual cortex: the logic of connections between molecularly distinct interneurons. Nat. Neurosci 16 , 1068–1076 doi: 10.1038/nn.3446.23817549 Pi H-J , Hangya B , Kvitsiani D , Sanders JI , Huang ZJ , and Kepecs A (2013). Cortical interneurons that specialize in disinhibitory control. Nature 503 , 521–524 doi: 10.1038/nature12676.24097352 Pinto L , and Dan Y (2015). Cell-Type-Specific Activity in Prefrontal Cortex during Goal-Directed Behavior. Neuron 87 , 437–450 doi: 10.1016/j.neuron.2015.06.021.26143660 Ramaswamy S , Colangelo C , and Markram H (2018). Data-driven modeling of cholinergic modulation of neural microcircuits: bridging neurons, synapses and network activity. Front. Neural Circuits 12 , 77 doi: 10.3389/fncir.2018.00077.30356701 Reimer J , McGinley MJ , Liu Y , Rodenkirch C , Wang Q , McCormick DA , and Tolias AS (2016). Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex. Nat. Commun 7 , 13289 doi: 10.1038/ncommsl3289.27824036 Rossi J , Balthasar N , Olson D , Scott M , Berglund E , Lee CE , Choi MJ , Lauzon D , Lowell BB , and Elmquist JK (2011). Melanocortin-4 receptors expressed by cholinergic neurons regulate energy balance and glucose homeostasis. Cell Metab. 13 , 195–204 doi: 10.1016/j.cmet.2011.01.010.21284986 Roth BL (2016). DREADDs for neuroscientists. Neuron 89 , 683–694 doi: 10.1016/j.neuron.2016.01.040.26889809 Rudy B , and Munoz W (2014). Spatiotemporal specificity in cholinergic control of neocortical function. Curr. Opin. Neurobiol 26 , 149–160 doi: 10.1016/j.conb.2014.02.015.24637201 Rudy B , Fishell G , Lee S , and Hjerling-Leffler J (2011). Three groups of interneurons account for nearly 100% of neocortical GABAergic neurons. Dev. Neurobiol 71 , 45–61 doi: 10.1002/dneu.20853.21154909 Saunders A , Macosko EZ , Wysoker A , Goldman M , Krienen FM , de Rivera H , Bien E , Baum M , Bortolin L , Wang S , (2018). Molecular diversity and specializations among the cells of the adult mouse brain. Cell 174 , 1015–1030.e16 doi: 10.1016/j.cell.2018.07.028.30096299 Schindelin J , Arganda-Carreras I , Frise E , Kaynig V , Longair M , Pietzsch T , Preibisch S , Rueden C , Saalfeld S , Schmid B , (2012). Fiji: An open-source platform for biological-image analysis. Nat. Methods 9 , 676–682 doi: 10.1038/nmeth.2019.22743772 Sciolino NR , Mazzone CM , Plummer NW , Evsyukova I , Amin J , Smith KG , McGee CA , Fry SA , Yang CX , Powell JM , (2019). A role for the locus coeruleus in the modulation of feeding. bioRxiv 881599 doi: 10.1101/2019.12.18.881599. Date: 2019.12.19 Taniguchi H , Fie M , Wu P , Kim S , Paik R , Sugino K , Kvitsani D , Fu Y , Lu J , Lin Y , (2011). A resource of ere driver lines for genetic targeting of GABAergic neurons in cerebral cortex. Neuron 71 , 995–1013 doi: 10.1016/j.neuron.2011.07.026.21943598 Tasic B , Yao Z , Graybuck LT , Smith KA , Nguyen TN , Bertagnolli D , Goldy J , Garren E , Economo MN , Viswanathan S , (2018). Shared and distinct transcriptomic cell types across neocortical areas. Nature 563 , 72–78 doi: 10.1038/s41586-018-0654-5.30382198 Tremblay R , Lee S , and Rudy B (2016). GABAergic interneurons in the neocortex: from cellular properties to circuits. Neuron 91 , 260–292 doi: 10.1016/j.neuron.2016.06.033.27477017 Vallentin D , Kosche G , Lipkind D , and Long MA (2016). Inhibition protects acquired song segments during vocal learning in zebra finches. Science 351 , 267–271 doi: 10.1126/science.aad3023.26816377 Wall NR , De La Parra M , Sorokin JM , Taniguchi H , Huang ZJ , and Callaway EM (2016). Brain-wide maps of synaptic input to cortical interneurons. J. Neurosci 36 , 4000–4009 doi:10.1523/JNEUROSCI.3967-15.2016.27053207 Williams LE , and Holtmaat A (2019). Higher-order thalamocortical inputs gate synaptic long-term potentiation via disinhibition. Neuron 101 , 91–102.e4 doi: 10.1016/j.neuron.2018.10.049.30472077 Woolf NJ (1991). Cholinergic systems in mammalian brain and spinal cord. Prog. Neurobiol 37 , 475–524 doi: 10.1016/0301-0082(91)90006-M.1763188 Xu T , Yu X , Perlik AJ , Tobin WF , Zweig JA , Tennant K , Jones T , and Zuo Y (2009). Rapid formation and selective stabilization of synapses for enduring motor memories. Nature 462 , 915–919 doi: 10.1038/nature08389.19946267 Xue M , Atallah BV , and Scanziani M (2014). Equalizing excitation–inhibition ratios across visual cortical neurons. Nature 511 , 596–600 doi: 10.1038/nature13321.25043046 Yang CF , Chiang MC , Gray DC , Prabhakaran M , Alvarado M , Juntti SA , Unger EK , Wells JA , and Shah NM (2013). Sexually dimorphic neurons in the ventromedial hypothalamus govern mating in both sexes and aggression in males. Cell 153 , 896–909 doi: 10.1016/j.cell.2013.04.017.23663785 Yao Z , Liu H , Xie F , Fischer S , Adkins RS , Aldridge AI , Ament SA , Bartlett A , Behrens MM , Van den Berge K , (2021). A transcriptomic and epigenomic cell atlas of the mouse primary motor cortex. Nature 598 , 103–110 doi: 10.1038/s41586-021-03500-8.34616066 Zeisel A , Hochgerner H , Lönnerberg P , Johnsson A , Memic F , van der Zwan J , Höring M , Braun E , Borm LE , La Manno G , (2018). Molecular architecture of the mouse nervous system. Cell 174 , 999–1014.e22 doi: 10.1016/j.cell.2018.06.021.30096314 Zhang S , Xu M , Kamigaki T , Do JPH , Chang W-C , Jenvay S , Miyamichi K , Luo L , and Dan Y (2014). Long-range and local circuits for top-down modulation of visual cortex processing. Science 345 , 660–665 doi: 10.1126/science.1254126.25104383

PMC009xxxxxx/PMC9308720.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101676030 44856 Cell Chem Biol Cell Chem Biol Cell chemical biology 2451-9456 2451-9448 35429459 9308720 10.1016/j.chembiol.2022.03.010 NIHMS1801179 Article Cellular signals converge at the NOX2-SHP-2 axis to induce reductive carboxylation in cancer cells Zhang Rukang 1348 Chen Dong 1378 Fan Hao 134 Wu Rong 134 Tu Jiayi 4 Zhang Freya Q. 4 Wang Mei 137 Zheng Hong 23 Qu Cheng-Kui 23 Elf Shannon E. 5 Faubert Brandon 4 He Yu-Ying 4 Bissonnette Marc B. 4 Gao Xue 1349 DeBerardinis Ralph J. 69 Chen Jing 134910 1 Department of Hematology and Medical Oncology, Emory University School of Medicine, Atlanta, GA 30322 2 Department of Pediatrics and Aflac Cancer and Blood Disorder Center, Emory University School of Medicine, Atlanta, GA 30322 3 Winship Cancer Institute, Emory University School of Medicine, Atlanta, GA 30322 4 Department of Medicine, The University of Chicago, Chicago, IL 60637 5 The Ben May Department for Cancer Research, The University of Chicago, Chicago, IL 60637 6 UT Southwestern Medical Center, Dallas, TX 75390, USA. 7 Current address: Soochow University, Suzhou, China, 215000 8 These authors contributed equally 9 Correspondence to: xgao30@uchicago.edu (X.G.), Ralph.Deberardinis@utsouthwestern.edu(R.J.D.) or jingchen@medicine.bsd.uchicago.edu(J.C.) 10 Lead Contact AUTHOR CONTRIBUTIONS Conception and design: R.J.D. and J.C. Development of methodology: R.Z., D.C., M.W., B.F., and X.G. Acquisition of data: R.Z., D.C., H.F., R.W., J.T., F.Q.Z., and M.W. Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.Z., D.C., H.F., R.W., J.T., F.Q.Z., and M.W. Writing, review, and/or revision of the manuscript: R.Z., X.G., R.J.D., and J.C. Administrative, technical, or material support: H.Z., C.-K.Q., Y-Y.H., M.B.B., and S.E.E. Study supervision: X.G., R.J.D., and J.C. 28 4 2022 21 7 2022 15 4 2022 21 7 2023 29 7 12001208.e6 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY Environmental stresses including hypoxia or detachment for anchorage independence, or attenuation of mitochondrial respiration through inhibition of electron transport chain induce reductive carboxylation in cells with an enhanced fraction of citrate arising through reductive metabolism of glutamine. This metabolic process contributes to redox homeostasis and sustains biosynthesis of lipids. Reductive carboxylation is often dependent on cytosolic isocitrate dehydrogenase 1 (IDH1). However, whether diverse cellular signals induce reductive carboxylation differentially, or through a common signaling converging node remains unclear. We found that induction of reductive carboxylation commonly requires enhanced tyrosine phosphorylation and activation of IDH1, which, surprisingly, is achieved by attenuation of a cytosolic protein-tyrosine phosphatase, Src homology region 2 domain-containing phosphatase-2 (SHP-2). Mechanistically, diverse signals induce reductive carboxylation by converging at upregulation of NADPH oxidase 2, leading to elevated cytosolic reactive oxygen species that consequently inhibit SHP-2. Together, our work elucidates the signaling basis underlying reductive carboxylation in cancer cells. Graphical Abstract eTOC Blurb The signaling basis underlying induction of reductive carboxylation remains unclear. Zhang et. al. report that diverse signals converge at the NOX2-SHP-2 axis, leading to enhanced tyrosine phosphorylation and activation of IDH1 that is commonly required for induction of efficient reductive carboxylation. pmcINTRODUCTION During the tricarboxylic acid (TCA) cycle, citrate can be generated from acetyl-CoA and oxaloacetate, which is at the center of cellular metabolism by breaking down glucose initiated in glycolysis and fueling ATP production (Williams and O'Neill, 2018). However, in cells where mitochondrial oxidation is suppressed due to defects in electron transport chain (ETC), or treatment with ETC inhibitors such as Complex I inhibitor rotenone, reductive carboxylation of glutamine-derived α-ketoglutarate (αKG)accounts for an increased fraction of cellular isocitrate and citrate (Mullen et al., 2012). Moreover, reductive carboxylation is also induced not only in cells experiencing hypoxia, where mitochondrial reactive oxygen species (ROS) accumulate due to inefficient electron transfer through the ETC (Metallo et al., 2012; Wise et al., 2011), but in cells during anchorage-independent growth, where detachment of cells from a monolayer also increases mitochondrial ROS, leading to aberrant oxidative stress that promotes reductive metabolism of glutamine (Jiang et al., 2016). Reductive carboxylation is highly dependent on cytosolic isocitrate dehydrogenase 1 (IDH1) to generate isocitrate from αKG via a non-canonical reverse reaction by IDH1. Subsequently, isocitrate is important for producing citrate and acetyl-CoA that are essential for lipid synthesis in cells experiencing hypoxia, as well as for reducing mitochondrial ROS to sustain redox homeostasis during hypoxia, or to support anchorage-independent growth and survival of cells during extracellular matrix detachment and metastatic spread (Jiang et al., 2016; Metallo et al., 2012; Wise et al., 2011). Thus, sustained enzyme activity of cytosolic IDH1 must play a crucial role during reductive carboxylation. We recently reported that tyrosine phosphorylation-enhanced IDH1 activation is important for glutamine-dependent reductive carboxylation and lipogenesis in cancer cells under hypoxia, which provides a metabolic advantage to cancer cell proliferation and tumor growth (Chen et al., 2019a). We found that Y42 and Y391 phosphorylation enhances IDH1 enzyme activity for both canonical (isocitrate→αKG) and non-canonical (αKG→isocitrate; reductive carboxylation) directions in cells. Diverse receptor tyrosine kinases including EGFR and FGFR1 activate Src to achieve Y42 and Y391 phosphorylation of IDH1, respectively, which contributes to reductive carboxylation and consequent cell proliferation and tumor growth. In addition, we demonstrated that IDH1 phosphorylation levels correlate with EGFR signal intensity and the contribution of reductive carboxylation to citrate metabolism in diverse lung cancer cell lines(Chen et al., 2019b). However, the signaling link between IDH1 activation and induction of reductive carboxylation remains unclear. RESULTS AND DISCUSSION Diverse signals commonly require enhanced tyrosine phosphorylation and activation of IDH1 to induce reductive carboxylation in cancer cells We found that phosphorylation levels of IDH1 at Y42 and Y391 were induced in diverse human cells treated with different reductive carboxylation-inducing stresses, including lung cancer A549 and H1299 cells experiencing hypoxia, anchorage-independent detachment or treatment with Complex I inhibitor rotenone (Figure 1A, left, middle and right, respectively), as well as leukemia K562 and MOLM-14 cells treated with rotenone (Figure S1A). Moreover, we examined “rescue” A549 cells with siRNA-mediated knockdown of endogenous IDH1, followed by rescue expression of siRNA-resistant FLAG-IDH1 WT or the catalytically less active, phosphor-deficient FLAG-IDH1 Y42F/Y391F double mutant (Figure 1B). We found that IDH1 knockdown abolished the enhanced levels of reductive carboxylation induced by hypoxia, detachment, or rotenone treatment (Figure 1B, leftfour bars in left, middle and right panels, respectively), which were completely rescued by expression of FLAG-IDH1 WT, but not by expression of the phosphor-deficient Y42F/Y391F (Figure 1B, right four bars in left, middle and right panels, respectively). These data together suggest that tyrosine phosphorylation and subsequent activation of IDH1 are commonly enhanced and required for effective reductive carboxylation in cancer cells induced by diverse signals. Induction of reductive carboxylation attenuates SHP-2 activity to enhance phosphorylation and activation of IDH1 We next sought to examine whether diverse reductive carboxylation-inducing stresses promote tyrosine phosphorylation of IDH1 by activating upstream tyrosine kinases. We previously demonstrated that the EGFR-Src cascade in A549 cells, the FGFR1-Src cascade in H1299 cells, and the BCR-ABL-Src cascade in K562 cells represent distinct upstream IDH1 tyrosine kinase cascades in different cells (Chen et al., 2019a). Surprisingly, we found that treatments with hypoxia, detachment or rotenone resulted in rather decreased EGFR activity assessed by reduced tyrosine phosphorylation while Src activity was unaltered in A549 cells (Figure S1B). In H1299 cells, hypoxia resulted in reduced tyrosine phosphorylation of both FGFR1 and Src, while Rotenone and detachment treatments reduced Src phosphorylation levels but did not alter phosphorylation levels of FGFR1 (Figure S1C). Moreover, rotenone treatment resulted in slightly reduced tyrosine phosphorylation levels of both BCR-ABL and Src. Note that the leukemogenic fusion tyrosine kinase BCR-ABL is constitutively active and cannot be further activated(Gilliland, 2001) (Figure S1D). These data together suggest that all three different stresses may commonly attenuate an upstream protein tyrosine phosphatase (PTP) of IDH1, leading to enhanced IDH1 phosphorylation and activation that are required for effective reductive carboxylation. We next examine three well known PTPs including Src homology region 2 domain-containing phosphatase-2 (SHP-2, a.k.a. tyrosine-protein phosphatase non-receptor type 11 (PTPN11)), PTP localized to the Mitochondrion 1 (PTPMT1), and PTP1B (a.k.a. tyrosine-protein phosphatase non-receptor type 1 (PTPN1)).We found that treatment with SHP099 (SHP-2 inhibitor), but not alexidine dihydrochloride (PTPMT1 inhibitor) or TCS401 (PTP1B inhibitor), resulted in elevated Y42 and Y391 phosphorylation levels of IDH1 and abolished rotenone-induced IDH1 phosphorylation in A549 cells (Figure 2A). Similar results were obtained in A549 cells treated with detachment or hypoxia (Figure S2A, upper and lower, respectively). In addition, treatment with SHP099, but not alexidine dihydrochloride or TCS401, resulted in elevated reductive carboxylation rate, which could not be further enhanced by hypoxia in A549 cells (Figure S2B). Consistent with these findings, knockdown of SHP-2, but not PTPMT1 or PTP1B, by specific siRNAs led to elevated Y42 and Y391 phosphorylation levels of IDH1 and abolished rotenone-induced IDH1 phosphorylation (Figure 2B). Moreover, rotenone treatment resulted in increased tyrosine phosphorylation of Y42 and Y391 of IDH1(Figure S2C) as well as increased reductive carboxylation in mouse embryonic fibroblasts (MEFs) assessed by elevated lipogenesis rate(Figure 2C), which was not observed in Shp-2 knockout MEFs (Shp-2Δ/Δ) generated from Ptpn11 conditional knockout embryos (Ptpn11 fl/fl/ER-Cre) (Yu et al., 2003), further supporting our overall hypothesis. We next confirmed SHP-2 as an IDH1 upstream phosphatase by performing an in vitro dephosphorylation assay using purified SHP-2 (rSHP-2) incubated with phosphorylated rIDH1as a substrate; rIDH1protein was pre-treated with rEGFR and/or rSrc to achieve phosphorylation at Y42 and/or Y391, respectively (Figure 2D). The results demonstrate that SHP-2 directly dephosphorylates IDH1.In addition, we found that SHP-2 phosphatase activity was commonly reduced in A549 (Figure 2E) and H1299 (Figure S2D) cells treated with rotenone, hypoxia, or detachment, which was assessed in an in vitro dephosphorylation assay using immunoprecipitated SHP-2 incubated with DiFMUP as a substrate. Finally, we found that treatment with SHP099resulted in elevated reductive carboxylation rate, which could not be further enhanced by rotenone in A549 cells (Figure 2F, left four bars), whileIDH1 knockdown abolished the enhanced reductive carboxylation induced by SHP99 and/or rotenone(Figure 2F, middle two bars), which was completely rescued by expression of FLAG-IDH1 WT, but not by expression of the phosphor-deficient Y42F/Y391F (Figure 2F, right four bars).Similar results were obtained using A549 cells treated with detachment (Figure S2E). These results together suggest that induction of reductive carboxylation is mediated through inhibition of SHP-2, leading to enhanced tyrosine phosphorylation and activation of IDH1. Reductive carboxylation elevates cytosolic ROS to inhibits SHP-2 It was reported that reductive carboxylation-inducing stresses including hypoxia, detachment and rotenone suppress mitochondrial oxidation and cause mitochondrial ROS accumulation (Jiang et al., 2016; Metallo et al., 2012; Mullen et al., 2012; Wise et al., 2011); and that PTPs including SHP-2 are important targets of ROS, which inactivates PTPs by oxidizing catalytic Cys residues to the sulfenic acid state (Tanner et al., 2011). Since SHP-2 primarily localizes in cytosol (Qu, 2000), we were curious about whether cytosolic ROS levels are also elevated during reductive carboxylation and how this might influence cytosolic SHP-2 phosphatase activity. We found that rotenone elevated both cytosolic ROS and mitochondrial ROS (Figure 3A, left and right, respectively). The cytosolic and mitochondrial ROS levels were assessed by confocal microscopic imaging to detect cytosolic Hyper-cyto or mitochondrial Hyper-mito, respectively, which are red fluorescent genetically encoded indicators (Belousov et al., 2006; Ermakova et al., 2014) being expressed in A549 cells with specific subcellular localizations (Figure S3A-S3B). Moreover, we found that treatment with antioxidant agent N-acetyl-L-cysteine (NAC; 1mM) effectively reversed elevated cytosolic ROS but not mitochondrial ROS in A549 cells treated with rotenone, whereas treatment with mitochondria-specific antioxidant agent mitoTEMPO (10μM) effectively reversed rotenone-elevated mitochondrial ROS but not cytosolic ROS in A549 cells (Figures 3A and S3A-S3B). We found that NAC but not MitoTEMPO treatment effectively reversed the decreased SHP-2 phosphatase activity by rotenone in A549 cells (Figure 3B), suggesting that cytosolic rather than mitochondrial ROS inhibit SHP-2 during the induction of reductive carboxylation. Consistent with this finding, NAC but not MitoTEMPO treatment effectively reversed the increased tyrosine phosphorylation of Y42 and Y391 of IDH1 (Figure 3C) as well as the increased reductive carboxylation assessed by increased lipogenesis rate in A549 cells treated with rotenone (Figure 3D).Similar results were obtained using H1299 cells treated with rotenone(Figures S3C) andA549 cells experiencing detachment (Figures S3D and 3E), or hypoxia (Figures S3E and 3F). Furthermore, treatment with another antioxidant agent, Ebselen, shows similar effects as NAC treatment in A549 cells treated with rotenone(Figures S3F).We also found that treatment with NAC abolished elevated reductive carboxylation rate by rotenone treatment (Figure 3G, first three bars on left), whileIDH1 knockdown abolished the effects on reductive carboxylation by rotenone or rotenone combined with NAC (Figure 3G, second three bars on left), which was completely rescued by expression of FLAG-IDH1 WT, but not by expression of the phosphor-deficient Y42F/Y391F (Figure 3G, right six bars).Moreover, we added indicated concentrations of hydrogen peroxide in an in vitro dephosphorylation assay using purified SHP-2 (rSHP-2) incubated with phosphorylated rIDH1 as a substrate and found that in vitro dephosphorylation of IDH1 by SHP-2 can be inhibited by hydrogen peroxide(Figure S3G).These data together suggest that different reductive carboxylation-inducing stresses commonly elevate cytosolic ROS for SHP-2 inhibition, leading to elevated tyrosine phosphorylation of IDH1in cancer cells. Induction of reductive carboxylation commonly upregulates NOX2 to achieve SHP-2 inhibition SHP-2 oxidation was suggested to require NADPH oxidases (NOXs) (Tsutsumi et al., 2017). There are five NOX family members (NOX1-5)(Bedard and Krause, 2007). We found that detachment (Figure S4A), hypoxia (Figure S4B), and rotenone treatment (Figure S4C) resulted in upregulated mRNA levels of multiple NOX family members, with common upregulation of NOX2 and NOX3 under all three conditions. In order to identify which NOX member is responsible for SHP-2 inhibition during reductive carboxylation, we tested a group of NOX inhibitors. We found that treatment with a pan-NOX inhibitor diphenyleneiodonium (DPI) reversed the elevated reductive carboxylation assessed by increased lipogenesis rate in A549 cells experiencing hypoxia in a dose-dependent manner; NAC treatment was included as a positive control (Figure 4A). In contrast, among several selective NOX inhibitors including GSK2795039 (NOX2), GKT137831 (NOX1/4), GLX351322 (NOX4) and ML090 (NOX5), only the NOX2 inhibitor GSK2795039 (GSK) effectively reversed elevated reductive carboxylation in A549 cells treated with rotenone (Figure 4B). Moreover, GSK treatment similarly reversed increased reductive carboxylation in a dose-dependent manner in A549 cells induced by hypoxia or detachment (Figure S4D, left and right, respectively). These results are consistent with findings that detachment, hypoxia or rotenone treatment resulted in increased protein expression levels of NOX2 in A549 cells (Figure 4C).Note that NOX2 requires the membrane subunit (p22phox), cytosolic subunits (p67phox and p47phox), and the Rac GTPase to form a complex for activation (Altenhofer et al., 2012; Brandes et al., 2014; Panday et al., 2015).Interestingly, we found that mRNA (Figure 4D) and protein (Figure 4E) levels of p67phox, p47phox, but not p22phox were elevated by rotenone treatment. Since p67phox and p47phox are NOX2-specific cytosolic subunits, while p22phox is a common binding partner for NOX1-4 in the membrane, these data not only further support our findings that reductive carboxylation activates NOX2 but also provide mechanistic insights into the selective activation of NOX2 that is ensured by upregulation of p67phox and p47phox induced by rotenone. We next checked NOX2 and its partners subcellular localization by immunocytochemistry (ICC) staining using confocal microscopy(Figure 4F). We observed co-localization of NOX2 and p67phox in both cell membrane and cytosol, which was enriched to cell membrane with increased protein intensity upon rotenone treatment in A549 cells(Figure 4F).Furthermore, knockdown of NOX2 by siRNA in A549 cells (Figure 4G,left), but not siRNAs targeting NOX1 or NOX3 (Figure S4E), reversed the rotenone-induced reductive carboxylation and elevated cytosolic ROS levels (Figure 4G, left and right, respectively), as well as increased tyrosine phosphorylation levels of IDH1 and reduced SHP-2 activity (Figure 4H, left and right, respectively). Similar results were obtained in K562 and H1299 cells treated with rotenone(Figure S4F-S4I). NOX2 knockdown in A549 cells treated with detachment or hypoxia also led to induced reductive carboxylation rates (Figure 4I), elevated cytosolic ROS levels (Figure S4J), increased tyrosine phosphorylation levels of IDH1 (Figure S4K) and reduced SHP-2 activity (Figure S4L). Lastly, we found that treatment with GSK abolished elevated reductive carboxylation rate by rotenone treatment (Figure 4J, first three bars on left), whileIDH1 knockdown abolished the effects on reductive carboxylation by rotenone or rotenone combined with GSK(Figure 4J, second three bars on left), which was completely rescued by expression of FLAG-IDH1 WT, but not by expression of the phosphor-deficient Y42F/Y391F (Figure 4J, right six bars).Similar results were obtained using A549 cells treated with detachment (Figure S4M). Thus, these data together strongly support our hypothesis thatNOX2 is commonly required by reductive carboxylation induced by different stresses in cancer cells. DISCUSSION Our studies elucidate the underlying signaling mechanism by which diverse cellular responses to hypoxia, detachment and attenuation of ETC commonly converge at upregulation of NOX2, leading to consequent inhibition of cytosolic protein tyrosine phosphatase SHP-2, which is upstream of IDH1 and plays a crucial role in inducing effective reductive carboxylation in cancer cells through enhanced tyrosine phosphorylation and activation of cytosolic IDH1 (Figure 4K). In contrast, reductive carboxylation enhances tyrosine phosphorylation of IDH1 independent of its upstream Group I and II tyrosine kinases (Chen et al., 2019a) (Figure 4K).Our results revealed that diverse reductive carboxylation-inducing stresses simultaneously promote productions of both mitochondrial and cytosolic ROS, while although mitochondrial ROS accumulation suppresses mitochondrial oxidation, it is the cytosolic ROS that are responsible for the inhibition of SHP-2 and consequent phosphorylation and activation of IDH1, which is crucial for effective reductive carboxylation in cancer cells (Figure 4K). Limitations of the study: It was reported that reductive carboxylation is enhanced by hypoxia in a HIF-1α-dependent manner, whereas detachment associated reductive metabolism is independent of hypoxia or HIF-1α (Jiang et al., 2016). This may suggest that hypoxia signal through HIF-1α to upregulate NOX2, whereas detachment might activate NOX2 in a different way. Rotenonedependent activation of NOX2 in human lung cancer cells were reported but the underlying mechanism remains unknown (Hu et al., 2016). It remains also unknown about the connection between detachment and NOX2. A previous report suggested that inhibition of NOX1/4 with GKT137831 attenuates retinal detachment-induced photoreceptor apoptosis (Deliyanti and Wilkinson-Berka, 2015), whereas GKT137831 did not affect reductive carboxylation induced by detachment in A549 cells (Figure 4B), suggesting that NOX1/4 might not be involved. Future studies are warranted to decipher the underlying mechanisms by which NOX2 is upregulated by different signals. STAR★Methods RESOURCE AVAILABILITY Lead contact Further information requests should be directed to the lead contact, Jing Chen (jingchen@medicine.bsd.uchicago.edu) Materials availability Materials will be shared by the lead contact upon request. Data and code availability Data will be shared by the lead contact upon request. This paper does not report original code. Additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. EXPERIMENTAL MODEL AND SUBJECT DETAILS Primary Cell culture MEF (from CK Qu’s lab; authentication and Mycoplasma not test) were cultured in DMEM medium with 10% FBS and 2 ng/ml recombinant human GM-CSF (R&D Systems). Cell lines HEK293T cells were cultured in Dulbecco Modified Eagle Medium (DMEM) with10% fetal bovine serum (FBS) (Sigma, F2442) and penicillin/streptomycin. NCI-H1299 (#CRL-5803, ATCC; purchased 2015; authentication and Mycoplasma tested 2018), K562, MOLM-14, and A549 cells were cultured in RPMI 1640 medium with 10% FBS. Please also refer to KEY RESOURCES TABLE for detailed information of each cell line. Cell lines experiments were conducted and designed according to protocols approved by Institutional Biosafety Committee (IBC) of the University of Chicago. METHOD DETAILS Antibodies Antibodies against DYKDDDDK (FLAG) tag, p-Tyrosine (p-Tyr-100) were from Cell Signaling Technology (CST). Antibodies against β-actin and were from Sigma-Aldrich. Antibody against IDH1 was from R&D SYSTEMS. Antibodies against SHP2, PTPMT1 and PTP1B was from PROTEINTECH. Goat anti-Mouse IgG (H+L) secondary antibody and goat anti-rabbit IgG (H+L) secondary antibody were from Thermo Fisher Scientific. Antibodies against p-IDH1 Y42 and p-IDH1 Y391 were custom-made by SHANGHAI GENOMICS, INC. Anti-NOX2 antibody was purchased from Abcam and Bioss. p67phox was purchased from Santa Cruz. Anti-rabbit IgG (H+L) F(ab')2 Fragment (Alexa Fluor 488 conjugate) and Anti-mouse IgG (H+L) F(ab')2 Fragment (Alexa Fluor 555 conjugate) was purchase from Cell Signaling Technology. Reagents Rotenone, NAC, GSK2795039,Ebselen, DPI, GSK2795039 and alexidine dihydrochloride were purchased from Sigma-Aldrich. Glutamine L-[5-14C] was from ARC. SHP099 and GKT137831 was purchased from Selleckchem. TCS401, ML090 was purchased from CAYMAN.GLX351322 was purchased from MCE. Mito-TEMPO, DiFMUP and hydrogen peroxide were purchase from Fisher Scientific. Si-IDH1, Si-NOX1, si-NOX3, si-SHP-2 were purchase from Sigma. si-PTP1B and si-PTPMT1 were purchased from Qiagen. si-NOX2 was synthesized by IDT. Primers was synthesized by IDT (The Sequence of siRNA are listed in Key Resources Table). Retrovirus production and stable cell lines construction Stable overexpression of IDH1 WT and mutants in A549 was conducted using retroviral vectors harboring RNAi-resistant FLAG-tagged IDH1 WT, and RNAi-resistant FLAG-tagged IDH1 Y42F/Y391F mutant. Briefly, to produce retrovirus, each construct was co-transfected with 0.1 μg VSVG, 0.9 μg EcoPak packaging plasmid, and 1 μg envelope plasmid (Addgene) into HEK293T cells seeded in 6 well plate using TransIT®-LT1 Transfection Reagent (Mirus) according to the manufacturer’s instructions. Retrovirus-containing supernatant medium was collected 48 hours after transfection and filtered by 0.45 μm filter before addition to the indicated host cell lines with 3 μL 10 mg/ml polybrene(1 mol/L HEPES was used to adjust the pH to 7.4 in culture medium). Twenty-four hours after infection, target cells were subjected to hygromycin selection (Invitrogen). The overexpression of proteins was confirmed by Western blotting using antibodies against IDH1. Purification of Prokaryotic Recombinant IDH1 and SHP-2 Proteins 6 × His-FLAG-IDH1were purified by sonication of high expression BL21(DE3) cells obtained from a 250 mL culture subjected to IPTG induction for 16 hours at 30°C. Bacteria cell lysates were obtained by centrifugations and loaded onto a Ni-NTA column within 20 mmol/L imidazole. The bound proteins were eluted with 250 mmol/L imidazole, followed by desalting using a PD-10 column. SHP-2 expression plasmid was purchased from Addgene (Addgene8322). GST-SHP-2 protein was purified by sonication of high expression BL21(DE3) cells obtained from a 250 mL culture subjected to IPTG induction for 16 hours at 16 °C. Bacteria cell lysates were obtained by centrifugations and loaded onto a GSTrap™ High Performance (Cytiva 17-5281-01). The bound proteins were eluted with25 mM HEPES, pH 7.2, 50 mM NaCl, 5mM GSH, followed by desalting using a PD-10 column. The recombinant IDH1 and SHP-2 protein were stored at −80 °C fridge with adding 5 % glycerol (v/v). In Vitro SHP-2 dephosphorylation assay First, Flag-IDH1 protein was treated with EGFR or Src kinase as previous described (Chen et al., 2019a). 1 μg recombinant Flag-IDH1 protein was incubated with diverse recombinant active form of 800 ng EGFR (Thermo Fisher) or 100 ng SRC (Thermo Fisher) in the thermomixer in the presence of 800 μM ATP (Sigma) at 30°C, 300 rpm for 90 minutes in the following assay buffers, respectively. SRC buffer, 50 mM HEPES (pH 7.5), 10 mM MgCl2, 10% glycerol, 2.5 mM DTT, and 0.01% Triton X-100 were used; for EGFR buffer, 20 mM Tris (pH 7.5), 10 mM MgCl2, 1 mM EGTA, 1 mM Na3VO4, 5 mM β-glycerophosphate, 2 mM DTT, and 0.02% Triton X-100 were used. After kinase reaction, the mixture was incubated with 30 μL of ANTI-FLAG M2 Affinity Gel (Sigma-Aldrich) for 4 hours at 4°C, followed by washing with phosphate-buffered saline (PBS) 3 times to remove unbound materials. Centrifuge the mixture at 3000 rpm for 3 min and discard the supernatants. Then the beads were resuspended using SHP-2 reaction buffer (25 mM HEPES, pH 7.2,50 mM NaCl, 2.5 mM EDTA, 5 mM DTT) including 10 μM SHP-2with or without hydrogen peroxide. The dephosphorylation assay was performed at room temperature for 120 min. Then the IDH1 phosphorylation was examined by Western Blotting. Small interfering RNA-mediated knockdown The transfection of small interfering RNA (siRNA) into A549 cells was carried out using Lipofectamine RNAiMAX Transfection Reagent (Thermo fisher), according to the manufacturer’s instructions. Briefly, 5 μL siRNA (10 μM) for each well and RNAiMAX reagent were mixed in Opti-MEM medium (Thermo fisher) and incubated for 5 min at room temperature to allow the complex formation. Then the cells seeded in 6 well plates were washed with Opti-MEM medium (Thermo fisher), and the mixtures were added. Twelve hours after transfection, the culture medium was replaced by fresh complete medium. The cells were harvested 72 hours after transfection, followed by further analysis (Chen et al., 2019a). The following siRNA sequences were used for knockdown: negative control siRNA (non-silencing; QIAGEN; SI03650325); PTP1B or PTPMT1 siRNA was purchased from QIAGEN; SHP2, IDH1 siRNA was synthesized from Sigma. NOX1, NOX2, NOX3 siRNA were synthesized from IDT. Cell culture treatment Treatment with Rotenone was performed by incubating 1 × 107 cells with 10 μM Rotenone for 16 hours (Li et al., 2003). Treatment with detachment was performed by placing 1 × 107 cells in Corning® 100 mm Ultra-Low Attachment Culture Dish for 24 hour (Lam et al., 2007). Treatment with hypoxia was performed by incubating 3 × 106 cells under hypoxia (5% CO2, 1% O2 and 94% N2) for 48 hours (Chen et al., 2019a). For PTP inhibitors treatment, SHP099 (10 μM), TCS401 (1 μM) and Alexidine dihydrochloride (1 μM) were added to medium synchronized with Rotenone, detachment or hypoxia (Du et al., 2015; Kenny et al., 2015; Niogret et al., 2019). For NAC, MitoTEMPO or Ebselen treatment, NAC (1 mM), MitoTEMPO (10 μM) or Ebselen (5 μM or 10 μM) was added to medium in synchronized with Rotenone, detachment or hypoxia (Sourbier et al., 2019; Yang et al., 2018). For MEF cells, 3 × 106 MEF cells or MEF SHP-2-ff/ER-cre cells were seeded in 6 well plate, after 16 hours cells treated with 4-OHT (4-hydroxytamoxifen) (0.5 μM) to induce Cre expression and excision of floxed Ptpn11 gene fragment (Wu et al., 2019). For NOXs inhibitors treatment, GSK2795039 (20 μM), GKT137831 (100 nM), GLX351322 (5 μM), ML090 (10 nM) was added to medium synchronized with Rotenone, detachment or hypoxia treatment (Anvari et al., 2015; Casas et al., 2019; Hirano et al., 2015; Moon et al., 2016). Lipid biosynthesis assay Lipid biosynthesis assay was performed following the protocol as previously described (Chen et al., 2019a). For 14C-Lipid biosynthesis assay, cells were incubated with 4 μM Glutamine L-[5-14C] (ARC) for 48 hours under hypoxia (5% CO2, 1% O2 and 94% N2). For rotenone treatment lipid biosynthesis measurement, cells were incubated with 4 μM Glutamine L-[5-14C] and rotenone. For detachment treatment lipid biosynthesis measurement, cells were incubated with 4 μM Glutamine L-[5-14C] in low attachment surface dish (Corning). Lipids were then extracted by the addition of 500 μL of hexane: isopropanol (3:2 v/v), air dried, resuspended in 50 μL of chloroform and transferred to 7 ml glass tubes followed by adding 5ml liquid scintillation cocktail to each tube. Then the bottles were gently inverted 5 times, and subjected to scintillation counting. SHP2 activity assay 1 × 107 cells that were treated by rotenone, detachment or hypoxia were collected and lysed by 800 μL NP-40 lysis buffer with protease inhibitor cocktail for 20 min under 4 °C. Then the mixture was centrifuged at 13000 rpm for 15 min at 4 °C and the supernatants were collected to 1.5 mL tubes. Then each tube of the cell lysates was incubated with 5 μL anti-SHP2 antibody (PROTEINTECH) at 4°C overnight, then wash with TBS for 3 times. Then 30 μL protein G Sepharose 4 Fast Flow beads were added to each tube following 3-hour incubation at 4 °C. The beads were centrifuged at 3000 rpm for 3min and discard the supernatants and washed by PBS for 3 times. Then the beads were resuspended by 150 μL SHP-2 reaction buffer (25 mM HEPES, pH 7.2,50 mM NaCl, 2.5 mM EDTA,5 mM DTT). 100 μM substrate DiFMUP (Thermo Fisher) were added to mixture to begin the reaction. Then the mixture was transferred to black 96-well plate (Corning). Gently shake the plate for 30 second and measure fluorescence at 455 nm every 1 min for 10 min under 358 nm excitation light using SpectralMax Plus spectrophotometer (Molecular Devices). The remaining beads were used for Western Blot. Subcellular ROS detection Cytosolic and mitochondrial ROS levels were measured with the organelle-specific HyPer system (Jiang et al., 2016). For rotenone or hypoxia treatment, 1 × 105A549 cells were planted in 35 mm dishes (Ibidi 81158) and transfected with 1 μg Hyper-cyto (Addgene 136467) or 1 μg Hyper-mito vectors (Addgene 136470). After 48 hours, transfected cells were treated with treated with 10 μM rotenone for or 16 hours or treated with hypoxia for 48 hours. For detachment treatment, 1 × 105cells were seeded in 6-well plate for vectors transfection. After 48 hours, transfected cells were trypsinized and transferred to low attachment 6-well plate (Corning 3471). After culturing 24 hours, cells were transferred to 35 mm dishes for imaging. Images were acquired using LeicaSP5 2photon microscope. ROS levels were calculated as the fluorescence intensity at 488 nm using Fiji software. Quantitative RT-PCR RNA was extracted by using TRIzol Reagent (Invitrogen). Quantitative RT-PCR was performed with PrimeScript 1st strand cDNA Synthesis Kit (Takara) and iTaq Universal SYBR Green Supermix (Bio-Rad). Real-time primers targeting NOX1, NOX2, NOX3, NOX4, NOX5, DUOX1, DUOX2, p22phox, p47phox, p67phox and actin were synthesized from Integrated DNA Technologies (IDT). The Sequence of primers are listed in Key Resources Table. Immunocytochemistry(ICC)staining For ICC staining, cells were seeded in 35 mm dishes for 24 hours and then treated with rotenone for 16 hours. Then the cells were washed by PBS for 3 times and fixed by 4% paraformaldehyde in PBS for 15 min at room temperature. Then discard the paraformaldehyde and washed the cells with PBS for 3 times and incubated the cells with 0.25% TritonX-100 in PBS for 10 min. After that, wash the cells with PBS 3 times for 5 min. Cells were blocked by 1% BSA in PBST for 30 min at room temperature. Then the NOX2 antibody and p67phox antibody were added to dish and incubated overnight at 4 °C. Discard the mixture solution and wash the cells three times in PBS, 5 mins each wash. Incubate cells with the mixture of anti-rabbit and anti-mouse secondary antibodies in 1% BSA for 1 hour at room temperature in dark. Discard the mixture of the secondary antibody solution and wash three times with PBS for 5 min each in dark. Incubate cells on 1 μg/ml DAPI for 10 min and then wash the cells three times in PBS. Cells were visualized and pictured using the LeicaSP5 2photon microscope. Quantification and Statistical Analysis Student’s t-test was used in studies in which statistical analyses were performed to generate p values. p values less than or equal to 0.05 were considered significant. Data with error bars represent mean ± s.d. There is no estimate of variation in each group of data, and the variance is similar between the groups. No statistical method was used to predetermine sample size. The investigators were not blinded to allocation during experiments and outcome assessment. All data are expected to have normal distribution. Statistical analysis and graphical presentation were performed using Prism 5.0 (Graph- Pad) and Microsoft Office Excel 2013. DATA AND SOFTWARE AVAILABILITY All software used in this study is listed in the Key Resource Table. Original imaging data have been deposited to Mendeley Data. Supplementary Material 1 ACKNOWLEDGEMENTS This work was supported in part by NIH grants including CA140515, CA174786 (J.C.), CA22044901 (R.J.D.) and CPRIT grant RP180778 (R.J.D.). J.C. is the Janet Davison Rowley Distinguished Service Professor in Cancer Research. Figure 1. Diverse signals commonly require enhanced tyrosine phosphorylation and activation of IDH1 to induce reductive carboxylation (A) Effect of hypoxia treatment for 48 hours (left), detachment treatment for 24 hours (middle), or 10 μM rotenone treatment for 16 hours (Right)on tyrosine phosphorylation of IDH1 assessed by Western blot. (B) Effect of “rescue” expression of siRNA-resistant FLAG-IDH1 WT or the catalytically less active FLAG-IDH1 Y42F/Y391F on reductive carboxylation indicated by 14C-lipid biosynthesis rate inIDH1 transient knockdown A549 cell with or without hypoxia (left), detachment (middle), or rotenone treatment (Right). The fold changes of the intensity ratios between phosphor-Y42 and total IDH1 or phosphor-Y391 and total IDH1 were indicated. The results were presented as mean ± s.d. of triplicate experiments. p values were obtained by a two-tailed Student’s t-test (*0.01<p<0.05; **0.001<p<0.01;***p<0.001; ns, not significant). Figure 2. Induction of reductive carboxylation attenuates SHP-2 activity to enhance phosphorylation and activation of IDH1 (A) Effect of treatment with SHP099 (left), Alexidine dihydrochloride (middle), or TCS401 (Right) on tyrosine phosphorylation of IDH1 in A549 cells treated with or without rotenone assessed by Western blot. (B) Effect of transient knockdown of endogenous SHP-2 (left), PTPMT1 (middle), or PTP1B (Right) on tyrosine phosphorylation of IDH1 in A549 cells treated with or without rotenone assessed by Western blot. (C) Effect of rotenone treatment on reductive carboxylation in Shp-2 knockout MEFs (Shp-2−/−). (D) Tyrosine kinase rEGFR and/or rSrc pretreatedrIDH1 was incubated with purified SHP-2 (rSHP-2) in an in vitro dephosphorylation assay. Tyrosine phosphorylation of IDH1 were determined by Western blot. (E) Effect of hypoxia, detachment, or rotenone treatment onSHP-2 phosphatase activity in A549 cells. (F) Effect ofSHP099 and/or rotenone treatment on reductive carboxylation in IDH1 transientknockdown A549 cells with or without “rescue” expression of FLAG-IDH1 WT orFLAG-IDH1 Y42F/Y391F. The results were presented as mean ± s.d. of triplicate experiments. p values were obtained by a two-tailed Student’s t-test (***p<0.001; ns, not significant). See also Figure S1 and S2. Figure 3. Reductive carboxylation elevates cytosolic ROS to inhibits SHP-2 (A) Effect of NAC or MitoTEMPO treatment on cytosolic ROS(left) and mitochondrial ROS (right) in A549 cells with rotenone treatment. (B-D) Effect of NAC or MitoTEMPO treatment on SHP-2 phosphatase activity (B), tyrosine phosphorylation of IDH1 (C), and reductive carboxylation (D) in A549 cells with rotenone treatment. (E-F) Effect of NAC or MitoTEMPO treatment on SHP-2 phosphatase activity (left), tyrosine phosphorylation of Y42 and Y391 of IDH1 (middle), and reductive carboxylation (right) in A549 cells with detachment (E)or hypoxia (F) treatment. (G) Effect of NAC treatment on reductive carboxylation in IDH1transient knockdown A549 cells with or without “rescue” expression of FLAG-IDH1 WT orFLAG-IDH1 Y42F/Y391Fwith rotenone treatment. The results were presented as mean ± s.d. of triplicate experiments. p values were obtained by a two-tailed Student’s t-test (*0.01<p<0.05; **0.001<p<0.01;***p<0.001; ns, not significant).See also Figure S3. Figure 4. Induction of reductive carboxylation commonly upregulates NOX2 to achieve SHP-2 inhibition (A) Effect of treatment with diphenyleneiodonium (DPI) with indicated concentrations on reductive carboxylation in A549 cells under hypoxia. (B) Effects of GSK2795039, GKT137831,GLX351322, or ML090treatments on reductive carboxylation in A549 cells with rotenone treatment. (C) Effect of detachment (left), hypoxia (middle), or rotenone (right) treatment on protein level of NOX2 in A549 cells assessed by Western blot. (D-E) Effect of rotenone treatment on NOX2 complex partner mRNA levels assessed by RT-qPCR (D), and protein level assessed by Western blot (E) in A549 cells. (F) Representative confocal microscopy images of NOX2 or p67phoxImmunocytochemistry(ICC)staining in A549 cells with or without rotenone treatment. (G-H) Effect of transient knockdown ofNOX2 on reductive carboxylation (G; left) and cytosolic ROS (G; right), tyrosine phosphorylation of IDH1 (H; left) and SHP-2 phosphatase activity (H; right)in A549 cells with or without rotenone treatment. (I) Effect of transient knockdown of NOX2 on reductive carboxylation in A549 cells with detachment treatment (left) or under hypoxia (right). (J) Effect of GSK2795039 treatment on reductive carboxylation in transient IDH1-knockdown A549 cells with or without “rescue” expression of FLAG-IDH1 WT or FLAG-IDH1 Y42F/Y391F with rotenone treatment. (K) Working model. The results were presented as mean ± s.d. of triplicate experiments. p values were obtained by a two-tailed Student’s t-test (*0.01<p<0.05; **0.001<p<0.01;***p<0.001; ns, not significant). See also Figure S4. KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Mouse monoclonal Anti-β-actin antibody Sigma-Aldrich Cat# A1978; Clone# AC-15; RRID: AB_476692 Mouse monoclonal ANTI-FLAG M2 antibody Sigma-Aldrich Cat# F3165; Clone# M2; RRID: AB_259529 NOX2 antibody Abcam Cat# Ab129068; Clone# N/A; RRID:AB_11144496 SHP-2 antibody Proteintech Cat# 20145-1-AP; Clone# N/A; RRID: AB_10699877 p-IDH1 Y42 SHANGHAI GENOMICS, INC.(Chen et al., 2019a) Cat# N/A; Clone# N/A; RRID: N/A p-IDH1 Y391 SHANGHAI GENOMICS, INC.(Chen et al., 2019a) Cat# N/A; Clone# N/A; RRID: N/A Human Isocitrate Dehydrogenase1/IDH1 Antibody R&D SYSTEMS Cat# MAB7049; Clone# 843219; RRID: AB_2811299 Goat anti-Mouse IgG (H+L) Secondary Antibody, HRP Thermo Fisher Scientific Cat# 31430; Clone# N/A; RRID: AB_228307 Goat anti-Rabbit IgG (H+L) Secondary Antibody, HRP Thermo Fisher Scientific Cat# 31460; Clone# N/A; RRID: AB_ 228341 DYKDDDDK Tag Antibody Cell Signaling Technology Cat#2368S; Clone# N/A; RRID: AB_2217020 Anti-rabbit IgG (H+L), F(ab')2 Fragment (Alexa Fluor® 488 Conjugate) Cell Signaling Technology Cat#4412 Clone# N/A; RRID:AB_1904025 Anti-mouse IgG (H+L), F(ab')2 Fragment (Alexa Fluor® 555 Conjugate) Cell Signaling Technology Cat#4409 Clone# N/A; RRID:AB_1904022 NOX2/gp91phox Polyclonal Antibody Bioss Cat#BS-3889R Clone# N/A; RRID:AB_10855911 p67phox antibody Santa Cruz Cat#sc-374510 Clone# N/A; RRID:AB_10988074 Bacterial and Virus Strains BL21(DE3) Chemically Competent Cells Sigma-Aldrich Cat# CMC0014 Chemicals, Peptides, and Recombinant Proteins Rotenone Sigma-Aldrich Cat# R8875 CAS: 83-79-4 NAC Sigma-Aldrich Cat# A7250 CAS: 616-91-1 alexidine dihydrochloride Sigma-Aldrich Cat# A8986 CAS: 1715-30-6 SHP099 Selleckchem Cat# S6388 CAS: 1801747-42-1 TCS401 CAYMAN Cat# 20393 CAS: 243966-09-8 EGFR Recombinant Human Protein Thermo Fisher Cat# PR7295B Src Recombinant Human Protein Thermo Fisher Cat# P3044 HEPES Sigma-Aldrich Cat# H4034 CAS: 7365-45-9 NaCl Sigma-Aldrich Cat# S9888 CAS: 7647-14-5 GSH Sigma-Aldrich Cat# G4251 CAS: 70-18-8 Glutamine L-[5-14C] ARC Cat# ARC 3562-50 uCi MitoTEMPO Thermo Fisher Cat# NC1037796 DPI Selleckchem Cat# S8639 CAS: 4673-26-1 GSK2795039 Sigma-Aldrich Cat# SML2770-5MG CAS: 1415925-18-6 GKT137831 Selleckchem Cat# S7171 CAS: 1218942-37-0 GLX351322 MCE Cat# HY-100111 CAS: 835598-94-2 ML090 CAYMAN Cat# 15172-10 CAS: 531-46-4 DiFMUP Thermo Fisher Cat# D6567 si-NOX1 Sigma-Aldrich Cat#EHU076051 si-NOX3 Sigma-Aldrich Cat#EHU051651 si-SHP-2 Sigma-Aldrich Cat#VPDSIRNA2D si-PTP1B Qiagen Cat#SI00043827 Si-PTPMT1 Qiagen Cat# SI4379921 iScript cDNA synthesize kit BIO-RAD Cat# 1708890 iTaq™ Universal SYBR® Green Supermix BIO-RAD Cat# 1725122 ANTI-FLAG® M2 Affinity Gel Sigma-Aldrich Cat# A2220 Protein G Sepharose® 4 Fast Flow Sigma-Aldrich Cat# GE17-0618-01 TransIT®-LT1 Transfection Reagent Mirus Cat# MIR 2305 TRIzol reagent Thermo Fisher Cat# 15596026 RNAi MAX Transfection Reagent Thermo Fisher Cat# 13778500 Ebselen Millipore Sigma Cat# E3520-25MG Hydrogen peroxide Thermo Fisher Cat# BP2633-500 Triton X-100 Millipore Sigma Cat# 64-846-410ML 4% Paraformaldehyde solution Thermo Scientific Cat# AAJ61899AK Experimental Models: Cell Lines Human: HEK293T cells ATCC Cat# CRL-3216; RRID: CVCL_0063 Human: A549 ATCC Cat# CCL-185; RRID: CVCL_0023 Human: NCI-H1299 ATCC Cat# CRL-5803; RRID: CVCL_0060 Human: MOLM-14 (Fan et al., 2016) Cat# N/A; RRID: N/A Human: K562 (Chen et al., 2019a) Cat# N/A; RRID: N/A Oligonucleotides si-IDH1 sequence: Sense: 5’-CGAAUCAUUUGGGAAUUGAUU -3’ Antisense: 5’-UCAAUUCCCAAAUGAUUCGUU-3’ Sigma-Aldrich N/A si-NOX2 sequence: Sense: 5’- CCCUGAGUAAACAAAGCA-3’ Antisense: 5’- UUGGAGAUGCUUUGUUUA-3’ IDT N/A Primers: β-actin Forward: 5’- CACCATTGGCAATGAGCGGTTC-3’ Reverse: 5’- AGGTCTTTGCGGATGTCCACGT-3’ IDT N/A Primers: NOX1 Forward: 5’- TTGTTTGGTTAGGGCTGAATGT-3’ Reverse: 5’- GCCAATGTTGACCCAAGGATTTT-3’ IDT N/A Primers: NOX2 Forward: 5’- CCCTTTGGTACAGCCAGTGAAGAT-3’ Reverse: 5’- CAATCCCGGCTCCCACTAACATCA-3’ IDT N/A Primers: NOX3 Forward: 5’- ACCGTGGAGGAGGCAATTAGA-3’ Reverse: 5’- TGGTTGCATTAACAGCTATCCC-3’ IDT N/A Primers: NOX4 Forward: 5’- CAGATGTTGGGGCTAGGATTG-3’ Reverse: 5’- GAGTGTTCGGCACATGGGTA-3’ IDT N/A Primers: NOX5 Forward: 5’-CCACCATTGCTCGCTATGAGTG -3’ Reverse: 5’- GCCTTGAAGGACTCATACAGCC-3’ IDT N/A Primers: DUOX1 Forward: 5’- TCTCTGGCTGACAAGGATGGCA-3’ Reverse: 5’- AGGCGAGACTTTTCCTCAGGAG-3’ IDT N/A Primers: DUOX2 Forward: 5’- CAATGGCTACCTGTCCTTCCGA-3’ Reverse: 5’- GTCCTTGGAGAGGAAGCCATTC-3’ IDT N/A Primers: p67phox Forward: 5’-CCCACTCCCGGATTTGCTTC -3’ Reverse: 5’-GTCTCGGTTAATGCTTCTGGTAA -3’ IDT N/A Primers: p47phox Forward: 5’-GGGGCGATCAATCCAGAGAAC -3’ Reverse: 5’-GTACTCGGTAAGTGTGCCCTG -3’ IDT N/A Primers: p22phox Forward: 5’-CCCAGTGGTACTTTGGTGCC -3’ Reverse: 5’-GCGGTCATGTACTTCTGTCCC -3’ IDT N/A Recombinant DNA pCS2+HyPer7-NES Addgene Cat# 136467 pCS2+MLS-HyPer7 Addgene Cat# 136470 pGEX-4T1 SHP-2 WT Addgene Cat# 8322 PET53-His-IDH1-Flag (Chen et al., 2019a) N/A PLHCX-IDH1-Flag (Chen et al., 2019a) N/A PLHCX-IDH1-Flag Y42/391F (Chen et al., 2019a) N/A PLHCX empty vector (Chen et al., 2019a) N/A Software and Algorithms GraphPad Prism 7 software GraphPad Software https://www.graphpad.com/ Image J NIH https://imagej.nih.gov/ij/index.html Highlights Tyrosine phosphorylation of IDH1 is commonly required for reductive carboxylation Reductive carboxylation attenuates SHP-2 to activate IDH1 Reductive carboxylation elevates cytosolic ROS to inhibit SHP-2 Diverse signals converge at the NOX2-SHP-2 axis to induce reductive carboxylation SIGNIFICANCE Reductive carboxylation describes an enhanced reductive formation of citrate from glutamine during diverse cellular stress processes including suppressed mitochondrial oxidation due to defects in ETC such as treatment with Complex I inhibitor rotenone, or hypoxia or detachment for anchorage-independent growth. Although lung cancer cells including A549 and H1299 have been extensively used for reductive carboxylation research, this phenomenon is not cancer-specific. Instead, reductive carboxylation is also important to support redox homeostasis as well as biosynthesis in quiescent fibroblasts, retinal pigment epithelium cells and many other types of “non-cancer” cells from different organs including liver, heart, and brown adipocytes. However, the signaling basis underlying induction of reductive carboxylation remains unclear. Here, we report that diverse signals converge at the NOX2-SHP-2 axis, leading to enhanced tyrosine phosphorylation and activation of IDH1 that is commonly required for induction of efficient reductive carboxylation. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. SUPPLEMENTAL INFORMATION Supplemental information includes four figures. DECLARATION OF INTERESTS The authors have no conflicts of interest to declare. REFERENCES Altenhofer S , Kleikers PW , Radermacher KA , Scheurer P , Rob Hermans JJ , Schiffers P , Ho H , Wingler K , and Schmidt HH (2012). The NOX toolbox: validating the role of NADPH oxidases in physiology and disease. Cell Mol Life Sci 69 , 2327–2343.22648375 Anvari E , Wikstrom P , Walum E , and Welsh N (2015). The novel NADPH oxidase 4 inhibitor GLX351322 counteracts glucose intolerance in high-fat diet-treated C57BL/6 mice. Free Radic Res 49 , 1308–1318.26118714 Bedard K , and Krause KH (2007). The NOX family of ROS-generating NADPH oxidases: physiology and pathophysiology. Physiol Rev 87 , 245–313.17237347 Belousov VV , Fradkov AF , Lukyanov KA , Staroverov DB , Shakhbazov KS , Terskikh AV , and Lukyanov S (2006). Genetically encoded fluorescent indicator for intracellular hydrogen peroxide. Nat Methods 3 , 281–286.16554833 Brandes RP , Weissmann N , and Schroder K (2014). Nox family NADPH oxidases: Molecular mechanisms of activation. Free Radic Biol Med 76 , 208–226.25157786 Casas AI , Kleikers PW , Geuss E , Langhauser F , Adler T , Busch DH , Gailus-Durner V , de Angelis MH , Egea J , Lopez MG , (2019). Calcium-dependent blood-brain barrier breakdown by NOX5 limits postreperfusion benefit in stroke. J Clin Invest 129 , 1772–1778.30882367 Chen D , Xia S , Wang M , Lin R , Li Y , Mao H , Aguiar M , Famulare CA , Shih AH , Brennan CW , (2019a). Mutant and Wild-Type Isocitrate Dehydrogenase 1 Share Enhancing Mechanisms Involving Distinct Tyrosine Kinase Cascades in Cancer. Cancer Discov 9 , 756–777.30862724 Chen PH , Cai L , Huffman K , Yang C , Kim J , Faubert B , Boroughs L , Ko B , Sudderth J , McMillan EA , (2019b). Metabolic Diversity in Human Non-Small Cell Lung Cancer Cells. Mol Cell 76 , 838–851 e835.31564558 Deliyanti D , and Wilkinson-Berka JL (2015). Inhibition of NOX1/4 with GKT137831: a potential novel treatment to attenuate neuroglial cell inflammation in the retina. J Neuroinflammation 12 , 136.26219952 Du ZD , Hu LT , Zhao GQ , Li Y , and Ma ZZ (2015). Protein tyrosine phosphatase 1B regulates the activity of retinal pigment epithelial cells. Mol Vis 21 , 523–531.25999679 Ermakova YG , Bilan DS , Matlashov ME , Mishina NM , Markvicheva KN , Subach OM , Subach FV , Bogeski I , Hoth M , Enikolopov G , (2014). Red fluorescent genetically encoded indicator for intracellular hydrogen peroxide. Nat Commun 5 , 5222.25330925 Fan J , Lin R , Xia S , Chen D , Elf SE , Liu S , Pan Y , Xu H , Qian Z , Wang M , (2016). Tetrameric Acetyl-CoA Acetyltransferase 1 Is Important for Tumor Growth. Mol Cell 64 , 859–874.27867011 Gilliland DG (2001). Hematologic malignancies. Curr Opin Hematol 8 , 189–191.11561153 Hirano K , Chen WS , Chueng AL , Dunne AA , Seredenina T , Filippova A , Ramachandran S , Bridges A , Chaudry L , Pettman G , (2015). Discovery of GSK2795039, a Novel Small Molecule NADPH Oxidase 2 Inhibitor. Antioxid Redox Signal 23 , 358–374.26135714 Hu W , Tian H , Yue W , Li L , Li S , Gao C , Si L , Qi L , Lu M , Hao B , (2016). Rotenone induces apoptosis in human lung cancer cells by regulating autophagic flux. IUBMB life 68 , 388–393.27015848 Jiang L , Shestov AA , Swain P , Yang C , Parker SJ , Wang QA , Terada LS , Adams ND , McCabe MT , Pietrak B , (2016). Reductive carboxylation supports redox homeostasis during anchorage-independent growth. Nature 532 , 255–258.27049945 Kenny HA , Lal-Nag M , White EA , Shen M , Chiang CY , Mitra AK , Zhang Y , Curtis M , Schryver EM , Bettis S , (2015). Quantitative high throughput screening using a primary human three-dimensional organotypic culture predicts in vivo efficacy. Nat Commun 6 , 6220.25653139 Lam JT , Hemminki A , Kanerva A , Lee KB , Blackwell JL , Desmond R , Siegal GP , and Curiel DT (2007). A three-dimensional assay for measurement of viral-induced oncolysis. Cancer Gene Ther 14 , 421–430.17235353 Li N , Ragheb K , Lawler G , Sturgis J , Rajwa B , Melendez JA , and Robinson JP (2003). Mitochondrial complex I inhibitor rotenone induces apoptosis through enhancing mitochondrial reactive oxygen species production. J Biol Chem 278 , 8516–8525.12496265 Metallo CM , Gameiro PA , Bell EL , Mattaini KR , Yang J , Hiller K , Jewell CM , Johnson ZR , Irvine DJ , Guarente L , (2012). Reductive glutamine metabolism by IDH1 mediates lipogenesis under hypoxia. Nature 481 , 380–384. Moon JS , Nakahira K , Chung KP , DeNicola GM , Koo MJ , Pabon MA , Rooney KT , Yoon JH , Ryter SW , Stout-Delgado H , (2016). NOX4-dependent fatty acid oxidation promotes NLRP3 inflammasome activation in macrophages. Nat Med 22 , 1002–1012.27455510 Mullen AR , Wheaton WW , Jin ES , Chen PH , Sullivan LB , Cheng T , Yang Y , Linehan WM , Chandel NS , and DeBerardinis RJ (2012). Reductive carboxylation supports growth in tumour cells with defective mitochondria. Nature 481 , 385–388. Niogret C , Miah SMS , Rota G , Fonta NP , Wang H , Held W , Birchmeier W , Sexl V , Yang W , Vivier E , (2019). Shp-2 is critical for ERK and metabolic engagement downstream of IL-15 receptor in NK cells. Nat Commun 10 , 1444.30926899 Panday A , Sahoo MK , Osorio D , and Batra S (2015). NADPH oxidases: an overview from structure to innate immunity-associated pathologies. Cell Mol Immunol 12 , 5–23.25263488 Qu CK (2000). The SHP-2 tyrosine phosphatase: signaling mechanisms and biological functions. Cell Res 10 , 279–288.11191350 Sourbier C , Ricketts CJ , Liao PJ , Matsumoto S , Wei D , Lang M , Railkar R , Yang Y , Wei MH , Agarwal P , (2019). Proteasome inhibition disrupts the metabolism of fumarate hydratase- deficient tumors by downregulating p62 and c-Myc. Sci Rep 9 , 18409.31804603 Tanner JJ , Parsons ZD , Cummings AH , Zhou H , and Gates KS (2011). Redox regulation of protein tyrosine phosphatases: structural and chemical aspects. Antioxid Redox Signal 15 , 77–97.20919935 Tsutsumi R , Harizanova J , Stockert R , Schroder K , Bastiaens PIH , and Neel BG (2017). Assay to visualize specific protein oxidation reveals spatio-temporal regulation of SHP2. Nat Commun 8 , 466.28878211 Williams NC , and O'Neill LAJ (2018). A Role for the Krebs Cycle Intermediate Citrate in Metabolic Reprogramming in Innate Immunity and Inflammation. Front Immunol 9 , 141.29459863 Wise DR , Ward PS , Shay JE , Cross JR , Gruber JJ , Sachdeva UM , Platt JM , DeMatteo RG , Simon MC , and Thompson CB (2011). Hypoxia promotes isocitrate dehydrogenase-dependent carboxylation of alpha-ketoglutarate to citrate to support cell growth and viability. Proc Natl Acad Sci U S A 108 , 19611–19616.22106302 Wu X , Xu G , Li X , Xu W , Li Q , Liu W , Kirby KA , Loh ML , Li J , Sarafianos SG , (2019). Small Molecule Inhibitor that Stabilizes the Autoinhibited Conformation of the Oncogenic Tyrosine Phosphatase SHP2. J Med Chem 62 , 1125–1137.30457860 Yang SG , Park HJ , Kim JW , Jung JM , Kim MJ , Jegal HG , Kim IS , Kang MJ , Wee G , Yang HY , (2018). Mito-TEMPO improves development competence by reducing superoxide in preimplantation porcine embryos. Sci Rep 8 , 10130.29973637 Yu WM , Hawley TS , Hawley RG , and Qu CK (2003). Catalytic-dependent and - independent roles of SHP-2 tyrosine phosphatase in interleukin-3 signaling. Oncogene 22 , 5995–6004.12955078

PMC009xxxxxx/PMC9308734.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9802571 20730 Mol Cell Mol Cell Molecular cell 1097-2765 1097-4164 35662398 9308734 10.1016/j.molcel.2022.05.014 NIHMS1808307 Article Stressful steps: progress and challenges in understanding stress-induced mRNA condensation and accumulation in stress granules Glauninger Hendrik 1 Wong Hickernell Caitlin J. 1 Bard Jared A.M. 1 Drummond D. Allan 1 1 Department of Biochemistry & Molecular Biology, University of Chicago, Chicago, Illinois 60673, USA Correspondence: dadrummond@uchicago.edu 16 5 2022 21 7 2022 03 6 2022 21 7 2023 82 14 25442556 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Stress-induced condensation of mRNA and protein into massive cytosolic clusters is conserved across eukaryotes. Known as stress granules when visible by imaging, these structures remarkably have no broadly accepted biological function, mechanism of formation or dispersal, or even molecular composition. As part of a larger surge of interest in biomolecular condensation, studies of stress granules and related RNA/protein condensates have increasingly probed the biochemical underpinnings of condensation. Here, we review open questions and recent advances, including the stages from initial condensate formation to accumulation in mature stress granules; mechanisms by which stress-induced condensates form and dissolve; and surprising twists in understanding the RNA components of stress granules and their role in condensation. We outline grand challenges in understanding stress-induced RNA condensation, centering on the unique and substantial barriers in the molecular study of cellular structures, such as stress granules, for which no biological function has been firmly established. Graphical Abstract eTOC blurb Stress granules (SGs), canonical membraneless organelles, have seen a surge of activity in identifying their protein/RNA components and biophysical underpinnings. Glauninger et al. review our understanding of SGs and their biomolecular condensate precursors, emphasizing conceptual and methodological challenges arising from the lack of established biological functions for these conserved structures. pmcFrom humans and other vertebrates to single-celled yeasts, from plants to protozoa, the onset of primordial stresses such as heat shock, oxidizing agents, hypoxia, and starvation is rapidly followed by the intracellular condensation and accumulation of myriad proteins and mRNAs in cytosolic clusters (Cherkasov et al., 2013; Decker and Parker, 2012; Farny et al., 2009; Jain et al., 2016; Kedersha et al., 2000, 1999; Kramer et al., 2008; Nover et al., 1989; Wallace et al., 2015). These enigmatic structures, called stress granules when they grow large enough to resolve by microscopy, have become standard examples of so-called membraneless organelles alongside nucleoli, processing (P) bodies, paraspeckles, and others (Alberti and Carra, 2018; Boeynaems et al., 2018; Brangwynne, 2013; Gomes and Shorter, 2019; Guo and Shorter, 2015; Lyon et al., 2020; Mitrea and Kriwacki, 2016). Stress granules and their condensed molecular precursors have become a nexus of extraordinary recent activity because of the involvement of protein and RNA liquid-liquid phase separation (LLPS) in their formation (Guillén-Boixet et al., 2020; Molliex et al., 2015; Riback et al., 2017; Sanders et al., 2020; Van Treeck et al., 2018; Wheeler et al., 2016; Yang et al., 2020) and hints that dysregulation of condensation and stress granule formation contribute to disease (Bosco et al., 2010; Patel et al., 2015). Yet despite sustained and vigorous inquiry, a remarkable array of foundational questions remain unanswered. What do stress granules do, if anything? What are the functional consequences of condensation, and what functions do specific mechanisms of condensation, such as LLPS, carry out? (Throughout this review we explicitly intend “condensate” to be a catch-all term for membraneless clusters without any further stipulation as to their structure, process of formation, or adaptive significance (Box 1), largely following standard usage (Banani et al., 2017; Lyon et al., 2020).) What biological roles are played by molecular-level condensation events versus subsequent merging of these condensates into larger, microscopically visible structures? How do condensation and accumulation occur, and are these processes mediated mainly by intrinsic molecular forces or extrinsic cellular machinery such as cytoskeleton-associated motors? To what extent are stress-triggered condensation and stress granule accumulation processes and participants conserved over evolutionary time? Among the deepest challenges in studying stress granules is that, in the absence of molecular functions and cellular phenotypes, the phenomenon itself is operationally rather than biologically defined: a stress granule consists of anything which forms microscopically visible foci which colocalize with established stress granule markers (cf. Box 1). Although these structures have been hypothesized to play a variety of cellular roles, their function remains unclear (Buchan et al., 2011; Ivanov et al., 2019; Kedersha and Anderson, 2002, 2009; Kedersha et al., 2000). That stress granules are termed “membraneless organelles,” where the latter word explicitly means a cellular structure which performs distinct functions, has served to create the unfortunate impression that this fundamental question has been answered. This question of function applies not only to stress granules, but also to the broader study of cytoplasmic ribonucleoprotein (RNP) foci including P-bodies, RNA transport granules, and P granules. In some cases, such as RNA transport granules in neurons, the question of function has been more directly addressed (Kiebler and Bassell, 2006; Pushpalatha and Besse, 2019). However, in many cases function is still presented as a model. P-bodies were long presumed to be sites of RNA degradation (Aizer et al., 2014; Franks and Lykke-Andersen, 2007; Sheth and Parker, 2003), but this model has been challenged (Eulalio et al. 2007; Hubstenberger et al. 2017). Additionally, work on G3BP1 aggregates in axons shows that condensates composed of canonical stress granule proteins may play a role under non stress conditions, introducing basal stress granule-like condensates (Sahoo et al., 2018, 2020). The questions and challenges regarding stress granules raised here apply to other biomolecular condensates, purported membraneless organelles, and contexts beyond cell stress. As efforts to develop a parts list for stress granules (Buchan et al., 2011; Cherkasov et al., 2015; Jain et al., 2016; Wallace et al., 2015) have proceeded alongside attempts to recapitulate in vitro certain molecular events such as stress-reactive condensation and RNA recruitment (Begovich and Wilhelm, 2020; Iserman et al., 2020; Riback et al., 2017; Van Treeck et al., 2018), evidence has emerged for multiple quasi-independent contributing pathways, multiple molecular stages, and multiple levels of organization in stress granules and their precursors. This will serve as our jumping-off point. Given the multiple levels of molecular organization known to contribute to stress-induced RNA condensation, how do these levels interrelate, and at what level are adaptive features best understood? Throughout this review, we intend a larger question to lurk in the reader’s mind. How can the characterization, interrogation, isolation, and reconstitution of stress-induced protein/RNA condensates and stress granules be effectively guided and evaluated in the absence of established functions, biological activities, or cellular phenotypes? Multiple stages of stress-induced RNA condensation and stress granule formation What is the relationship between protein/mRNA biomolecular condensation and stress granule formation? Although these processes are sometimes considered synonymous, and although how initial condensates accumulate in microscopically visible foci remains largely unknown, the existence of multiple stages in stress granule formation has long been understood (Figure 1). Existing models commonly reflect hierarchical organization in stress granules, with some stable components (“core”) surrounded by more dynamic components (“shell”) (Jain et al., 2016; Wheeler et al., 2016), or nanoscopic “seeds” interacting and merging to form stress granules (Padrón et al., 2019; Panas et al., 2016). Evidence for these multiple stages comes from several independent sources. First, individual core markers for stress granules such as poly(A)-binding protein, G3BP, and Ded1 can be purified recombinantly and will autonomously condense in response to stress-associated physiological cues (e.g. heat shock, presence of long ribosome-free mRNA) in vitro (Guillén-Boixet et al., 2020; Iserman et al., 2020; Kroschwald et al., 2018; Riback et al., 2017; Yang et al., 2020). These in vitro results suggest that condensation in vivo may not depend on interactions between a large set of stress granule components, at least at initial stages. Second, although formation of canonical microscopically visible stress granules can be blocked by translation elongation inhibitors (Kedersha et al., 2000; Nadezhdina et al., 2010; Namkoong et al., 2018; Wallace et al., 2015), the stress-triggered condensation, as measured by biochemical fractionation, of stress granule components such as poly(A)-binding protein proceeds virtually unaffected by such inhibition, indicating that accumulation of condensates into stress granules is a separate step (Wallace et al., 2015). This suggests that formation of canonical stress granules involves cell-biological transport processes which bring multiple components together in the cytosol (Panas et al., 2016). In support of this model, depolymerization of microtubules disrupts stress granule accumulation (Ivanov et al., 2003a, 2003b), and stress granules tether to the endoplasmic reticulum and lysosomes using specific factors for intracellular transport (Liao et al., 2019). Similarly, in contrast to in vitro ATP-independent condensation processes, ATP-driven mechanisms are required for stress granule formation in cells (Jain et al., 2016). Transport and accumulation of small condensates and other components is a separate process from the initial condensation events which also accompany stress. Finally, the appearance of canonical stress granulesstress granules generally depends on stress intensity and duration, and in important cases, low levels of stress cause condensation of protein constituents but not their stress granule accumulation. For example, heat shock in budding yeast leads to biochemically detectable condensation of certain proteins after 8 minutes at 37 °C or 42 °C, and accumulation of certain proteins in cytosolic foci, but formation of classic stress granules marked by poly(A)-binding protein requires pushing temperatures to 44–46 °C at this timescale (Cherkasov et al., 2013; Wallace et al., 2015). Limitations of imaging techniques may contribute to this discrepancy to some degree (see our discussion of grand challenges below), and exciting developments of improved microscopy-based methods—such as lattice light-sheet microscopy or fluorescence cross-correlation spectroscopy—may help minimize these concerns in the future (Guillén-Boixet et al., 2020; Peng et al., 2020). But the differential accumulation of protein factors at different levels of stress intensity (Grousl et al., 2013) rules out simplistic notions that, for example, stress granules are merely small at first and grow larger with intensifying stress. More evidence for an ordered assembly of stress granules comes from time-resolved proximity labeling experiments, which identified the interactome of the stress granule component eIF4A1 during heat shock of HEK293 cells (Padrón et al., 2019). This study found that certain canonical stress granule components interacted with eIF4A1 before others. Thus, assembly proceeds in separable stages, ending with accumulation in large foci under severe stress. The existence of assembly stages naturally raises the question: at what stages might specific functions be carried out? A deeper question haunting the field is: what do stress granules actually do? Elusive functions of stress granules and stress-triggered RNA condensation No commonly accepted function for stress granules yet exists. Many functions have been proposed, implicating stress granules in a range of roles, including sequestration of mRNAs and proteins; protection of mRNAs and proteins from degradation; promotion of enzymatic activities by increasing local concentration; minimization of cellular energy expenditure; and acting in translational quality control, signaling, and cargo delivery (Aronov et al., 2015; Buchan and Parker, 2009; Escalante and Gasch, 2021; Ivanov et al., 2019; Kedersha and Anderson, 2002; Kedersha et al., 2013; Mahboubi and Stochaj, 2017; Moon et al., 2020). Stress granules have also been implicated in suppressing cell death by sequestering pro-apoptotic factors such as receptor of activated C kinase 1 (RACK1) (Arimoto et al., 2008; Tsai and Wei, 2010). Similarly, a recent study found that stress granule formation suppressed pyroptosis, a form of cell death associated with inflammation, by sequestering the protein DEAD-box helicase 3 X-linked (DDX3X) (Samir et al., 2019). However, the large variety of functions proposed for stress granules, combined with some conflicting findings, have made it difficult to form an overarching model of stress granule function (Mateju and Chao, 2021). For instance, an oft-speculated function for RNA condensation is transiently protecting transcripts from degradation during stress (Hubstenberger et al., 2017; Moon et al., 2019; Sorenson and Bailey-Serres, 2014), yet other work finds no effect on mRNA half-life following stress granuleinhibition (Bley et al., 2015). Another model holds that RNA condensation contributes to selective translation of non-condensed transcripts. Stress-induced transcripts are often translated in the midst of global translational shutoff. Some transcripts that are highly translated during stress, such as HSP70 and HSP90, do not associate with stress granules, suggesting a connection between translation and escaping condensation (Kedersha and Anderson, 2002; Stöhr et al., 2006; Zid and O’Shea, 2014). Certain translation initiation factors also condense, raising the possibility that a combination of protein and RNA sequestration can help promote selective translation during stress (Iserman et al., 2020; Wallace et al., 2015). However, stress granules are not required for global translational shutoff, so this selective translation would occur on top of a more dominant effect (Escalante and Gasch, 2021). Additionally, translation has been observed inside stress granules, complicating this model (Mateju et al., 2020). A potential resolution to these conflicting results may be that particular functions are carried out at specific stages of organization. For example, stabilization of RNA by sequestration can conceivably occur at the pre-microscopic condensate level whereas other proposed functions may require collection of components into a larger and more molecularly diverse body (Figure 2). Hypothetically, a study in which perturbations block stress granule accumulation but not initial condensation, with no effect on RNA stabilization, would reach different conclusions than a study in which perturbations block both processes. An expanded understanding of assembly stages, a deepened grasp of the molecular drivers of these stages, and a widened array of perturbations capable of targeting specific stages and molecular determinants will be needed to sort out these questions. Less discussed in the field are the issues inherent in studying biological phenomena whose functional contributions, if any, are unclear. Purification and reconstitution strategies, deprived of an activity-based standard for measuring success, must instead rely on morphological or compositional metrics whose relationship with biological function remains to be established (Begovich and Wilhelm, 2020; Freibaum et al., 2021). The lack of functional insight is compounded by the remarkable lack of standard cellular phenotypes in the study of stress granules. Because not all of a given protein or RNA localizes to stress granules, determining a function must come from specifically perturbing condensation behavior without influencing activity, localization, or expression level. Even at the condensate level, phenotypes have been difficult to establish, although an allelic series of mutations which suppress poly(A)-binding protein’s heat-triggered condensation in vitro and in vivo also suppress growth during heat stress (Riback et al., 2017). The rarity of such phenotypes, particularly for stress granules, has led to a lingering question of whether stress granules may often simply be byproducts of other cellular changes (Mateju and Chao, 2021). Informing functions of stress-triggered condensation through the lens of disease Some promising directions in uncovering stress granule function have come through study of disease contexts. Stress granules are induced by viral infection, where their formation has been proposed to help restrict viral replication (Eiermann et al., 2020). In fact, many viruses have developed strategies for preventing stress granule formation by, for instance, sequestering or cleaving key stress granule components (Katoh et al., 2013; White et al., 2007). What function do stress granules serve that viruses are so intent on disrupting? One possibility is that stress granules could sequester viral RNA, similar to their proposed function in storing cellular mRNAs (Burgess and Mohr, 2018; Law et al., 2019). However, as discussed above, it is difficult to conclude whether recruitment of viral RNA to stress granules is required for proposed functions without mutations that specifically perturb stress granule formation while preserving separate molecular functions of stress granule components. One such perturbation comes from recent work showing that chikungunya virus promotes stress granule disassembly through the ADP-ribosylhydrolyase activity of nonstructual protein 3 (nsP3) (Abraham et al., 2018; Akhrymuk et al., 2018; Jayabalan et al., 2021). Removing this activity from nsP3 preserves stress granules during infection, providing a manipulatable system for future studies of stress granule function without deletion of any host machinery. The stressful environments inhabited by tumors—such as nutrient deprivation, hypoxia, increased reactive oxygen species, and perturbed protein folding resulting from the dysregulation of metabolism and growth in malignancy—makes cancer biology a useful model for studying the functions of stress-induced condensation (Ackerman and Simon, 2014; Anderson et al., 2015; Clarke et al., 2014; Gorrini et al., 2013). Moreover, certain chemotherapy drugs trigger cancer cells to form stress granules, which are generally thought to be pro-survival, leading to condensation modulation as a potential target for therapeutics (Fournier et al., 2010; Gao et al., 2019; Kaehler et al., 2014). In contrast, another chemotherapy agent, sodium selenite, triggers non-canonical stress granules lacking certain components whose stress granule localization has been linked to cell survival. These non-canonical stress granules have thus been suggested to be less functional in the stress response (Fujimura et al., 2012). Additional work aimed at understanding the precise differences in stress-induced condensation between the considered pro-survival canonical and the non-canonical stress granules, at both the stress granule and pre-microscopic condensate level, will help inform the functions of condensation in response to stress and perhaps even inform the importance of its organization at the size/spatial levels. Further underscoring the potential role of condensation in the pathogenesis of cancer, recent work studying myeloid malignancies has shown that specific driver mutations upregulate stress granule formation, which is linked to increased stress adaptation and cancer development (Biancon et al., 2022). Additionally, work with disease mutations related to neurodegenerative diseases suggests a relationship between maladaptive protein aggregates and adaptive condensates like stress granules, suggesting that maladaptive aggregates may occur when stress granules are not properly disassembled (Gal et al., 2016; Gwon et al., 2021; Mackenzie et al., 2017). Even so, our understanding of these maladaptive protein aggregates will be limited without a deeper understanding of the function of adaptive condensates. Without understanding the functions of stress-induced condensation, we can only speculate on the pathophysiology of persistent stress granules. While many studies of stress granules focus on proteins which, when fluorescently tagged, are easily visible microscopically, RNA sits at the center of stress granule formation and function. We thus begin with a consideration of how our understanding of RNA’s role has changed as new methods have come into use. The role of RNA: old observations and emerging results The accumulation of poly(A)-RNA is among the defining features of stress granules. Moreover, the role of mRNA in stress granule formation has long been known. Among the most crucial experiments is the demonstration that translational inhibition affects stress granule formation in a mechanistically specific way: elongation inhibitors such as cycloheximide and emetine, which freeze ribosomes on mRNA, block stress granule formation, whereas puromycin, which prematurely terminates translation and frees mRNA of ribosomes, promotes stress granule formation (Bounedjah et al., 2014; Kedersha et al., 2000; Namkoong et al., 2018; Wallace et al., 2015). Inhibition of transcription also inhibits stress granule formation (Bounedjah et al., 2014; Khong et al., 2017a), further underscoring the role of RNA, at least at the accumulation stage. But which RNAs? How does RNA contribute to condensation and stress granule formation? To what extent does RNA drive condensation or accumulation, and to what extent is it passively dragged along? Early important results showed that prominent stress-induced mRNAs are selectively excluded from stress granules in both plant and mammalian cells (Kedersha and Anderson, 2002; Nover et al., 1989; Stöhr et al., 2006; Zid and O’Shea, 2014). Because stress granules are, by most metrics, accumulation sites for translationally repressed mRNAs, and because it is both biologically appealing and empirically established in some systems that stress-induced transcripts are well-translated (Preiss et al., 2003; Zid and O’Shea, 2014), these early results placed stress granules at the center of translational regulation during stress. But these foundational results have not survived into the recent era dominated by high-throughput studies, where transcriptome-scale effects can be observed. Modern studies do not find substantial depletion of stress-induced mRNAs from stress granules; instead, recent studies employing diverse approaches have converged on transcript length as the key correlate of mRNA recruitment to stress granules. Messenger RNA length is the dominant correlate of their enrichment in the transcriptome associated with purified stress granule cores and stress-associated RNA granules (Khong et al., 2017b; Matheny et al., 2019, 2021; Namkoong et al., 2018); in in vitro systems, increasing RNA length promotes RNA/protein phase separation organized by the stress-granule hub G3BP1 (Guillén-Boixet et al., 2020; Yang et al., 2020); and single-molecule studies show that mRNA length correlates with the dwell-time of mRNAs on stress granules and other condensed structures (Moon et al., 2019). An increased concentration of ribosome-free mRNA following stress-induced translational shutdown is considered the key trigger for stress granule formation (Hofmann et al., 2020), and inhibition of translation initiation triggers condensation, such as in stress, eIF2α phosphorylation, or inhibition of the initiation factor eIF4A (Buchan et al., 2008; Iserman et al., 2020; Kedersha et al., 1999; Mazroui et al., 2006; Riback et al., 2017). (Fig. 3). This model is supported by several lines of evidence: 1) global translation initiation downregulation and subsequent polysome collapse is associated with RNA condensation during stress (Cherkasov et al., 2013), 2) prevention of polysome collapse during stress blocks stress granule formation (Kedersha et al., 2000), 3) transfection of translationally arrested cells with free mRNA triggers stress granule formation (Bounedjah et al., 2014), 4) inhibiting eIF4A, an essential translation initiation factor, promotes stress granule formation (Dang et al., 2006; Low et al., 2005; Mazroui et al., 2006; Tauber et al., 2020a). Alongside these data, early and still-current alternative models in which RNA length plays a minimal role exist. For example, stalled preinitiation complexes (PICs) which accumulate during stress may in part form the core of stress granules (Kedersha et al., 2002) (Fig. 3). Beyond ribosome-free RNA, a role of RNA length makes intuitive biophysical sense, because the number of opportunities for either RNA-RNA or protein-RNA interactions—i.e., valence—naturally scales with length, all else being equal (Jain and Vale, 2017). Evidence for a role from RNA-RNA interactions is circumstantial, resting on partial recapitulation of some stress granule transcriptome features in vitro using only purified RNA (Van Treeck et al., 2018), the dependence of in vitro phase separation on long, unfolded RNAs (Guillén-Boixet et al., 2020; Yang et al., 2020) and RNA helicases (Tauber et al., 2020a). Further discussion of the available evidence supporting the roles of RNA-RNA or protein-RNA interactions can be found in several informative reviews (Campos-Melo et al., 2021; Hofmann et al., 2020; Ripin and Parker, 2021; Van Treeck and Parker, 2018). Even though a dominant role for RNA length is sensible biophysically, it is puzzling biologically. The overwhelming consensus holds that stress granules are accumulation sites for mRNA whose translation is suppressed during stress. Yet the length-driven model (and existing results supporting it) suggests that induction of long transcripts during stress would be futile for protein production, because long transcripts would be immediately recruited into translationally silent stress granules. However, while evidence that long transcripts are translationally silenced during stress after their stress granule recruitment is lacking, it has been hypothesized that shorter transcripts may be associated with rapid responses, which could help resolve the paradox (Lopes et al., 2021). However, an important caveat is that mRNA length is also a natural confounding variable in experiments and analyses. Sedimentation by centrifugation is employed in most transcriptome-scale studies aimed at isolating stress granule-associated mRNAs, mirroring the use of sedimentation in proteome-scale studies of stress granule-associated proteins (Cherkasov et al., 2015; Jain et al., 2016; Wallace et al., 2015). But unlike proteins, long RNAs, due to their size—an mRNA weighs roughly an order of magnitude more than the protein it encodes—will tend to sediment whether or not they are in a condensate. Consequently, comparing stress and non-stress conditions is crucial to determining the extra sedimentation due to stress. However, as others have pointed out (Namkoong et al., 2018), the original study (Khong et al., 2017b) reporting yeast and mammalian stress granule transcriptomes, and reporting the profound effect of length, did not include non-stress controls. Long RNAs may stick nonspecifically to affinity reagents in pulldowns due to their valence or increased structure (Sanchez de Groot et al., 2019). Although subsequent controlled work in mammalian cells has confirmed the accumulation of longer RNAs in granules following ER or oxidative stress (Matheny et al., 2019; Namkoong et al., 2018), the effects are more modest, and no non-stress control is yet available in yeast. Reduced translational efficiency (TE) has also been reported to be a major contributor to stress granule RNA accumulation. However, the two measures of TE used—codon optimality and ribosome density—have long been known to be inversely correlated with transcript length (Arava et al., 2005; Duret and Mouchiroud, 1999; Weinberg et al., 2016), raising the question of whether TE is a causal contributor to mRNA recruitment or a spurious correlation. Sedimentation-independent methods to examine recruitment of mRNAs, such as mRNA fluorescence in situ hybridization (FISH) in intact cells, have covered only a handful of targets (Khong et al., 2017b; Matheny et al., 2019), reported only a modest stress granule recruitment effect from length, and concluded that “length, per se, is not the major driving force in stress granule enrichment” (Matheny et al., 2021). Large-scale, well-controlled, systematic studies of the effect of length will be useful in resolving lingering uncertainty. Given the sharp change in the apparent biology of RNA recruitment to stress granules from early to present-day studies, the limited set of transcriptome-scale studies available at this writing, and the challenging nature of isolating molecular components of functionally ill-defined structures, the RNA components of stress-induced condensates and stress granules will continue to be an area of intense investigation. Mechanisms of dissolution How do stress-induced RNA condensates dissolve after stress, as cells return to basal operations? Dissolution appears to be a regulated, controlled process that relies on specific proteins (Hofmann et al., 2020; Marmor-Kollet et al., 2020). Proteins categorized as molecular chaperones and autophagic proteins have been implicated in stress granule dissolution, as have proteins associated with post-translational modifications (PTMs) such as sumoylation, ubiquitination, and phosphorylation (Buchan et al., 2013; Cherkasov et al., 2013; Gwon et al., 2021; Keiten-Schmitz et al., 2020; Marmor-Kollet et al., 2020; Maxwell et al., 2021; Shattuck et al., 2019; Yoo et al., 2021). Work in yeast has revealed that heat-induced (42 °C) protein aggregates are entirely reversible, which is incompatible with autophagy and suggests that different fates occur in different stresses (Wallace et al., 2015). Recent work shows that molecular chaperones can dissolve stress-triggered protein condensates orders of magnitude more efficiently than misfolded reporter proteins in vitro, suggesting that molecular chaperones may have evolved to interact with stress-induced condensates (Yoo et al., 2021). Additionally, recent work in mammalian cells has shown that stress granules can be eliminated through either an autophagy-independent disassembly process or autophagy-dependent degradation, depending on the severity and acuteness of the initial stress (Gwon et al., 2021; Maxwell et al., 2021). This work suggests that the disassembly of stress granules is related to the initial stress, suggesting that different methods of assembly may require different methods of disassembly. The kinetics of stress granule dissolution may be tied to a functional role, such as translational control. If stress-induced condensates are sites of storage, the contents must be disassembled in a timely manner. It has been proposed that stress granules dissolve in discrete steps, where an initial shell is pulled away followed by a core, with particular proteins being recruited at distinct stages (Wheeler et al., 2016). Proteins necessary for cell recovery from stress, such as translation initiation factors, may need to be dispersed earlier than other stress granule core proteins that are dissolved more slowly. In fact, proper disassembly of stress granules was shown to be required for recovering cellular activities, such as translation, after stress (Maxwell et al., 2021). The dissolution of stress-induced condensates may be related to maladaptive insoluble protein aggregates that are often associated with diseases, motivating a further understanding of the mechanism and function of dissolution (Hofmann et al., 2020). However, as the function of stress granules remains unclear, the lack of functional assays demands careful experimental perturbations and cautious conclusions. For example, condensates that are no longer visible by microscopy may still occupy a conformation distinct from a monomeric form. New findings about the material state and assembly process of stress-induced condensates will illuminate the dissolution process, addressing questions such as whether the multiple steps of dissolution are equivalent to the stages of assembly or if a change in material state may lead to a different dissolution process. On this front, the role of liquid-liquid phase separation in stress granule formation may have crucial consequences for how these structures dissolve. Examining the role of liquid-liquid phase separation in stress-induced condensation Liquid-liquid phase separation (LLPS) is a thermodynamically driven mechanism by which a solution of a compound demixes into a dilute and a dense phase above a certain critical concentration (Hyman et al., 2014). A host of stress granule-associated proteins have been shown to undergo phase separation in vivo and in vitro (Guillén-Boixet et al., 2020; Iserman et al., 2020; Kroschwald et al., 2018; Molliex et al., 2015; Riback et al., 2017; Sanders et al., 2020; Yang et al., 2020), and it is widely held that stress granule assembly is driven by LLPS (reviewed in Hofmann et al., 2020). Recent work has converged on G3BP as a central node in LLPS-driven stress granule formation (Guillén-Boixet et al., 2020; Sanders et al., 2020; Yang et al., 2020), yet G3BP is dispensable for stress granule formation in response to certain stressors, such as heat and osmotic shock (Kedersha et al., 2016; Matheny et al., 2021). Thus, G3BP-focused models of stress granule formation may overly simplify the complex process of stress-induced condensation. Using LLPS as an assembly mechanism provides key advantages beneficial for responding to stress. The ultra-cooperativity of LLPS enables proteins to precisely sense and respond to small changes in their environments (Yoo et al., 2019). For instance, in yeast Ded1 autonomously condenses in response to temperature stress. Ded1 from a cold-adapted yeast condenses at lower temperatures than that of S. cerevisiae, while Ded1 from a thermophilic yeast condenses at higher temperatures (Iserman et al., 2020). This correlates with the fact that each yeast species has evolved to trigger its heat shock response relative to its environmental niche. Other key advantages of LLPS include that it enables passive (energy-independent) cellular reorganization and that it is reversible. Following the removal of the stress stimulus, LLPS would no longer be energetically favored, and the system would spontaneously return to basal conditions. Biomolecular condensation can result in the concentration of protein and RNA molecules into phases with a variety of material states. How could a condensate’s material state—how liquid-like or solid-like it is—affect its function? More solid-like condensates have been linked to disease, as pathogenic mutations of certain condensing proteins such as FUS increase aging and a loss of liquid-like properties over time (Patel et al., 2015). This thinking extends to RNA condensates as well, as it has been proposed that RNA helicases prevent RNA-RNA entanglement to maintain a liquid-like condensed state (Tauber et al., 2020a, 2020b). Further, the viscoelasticity of the nucleolus has been linked with enabling the vectorial release of properly folded ribosomes (Riback et al., 2022). Yet, the material state of stress-induced condensates does not appear to be widely conserved across eukaryotes, which like other evolutionarily variable features would usually be taken as evidence that the material state is not central to function. For instance, yeast stress granules are more solid-like than those of metazoa (Kroschwald et al., 2015), although there are methodological caveats (Wheeler et al., 2016). Reconstituted heat-induced condensates of the yeast stress granule protein Pab1 are solids (Riback et al., 2017) which are not spontaneously reversible, even though these condensates are readily dispersed by endogenous molecular chaperones (Yoo et al., 2021). Even within an organism, pH-induced condensates of the yeast stress granule protein Pub1 are more liquid-like than those induced by heat shock—and only the heat-induced condensates depend on chaperones (Kroschwald et al., 2018)—yet both conditions are thought to be physiologically relevant. The apparent lack of conservation of the material state can be rationalized when we consider that a condensate’s material state appears irrelevant for many of the functions ascribed to stress granules. For example, if the role of stress-induced condensation is to temporarily store housekeeping mRNA to enable the preferential translation of stress-response messages, how liquid-like the storage compartment is may be of minor importance. Additionally, if the function is to sequester certain proteins to perturb a given signaling pathway in the cytoplasm, the key feature is to deplete the protein from the dilute phase, and the liquidity of the dense phase is less relevant. On the other hand, if the material state is particularly relevant for the potential pathogenicity of condensates, then the evolutionary pressures on material state in different organisms may differ substantially even if stress granules have a conserved cellular function. Hazards in defining stress granule composition Defining the composition of stress granules is complicated by a number of factors, even setting aside the existential problem of what constitutes a biologically important structure in the absence of well-established functions and phenotypes. Nevertheless, the obvious consistency and evolutionary conservation of the accumulation of some proteins and RNAs into large foci has led to a sustained effort to identify lists of molecular components involved in the lifecycle of stress granules. Individual mRNAs and proteins can be localized to microscopically visible foci of stress granule markers (Cherkasov et al., 2015; Khong et al., 2017b; Mateju et al., 2020; Moon et al., 2019, 2020; Wallace et al., 2015; Wilbertz et al., 2019). On a larger scale, the stress granule interactome has been defined using a variety of techniques, many of which rely on using individual stress granule components, such as poly(A)-binding protein, G3BP1, TIA1, and eIF4A, as bait proteins and then assessing the mRNAs and proteins which interact with that bait. The interactors have been identified through immunoprecipitations, purification of particles containing a bait fused to a fluorescent protein, and by biotin proximity labeling (Hubstenberger et al., 2017; Khong et al., 2017b; Namkoong et al., 2018; Padrón et al., 2019; Somasekharan et al., 2020). Additionally, proximity labeling methods have found similar interactomes between stress granule proteins prior to stress and during stress (Markmiller et al., 2018; Youn et al., 2018). This may indicate that stress granules are mainly stabilized by enhancements of basal interactions, or that the interactions which distinguish stress granules are labile or refractive to these methods. The different levels of organization in stress-triggered condensation and stress granule formation, along with diverse methods whose relative accuracy can be difficult to establish given the ill-defined nature of the target, combine to create a challenging experimental landscape (Fig. 4). Unlike a membrane-bound mitochondrion or a relatively compositionally stable ribosome, stress-induced condensates and stress granules lack features which might simplify their description. A hallmark of biomolecular condensation is that many of the components of the condensate individually associate through weak, dynamic interactions (Alberti and Hyman, 2021). No biologically clear cutoff for interaction strength exists, making it unclear how to decide if a given component is part of the structure or not. For instance, many transcripts have been observed to associate only briefly with stress granule proteins (Wilbertz et al., 2019). How long must an mRNA reside at a stress granule to be considered a component? Additionally, consistent but weak associations may be lost during the isolation steps necessary for sequencing, mass spectrometry, or other biochemical methods. Perhaps certain molecular components form a scaffold to which client proteins are recruited (Campos-Melo et al., 2021; Shiina, 2019; Zhang et al., 2019). Differences in interaction strength may reveal biologically important differences; for example, major molecular chaperones associate with stress granules by colocalization (Cherkasov et al., 2013), but do not co-fractionate with stress-triggered condensates (Wallace et al., 2015). Should such chaperones be considered a component of stress granules, merely associates, or something else? Here again, functional assays would sharpen these distinctions in crucial ways. Because stress granules are operationally defined as microscopic foci marked by specific proteins, the definition of the structure is unfortunately entwined with technical limitations and with compositional preconceptions. Failure to observe foci microscopically, for example at low levels of stress, are consistent with two distinct biological possibilities: the absence of condensates entirely, or the formation of structures below the diffraction limit which still retain key properties of larger condensates (Guzikowski et al., 2019). Likewise, failure to observe colocalization with a specific marker molecule may reflect legitimate biological variation either in the marker itself or in the structure being marked. Finally, the composition of stress granules is not static, but depends on the nature of the stress and also changes over time (Aulas et al., 2017; Buchan et al., 2011; Padrón et al., 2019; Reineke and Neilson, 2019; Zhang et al., 2019). Cells have evolved a variety of strategies to deal with changing environments. In the face of brief stresses, it may be advantageous to store transcripts until the stress has passed, allowing for a faster restoration of growth, whereas prolonged stress may necessitate more drastic reprogramming of cellular processes (Arribere et al., 2011). Consequently, deciding whether a molecular species is or is not a part of the stress granule transcriptome/proteome, reducing the problem to a yes or no, may obscure more biology than it illuminates. Grand challenges in studying stress-induced protein/mRNA condensation As is now apparent, stress granules and their molecular precursors represent an exemplary system in which field-level challenges find crisp expression. Here we identify grand challenges in the study of these structures (Figure 5). The first central challenge is to identify the functions of stress-induced condensates and stress granules, and determine how these functions are executed. Of particular importance is the identification of fitness-related cellular phenotypes. The near-total reliance on molecular or imaging phenotypes, in the absence of function- and fitness-related phenotypes (growth, survival, differentiation, activity), has become tolerated in ways that may hinder progress. For example, given that canonical stress granules only become microscopically visible during severe stress in some important cases (Grousl et al., 2009; Wallace et al., 2015), the reliance on microscopic methods may blind us to wide swaths of functional phenomena. In addition, the identification of a cellular phenotype would make it possible to design genetic screens that search for factors that are not just involved in focus formation but are integral to stress granule function. Similarly, the use of inducers which robustly and reliably produce stress granules but are of uncertain physiological relevance, such as the broadly popular sodium arsenite, may have hidden disadvantages. If cells have not evolved to respond to a trigger, the cellular response is likely to lack organizational and molecular features which characterize responses to more physiological triggers such as heat, hypoxia, and osmotic shock. Even for these stresses, intensities which exceed physiological levels are in routine experimental use. Moreover, to validate a potent inducer such as sodium arsenite phenotypically against physiological inducers remains challenging until a phenotype or function of physiological stress granules is itself firmly established. Surmounting this central functional challenge will require sustained searches, a focus on physiology to match the extraordinary attention given to biophysics, and perhaps new thinking to identify a set of standardized phenotypes for functional studies. Surrounding this central challenge lurk many other intertwined grand challenges (Fig. 5). Some are well-established: determining the molecular bases of condensation and accumulation, and measuring molecular-scale condensation in living cells. Success on the latter would allow us, for the first time, to observe all the stages of stress-triggered condensation in vivo, even under mild stress conditions where large canonical stress granules do not form (Fig. 1). In attempting to discern the molecular determinants of condensation and stress granule formation, less discussed is the crucial difficulty—another grand challenge—of perturbing these phenomena cleanly, that is, without disrupting other activities. By analogy, study of an enzyme might involve, in order of decreasing disruption, a gene knockout, a temperature-sensitive mutation, a catalytic mutation, or development of a specific and reversible inhibitor. Despite considerable strides in this direction for stress granules (including screens for gene knockouts which disrupt stress granules (Yang et al., 2014)), at this moment the search for clean perturbations remains almost entirely open. In the absence of defined functions, another clear grand challenge looms: biochemical reconstitution of stress granule activities and functions. Reconstitution demonstrates the sufficiency of specific molecules and conditions to recapitulate cellular behavior. At present, all efforts have necessarily focused on reconstitution of traits without any unambiguous link to cellular fitness or adaptive function. Our situation in the stress granule field is remarkably different from historical efforts to purify specific biochemical fractions or molecules which could recapitulate an observed cellular activity. Finally, the evolutionary conservation of stress granules provides powerful motivation for their study. But how conserved are they? To what degree are the following conserved: specific components and stages; molecular determinants such as domains; biophysical forces; formation and dispersal pathways; regulators; and ultimate functions? Answering these questions would meet our final grand challenge (Fig. 5). Serious efforts to use evolutionary approaches, and to move beyond a handful of model organisms, has the potential to dramatically accelerate progress in our understanding of these enigmatic structures and processes. To the extent that stress granules are not merely reliable side-effects of some other biological process, consistent contributions to cellular and organismal fitness will be the decisive factors in their preservation across the tree of life. These grand challenges underscore that the field of stress granule biology is at a pivotal point. As we approach the 40-year mark since stress granules were first observed in tomato plants (Nover et al., 1983), we are due to move toward a deeper understanding of stress granules. Armed with clearly defined challenges, we can tackle the fundamental unknowns that still remain. Massive parallel surges in our understanding of composition and assembly mechanisms, both cell-biologically and biophysically, appear poised to drive a positive feedback loop of research integrating studies of assembly at multiple biological scales, mechanistic studies of the impact of condensation on mRNA lifecycles, and finally the fitness advantages that stress-induced condensation imparts. Acknowledgements We thank members of the Drummond and Sosnick labs for critical comments on the manuscript. Research reported in this publication was supported by the National Institutes of Health through the National Institute of General Medical Sciences (awards GM126547, GM127406, GM144278) and the National Institute of Environmental Health Sciences (award F30ES032665). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Figure 1: Stress-triggered protein/mRNA condensation and stress granule formation occur in stages, depend on stress intensity and identity, and involve multiple types of molecular interactions. Severe stress causes the accumulation of diverse small condensates into stress granules observable as cytosolic foci by standard microscopy. Figure 2: Formation of canonical stress granules (visible by standard microscopy, composed of a large number of components) may not be required for many attributed functions. Many roles could in principle be accomplished by small RNA/protein condensates consisting of a sharply restricted subset of components assembled into submicroscopic condensates. The diagram provides speculative positioning of functions on the size spectrum because strong hypotheses regarding which functions require large foci are lacking. Figure 3: The mechanisms of stress-triggered condensation and stress granule formation remain an area of active inquiry. Treatments which inhibit translation initiation (often by phosphorylation of eIF2ɑ), producing ribosome-free mRNA, cause stress granule formation in a wide range of systems and circumstances. Substantial recent work implicates long RNAs in condensation and formation of stress granules, a result which is biophysically plausible yet functionally puzzling. Figure 4: Different methods used to probe stress-induced condensation capture and report on different stages of stress-induced condensation and stress granule formation, providing complementary information. Figure 5: Grand challenges in the study of stress granules and stress-induced condensation. Box 1: What is a condensate? Biomolecular condensates are membraneless clusters of biomolecules such as proteins and nucleic acids. Classic examples are nucleoli, stress granules, P bodies, and germline P granules, among many others. “Biomolecular condensate” serves as an umbrella term for these structures which is agnostic as to their specific size, function, mechanism of formation, material state, or method of experimental study. The term arose, in part, due to the growing realization that more specific terms referring to mechanism (e.g., liquid-liquid phase separation [LLPS]), material state (e.g. droplet, hydrogel), or function (compartment, membraneless organelle) often implied more than is presently known. Importantly, many biomolecular condensates have been near-exclusively studied by specific methods. Stress granules, for example, are operationally defined by formation of foci resolvable by fluorescence microscopy that contain specific marker proteins and poly(A)+ RNA. Failure to detect microscopic foci is routinely taken to indicate the absence of stress granules, even though submicroscopic assemblies may be present. Rather than overturn this well-established operational definition, here we use the umbrella term condensates to refer to assemblies whether or not they are visible by microscopy. We use “accumulation” as a general term for processes in which smaller condensates are brought together to form larger structures. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Declaration of Interests The authors declare no competing interests. References Abraham R , Hauer D , McPherson RL , Utt A , Kirby IT , Cohen MS , Merits A , Leung AKL , and Griffin DE (2018). ADP-ribosyl-binding and hydrolase activities of the alphavirus nsP3 macrodomain are critical for initiation of virus replication. Proc. Natl. Acad. Sci. U. S. A 115 , E10457–E10466.30322911 Ackerman D , and Simon MC (2014). Hypoxia, lipids, and cancer: surviving the harsh tumor microenvironment. Trends Cell Biol. 24 , 472–478.24985940 Aizer A , Kalo A , Kafri P , Shraga A , Ben-Yishay R , Jacob A , Kinor N , and Shav-Tal Y (2014). Quantifying mRNA targeting to P bodies in living human cells reveals a dual role in mRNA decay and storage. Journal of Cell Science. Akhrymuk I , Frolov I , and Frolova EI (2018). Sindbis Virus Infection Causes Cell Death by nsP2-Induced Transcriptional Shutoff or by nsP3-Dependent Translational Shutoff. J. Virol 92 . Alberti S , and Carra S (2018). Quality Control of Membraneless Organelles. J. Mol. Biol 430 , 4711–4729.29758260 Alberti S , and Hyman AA (2021). Biomolecular condensates at the nexus of cellular stress, protein aggregation disease and ageing. Nat. Rev. Mol. Cell Biol 22 , 196–213.33510441 Anderson P , Kedersha N , and Ivanov P (2015). Stress granules, P-bodies and cancer. Biochim. Biophys. Acta 1849 , 861–870.25482014 Arava Y , Boas FE , Brown PO , and Herschlag D (2005). Dissecting eukaryotic translation and its control by ribosome density mapping. Nucleic Acids Res. 33 , 2421–2432.15860778 Arimoto K , Fukuda H , Imajoh-Ohmi S , Saito H , and Takekawa M (2008). Formation of stress granules inhibits apoptosis by suppressing stress-responsive MAPK pathways. Nat. Cell Biol 10 , 1324–1332.18836437 Aronov S , Dover-Biterman S , Suss-Toby E , Shmoish M , Duek L , and Choder M (2015). Pheromone-encoding mRNA is transported to the yeast mating projection by specific RNP granules. J. Cell Biol 209 , 829–842.26101218 Arribere JA , Doudna JA , and Gilbert WV (2011). Reconsidering Movement of Eukaryotic mRNAs between Polysomes and P Bodies. Mol. Cell 44 , 745–758.22152478 Aulas A , Fay MM , Lyons SM , Achorn CA , Kedersha N , Anderson P , and Ivanov P (2017). Stress-specific differences in assembly and composition of stress granules and related foci. J. Cell Sci 130 , 927–937.28096475 Banani SF , Lee HO , Hyman AA , and Rosen MK (2017). Biomolecular condensates: organizers of cellular biochemistry. Nat. Rev. Mol. Cell Biol Begovich K , and Wilhelm JE (2020). An In Vitro Assembly System Identifies Roles for RNA Nucleation and ATP in Yeast Stress Granule Formation. Mol. Cell Biancon G , Joshi P , Zimmer JT , Hunck T , Gao Y , Lessard MD , Courchaine E , Barentine AES , Machyna M , Botti V , (2022). Precision analysis of mutant U2AF1 activity reveals deployment of stress granules in myeloid malignancies. Mol. Cell 82 , 1107–1122.e7.35303483 Bley N , Lederer M , Pfalz B , Reinke C , Fuchs T , Glaß M , Möller B , and Hüttelmaier S (2015). Stress granules are dispensable for mRNA stabilization during cellular stress. Nucleic Acids Res. 43 , e26.25488811 Boeynaems S , Alberti S , Fawzi NL , Mittag T , Polymenidou M , Rousseau F , Schymkowitz J , Shorter J , Wolozin B , Van Den Bosch L , (2018). Protein Phase Separation: A New Phase in Cell Biology. Trends Cell Biol. 28 , 420–435.29602697 Bosco DA , Lemay N , Ko HK , Zhou H , Burke C , Kwiatkowski TJ Jr , Sapp P , McKenna-Yasek D , Brown RH Jr , and Hayward LJ (2010). Mutant FUS proteins that cause amyotrophic lateral sclerosis incorporate into stress granules. Hum. Mol. Genet 19 , 4160–4175.20699327 Bounedjah O , Desforges B , Wu T-D , Pioche-Durieu C , Marco S , Hamon L , Curmi PA , Guerquin-Kern J-L , Piétrement O , and Pastré D (2014). Free mRNA in excess upon polysome dissociation is a scaffold for protein multimerization to form stress granules. Nucleic Acids Res. 42 , 8678–8691.25013173 Brangwynne CP (2013). Phase transitions and size scaling of membrane-less organelles. J. Cell Biol 203 , 875–881.24368804 Buchan JR , and Parker R (2009). Eukaryotic stress granules: the ins and outs of translation. Mol. Cell 36 , 932–941.20064460 Buchan JR , Muhlrad D , and Parker R (2008). P bodies promote stress granule assembly in Saccharomyces cerevisiae. J. Cell Biol 183 , 441–455.18981231 Buchan JR , Yoon J-H , and Parker R (2011). Stress-specific composition, assembly and kinetics of stress granules in Saccharomyces cerevisiae. J. Cell Sci 124 , 228–239.21172806 Buchan JR , Kolaitis R-M , Taylor JP , and Parker R (2013). Eukaryotic stress granules are cleared by autophagy and Cdc48/VCP function. Cell 153 , 1461–1474.23791177 Burgess HM , and Mohr I (2018). Defining the Role of Stress Granules in Innate Immune Suppression by the Herpes Simplex Virus 1 Endoribonuclease VHS. J. Virol 92 . Campos-Melo D , Hawley ZCE , Droppelmann CA , and Strong MJ (2021). The Integral Role of RNA in Stress Granule Formation and Function. Front Cell Dev Biol 9 , 621779.34095105 Cherkasov V , Hofmann S , Druffel-Augustin S , Mogk A , Tyedmers J , Stoecklin G , and Bukau B (2013). Coordination of translational control and protein homeostasis during severe heat stress. Curr. Biol 23 , 2452–2462.24291094 Cherkasov V , Grousl T , Theer P , Vainshtein Y , Gläßer C , Mongis C , Kramer G , Stoecklin G , Knop M , Mogk A , (2015). Systemic control of protein synthesis through sequestration of translation and ribosome biogenesis factors during severe heat stress. FEBS Lett. Clarke HJ , Chambers JE , Liniker E , and Marciniak SJ (2014). Endoplasmic reticulum stress in malignancy. Cancer Cell 25 , 563–573.24823636 Dang Y , Kedersha N , Low W-K , Romo D , Gorospe M , Kaufman R , Anderson P , and Liu JO (2006). Eukaryotic initiation factor 2alpha-independent pathway of stress granule induction by the natural product pateamine A. J. Biol. Chem 281 , 32870–32878.16951406 Decker CJ , and Parker R (2012). P-bodies and stress granules: possible roles in the control of translation and mRNA degradation. Cold Spring Harb. Perspect. Biol 4 , a012286.22763747 Duret L , and Mouchiroud D (1999). Expression pattern and, surprisingly, gene length shape codon usage in Caenorhabditis, Drosophila, and Arabidopsis. Proc. Natl. Acad. Sci. U. S. A 96 , 4482–4487.10200288 Eiermann N , Haneke K , Sun Z , Stoecklin G , and Ruggieri A (2020). Dance with the Devil: Stress Granules and Signaling in Antiviral Responses. Viruses 12 . Escalante LE , and Gasch AP (2021). The role of stress-activated RNA-protein granules in surviving adversity. RNA. Farny NG , Kedersha NL , and Silver PA (2009). Metazoan stress granule assembly is mediated by P-eIF2alpha-dependent and -independent mechanisms. RNA 15 , 1814–1821.19661161 Fournier M-J , Gareau C , and Mazroui R (2010). The chemotherapeutic agent bortezomib induces the formation of stress granules. Cancer Cell Int. 10 , 12.20429927 Franks TM , and Lykke-Andersen J (2007). TTP and BRF proteins nucleate processing body formation to silence mRNAs with AU-rich elements. Genes Dev. 21 , 719–735.17369404 Freibaum BD , Messing J , Yang P , Kim HJ , and Taylor JP (2021). High-fidelity reconstitution of stress granules and nucleoli in mammalian cellular lysate. J. Cell Biol 220 . Fujimura K , Sasaki AT , and Anderson P (2012). Selenite targets eIF4E-binding protein-1 to inhibit translation initiation and induce the assembly of non-canonical stress granules. Nucleic Acids Res. 40 , 8099–8110.22718973 Gal J , Kuang L , Barnett KR , Zhu BZ , Shissler SC , Korotkov KV , Hayward LJ , Kasarskis EJ , and Zhu H (2016). ALS mutant SOD1 interacts with G3BP1 and affects stress granule dynamics. Acta Neuropathol. 132 , 563–576.27481264 Gao X , Jiang L , Gong Y , Chen X , Ying M , Zhu H , He Q , Yang B , and Cao J (2019). Stress granule: A promising target for cancer treatment. Br. J. Pharmacol 176 , 4421–4433.31301065 Gomes E , and Shorter J (2019). The molecular language of membraneless organelles. J. Biol. Chem 294 , 7115–7127.30045872 Gorrini C , Harris IS , and Mak TW (2013). Modulation of oxidative stress as an anticancer strategy. Nat. Rev. Drug Discov 12 , 931–947.24287781 Grousl T , Ivanov P , Frydlova I , Vasicova P , Janda F , Vojtova J , Malinska K , Malcova I , Novakova L , Janoskova D , (2009). Robust heat shock induces eIF2alphaphosphorylation-independent assembly of stress granules containing eIF3 and 40S ribosomal subunits in budding yeast, Saccharomyces cerevisiae. J. Cell Sci 122 , 2078–2088.19470581 Grousl T , Ivanov P , Malcova I , Pompach P , Frydlova I , Slaba R , Senohrabkova L , Novakova L , and Hasek J (2013). Heat shock-induced accumulation of translation elongation and termination factors precedes assembly of stress granules in S. cerevisiae. PLoS One 8 , e57083.23451152 Guillén-Boixet J , Kopach A , Holehouse AS , Wittmann S , Jahnel M , Schlüßler R , Kim K , Trussina IREA , Wang J , Mateju D , (2020). RNA-Induced Conformational Switching and Clustering of G3BP Drive Stress Granule Assembly by Condensation. Cell 181 , 346–361.e17.32302572 Guo L , and Shorter J (2015). It’s Raining Liquids: RNA Tunes Viscoelasticity and Dynamics of Membraneless Organelles. Mol. Cell 60 , 189–192.26474062 Guzikowski AR , Chen YS , and Zid BM (2019). Stress-induced mRNP granules: Form and function of processing bodies and stress granules. Wiley Interdiscip. Rev. RNA 10 , e1524.30793528 Gwon Y , Maxwell BA , Kolaitis R-M , Zhang P , Kim HJ , and Taylor JP (2021). Ubiquitination of G3BP1 mediates stress granule disassembly in a context-specific manner. Science 372 , eabf6548.34739333 Hofmann S , Kedersha N , Anderson P , and Ivanov P (2020). Molecular mechanisms of stress granule assembly and disassembly. Biochim. Biophys. Acta Mol. Cell Res 1868 , 118876.33007331 Hubstenberger A , Courel M , Bénard M , Souquere S , Ernoult-Lange M , Chouaib R , Yi Z , Morlot J-B , Munier A , Fradet M , (2017). P-Body Purification Reveals the Condensation of Repressed mRNA Regulons. Mol. Cell 68 , 144–157.e5.28965817 Hyman AA , Weber CA , and Jülicher F (2014). Liquid-liquid phase separation in biology. Annu. Rev. Cell Dev. Biol 30 , 39–58.25288112 Iserman C , Altamirano CD , Jegers C , Friedrich U , Zarin T , Fritsch AW , Mittasch M , Domingues A , Hersemann L , Jahnel M , (2020). Condensation of Ded1p Promotes a Translational Switch from Housekeeping to Stress Protein Production. Cell 0 . Ivanov P , Kedersha N , and Anderson P (2019). Stress Granules and Processing Bodies in Translational Control. Cold Spring Harb. Perspect. Biol 11 . Ivanov PA , Chudinova EM , and Nadezhdina ES (2003a). Disruption of microtubules inhibits cytoplasmic ribonucleoprotein stress granule formation. Exp. Cell Res 290 , 227–233.14567982 Ivanov PA , Chudinova EM , and Nadezhdina ES (2003b). RNP stress-granule formation is inhibited by microtubule disruption. Cell Biol. Int 27 , 207–208.12681309 Jain A , and Vale RD (2017). RNA phase transitions in repeat expansion disorders. Nature 546 , 243–247.28562589 Jain S , Wheeler JR , Walters RW , Agrawal A , Barsic A , and Parker R (2016). ATPase-Modulated Stress Granules Contain a Diverse Proteome and Substructure. Cell 164 , 487–498.26777405 Jayabalan AK , Adivarahan S , Koppula A , Abraham R , Batish M , Zenklusen D , Griffin DE , and Leung AKL (2021). Stress granule formation, disassembly, and composition are regulated by alphavirus ADP-ribosylhydrolase activity. Proc. Natl. Acad. Sci. U. S. A 118 . Kaehler C , Isensee J , Hucho T , Lehrach H , and Krobitsch S (2014). 5-Fluorouracil affects assembly of stress granules based on RNA incorporation. Nucleic Acids Res. 42 , 6436–6447.24728989 Katoh H , Okamoto T , Fukuhara T , Kambara H , Morita E , Mori Y , Kamitani W , and Matsuura Y (2013). Japanese encephalitis virus core protein inhibits stress granule formation through an interaction with Caprin-1 and facilitates viral propagation. J. Virol 87 , 489–502.23097442 Kedersha N , and Anderson P (2002). Stress granules: sites of mRNA triage that regulate mRNA stability and translatability. Biochem. Soc. Trans 30 , 963–969.12440955 Kedersha N , and Anderson P (2009). Regulation of translation by stress granules and processing bodies. Prog. Mol. Biol. Transl. Sci 90 , 155–185.20374741 Kedersha N , Cho MR , Li W , Yacono PW , Chen S , Gilks N , Golan DE , and Anderson P (2000). Dynamic shuttling of TIA-1 accompanies the recruitment of mRNA to mammalian stress granules. J. Cell Biol 151 , 1257–1268.11121440 Kedersha N , Chen S , Gilks N , Li W , Miller IJ , Stahl J , and Anderson P (2002). Evidence that ternary complex (eIF2-GTP-tRNA(i)(Met))-deficient preinitiation complexes are core constituents of mammalian stress granules. Mol. Biol. Cell 13 , 195–210.11809833 Kedersha N , Ivanov P , and Anderson P (2013). Stress granules and cell signaling: more than just a passing phase? Trends Biochem. Sci 38 , 494–506.24029419 Kedersha N , Panas MD , Achorn CA , Lyons S , Tisdale S , Hickman T , Thomas M , Lieberman J , McInerney GM , Ivanov P , (2016). G3BP-Caprin1-USP10 complexes mediate stress granule condensation and associate with 40S subunits. J. Cell Biol 212 , 845–860.27022092 Kedersha NL , Gupta M , Li W , Miller I , and Anderson P (1999). RNA-binding proteins TIA-1 and TIAR link the phosphorylation of eIF-2 alpha to the assembly of mammalian stress granules. J. Cell Biol 147 , 1431–1442.10613902 Keiten-Schmitz J , Wagner K , Piller T , Kaulich M , Alberti S , and Müller S (2020). The Nuclear SUMO-Targeted Ubiquitin Quality Control Network Regulates the Dynamics of Cytoplasmic Stress Granules. Mol. Cell 79 , 54–67.e7.32521226 Khong A , Kerr CH , Yeung CHL , Keatings K , Nayak A , Allan DW , and Jan E (2017a). Disruption of Stress Granule Formation by the Multifunctional Cricket Paralysis Virus 1A Protein. J. Virol 91 . Khong A , Matheny T , Jain S , Mitchell SF , Wheeler JR , and Parker R (2017b). The Stress Granule Transcriptome Reveals Principles of mRNA Accumulation in Stress Granules. Mol. Cell 68 , 808–820.e5.29129640 Kiebler MA , and Bassell GJ (2006). Neuronal RNA granules: movers and makers. Neuron 51 , 685–690.16982415 Kramer S , Queiroz R , Ellis L , Webb H , Hoheisel JD , Clayton C , and Carrington M (2008). Heat shock causes a decrease in polysomes and the appearance of stress granules in trypanosomes independently of eIF2(alpha) phosphorylation at Thr169. J. Cell Sci 121 , 3002–3014.18713834 Kroschwald S , Maharana S , Mateju D , Malinovska L , Nüske E , Poser I , Richter D , and Alberti S (2015). Promiscuous interactions and protein disaggregases determine the material state of stress-inducible RNP granules. Elife 4 , e06807.26238190 Kroschwald S , Munder MC , Maharana S , Franzmann TM , Richter D , Ruer M , Hyman AA , and Alberti S (2018). Different Material States of Pub1 Condensates Define Distinct Modes of Stress Adaptation and Recovery. Cell Rep. 23 , 3327–3339.29898402 Law LMJ , Razooky BS , Li MMH , You S , Jurado A , Rice CM , and MacDonald MR (2019). ZAP’s stress granule localization is correlated with its antiviral activity and induced by virus replication. PLoS Pathog. 15 , e1007798.31116799 Liao Y-C , Fernandopulle MS , Wang G , Choi H , Hao L , Drerup CM , Patel R , Qamar S , Nixon-Abell J , Shen Y , (2019). RNA Granules Hitchhike on Lysosomes for Long-Distance Transport, Using Annexin A11 as a Molecular Tether. Cell 179 , 147–164.e20.31539493 Lopes I , Altab G , Raina P , and de Magalhães JP (2021). Gene Size Matters: An Analysis of Gene Length in the Human Genome. Front. Genet 12 , 559998.33643374 Low W-K , Dang Y , Schneider-Poetsch T , Shi Z , Choi NS , Merrick WC , Romo D , and Liu JO (2005). Inhibition of eukaryotic translation initiation by the marine natural product pateamine A. Mol. Cell 20 , 709–722.16337595 Lyon AS , Peeples WB , and Rosen MK (2020). A framework for understanding the functions of biomolecular condensates across scales. Nat. Rev. Mol. Cell Biol Mackenzie IR , Nicholson AM , Sarkar M , Messing J , Purice MD , Pottier C , Annu K , Baker M , Perkerson RB , Kurti A , (2017). TIA1 Mutations in Amyotrophic Lateral Sclerosis and Frontotemporal Dementia Promote Phase Separation and Alter Stress Granule Dynamics. Neuron 95 , 808–816.e9.28817800 Mahboubi H , and Stochaj U (2017). Cytoplasmic stress granules: Dynamic modulators of cell signaling and disease. Biochim. Biophys. Acta Mol. Basis Dis 1863 , 884–895.28095315 Markmiller S , Soltanieh S , Server KL , Mak R , Jin W , Fang MY , Luo E-C , Krach F , Yang D , Sen A , (2018). Context-Dependent and Disease-Specific Diversity in Protein Interactions within Stress Granules. Cell 172 , 590–604.e13.29373831 Marmor-Kollet H , Siany A , Kedersha N , Knafo N , Rivkin N , Danino YM , Moens TG , Olender T , Sheban D , Cohen N , (2020). Spatiotemporal Proteomic Analysis of Stress Granule Disassembly Using APEX Reveals Regulation by SUMOylation and Links to ALS Pathogenesis. Mol. Cell 80 , 876–891.e6.33217318 Mateju D , and Chao JA (2021). Stress granules: Regulators or by-products? FEBS J. Mateju D , Eichenberger B , Voigt F , Eglinger J , Roth G , and Chao JA (2020). Single-Molecule Imaging Reveals Translation of mRNAs Localized to Stress Granules. Cell 183 , 1801–1812.e13.33308477 Matheny T , Rao BS , and Parker R (2019). Transcriptome-Wide Comparison of Stress Granules and P-Bodies Reveals that Translation Plays a Major Role in RNA Partitioning. Mol. Cell. Biol 39 . Matheny T , Van Treeck B , Huynh TN , and Parker R (2021). RNA partitioning into stress granules is based on the summation of multiple interactions. RNA 27 , 174–189.33199441 Maxwell BA , Gwon Y , Mishra A , Peng J , Nakamura H , Zhang K , Kim HJ , and Taylor JP (2021). Ubiquitination is essential for recovery of cellular activities after heat shock. Science 372 , eabc3593.34739326 Mazroui R , Sukarieh R , Bordeleau M-E , Kaufman RJ , Northcote P , Tanaka J , Gallouzi I , and Pelletier J (2006). Inhibition of ribosome recruitment induces stress granule formation independently of eukaryotic initiation factor 2alpha phosphorylation. Mol. Biol. Cell 17 , 4212–4219.16870703 Mitrea DM , and Kriwacki RW (2016). Phase separation in biology; functional organization of a higher order. Cell Commun. Signal 14 , 1.26727894 Molliex A , Temirov J , Lee J , Coughlin M , Kanagaraj AP , Kim HJ , Mittag T , and Taylor JP (2015). Phase Separation by Low Complexity Domains Promotes Stress Granule Assembly and Drives Pathological Fibrillization. Cell 163 , 123–133.26406374 Moon SL , Morisaki T , Khong A , Lyon K , Parker R , and Stasevich TJ (2019). Multicolour single-molecule tracking of mRNA interactions with RNP granules. Nat. Cell Biol 21 , 162–168.30664789 Moon SL , Morisaki T , Stasevich TJ , and Parker R (2020). Coupling of translation quality control and mRNA targeting to stress granules. J. Cell Biol 219 . Nadezhdina ES , Lomakin AJ , Shpilman AA , Chudinova EM , and Ivanov PA (2010). Microtubules govern stress granule mobility and dynamics. Biochim. Biophys. Acta 1803 , 361–371.20036288 Namkoong S , Ho A , Woo YM , Kwak H , and Lee JH (2018). Systematic Characterization of Stress-Induced RNA Granulation. Mol. Cell 70 , 175–187.e8.29576526 Nover L , Scharf KD , and Neumann D (1983). Formation of cytoplasmic heat shock granules in tomato cell cultures and leaves. Mol. Cell. Biol 3 , 1648–1655.6633535 Nover L , Scharf KD , and Neumann D (1989). Cytoplasmic heat shock granules are formed from precursor particles and are associated with a specific set of mRNAs. Mol. Cell. Biol 9 , 1298–1308.2725500 Padrón A , Iwasaki S , and Ingolia NT (2019). Proximity RNA Labeling by APEX-Seq Reveals the Organization of Translation Initiation Complexes and Repressive RNA Granules. Mol. Cell 75 , 875–887.e5.31442426 Panas MD , Ivanov P , and Anderson P (2016). Mechanistic insights into mammalian stress granule dynamics. J. Cell Biol 215 , 313–323.27821493 Patel A , Lee HO , Jawerth L , Maharana S , Jahnel M , Hein MY , Stoynov S , Mahamid J , Saha S , Franzmann TM , (2015). A Liquid-to-Solid Phase Transition of the ALS Protein FUS Accelerated by Disease Mutation. Cell 162 , 1066–1077.26317470 Peng S , Li W , Yao Y , Xing W , Li P , and Chen C (2020). Phase separation at the nanoscale quantified by dcFCCS. Proc. Natl. Acad. Sci. U. S. A 117 , 27124–27131.33087563 Preiss T , Baron-Benhamou J , Ansorge W , and Hentze MW (2003). Homodirectional changes in transcriptome composition and mRNA translation induced by rapamycin and heat shock. Nat. Struct. Biol 10 , 1039–1047.14608375 Pushpalatha KV , and Besse F (2019). Local Translation in Axons: When Membraneless RNP Granules Meet Membrane-Bound Organelles. Front Mol Biosci 6 , 129.31824961 Reineke LC , and Neilson JR (2019). Differences between acute and chronic stress granules, and how these differences may impact function in human disease. Biochem. Pharmacol 162 , 123–131.30326201 Riback JA , Katanski CD , Kear-Scott JL , Pilipenko EV , Rojek AE , Sosnick TR , and Drummond DA (2017). Stress-Triggered Phase Separation Is an Adaptive, Evolutionarily Tuned Response. Cell 168 , 1028–1040.e19.28283059 Riback JA , Eeftens JM , Lee DSW , Quinodoz SA , Beckers L , Becker LA , and Brangwynne CP (2022). Viscoelastic RNA entanglement and advective flow underlie nucleolar form and function. Ripin N , and Parker R (2021). Are stress granules the RNA analogs of misfolded protein aggregates? RNA. Sahoo PK , Lee SJ , Jaiswal PB , Alber S , Kar AN , Miller-Randolph S , Taylor EE , Smith T , Singh B , Ho TS-Y , (2018). Axonal G3BP1 stress granule protein limits axonal mRNA translation and nerve regeneration. Nat. Commun 9 , 3358.30135423 Sahoo PK , Kar AN , Samra N , Terenzio M , Patel P , Lee SJ , Miller S , Thames E , Jones B , Kawaguchi R , (2020). A Ca2+-Dependent Switch Activates Axonal Casein Kinase 2α Translation and Drives G3BP1 Granule Disassembly for Axon Regeneration. Curr. Biol 30 , 4882–4895.e6.33065005 Samir P , Kesavardhana S , Patmore DM , Gingras S , Malireddi RKS , Karki R , Guy CS , Briard B , Place DE , Bhattacharya A , (2019). DDX3X acts as a live-or-die checkpoint in stressed cells by regulating NLRP3 inflammasome. Nature. Sanchez de Groot N , Armaos A , Graña-Montes R , Alriquet M , Calloni G , Vabulas RM , and Tartaglia GG (2019). RNA structure drives interaction with proteins. Nat. Commun 10 , 3246.31324771 Sanders DW , Kedersha N , Lee DSW , Strom AR , Drake V , Riback JA , Bracha D , Eeftens JM , Iwanicki A , Wang A , (2020). Competing Protein-RNA Interaction Networks Control Multiphase Intracellular Organization. Cell 181 , 306–324.e28.32302570 Shattuck JE , Paul KR , Cascarina SM , and Ross ED (2019). The prion-like protein kinase Sky1 is required for efficient stress granule disassembly. Nat. Commun 10 , 3614.31399582 Sheth U , and Parker R (2003). Decapping and Decay of Messenger RNA Occur in Cytoplasmic Processing Bodies. Science. Shiina N (2019). Liquid- and solid-like RNA granules form through specific scaffold proteins and combine into biphasic granules. J. Biol. Chem 294 , 3532–3548.30606735 Somasekharan SP , Zhang F , Saxena N , Huang JN , Kuo I-C , Low C , Bell R , Adomat H , Stoynov N , Foster L , (2020). G3BP1-linked mRNA partitioning supports selective protein synthesis in response to oxidative stress. Nucleic Acids Res. 48 , 6855–6873.32406909 Sorenson R , and Bailey-Serres J (2014). Selective mRNA sequestration by OLIGOURIDYLATE-BINDING PROTEIN 1 contributes to translational control during hypoxia in Arabidopsis. Proc. Natl. Acad. Sci. U. S. A 111 , 2373–2378.24469793 Stöhr N , Lederer M , Reinke C , Meyer S , Hatzfeld M , Singer RH , and Hüttelmaier S (2006). ZBP1 regulates mRNA stability during cellular stress. J. Cell Biol 175 , 527–534.17101699 Tauber D , Tauber G , Khong A , Van Treeck B , Pelletier J , and Parker R (2020a). Modulation of RNA Condensation by the DEAD-Box Protein eIF4A. Cell 0 . Tauber D , Tauber G , and Parker R (2020b). Mechanisms and Regulation of RNA Condensation in RNP Granule Formation. Trends Biochem. Sci Tsai N-P , and Wei L-N (2010). RhoA/ROCK1 signaling regulates stress granule formation and apoptosis. Cell. Signal 22 , 668–675.20004716 Van Treeck B , and Parker R (2018). Emerging Roles for Intermolecular RNA-RNA Interactions in RNP Assemblies. Cell 174 , 791–802.30096311 Van Treeck B , Protter DSW , Matheny T , Khong A , Link CD , and Parker R (2018). RNA self-assembly contributes to stress granule formation and defining the stress granule transcriptome. Proc. Natl. Acad. Sci. U. S. A 201800038. Wallace EWJ , Kear-Scott JL , Pilipenko EV , Schwartz MH , Laskowski PR , Rojek AE , Katanski CD , Riback JA , Dion MF , Franks AM , (2015). Reversible, Specific, Active Aggregates of Endogenous Proteins Assemble upon Heat Stress. Cell 162 , 1286–1298.26359986 Weinberg DE , Shah P , Eichhorn SW , Hussmann JA , Plotkin JB , and Bartel DP (2016). Improved Ribosome-Footprint and mRNA Measurements Provide Insights into Dynamics and Regulation of Yeast Translation. Cell Rep. 14 , 1787–1799.26876183 Wheeler JR , Matheny T , Jain S , Abrisch R , and Parker R (2016). Distinct stages in stress granule assembly and disassembly. Elife 5 . White JP , Cardenas AM , Marissen WE , and Lloyd RE (2007). Inhibition of cytoplasmic mRNA stress granule formation by a viral proteinase. Cell Host Microbe 2 , 295–305.18005751 Wilbertz JH , Voigt F , Horvathova I , Roth G , Zhan Y , and Chao JA (2019). Single-Molecule Imaging of mRNA Localization and Regulation during the Integrated Stress Response. Mol. Cell 73 , 946–958.e7.30661979 Yang P , Mathieu C , Kolaitis R-M , Zhang P , Messing J , Yurtsever U , Yang Z , Wu J , Li Y , Pan Q , (2020). G3BP1 Is a Tunable Switch that Triggers Phase Separation to Assemble Stress Granules. Cell 181 , 325–345.e28.32302571 Yang X , Shen Y , Garre E , Hao X , Krumlinde D , Cvijović M , Arens C , Nyström T , Liu B , and Sunnerhagen P (2014). Stress Granule-Defective Mutants Deregulate Stress Responsive Transcripts. PLoS Genet. 10 , e1004763.25375155 Yoo H , Triandafillou C , and Drummond DA (2019). Cellular sensing by phase separation: Using the process, not just the products. J. Biol. Chem 294 , 7151–7159.30877200 Yoo H , Bard JAM , Pilipenko E , and Allan Drummond D (2021). Chaperones directly and efficiently disperse stress-triggered biomolecular condensates. Youn J-Y , Dunham WH , Hong SJ , Knight JDR , Bashkurov M , Chen GI , Bagci H , Rathod B , MacLeod G , Eng SWM , (2018). High-Density Proximity Mapping Reveals the Subcellular Organization of mRNA-Associated Granules and Bodies. Mol. Cell 69 , 517–532.e11.29395067 Zhang P , Fan B , Yang P , Temirov J , Messing J , Kim HJ , and Taylor JP (2019). Chronic optogenetic induction of stress granules is cytotoxic and reveals the evolution of ALS-FTD pathology. Elife 8 . Zid BM , and O’Shea EK (2014). Promoter sequences direct cytoplasmic localization and translation of mRNAs during starvation in yeast. Nature 514 , 117–121.25119046

PMC009xxxxxx/PMC9308753.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101676030 44856 Cell Chem Biol Cell Chem Biol Cell chemical biology 2451-9456 2451-9448 35654040 9308753 10.1016/j.chembiol.2022.05.004 NIHMS1812992 Article Visualizing Staphylococcus aureus pathogenic membrane modification within the host infection environment by multimodal imaging mass spectrometry Perry William J. 123 Grunenwald Caroline M. 34 Van de Plas Raf 156 Witten James C. 7 Martin Daniel R. 8 Apte Suneel S. 8 Cassat James E. 34910 Pettersson Gösta B. 7 Caprioli Richard M. 1261112 Skaar Eric P. 34* Spraggins Jeffrey M. 12613*‡ 1 Mass Spectrometry Research Center, Vanderbilt University, Nashville, TN, 37232, USA 2 Department of Chemistry, Vanderbilt University, Nashville, TN, 37212, USA 3 Vanderbilt Institute for Infection, Immunology, and Inflammation, Vanderbilt University, Nashville, TN, 37232, USA 4 Department of Pathology, Microbiology, and Immunology, Vanderbilt University Medical Center, Nashville, TN, 37212, USA 5 Delft Center for Systems and Control, Delft University of Technology - TU Delft, Delft, The Netherlands 6 Department of Biochemistry, Vanderbilt University, Nashville, TN, 37212, USA 7 Department of Thoracic and Cardiovascular Surgery, Cleveland Clinic Heart and Vascular Institute, Cleveland, OH, 44195, USA 8 Department of Biomedical Engineering, Cleveland Clinic Lerner Research Institute, Cleveland, OH, 44195, USA 9 Department of Pediatrics, Division of Pediatric Infectious Diseases, Vanderbilt University Medical Center, Nashville, TN, 37232, USA 10 Department of Biomedical Engineering, Vanderbilt University Medical Center, Nashville, TN, 37232, USA 11 Department of Pharmacology, Vanderbilt University, Nashville, TN, 37212, USA 12 Department of Medicine, Vanderbilt University, Nashville, TN, 37212, USA 13 Department of Cell & Developmental Biology, Vanderbilt University, Nashville, TN, 37232, USA Author Contributions: W.J.P., J.M.S., R.M.C., and E.P.S. designed experiments. C.M.G. and J.E.C. performed animal infection models. W.J.P. completed Bright-field microscopy, fluorescence microscopy, and LC-MS/MS lipidomics experiments. W.J.P. and J.M.S. completed MALDI IMS experiments. R.V.P. performed and analyzed the data-driven image fusion results. J.C.W. and G.B.P. provided the human tissue. D.R.M. and S.S.A. contributed LC-MS/MS proteomics data. W.J.P., J.M.S., and E.P.S. completed the original manuscript draft. All authors contributed to critical review and editing of the manuscript. ‡ Lead Contact: Jeffrey M. Spraggins (jeff.spraggins@vanderbilt.edu) * Corresponding Authors: Jeffrey M. Spraggins (jeff.spraggins@vanderbilt.edu) and Eric P. Skaar (eric.skaar@vumc.org) 12 6 2022 21 7 2022 01 6 2022 21 7 2023 29 7 12091217.e4 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Summary: Bacterial pathogens have evolved virulence factors to colonize, replicate, and disseminate within the vertebrate host. Although there is an expanding body of literature describing how bacterial pathogens regulate their virulence repertoire in response to environmental signals, it is challenging to directly visualize virulence response within the host tissue microenvironment. Multimodal imaging approaches enable visualization of host-pathogen molecular interactions. Here we demonstrate multimodal integration of high spatial resolution imaging mass spectrometry and microscopy to visualize Staphylococcus aureus envelope modifications within infected murine and human tissues. Data-driven image fusion of fluorescent bacterial reporters and MALDI FT-ICR IMS uncovered S. aureus lysyl-phosphatidylglycerol lipids, localizing to select bacterial communities within infected tissue. Absence of lysyl-phosphatidylglycerols is associated with decreased pathogenicity during vertebrate colonization as these lipids provide protection against the innate immune system. The presence of distinct staphylococcal lysyl-phosphatidylglycerol distributions within murine and human infections suggests a heterogeneous, spatially oriented microbial response to host defenses. Graphical Abstract eTOC Blurb: Perry et al. combine imaging mass spectrometry (IMS) and microscopy to reveal Staphylococcus aureus envelope modifications within infected murine and human tissues. Their technique provides bacterial lipid profiles that highlight varied lysyl-phosphatidylglycerol distributions suggesting a heterogeneous, spatially oriented microbial response to host defenses in situ. pmcIntroduction: The increased incidence of antimicrobial-resistance among bacterial pathogens has emphasized the need to discover critical pathogenic adaptations that can be targeted to develop alternative or adjunctive therapeutics (Beceiro, Tomás, and Bou, 2013). Staphylococcus aureus is an increasingly antibiotic resistant human pathogen, capable of causing a variety of life-threatening illnesses that range from soft tissue infections to serious systemic infections (Casadevall and Pirofski, 2000). Antimicrobials used to combat bacterial infections commonly function by disrupting critical microbial processes such as protein synthesis, energy production, and envelope stability and biogenesis (Kohanski, Dwyer, and Collins, 2010). Cationic compounds, such as daptomycin, are one class of antimicrobial compounds that target envelope stability and are frequently used to combat Gram-positive bacterial infections. Moreover, host-generated cationic antimicrobial peptides (CAMPs) also bind or insert into bacterial membranes to compromise structural integrity (Muthaiyan, Silverman, Jayaswal, and Wilkinson, 2008). In response to these compounds, bacterial pathogens can aminoacylate anionic phosphatidylglycerol (PG) membrane lipids, thereby reducing electrostatic interactions as a mechanism to evade cationic compounds (Slavetinsky, Kuhn, and Peschel, 2017). S. aureus, in particular, is known to modify PG membrane lipids with lysine residues (lysyl-PG) through the action of Multiple peptide resistance Factor (MprF), providing resistance to cationic compounds (Peschel et al., 2001). Inflammatory lesions or abscesses within soft tissue are hallmarks of S. aureus infection. Abscess morphology consists of Staphylococcal Abscess Communities (SACs) segregated from normal host tissue by layers of necrotic and healthy polymorphonuclear neutrophil (PMN) innate immune cells (Cheng, DeDent, Schneewind, and Missiakas, 2011). Recruited PMNs phagocytose bacteria and release numerous antimicrobial effectors including reactive oxygen and nitrogen species, antimicrobial peptides, and digestive enzymes (Kobayashi, Malachowa, and DeLeo, 2015). Staphylococcal dissemination within a vertebrate host relies on a diverse array of virulence factors (Cheng, Kim, Burts, Krausz, Schneewind, and Missiakas, 2009). Many of these virulence factors function in evasion or resistance against immune-mediated killing. One such example is the production of lysyl-PGs by MprF and the subsequent incorporation of these modified PGs into the staphylococcal envelope. Comprised of synthase and flippase domains, MprF facilitates the transfer of lysine from lysyl-transfer RNA to phosphatidylglycerol lipids to then be translocated to the outer membrane leaflet (Ernst, Staubitz, Mishra, Yang, Hornig, Kalbacher, Bayer, Kraus, and Peschel, 2009). The decreased net negative charge of the altered membrane decreases electrostatic attraction from positively charged antimicrobials providing resistance against these compounds (Ernst and Peschel, 2019). Absence of lysyl-PG production results in increased staphylococcal killing by human PMNs and attenuated virulence in a murine model of systemic infection (Peschel et al., 2001). Genetic modifications to MprF resulting in increased activity of the synthase or flippase domains have been identified that result in increased lysyl-PG abundance. Elevated lysyl-PG production results in altered resistance and susceptibility to daptomycin (Ernst and Peschel, 2019; Baltz, 2009; Mishra and Bayer, 2013). Emerging literature suggests that the staphylococcal abscess is a molecularly heterogenous environment and therefore SACs elaborate differential gene expression across distinct abscesses and even within a single abscess (Cassat et al., 2018; Perry, Spraggins, Sheldon, Grunenwald, Heinrichs, Cassat, Skaar, and Caprioli, 2019). The abundance and distribution of lysyl-PGs produced by staphylococcal communities within infected tissue has never been determined. The spatial structure and unique interplay of host and bacterial factors that comprise abscesses make in vivo studies of this interface challenging, and most technologies do not allow for the selection or enrichment of bacteria-specific metabolic factors from the complex chemical matrix of tissue. Isolating and identifying unique or heterogeneous molecular distributions within these infections can provide insight into potential pathogenic adaptations or resistant S. aureus subpopulations that are more likely to promote chronicity of infection. Traditional discovery approaches for molecular investigation of tissue samples (e.g. LC-MS) require solubilization, eliminating spatial information and diluting low abundance analytes of interest. Matrix-assisted laser desorption/ionization imaging mass spectrometry (MALDI IMS) has proven its utility as an ex vivo technology to study host-pathogen chemical interactions at sites of infection (Cassat et al., 2018; Perry, Spraggins, Sheldon, Grunenwald, Heinrichs, Cassat, Skaar, and Caprioli, 2019; Moore, Caprioli, and Skaar, 2014; Caprioli, Farmer, and Gile, 1997; Blanc, Lenaerts, Dartois, and Prideaux, 2018; Scott, Post, Lerner, Ellis, Lieberman, Shirey, Heeren, Bindila, and Ernst, 2017; Juttukonda et al., 2019; Kehl-Fie, Zhang, Moore, Farrand, Hood, Rathi, Chazin, Caprioli, and Skaar, 2013; Spraggins, Rizzo, Moore, Rose, Hammer, Skaar, and Caprioli, 2015; Wakeman et al., 2016). The distribution and abundance of any mass-to-charge ratio (m/z) can be mapped across the measurement area as an ion image (Caprioli, Farmer, and Gile, 1997). Recent advancements in IMS instrumentation allow for unprecedented molecular specificity and spatial fidelity, in some instances approaching sub-cellular resolution (Spraggins et al., 2019; Niehaus, Soltwisch, Belov, and Dreisewerd, 2019; Prentice, Ryan, Van de Plas, Caprioli, and Spraggins, 2019). Despite advances in imaging technologies, there is no universal technology or modality that can capture all molecular and morphological information in a single experiment. Thus, the incorporation of multiple co-registered and computationally fused imaging modalities can provide a deeper understanding of how molecular signatures are linked to specific tissue features (Patterson, Tuck, Van de Plas, and Caprioli, 2018; Van de Plas, Yang, Spraggins, and Caprioli, 2015). Application of various endogenous fluorescent markers, such as transgenic host fluorophores or bacterial fluorescent reporters, can drive IMS data mining strategies and interpretation (Jones, Cho, Patterson, Van de Plas, Spraggins, Boothby, and Caprioli, 2020). In this workflow, fluorescence microscopy data can be obtained prior to IMS without influencing the molecular signatures detected by mass spectrometry. Furthermore, both datasets can be acquired from a single tissue section, eliminating any issues with registration between serial sections. Herein, we demonstrate the capabilities of multimodal molecular imaging for interrogating SAC molecular architecture by leveraging high performance MALDI Fourier transform ion cyclotron resonance (FT-ICR) IMS computationally fused with fluorescence micrographs of bacterial reporters to investigate the presence and distributions of S. aureus lysyl-PG lipids within soft tissue infections. Results: S. aureus lysyl-PGs are differentially abundant within murine tissue abscesses To investigate molecular species co-localizing with SACs, S. aureus infected tissues were analyzed using a multimodal approach integrating fluorescence microscopy, MALDI IMS of lipids, and hematoxylin and eosin (H&E) staining (Fig. 1). In a systemic model of staphylococcal infection, mice were intravenously inoculated with S. aureus PmntArfp, where red fluorescent protein (RFP) expression is driven by the mntA promoter (PmntA), a gene whose expression increases upon manganese starvation (Perry, Spraggins, Sheldon, Grunenwald, Heinrichs, Cassat, Skaar, and Caprioli, 2019). Seven days post infection (DPI), tissues were harvested, frozen on dry ice, and thinly sectioned for analyses. Visual comparison of the H&E stain and the fluorescence micrograph show co-localization of the RFP fluorophores to the SACs (Fig. 1A & Fig. 1B). All SACs present within the abscess express the RFP reporter. Fluorescence microscopy of the RFP reporter is a chemically non-destructive approach to distinguish bacterial foci prior to IMS. MALDI IMS (Fig.1C, Fig. S1 & S2) of various murine and/or bacterial molecular species highlights the power of IMS to molecularly interrogate complex tissue environments without the use of tags or labels. To isolate and explore relationships of ions that localize to bacterial foci, data-driven image fusion28 was employed. This computational method of data analysis connects the spatial and informational content of two imaging modalities by constructing a mathematical cross-modality model using multivariate (partial least squares) linear regression (Van de Plas, Yang, Spraggins, and Caprioli, 2015). Data-driven image fusion has previously been utilized for predictive applications such as spatial sharpening and out-of-sample predictions (Van de Plas, Yang, Spraggins, and Caprioli, 2015; Neumann, Comi, Spegazzini, Mitchell, Rubakhin, Gillette, Bhargava, and Sweedler, 2018; Lin et al., 2020). However, in this case, data-driven image fusion is used for a discovery-oriented application instead of a predictive one, utilizing only the first model-building phase of the fusion framework to mine relationships between the modalities, foregoing the second predictive phase that usually follows. The objective is to learn which relationships can be detected between the observations made in two distinct modalities. Since these relationships can be retrieved by opening up an empirically learned cross-modal model, the prediction portion of data-driven fusion is not necessary for this approach, and any uncertainties inherent to predictions are avoided. In this study, IMS images are tied to a fluorescence micrograph using fusion-learned multivariate linear models, isolating correlative relationships between the two modalities. Specifically, we searched for relationships between IMS data and the RFP bacterial reporter to isolate ions of bacterial origin. Fusion of the IMS and fluorescence microscopy datasets revealed many ions with high correlations to the fluorescent reporters (Table 1). In this process, a fusion-generated, multivariate linear model that ties fluorescent reporters to IMS-reported ions is examined post-training phase, and the model’s slope coefficients are used as heuristic measures for the strength of the relationship between a specific fluorescent reporter and a specific ion. If there is little to no relationship between the microscopy measurement and a variable (i.e. m/z) in the IMS data, then the slope that connects the two is relatively small or zero. If the slope is relatively large, a change in one microscopy variable correlates with significant change in the other IMS variable. Database searching using The LIPID MAPS Lipidomics Gateway (http://www.lipidmaps.org/)) resulted in tentative identification of many, but not all, ions of interest. MALDI MS/MS experiments to profile fluorescent bacterial foci resulted in the molecular identification of lysyl-PGs (Fig. S3). Subsequent experiments using liquid chromatography coupled to MS/MS of S. aureus culture identified the presence of many PG and lysyl-PG lipids (Fig. S3). Interestingly, lysyl-PGs in Fig. 1D are not present at all locations of bacterial fluorophores. A more detailed comparison of heterogenous molecular profiles of SACs is presented in Fig. S4. This suggests differential production of lysyl-PG across distinct SACs within the same abscess. An extracted MALDI MS spectrum (Fig. 1E) from a single pixel co-localizing with a SAC is annotated for three lysyl-PGs. Notably, tissue abscesses were not observed when mice were infected with S. aureus ΔmprF (Fig. S1). By employing multiple imaging modalities, heterogenous distributions of lysyl-PGs were mapped to staphylococcal communities within a murine infection model. S. aureus lysyl-phosphatidylglycerols and host cationic antimicrobial peptides are present in human infective endocarditis tissue samples. Infective endocarditis is an infection of the endocardium, the inner tissue lining of cardiac chambers and valves. These infections are associated with both high morbidity and mortality among human patients and are characterized by bulky lesions that form on heart valves, called vegetations (Holland, Baddour, Bayer, Hoen, Miro, and Fowler, 2016). Like soft tissue abscesses, endocarditis vegetations are composed of recruited innate immune cells and SACs, as well as coagulation proteins and other blood components (Martin, Witten, Tan, Rodriguez, Blackstone, Pettersson, Seifert, Willard, and Apte, 2020). S. aureus is the most common causative pathogen for this disease (Thuny, Grisoli, Collart, Habib, and Raoult, 2012; Cahill, Baddour, Habib, Hoen, Salaun, Pettersson, Schäfers, and Prendergast, 2017; Pettersson, Hussain, Shrestha, Gordon, Fraser, Ibrahim, and Blackstone, 2014). As such, describing S. aureus human-specific virulence factors is critical for understanding the molecular architecture of infectious lesions (Onyango and Alreshidi, 2018; Dastgheyb and Otto, 2015). To investigate the presence of lysyl-PGs within these infections, infected valve tissue and vegetations from patients with community-acquired methicillin sensitive S. aureus (CA-MSSA) infective endocarditis were subjected to MALDI IMS of lipids. Fig. 2A shows a graphic of an infected aortic valve within a human heart. This highlights the location of the excised tissue shown in the H&E stain. The H&E stain is annotated for vegetation and valve regions. Ion images from MALDI IMS performed on a serial tissue section (Fig. 2B) show localizations of both S. aureus PGs and lysyl-PGs. The selected ion images show that the distributions of the two lipids are not overlapping in this case. Furthering this investigation of CA-MSSA endocarditis, MALDI IMS data of intact proteins were acquired from a serial tissue section. Spectral examination of the averaged IMS spectrum revealed the presence of two high intensity protein distributions at m/z 3371.47 and m/z 3442.54. Ion images of the two species revealed distributions that span the entirety of the vegetation regions (Fig. 3a). Subsequent bottom-up LC-MS/MS workflows identified these analytes as α-Defensin 2 and α-Defensin 1 CAMPs, Fig. 3b. Accurate mass measurements from FT-ICR IMS were used to link imaging data to LC-MS/MS identifications with low ppm mass errors. Notably, the CAMPs are absent from the valve tissue. It is hypothesized that the presence of CAMPs provokes lysyl-PG production as a stress response, potentially influencing the location of lysyl-PGs in the MALDI IMS of lipids from the serial tissue section (Roy, 2009). Expanding on the observations from a murine model of S. aureus infection, heterogeneous distributions of lysyl-PGs as well as host CAMPs were mapped to human infected tissue. Discussion: In this work, we mapped host and staphylococcal molecular responses within infection environments. A multimodal imaging approach combining fluorescence microscopy and MALDI IMS was mined using a computational strategy to isolate ions with correlative spatial relationships to a bacterial fluorescent reporter. A subsequent experiment identified many of the highly correlated ions to be lysyl-PG lipids. Lysyl-PGs and CAMPs were subsequently mapped to endocarditis infection within human tissue. Heterogeneous distributions of lysyl-PGs were observed from both experiments, highlighting the potential for niche-specific responses during infection. These results provide insight into mechanisms of host innate immunity and bacterial pathogenesis at the site of infection. While the production of lysyl-PGs by S. aureus is attributed to virulence and antimicrobial resistance, this evasive response had yet to be observed in vivo (Ernst and Peschel, 2019; Baltz, 2009; Mishra and Bayer, 2013). Lysyl-PGs are not present at all SACs within a murine infection model. This suggests differential molecular responses to host stresses or mechanisms of innate immunity. Lysyl-PG distributions from a human endocarditis tissue sample show the heterogeneous presence of lysyl-PGs with S. aureus PGs and suggest variation in molecular responses, potentially relating to fitness. Distributions of α-defensin 1 and α-defensin 2 CAMPs localize to vegetations and are absent from valve tissue. However, increased presence of lysyl-PGs localizes near the valve tissue. Presence of the lipids at that location as well as the lack of CAMPs could be due to bacterial evasion of the innate immune system. Defensins were not observed in murine systemic infection models due to the lack of production by mice. Many studies have highlighted the molecular heterogeneity present in S. aureus tissue infections (Cassat et al., 2018; Perry, Spraggins, Sheldon, Grunenwald, Heinrichs, Cassat, Skaar, and Caprioli, 2019; Yao, Yu, Qin, Chen, He, Chen, Zhang, and Wang, 2010). While the in vitro roles of lysyl-PGs and other various pathogenic adaptations are known, it is not clear when and where these adaptations are most important during infection (Peschel et al., 2001; Ernst, Staubitz, Mishra, Yang, Hornig, Kalbacher, Bayer, Kraus, and Peschel, 2009; Baltz, 2009; Mishra, and Bayer, 2013; Staubitz, Neumann, Schneider, Wiedemann, and Peschel, 2004). Molecular heterogeneity across infectious foci could be explained by differential responses or mechanisms of host-pathogen interactions, highlighting the potential for niche-specific S. aureus pathogenic strategies. Organ-specific responses to infection may also exist. Increased understanding of when and where these pathogenic adaptations are required for bacterial proliferation can drive the search for alternative, novel therapeutics. Spatially targeted and molecularly specific technologies such as MALDI IMS will assist in relating molecular heterogeneity to conditions, environments, or other previously uncharacterized molecular distributions to further elucidate host-pathogen interactions. Due to the lack of tags or labels, IMS allows for untargeted chemical information from a tissue surface, not achievable by other technologies. Furthermore, untargeted molecular analysis of staphylococcal infections, as provided by MALDI IMS, can drive the search for uniformly expressed bacterial factors, as these factors can serve as candidate vaccine targets. Incorporation of other imaging modalities to MALDI IMS, as exhibited here, allows for increased information about a tissue surface, expanding testable hypotheses. Application of automated data mining strategies, such as data-driven image fusion, can help isolate relationships between datasets. Leveraging this spatial information, chemical mechanisms of bacterial pathogenicity and antibiotic resistance as well as host innate immunity were mapped across infections in murine and human tissue. Significance: Bacterial pathogens have evolved virulence factors to colonize, replicate, and disseminate within the vertebrate host. Although there is an expanding body of literature describing how bacterial pathogens regulate virulence repertoire in response to environmental signals, it is challenging to directly visualize virulence response within the host tissue microenvironment. Multimodal integration of high spatial resolution imaging mass spectrometry and microscopy allow visualization of modifications to the Staphylococcus aureus envelope within infected murine and human tissues. Data mining strategies leveraging multimodal data-driven image fusion of fluorescent bacterial reporters and MALDI FT-ICR imaging mass spectrometry revealed S. aureus lysyl-phosphatidylglycerol lipids localized to distinct bacterial communities within infected tissue. Varied staphylococcal lysyl-phosphatidylglycerol lipid distributions within murine and human infections suggests a heterogeneous, spatially oriented microbial response to host defenses. Limitations of the study: We predict that the detection of heterogenous levels of bacterial lipids within the abscess is due to heterogenous abundance of these lipids throughout the staphylococcal microcolony. However, we are not able to quantify the specific number of bacteria within each microcolony therefore it is possible that differences in the observed lipid levels are partially driven by differences in bacterial abundance throughout the lesion. We think this is unlikely since, qualitatively, there does not appear to be any correlation between lipid signals and bacterial abundance in a microcolony as determined by fluorescence. A second limitation involves the small number of samples imaged in these experiments. Obtaining infected human tissue is particularly challenging. Future work should expand these studies to include larger numbers of samples, and distinct types of lesions, to determine the generalizability of these findings. Finally, like all analytical technologies, there are limitations to the capabilities of the MALDI imaging experiments performed here. There are likely multiple isomers for some of the bacterial lipid species detected due to the variety of potential acyl chain compositions that are possible. We do not think this detracts from the observed molecular heterogeneity because this would only mask potential variability. Improved specificity would further enhance our ability to detect molecular differences between bacterial colonies. STAR Methods text RESOURCE AVAILABILITY Lead Contact: Further information and requests for resources and reagents should be directed to and will be fulfilled by Jeffrey M. Spraggins (jeff.spraggins@vanderbilt.edu). Materials Availability: This study did not generate new unique reagents. Data Code and Availability: All data have been deposited at Figshare and are publicly available as of the date of publication. DOIs are listed in the key resources table. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. EXPERIMENTAL MODEL AND SUBJECT DETAILS Bacterial strains and growth conditions Bacterial strains, primers, and plasmids used in this study are listed in Table 2. Bacteria were routinely cultured at 37°C in Difco Tryptic Soy Broth (TSB) or on Tryptic Soy Agar (TSA) with 10 μg/mL erythromycin or 10 μg/mL chloramphenicol antibiotics supplemented as needed. All growth in liquid medium occurred in an Innova 44 incubator shaking at 180 rpm, unless otherwise noted. Fifteen-milliliter round-bottom polypropylene tubes with aeration lids incubated at a 45° angle were used for all standard cultures of 5 mL. The S. aureus clinical isolate strain Newman served as the genetic background for all experiments (Duthie and Lorenz, 1952). The construction of S. aureus strain Newman p.PmntArfp has been previously described (Cassat et al., 2019). The strain inactivated for lysyl-PG production (mprF::erm) has been previously described (Peschel et al., 2001). Murine Model of Systemic S. aureus Infection All animal experimental protocols were reviewed and approved by the Vanderbilt University Institutional Animal Care and Use Committee (IACUC) and are in compliance with institutional policies, NIH guidelines, the Animal Welfare Act, and American Veterinary Medical Association guidelines on euthanasia. S. aureus strains were streaked from freezer stocks onto TSA with antibiotics, as required, and grown for 24 h at 37°C. Isolated colonies were used to prepare overnight cultures in 5 mL TSB. For all experiments, 6–8 week old female BALB/cJ mice (Jackson Laboratory) were retro-orbitally infected with 1 × 107 CFU in 100 μl of sterile phosphate-buffered saline as previously described (Corbin et al., 2008). Following infection, mice were humanely euthanized on day 4, 7, or 10. The kidney, hearts and livers were removed and immediately frozen on a bed of dry ice. Tissues were stored at −80°C until further processing. Infective Endocarditis Vegetations Vegetations were collected from a single patient prospectively during open-heart surgery for infective endocarditis under an approved Cleveland Clinic Institutional Review Board protocol with verbal patient consent (Protocol #16–1521). Visible blood clots on vegetations were removed with forceps and residual blood was reduced by extensive rinsing in phosphate buffered saline. Vegetations analyzed using mass spectrometry were snap-frozen in liquid nitrogen and stored at −80°C. METHOD DETAILS Materials Ammonium formate, hematoxylin, and eosin were purchased from Sigma Aldrich (St. Louis, MO, USA). 1,5-diaminonapthaline (DAN) and 2,5-dihydroxyacetophenone (DHA) were also purchased from Sigma Aldrich (St. Louis, MO, USA) then purified by recrystallization. Indium tin oxide (ITO) coated slides were purchased from Delta Technologies, Limited (Loveland, CO, USA). All solvents (methanol, ethanol, acetonitrile, acetic acid, chloroform, and xylenes) and optimal cutting temperature compound were purchased from Fisher Scientific (Kalamazoo, MI, USA). Sample Preparation Fresh frozen tissues were serially sectioned at 10 μm using a Leica CM3050s cryostat (Leica Biosystems, Buffalo Grove, IL, USA) and thaw-mounted on glass microscope slides (Fisher Scientific, Kalamazoo, MI, USA) or indium tin oxide (ITO)-coated glass slides (Delta Technologies, Loveland, CO, USA. Sections were stored at −80°C until thawed for analysis under vacuum for approximately 15 min. MALDI matrix for lipid analysis was applied by a robotic aerosol sprayer (TM Sprayer, HTX Technologies, Chapel Hill, NC, USA). Samples for the analysis of lipids were washed using ammonium formate buffer as reported previously then homogeneously coated with DAN using optimized conditions at a surface density of 3.6 μg/mm2 (Angel, Spraggins, Baldwin, and Caprioli, 2012). Samples for the analysis of intact proteins were washed of lipids and salts as previously reported then homogenously coated with DHA using optimized conditions at a surface density of 3.1 μg/mm2 (Zavalin, Yang, Hayden, Vestal, and Caprioli, 2015). Samples for protein analysis were then recrystallized as previously reported (Yang and Caprioli, 2011). Post-MALDI IMS, tissue sections were washed of matrix using methanol and stained with hematoxylin and eosin (Fisher Scientific, Kalamazoo, MI, USA). Bright-field images of stained sections were acquired using a Leica SCN-400 optical slide scanner (Leica Biosystems, Buffalo Grove, IL, USA) at 20X magnification. Fluorescence Microscopy Image Acquisition Fluorescence microscopy images were acquired from tissue sections on ITO-coated slides before tissue washing or matrix application using a fluorescence microscope (Nikon Eclipse 90i, Nikon Instruments Inc., Melville, NY, USA) equipped with a motorized stage and a 10X objective. Resulting pixel resolutions were 0.92 μm/pixel. A TRITC (excitation = 528–553, emission = 590–650) specific epifluorescence filter was used to visualize the RFP reporter at an exposure time of 40 ms. DAPI (excitation = 340–380, emission = 435–485) and FITC (excitation = 465–495, emission = 515–555) specific epifluorescence filters were used to allow visualization of tissue morphology by autofluorescence at exposure times of 100 ms and 150 ms, respectively. Molecular Image Acquisition MALDI IMS of lipids was performed using a modified 9.4T Fourier transform-ion cyclotron resonance mass spectrometer (FT-ICR MS) (SolariX, Bruker Daltonics, Billerica, MA, USA) equipped with an Apollo II dual MALDI/ESI source and a dynamically harmonized ParaCell. The source region of this instrument is modified with a secondary Gaussian laser, resulting in incident spots sizes of ~10 μm. Molecular images of lipids were acquired at pixel spacings of 15 μm for murine tissue (~10 μm laser beam size, 500 laser shots) and 40 μm for human tissue (~30 μm laser beam size, 500 laser shots) in both the x and y directions. Data were collected from m/z 200 to 2,000 in negative ionization mode with a resolving power (m/Δm) of ~50,000 at m/z 500. The time domain file size was set to 512k (free induction decay: 0.28 s). The ion optics were tuned to maximize transmission at the defined m/z range including the funnel RF amplitude (250 Vpp), source octopole (5 MHz, 400 Vpp), collision cell (collision cell voltage: 2.0 V, cell: 2 MHz, 1200 Vpp), time-of-flight delay (0.9 ms), transfer optics (4 MHz, 325 Vpp), and quadrupole (Q1 mass: m/z 350). The source DC optics were tuned to maximize ion transmission and observed intensities (capillary exit −190 V, deflector plate: −200 V, plate offset: −100 V, funnel 1: −100 V, skimmer 1: −20 V) as well as the ICR cell parameters (transfer exit lens: 10 V, analyzer entrance: 10 V, side kick: 0 V, side kick offset: 6.0 V, front trap plate: −1.200 V, and back trap plate: −1.205 V). Ion detection was performed using a sweep excitation power of 22%. MALDI MS/MS of fluorescent foci was performed using a 15T FT-ICR MS (SolariX, Bruker Daltonics, Billerica, MA, USA) equipped with an Apollo II dual MALDI/ESI source and a dynamically harmonized ParaCell. The MALDI source employs a Smartbeam II 2 kHz, frequency tripled Nd:YAG (355 nm) laser. A spectrum was acquired using ~5,000 laser shots, rastered across fluorescent bacterial foci when compared to a fluorescent micrograph. Data were collected from m/z 49 to 900 in negative ionization mode. An isolation window of 5 Da and collision induced dissociation (CID) voltage of 23 V was used for MALDI MS/MS experiments. MALDI IMS of intact proteins was performed using a 15T FT-ICR MS (SolariX, Bruker Daltonics, Billerica, MA, USA) equipped with an Apollo II dual MALDI/ESI source and a dynamically harmonized ParaCell. The MALDI source employs a Smartbeam II 2 kHz, frequency tripled Nd:YAG (355 nm) laser. Molecular images of intact proteins were acquired at pixel spacings of 100 μm for murine tissue with a ~50 μm laser beam size (750 laser shots) in both the x and y directions. Data were collected from m/z 2,000 to 30,000 in positive ionization mode with a resolving power (m/Δm) of ~60,000 at m/z 10,000. The time domain file size was set to 1M (free induction decay: 4.61 s). The ion optics were tuned to maximize transmission at the defined m/z range including the funnel RF amplitude (285 Vpp), source octopole (2 MHz, 525 Vpp), collision cell (collision cell voltage: −8.0 V, cell: 1.4 MHz, 1850 Vpp), time-of-flight delay (3.00 ms), transfer optics (1 MHz, 410 Vpp), and quadrupole (Q1 mass: m/z 1,000). The source DC optics were tuned to maximize ion transmission and observed intensities (capillary exit 250 V, deflector plate: 200 V, plate offset: 100 V, funnel 1: 150 V, skimmer 1: 60 V) as well as the ICR cell parameters (transfer exit lens: −20 V, analyzer entrance: −10 V, side kick: 0 V, side kick offset: −1.5 V, shimming DC bias: 1.5 V, and gated injection DC bias: 1.5 V, back trap plate quench: −30 V). The gated trapping and detection function of the ICR cell was enabled (front and back plates for trapping: 2.5 V, front and back plates for detection: 1.1 V, ramp time: 0.01 s). Ion detection was performed using a sweep excitation power of 45%. FlexImaging 5.0 (Bruker Daltonics, Billerica, MA, USA) and SCiLS Lab (version 2016b, Bruker Daltonics, Billerica, MA, USA) were used to normalize intensities (root mean squared) and visualize ion images. Protein Extraction for LC-MS/MS Methods for LC-MS/MS analysis were previously published with extensive details (Martin, Witten, Tan, Rodriguez, Blackstone, Pettersson, Seifert, Willard, and Apte, 2020). Approximately 150–200 mg of each vegetation was homogenized in 1 mL of T-Per (ThermoFisher Scientific, Waltham, MA, USA) extraction buffer supplemented with protease inhibitor (cOmplete tablets, Roche, Millipore Sigma, Burlington, MA, USA) using a T10 ultra Turrax (IKA, Staufen, Germany) homogenizer, boiled at 95°C for 5 min and cooled to room temperature before adding 1 μL of benzonase (Millipore Sigma, Burlington, MA, USA). The homogenates were tip-probe sonicated at 20% amperage with 3 sec on/off intervals a total of six times using a Q-500 sonicator (Qsonica, Newtown, CT, USA), centrifuged at 15,000 xg for 10 min, and the supernatant was retained for analysis as the soluble fraction (T-Per fraction). The pellet was washed twice with cold PBS and resuspended in a chaotropic buffer (5 M GuHCL, 1% CHAPS, 25 mM NAC2H3O2, 50 mM aminocaproic acid, 5 mM NA2EDTA) supplemented with protease inhibitor (cOmplete tablets, Roche, Millipore Sigma, Burlington, MA, USA) at 4°C with rotation for 72 h, centrifuged at 17,000 xg for 10 min and the supernatant was retained for analysis as the matrix fraction. Fractions were analyzed separately, and data were combined by sample post database search analysis. LC-MS/MS Proteomics Data Acquisition and Analysis Peptide samples were analyzed using a ThermoFisher Scientific Fusion Lumos tribrid mass spectrometer system interfaced with a Thermo Ultimate 3000 UHPLC. The HPLC column was a Dionex 15 cm × 75 μm id Acclaim Pepmap C18, 2 μm, 100 Å reversed phase capillary chromatography column. Five μL volumes of the trypsin-digested extract were injected, peptides were eluted from the column by an acetonitrile/0.1% formic acid gradient at a flow rate of 0.3 μL/min and introduced in-line into the mass spectrometer over a 120 min gradient. The nanospray ion source was operated at 1.9 kV. The digest was analyzed using a data-dependent method with 35% collision-induced dissociation fragmentation of the most abundant peptides every 3 s and an isolation window of 0.7 m/z. Scans were conducted at a maximum resolution of 120,000 at m/z 200 for full MS and 60,000 at m/z 200 for MS/MS. Individual LC-MS/MS raw files were searched against a human database (Uniprot.org) using Proteome Discoverer 2.2 (ThermoFisher Scientific, Waltham, MA, USA). Peptides were identified using a precursor mass tolerance of 10 ppm, and fragment mass tolerance of 0.6 Da. The only static modification was carbamidomethyl (C), whereas dynamic modifications included the light (28.03 Da) dimethyl formaldehyde (N-terminal, K), oxidation (M), deamidation (N, R (citrullination)), acetylation (N-terminal), and Gln to pyro-Glu N-terminal cyclization. Peptides were validated using a false discovery rate (FDR) of 1% against a decoy database. Only high confidence proteins (containing peptides at a 99% confidence level or higher) were recorded from each sample for data analysis. When a peptide corresponding to multiple proteins was identified, all protein accession numbers were reported. However, when labeled as protein groups, peptides corresponding to multiple accession numbers were assigned to one master protein accession (based the number of unique peptides in each accession and overall sequence length of the protein, e.g., ACTG2, ACTB, ACTA1, and ACTA2 have peptides with same sequence, but only the protein which has a unique peptide(s) found is considered a master protein, and all others are combined into the same protein group). LC-MS/MS of lipids from S. aureus Culture LC-MS/MS of lipids from S. aureus culture was completed to relate identifications to ions within IMS data. Separations and analyses were adapted from previously published methods (Laut, Perry, Metzger, Weiss, Stauff, Walker, Caprioli, and Skaar, 2020). Lipids were extracted from cell pellets (normalized by OD600) using the Bligh-Dyer method (Bligh and Dyer, 1959). Extracts were dried under nitrogen and reconstituted in 100μl of 65% acetonitrile, 30% isopropyl alcohol, and 5% water. A 15μl volume of each sample was injected into an Acquity Arc ultra-high-performance liquid chromatography (UPLC) system (Waters Corporation, Milford, MA, USA). Lipids were separated using an Acquity UPLC HSS C18 column (Waters Corporation, Milford, MA, USA) with 1.8μm particle size and dimensions of 2.1 mm by 150 mm. The aqueous solvent system (solvent A) consisted of 60% acetonitrile, 40% water, 0.1% formic acid, and 10 mM ammonium acetate. The organic solvent system (solvent B) consisted of 90% isopropyl alcohol, 10% acetonitrile, 10 mM ammonium acetate, and 0.1% formic acid (100). The following gradient was used: 0 min, 70% solvent A; 0 to 5 min, 70% to 57% solvent A; 5 to 5.1 min, 57% to 50% solvent A; 5.1 to 14 min, 50% to 30% solvent A; 14 to 21 min, 30% to 1% solvent A; 21 to 30 min, 1% solvent A; 30 to 30.1 min, 1% to 70% A. The column was allowed to equilibrate at 70% solvent A for 9.9 min prior to the next injection. The column heater was set at 40°C, and the flow rate was 0.22 ml/min. After separation, samples were introduced by electrospray ionization (2.5-kV capillary; 100°C source temperature; 40-V sampling cone) into a quadrupole-time of flight mass spectrometer (Waters Synapt G2-Si; Waters Corporation, Milford, MA, USA) for analysis in negative-ionization mode (trap and transfer collision energies, 15 V; resolution mode; ion mobility not enabled). Samples were analyzed in data-dependent mode with a survey window at mass-to-charge (m/z) ratios of 500 to 1,750 with a scan time of 0.2 s. Fragmentation data were acquired using a collision energy ramp of 6 to 147 eV (depending on the m/z value selected) with a 30-s exclusion window. The instrument was calibrated using sodium formate prior to analysis. PG and lysyl-PG lipids eluted between 17 and 24 min. QUANTIFICATION AND STATISTICAL ANALYSIS Biological (2) replicates of IMS experiments were completed for the S. aureus murine infection model to ensure reproducibility. Experiments were only completed twice due to large data sizes and lengthy analysis times. Ion intensity noise in MALDI IMS experiments was addressed using root mean squared (RMS) intensity normalization. Data-driven Image Fusion For image fusion analysis with IMS data sets, the mass spectrometry data were treated as a data cube in which the x and y coordinates are pixel dimensions and the z coordinate is m/z. Analogously, the microscopy data map pixel dimensions are x and y, but the z coordinate was the color channels. Image fusion algorithms were used to train a cross-modality model. The model was subsequently opened up to retrieve relative slope coefficients as heuristics for correlations between the microscopy and IMS channels, effectively highlighting image pairs of IMS and fluorescence micrograph data that seem to exhibit potential relationships. In the present work, relationships were mined to identify ions of interest that relate to the red color channel, to discover correlations between IMS-reported ions and the bacterial RFP fluorophore reporter after image registration (Fig. S2). Further information on this algorithm can be found in previous work (Van de Plas, Yang, Spraggins, and Caprioli, 2015). Supplementary Material 2 Acknowledgements: The authors thank Dr. Andreas Peschel (University of Tübingen, Germany) for contribution of the S. aureus strain inactivated for lysyl-PG production, S. aureus ΔmprF, and the Skaar Laboratory and MSRC for critical review of this manuscript. This work was funded by the NIH/National Institute of General Medical Sciences (2P41 GM103391-07 awarded to R.M.C.), the NIH National Institute of Allergy and Infectious Diseases (R01AI138581 awarded to E.P.S and J.M.S.; R01AI145992 awarded to E.P.S, J.M.S, and J.E.C.; R01AI132560 awarded to J.E.C; and R01AI069233, R01AI073843, and R01AI150701 awarded to E.P.S.), and the Burroughs Wellcome Fund. J.E.C. is also supported a Career Award for Medical Scientists from the Burroughs Wellcome Fund. S.S.A. and D.R.M. are supported by the Allen Distinguished Investigator Program, through support made by The Paul G. Allen Frontiers Group and the American Heart Association (to S.S.A). The 15T FT-ICR MS in the Mass Spectrometry Research Center at Vanderbilt University was acquired through the NIH Shared Instrumentation Grant Program (1S10OD012359). The Fusion Lumos instrument at the Lerner Research Institute Proteomics and Metabolomics Core was purchased via an NIH shared instrument grant, 1S10OD023436-0. Fig. 1: MALDI IMS spatially informed by fluorescence microscopy of S. aureus, PmntArfp maps chemical information from staphylococcal colonies. A: H&E staining of a 7 DPI murine renal tissue section shows an abscess at the top right. B: A fluorescence micrograph of the same tissue section as A and C prior to MALDI IMS highlights the location of SACs (red fluorescent protein (RFP)). C: An ion overlay image highlights the various molecular constituents of the abscess and chemical complexity of the tissue section. These distributions also highlight kidney tissue substructures, such as the medulla and cortex. D: Lysyl-phosphatidylglycerol lipid species were isolated using image fusion data mining strategies (relative slopes shown) and identified by subsequent MALDI MS/MS experiments. The red crosshair is present to use as a reference point across the fluorescence micrograph zoom and ion images. E: A zoom of single extracted mass spectrum shows signals corresponding to lysyl-PG species. See also Figs. S1 – S4. Fig. 2: MALDI IMS of lipids from infective endocarditis tissue reveals S. aureus PGs and lysyl-PGs. A) A graphic depicts infected human heart tissue lesions. A tissue stain of an infected heart lesion is annotated for normal and diseased tissue. B) Ion images of S. aureus PGs show distributions across the diseased tissue. Increased presence of lysyl-PGs can be observed near valve tissue (zoom). Fig. 3: MALDI IMS of intact proteins from an infective endocarditis vegetation reveals CAMPs colocalizing with diseased tissue A) Ion images of host CAMPs, α-defensin 2 and α-defensin 1 show high abundances within vegetations. B) LC-MS/MS spectra from bottom-up proteomics workflows show peptides from α-defensin 2 and α-defensin 1. Table 1: Selected ions isolated by image fusion of IMS data and fluorescence microscopy of the RFP bacterial reporter. Lipid identifications are based on accurate measurements. Asterisks (*) indicate final identification using MS/MS data and was not present within the database. Part per million (ppm) errors are calculated using theoretical m/z values. Error values were calculated by comparison to a single extracted spectrum, calibrated internally using known lipid species. The image fusion relative slope is for the red channel, corresponding to the bacterial fluorescent reporter. Other ions and relative slopes for all fluorescence channels are in the supplemental materials. Ion Rank m/z Lipid Identification Database Matches ppm Error Image Fusion Relative Slope 2 735.5178 PG(33:0) 3 0.6 544.1 7 797.6531 PE-Cer(d44:2) 1 1.4 354.2 8 616.4720 CerP(d34:1) 2 1.3 346.9 9 707.4871 PG(31:0) 3 0.3 343.9 10 835.5811 Lysyl-PG(31:0) 2* 0.9 336.4 12 863.6132 Lysyl-PG(33:0) 3* 0.1 305.8 15 747.5162 PG(34:1) 4 2.7 262.7 36 699.4974 PA(36:2) 4 0.5 174.7 44 721.5044 PG(32:0) 4 2.7 140.7 63 849.5987 Lysyl-PG(32:0) 1* 1.5 102.2 Table 2. S. aureus strains and plasmids used in this study. Strain Genotype Description Reference Newman WT Wild-type, methicillin-sensitive clinical isolate (Duthie and Lorenz, 1952) Newman mprF::erm In frame deletion of mprF gene marked with erythromycin resistance cassette (Peschel et al., 2001) Plasmid Description pOS1 PmntA rfp mntA promoter driving expression of a red fluorescent protein, responds to manganese starvation. (Cassat et al., 2019) KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Bacterial and Virus Strains Staphylococcus aureus Newman PmntArfp Cassat et al., 2018 Newman PmntArfp Staphylococcus aureus Newman Clinical Isolate, Duthie and Lorenz, 1952 Newman Staphylococcus aureus ΔmprF Peschel et al., 2001 mprF::erm Biological Samples Human Infective Endocarditis Vegetations Cleveland Clinic Protocol #16–1521 Chemicals, Peptides, and Recombinant Proteins 1,5-diaminonapthalene (DAN) Sigma Aldrich D21200-25G, CAS: 2243-62-1 2,5-dihydroxyacetophenone (DHA) Sigma Aldrich D107603-25G, CAS: 490-78-8 Critical Commercial Assays Deposited Data Data are available at Figshare Figshare https://doi.org/10.6084/m9.figshare.12780077.v2 Experimental Models: Cell Lines Experimental Models: Organisms/Strains Female BALB/c mice (6–8 week old) The Jackson Laboratory JAX: 000651 Oligonucleotides Recombinant DNA Software and Algorithms Data Driven Image Fusion Van de Plas et al., 2015 https://fusion.vueinnovations.com/ SCiLS Lab, version 2019b Bruker Daltonics, Billerica, MA, USA https://www.bruker.com/products/mass-spectrometry-and-separations/ms-software/scils/overview.html Other Highlights: Multimodal imaging shows S. aureus envelope modification in situ Microscopy-IMS image fusion reveals S. aureus lysyl-PGs linked to bacterial foci Varied lipid profiles imply spatially driven S. aureus response to host defenses This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Declaration of Interests: The authors declare no interests. References: Beceiro A ; Tomás M ; Bou G (2013). Antimicrobial Resistance and Virulence: a Successful or Deleterious Association in the Bacterial World? Clin. Microbiol. Rev 26 (2 ), 185–230.23554414 Casadevall A ; Pirofski L.-a. (2000). Host-Pathogen Interactions: Basic Concepts of Microbial Commensalism, Colonization, Infection, and Disease. Infect. Immun 68 (12 ), 6511–6518.11083759 Kohanski MA ; Dwyer DJ ; Collins JJ (2010). How antibiotics kill bacteria: from targets to networks. Nat. Rev. Microbiol 8 (6 ), 423–435.20440275 Muthaiyan A ; Silverman JA ; Jayaswal RK ; Wilkinson BJ (2008). Transcriptional Profiling Reveals that Daptomycin Induces the Staphylococcus aureus Cell Wall Stress Stimulon and Genes Responsive to Membrane Depolarization. Antimicrob. Agents Chemother 52 (3 ), 980–990.18086846 Slavetinsky C ; Kuhn S ; Peschel A (2017). Bacterial aminoacyl phospholipids – Biosynthesis and role in basic cellular processes and pathogenicity. Biochim. Biophys. Acta – Mol. Cell Biol. Lipids 1862 (11 ), 1310–1318. Peschel A ; Jack RW ; Otto M ; Collins LV ; Staubitz P ; Nicholson G ; Kalbacher H ; Nieuwenhuizen WF ; Jung G ; Tarkowski A ; van Kessel KPM ; van Strijp JAG (2001). Staphylococcus aureus Resistance to Human Defensins and Evasion of Neutrophil Killing via the Novel Virulence Factor Mprf Is Based on Modification of Membrane Lipids with Lysine. J. Exp. Med 193 (9 ), 1067–1076.11342591 Cheng AG ; DeDent AC ; Schneewind O ; Missiakas D (2011). A play in four acts: Staphylococcus aureus abscess formation. Trends. Microbiol 19 (5 ), 225–232.21353779 Kobayashi SD ; Malachowa N ; DeLeo FR (2015). Pathogenesis of Staphylococcus aureus Abscesses. Am. J. Path 185 (6 ), 1518–1527.25749135 Cheng AG ; Kim HK ; Burts ML ; Krausz T ; Schneewind O ; Missiakas DM (2009). Genetic requirements for Staphylococcus aureus abscess formation and persistence in host tissues. FASEB J. 23 (10 ), 3393–3404.19525403 Ernst CM ; Staubitz P ; Mishra NN ; Yang S-J ; Hornig G ; Kalbacher H ; Bayer AS ; Kraus D ; Peschel A (2009). The Bacterial Defensin Resistance Protein MprF Consists of Separable Domains for Lipid Lysinylation and Antimicrobial Peptide Repulsion. PLoS Pathog. 5 (11 ), e1000660.19915718 Ernst CM ; Peschel A (2019). MprF-mediated daptomycin resistance. Int. J. Med. Microbiol 309 (5 ), 359–363.31182276 Baltz RH (2009). Daptomycin: mechanisms of action and resistance, and biosynthetic engineering. Curr. Opin. Chem. Biol 13 (2 ), 144–151.19303806 Mishra NN ; Bayer AS (2013). Correlation of Cell Membrane Lipid Profiles with Daptomycin Resistance in Methicillin-Resistant Staphylococcus aureus. Antimicrob. Agents Chemother 57 (2 ), 1082–1085.23254419 Cassat JE ; Moore JL ; Wilson KJ ; Stark Z ; Prentice BM ; Van de Plas R ; Perry WJ ; Zhang Y ; Virostko J ; Colvin DC ; Rose KL ; Judd AM ; Reyzer ML ; Spraggins JM ; Grunenwald CM ; Gore JC ; Caprioli RM ; Skaar EP (2018). Integrated molecular imaging reveals tissue heterogeneity driving host-pathogen interactions. Sci. Transl. Med 10 (432 ). Perry WJ ; Spraggins JM ; Sheldon JR ; Grunenwald CM ; Heinrichs DE ; Cassat JE ; Skaar EP ; Caprioli RM (2019). Staphylococcus aureus exhibits heterogeneous siderophore production within the vertebrate host. Proc. Natl. Acad. Sci 116 (44 ), 21980–21982.31611408 Moore JL ; Caprioli RM ; Skaar EP (2014). Advanced mass spectrometry technologies for the study of microbial pathogenesis. Curr. Opin. Microbiol 19 , 45–51.24997399 Caprioli RM ; Farmer TB ; Gile J (1997). Molecular Imaging of Biological Samples: Localization of Peptides and Proteins Using MALDI-TOF MS. Anal. Chem 69 (23 ), 4751–4760.9406525 Blanc L ; Lenaerts A ; Dartois V ; Prideaux B (2018). Visualization of Mycobacterial Biomarkers and Tuberculosis Drugs in Infected Tissue by MALDI-MS Imaging. Anal. Chem 90 (10 ), 6275–6282.29668262 Scott AJ ; Post JM ; Lerner R ; Ellis SR ; Lieberman J ; Shirey KA ; Heeren RMA ; Bindila L ; Ernst RK (2017). Host-based lipid inflammation drives pathogenesis in Francisella infection. Proc. Natl. Acad. Sci 114 (47 ), 12596–12601.29109289 Juttukonda LJ ; Berends ETM ; Zackular JP ; Moore JL ; Stier MT ; Zhang Y ; Schmitz JE ; Beavers WN ; Wijers CD ; Gilston BA ; Kehl-Fie TE ; Atkinson J ; Washington MK ; Peebles RS ; Chazin WJ ; Torres VJ ; Caprioli RM ; Skaar EP (2017). Dietary Manganese Promotes Staphylococcal Infection of the Heart. Cell Host & Microbe 22 (4 ), 531–542.28943329 Kehl-Fie TE ; Zhang Y ; Moore JL ; Farrand AJ ; Hood MI ; Rathi S ; Chazin WJ ; Caprioli RM ; Skaar EP (2013). MntABC and MntH Contribute to Systemic Staphylococcus aureus Infection by Competing with Calprotectin for Nutrient Manganese. Infect. Immun 81 (9 ), 3395–3405.23817615 Spraggins JM ; Rizzo DG ; Moore JL ; Rose KL ; Hammer ND ; Skaar EP ; Caprioli RM (2015). MALDI FTICR IMS of Intact Proteins: Using Mass Accuracy to Link Protein Images with Proteomics Data. J. Am. Soc. Mass Spectrom 26 (6 ), 974–985.25904064 Wakeman CA ; Moore JL ; Noto MJ ; Zhang Y ; Singleton MD ; Prentice BM ; Gilston BA ; Doster RS ; Gaddy JA ; Chazin WJ ; Caprioli RM ; Skaar EP (2016). The innate immune protein calprotectin promotes Pseudomonas aeruginosa and Staphylococcus aureus interaction. Nat. Commun 7 , 11951.27301800 Spraggins JM ; Djambazova KV ; Rivera ES ; Migas L ; Neumann EK ; Fuetterer A ; Suetering J ; Goedecke N ; Ly A ; Van de Plas R ; Caprioli RM (2019). High Performance Molecular Imaging with MALDI Trapped Ion Mobility Time-of-Flight (timsTOF) Mass Spectrometry. Anal. Chem 91 (22 ) 14552–14560.31593446 Niehaus M ; Soltwisch J ; Belov ME ; Dreisewerd K (2019). Transmission-mode MALDI-2 mass spectrometry imaging of cells and tissues at subcellular resolution. Nat. Methods 16 (9 ), 925–931.31451764 Prentice BM ; Ryan DJ ; Van de Plas R ; Caprioli RM ; Spraggins JM (2019). Enhanced Ion Transmission Efficiency up to m/z 24 000 for MALDI Protein Imaging Mass Spectrometry. Anal. Chem 90 (8 ), 5090–5099. Patterson NH ; Tuck M ; Van de Plas R ; Caprioli RM (2018). Advanced Registration and Analysis of MALDI Imaging Mass Spectrometry Measurements through Autofluorescence Microscopy. Anal. Chem 90 (21 ),12395–12403 30272960 Van de Plas R ; Yang J ; Spraggins J ; Caprioli RM (2015). Fusion of mass spectrometry and microscopy: a multi-modality paradigm for molecular tissue mapping. Nat. Methods 12 (4 ), 366–372.25707028 Jones MA ; Cho SH ; Patterson NH ; Van de Plas R ; Spraggins JM ; Boothby MR ; Caprioli RM (2020). Discovering new lipidomic features using cell type specific fluorophore expression to provide spatial and biological specificity in a multimodal workflow with MALDI Imaging Mass Spectrometry. Anal. Chem 92 (10 ), 7079–7086.32298091 Neumann EK ; Comi TJ ; Spegazzini N ; Mitchell JW ; Rubakhin SS ; Gillette MU ; Bhargava R ; Sweedler JV (2018). Multimodal Chemical Analysis of the Brain by High Mass Resolution Mass Spectrometry and Infrared Spectroscopic Imaging. Anal. Chem 90 (19 ), 11572–11580.30188687 Lin L-E ; Chen C-L ; Huang Y-C ; Chung H-H ; Lin C-W ; Chen K-C ; Peng Y-J ; Ding S-T ; Wang M-Y ; Shen T-L ; Hsu C-C (2020). Precision biomarker discovery powered by microscopy image fusion-assisted high spatial resolution ambient ionization mass spectrometry imaging. Anal. Chim. Acta 1100 , 75–87.31987155 Holland TL ; Baddour LM ; Bayer AS ; Hoen B ; Miro JM ; Fowler VG Jr. (2016). Infective endocarditis. Nat. Rev. Dis. Primers 2 , 16059.27582414 Martin DR ; Witten JC ; Tan CD ; Rodriguez ER ; Blackstone EH ; Pettersson GB ; Seifert DE ; Willard BB ; Apte SS (2020). Proteomics identifies a convergent innate response to infective endocarditis and extensive proteolysis in vegetation components. JCI Insight 5 (14 ). Thuny F ; Grisoli D ; Collart F ; Habib G ; Raoult D (2012). Management of infective endocarditis: challenges and perspectives. Lancet 379 (9819 ), 965–975.22317840 Cahill TJ ; Baddour LM ; Habib G ; Hoen B ; Salaun E ; Pettersson GB ; Schäfers HJ ; Prendergast BD (2017). Challenges in Infective Endocarditis. J. Am. Coll. Cardiol 69 (3 ), 325–344.28104075 Pettersson GB ; Hussain ST ; Shrestha NK ; Gordon S ; Fraser TG ; Ibrahim KS ; Blackstone EH (2014). Infective endocarditis: An atlas of disease progression for describing, staging, coding, and understanding the pathology. J. Thorac. Cardiovasc. Surg 147 (4 ), 1142–1149.24507402 Onyango LA ; Alreshidi MM (2018). Adaptive Metabolism in Staphylococci: Survival and Persistence in Environmental and Clinical Settings. J. Pathog 2018 (378 ), 1–11. Dastgheyb SS ; Otto M (2015). Staphylococcal adaptation to diverse physiologic niches: an overview of transcriptomic and phenotypic changes in different biological environments. Future Microbiol. 10 (12 ), 1981–1995.26584249 Roy H (2009). Tuning the properties of the bacterial membrane with aminoacylated phosphatidylglycerol. IUBMB Life 61 (10 ), 940–953.19787708 Yao D ; Yu F.-y. ; Qin Z.-q. ; Chen C ; He S.-s. ; Chen Z.-q. ; Zhang X.-q. ; Wang L.-x. (2010). Molecular characterization of Staphylococcus aureus isolates causing skin and soft tissue infections (SSTIs). BMC Infect. Dis 10 (1 ), 133.20500885 Staubitz P ; Neumann H ; Schneider T ; Wiedemann I ; Peschel A (2004). MprF-mediated biosynthesis of lysylphosphatidylglycerol, an important determinant in staphylococcal defensin resistance. FEMS Microbiol. Lett 231 (1 ), 67–71.14769468 Duthie ES ; Lorenz LL (1952). Staphylococcal Coagulase: Mode of Action and Antigenicity. Microbiology 6 (1–2 ), 95–107. Corbin BD ; Seeley EH ; Raab A ; Feldmann J ; Miller MR ; Torres VJ ; Anderson KL ; Dattilo BM ; Dunman PM ; Gerads R ; Caprioli RM ; Nacken W ; Chazin WJ ; Skaar EP (2008). Metal Chelation and Inhibition of Bacterial Growth in Tissue Abscesses. Science 319 (5865 ), 962–965.18276893 Angel PM ; Spraggins JM ; Baldwin HS ; Caprioli R (2012). Enhanced Sensitivity for High Spatial Resolution Lipid Analysis by Negative Ion Mode Matrix Assisted Laser Desorption Ionization Imaging Mass Spectrometry. Anal. Chem 84 (3 ), 1557–1564.22243218 Zavalin A ; Yang J ; Hayden K ; Vestal M ; Caprioli RM (2015). Tissue protein imaging at 1 μm laser spot diameter for high spatial resolution and high imaging speed using transmission geometry MALDI TOF MS. Anal. Bioanal. Chem 407 (8 ), 2337–2342.25673247 Yang J ; Caprioli RM (2011). Matrix sublimation/recrystallization for imaging proteins by mass spectrometry at high spatial resolution. Anal. Chem 83 (14 ), 5728–34.21639088 Laut CL ; Perry WJ ; Metzger AL ; Weiss A ; Stauff DL ; Walker S ; Caprioli RM ; Skaar EP (2020). Bacillus anthracis Responds to Targocil-Induced Envelope Damage through EdsRS Activation of Cardiolipin Synthesis. MBio. 11 (2 ), e03375–19.32234818 Bligh EG ; Dyer WJ (1959). A Rapid Method of Total Lipid Extraction and Purification. Can. J. Biochem. Physiol 37 (8 ), 911–917.13671378

PMC009xxxxxx/PMC9308757.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9802571 20730 Mol Cell Mol Cell Molecular cell 1097-2765 1097-4164 35588748 9308757 10.1016/j.molcel.2022.04.022 NIHMS1800629 Article Sexually dimorphic RNA helicases DDX3X and DDX3Y differentially regulate RNA metabolism through phase separation Shen Hui 17 Yanas Amber 127 Owens Michael C. 12 Zhang Celia 1 Fritsch Clark 3 Fare Charlotte M. 12 Copley Katie E. 14 Shorter James 12 Goldman Yale E. 256 Liu Kathy Fange 128* 1 Department of Biochemistry and Biophysics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA 2 Graduate Group in Biochemistry and Molecular Biophysics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA 3 Graduate Group in Cellular and Molecular Biology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA 4 Graduate Group in Neuroscience, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA 5 Department of Physiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA 6 Pennsylvania Muscle Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA 7 These authors contributed equally 8 Lead contact AUTHOR CONTRIBUTION H.S. and K.F.L. designed the experiments. H.S., and K.F.L. processed the APEX-seq data. H.S., A.Y., M.C.O., and C.Z. generated the constructs and the mammalian cell lines used in this study. H.S., A.Y., and C.Z. purified the recombinant DDX3X and DDX3Y (full-length proteins and the truncation variants) used in this study. H.S. and A.Y. performed the immunofluorescence imaging experiments and analyzed the data. A.Y., M.C.O., and C.Z. performed the ATPase activity assays. C.M.F. performed FUS aggregation assays and K.E.C performed TDP-43 aggregation assays with the suggestions from J.S. A.Y., and C.F. performed the smFRET studies under the direction of Y.E.G. H.S., M.C.O., and A.Y. wrote the manuscript together with K.F.L. All authors participated in the discussion and editing of the manuscript. * Correspondence: liufg@pennmedicine.upenn.edu 15 5 2022 21 7 2022 18 5 2022 21 7 2023 82 14 25882603.e9 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY Sex differences are pervasive in human health and disease. One major key to sex-biased differences lies in the sex chromosomes. Although the functions of the X-chromosome proteins are well appreciated, how they compare with their Y-chromosome homologs remains elusive. Herein, using ensemble and single-molecule techniques, we report that the sex chromosome-encoded RNA helicases DDX3X and DDX3Y are distinct in their propensities for liquid-liquid phase separation (LLPS), dissolution, and translation repression. We demonstrate that the N-terminal intrinsically disordered region of DDX3Y more strongly promotes LLPS than the corresponding region of DDX3X and that the weaker ATPase activity of DDX3Y compared to DDX3X contributes to the slower disassembly dynamics of DDX3Y-positive condensates. Interestingly, DDX3Y-dependent LLPS represses mRNA translation and enhances aggregation of FUS more strongly than DDX3X-dependent LLPS. Our study provides a platform for future comparisons of sex chromosome-encoded protein homologs, providing insights into sex differences in RNA metabolism and human disease. eTOC Blurb: Shen et al. report that the Y chromosome-encoded RNA helicase DDX3Y has stronger propensity for liquid-liquid phase separation compared to its X chromosome-encoded homolog DDX3X, which results in the stronger repression of mRNA translation and promotion of FUS aggregation under stress conditions. Graphical Abstract Sex chromosome homolog proteins DDX3X DDX3Y RNA helicase liquid-liquid phase separation condensates ATPase activity translation repression pmcINTRODUCTION Many human disorders manifest in a sex-biased manner, yet the molecular mechanisms responsible for these biases are not fully understood (Mauvais-Jarvis et al., 2020). On a genetic level, the most significant differences between males and females lie in the sex chromosomes. Although the X and Y chromosomes are, by and large, not homologous, a handful of X chromosome-encoded proteins have Y chromosome-encoded homologs (Bellott et al., 2014). These Y chromosome-encoded homologs were historically thought to only be expressed and to function in the reproductive system. However, a growing body of evidence suggests that Y chromosome-encoded homologs are not only expressed throughout the body at the transcript and protein levels but are also evolutionarily conserved (Godfrey et al., 2020). However, the functional differences between the protein homologs encoded on the X and Y chromosomes have not been thoroughly investigated. Emerging evidence is starting to reveal that Y chromosome-encoded proteins may function differently from their X chromosome-encoded homologs (Gozdecka et al., 2018; Nguyen et al., 2020; Shi et al., 2021). One pair of sex chromosome-encoded proteins are the RNA helicases DDX3X and DDX3Y. In females (specified as XX individuals), DDX3X escapes X-inactivation, and two copies of the protein are expressed, whereas in males (specified as XY individuals), one copy of each of the X- and Y-linked proteins is expressed (Cotton et al., 2015; Ditton et al., 2004). The DDX3X gene (Xp11.4) encodes a DEAD-box RNA helicase (Sharma and Jankowsky, 2014), which is evolutionarily conserved in C. elegans (LAF-1), yeast (Ded1p), Drosophila (Belle), and humans (Sharma and Jankowsky, 2014). The functions of DDX3X are more well studied than those of DDX3Y, and include its activity as a translation-initiation factor (Lai et al., 2008; Lee et al., 2008) for a set of mRNAs with highly structured 5’-UTRs (Ku et al., 2019; Lai et al., 2008; Lee et al., 2008; Phung et al., 2019; Soto-Rifo et al., 2012), and at repeat-associated non-AUG (RAN) start sites (Cheng et al., 2019; Linsalata et al., 2019). In addition, DDX3X (and its homologs) undergoes liquid-liquid phase separation (LLPS) and is a conserved component of stress granules (SGs) (Beckham et al., 2008; Elbaum-Garfinkle et al., 2015; Iserman et al., 2020; Shih et al., 2012; Valentin-Vega et al., 2016). SGs, which are composed of RNAs and proteins and form as part of the stress response, are correlated with changes to mRNA metabolism, including translational repression (Kimball et al., 2003; Moon et al., 2019). Dysregulation of SGs has been implicated in a wide range of human disorders, including cardiomyopathy, Alzheimer’s disease, and amyotrophic lateral sclerosis (ALS), many of which present with a varying degree of sex-biased incidences, progressions, and outcomes (Ash et al., 2014; Baradaran-Heravi et al., 2020; Ferretti et al., 2018; Manjaly et al., 2010; McCombe and Henderson, 2010; Meyer et al., 2014; Schneider et al., 2020; Watkins et al., 2020). Additionally, DDX3X mutations can lead to DDX3X syndrome, a disorder which accounts for 1–3% of intellectual disability cases and is more prevalent in females than in males (Iossifov et al., 2014; Lennox et al., 2020; Ruzzo et al., 2019; Scala et al., 2019; Snijders Blok et al., 2015; Wang et al., 2018b). Notably, the presence of the DDX3Y gene can only sometimes compensate for the loss of the DDX3X gene (Chen et al., 2016). For example, Ddx3y cannot fully compensate for the loss of Ddx3x during embryonic and neuronal development in mice (Chen et al., 2016; Patmore et al., 2020). Furthermore, male mice, but not female mice, survive conditional knockout of Ddx3x in the bone marrow, but these male mice present with distinct deficiencies in innate antimicrobial immunity compared to Ddx3x conditional knockout female mice (Szappanos et al., 2018). These findings highlight the fact that DDX3X and DDX3Y have roles beyond the reproductive system and suggest that Ddx3x and Ddx3y possibly have distinct functions in immune cells (Szappanos et al., 2018). Indeed, recently compiled proteomic databases show that both DDX3X and DDX3Y proteins are expressed in the human immune system, including in T-cells, B-cells, and NK-cells (Bryk and Wisniewski, 2017; Joshi et al., 2019). Moreover, previous studies revealed that the DDX3Y protein is present in the enteric nervous system and the human heart (Cardinal et al., 2020; Godfrey et al., 2020; Vakilian et al., 2015). Still, potential functional differences between DDX3X and DDX3Y and their contributions to sex-biased human diseases are largely unknown. Like other DEAD-box helicases, DDX3X and DDX3Y contain a helicase core composed of two RecA-like domains and one intrinsically disordered region (IDR) on each of the N- and C-termini. IDRs are frequently involved in the process of LLPS, driven by weak multivalent interactions (Figure 1A). Although DDX3X and DDX3Y share 92% amino acid sequence identity overall, the N-terminal IDRs (IDR1) are more divergent, accounting for 60% difference between DDX3X and DDX3Y (Figure S1A and S1B). IDR1 of DDX3X specifically is known to be essential for its LLPS in vitro and inside cells (Saito et al., 2019; Shih et al., 2012). Since many of the differences between the sequences of DDX3X and DDX3Y are concentrated in IDR1, we wondered whether DDX3X and DDX3Y differ in their propensity to LLPS and consequently differ in responding to cellular stress. Here, we show that DDX3Y has a greater LLPS propensity than DDX3X. DDX3Y-positive SGs are less dynamic (less able to exchange particles with the light phase) than DDX3X-positive SGs. Furthermore, we demonstrate that while DDX3X-positive and DDX3Y-positive SGs share a large overlap of RNA constituents, there are also RNA components that are unique to either SG. We also show that the condensation of either DDX3X or DDX3Y represses the translation of RNAs, with DDX3Y condensation showing a significantly stronger inhibitory effect. Additionally, we show that both helicases specifically augment the aggregation of FUS in vitro and in cells, with DDX3Y having a more profound effect. Collectively, our results suggest that these sexually dimorphic RNA helicases differentially regulate RNA metabolism through their distinct biochemical and biophysical properties, which might contribute to sex bias in human diseases. RESULTS DDX3Y has a stronger propensity than DDX3X for in vitro and cellular phase separation Given that DDX3X is known to undergo LLPS in vitro, we wanted to establish whether DDX3Y could also phase separate, and how it is compared to DDX3X. Thus, we purified mCherry-tagged full-length DDX3X and DDX3Y to near homogeneity (Figure S1C and S1D). Full-length mCherry-tagged DDX3X and DDX3Y formed noticeable droplets in vitro, and the addition of poly(U)-RNA greatly stimulated this process (Figure 1B and 1C). Strikingly, DDX3Y phase separation was more strongly enhanced by the addition of poly(U)-RNA than DDX3X in vitro; the average integrated fluorescence intensity of DDX3Y droplets was ~10-fold higher than an equal amount of DDX3X (Figure 1B and 1C). As shown in Figure S1E, RNase treatment had no influence on the in vitro droplet formation, indicating negligible RNA carry-over during protein purification. Furthermore, we determined the saturation concentration (Csat) of DDX3X and DDX3Y using sedimentation analysis. The Csat of DDX3X was ~5 μM whereas DDX3Y was ~3 μM, indicating that DDX3Y undergoes LLPS at lower concentrations than DDX3X (Figure 1D). In parallel, we performed a turbidity assay; the results consistently showed that DDX3Y gave substantially higher turbidity at the same total protein concentrations compared to DDX3X (Figure 1E), suggesting a greater degree of phase separation. We next studied the dynamics of DDX3X and DDX3Y droplets using fluorescent recovery after photobleaching (FRAP). The recovery halftime for DDX3X droplets was ~21.9 seconds, which was significantly faster than the ~83.4 seconds for DDX3Y droplets (Figures 1F, 1G, S1F, and Movies S1 – S4). Also, we observed a larger mobile fraction for DDX3X droplets (~99%) than for DDX3Y droplets (~78%) (Figure 1H). These results demonstrate that DDX3X is less prone to phase separation and that DDX3X droplets are more dynamic than DDX3Y droplets in vitro. To assess if DDX3Y enters cellular SGs, we expressed DDX3X or DDX3Y in several mammalian cell types, each of which lacks endogenous DDX3Y (HeLa, N2a, and HEK 293T cells). Before expressing either protein, we performed a transient knockdown of endogenous DDX3X with >80% knockdown efficiency (Figure S2A). Exogenous expressions of DDX3X and DDX3Y were at a similar level (Figure S2B and S2C). DDX3X and DDX3Y were diffuse throughout the cytoplasm in unstressed HeLa cells (Figure S2D). Upon arsenite treatment (a commonly used oxidative-stress inducer (Markmiller et al., 2018; Protter and Parker, 2016)), both DDX3X and DDX3Y colocalized with the stress granule marker, G3BP1 (Markmiller et al., 2018) (Figure 2A). Strikingly, the total area of DDX3Y-positive SGs was larger than DDX3X-positive SGs in HeLa cells (1.6-fold), N2a cells (1.3-fold), and HEK293T cells (1.7-fold) (Figure 2B). To control for possible differences in protein concentrations, we expressed DDX3X and DDX3Y across a range of concentrations in HeLa cells with endogenous DDX3X depleted. The results consistently showed that DDX3Y-positive granules were significantly larger than DDX3X-positive granules at similar expressed concentrations (Figure S2E – S2G). Additionally, we quantified the protein half-life of DDX3X and DDX3Y in HeLa cells using cycloheximide (CHX) chase experiments. As shown in Figure 2C and 2D, the cellular half-life of DDX3X is indistinguishable from DDX3Y. FRAP experiments performed on live HeLa cells expressing DDX3X-EGFP or DDX3Y-EGFP (Figure S2H) reveal that the average recovery halftime of DDX3X-positive granules was ~7.6 s, which was significantly faster than the ~10.2 s measured for DDX3Y-positive SGs (Figure 2E – 2G, Movies S5 and S6). Additionally, a larger mobile fraction was observed in DDX3X-positive SGs (86%) than in DDX3Y-positive SGs (77%) (Figure 2G), consistent with our in vitro FRAP experiments. These results together suggest that DDX3X-positive SGs are more dynamic than DDX3Y-positive SGs. IDR1 is a major contributor to the higher phase separation propensity of DDX3Y We constructed several truncated variants of DDX3X and DDX3Y to study the effect of each domain on SG partitioning (Figure S3A and S3B). Deletion of IDR1 from DDX3X (DDX3XΔIDR1) and DDX3Y (DDX3YΔIDR1) led to the sequestration of the variants into cell nuclei and thus completely prevented either protein from entering cytoplasmic SGs, which is consistent with previous work showing that IDR1 of DDX3X contains a nuclear export sequence (Yedavalli et al., 2004) (Figures S1A and S3C). Both DDX3XΔIDR1 and DDX3YΔIDR1 were diffuse in the nucleus, even with arsenite treatment (Figure S3C), so we hypothesized that the high nuclear RNA concentrations prevented these truncations from forming condensates, since this phenomenon has been seen with other RNA-binding proteins with IDRs (Maharana et al., 2018). To test this, we treated cells expressing DDX3XΔIDR1 or DDX3YΔIDR1 with actinomycin D (ActD), which inhibits transcription and thus decreases RNA levels. ActD treatment led to the formation of DDX3X and DDX3Y puncta inside the nucleus, consistent with an RNA-dependent LLPS buffering mechanism (Figure S3D). Deleting any individual domain in DDX3X and DDX3Y dramatically decreased SG area compared to the full-length (Figures 3A, 3B, S3A, and S3B). Consistent with previous results, IDR1 of either DDX3X or DDX3Y as a standalone protein entered SGs and colocalized with G3BP1. Furthermore, a noticeable fraction of IDR1 from either protein remained diffuse throughout the cytoplasm upon arsenite treatment, indicating that IDR1 alone is less prone to enter SGs. For DDX3X, the deletion of IDR2 (DDX3XΔIDR2) significantly decreased the total SG area of DDX3XΔIDR2-positive SGs per cell (by ~1.4 fold) compared to the wild-type-DDX3X SGs. Deletion of the helicase domains (both DEAD1 and DEAD2) from DDX3X also decreased the total area of DDX3XΔhelicase-SGs (~1.2 fold) per cell, but to a lesser extent than the deletion of IDR1 or IDR2. Similarly, for DDX3Y, deletion of IDR2 or the helicase domains both lead to a ~1.2-fold decrease in the size of DDX3YΔIDR2-positive or DDX3YΔhelicase-positive SGs compared to the full-length DDX3Y SGs. Notably, the total SG area for each DDX3Y truncation was still larger than their DDX3X truncation counterparts (~1.3 to 1.85-fold), suggesting that subtle sequence variation throughout the protein contributes to the distinct behaviors of DDX3X and DDX3Y (Figure 3A and 3B). When we swapped IDRs between DDX3X and DDX3Y, we observed that any domain-swapped variants containing IDR1 of DDX3Y formed significantly larger SGs than the hybrid variants that did not contain IDR1 of DDX3Y, whereas swapping helicase domains or IDR2 did not significantly affect SG sizes (Figures 3C, 3D, S3E, and S3F). When we performed in vitro droplet formation assays with wild-type and domain-swapped variants (which were purified to a similar level of homogeneity, Figure S3G and S3H), our results recapitulated the finding that any domain-swapped variants that contained IDR1 of DDX3Y more readily phase-separated in the presence of RNA than variants which did not contain IDR1 of DDX3Y (Figure 3E and 3F). Our results also suggest that the IDR2 and helicase domains may also contribute to the differences in propensity to undergo LLPS, as the XIDR1YhelicaseYIDR2 variant more readily formed droplets compared to wild-type DDX3X, XIDR1YhelicaseXIDR2, and XIDR1XhelicaseYIDR2 (Figure 3D and 3F). To dissect the contributions of RNA binding and IDRs in phase separation of the proteins, we next performed the in vitro droplet assays with wild-type and domain-swapped variants in the absence of poly(U)-RNA. As shown in Figure S3I, proteins containing IDR1 of DDX3Y formed larger droplets than proteins containing IDR1 of DDX3X. These results suggest that although each of the domains in DDX3 proteins plays a distinct role in facilitating LLPS and in their accumulation into SGs, the sequence of IDR1 is a significant factor in determining the relative droplet sizes. The weaker ATPase activity of DDX3Y compared to DDX3X weakens its condensate disassembly DDX3X and DDX3Y harness the energy of ATP hydrolysis to control RNA binding, unwinding, and release (Hondele et al., 2019). Because LLPS droplets assemble through many multivalent, weak interactions between RNAs and proteins (Wang et al., 2018a), we stimulated the ATPase activity of DDX3X and DDX3Y to see if this could induce condensation disassembly by affecting RNA-protein interactions. As shown in Figure 4A and 4B, the addition of 4 mM ATP to protein-RNA droplets induced the disassembly of both DDX3X and DDX3Y condensates. To control for the hydrotropic effect of ATP (Patel et al., 2017) in dissolving the droplets, we repeated the droplet assay using UTP (which is not hydrolyzed by DDX3 and thus could only act as a hydrotrope in this context (Patel et al., 2017)). The addition of UTP had no significant effect on the droplets formed by DDX3X and DDX3Y with RNA (Figure 4A and 4B). These results suggest that the ATPase activity of DDX3X and DDX3Y is a major factor in dissolving DDX3-RNA condensation, possibly by breaking the multivalent protein-RNA interactions. We next studied whether there are intrinsic differences in ATPase activity between DDX3X and DDX3Y, as this could explain the differences observed in their condensate disassembly in response to ATP. Thus, we measured the ATPase activities of both DDX3X and DDX3Y using a malachite green ATPase assay. In this assay, we used MBP-tagged proteins purified to near homogeneity (Figure S4A and S4B). Both of the MBP-tagged DDX3X and DDX3Y displayed the same melting temperatures in differential scanning fluorimetry measurements, and the melting curves are indicative of well-folded proteins devoid of aggregates (Figure S4C and S4D) (Gao et al., 2020). To ensure that any potential RNA carry-over would not confound our measurements, we performed the malachite green ATPase assays in the absence of additional RNA or with 100 ng/μL total HeLa RNA. In reactions with no added RNA but with 2 mM ATP, DDX3X and DDX3Y reactions had nearly identical amounts of free phosphate after a 30-minute reaction time (Figure S4E). This phosphate is likely due to spontaneous hydrolysis of ATP, as it is also present in the controls with buffer only. When RNA was added, DDX3X reactions produced more free phosphates (indicative of more ATP hydrolysis) than DDX3Y reactions. The malachite green ATPase assays were also performed using RNase treated and untreated MBP-tagged DDX3X and DDX3Y respectively. As shown in Figure S4F, RNase treatment has no noticeable effects on the ATPase activities. Collectively, these results suggest that DDX3X is a more robust ATPase than DDX3Y, and that the possible trace amount of RNA carry-over in the purified MBP-tagged DDX3X and DDX3Y was not responsible for the ATPase activity differences. To obtain a more complete picture of the ATPase differences between DDX3X and DDX3Y, we performed a continuous ATPase assay as previously reported using the MBP-tagged DDX3X and DDX3Y constructs (Song and Ji, 2019). DDX3X reached a Vmax of 7.9 μM/min while DDX3Y only achieved a Vmax of 5.9 μM/min (Figure 4C). Both DDX3X and DDX3Y hydrolyzed ATP cooperatively (Hill coefficient ≈ 2), as has been previously reported for DDX3X (Song and Ji, 2019). While the ATPase activity of DDX3Y has not been previously investigated, our Vmax values for the full-length DDX3X were higher than those reported for a truncation lacking both IDRs (amino acids 132 – 607) (7.9 vs. 3.1 μM/min, 2.54-fold higher) (Song and Ji, 2019). Our data suggest that ATP hydrolysis activity of DDX3Y is significantly slower than that of DDX3X (Figure 4C). DDX3X is a non-processive helicase: it binds its dsRNA substrate, binds ATP, unwinds approximately 13 – 19 bp of dsRNA, and releases the two RNA strands upon ATP hydrolysis (Song and Ji, 2019). Given that ATP hydrolysis is a crucial step in this catalytic cycle and that our ATPase data suggest that DDX3X is a more efficient ATPase than DDX3Y, we next investigated how the differences in ATPase activities may affect the dynamics of these enzymes. To this end, we employed smFRET assays with MBP-DDX3X and MBP-DDX3Y to compare their RNA binding abilities and the dynamics of their ATP-dependent interactions with RNA as described previously (Figure 4D) (Kim and Myong, 2016). At protein concentrations of 2, 4, or 8 μM, DDX3X and DDX3Y tightly bound the RNA substrate, resulting in high FRET, with a peak FRET efficiency (E) of 0.81 – 0.85 (Figures 4E, S4G, and Table S1). This binding is specific to DDX3X and DDX3Y, as MBP (a non-RNA interacting protein) gave low FRET, like RNA alone (Figure S4H and S4I). The addition of ATP to DDX3X and DDX3Y led to a low FRET population that was larger for DDX3X than it was for DDX3Y (Figures 4E, S4J, S4K, and Table S1). Of note, this low-FRET population of RNA molecules does not represent full strand separation of the RNA duplex because alternating direct illumination of the Cy3 and Alexa Fluor 647 probes (lowest traces in Figure S4J and S4K) showed that both labeled strands are present in each individual high- and low-FRET particles. This change in FRET from high to low efficiency likely represents partial unwinding activity. As shown in Figure 4F, the relative areas of the low FRET peak induced by addition of ATP increased along with increasing protein concentrations which exemplifies the cooperativity of the DDX3X- and DDX3Y-catalyzed reactions. Importantly, the smFRET histograms suggest that DDX3X had higher (partial) unwinding activity than DDX3Y, consistent with the kinetics data (Figure 4C). DDX3X also showed a larger proportion of dynamic FRET recordings upon addition of ATP than DDX3Y (Figure S4L). Together, the kinetics and smFRET data support the conclusion that DDX3Y has slower ATPase and (partial) unwinding activities, leading to the less dynamic characteristics of RNA-DDX3Y complexes. Decreased dynamics may contribute to the weaker disassembly of DDX3Y condensates compared to those of DDX3X, which in turn contributes to the observation that DDX3Y condensates persist in the presence of ATP to a greater degree than DDX3X condensates (Figure 4A). Translation is differentially modulated by DDX3X and DDX3Y Given that SGs are formed concomitant with translational repression and are thought to harbor proteins that regulate translation (Kimball et al., 2003), we employed a reticulocyte in vitro translation assay to investigate how phase-separated DDX3X and DDX3Y influence the translation of a luciferase reporter (Figure 5A). Upon titration of 4 – 10 μM mCherry-DDX3X or mCherry-DDX3Y into the lysate, there was a dramatic, dose-dependent decrease of luciferase signal with increased concentrations of DDX3X or DDX3Y, indicating a decrease in the in vitro translation (Figure 5B). Indeed, we observed a concentration-dependent increase in protein condensation of DDX3X or DDX3Y within the lysate (Figure 5C and 5D). These results indicate that the condensation of DDX3X and DDX3Y is correlated with repressed mRNA translation. Of note, at each concentration tested, DDX3Y induced a more pronounced decrease in luciferase signal compared to DDX3X (Figure 5B). To ensure that the observed effects were specific to DDX3X and DDX3Y, we also titrated truncated versions of DDX3X and DDX3Y, which only contained the minimally active helicase domain (both RecA-like domains and the N- and C-terminal extensions) (Floor et al., 2016; Song and Ji, 2019) to serve as negative controls. As shown in Figure S5A, the luciferase signal remained steady across all concentrations of the truncated versions of DDX3X and DDX3Y, likely because these constructs lack IDRs. We next studied whether adding ATP could alleviate translation repression by disassembling the DDX3X and DDX3Y condensates in the in vitro translation assays. To this end, the assays were repeated with the addition of 1 mM ATP. Upon ATP addition, in vitro translation was partially restored in both DDX3X and DDX3Y reactions; however, DDX3Y reactions remained less translationally active than the corresponding DDX3X reactions (Figure S5B). These findings support the notion that the decreased ATPase activity of DDX3Y decelerates the dispersal of DDX3Y-containing condensates, and thus ATP does not fully restore in vitro translation. Next, we investigated whether DDX3Y condensation had a stronger translation repression impact than DDX3Y condensation in cells using a puromycin incorporation assay in HeLa cells after transient transfection of FLAG-DDX3X or FLAG-DDX3Y with endogenous DDX3X depleted. Puromycin was incorporated into newly synthesized proteins, and, as shown in Figure 5E and 5F, the formation of DDX3X-positive SGs and DDX3Y-positive SGs significantly reduced puromycin signals compared to the neighboring cells which were not transfected and thus lacked DDX3X-positive or DDX3Y-positive SGs. Notably, the puromycin signals were weaker in cells with DDX3Y-positive SGs than in cells with DDX3X-positive SGs (Figure 5F). These results suggest a potential mechanism by which DDX3Y more effectively inhibit translation through its unique biophysical properties. Phase separation can inhibit translation by sequestering translational machinery and/or mRNA transcripts into the phase-separated compartment (Kim et al., 2019). DDX3Y-containing condensates and SGs are larger and less mobile (Figures 1 and 2), which may result in stronger translation inhibition than the smaller and more dynamic DDX3X condensates which exchange material more rapidly. APEX-seq captures the protein-RNA interaction patterns of DDX3X and DDX3Y SGs formed under different stress conditions contain distinct proteins and RNA constituents (Markmiller et al., 2018), which raises the possibility that SG contents vary based on the type of stress presented to a cell. Neither the RNA composition nor the differential effects DDX3X or DDX3Y may exert on resident SG RNAs is known. Because SGs are largely thought to be a mechanism of regulating mRNA metabolism (Buchan and Parker, 2009; Jain et al., 2016; Khong et al., 2017), we sought to define the RNA content of DDX3X- and DDX3Y-positive SGs. To this end, we employed an adapted ascorbate-peroxidase (APEX2)-based proximity labeling method (Fazal et al., 2019; Marmor-Kollet et al., 2020; Padron et al., 2019) (Figure S5C). First, to examine the specificity of APEX2-catalyzed RNA biotinylation, we generated a mito-APEX2 fusion protein consisting of APEX2 fused to a mitochondrial matrix localization signal (Figure S5D) (Fazal et al., 2019; Mercer et al., 2011). Using a previously described protocol (Fazal et al., 2019), followed by RT-qPCR analysis, we could reliably identify mitochondrial-specific RNAs (ND1 and ND2) labeled by mito-APEX2 (Figure S5E – S5G). Next, using this validated approach, we expressed APEX2-DDX3X, APEX2-DDX3Y, or APEX2-EGFP (control) in DDX3X knockdown HeLa cells. Cells were then treated with arsenite or DMSO for 1 hr to generate three experimental cell populations. We consistently observed that the total area of APEX2-DDX3Y-positive SGs was significantly larger (1.5-fold) than APEX2-DDX3X-positive SGs (Figure S5H and S5I). The results suggest that the fusion of APEX2 to DDX3X or DDX3Y did not significantly interfere with the ability of either protein to colocalize with SGs in cells upon arsenite treatment. The above cells were then incubated with biotin-phenol for 30 min, followed by H2O2 treatment for 1 min to activate the APEX2 enzyme and covalently link biotin to RNAs (Figure 5G). To confirm biotinylating, small aliquots of cell lysate for each condition were blotted using a streptavidin antibody; this antibody detected multiple protein bands (consistent with the previous results (Fazal et al., 2019)), confirming that the APEX2 enzyme was active (Figure S5J and S5K). Subsequently, we performed biotin pulldown and poly(dT) extraction to enrich biotinylated poly(A)-RNAs. Enriched RNAs were subjected to next-generation high-throughput RNA-seq. The RNA-seq data from the biological replicates in each group correlated well (Figure S5L). To determine whether DDX3X- and DDX3Y-positive SGs harbor unique mRNAs, we compared the levels of transcripts between APEX2-DDX3X or APEX2-DDX3Y libraries to APEX2-EGFP libraries. Any transcript for which expression in either the APEX2-DDX3X or APEX2-DDX3Y libraries was at least 2-fold higher than in the APEX2-EGFP dataset (log2 -fold enrichment > 1) was defined as “enriched.” We found that DDX3X-positive SGs enriched 562 RNAs (Figure 5H and Table S2), and Gene Ontology (GO) analysis indicated the encoded proteins were mainly involved in the regulation of glycolipid and nucleic acid metabolism (Figure 5I). DDX3Y-positive SGs enriched 1020 RNAs (Figure 5H); GO analysis suggested that some of the encoded proteins were also involved in the regulation of glycolipid metabolism, while others play roles in transcriptional regulation (Figure 5I). Interestingly, while there was a large pool of RNAs enriched in both DDX3X- and DDX3Y-positive SGs (where 61% of DDX3X- and 34% of DDX3Y-positive SG enriched RNAs are shared), there are also RNA targets that were specific to each helicase (Figure 5J). In the absence of arsenite treatment, there was a small number of transcripts with fold change log2 > 1 or log2 < −1 in APEX2-DDX3X or APEX2-DDX3Y compared to APEX2-EGFP (Figure S5H and S5M). We validated a list of RNA targets specifically enriched by either APEX2-DDX3X or APEX2-DDX3Y, ranging from lower to upper enrichment, using RT-qPCR (Figure 5K). These results indicate that DDX3X- and DDX3Y-positive SGs regulate distinct mRNA targets. However, DDX3X and DDX3Y may act on shared transcripts in divergent ways as these two enzymes inhibit translation to different degrees (Figure 5A – 5F). DDX3X and DDX3Y co-phase separate into SGs While our study up to this point has focused on either DDX3X or DDX3Y individually, XY individuals express both proteins simultaneously (Cotton et al., 2015; Ditton et al., 2004; Godfrey et al., 2020). To examine whether DDX3X and DDX3Y go into the same SGs, we expressed DDX3X and DDX3Y together in cells with endogenous DDX3X transiently knocked down. We found that all the antibodies we tested were highly cross-reactive (Figure S6A), likely due to the high similarity of these two proteins. Thus, we used proteins with different tags (FLAG vs. HA), and tagged homologs were expressed to a similar extent (Figure S6B). Given that SG size and composition are sensitive to different types of stressors (Fujimura et al., 2012; Markmiller et al., 2018; Saito et al., 2019; Szaflarski et al., 2016), we studied SGs triggered by a range of stressors, including energy depletion (CCCP), osmotic stress (sorbitol), translation inhibition (puromycin), proteasome inhibition (MG132), and ER stress (thapsigargin). Under these conditions, MG132 and thapsigargin did not induce SG formation (Figure S6C). However, when CCCP, sorbitol, or puromycin were used to stress cells, DDX3X, DDX3Y, and G3BP1 colocalize, although the fluorescence intensity of FLAG-DDX3X was much lower than HA-DDX3Y (Figure 6A). To ensure that our observations were not due to differential recognition by the anti-FLAG and anti-HA antibodies, we repeated the experiments using HA-DDX3X and FLAG-DDX3Y. We observed the same lower intensity of DDX3X, even though total protein levels for each DDX3 homolog were similar and were not affected by either tag (Figures S6D). These results suggest that, while DDX3X and DDX3Y phase separate to SGs together, DDX3Y has a stronger propensity to go into SGs than DDX3X, in line with our previous observations. Given that DDX3X and DDX3Y can co-phase separate into SGs, we wondered how mixtures of DDX3X and DDX3Y affected translation compared to DDX3X or DDX3Y alone. As shown in Figure 6B, a 3:1 ratio of DDX3X (6 μM) and DDX3Y (2 μM), which is close to the physiological ratios between DDX3X and DDX3Y (Godfrey et al., 2020), led to a 1.4-fold more robust repression of translation compared to DDX3X (8 μM) alone, and a 3.6-fold more robust repression of translation compared to a mixture of DDX3X (6 μM) and MBP (2 μM). Moreover, a mixture of DDX3X (2 μM) and DDX3Y (6 μM) less efficiently repressed translation relative to DDX3Y (8 μM) alone. Furthermore, the in vitro LLPS assays were performed with different ratios of DDX3X and DDX3Y in the presence of RNA. The addition of DDX3Y to DDX3X stimulated droplet formation, and condensation was positively correlated with the relative amount of DDX3Y (Figure 6C). DDX3Y more strongly promotes FUS aggregation than DDX3X Dysregulation of SGs can promote FUS aggregation, leading to cell death (Bentmann et al., 2012; Guo et al., 2018; Kamelgarn et al., 2016; Silva et al., 2019). Thus, we studied how DDX3X and DDX3Y influence in vitro FUS fibrillization. When we added 1 μM wild-type FUS to either DDX3X or DDX3Y, we saw that the aggregation of FUS, detected by light scattering at 395 nm, was enhanced by both helicases but was more extensively aggravated in the presence of DDX3Y (Figure S6E). To quantify the effect of DDX3X and DDX3Y on FUS aggregation over time, we measured the area under the curve (AUC) for each scattering time course. Both DDX3X and DDX3Y enhance FUS aggregation but that DDX3Y had a stronger enhancement, even at high concentrations of either protein (Figures 6D and S6E). We also studied the colocalization of FUS and DDX3X or DDX3Y in HeLa cells upon arsenite treatment. We constructed a DOX-inducible stable cell line that expressed FUS. As shown in Figure 6E – 6F, while wild-type FUS was mainly located in the nucleus with transfection of empty vector, a portion of FUS formed puncta which colocalize with DDX3X-positive granules and DDX3Y-positive granules in the cytoplasm with transfection of DDX3X and DDX3Y expression plasmids. Furthermore, DDX3Y-FUS puncta were significantly larger than DDX3X-FUS puncta (Figure 6G). As XY individuals are at a higher risk for developing ALS (Manjaly et al., 2010), these data not only suggest that DDX3X and DDX3Y affect FUS aggregation but also suggest that the enhanced propensity to promote LLPS by DDX3Y might lead to a XY-specific increase of FUS aggregation. DDX3Y more strongly accelerates TDP-43 aggregation than DDX3X Finally, we tested whether DDX3X or DDX3Y might also stimulate the aggregation of TDP-43, another prominent RNA-binding protein connected to ALS and FTD (Portz et al., 2021; Tan et al., 2017). We found that DDX3X or DDX3Y did not enhance the total amount of TDP-43 aggregation (Figure S6F and S6G). However, both helicases accelerated TDP-43 aggregation (Figure S6F and S6H). The halftime t1/2 (i.e. the time at which 50% TDP-43 aggregation had occurred) was reduced from ~6.9 hrs in the absence of helicase to ~5.7 hrs in the presence of DDX3X and ~ 4.4 hrs in the presence of DDX3Y (Figure S6F and S6H). These findings suggest that DDX3Y can accelerate TDP-43 aggregation more potently than DDX3X. DISCUSSION Herein, we report the first comparative study of the sexually dimorphic RNA helicases DDX3X and DDX3Y in the regulation of translation via distinct phase separation behaviors (Figure 6H). Importantly, we reveal the molecular mechanism underpinning the higher propensity of DDX3Y to phase separate and its lower propensity to disassemble once condensed. Firstly, we find that the condensation propensity differences of DDX3X and DDX3Y are most likely due to the sequence composition of both IDR1s. Overall, the percentages of both negatively and positively charged amino acids in YIDR1 (17.3% and 16.1% respectively) are higher than in XIDR1(16.7% and 14.9% respectively), whereas the percentage of charged amino acids is similar between the other domains of DDX3X and DDX3Y (Figure S6I). This finding suggests that YIDR1 can form more charge-charge interactions to support phase separation. In addition to these electrostatic interactions, cation-π and π-π (Vernon et al., 2018) interactions are known to facilitate LLPS (Qamar et al., 2018). As such, Tyr to Phe and Arg to Lys mutations in IDRs dramatically impair phase separation (Schuster et al., 2020). In line with this, Phe84 and Lys118 in XIDR1 correspond to Tyr83 and Arg116 in YIDR1, suggesting that YIDR1 is capable of more cation-π and π-π interactions than XIDR1, which might result in the stronger phase separation of DDX3Y found in these studies (Figure S1A). Indeed, Lys118 in DDX3X is a known site of post-translational acetylation (Saito et al., 2019). Acetylation at this site decreases the phase separation of DDX3X in vitro and inside of cells through the disruption of cation-π interactions. Given that the analogous position in DDX3Y is an arginine (Arg116), which is often thought of as a “non-acetyl” mimetic, this position is likely a key source of difference between DDX3X and DDX3Y and merits future study. Second, our data suggest that the distinct dynamics of DDX3X- and DDX3Y-positive SGs may also be related to differences in their ATPase-driven SG remodeling activity (Jain et al., 2016; Tauber et al., 2020) (Figure 4). Furthermore, the differences in dynamics of DDX3X and DDX3Y condensates may also explain why DDX3Y more strongly repressed translation than DDX3X. DDX3X-positive SGs and DDX3Y-positive SGs sequester distinct mRNAs in addition to a shared pool of transcripts, suggesting that the differences in ATPase activity and translational repression may have functional consequences, especially under stress (Figure 5G – 5K). The results shared here suggest that DDX3X and DDX3Y might influence the translation of the overlapping mRNA targets to different degrees in addition to exerting differential regulation of distinct RNA components. They might also differentially sequester other translational components. One important question that remains to be answered is the degree to which DDX3X and DDX3Y overlap functionally. Studies by others have suggested that DDX3X and DDX3Y are redundant in protein synthesis under unstressed conditions (Venkataramanan et al., 2021). Our data suggest that their divergent roles may not be apparent until they are driven to phase separate during the stress response. Stress leads to various human disorders, many of which display sex-biased features. For instance, ALS is ~20% more common in males than females (Manjaly et al., 2010). We showed that DDX3Y more strongly promotes FUS self-assembly in vitro and formed larger FUS granules in cells (Figure 6D – 6G), possibly through its stronger phase separation propensity compared to DDX3X. Furthermore, DDX3Y more strongly accelerated TDP-43 aggregation than DDX3X (Figure S6F – S6H). Other known SG resident proteins such as TIA-1, hnRNPA1, and hnRNPA2 have been previously indicated in ALS (Fernandes et al., 2020; Gilks et al., 2004; Harrison and Shorter, 2017; Khalfallah et al., 2018; Molliex et al., 2015). Future investigation on the scope and extent of DDX3X’s and DDX3Y’s impact on these proteins in sex specificity in neurodegenerative diseases will be exciting. Limitations of the Study HeLa cells may not be perfectly suited to the task of uncovering the biological targets in DDX3Y-specific SGs, because they lack a Y chromosome and thus may also lack some of the specific transcripts that may be targeted by DDX3Y. However, we feel that reconstituting HeLa cells (with endogenous DDX3X depleted) with ectopically expressed DDX3X or DDX3Y provided a clean system to reveal the different RNA sequestration impacts of DDX3X and DDX3Y due to their intrinsic differences in phase separation. Thus, we believe our findings are an excellent proof of principle. Additionally, while our recombinant proteins were purified to apparent homogeneity and are not expected to have unequal carryover of RNA, there could be some effect of potential impurities on the magnitude of the differences we measure. However, we believe that the difference itself is a true observation about these proteins. The apparent unwinding activity observed in the smFRET assays is not an indication of complete unwinding by DDX3X/DDX3Y and this assay is not a conventional assay to study helicase activity. Both the donor and acceptor fluorophores being present in all molecules compiled in the smFRET analysis indicates only partial unwinding of the RNA substrate. We believe our smFRET assay is mainly reflecting the ATPase and dynamic activity of the proteins with RNA and give insight into the differences between DDX3X and DDX3Y in these features. Our experiments should serve as a crucial first step towards understanding the role of sexually dimorphic proteins in disease. STAR METHODS RESOURCE AVAILABILITY Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Kathy Fange Liu (liufg@pennmedicine.upenn.edu). Materials availability All materials generated in this study are available on request to Lead Contact. Data and code availability RNA-seq data have been deposited at GEO and are publicly available as of the date of publication. Accession numbers are listed in the key resources table. Original western blot images have been deposited at Mendeley and are publicly available as of the date of publication. The DOI is listed in the key resources table. Microscopy data reposted in this paper will be shared by the lead contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. EXPERIMENTAL MODEL AND SUBJECT DETAILS Cell culture, transfection and Escherichia coli strains HeLa, HEK293T, and N2a cells were cultured in DMEM + GlutaMAX (GIBCO) with 10% FBS (GIBCO) and 1% Pen/Strep (Corning) in a humidified incubator with 5% CO2 at 37°C. For bacterial cell culture, TOP10 and BL21(DE3)-RIL chemically competent bacterial strains grew in lysogeny broth containing the corresponding antibiotics at 200 rpm, 37°C. The negative control siRNA from Ambion (AM4611) was used as a control siRNA in the knockdown experiments. DDX3X siRNA was purchased from Ambion (Assay ID 145803). Lipofectamine 2000 (Invitrogen) and Lipofectamine RNAiMax (Invitrogen) from Invitrogen were used for plasmids and siRNA transfection, respectively. It took 48 hrs for siRNA knockdown and 24 hrs for plasmid expression. METHOD DETAILS Constructs For recombinant MBP-DDX3X and MBP-DDX3Y protein expression: DNA fragments encoding human DDX3X were PCR-amplified from the HeLa cDNA library, and the human DDX3Y coding sequence was PCR-amplified from pCMV6-DDX3Y (Origene RC226072). These DNA fragments were then inserted into the pMAL-c2x vector (Walker et al., 2010) (Addgene Plasmid #75286) using restriction enzymes BamHI and SalI. A TEV enzyme digestion sequence (protein sequence ENLYFQG; DNA sequence GAAAACCTGTACTTCCAGGGA) was added to the forward primers, and a His-tag sequence (protein sequence HHHHHH; DNA sequence CATCATCACCATCACCAC) was added to the reverse primers. For in vitro LLPS experiments, the pETMCN_His-TEV_V5-DDX3Y-mCherry construct was made by swapping the DDX3Y with the DDX3X in the pETMCN_His-TEV_V5-DDX3X-mCherry (Hondele et al., 2019) using NdeI and BamHI restriction enzymes. For expression of DDX3X and DDX3Y in mammalian cells, DNA fragments encoding the full-length DDX3X and full-length DDX3Y were inserted into the pPB vector with FLAG (DYKDDDDK) and HA (YPYDVPDYA) tags before the N-terminus of the proteins using MfeI and SalI (XhoI was used for digesting the pPB vector) restriction enzymes. The plasmids expressing FLAG or HA single-tagged DDX3X and DDX3Y were made by inserting the DNA fragments encoding full-length DDX3X or full-length DDX3Y into the modified pcDNA3 vector with FLAG or HA tag in frame at the N-terminus. For the truncation variants of DDX3X and DDX3Y, the following truncations were put into the pPB vector (domain prediction based on PONDR (Xue et al., 2010)): IDR1 of DDX3X (amino acids 1–168); IDR1 of DDX3Y (amino acids 1 – 164); IDR2 of DDX3X (amino acids 580 – 662); IDR2 of DDX3Y (amino acids 579 – 660); helicase domain of DDX3X (amino acids 169 – 579); helicase domain of DDX3Y (amino acids 165 – 578); ΔIDR1 truncation variant of DDX3X (amino acid 169 – 662); ΔIDR1 truncation variant of DDX3Y (amino acids 165 – 660); ΔIDR2 truncation variant of DDX3X (amino acids 1 – 579); ΔIDR2 truncation variant of DDX3Y (amino acids 1 – 578); ΔHelicase truncation variant of DDX3X (amino acids Δ169 – 579); ΔHelicase truncation variant of DDX3Y (amino acids Δ165 – 578). For domain-swap variants of DDX3X and DDX3Y, the following constructs were inserted into the pPB vector: DDX3XIDR1-DDX3YHelicae-DDX3YIDR2; DDX3YIDR1-DDX3XHelicae-DDX3XIDR2; DDX3XIDR1-DDX3Xhelicae-DDX3YIDR2; DDX3YIDR1-DDX3Yhelicae-DDX3XIDR2; DDX3XIDR1-DDX3Yhelicae-DDX3XIDR2; DDX3YIDR1-DDX3Xhelicae-DDX3YIDR2. DDX3X-EGFP, DDX3X-mCherry, DDX3Y-EGFP, and DDX3Y-mCherry with a linker sequence (amino acids GlyGlySerGly) inserted between DDX3X/DDX3Y and EGFP/mCherry were inserted into the pPB vector using MfeI and SalI restriction enzymes. Those domain-swap variants were cloned into the pETMCN_His-TEV_V5-mCherry vector to express mCherry-tagged proteins in E.coli as well. For APEX2-seq experiments, the DNA fragment encoding APEX2 was PCR amplified from pcDNA5/FRT/TO APEX2-GFP (Addgene, 129640) and fused to DDX3X and DDX3Y using fusion PCR. APEX2-DDX3X and APEX2-DDX3Y were inserted into the pPB vector using MfeI and Sal I restriction enzymes, and APEX2-EGFP was inserted into the pPB vector using MfeI and XhoI restriction enzymes. To validate the biotin labeling efficiency of APEX2, the plasmid pPB mito-APEX2 was constructed. The DNA fragment encoding a mitochondria matrix localization sequence (amino acids MLATRVFSLVGKRAISTSVCVRAH) derived from COX4 was added to the APEX2 forward primer for the PCR reaction. The PCR product was subsequently inserted into the pcDNA3 vector by using BamHI and XhoI restriction enzymes. All the sequences of the primers used for these clones are summarized in Table S3, and each plasmid was validated by Sanger sequencing. Protein purification The pMAL-c2X-DDX3X, pMAL-c2X-DDX3Y plasmids were transformed into Escherichia coli strain BL21-RIL to express the MBP-tagged recombinant proteins. The pETMCN_His-TEV_V5-DDX3X-mCherry (Hondele et al., 2019), pETMCN_His-TEV_V5-DDX3Y-mCherry and other domain-swap variants in the pETMCN_His-TEV_V5-mCherry vectors were transformed into E. coli strain BL21-RIL to express the mCherry-tagged recombinant proteins. The bacteria were cultured in lysogeny broth at 37°C till OD600 nm = 0.8 before administration of 1 mM IPTG at 16°C for 16 hrs. The pellets from 2 L bacterial culture were resuspended with 80 mL binding buffer (25 mM Tris-HCl, pH 7.5, 500 mM NaCl) and sonicated. After centrifuging at 12,000 rpm for 30 min to remove the cell debris, the supernatant was loaded to a Ni-NTA column. Next, 10 column volumes of the binding buffer supplemented with 50 mM imidazole was used as buffer A to wash away the non-specific binding proteins. Another 10 column volumes of a high salt buffer (25 mM Tris-HCl, pH 7.5, and 2 M NaCl) was used to decrease the amount of bound RNAs from DDX3X or DDX3Y. Finally, four column volumes of the binding buffer supplemented with 500 mM imidazole was used to elute the bound proteins. The DDX3X-mCherry and DDX3Y-mCherry recombinant proteins were dialyzed into the storage buffer (25 mM Tris-HCl, pH 7.5, 500 mM NaCl, 10% glycerol, and 2 mM DTT) and the His-tag was cleaved using TEV enzyme simultaneously with the dialysis. Then, the proteins were concentrated using Amicon Ultra-15 (Millipore) tubes before loading to a Superdex 200 column, with buffer (25 mM Tris-HCl, pH 7.5, and 500 mM NaCl) for mCherry tagged proteins and buffer (25 mM Tris-HCl, pH 7.5, 200 mM NaCl) for MBP tagged proteins. MBP-tagged proteins were purified at 4°C, while mCherry-tagged proteins were purified at room temperature. The purity of the proteins was analyzed by SDS-PAGE. Purified proteins were aliquoted, snap-frozen in liquid nitrogen and stored at −80°C. Once thawed, aliquots were never refrozen. GST-TEV-FUSWT was purified as described (Sun et al., 2011). Briefly, E. coli cells were lysed by sonication on ice in PBS with protease inhibitors (cOmplete, EDTA-free, Roche Applied Science). The protein was purified over Glutathione Sepharose 4 Fast Flow (GE Healthcare) and eluted from the beads using 50 mM Tris-HCl pH 8, 20 mM trehalose, and 20 mM glutathione. Purified protein was snap-frozen in liquid nitrogen and stored at −80°C. pJ4M/TDP-43 was a gift from Nicolas Fawzi (Addgene plasmid # 104480). TDP-43 was purified as previously described (Hallegger et al., 2021). Briefly, the plasmid was transformed into BL21(DE3) RIL E. coli. Cells were harvested by centrifugation and lysed by lysozyme (1 mg/mL) and sonication in wash buffer (20 mM Tris-HCl pH 8.0, 1 M NaCl, 10 mM imidazole, 10% glycerol, 1 mM DTT, 5 μM Pepstatin A, 100 μM PMSF, and cOmplete, EDTA-free, Roche Applied Science protease inhibitors). The protein was purified over Ni-NTA agarose beads (QIAGEN) and eluted from the beads using elution buffer (wash buffer except with 300 mM imidazole rather than 10 mM imidazole). The protein was further purified over amylose resin (NEB) and eluted with 20 mM Tris-HCl pH 8.0, 1 M NaCl, 10 mM imidazole, 10% glycerol, 1 mM DTT, 5 μM Pepstatin A, 100 μM PMSF, and 10 mM maltose. The protein was concentrated, snap-frozen in liquid nitrogen, and stored at −80°C. Differential scanning fluorimetry assay Purified MBP-DDX3X and MBP-DDX3Y proteins were diluted to 0.25 mg/mL. 19 μL of each protein was transferred to a well of a 384-well plate, and 1 μL of 5-fold SYPRO Orange (Thermo Fisher) was added to each well. The plate was sealed and spun at 3,600 g for 2 min. The fluorescent signal at 570 nm was collected using a RT-qPCR machine with the temperature ramping from 20 to 95°C. The data were analyzed using DSF World (Wu et al., 2020). In vitro LLPS assay The in vitro LLPS was set up at room temperature with total volume of 20 μL in PCR tube. Proteins were diluted to 100 μM (DDX3X-mCherry and DDX3Y-mCherry) or 75 μM (DDX3X-mCherry, DDX3Y-mCherry and all the domain-swap mutants) using the storage buffer. 10 μL buffer (25 mM Tris-HCl, pH 7.5, 500 mM NaCl) and 2 μL diluted protein were transferred to the PCR tube. Then, 8 μL water was added to the tube to observe the LLPS of protein alone; 2 μL polyU-RNA (2 mg/mL, dissolved in water) and 6 μL water were added to the tube to observe the LLPS of protein with RNA; 2 μL polyU-RNA (2 mg/mL, dissolved in water), 2 μL ATP buffer (40 mM ATP and 50 mM MgCl2) or UTP buffer (40 mM UTP and 50 mM MgCl2) and 4 μL water were added to the tube to observe the LLPS of protein with RNA and ATP or UTP. To observe the LLPS of different combinations of DDX3X-mCherry and DDX3Y-mCherry, the proteins were diluted to the proper concentrations. Then 10 μL buffer, 2 μL diluted protein and 8 μL water were transferred to a PCR tube. The mixtures were mixed by pipetting and transferred to 384-well glass-bottomed plate. After incubating at room temperature for 1 hr, the plate was spun at 100 g for 1 min. Then the images were taken using a Zeiss LSM 880 confocal microscope under a 63 × oil lens. Immunofluorescence cell staining Cells were passaged to a 6-well plate with a coverslip in each well and cultured overnight. The cells were washed once in PBS and then fixed using 4% paraformaldehyde in PBST (PBS with 0.05% Tween-20) at room temperature for 15 min. Then, the cells were washed twice by PBST and permeabilized by 0.5% Triton at room temperature for 20 min. After being washed once by PBST, the cells were blocked with 1% BSA in PBST at room temperature for 30 min. Then, the blocking solution was replaced with 1 mL blocking solution supplemented with desired primary antibodies (at 1:1000 dilution) and incubated at room temperature for 1 hr or 4 °C overnight. After 4 washes with PBST, the corresponding Alexa Fluor-conjugated secondary antibodies were applied (1:1,000 diluted in the blocking solution) and incubated at room temperature for 1 hr. After being washed three times by PBST, the cells were incubated with 0.5 μg/mL DAPI for 1 min. After 4 times PBST washes, an antifade reagent (Invitrogen) was used to mount the slides. The images were taken using a Leica TCS SP8 confocal microscope. The “analyze particles tool” in Fiji (Schindelin et al., 2012) was utilized to quantify the sizes of the stress granules in mammalian cells in about 50 different cells per condition. The sizes of the stress granule in cells were analyzed by two researchers independently. Fluorescence recovery after photobleaching The FRAP assays were conducted using the bleaching module of the Zeiss LSM 880 confocal microscope for DDX3X and DDX3Y droplets individually. The 488 nm laser was used to bleach the EGFP signal, and the 561 nm laser was used to bleach the mCherry signal. Bleaching was focused on a circular region of interest (ROI) using 100% laser power, and time-lapse images were collected afterward. A same-sized circular area away from the bleaching point was selected as an unbleached control. The fluorescence intensity was directly measured in the Zen software. The values were reported as relative to pre-bleaching time points. GraphPad Prism was used to plot the data. The halftime for each replicate was calculated using the following formula: y=a•(1-exp(−b•x)) + c, in which a is the slow recovery fraction, c is the rapid diffusion fraction, and b is the recovery rate. The halftime is ln2 / b, and a mobile fraction is a + c. The two-tail t-test was used to calculate the p-values. For FRAP of the live cells, cells expressing DDX3X-EGFP or DDX3Y-EGFP were cultured in 35 mm poly-D-lysine coated glass-bottomed dishes (Mattek). Before taking the images, cells were treated with 500 μM sodium arsenite for 1 hr in the FluoroBrite DMEM medium with 10% FBS. A 20 × lens was used at zoom scale 6. For FRAP of in vitro LLPS, a 63 × oil lens was used. Formation of stress granules in cells Cells were seeded in a 6-well plate and cultured overnight. The following stressors: 500 μM sodium arsenite (1 hr), 1 M sorbitol (1 hr), 50 μM thapsigargin (1 hr), 40 μg/mL puromycin (3 hrs), and 10 μM MG132 (3 hrs) were added to the cell culture media (DMEM + 10% FBS) each in a separate well. For the stress condition of 60 μM CCCP, CCCP was added to glucose-free DMEM with 10% FBS. After the treatment, cells were fixed and subjected to the immunofluorescence imaging procedure detailed above. Sequence alignment The amino acid sequences of DDX3X and DDX3Y were downloaded from CCDS Database and aligned using Clustal Omega. The alignment results were redrawn using ESPript 3.0 (Robert and Gouet, 2014). The sequences were used to predict the natural disordered regions by PONDR (Xue et al., 2010), prion-like amino acid regions by PLAAC (Lancaster et al., 2014) and LLPS propensity by catGRANULE (Mitchell et al., 2013). Turbidity assay DDX3X-mCherry and DDX3Y-mCherry proteins were diluted to 10, 20, 30, 50, 75, 100, 125 and 150 μM using the storage buffer. 15 μL buffer (25 mM Tris-HCl, pH 7.5, 200 mM NaCl), 3 μL diluted protein were transferred to the PCR tube, 3 μL polyU-RNA (2 mg/mL, dissolved in water), and 9 μL water were mixed in a PCR tube. After incubating for 20 min at room temperature, the mixtures were transferred to a 384-well black plate with a clear flat bottom. The turbidity was measured by using a Tecan plate reader at OD600 nm. Then, the solution was transferred to a clean Eppendorf tube, and spun at 16,000 g for 2 min. The supernatant was used to measure protein concentration using Bradford method. Construction of DOX-inducible cell lines 5 μg of the lentiviral vectors expressing mClover3-FUS, 2.5 μg VSVG plasmid, and 3.75 μg pPAX2 plasmid were co-transfected to 100% confluent HEK 293T cells in 6-well plate with 40 μL polyethylenimine. The medium was changed after 6 to 8 hrs. 500 μL medium containing the virus was collected twice a day and 500 μL fresh medium was replenished at each time. The virus was collected in a total three-day period, and then spun down at 3000 rpm for 5 min at room temperature. The HeLa cells were infected with lentivirus with 8 μg/mL polybrene. After induction with 100 ng/mL of doxycycline (DOX) for two days, the cells were subjected to cell sorting. Western blots were performed to validate the expression of mClover3-FUS by using anti-GFP antibody. Colocalization of DDX3X/Y with FUS in cells The DOX-inducible HeLa cells expressing mClover3-FUS were seeded in a 6-well plate with 100 ng/mL of DOX and a coverslip in the well. 1 μg of empty vector, DDX3X-pPB, or DDX3Y-pPB were transfected to those cells respectively using lipofectamine 2000. 24 hrs later, sodium arsenite was used to each well at a final concentration of 500 μM for 1 hr. The cells were then subjected to the immunofluorescence cell stain protocol as discussed above. Cycloheximide chase assay The cycloheximide (CHX) chase assay was performed to explore the half-life of DDX3X and DDX3Y proteins in cells. HeLa cells were seeded in 6-well plate and transfected with 1 μg of DDX3X-pPB and DDX3Y-pPB respectively. 24 hrs later, CHX was added to the cells at a final concentration of 100 μg/mL and incubated for different time intervals. Then, the cells were collected and lysed with RIPA buffer (25 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, and 1% NP-40). The protein concentrations were measured using a Bradford assay and the same amount of cell lysate was used in Western blot analyses. The intensity for each band was quantified using Fiji. The intensity for each DDX3X and DDX3Y band was normalized to the corresponding GAPDH intensity firstly, and then for each replicate at different time points were normalized to time 0 hr. The data were plotted in Prism and an exponential decay formula was used to determine the half-life of DDX3X and DDX3Y. The degradation rate Kdecay was estimated by ln(At/A0) = − Kdecay t where At and A0 stand for the quantity at time t and time 0. Thus, the half-life (t1/2), when 50% of the protein is decayed is described by t1/2 = ln2 / Kdecay. Puromycin incorporation assay HeLa cells were seeded in a 6-well plate with a coverslip in it, and then transfected with 1 μg of pPB-DDX3X and pPB-DDX3Y respectively. After 24 hrs, a final concentration of 500 μM sodium arsenite was added to the DMEM and incubated for 5 min to trigger stress granule formation. Then, the medium was replaced by fresh DMEM medium with puromycin (1 μg/mL). After incubating for 30 min in a humidified incubator with 5% CO2 at 37°C, the cells were then washed with 1 × PBS and subjected to the immunofluorescence cell staining protocol as described above using anti-puromycin and anti-FLAG antibodies. Malachite green ATPase assay ATPase measurements were taken using the Malachite Green Phosphate Assay Kit according to the manufacturer’s instructions. Briefly, 1.0 μM MPB-DDX3X or MBP-DDX3Y was incubated with 100 ng/μL total RNA extracted from HeLa cells for 15 min in the reaction buffer (25 mM Tris-HCl, pH 8, 200 mM NaCl, 1 mM DTT, and 2 mM MgCl2) before the addition of 2 mM ATP. The reaction was incubated at room temperature for 30 min. Then the reaction was quenched by the addition of malachite green mixture and left for an additional 30 min to develop the color. The samples were then loaded into a clear-bottom 384-well plate and absorbance at OD600 nm was measured. Values were converted from absorbance units to μM free phosphate using a standard curve generated with the kit’s phosphate standard. Background values (free phosphates detected from reactions lacking RNA) were subtracted from the value from reactions with RNA. Data were plotted in Prism. Significance was calculated using Student’s two-tailed t test. Continuous ATPase assay ATPase measurements were taken using the EnzChek Phosphate Assay Kit (ThermoFisher) as previously described (Song and Ji, 2019). Briefly, MBP-DDX3X or MBP-DDX3Y was titrated at the indicated concentrations into 500 μL reactions containing 40 mM Tris-HCl pH 7.5, 50 mM NaCl, 0.5 mM MgCl2, 2 mM DTT, 0.5 units Super RNaseIN, 0.01% NP-40 substitute and 100 nM annealed duplex RNA (sequence in Table S3). After 10 min of incubation at room temperature, the reaction was started by adding 2 mM magnesium ATP. The reaction progress was monitored over five min at 30 sec intervals as absorbance at 360 nm. Absorbance values were the converted to μM/min using a standard curve and initial rates were plotted and fit to Hill kinetics using Prism. smFRET PEG-passivated slides were prepared according to previously published protocols with minor modifications (Jamiolkowski et al., 2017). Briefly, glass coverslips and slides were sonicated at 40°C twice with acetone for 15 min followed by sonication with methanol (25 min), 1 M KOH (40 min), and ethanol (15 min). Plasma cleaning was then used to remove any remaining contaminants from the surfaces. Plasma cleaned slides and coverslips were then incubated in a solution composed of 1.4 mL of 3-aminopropyltriethoxysilane (APTES), 2.3 mL of glacial acetic acid, and 46 mL of methanol overnight at room temperature. Silicated slides and coverslips were then incubated with polyethylene glycol (PEG, Laysan Bio, Inc., containing 20% (w/w) mPEG succinimidyl valerate, MW 2000 and 1% biotin-PEG-SC, MW 2000) in 0.1 M sodium bicarbonate (pH 8.3) for four hrs followed by an additional incubation overnight in a humidifying chamber. Slides and coverslips were then washed with MilliQ water and used to construct flow chambers for single molecule experiments with double-sided sticky tape. Double stranded RNA with a 5’ overhang was obtained as follows: 5’-biotin/ACCGCUGCCGUCGCUCCG/AlexF647N/−3’; and 5’-/Cy3/UUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUCGGAGCGACGGCAGCGGU-3’ were ordered from IDT; the two strands were annealed by heating the sample at 65°C for 5 min, then slowly cooling the sample to room temperature over 3 hrs. 2 mM protocatechuic acid (PCA), 0.05 uM protocatechuate-3,4,dioxygenase, 2.45 mM 6-hydroxy-2,5,7,8- tetramethylchromane-2-carboxylic (Trolox), 1 mM cyclooctatetraene (COT), and 1 mM 4-nitrobenzyl alcohol (NBA) were added to the imaging buffer (125 mM NaCl and 50 mM Tris-HCl pH 7.5). 70 pM biotinylated, fluorescent dsRNA was added to the flow cell and incubated for 6 min. Unbound RNA was washed out and recording either commenced or DDX3X or DDX3Y were first added as indicated in the text. The TIRF microscope, optics and camera were used to record smFRET as in previous report (Jamiolkowski et al., 2017). All FRET measurements were carried out at room temperature (~23°C) using a frame interval of 100 ms with alternating 532 nm/640 nm laser excitation (ALEX; effective frame interval 200 ms). Data analyses were carried out using custom scripts written in either Python or Java. Background-corrected fluorescence intensities and FRET distributions were corrected for differential quantum yield, differential detector sensitivity, direct excitation of the A647 by the 532 nm laser and leakage of Cy3 fluorescence into the acceptor detector channel as in previous report (Hellenkamp et al., 2018; Jamiolkowski et al., 2017). Only recordings exhibiting a single step photobleaching event in the direct acceptor excitation (ALEX) channel were further analyzed and included in FRET distributions. RNA isolation A Direct-zol RNA MicroPrep kit (Zymo Research) was used to isolate total RNA following the manufacturer’s instructions. DNA digestion was performed on the column at room temperature for 15 min to remove the DNA contamination of the extracted RNA. For a larger scale of RNA purification, total RNA was purified by TRIzol reagent (Invitrogen), following the manufacturer’s instruction. RT-qPCR RT-qPCR was performed using a Luna Universal One-Step RT-qPCR kit (NEB) to assess the relative abundance of RNA in each sample. All samples used for RT-qPCR were treated with DNase I to remove possible DNA contamination. The primers used for RT-qPCR are listed in Table S3. Cellular APEX labeling To enable APEX labeling in mammalian cells, we followed a previously established protocol (Fazal et al., 2019). Briefly, the cells were treated with or without 500 μM sodium arsenite in DMEM media for 30 min. Then, the DMEM media with 500 μM sodium arsenite was replaced with DMEM media supplemented with 500 μM sodium arsenite and 500 mM biotin-phenol for another 30 min. Next, H2O2 (Sigma-Aldrich) was added to each cell culture dish at 1 mM final concentration for exactly 1 min with gentle agitation. To stop the labeling, the culture media was removed, and the quenching solution (10 mM sodium ascorbate, 10 mM sodium azide, and 5 mM Trolox in PBS) was immediately used to wash the cells three times. Finally, 1 mL quenching solution was applied to cover the cells. Cells were then collected with a cell scraper. The unlabeled control samples were prepared in parallel under the same procedure as aforementioned, without the addition of the H2O2. Validation of APEX labeling To validate APEX labeling in cells, after the cellular labeling reaction, the cells were lysed using RIPA buffer (25 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, and 1% NP-40) supplemented with 10 mM sodium ascorbate, 10 mM sodium azide, 5 mM Trolox, and proteinase inhibitor on ice for 15 min. After centrifugation, the supernatant of the cell lysate was loaded on an SDS-PAGE gel and transferred to a PVDF membrane (Millipore) by a semi-dry transfer instrument. The membrane was blocked with 5% BSA in PBST at 4 °C overnight. 1:20,000 diluted streptavidin-HRP antibody (cell signaling) in 5% BSA was used at room temperature for 1 hr to detect the biotinylated proteins. After washing three times with PBST, the membrane was visualized by ECL Western Blotting Detection Kit (Thermo Fisher). The endogenous biotinylated proteins were also visible at 130, 75, and 72 kDa in both the labeled and unlabeled cells (as expected). Purification and sequencing biotinylated RNA After the cellular APEX-labeling, total RNA was extracted using TRIzol following the manufacturer’s instruction. To enrich biotinylated RNAs, Pierce streptavidin magnetic beads (Thermo Fisher) were used (20 μL beads per 50 μg RNA). The beads were washed 3 times with 500 μL B&W buffer (5 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 1 M NaCl, and 0.1% TWEEN 20), followed by 2 times with 500 μL solution A (0.1 M NaOH and 50 mM NaCl), and once with 500 μL solution B (100 mM NaCl). The beads were then suspended in 125 μL solution B. 50 μg RNA diluted in 125 μL water was mixed with the beads and incubated at 4 °C for 2 hrs with rotation. Then, the beads were placed on a magnetic stand to remove the solution and washed 3 times with 500 μL B&W buffer. Finally, the beads were resuspended in 54 μL water. To release the biotinylated RNA, 33 μL proteinase buffer (3 × PBS, 6% N-Lauryl sarcosine sodium solution (Sigma-Aldrich), 30 μM EDTA, and 15 μM DTT) was added to the beads with 10 μL proteinase K (Thermo Fisher) and 5 μL RNase inhibitor. The beads were then incubated at 42 °C for 1 hr and 55°C for 1 hr on a shaker at 600 rpm. The RNA was then purified using a RNA Clean and Concentrator 5 kit (Zymo Research) following the manufacturer’s instruction. The resulting RNA was then used for RNA-seq library construction using a TruSeq Stranded mRNA Kit (Illumina). The concentrations for all the libraries were determined by KAPA Library Quantification Kit (KAPA Biosystems) following the manufacturer’s protocol and subjected to Next-generation high-throughput sequencing using an Illumina NextSeq 550 with a single-end 75-bp read length. Validation of the APEX labeling of RNA To test the specificity of APEX2 labeling, APEX2 was fused with a mitochondrial localization signal (MLATRVFSLVGKRAISTSVCVRAH, derived from COX4) at its N terminus. The localization of the fused protein was validated by immunofluorescence. Total RNA was first extracted from the labeled and unlabeled cells, which was then followed by enrichment of biotinylated RNA using the streptavidin magnetic beads as described above. To test for the RNA enrichment, primers against mitochondria-translated transcripts ND1 and ND2, and cytoplasm translated transcripts GAPDH and TRMT10A were designed and listed in Table S3. RT-qPCR was performed in the labeled and unlabeled controls. The ratios of RNA recovered in the labeled samples relative to unlabeled controls were calculated. High throughput data analysis The high throughput sequencing reads were adaptor and quality trimmed using Trimmomatic (Bolger et al., 2014) with the following command: trimmomatc SE -phred33 ILLUMINACLIP:TruSeq3-SE.fa:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:30. Then, the reads were aligned to the GRCh38 human genome reference using HISAT2(Kim et al., 2015). The default parameters were used. Aligned reads were quantified using featureCounts (Liao et al., 2014). Read counts were further analyzed using DESeq2 (Love et al., 2014). Gene ontology (GO) analysis was carried out with the Metascape (Zhou et al., 2019). In Vitro translation In vitro translation assays were performed using the Flexi Rabbit Reticulocyte Lysate System (Promega). Each reaction (25 μL) contains 10 μL rabbit reticulocyte lysate, 0.25 μL amino acid mixture minus leucine (1 mM), 0.25 μL of amino acid mixture minus methionine (1 mM), 1 μL Mg(OAc)2, 0.25 μL luciferase mRNA (1 mg/mL), 0.5 μL RNase inhibitor, 0.25 μL DTT (1 M), and 12.5 μL of different concentrations of DDX3X-mCherry, DDX3Y-mCherry, or buffer (25 mM HEPES-KOH, pH 7.4, 150 mM KCl, and 2 mM DTT). Assembled reactions were incubated at 30°C for 90 mins. A standard reaction containing 75 μL of the luciferase substrate mixed with 5 μL of the unpurified translation mixture in a white 96-well plate. The luminescence was measured using a luminometer (Promega). To observe any LLPS in the lysate, the lysate from the translation assays was transferred to 384-well glass-bottomed plate. Then, the plate was spun at 100 g for 1 min. The images were taken using a Zeiss LSM 880 confocal microscope under a 63 × oil lens. FUS aggregation assay First, 0, 0.25, or 0.5 μM MBP-TEV-DDX3X or MBP-TEV-DDX3Y was incubated with 1 μg TEV protease for 30 min at room temperature. Then, 1 μM GST-TEV-FUSWT (or an equal volume of elution buffer) was added to the reaction and turbidity was used to assess aggregation by measuring absorbance at 395 nm in a Tecan plate reader. Readings from DDX3X or DDX3Y alone were subtracted from the appropriate FUS conditions. Area under the curve was used to compare the extent of aggregation for each condition (GraphPad Prism). TDP-43 aggregation assay Firstly, 0, 0.25, or 0.5 μM MBP-TEV-DDX3X or MBP-TEV-DDX3Y in buffer (200 mM NaCl, 25 mM Tris-HCl pH 8.0) was incubated in TDP-43 assay buffer (150 mM NaCl, 20 mM HEPES-NaOH pH 7.0, 1mM DTT) with 0.5 μg TEV protease for 30 minutes at room temperature. Turbidity was assessed by measuring absorbance at 395 nm in a Tecan plate reader. TDP-43 was buffer exchanged into TDP-43 assay buffer (Bio-Rad Micro Bio-Spin Chromatography Columns, following manufacturer’s instructions) and concentration was determined via NanoDrop. Turbidity measurements were paused after 30 minutes in order to add 4.0 μM TDP-43 (or an equal volume of TDP-43 assay buffer) to the reaction. Then, turbidity measurements were resumed for an additional 16 hours. The data was standardized by subtracting out the initial reading at t = 1 min. from each respective condition. Values from conditions with DDX3X or DDX3Y alone were then subtracted from the appropriate conditions with TDP-43. Area under the curve analysis was used to compare the extent of aggregation for each condition. The t1/2 of aggregation was determined by performing a nonlinear regression (asymmetric sigmoidal) on the data starting after TDP-43 addition (t = 31 min.) (GraphPad Prism). Protein quantification and Western blot Protein concentrations of the samples were calculated using the Bradford Assay (Bio-Rad). Protein samples were boiled at 95°C in Laemmli sample buffer for 10 min. After brief centrifugation, the samples were loaded onto SDS-PAGE gels. After running at 180 V for 1 hr, the gels were transferred to the PVDF membranes (Millipore) by semi-dry transfer apparatus at 20 V for 50 min. Then, the PVDF membranes were blocked with 5% milk or BSA in 1 × PBST for 30 mins at room temperature or 4°C overnight. The membranes were then incubated in 3% milk or BSA in 1 × PBST containing the corresponding primary antibodies overnight at 4°C. After washing three times with 1 × PBST, the horseradish peroxidase (HRP)-conjugated secondary antibodies (1:20,000) in 1% of milk were applied and incubated at room temperature for 1 hr. After washing three times with 1 × PBST, the membranes were visualized using ECL Western Blotting Detection Kit (Thermo Fisher). QUANTIFICATION AND STATISTICAL ANALYSIS Images were analyzed using Fiji. All data are presented as the mean ± standard error of mean (s.e.m.) or standard deviation (s.d.) from the independent determinations. The statistical analyses were performed using the GraphPad Prism (GraphPad Software, Inc,; La Jolla, CA, USA). Differences of means were tested for statistical significance with unpaired two-tailed Student’s t-test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; n.s. means p > 0.05. Supplementary Material 1 2 Table S2. Summary of DDX3X-specific, DDX3Y-specific, and overlapped transcripts from APEX-seq, related to Figure 5. 3 Table S3. Summary of the cloning and RT-qPCR primers used in this study, related to Figures 1, 2, 3, 4, 5 and 6. 4 Movie S1. Time-lapse imaging video of FRAP experiment with DDX3X-mCherry LLPS, related to Figure 1. 5 Movie S2. Time-lapse imaging video of FRAP experiment with DDX3Y-mCherry LLPS, related to Figure 1. 6 Movie S3. A biological replicate of time-lapse imaging video of FRAP experiment with DDX3X-mCherry LLPS, related to Figures 1 and S1. 7 Movie S4. A biological replicate of time-lapse imaging video of FRAP experiment with DDX3Y-mCherry LLPS, related to Figures 1 and S1. 8 Movie S5. Time-lapse imaging video of FRAP experiments with DDX3X-EGFP Stress Granule in HeLa Cells, related to Figure 2. 9 Movie S6. Time-lapse imaging video of FRAP experiments with DDX3Y-EGFP Stress Granule in HeLa Cells, related to Figure 2. ACKNOWLEDGEMENTS This work was supported by the National Institutes of Health (R35GM133721 to K.F.L., R35GM133721–03S1 to A.Y., T32GM132039 to A.Y. and M.C.O., T32GM008275 to K.E.C. and C.M.F., F31NS111870. to C.M.F., R01GM099836, R21AG061784, and R21AG065854 to J.S., and R35GM118139 to Y.E.G.). K.F.L. is supported by the American Cancer Society (RSG-22–064-01-RMC). J.S. is supported by grants from Target ALS, the Amyotrophic Lateral Sclerosis Association, the Office of the Assistant Secretary of Defense for Health Affairs through the Amyotrophic Lateral Sclerosis Research Program (W81XWH-20–1-0242), the G. Harold and Leila Y. Mathers Foundation, and Sanofi. We thank Dr. Bede Portz for the pMAL-c2X-DDX3X plasmid, Kollin Schultz for his help with DSF experiments, Dr. Matthew Kayser for sharing the Leica SP8 confocal microscope, Dr. Brain Capell for sharing the Nextseq550 sequencer, Dr. Karsten Weis for sharing the pKW4557 (DDX3X-mCherry) plasmid, and Dr. Ophir Shalem for sharing the plasmid expressing mClover3-FUS and mClover3-TDP-43 with us. We thank Drs. Ronen Marmorstein, Hillary Nelson, Gregory Van Duyne, and Kristen Lynch for their constructive discussions and for editing this manuscript. Figure 1. DDX3Y has a stronger LLPS propensity compared to DDX3X in vitro. (A) Structural prediction of DDX3X and DDX3Y using PONDR (natural disordered regions), PLAAC (prion-like amino acid regions), and catGRANULE (LLPS propensity). (B) In vitro droplet formation of 10 μM recombinant DDX3X-mCherry or DDX3Y-mCherry in the absence or presence of 200 ng/μL poly(U)-RNA. Scale bar, 25 μm. (C) Quantification of the total integrated intensity of DDX3X condensation and DDX3Y condensation in Figure 1B. A two-tailed t-test was used to calculate the p-value. ***p < 0.001, **** p < 0.0001. (D) Concentrations of DDX3 proteins in the light phase (supernatant after centrifugation) vs. input protein concentrations. (E) Turbidity (absorbance at 600 nm) of DDX3X-mCherry and DDX3Y-mCherry LLPS. The mean value of turbidity and protein concentration for each condition from three separate protein purifications and three technical repeats were plotted in Figure 1D and 1E. (F) Time-lapse images of in vitro FRAP experiments. The FRAP experiments were performed identically for DDX3X-mCherry and DDX3Y-mCherry droplets. (G) FRAP curves for in vitro droplets of DDX3X-mCherry (red) and DDX3Y-mCherry (blue). The traces of the FRAP data represent mean ± s.e.m (n = 3, from three independent experiments). (H) Halftime and mobile fractions from Figure 1G. A two-tailed t-test was used to calculate the p-value. **p < 0.01. See also Figure S1. Figure 2. DDX3Y has a stronger LLPS propensity compared to DDX3X in cells. (A) Representative images of co-localization of DDX3X and DDX3Y with G3BP1 in HeLa, N2a, and HEK 293T cells with the endogenous DDX3X knocked down upon arsenite treatment (500 μM, 1 hr). Scale bar, 10 μm. (B) Violin plots of the total SG area of DDX3X- or DDX3Y-positive SGs per cell across 50 cells upon arsenite treatment (500 μM, 1 hr) in endogenous DDX3X-depleted HeLa, N2a, or HEK 293T cells (n = 50 cells in total, from 3 biologically independent experiments). p values were determined by a two-tailed t-test; **p < 0.01, ***p < 0.001, **** p < 0.0001. (C) Cycloheximide (CHX) chase assay to determine the cellular half-life of DDX3X and DDX3Y in HeLa cells. Two biological replicates for DDX3X and for DDX3Y were performed. “M” represents the protein ladder on the Western blot membranes (markers: upper 75 kDa, lower 37 kDa). (D) Quantification of the protein levels in Figure 2C. The intensity of each DDX3X and DDX3Y band was normalized to the corresponding GAPDH intensity before being normalized to the intensity at the corresponding 0 hr time point. The half-lives of DDX3X and DDX3Y were 4.49 hrs and 4.41 hrs, respectively. (E) Time-lapse images of photobleached SGs in HeLa cells expressing DDX3X-EGFP (left) or DDX3Y-EGFP (right) from in-cell FRAP experiments. The photobleaching events and fluorescence recovery by DDX3X-EGFP- and DDX3Y-EGFP-positive SGs are highlighted by the arrow in each outlined box. (F) FRAP curves for DDX3X-EGFP (red) and DDX3Y-EGFP (blue) in HeLa cells. The trace of the FRAP data represents mean ± s.e.m. (n = 20 independent measurements, from 3 biologically independent experiments). (G) The halftime and mobile fractions in Figure 2E & 2F. A two-tailed t-test was used to calculate the p-value. **p < 0.01, **** p < 0.0001. See also Figure S2. Figure 3. IDR1 of DDX3Y more strongly promotes phase separation than IDR1 of DDX3X. (A) Co-localization of DDX3X or DDX3Y domain truncation variants in HeLa cells with G3BP1 upon 500 μM arsenite treatment for 1 hr. Scale bar, 10 μm. (B) Violin plots of the total SG area of truncation SGs per cell (n = 50 cells in total, from three biologically independent experiments). The median of the total SG areas per cell of wild-type DDX3X-SGs and DDX3Y-SGs are indicated by red and blue dashed lines, respectively. p values were determined by a two-tailed t-test; **p < 0.01, ***p < 0.001, ****p < 0.0001. (C) Co-localization of DDX3X or DDX3Y domain swap variants in HeLa cells with G3BP1 upon 500 μM arsenite treatment for 1 hr. Scale bar, 10 μm. (D) Violin plots of the total SG area of DDX3X or DDX3Y domain swap SGs per cell (50 cells in total from three biologically independent experiments). The median sum of SG areas per cell of wild-type DDX3X-SGs and DDX3Y-SGs is indicated by red and blue dashed lines, respectively. p values were determined by nested t-test to compare all domain-swapped variants with XIDR1 versus all domain-swapped variants with YIDR1; ****p < 0.0001. (E) In vitro droplet formation of 7.5 μM recombinant DDX3X-mCherry, DDX3Y-mCherry, or domain swap variants of DDX3X and DDX3Y in the presence of 200 ng/μL poly(U)-RNA. Scale bar, 25 μm. (F) Quantification of the total integrated intensity of the different conditions shown in Figure 3E. Error bars represent s.d. from three repeats at each condition. p values were determined by nested t-test to compare all domain-swapped variants with XIDR1 versus all domain-swapped variants with YIDR1; **p < 0.01. See also Figure S3. Figure 4. The weaker ATPase activity of DDX3Y compared to DDX3X weakens its condensate disassembly. A) In vitro droplet formation of 10 μM recombinant DDX3X-mCherry and DDX3Y-mCherry in the absence and presence of 200 ng/μL poly(U)-RNA with and without the addition of 4 mM ATP or UTP. Scale bar, 25 μm. (B) Quantification of the total integrated intensity of different groups of condensates shown in Figure 4A. A two-tailed t-test was used to calculate the p-value. n.s. means p > 0.05, **p < 0.01, ***p < 0.001, and **** p < 0.0001. (C) ATPase activity of DDX3X and DDX3Y, as measured by release of free phosphate, in μM/min. Error bars represent ± s.d. from 6 individual replicates. The background value (initial rate when no protein was added) was subtracted from each point before plotting and curve fitting to the Hill equation. DDX3X: Vmax = 7.9 ± 0.4 μM/min, H = 1.9 ± 0.2, K1/2 = 297.2 ± 22.6 nM. DDX3Y: Vmax = 5.9 ± 0.7 μM/min, H = 1.5 ± 0.2, K1/2 = 617.9 ± 123.6 nM. All fitting parameter uncertainties are ± s.e.m. (D) Schematic of smFRET RNA probe. (E) FRET efficiency distributions with increasing protein concentrations (2, 4, and 8 μM) of DDX3X and DDX3Y in the presence or absence of ATP. FRET values were collected from over 1000 molecules to build the histograms. (F) Fitting of the Hill equation to relative values of the low FRET peak area (with background subtracted) in Figure 4E for DDX3X and DDX3Y. See also Figure S4. Figure 5. DDX3X and DDX3Y condensation inhibit the translation of luciferase RNA, and DDX3X- and DDX3Y-positive SGs have shared and unique RNA constituents in cells. (A) Schematic illustration of the in vitro translation assay. (B) In vitro translation inhibition at the indicated concentrations of DDX3X-mCherry or DDX3Y-mCherry. p values were determined by two-tailed t-test; ****p < 0.0001. (C) DDX3X-mCherry and DDX3Y-mCherry phase separation in the reticulocyte assay at each indicated concentration; scale bar, 20 μm. (D) Quantification of the total integrated intensity of different condensates shown in Figure 5C. (E) Puromycin incorporation assay to determine the extent of translation repression in cells with DDX3X-positive SGs and DDX3Y-positive SGs. The white outlines indicate the cells which express exogenous DDX3X or DDX3Y. (F) Quantification of puromycin signal in Figure 5E. Only cells expressing exogenous DDX3X or DDX3Y were selected, and the total puromycin signal in each cell was quantified. The total puromycin signals in neighboring cells not expressing exogenous DDX3X or DDX3Y were quantified similarly and used to normalize the data. p-value was determined by two-tailed t-test; **p < 0.01. (G) Schematic illustration of APEX2-mediated proximity labeling reaction. (H) Volcano plots showing differential RNA enrichment in streptavidin pull-downs from APEX2-DDX3X (left) and APEX2-DDX3Y (right) expressing cells compared to APEX2-EGFP expressing cells after 500 μM arsenite treatment for 1 hr. Differentially expressed genes are shown in red and blue for APEX2-DDX3X-enriched and APEX2-DDX3Y-enriched RNAs respectively (adjusted p < 0.05, log2 fold change > 1) and dark gray for APEX2-DDX3X-depleted and APEX2-DDX3Y-depleted RNAs (adjusted p < 0.05, log2 fold change < −1). The rest of the RNAs are shown in light gray for both DDX3X and DDX3Y. The triangles represent the transcripts with log2 fold change > 13 or log2 fold change < −13 in the X-axis; −log10 adjusted p >15 in the Y-axis. (I) Gene Ontology (GO) analysis of the differentially enriched RNA groups in Figure 5E; red: APEX2-DDX3X-enriched RNAs; and blue: APEX2-DDX3Y-enriched mRNAs. (J) Venn diagram quantifying the overlapping and distinct RNA clients enriched by APEX2-DDX3X and APEX2-DDX3Y after arsenite treatment. (K) Relative fold change of select RNAs used to conduct RT-qPCR validation of the sequencing results. See also Figure S5. Figure 6. A combination of DDX3X and DDX3Y shows a stronger propensity of LLPS and translation repression than DDX3X alone; and DDX3Y enhances FUS aggregation and accelerates TDP-43 aggregation more than DDX3X. (A) Immunofluorescence images of SGs containing both FLAG-DDX3X and HA-DDX3Y in HeLa cells treated with sorbitol, puromycin, or CCCP. Below each image, traces of fluorescence intensity profiles through positions denoted by the white lines in the merged images. The area under the curve (AUC) normalized to that of FLAG-DDX3X is plotted for each intensity profile and shows that the signal from HA-DDX3Y is consistently higher than the signal from FLAG-DDX3X. (B) A mixture of DDX3X-mCherry and DDX3Y-mCherry at the annotated concentrations differentially repress in vitro translation of luciferase RNA. Overall, DDX3Y represses translation more than DDX3X. p values were determined by two-tailed t-test; ****p < 0.0001. (C) In vitro droplet formation of recombinant DDX3X-mCherry and DDX3Y-mCherry at the annotated ratios in the presence of 200 ng/μL poly(U)-RNA (top panel). Scale bar, 25 μm. Quantification of the total integrated intensity of each type of condensate (bottom panel). p values were determined by two-tailed t-test; *p < 0.05, ****p < 0.0001. (D) DDX3Y more strongly promotes FUS aggregation than DDX3X in vitro. The area under the curve for each time course of light scattering at 395 nm in Figure S6E was used to compare the extent of aggregation for each condition. p values were determined by two-tailed t-test; *p < 0.05, ***p < 0.001. (E) Western blots showing the expression of DDX3X and DDX3Y in DOX-inducible stable cell lines expressing mClover3-FUS. (F) Representative images showing the localization of Flag-DDX3X, Flag-DDX3Y, and mClover-FUS upon arsenite treatment (500 μm, 1 hr). Scale bar, 10 μm. (G) The total area of granules containing mClover3-FUS and Flag-DDX3X or granules containing mClover3-FUS and Flag-DDX3Y per cell was quantified (n = 50 cells in total, from three biologically independent experiments). p values were determined by a two-tailed t-test; ****p < 0.0001. (H) Schematic illustration of how sexually dimorphic RNA helicases DDX3X and DDX3Y differentially regulate RNA translation through phase separation. See also Figure S6. Key resources table REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Rabbit polyclonal anti-G3BP1 Proteintech Cat# 13057-2-AP, RRID:AB_2232034 Rabbit polyclonal anti-HA Abcam Cat# ab9110, RRID:AB_307019 Mouse monoclonal anti-FLAG Sigma-Aldrich Cat# F3165, RRID:AB_259529 Rat monoclonal anti-FLAG Invitrogen Cat# MA1-142, RRID:AB_2536846 HRP conjugated anti-FLAG Invitrogen Cat# MA1-91878-HRP, RRID:AB_2537626 Rabbit polyclonal anti-GAPDH Invitrogen Cat# PA1-16777, RRID:AB_568552 Mouse monoclonal anti-GFP Santa Cruz Cat# sc-996, RRID:AB_2187785 Mouse anti-puromycin, clone 12D10, Alexa Fluor 488 conjugated antibody Sigma-Aldrich Cat# MABE343-AF488, RRID:AB_2736875 Goat anti-mouse IgG (H&L) (HRP) Abcam Cat# ab6789, RRID:AB_955439 Mouse monoclonal anti-ATP5A Abcam Cat# ab14748, RRID:AB_301447 Goat anti-Mouse IgG H&L (Alex Fluor® 647) Abcam Cat# ab150115, RRID:AB_2687948 Goat anti-Rabbit IgG H&L (Alexa Fluor® 594) Abcam Cat# ab150080, RRID:AB_2650602 Goat anti-Mouse IgG H&L (Alexa Fluor® 488) Abcam Cat# ab150113, RRID:AB_2576208 Goat Anti-Rat IgG H&L (Alexa Fluor® 488) Abcam Cat# ab150157, RRID:AB_2722511 Mouse monoclonal anti-DDX3 Abcam Cat# ab196032 Rabbit polyclonal anti-DDX3Y Invitrogen Cat# PA5-90055, RRID:AB_2805908 Bacterial and Virus Strains One Shot™ TOP10 Chemically Competent E. coli Invitrogen Cat# 404003 BL21(DE3)-RIL Competent E. coli Agilent Cat# 230204 Chemicals, Peptides, and Recombinant Proteins Streptavidin-HRP Cell Signaling Technology Cat# 3999, RRID:AB_10830897 DDX3X-mCherry recombinant protein This paper N/A DDX3Y-mCherry recombinant protein This paper N/A MBP-DDX3X recombinant protein This paper N/A MBP-DDX3Y recombinant protein This paper N/A N-Lauroylsarcosine sodium salt solution Sigma-Aldrich Cat# L7414-10ML SUPERase·In™ RNase Inhibitor Invitrogen Cat# AM2694 Proteinase K solution (20 mg/mL) Life Technologies Cat# AM2548 Lipofectamine 2000 Thermo Fisher Cat# 11668019 Pierce streptavidin magnetic beads Thermo Fisher Cat# 88816 Agencourt AMPure XP Beckman Coulter Cat# A63881 TRIzol Invitrogen Cat# 15596026 IPTG Thermo Fisher Cat# 34060 DAPI Sigma-Aldrich Cat# D9542 Fisher BioReagents™ Bovine Serum Albumin, Heat Shock Treated Fisher Scientific Cat# BP1600-100 Polyuridylic acid potassium salt Sigma-Aldrich Cat# P9528-10MG Sodium arsenite Spectrum Chemical Cat# S-222 Sorbitol solution Spectrum Chemical Cat# S-1525 Thapsigargin ACROS Organics Cat# AC328570010 CCCP Alfa Aesar Cat# AAL06932MC Puromycin Takara Bio USA Cat# 631305 MG132 Sigma-Aldrich Cat# 474790-1 MG Hydrogen peroxide solution Sigma-Aldrich Cat# H1009 Biotin-phenol Iris Biotech Cat# LS-3500.1000 Sodium azide Sigma-Aldrich Cat# S2002 Sodium ascorbate Sigma-Aldrich Cat# A7631 Trolox Sigma-Aldrich Cat# 238813 Cycloheximide Sigma-Aldrich Cat# C7698 Critical Commercial Assays Malachite Green Phosphate Assay Kit Sigma-Aldrich Cat# MAK307-1KT Flexi® Rabbit Reticulocyte Lysate System Promega Cat# L4540 Luna® Universal One-Step RT-qPCR Kit NEB Cat# M3005L Truseq Stranded mRNA Library Prep Illumina Cat# 20020594 Dynabeads™ mRNA Purification Kit (for mRNA purification from total RNA preps) Invitrogen Cat# 61006 EnzChek™ Phosphate Assay Kit Thermo Fisher Cat# E6646 RNA Clean and Concentrator-5 Zymo Research Cat# R1016 Deposited Data APEX-seq with stress treatment This paper GEO: GSE171792 APEX-seq without stress treatment This paper GEO: GSE193783 Raw data This paper; Mendeley Data https://dx.doi.org/10.17632/9hs5d4fvgd.1 Experimental Models: Cell Lines Human: HeLa ATCC CCL-2, RRID:CVCL_0030 Human: HEK 293T ATCC CRL-3216, RRID:CVCL_0063 Mouse: N2a ATCC CCL-131, RRID:CVCL_0470 Oligonucleotides Primers for all the recombinant DNA This paper Table S3 Recombinant DNA pPB DDX3X This paper N/A pPB DDX3Y This paper N/A pPB DDX3X-IDR1 This paper N/A pPB DDX3Y-IDR1 This paper N/A pPB DDX3X-Helicae This paper N/A pPB DDX3Y-Helicae This paper N/A pPB DDX3X-IDR2 This paper N/A pPB DDX3Y-IDR2 This paper N/A pPB DDX3X-ΔIDR1 This paper N/A pPB DDX3Y-ΔIDR1 This paper N/A pPB DDX3X-ΔIDR2 This paper N/A pPB DDX3Y-ΔIDR2 This paper N/A pPB DDX3X-ΔHelicase This paper N/A pPB DDX3Y-ΔHelicase This paper N/A pPB DDX3XIDR1-DDX3YHelicae-DDX3YIDR2 This paper N/A pPB DDX3YIDR1-DDX3XHelicae-DDX3XIDR2 This paper N/A pPB DDX3XIDR1-DDX3Xhelicae-DDX3YIDR2 This paper N/A pPB DDX3YIDR1-DDX3Yhelicae-DDX3XIDR2 This paper N/A pPB DDX3XIDR1-DDX3Yhelicae-DDX3XIDR2 This paper N/A pPB DDX3YIDR1-DDX3Xhelicae-DDX3YIDR2 This paper N/A pcDNA3 mito-APEX2 This paper N/A pPB DDX3X-EGFP This paper N/A pPB DDX3Y-EGFP This paper N/A pcDNA3 FLAG-DDX3X This paper N/A pcDNA3 FLAG-DDX3Y This paper N/A pcDNA3 HA-DDX3X This paper N/A pcDNA3 HA-DDX3Y This paper N/A pET-MCN_His-TEV_V5-DDX3X-mCherry (Hondele et al., 2019) N/A pET-MCN_His-TEV_V5-DDX3Y-mCherry This paper N/A pET-MCN_His-TEV_V5-DDX3XIDR1-DDX3YHelicae-DDX3YIDR2-mCherry This paper N/A pET-MCN_His-TEV_V5-DDX3YIDR1-DDX3XHelicae-DDX3XIDR2-mCherry This paper N/A pET-MCN_His-TEV_V5-DDX3XIDR1-DDX3Xhelicae-DDX3YIDR2-mCherry This paper N/A pET-MCN_His-TEV_V5-DDX3YIDR1-DDX3Yhelicae-DDX3XIDR2-mCherry This paper N/A pET-MCN_His-TEV_V5-DDX3XIDR1-DDX3Yhelicae-DDX3XIDR2-mCherry This paper N/A pET-MCN_His-TEV_V5-DDX3YIDR1-DDX3Xhelicae-DDX3YIDR2-mCherry This paper N/A pMAL-c2X (Walker et al., 2010) Addgene Plasmid #75286, RRID:Addgene_75286 pMAL-c2X-DDX3X This paper N/A pMAL-c2X-DDX3Y This paper N/A pcDNA5/FRT/TO APEX2-GFP (Padron et al., 2019) Addgene Plasmid #129640, RRID:Addgene_129640 pJ4M/TDP-43 (Hallegger et al., 2021) Addgene plasmid #104480, RRID:Addgene_104480 pPB APEX2-DDX3X This paper N/A pPB APEX2-DDX3Y This paper N/A pPB APEX2-EGFP This paper N/A Software and Algorithms HISAT2 (Kim et al., 2015) https://ccb.jhu.edu/software/hisat2/index.shtml, RRID:SCR_015530 DESeq2 (Love et al., 2014) https://bioconductor.org/packages/release/bioc/html/DESeq2.html, RRID:SCR_015687 featureCounts (Liao et al., 2014) http://bioinf.wehi.edu.au/featureCounts/, RRID:SCR_012919 trimmomatic (Bolger et al., 2014) http://www.usadellab.org/cms/index.php?page=trimmomatic, RRID:SCR_011848 R https://www.r-project.org https://www.r-project.org, RRID:SCR_000036 Fiji (Schindelin et al., 2012) https://imagej.net/Fiji, RRID:SCR_002285 GraphPad Prism GraphPad Software https://graphpad.com/scientificsoft, RRID:SCR_002798 Other 384-well microscopy plates Brooks Life Sciences Cat#4ti-0203 Amicon ultra-15 EMD Millipore Cat# UFC901024 Highlights: The N-terminal IDR of DDX3Y more strongly promotes condensation than DDX3X. Slower ATPase activity of DDX3Y contributes to weaker dissolution of condensates. Stronger phase separation of DDX3Y than DDX3X leads to more translation repression. DDX3Y condensates enhance aggregation of FUS more strongly than DDX3X. DECLARATION of INTERESTS J.S. is a consultant for Dewpoint Therapeutics, Vivid Sciences, Korro Bio, Neumora, and ADRx. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. REFERENCES Ash PE , Vanderweyde TE , Youmans KL , Apicco DJ , and Wolozin B (2014). Pathological stress granules in Alzheimer’s disease. Brain Res. 1584 , 52–58.25108040 Baradaran-Heravi Y , Van Broeckhoven C , and van der Zee J (2020). Stress granule mediated protein aggregation and underlying gene defects in the FTD-ALS spectrum. Neurobiol. Dis. 134 , 104639.31626953 Beckham C , Hilliker A , Cziko AM , Noueiry A , Ramaswami M , and Parker R (2008). The DEAD-box RNA helicase Ded1p affects and accumulates in Saccharomyces cerevisiae P-bodies. Mol. Biol. Cell 19 , 984–993.18162578 Bellott DW , Hughes JF , Skaletsky H , Brown LG , Pyntikova T , Cho TJ , Koutseva N , Zaghlul S , Graves T , Rock S , (2014). Mammalian Y chromosomes retain widely expressed dosage-sensitive regulators. Nature 508 , 494–499.24759411 Bentmann E , Neumann M , Tahirovic S , Rodde R , Dormann D , and Haass C (2012). Requirements for stress granule recruitment of fused in sarcoma (FUS) and TAR DNA-binding protein of 43 kDa (TDP-43). J. Biol. Chem. 287 , 23079–23094.22563080 Bolger AM , Lohse M , and Usadel B (2014). Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30 , 2114–2120.24695404 Bryk AH , and Wisniewski JR (2017). Quantitative analysis of human red blood cell proteome. J. Proteome Res. 16 , 2752–2761.28689405 Buchan JR , and Parker R (2009). Eukaryotic stress granules: the ins and outs of translation. Mol. Cell 36 , 932–941.20064460 Cardinal T , Bergeron KF , Soret R , Souchkova O , Faure C , Guillon A , and Pilon N (2020). Male-biased aganglionic megacolon in the TashT mouse model of Hirschsprung disease involves upregulation of p53 protein activity and Ddx3y gene expression. PLoS Genet. 16 , e1009008.32898154 Chen CY , Chan CH , Chen CM , Tsai YS , Tsai TY , Wu Lee YH , and You LR (2016). Targeted inactivation of murine Ddx3x: essential roles of Ddx3x in placentation and embryogenesis. Hum. Mol. Genet. 25 , 2905–2922.27179789 Cheng W , Wang S , Zhang Z , Morgens DW , Hayes LR , Lee S , Portz B , Xie Y , Nguyen BV , Haney MS , (2019). CRISPR-Cas9 screens identify the RNA helicase DDX3X as a repressor of C9ORF72 (GGGGCC)n repeat-associated non-AUG translation. Neuron 104 , 885–898.31587919 Cotton AM , Price EM , Jones MJ , Balaton BP , Kobor MS , and Brown CJ (2015). Landscape of DNA methylation on the X chromosome reflects CpG density, functional chromatin state and X-chromosome inactivation. Hum. Mol. Genet. 24 , 1528–1539.25381334 Ditton HJ , Zimmer J , Rajpert-De Meyts E , and Vogt PH (2004). The AZFa gene DBY (DDX3Y) is widely transcribed but the protein is limited to the male germ cells by translation control. Hum. Mol. Genet. 13 , 2333–2341.15294876 Elbaum-Garfinkle S , Kim Y , Szczepaniak K , Chen CC , Eckmann CR , Myong S , and Brangwynne CP (2015). The disordered P granule protein LAF-1 drives phase separation into droplets with tunable viscosity and dynamics. Proc. Natl. Acad. Sci. U.S.A. 112 , 7189–7194.26015579 Fazal FM , Han S , Parker KR , Kaewsapsak P , Xu J , Boettiger AN , Chang HY , and Ting AY (2019). Atlas of subcellular RNA localization revealed by APEX-Seq. Cell 178 , 473–490.31230715 Fernandes N , Nero L , Lyons SM , Ivanov P , Mittelmeier TM , Bolger TA , and Buchan JR (2020). Stress granule assembly can facilitate but is not required for TDP-43 cytoplasmic aggregation. Biomolecules 10 . Ferretti MT , Iulita MF , Cavedo E , Chiesa PA , Schumacher Dimech A , Santuccione Chadha A , Baracchi F , Girouard H , Misoch S , Giacobini E , (2018). Sex differences in Alzheimer disease - the gateway to precision medicine. Nat. Rev. Neurol. 14 , 457–469.29985474 Floor SN , Condon KJ , Sharma D , Jankowsky E , and Doudna JA (2016). Autoinhibitory interdomain interactions and subfamily-specific extensions redefine the catalytic core of the human DEAD-box protein DDX3. J. Biol. Chem. 291 , 2412–2421.26598523 Fujimura K , Sasaki AT , and Anderson P (2012). Selenite targets eIF4E-binding protein-1 to inhibit translation initiation and induce the assembly of non-canonical stress granules. Nucleic Acids Res. 40 , 8099–8110.22718973 Gao K , Oerlemans R , and Groves MR (2020). Theory and applications of differential scanning fluorimetry in early-stage drug discovery. Biophys. Rev. 12 , 85–104.32006251 Gilks N , Kedersha N , Ayodele M , Shen L , Stoecklin G , Dember LM , and Anderson P (2004). Stress granule assembly is mediated by prion-like aggregation of TIA-1. Mol. Biol. Cell 15 , 5383–5398.15371533 Godfrey AK , Naqvi S , Chmatal L , Chick JM , Mitchell RN , Gygi SP , Skaletsky H , and Page DC (2020). Quantitative analysis of Y-Chromosome gene expression across 36 human tissues. Genome Res. 30 , 860–873.32461223 Gozdecka M , Meduri E , Mazan M , Tzelepis K , Dudek M , Knights AJ , Pardo M , Yu L , Choudhary JS , Metzakopian E , (2018). UTX-mediated enhancer and chromatin remodeling suppresses myeloid leukemogenesis through noncatalytic inverse regulation of ETS and GATA programs. Nat. Genet. 50 , 883–894.29736013 Guo L , Kim HJ , Wang H , Monaghan J , Freyermuth F , Sung JC , O’Donovan K , Fare CM , Diaz Z , Singh N , (2018). Nuclear-import receptors reverse aberrant phase transitions of RNA-binding proteins with prion-like domains. Cell 173 , 677–692.29677512 Hallegger M , Chakrabarti AM , Lee FCY , Lee BL , Amalietti AG , Odeh HM , Copley KE , Rubien JD , Portz B , Kuret K , (2021). TDP-43 condensation properties specify its RNA-binding and regulatory repertoire. Cell 184 , 4680–4696.34380047 Harrison AF , and Shorter J (2017). RNA-binding proteins with prion-like domains in health and disease. Biochem. J. 474 , 1417–1438.28389532 Hellenkamp B , Schmid S , Doroshenko O , Opanasyuk O , Kuhnemuth R , Rezaei Adariani S , Ambrose B , Aznauryan M , Barth A , Birkedal V , (2018). Precision and accuracy of single-molecule FRET measurements-a multi-laboratory benchmark study. Nat. Methods 15 , 669–676.30171252 Hondele M , Sachdev R , Heinrich S , Wang J , Vallotton P , Fontoura BMA , and Weis K (2019). DEAD-box ATPases are global regulators of phase-separated organelles. Nature 573 , 144–148.31435012 Iossifov I , O’Roak BJ , Sanders SJ , Ronemus M , Krumm N , Levy D , Stessman HA , Witherspoon KT , Vives L , Patterson KE , (2014). The contribution of de novo coding mutations to autism spectrum disorder. Nature 515 , 216–221.25363768 Iserman C , Desroches Altamirano C , Jegers C , Friedrich U , Zarin T , Fritsch AW , Mittasch M , Domingues A , Hersemann L , Jahnel M , (2020). Condensation of Ded1p promotes a translational switch from housekeeping to stress protein production. Cell 181 , 818–831.32359423 Jain S , Wheeler JR , Walters RW , Agrawal A , Barsic A , and Parker R (2016). ATPase-modulated stress granules contain a diverse proteome and substructure. Cell 164 , 487–498.26777405 Jamiolkowski RM , Chen C , Cooperman BS , and Goldman YE (2017). tRNA fluctuations observed on stalled ribosomes are suppressed during ongoing protein synthesis. Biophys. J. 113 , 2326–2335.29211986 Joshi RN , Stadler C , Lehmann R , Lehtio J , Tegner J , Schmidt A , and Vesterlund M (2019). TcellSubC: an atlas of the subcellular proteome of human T cells. Front. Immunol. 10 , 2708.31849937 Kamelgarn M , Chen J , Kuang L , Arenas A , Zhai J , Zhu H , and Gal J (2016). Proteomic analysis of FUS interacting proteins provides insights into FUS function and its role in ALS. Biochim. Biophys. Acta 1862 , 2004–2014.27460707 Khalfallah Y , Kuta R , Grasmuck C , Prat A , Durham HD , and Velde CV (2018). TDP-43 regulation of stress granule dynamics in neurodegenerative disease-relevant cell types. Sci. Rep. 8 . Khong A , Matheny T , Jain S , Mitchell SF , Wheeler JR , and Parker R (2017). The stress granule transcriptome reveals principles of mRNA accumulation in stress granules. Mol. Cell. 68 , 808–820.29129640 Kim D , Langmead B , and Salzberg SL (2015). HISAT: a fast spliced aligner with low memory requirements. Nat. Methods 12 , 357–360.25751142 Kim TH , Tsang B , Vernon RM , Sonenberg N , Kay LE , and Forman-Kay JD (2019). Phospho-dependent phase separation of FMRP and CAPRIN1 recapitulates regulation of translation and deadenylation. Science 365 , 825–829.31439799 Kim Y , and Myong S (2016). RNA remodeling activity of DEAD box proteins tuned by protein concentration, RNA length, and ATP. Mol. Cell 63 , 865–876.27546789 Kimball SR , Horetsky RL , Ron D , Jefferson LS , and Harding HP (2003). Mammalian stress granules represent sites of accumulation of stalled translation initiation complexes. Am. J. Physiol. Cell Physiol. 284 , C273–C284.12388085 Ku YC , Lai MH , Lo CC , Cheng YC , Qiu JT , Tarn WY , and Lai MC (2019). DDX3 Participates in Translational Control of Inflammation Induced by Infections and Injuries. Mol. Cell Biol. 39 . Lai MC , Lee YH , and Tarn WY (2008). The DEAD-box RNA helicase DDX3 associates with export messenger ribonucleoproteins as well as tip-associated protein and participates in translational control. Mol. Biol. Cell 19 , 3847–3858.18596238 Lancaster AK , Nutter-Upham A , Lindquist S , and King OD (2014). PLAAC: a web and command-line application to identify proteins with prion-like amino acid composition. Bioinformatics 30 , 2501–2502.24825614 Lee CS , Dias AP , Jedrychowski M , Patel AH , Hsu JL , and Reed R (2008). Human DDX3 functions in translation and interacts with the translation initiation factor eIF3. Nucleic Acids Res. 36 , 4708–4718.18628297 Lennox AL , Hoye ML , Jiang R , Johnson-Kerner BL , Suit LA , Venkataramanan S , Sheehan CJ , Alsina FC , Fregeau B , Aldinger KA , (2020). Pathogenic DDX3X mutations impair RNA metabolism and neurogenesis during fetal cortical development. Neuron 106 , 404–420.32135084 Liao Y , Smyth GK , and Shi W (2014). featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30 , 923–930.24227677 Linsalata AE , He F , Malik AM , Glineburg MR , Green KM , Natla S , Flores BN , Krans A , Archbold HC , Fedak SJ , (2019). DDX3X and specific initiation factors modulate FMR1 repeat-associated non-AUG-initiated translation. EMBO Rep. 20 , e47498.31347257 Love MI , Huber W , and Anders S (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15 , 550.25516281 Maharana S , Wang J , Papadopoulos DK , Richter D , Pozniakovsky A , Poser I , Bickle M , Rizk S , Guillen-Boixet J , Franzmann TM , (2018). RNA buffers the phase separation behavior of prion-like RNA binding proteins. Science 360 , 918–921.29650702 Manjaly ZR , Scott KM , Abhinav K , Wijesekera L , Ganesalingam J , Goldstein LH , Janssen A , Dougherty A , Willey E , Stanton BR , (2010). The sex ratio in amyotrophic lateral sclerosis: A population based study. Amyotroph. Lateral Scler. 11 , 439–442.20225930 Markmiller S , Soltanieh S , Server KL , Mak R , Jin W , Fang MY , Luo EC , Krach F , Yang D , Sen A , (2018). Context-dependent and disease-specific diversity in protein interactions within stress granules. Cell 172 , 590–604.29373831 Marmor-Kollet H , Siany A , Kedersha N , Knafo N , Rivkin N , Danino YM , Moens TG , Olender T , Sheban D , Cohen N , (2020). Spatiotemporal proteomic analysis of stress granule disassembly using APEX reveals regulation by SUMOylation and links to ALS pathogenesis. Mol. Cell 80 , 876–891.33217318 Mauvais-Jarvis F , Bairey Merz N , Barnes PJ , Brinton RD , Carrero JJ , DeMeo DL , De Vries GJ , Epperson CN , Govindan R , Klein SL , (2020). Sex and gender: modifiers of health, disease, and medicine. Lancet 396 , 565–582.32828189 McCombe PA , and Henderson RD (2010). Effects of gender in amyotrophic lateral sclerosis. Gend. Med. 7 , 557–570.21195356 Mercer TR , Neph S , Dinger ME , Crawford J , Smith MA , Shearwood AM , Haugen E , Bracken CP , Rackham O , Stamatoyannopoulos JA , (2011). The human mitochondrial transcriptome. Cell 146 , 645–658.21854988 Meyer S , van der Meer P , van Tintelen JP , and van den Berg MP (2014). Sex differences in cardiomyopathies. Eur. J. Heart Fail 16 , 238–247.24464619 Mitchell SF , Jain S , She M , and Parker R (2013). Global analysis of yeast mRNPs. Nat. Struct. Mol. Biol. 20 , 127–133.23222640 Molliex A , Temirov J , Lee J , Coughlin M , Kanagaraj AP , Kim HJ , Mittag T , and Taylor JP (2015). Phase separation by low complexity domains promotes stress granule assembly and drives pathological fibrillization. Cell 163 , 123–133.26406374 Moon SL , Morisaki T , Khong A , Lyon K , Parker R , and Stasevich TJ (2019). Multicolour single-molecule tracking of mRNA interactions with RNP granules. Nat. Cell Biol. 21 , 162–168.30664789 Nguyen TA , Wu K , Pandey S , Lehr AW , Li Y , Bemben MA , Badger JD 2nd , Lauzon JL , Wang T , Zaghloul KA , (2020). A cluster of autism-associated variants on X-linked NLGN4X functionally resemble NLGN4Y. Neuron 106 , 759–768.32243781 Padron A , Iwasaki S , and Ingolia NT (2019). Proximity RNA labeling by APEX-seq reveals the organization of translation initiation complexes and repressive RNA granules. Mol. Cell 75 , 875–887.31442426 Patel A , Malinovska L , Saha S , Wang J , Alberti S , Krishnan Y , and Hyman AA (2017). ATP as a biological hydrotrope. Science 356 , 753–756.28522535 Patmore DM , Jassim A , Nathan E , Gilbertson RJ , Tahan D , Hoffmann N , Tong Y , Smith KS , Kanneganti TD , Suzuki H , (2020). DDX3X suppresses the susceptibility of hindbrain lineages to medulloblastoma. Dev. Cell 54 , 455–470.32553121 Phung B , Ciesla M , Sanna A , Guzzi N , Beneventi G , Cao Thi Ngoc P , Lauss M , Cabrita R , Cordero E , Bosch A , (2019). The X-linked DDX3X RNA helicase dictates translation reprogramming and metastasis in melanoma. Cell Rep. 27 , 3573–3586.31216476 Portz B , Lee BL , and Shorter J (2021). FUS and TDP-43 phases in health and disease. Trends Biochem. Sci. 46 , 550–563.33446423 Protter DSW , and Parker R (2016). Principles and properties of stress granules. Trends Cell Biol. 26 , 668–679.27289443 Qamar S , Wang GZ , Randle SJ , Ruggeri FS , Varela JA , Lin JQ , Phillips EC , Miyashita A , Williams D , Strohl F , (2018). FUS phase separation is modulated by a molecular chaperone and methylation of arginine cation-pi interactions. Cell 173 , 720–734.29677515 Robert X , and Gouet P (2014). Deciphering key features in protein structures with the new ENDscript server. Nucleic Acids Res. 42 , W320–324.24753421 Ruzzo EK , Perez-Cano L , Jung JY , Wang LK , Kashef-Haghighi D , Hartl C , Singh C , Xu J , Hoekstra JN , Leventhal O , (2019). Inherited and de novo genetic risk for autism impacts shared networks. Cell 178 , 850–866.31398340 Saito M , Hess D , Eglinger J , Fritsch AW , Kreysing M , Weinert BT , Choudhary C , and Matthias P (2019). Acetylation of intrinsically disordered regions regulates phase separation. Nat. Chem. Biol. 15 , 51–61.30531905 Scala M , Torella A , Severino M , Morana G , Castello R , Accogli A , Verrico A , Vari MS , Cappuccio G , Pinelli M , (2019). Three de novo DDX3X variants associated with distinctive brain developmental abnormalities and brain tumor in intellectually disabled females. Eur. J. Hum. Genet. 27 , 1254–1259.30936465 Schindelin J , Arganda-Carreras I , Frise E , Kaynig V , Longair M , Pietzsch T , Preibisch S , Rueden C , Saalfeld S , Schmid B , (2012). Fiji: an open-source platform for biological-image analysis. Nat. Methods 9 , 676–682.22743772 Schneider JW , Oommen S , Qureshi MY , Goetsch SC , Pease DR , Sundsbak RS , Guo W , Sun M , Sun H , Kuroyanagi H , (2020). Dysregulated ribonucleoprotein granules promote cardiomyopathy in RBM20 gene-edited pigs. Nat. Med. 26 , 1788–1800.33188278 Schuster BS , Dignon GL , Tang WS , Kelley FM , Ranganath AK , Jahnke CN , Simpkins AG , Regy RM , Hammer DA , Good MC , (2020). Identifying sequence perturbations to an intrinsically disordered protein that determine its phase-separation behavior. Proc. Natl. Acad. Sci. U.S.A. 117 , 11421–11431.32393642 Sharma D , and Jankowsky E (2014). The Ded1/DDX3 subfamily of DEAD-box RNA helicases. Crit. Rev. Biochem. Mol. 49 , 343–360. Shi B , Li W , Song YS , Wang ZJ , Ju R , Ulman A , Hu J , Palomba F , Zhao YF , Le JP , (2021). UTX condensation underlies its tumour-suppressive activity. Nature 597 , 726–731.34526716 Shih JW , Wang WT , Tsai TY , Kuo CY , Li HK , and Wu Lee YH (2012). Critical roles of RNA helicase DDX3 and its interactions with eIF4E/PABP1 in stress granule assembly and stress response. Biochem. J. 441 , 119–129.21883093 Silva JM , Rodrigues S , Sampaio-Marques B , Gomes P , Neves-Carvalho A , Dioli C , Soares-Cunha C , Mazuik BF , Takashima A , Ludovico P , (2019). Dysregulation of autophagy and stress granule-related proteins in stress-driven Tau pathology. Cell Death Differ. 26 , 1411–1427.30442948 Snijders Blok L , Madsen E , Juusola J , Gilissen C , Baralle D , Reijnders MR , Venselaar H , Helsmoortel C , Cho MT , Hoischen A , (2015). Mutations in DDX3X are a common cause of unexplained intellectual disability with gender-specific effects on Wnt signaling. Am. J. Hum. Genet. 97 , 343–352.26235985 Song H , and Ji X (2019). The mechanism of RNA duplex recognition and unwinding by DEAD-box helicase DDX3X. Nat. Commun. 10 , 3085.31300642 Soto-Rifo R , Rubilar PS , Limousin T , de Breyne S , Decimo D , and Ohlmann T (2012). DEAD-box protein DDX3 associates with eIF4F to promote translation of selected mRNAs. EMBO J. 31 , 3745–3756.22872150 Sun Z , Diaz Z , Fang X , Hart MP , Chesi A , Shorter J , and Gitler AD (2011). Molecular determinants and genetic modifiers of aggregation and toxicity for the ALS disease protein FUS/TLS. PLoS Biol. 9 , e1000614.21541367 Szaflarski W , Fay MM , Kedersha N , Zabel M , Anderson P , and Ivanov P (2016). Vinca alkaloid drugs promote stress-induced translational repression and stress granule formation. Oncotarget 7 , 30307–30322.27083003 Szappanos D , Tschismarov R , Perlot T , Westermayer S , Fischer K , Platanitis E , Kallinger F , Novatchkova M , Lassnig C , Muller M , (2018). The RNA helicase DDX3X is an essential mediator of innate antimicrobial immunity. Plos Pathog. 14 , e1007397.30475900 Tan RH , Ke YD , Ittner LM , and Halliday GM (2017). ALS/FTLD: experimental models and reality. Acta Neuropathol. 133 , 177–196.28058507 Tauber D , Tauber G , Khong A , Van Treeck B , Pelletier J , and Parker R (2020). Modulation of RNA condensation by the DEAD-box protein eIF4A. Cell 180 , 411–426.31928844 Vakilian H , Mirzaei M , Tabar MS , Pooyan P , Rezaee LH , Parker L , Haynes PA , Gourabi H , Baharvand H , and Salekdeh GH (2015). DDX3Y, a male-specific region of Y chromosome gene, may modulate neuronal differentiation. J. Proteome Res. 14 , 3474–3483.26144214 Valentin-Vega YA , Wang YD , Parker M , Patmore DM , Kanagaraj A , Moore J , Rusch M , Finkelstein D , Ellison DW , Gilbertson RJ , (2016). Cancer-associated DDX3X mutations drive stress granule assembly and impair global translation. Sci. Rep. 6 , 25996.27180681 Venkataramanan S , Gadek M , Calviello L , Wilkins K , and Floor SN (2021). DDX3X and DDX3Y are redundant in protein synthesis. RNA 27 , 1577–1588.34535544 Vernon RM , Chong PA , Tsang B , Kim TH , Bah A , Farber P , Lin H , and Forman-Kay JD (2018). Pi-Pi contacts are an overlooked protein feature relevant to phase separation. Elife 7 , e31486.29424691 Walker IH , Hsieh PC , and Riggs PD (2010). Mutations in maltose-binding protein that alter affinity and solubility properties. Appl. Microbiol. Biotechnol. 88 , 187–197.20535468 Wang J , Choi JM , Holehouse AS , Lee HO , Zhang X , Jahnel M , Maharana S , Lemaitre R , Pozniakovsky A , Drechsel D , (2018a). A molecular grammar governing the driving forces for phase separation of prion-like RNA binding proteins. Cell 174 , 688–699.29961577 Wang X , Posey JE , Rosenfeld JA , Bacino CA , Scaglia F , Immken L , Harris JM , Hickey SE , Mosher TM , Slavotinek A , (2018b). Phenotypic expansion in DDX3X - a common cause of intellectual disability in females. Ann. Clin. Transl. Neurol. 5 , 1277–1285.30349862 Watkins J , Ghosh A , Keerie AFA , Alix JJP , Mead RJ , and Sreedharan J (2020). Female sex mitigates motor and behavioural phenotypes in TDP-43(Q331K) knock-in mice. Sci. Rep. 10 , 19220.33154447 Wu T , Yu J , Gale-Day Z , Woo A , Suresh A , Hornsby M , and Gestwicki JE (2020). Three essential resources to improve differential scanning fluorimetry (DSF) experiments. bioRxiv. Xue B , Dunbrack RL , Williams RW , Dunker AK , and Uversky VN (2010). PONDR-FIT: a meta-predictor of intrinsically disordered amino acids. Biochim. Biophys. Acta. 1804 , 996–1010.20100603 Yedavalli VSRK , Neuveut C , Chi YH , Kleiman L , and Jeang KT (2004). Requirement of DDX3 DEAD box RNA helicase for HIV-1 Rev-RRE export function. Cell 119 , 381–392.15507209 Zhou Y , Zhou B , Pache L , Chang M , Khodabakhshi AH , Tanaseichuk O , Benner C , and Chanda SK (2019). Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat. Commun. 10 , 1523.30944313

PMC009xxxxxx/PMC9325914.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101133858 29668 J Clin Child Adolesc Psychol J Clin Child Adolesc Psychol Journal of clinical child and adolescent psychology : the official journal for the Society of Clinical Child and Adolescent Psychology, American Psychological Association, Division 53 1537-4416 1537-4424 35084265 9325914 10.1080/15374416.2022.2025597 NIHMS1771226 Article Engagement Barriers to Behavior Therapy for Adolescent ADHD Sibley Margaret H. 1 Link Kara 2 Antunez Gissell Torres 2 Greenwood Lydia 2 1 University of Washington School of Medicine, Seattle Children’s Research Institute, Seattle, WA 2 University of Washington Department of Psychology, Seattle Children’s Research Institute, Seattle, WA Address Correspondence to: Margaret H. Sibley, Ph.D., Seattle Children’s Research Institute, 1920 Terry Ave, Seattle, WA 98101, Phone: (206) 884-1424, margaret.sibley@seattlechildrens.org 21 1 2022 27 1 2022 27 7 2023 116 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Objective: To identify barriers to behavior therapy for adolescent ADHD (Supporting Teens’ Autonomy Daily; STAND) and understand the relationship between barriers and treatment engagement. Method: A mixed-method design with qualitative coding of 822 audio-recorded therapy sessions attended by 121 adolescents with ADHD (ages 11-16; 72.7% male, 77.7% Latinx, 7.4% African-American, 11.6% White, non-Latinx) and parents. Grounded theory methodology identified barriers articulated by parents and adolescents in session. Barriers were sorted by subtype (cognitive/attitudinal, behavioral, logistical) and subject (parent, teen, dyad). Frequency and variety of barriers were calculated by treatment phase (engagement, skills, planning). Generalized linear models and generalized estimating equations examined between-phase differences in frequency of each barrier and relationships between barriers frequency, subtype, subject, and phase on engagement (attendance and homework completion). Results: Coding revealed twenty-five engagement barriers (ten cognitive/attitudinal, eleven behavioral, four logistical). Common barriers were: low adolescent desire (72.5%), parent failure to monitor skill application (69.4%), adolescent forgetfulness (60.3%), and adolescent belief that no change is needed (56.2%). Barriers were most commonly cognitive/attitudinal, teen-related, and occurring in STAND’s planning phase. Poorer engagement was associated with cognitive/attitudinal, engagement phase, and dyadic barriers. Higher engagement in treatment was predicted by more frequent behavioral, logistical, parent, and skills/planning phase barriers. Conclusions: Baseline assessment of barriers may promote individualized engagement strategies for adolescent ADHD treatment. Cognitive/attitudinal barriers should be targeted at treatment outset using evidence-based engagement strategies (e.g., Motivational Interviewing). Behavioral and logistical barriers should be addressed when planning and reviewing application of skills. pmcEvidence-based behavioral treatments (EBTs) for adolescent Attention Deficit Hyperactivity Disorder (ADHD) typically teach coping skills to enhance youth executive functioning while leveraging parent-delivered motivational strategies, such as contingency management, to promote application of learned techniques (Langberg et al., 2012; Sibley et al., 2016a; Sprich et al., 2016). Compared to school-aged children, adolescents with ADHD are less likely to utilize EBTs (Danielson et al., 2018), with a minority receiving any form of ADHD treatment (behavior therapy or medication) in the past year (Bussing et al., 2011; Molina et al., 2009). Adolescents demonstrate lower willingness to participate in behavior therapy for ADHD compared to their parents and teachers (Bussing et al., 2012). Although reasons for adolescent medication disinterest are well-known (Brinkman et al., 2018), less is known about factors that reduce utilization of behavior therapy among adolescents with ADHD. Poor utilization of mental health treatment is likely a consequence of both access and engagement barriers. With respect to ADHD, a large body of research characterizes treatment access difficulties (for review see Sayal et al., 2017; Wright et al., 2015). In line with the general mental health treatment literature, patient and contextual factors such as poverty, ethnic minority status, female gender, low parent motivation, parent mental health difficulties, and poor availability of treatments often prevent initial receipt of care (Bussing et al., 2012; Sibley et al., 2016b; Smith et al., 2014). However, far less is known about barriers to engagement in EBTs for ADHD—or continued participation in services after initial uptake. In community clinics, ADHD diagnosis is a risk factor for premature discontinuation of youth services (Johnson et al., 2008). When adolescents are engaged in family-based behavior therapy for ADHD, dropout is traditionally high (Barkley et al., 2001). Although the adult caregiver is a critical participant in behavior therapy for adolescent ADHD, their treatment attendance is poor in community settings (Sibley, Graziano, Coxe, et al., 2021). Providers perceive logistical barriers (i.e., transportation, scheduling difficulties) to undermine parent engagement (Sibley, Ortiz, et al., 2021), and teens with higher family adversity show lower engagement in behavior therapy for ADHD (Sibley, Coxe, et al., 2021); yet parent and youth perspectives on these challenges are lacking. Therapy engagement is multifaceted and includes indices such as attendance, alliance, treatment credibility, active participation in sessions, and homework completion (Becker et al., 2015, 2018; Garland et al., 2012; Stevens et al., 2006). Like treatment access barriers (Brown et al., 2016; Owens et al., 2002; Reardon et al., 2017), engagement barriers may be cognitive (e.g., negative beliefs about treatment or one’s ability to improve, cultural worldviews), behavioral (e.g., parental interference behaviors, youth forgetfulness), and/or logistical (e.g., insufficient time to complete homework, transportation needs; Aggarwal et al., 2015; Baker-Ericzen et al., 2013; Becker et al., 2021; Buckingham et al., 2016; Kazdin & Mazurek, 1994; Kerkorian et al., 2006; McKay & Bannon, 2004). Chacko et al. (2017) demonstrated that cognitive factors such as attitudes about treatment, one’s parenting abilities, and one’s child predict engagement in EBT for childhood ADHD. Schneider et al. (2013) reported that child behavioral factors, such as problem severity, also undermined EBT engagement for school-aged youth with ADHD. When patients and their parents experience frequent barriers, resulting engagement problems predict poorer response to care (Clarke et al., 2015; Dvorsky et al., 2021; Haine-Schlagel & Walsh, 2015; Kazdin & McWhinney, 2018; Kazdin & Wassell, 1999; Salloum et al., 2016). Although several studies investigate barriers related to childhood ADHD treatments, this topic is largely uninvestigated in adolescence, when treatment shifts dramatically from a parent-focused to a youth-engaged approach (Evans et al., 2017). Knowledge of population and treatment-specific barriers is valuable to clinicians implementing EBTs and promotes selection of patient-specific engagement strategies (Becker et al., 2015, 2018, 2021; Lindsey et al., 2014). Supporting Teens’ Autonomy Daily (STAND; Sibley, 2016; Sibley et al., 2016) is an engagement-focused EBT for adolescent ADHD that was designed to mitigate known population-specific barriers such as adolescent disinterest in treatment (Bussing et al., 2012) and parent under-involvement (Sibley, Graziano, Coxe et al., 2021). STAND utilizes a range of evidence-based engagement strategies designed to address cognitive, behavioral, and logistical barriers during treatment (Becker et al., 2015, 2018). For example, cognitive barriers are directly addressed through Motivational Interviewing and a strength-based communication style (Miller & Rollnick, 2013). Logistical barriers are directly addressed through therapist progress monitoring and cognitive walkthrough of weekly homework assignments, with an emphasis on problem-solving potential barriers during assignment of homework. Additional engagement strategies that may mitigate cognitive, behavioral, and logistical barriers include shared parent-teen decision-making, patient and parent goal setting, assessment feedback, psychoeducation, and modular/menu-based treatment (Sibley, 2016). Thus, engagement is targeted in all phases of treatment, and not just at the outset. STAND includes a two-session engagement phase designed to identify meaningful patient and parent goals, cultivate motivation to change, provide clarity on the treatment process, and select treatment modules from a menu of options. During STAND’s four-session skills phase, adolescents receive coping skills training and parents are trained to oversee and reward skills practice outside of session. STAND’s final four sessions are a planning phase that promotes integration of new skills into a daily routine with continued parent reinforcement of practice and progress (Sibley, 2016; Sibley, Johansson et al., 2021). MI is blended throughout treatment to mobilize behavior change, particularly when homework is assigned and reviewed, and new skills are introduced. In spite of STAND’s engagement-focused approach, patient and parent engagement remains variable. In a randomized community-based trial of STAND, parents attended an average of 74% of parent-teen sessions, while adolescent satisfaction with treatment has remained moderate across university-based, telehealth, and community-based trials (Sibley et al., 2013, 2016, 2020; Sibley, Graziano, Coxe, et al., 2021). Therefore, continued analysis of population and treatment-specific engagement barriers is warranted. Identification of key factors that influence treatment engagement will promote refinement of embedded engagement strategies and recommendations for devising patient-specific engagement plans in the treatment of adolescent ADHD more broadly. The current study is a mixed-methods investigation of engagement barriers encountered during STAND. Participants were 121 adolescents with ADHD and their parents who received STAND at a university clinic as a part of two randomized controlled trials (Sibley et al., 2016; Sibley, Rodriguez, et al., 2020). We conducted content analysis of 822 audio-recorded STAND sessions to detect treatment barriers uttered by parents and teens during sessions and to sort these barriers into thematic categories. Following utterance extraction, category construction, and qualitative coding, we examined frequency and variety of barriers experienced by families across STAND’s three phases (engagement, skills, and planning), distribution of these barriers across three subtypes (cognitive/attitudinal, behavioral, logistical), and three subjects (i.e., whether barriers were parent-specific, teen-specific, or mutual to the dyad). Finally, we examined the relationship between two indices of engagement (i.e., attendance and homework completion) and barriers frequency across phase, subtype, and subject. Consistent with the general literature on mental health treatment engagement, we hypothesized that a range of barriers would be noted across phases, subjects, and subtypes. We hypothesized that cognitive barriers would be more frequent in the engagement phase and would reduce over time as participants entered the skills and planning phases, whereas behavioral and logistical barriers would be more frequent in the skills and planning phases than the engagement phase, when participants engaged in skill application outside of session (i.e., therapy homework). Finally, we hypothesized that higher barriers frequency would predict lower engagement across phases of treatment and barriers subtypes and subjects. Ultimately, the goal of this project was to identify common barriers to treatment for adolescents with ADHD, understand how these barriers impact engagement across the various phases behavior therapy, and support the development a menu of individualizable engagement strategies for adolescents engaging in ADHD treatment. In addition, it is hoped that the engagement barriers detected herein will inform a broader focus on engagement across evidence-based mental health treatments in the adolescent developmental phase. Method Participants Participants were 121 adolescents with ADHD (ages 11-16), and their parents, who participated in one of two randomized controlled trials (RCTs) that evaluated the efficacy of STAND between 2012-2016. This subgroup of participants possessed at least one full-length, audible audio recording of a STAND session and all available recorded sessions were utilized in the current study. The number of tapes per participant ranged from 1 to 10 (M=6.79, SD=2.44). A total of 822 tapes were analyzed. As a part of respective study inclusion criteria, all participants met DSM criteria for ADHD during a comprehensive psychiatric evaluation conducted by the research team that included a structured parent interview (Shaffer et al., 2001) and parent and teacher symptom and impairment ratings (Fabiano et al., 2006; Sibley & Kuriyan, 2016). Informant reports were integrated and reviewed by licensed clinical psychologists using an “or rule” that established ADHD symptoms to be present if endorsed by any informant (Bird et al., 1992). Autism spectrum disorders and intellectual disability were exclusionary in both trials. Participants with all other comorbidities were included. All study participants were permitted to continue stimulant medication and special education interventions at school during the trials, which were monitored for inclusion in study analyses when relevant. Demographic characteristics of the current subsample (N=121) are presented in Table 1. This subsample excluded nine participants from the eligible participant pool (N=130) who did not possess any available audio recordings due to human error, inaudible recordings, and/or failure to complete sessions. 81.9% of sessions that occurred in the study were available for analysis. Tapes were unavailable for 17.6% of engagement phase sessions, 13.2% of skills phase sessions, and 21.4% of planning phase sessions. Procedures Recruitment and Data Collection. All procedures were approved by the university’s institutional review board. In both studies, participants were recruited through referral from local schools and parent inquiry at the university clinic. Prospective participants completed a screener containing DSM ADHD symptoms and questions about impairment that was administered to a parent. Families were invited for an eligibility assessment if the parent endorsed four or more of either inattention or hyperactivity/impulsivity symptoms and clinically significant impairment (at least a 3 on a 0 to 6 impairment scale). At the eligibility assessment, informed parental consent and youth assent were obtained and participant demographic and clinical information was assessed. All participants in both studies were randomized to receive 10 sessions of STAND from a clinician in a university clinic (pre or post-doctoral trainee, clinical psychologist, or master’s level clinician). For more information about the trials that contributed data to the present study, see Sibley et al., 2016 and Sibley, Rodriguez, et al., 2020. Clinicians in both trials were instructed to audio record all therapy sessions and submit them to a secure computer drive immediately post-session. Treatment Description. STAND is manualized and consists of ten one-hour weekly sessions attended by the adolescent and parent. Treatment was offered in English or Spanish. 19.7% of recorded sessions (n=162) were delivered in Spanish. Therapists received three days of training prior to implementing treatment. Training consisted of small group discussions, role plays, video demonstrations, skill building games, didactic instruction, and individualized feedback. The treatment is fully described elsewhere (Sibley, 2016). In STAND, client skill instruction is blended with Motivational Interviewing (MI; Miller & Rollnick, 2013) and guided parent-teen behavioral contracting. Sessions move through three phases. In the engagement phase, MI increases awareness of personal values and goals, identifies strengths, and recognizes ways to achieve personal goals and act consistently with values. The skills phase teaches parent behavioral strategies and OTP skills applied to homework, school, and chores. Treatment is modular; families and therapists collaboratively select skills relevant to families’ personal goals. Rather than a didactic approach, the skills phase introduces strategies (see Table 2) using MI strategies that emphasize dyadic autonomy and shared-decision making. This phase includes guided parent-teen contracting, in which MI is incorporated to build commitment to weekly skill practice outside of session. As part of this process, therapists review adherence to contracts, building recognition of progress, confidence in treatment, and awareness of consequences of parent and teen choices. Planning sessions teach families to integrate skills into a daily routine, transfer new habits to school settings, and build a final parent-teen contract. In the current study, the pace of treatment was standardized such that all therapists had one session to complete each module. The sequence of modules was fixed for the engagement and planning phases, but was allowed to vary within the skills phase. Session Coding. Although sessions were designed to be completed in 50- to 60-minute blocks, available sessions varied slightly in length. Thus, to standardize the number of minutes per session, the first twenty minutes and the last twenty minutes were coded for each STAND session, leading to 40 minutes of coded content per session. Extraction of utterances for each session was completed using the CASAA Application for Coding Treatment Interactions (CACTI; Glynn et al., 2012). CACTI facilitates the real-time sequential coding of behavioral interactions using digital audio files. It operates transcript free and assists in moment-to-moment parsing, sequential coding, and global rating of audio-recorded psychotherapy sessions (Glynn et al., 2012). This study coded parent and teen reported barriers to treatment engagement. Bachelor’s level research assistants extracted relevant units of parent and teen speech in CACTI utilizing a standardized procedure that specified how to identify discrete units of speech (i.e., a complete thought, or a thought unit; Houck et al., 2010), discriminate speech units that were relevant versus irrelevant to research questions, extract and supplement contextual information needed for the interpretation of quotes, and apply decision rules. Spanish speaking coders extracted information from therapy sessions conducted in Spanish. Three coders contributed to category construction using a grounded theory framework (Charmaz, 2014). Coding began with open coding. The focus of this procedure is creating a consensus list of codes that saturate available data (for examples see: Abba et al., 2008; Schraw et al., 2007). First, the three coders independently sorted all extracted speech units into initial groupings based on comparing speech units and asking the question, “is this the same or different?” This process was completed for the first 50% therapy sessions, with each coder taking careful notes on the code generation process. Next, the three coders met with a fourth coder (first author) and reconciled the proposed categories to create a consensus list of categories and codebook with category names, definitions, and rare, challenging, and typical examples of each code. The three coders coded the second 50% of interviews using the new codebook, refining the codebook as necessary. In the final step, all statements were recoded using the finalized codebook, which organized the final list of codes into three types. Cognitive codes were barriers that represented attitudes or beliefs. Logistical codes were barriers that represented environmental challenges external to the patient or their parent. Behavioral codes were barriers that represented patient or parent behaviors. Codes were further sorted with respect to whether they referred to the teen only, the parent only, or the dyad mutually. Codes were orthogonal meaning that each unit of speech was offered a unique code. Twenty percent of speech units were double coded to assess inter-rater reliability. Intraclass correlations (ICCs) across categories ranged from .63 to 1.00; average ICC was .87 indicating excellent inter-rater reliability (Cicchetti, 1994). Full information about code-specific ICCs is reported in an online supplement. Role of Researchers in Coding. All authors of this paper contributed to category construction and completion of the codebook. As a researcher in the field of ADHD, the first author may possess presuppositions about the constructs of interest that may influence the lens through which they view the data. They may have preferences for certain informant statements that support these presuppositions, which may lead to blind spots or a priori judgments. Validity enhancing procedures included independent coding of interviews by three coders who were not informed of study hypotheses, linking codes to direct quotations (see Results), and recoding interviews to ensure full saturation of data. Measures Engagement. Providers in each study kept detailed attendance records for all sessions. Providers also marked whether homework was fully completed (1 point), partially completed (.5 points), or not completed (0 points) each week. Attendance was the percentage of study-assigned sessions attended by the recipient. Homework completion was the average homework completion score for the sessions attended by the family. Analytic Plan All analyses were performed in SPSS version 26. Barriers Prevalence, Frequency, and Variety. After defining the categories, we calculated the prevalence of each and sorted them into three empirically-informed subtypes: cognitive/attitudinal, behavioral, and logistical and subjects: parent, teen, and dyadic. Next, we calculated mean per session frequency and variety scores by dividing the total frequency (or variety) of barriers across treatment by the total number of sessions attended by the dyad. We also compared within-subject differences in barriers frequency by subtype (cognitive, behavioral, or logistical) and subject (parent, teen, dyad) using two separate repeated-measures general linear models. Barriers by Phase. Average per-session frequency and variety scores were also calculated for STAND’s three phases: engagement (two sessions), skills (four sessions), and planning (four sessions). Generalized estimating equations using a linear distribution, identity link function, and a maximum likelihood estimator were utilized to model within-subject effects of phase on frequency and variety of engagement barriers. We also utilized generalized estimating equations with a binomial distribution, a logit-link function, and a maximum likelihood estimator to examine between-phase differences in the incidence of each barrier, with a phase-varying covariate representing the number of sessions a participant received in each phase. Predictors of Engagement. We examined the influence of three sets of predictors: (a) barriers subtype (cognitive/attitudinal, behavioral, logistical), (b) subject (parent, teen, dyad), and (c) phase (engagement, skills/planning) on two indices of engagement: (1) attendance and (2) homework completion using generalized linear models. Six separate models were conducted (two dependent measures x three sets of predictors). Attendance was modeled as a count variable with a Poisson distribution, log-link function, and maximum likelihood estimator. Homework completion was modeled using a linear distribution, identity link function, and a maximum likelihood estimator. Average per session barriers frequency across the skills and planning phases were combined to minimize the impact of missing data. Only participants with at least one skills/planning phase recording (n=108) were included in the two models that examined phase. Results Barriers Categories Coding revealed twenty-five categories of barriers to STAND engagement that included ten categories of cognitive/attitudinal barriers, eleven behavioral, and four categories of logistical barriers. There were six categories of parent barriers, 14 teen barriers, and 5 dyadic barriers. Each category is listed in Table 2 by subtype and subject and described below. Low teen desire for treatment. Family members described low adolescent desire for treatment as interfering with engagement. For example, some teens noted not liking writing in the planner, not wanting to get organized, and disliking attending sessions. One teen stated: “No, I never wanted to do a program like this.” Parental failure to monitor teen practice. Parents struggled to monitor their teen in different ways, including not monitoring homework or grades, forgetting to be consistent with monitoring or participating in skills, as well as not checking the agenda. There were five sub-categories for this code that indicated the reasons for monitoring failures: parent forgetfulness, impatience, too busy, too tired, and unspecified reason. Some comments from parents included: “I forgot to give him his [reward] money.” “Like last night, I just sent him to bed because I got tired of waiting [for him to finish].” “I don’t have the time to do it every day.” “I feel stressed out when I have to be on top of him all of the time.” “I find myself not checking his bookbag.” Forgetfulness. Adolescents did not remember to complete a task or apply a skill from treatment. Teens reported that they forgot to complete an assignment, to bring necessary supplies to session, and to utilize skills such as writing in their agenda. Some teens described forgetfulness as a recurring issue: “The problem is that I always forget to write in it [the planner].” Teen belief that no change is needed. Some teens did not see a need to change their behavior in treatment. Participants were satisfied with their academic performance or work ethic and viewed treatment as unnecessary. For example, some stated that they did not need to take notes, to use a planner, or to write in their agenda. One student was satisfied with their grades as long as they were passing. Low teen self-efficacy. Adolescents described disbelief in their ability to improve presenting problems or integrate skills. One teen stated that “there’s not much [he] can do” to improve his organization. Another student rated the likelihood of meeting his goals as a 1 on a scale from 1 (low confidence) to 10 (high confidence). Parent Intrusiveness. Some parental behaviors inhibited teens from completing the skills independently. Examples of this included some parents who would complete activities for the teen, such as finishing their homework or projects, giving constant reminders for them to do their homework or write in an agenda, and managing the teen’s schedule for them. Low teen confidence in treatment. Some teens viewed treatment activities as ineffective and unnecessary, which decreased desire to participate. For example, some teens did not believe they would benefit from note taking, that therapy would help them, or that utilizing a planner or agenda would improve organization. For example, one teen stated: “Even though it’s some trouble looking to find what you need, you’ll always find it, so there’s no point in a binder” Schedule Conflicts. Familial schedule conflicts limited practice of activities or skills, including attending treatment. Recurrent reasons for the lack of task completion were, “we’ve had a busy week,” “we’ve had a lot going on,” and “we were on vacation.” Insufficient time. A lack of sufficient time to complete planned practice activities prevented teens from completing tasks or skills. Teens described not having enough time to practice treatment activities due to busyness and competing school assignments. Some students did not have sufficient time to use treatment skills during class due to the fast pace of the class or because other time-sensitive assignments were prioritized. Teen side-tracked during practice. Families described distraction and trouble focusing as barriers to skills practice. More desirable activities interfered with adolescents’ ability to focus on and complete therapy homework and skills practice, and this distraction took place both at home and in school. One teen noted: “Sometimes when I’m doing my homework, I get distracted; I’ll usually come out and cook dinner, watch Netflix, stay there for like an hour, then go back to doing my homework.” Failure to Find Common Ground. Some parents and teens failed to find mutual understanding which prevented them from successfully completing tasks. Some examples of this included arguments about why they were in the program and disagreements on how to complete tasks. A parent mentioned: “When it came to the project, I wanted it my way now because her way obviously didn’t work.” Teen skill deficit. Adolescents were unable to complete tasks due to a lack of understanding or ability. Some teens described a lack of understanding of class material and difficulties with skills like note taking or reading comprehension. For example: “No like, sometimes I don’t understand it. And I look back at my notes and I don’t understand them either.” “I am not good with reading abbreviations.” Inconsistent contingencies. Some parents failed to withhold rewards and/or privileges from teen when the teen had failed to do their tasks. Parents recalled allowing teens to receive rewards even if the teen had not completed their end of the contract. Parents reported not taking away electronics and “giving in” when the teen rebelled or threw a tantrum. Dyadic Conflict. Arguments between the teen and the parents prevented them from successfully completing tasks. Both the teens and parents describe getting into arguments and fights about homework, tasks, and contract rules. For example, one parent described: “Because of the argument that ensued, I walked away and didn’t supervise the rest of the process” Tired. A lack of energy interfered with treatment participation. Teens stated that they were too tired to complete their homework at the end of the day or to practice skills outside of sessions. Teens’ tiredness was attributed to multiple factors such as the demands of school, a lack of sleep, or having too many outside commitments. Fatigue affected motivation for some participants. One teen described this effect: “Yesterday was one of the times where I didn’t want to do anything because I was super tired.” Dislike of materials. For some teens, they did not like using some of the physical materials, including planners, bookbags, folders, or other items needed for different skills or tasks. For example, some disliked the number of folders they needed, complaining that they took up too much space or would take too much time to use. Another teen complained about the increased weight of their bookbag when they had to bring the extra materials recommended in STAND. Lack of Supplies. Some teens were unable to successfully complete tasks due to the lack of correct materials. Parents and teens expressed not having a planner, notebooks, and other supplies to complete the activity. One parent stated: “I didn't buy these right away after the session last week after work because I was very busy and I don’t live next to a (retail store), so I have to drive either north or south to (retail store), none of it was on my way” Electronics Overuse. Electronics affected the teens’ engagement in treatment by creating a distraction that teens chose instead of working on their skills. Sometimes the teens spent too much time on their electronics, decreasing the amount of time spent on skills practice. One teen’s statement exemplified this: “What kinds of things do you do instead of what you need to do? Video games.” Parental Low Confidence in Treatment. Some parents expressed that the treatment plan was ineffective or too complicated to successfully complete. Parents stated not trusting the reward system and/or not believing that the treatment was going to make a lasting impact. Some of this was due to past treatment failures. Demands on Parents. Parents sometimes found the treatment plans too complicated to implement into their routines. Parents described difficulties in following through with contracts and monitoring the teen’s tasks. For example, one parent described: “It’s too many steps. I have a feeling it’s not going to make it here or not going to make it back.” Ineffective rewards. Rewards offered by their parents did not motivate some teens to practice skills. For some, the teen did not like the reward offered or they did not believe that getting the rewards mattered. One parent stated: “I never rewarded him; he didn’t seem to care.” Poor school engagement. Teens and parents described difficulties with the school failing to cooperate on treatment interventions. Some teens stated teachers refusing to sign their agenda, lack of communication from teachers, and teacher unwillingness to work with the teen. One teen stated: “It’s not me it’s the teacher. I do the assignments, some I don’t do because I don’t know how to do them, but others I do, but she doesn’t understand that I’m Hispanic and don’t know how to speak English, so she goes and gives me bad grades.” Embarrassed to get school involved. Some teens did not want their teachers or peers to know that they were in the program or to witness them participating in the extra tasks, as the teens thought it might be embarrassing or awkward. This would often lead to them to not participate in the tasks that might be visible to others, such as having the teacher sign their agenda or writing due dates for assignments. For example, one teen said: “I don’t want to [have teacher sign their agenda], I don’t want to be embarrassed.” Cheating. Teens were reported to cheat on parent-teen contracts by secretly breaking the rules, using electronics without permission, and forging their parents’ signatures, among other behaviors. Examples of this were: “The only thing I do in secret is play games.” “I did the thing that I wanted to do before my homework.” Anti-rewards belief. Some parents expressed their beliefs against the use of rewards for the teen. They believed the rewards were excessive and/or that the teen would not respond to the rewards. Some parents said: “If he does all his homework like how he should, I don’t think he should receive any reward.” “I don’t believe that he should constantly get rewarded for doing stuff that is necessary.” Barriers Prevalence, Frequency, and Variety With respect to prevalence (see Table 2), engagement barriers articulated by a majority of families included adolescent low desire for treatment (72.5%), parent failure to monitor teen practice (69.4%), adolescent forgetfulness (60.3%), and adolescent belief that no change is needed (56.2%). On average, an engagement barrier was articulated 3.23 times per 40-minute block of coded session audio (SD=2.36), encompassing an average of 1.77 categories per session (SD=.99). Participants experienced a significantly higher frequency of cognitive/attitudinal barriers per session (M=1.59, SD=1.78) than behavioral [M=1.22, SD=.86, F(1, 120)=5.05, p=.027, d=.28] and logistical barriers [M=.16, SD=.20, F(1, 120)=80.57, p<.001, d=1.44]. Participants also experienced a higher frequency of behavioral versus logistical barriers per session [F(1, 120)=186.45, p<.001, d=2.00]. There were more frequent teen barriers observed per session (M=1.98, SD=1.85) than parent barriers [M=.78, SD=.69, F(1, 120)=45.99, p<.001, d=.94] and dyadic barriers [M=.22, SD=.38, F(1, 120)=118.43, p<.001, d=1.57]. There also were more frequent parent barriers observed per session than dyadic barriers [F(1, 120)=78.39, p<.001, d=1.04]. Barriers by Phase Generalized estimating equations revealed greater frequency of barriers per session in the planning phase (model-estimated M=4.37, SD=3.69) versus the skills phase (model-estimated M=3.19, SD=2.52; b=−1.18, SE=.32, p<.001, d=.47) and the engagement phase (model-estimated M=2.04, SD=2.68; b=−2.33, SE=.42, p<.001, d=.87). Barriers per session were also more frequent in the skills phase compared to the engagement phase (b=1.14, SE=.28, p<.001, d=.39). There was also lower variety of barriers per session in the engagement phase (model-estimated M=1.20, SD=1.19) versus the skills phase (model-estimated M=1.96, SD=1.17; b=.75, SE=.12, p<.001, d=.64) and the planning phase (model-estimated M=1.83, SD=1.15; b=.62, SE=.15, p<.001, d=.53). The skills and planning phases did not differ in variety of barriers per session (p=.273). Table 3 displays model estimates of the prevalence of each category of barrier by phase. Model-estimated results indicated lower prevalence of forgetfulness (18%) in the engagement phase versus the skills phase (35%; b=.91, SE=.39, p=.019) and planning phase (37%; b=.97, SE=.37, p=.008), as well as lower parent failure to monitor skill practice in the engagement phase (20% vs. skills: 43%, b=1.07, SE=.45, p=.017; vs. planning: 57%, b=1.64, SE=.39, p<.001) and fewer incidences of inconsistent contingency management in the engagement phase (3% vs. skills: 13%, b=1.55, SE=.77, p=.044; vs. planning: 13%; b=1.54, SE=.72, p=.032). Compared to the skills phase, the planning phase demonstrated a higher incidence of parental intrusiveness (24% vs. 36%; b=−.62, SE=.28, p=.027) and fewer instances of the adolescent getting side-tracked during skill practice (16% vs. 6%; b=1.05, SE=.46, p=.022). Compared to the engagement phase, the planning phase demonstrated more frequent schedule conflicts (4% vs. 15%; b=1.55, SE=.66, p=.019) and lower incidence of the adolescent belief that no change is needed (44% vs. 25%; b=−.88, SE=.33, p=.007). Prediction of Engagement Attendance. Attendance ranged from 1 to 10 sessions (M=8.71, SD=2.53) and 72.7% of participants completed all ten STAND sessions. For the subtype model, the overall likelihood ratio chi-square test was significant [χ2(3)=49.87, p<.001]. Lower frequency of cognitive/attitudinal barriers (b=−.02, SE=.01, p<.001) and higher frequency of behavioral (b=.02, SE=.01, p=.042) and logistical barriers (b=.31, SE=.05, p<.001) predicted higher attendance. For the subject model, the overall likelihood ratio chi-square test was significant [χ2(3)=31.35, p<.001]. Higher frequency of parent barriers (b=.05, SE=.01, p=.002) and lower frequency of dyadic barriers (b=−.16, SE=.03, p<.001) predicted higher attendance. Frequency of teen barriers did not predict attendance (p=.184). For the phase model, the overall likelihood ratio chi-square test was significant [χ2(3)=26.91, p<.001]. Lower frequency of barriers in the engagement phase (b=−.20, SE=.00, p<.001) and higher frequency of barriers in the skills/planning phases (b=.02, SE=.00, p<.001) predicted higher attendance. Homework Completion. Homework completion scores ranged from 0 to 1 (M=.71, SD=.20). For the subtype model [χ2(3)=5.23, p=.156] and the subject model [χ2(3)=6.65, p=.084] the overall likelihood ratio chi-square tests were non-significant; all predictors were also non-significant. For the phase model, the overall likelihood ratio chi-square test was significant [χ2(3)=7.60, p=.022]. Lower frequency of barriers in the engagement phase (b=−.19, SE=.01, p=.008) predicted higher homework completion. Frequency of barriers in the skills/planning phase did not predict homework completion (p=.947). Discussion The primary findings of this study were as follows: (a) a wide range of cognitive/attitudinal, behavioral, and logistical barriers were observed during behavior therapy sessions, which were either parent-specific, teen-specific, or mutual to the dyad; (b) observed barriers were most frequently teen-related (d=.94 to d=1.57), cognitive/attitudinal (d=.28 to d=1.44), and occurring during STAND’s planning phase (d=.47 to d=.87); (c) common barriers to adolescent ADHD treatment included teen low desire to change, forgetfulness, and belief that no change is needed, as well as parent failure to monitor teen skill practice; (d) teen forgetfulness and parent problems with monitoring and contingency management were most relevant in the skills and planning phase, while teen becoming side-tracked during practice and belief that no change was needed decreased over the phases of treatment; (e) lower engagement in treatment was associated with more frequent cognitive/attitudinal barriers, engagement phase barriers, and dyadic barriers, while conversely, higher engagement in treatment was predicted by more frequent behavioral, logistical, parent, and skills/planning phase barriers. We discuss these findings in turn below. Twenty-five unique barriers to adolescent ADHD behavior therapy engagement were identified in this investigation. Most families experienced low teen desire to change (75.2%), parent failures to monitor teen skill practice (69.4%), teen forgetfulness (60.3%), and teen beliefs that no change is needed (56.2%). These difficulties highlight how teen motivation difficulties and executive function deficits are both the target of and primary barriers to successful behavior therapy for adolescent ADHD. The ubiquity of teen cognitive/attitudinal barriers is not surprising and is consistent with past observed barriers to adolescent ADHD behavior therapy (Barkley et al., 2001), survey data on adolescent attitudes toward ADHD treatment (Bussing et al., 2012), and general findings on barriers to youth mental health treatment (Buckingham et al., 2016). Motivational Interviewing (MI; Miller & Rollnick, 2013) in STAND addresses adolescent attitudinal barriers by increasing adolescent openness to behavior change. In the present study, teen belief that no change was needed decreased significantly from the engagement phase (44%) to the planning phase (25%), indicating that these efforts may be effective for some youth. Improvement in teen attitudes toward treatment is consistent with data from a study of telehealth-delivered STAND, which reported large improvements in adolescent readiness to change over ten weeks of treatment (Sibley et al., 2017). There is also evidence that early-treatment MI sessions enhance engagement in adolescent mental health interventions more broadly (Dean et al., 2016), particularly when engaging parents in treatment (Ingoldsby, 2010). Forgetfulness is an executive functioning challenge that is commonly associated with ADHD (American Psychiatric Association, 2013) and was previously identified as a primary barrier to engagement in school-based treatment for adolescent ADHD (Sibley, Morley, et al., 2020). In a study of a peer-delivered intervention for high school students with ADHD, many teens initially forgot to attend to intervention sessions until peer interventionists were trained to fetch students from their classes (Sibley, Morley et al., 2020). In family-based behavior therapy for adolescent ADHD, parent monitoring and contingency management is designed to mitigate adolescent forgetfulness; however, this approach hinges upon consistent parent implementation of these strategies—which appeared to be challenging for many parents (see Table 2). To address this challenge, STAND’s MI approach targets increased parent commitment to implementation of behavioral strategies. Although parents frequently articulated challenges with monitoring and contingency management, STAND still produced large increases in these strategies compared to treatment as usual (Sibley et al., 2016). Behavioral barriers (e.g., teen forgetfulness, parent difficulties with monitoring and contingency management) were most frequent in the skills and planning phase. In the planning phase, families experienced the highest frequency and variety of treatment barriers as they created and implemented behavior plans designed to generalize new skills to home and school. The planning phase is also when community clinicians demonstrated lowest fidelity to STAND’s manualized procedures (Sibley, Graziano, Bickman, et al., 2021) and may represent the most challenging treatment phase for clinicians and families. The planning phase also demonstrated higher frequency of schedule conflicts and parental interference behaviors. However, adolescents appeared to be sidetracked less frequently in the planning phase (compared to the skills phase). It is possible that increased mastery of skills promotes habitual practice in the later sessions of treatment, as is the case for other behavior therapies for adolescent ADHD (Evans et al., 2009; Langberg et al., 2013). It is also possible that behavioral barriers are particularly low in the engagement phase because there are almost no skill application demands that might elicit these barriers in early sessions. Cognitive/attitudinal and dyadic barriers observed during STAND sessions significantly predicted attendance problems, while engagement phase barriers significantly predicted both attendance problems and difficulties with homework completion. These results extend upward the findings of Chacko et al. 2017, who demonstrated the key role of parent cognitive beliefs in EBT engagement for children with ADHD. They also support the suspicions of Barkley et al. (2001), who speculated that negative teen attitudes toward treatment likely were responsible for higher dropout in a parent + teen behavior therapy condition compared to a parent only treatment for adolescent ADHD. The relationship between dyadic barriers (e.g., parent-teen conflict, failure to find common ground, schedule conflicts) and engagement underscores the importance of parent-teen collaboration in family-based behavior therapy for adolescent ADHD. STAND demonstrates stronger efficacy than traditional group parent training and teen organization skills training models when parent-teen conflict is high (Sibley, Rodriguez, et al., 2020). However, treatment disengagement may still occur in STAND when parents and teens struggle to collaborate effectively. The salience of engagement phase barriers in predicting future engagement problems suggests that barriers that are unresolved in early sessions (particularly cognitive/attitudinal) may undermine engagement in later sessions. For example, a parent who leaves the engagement phase with a continued belief that treatment is unlikely to be effective may fail to engage in subsequent weekly homework assignments. This possibility supports a need for manualized mental health interventions to offer flexibility in the pace of treatment. Families with cognitive barriers near the end of the standard engagement phase might benefit from extra engagement sessions to resolve these barriers prior to progressing to the skills or planning phases of treatment. Contrary to our expectations, higher treatment engagement was associated with greater behavioral, logistical, parent, and skills/planning phase barriers. We speculate that highly engaged parents may face higher frequency behavioral and logistical barriers as they iteratively test various monitoring and contingency management systems in the skills and planning phase. Thorough discussion of these barriers may be elicited from STAND’s processes of: (1) anticipating and walking through possible barriers during homework assignment and (2) problem-solving barriers during homework review. Perhaps only families who are actively engaged in treatment will fully confront the behavioral and logistical barriers that stem from genuine trialing of various behavioral plans. It is hoped that working through these barriers collaboratively with a clinician will promote long-term adherence of well-designed behavioral plans. There are several limitations of this study. First, by definition, we had fewer audio recordings available for families who dropped out in early phases of treatment. While our analytic strategy attempted to address this concern, this limitation may introduce bias into analyses that model the relationship between barriers and engagement. Second, there were other indices of engagement (e.g., alliance, treatment credibility; Becker et al., 2018) that we could not examine because we did not have a common measure of these indices across the two RCTs included in this study. Though our sample represented a broad socioeconomic distribution, it was primarily comprised of ethnic minority youth, who characteristically demonstrate poorer treatment engagement than White peers (Aggarwal et al., 2015). Therefore, our findings may not generalize to non-minority samples. Although our sample age range was 11–16, our study under-represented older teens relative to younger teens. The observational approach utilized in this study is a strength and may reduce biases associated with self or informant reports of treatment behaviors. However, there also may have been barriers that were experienced, but not articulated, by families and were therefore undetectable using our observational approach. Therapist behaviors also may have influenced the degree to which family members discussed barriers (Moyers & Martin, 2006), just as each family member may have influenced the degree to which the other discussed barriers in treatment (Apodaca et al., 2013). Clinical Implications Based on the findings of this study, we offer several recommendations for clinical practice. First, baseline assessment of the treatment barriers noted herein may assist clinicians in developing an individualized engagement strategy for each adolescent with ADHD. Knowledge of family-specific barriers may also promote selection of treatment packages that circumvent barriers that may be intractable. For example, if parent engagement is deemed improbable, a model that leverages other adults as purveyors of contingency management (e.g., school staff) may be preferable (Breaux et al., 2019). Second, as a standard practice, teen negative attitudes toward treatment should be targeted at the outset of treatment using an evidence-based engagement strategy such as MI. In STAND specifically, extended engagement sessions might be considered if these beliefs remain after standard engagement phase tasks are completed. Similarly, behavior therapy for adolescent ADHD that engages both the parent and the teen should include parent-teen collaboration skills training when dyadic barriers are high (e.g., Barkley et al., 2001). These skills should be reinforced throughout treatment to ensure that dyadic barriers do not undermine engagement. Teen forgetfulness is a critical engagement challenge for adolescent ADHD treatment. This finding underscores the reality that adolescents still require adult stakeholder participation in behavior therapy (preferably a caregiver), whether as a primary (e.g., STAND) or secondary (e.g., Langberg et al., 2012; Sprich et al., 2016) treatment participant. Finally, there may be benefits to addressing behavioral and logistical barriers head-on in treatment, using walkthrough techniques that anticipate barriers during homework assignment and problem-solving discussions during homework review. Research Implications Initial uncontrolled findings suggest that behavior therapy for adolescent ADHD that incorporates engagement-focused components is associated with stronger engagement and outcomes than behavior therapies without them (Sibley et al., 2021b). However, future research should systematically evaluate whether integration of the engagement strategies above may reduce the impact of barriers on outcomes for adolescents with ADHD. Future research should also investigate the utility of decision rules that indicate readiness to shift from the engagement to skills to planning phases that are based on the assessment of barriers. Developing a scale that measures common barriers to treatment among adolescents with ADHD could be an important future direction in this process. Research should examine whether allowing the pace and sequencing of treatment to be measurement-based, rather than standardized, might allow for a better tailored, and more engaging, course of treatment. Furthermore, research should develop and validate a population-specific menu of engagement strategies that might be utilized by providers across various psychosocial and pharmacological treatments for adolescent ADHD. Conclusions The results of this study highlight that engagement is a dynamic process that unfolds throughout a course of mental health treatment and can be influenced by a range of barriers. In the case of adolescent ADHD, primary barriers are adolescent attitudes toward treatment, parent behavioral barriers, and logistical challenges during the planning phases of treatment. Behavior therapies for adolescent ADHD that include engagement-focused components are critical to overcoming barriers and ensuring a successful course of treatment. Supplementary Material Supp 1 Funding Details: This study was funded by the National Institute of Mental Health (R21 MH116499). It was also funded, in part, R34MH092466 and the Klingenstein Third Generation Foundation. Data availability: The quantitative (but not qualitative) data utilized in this study is publicly available through the National Data Archive: (https://nda.nih.gov), including a data dictionary. Table 1. Demographic and Clinical Characteristics of the Sample Age M (SD) 12.98 (1.19) Sex % (n)  Male 72.7 (88)  Female 27.3 (33) Race/Ethnicity % (n)  White Non-Latinx 11.6 (14)  African-American 7.4 (9)  Latinx Any Race 77.7 (94)  Asian-American 1.7 (2)  Mixed Race 1.7 (2) Receives ADHD Medication % (n) 39.7 (48) Primary Caregiver % (n)  Mother 89.3 (108)  Father 9.1 (11)  Grandmother 1.7 (2) Parent Education Level % (n)  High School or less 11.6 (14)  Some college or Associate’s 28.9 (35)  Bachelor’s degree 36.4 (44)  Master’s degree or higher 22.3 (27)  Undisclosed 0.8 (1) Single Parent % (n) 38.8 (47) Parent English Proficiency % (n) 83.5 (101) School Services  Individualized Education Plan 24.8 (30)  Section 504 Plan 23.1 (28)  None 49.6 (60)  Undisclosed 2.5 (3) ADHD Subtype % (n)  ADHD-Predominantly Inattentive Type 47.9 (58)  ADHD-Combined Type 52.1 (63) Oppositional Defiant Disorder % (n) 47.9 (58) Table 2. Observed barriers to ADHD treatment engagement Category Type % (n) Definition Low teen desire for treatment C/T 75.2 (91) Disinterest in treatment, expressing desire for therapy to discontinue, dissatisfaction with treatment, expressing that treatment is boring, low desire to learn or practice skills Parental failure to monitor teen practice B/P 69.4 (84) Failure to systematically and consistently check or monitor teen use of skills including writing in the agenda, monitoring grades or homework completion, checking organization of bookbag, or other discrete practice activities Teen forgetfulness B/T 60.3 (73) Forgot to complete weekly skill practice assignments Teen believes no change is needed C/T 56.2 (68) Satisfied with current behavior, performance, or effort. Includes failure to acknowledge difficulties. Low teen self-efficacy C/T 49.6 (60) Lack of confidence in ability to complete treatment activities or skill practice, expectations for failure, expression of helplessness, indication that teen does not believe they can improve presenting problems Parent intrusiveness B/P 49.6 (60) Intrusive parental behaviors that prevent autonomous teen skill practice. Ex: doing a treatment activity for the teen, repeatedly prompting and reminding the teen to complete tasks, taking over or micromanaging teen assignments Low teen treatment confidence C/T 34.7 (42) Belief that the treatment itself is not effective, expression that there is no point in trying skills, expectation that plans will be ineffective Schedule conflicts L/D 31.4 (38) Competing activities prevented dyad from completing practice activities or attending treatment (e.g., dance practice prevented studying, canceled session due to parent work commitment) Insufficient time L/T 30.6 (37) Teen ran out of time when practicing tasks or skill application Teen side-tracked during practice B/T 29.8 (36) Indication of distractions, day dreaming, or focus problems undermining ability to complete skill practice Failure to find common ground C/D 25.6 (31) Failure to find common ground that prevents treatment tasks. Ex: lack of consensus on how a task should be completed, declining to collaborate with family member until they change their behavior Teen skill deficit B/T 24.8 (30) Lacking the skills needed to perform practice activities Inconsistent contingencies B/P 24.0 (29) Parent allows teen to access rewards without earning them, gives in to inappropriate teen demands, or has difficulty removing electronics when time limit ends Dyadic conflict B/D 14.9 (18) Eruption of arguments between the dyad prevented skill practice or treatment activities Tired B/T 14.0 (17) Lack of sleep, too tired, not enough energy to practice skills Dislike of materials C/T 12.4 (15) Dislike of physical features of supplies provided or suggested by therapist such as binder or planner Lack of supplies L/D 12.4 (15) Failure to practice skills because parent did not obtain supplies (e.g., did not locate a planner to use) Electronics Overuse B/T 11.6 (14) Deciding to utilize electronics instead of practicing skills. Difficulty stopping electronics use when it’s time to do complete skill practice. Parent low treatment confidence C/P 9.9 (12) Low parent confidence in treatment, sometimes due to past treatment failures Demands on parent C/P 9.9 (12) Parenting demands of treatment are too complicated or burdensome Ineffective Rewards B/P 8.3 (10) Teen not motivated by rewards offered by parents as a part of contingency management Poor school engagement L/D 7.4 (9) Teacher refusal to sign planner, teacher lack of communication, school failing to implement interventions Embarrassed to involve school C/T 5.0 (6) Concern about teachers or classmates learning of treatment Cheating B/T 5.0 (6) Breaking rules covertly, sneaking unearned privileges Anti-rewards belief C/P 4.1 (5) Philosophical stance against use of rewards, belief that teen does not respond to rewards Note. C=cognitive/attitudinal barrier, B=behavioral barrier, L=logistical barrier, P=parent barriers, T=teen barrier, D=dyadic barrier Table 3. Estimated prevalence of each barrier by treatment phase (engagement, skills, planning) Estimated Marginal Means Engagement vs. Skills (1 vs 2) Skills vs. Planning (2 vs. 3) Engagement vs. Planning (1 vs. 3) Engagement Skills Planning b SE p b SE p b SE p Low teen desire for treatment .51 .54 .51 .11 .35 .760 .13 .28 .646 −.02 .31 .953 Parental failure to monitor teen practice .20 .43 .57 1.07 .45 .017 −.56 .29 .056 1.64 .39 .000 Teen forgetfulness .18 .35 .37 .91 .39 .019 −.06 .31 .853 .97 .37 .008 Teen believes no change is needed .44 .33 .25 −.46 .37 .209 .42 .31 .172 −.88 .33 .007 Low teen self-efficacy .21 .21 .27 −.01 .49 .988 −.31 .33 .350 .30 0.41 .463 Parent intrusiveness .25 .24 .36 −.06 .40 .889 −.62 .28 .027 .56 .32 .081 Low teen treatment confidence .18 .13 .12 −.42 .52 .420 .10 .36 .792 −.51 .52 .322 Schedule conflicts .04 .13 .15 1.33 .74 .072 −.22 .43 .613 1.55 .66 .019 Insufficient time .10 .12 .13 .23 .59 .699 −.12 .42 .766 .35 .57 .538 Teen side-tracked during practice .16 .16 .06 .01 .47 .979 1.05 .46 .022 −1.04 .56 .063 Failure to find common ground .11 .11 .10 .03 .50 .955 .08 .40 .847 −.05 .53 .952 Teen skill deficit .09 .10 .13 .08 .55 .891 −.25 .52 .626 .33 .44 .452 Inconsistent contingencies .03 .13 .13 1.55 .77 .044 .01 .42 .982 1.54 .72 .032 Dyadic conflict .03 .07 .08 1.07 .87 .215 NE NE NE 1.18 .80 .142 Tired .03 .05 .05 .62 .90 .488 .01 .48 .986 .61 .83 .463 Dislike of materials .00 .06 .03 NE NE NE .87 .67 .191 22.05 .67 .00 Lack of supplies .01 .09 .04 2.16 1.18 .067 −.77 .49 .113 1.39 1.25 .264 Electronics Overuse .05 .02 .05 −.78 1.15 .498 −.88 .70 .211 .10 .96 .916 Parent low treatment confidence .03 .03 .05 −.20 1.02 .846 −.69 .60 .246 .49 1.01 .626 Demands on parent .02 .02 .04 .37 1.08 .735 −.67 .65 .301 1.04 1.12 .355 Ineffective Rewards .02 .02 .01 .22 1.71 .896 1.13 1.21 .350 −.91 1.51 .547 Poor school engagement .04 .02 .02 −.71 1.18 .544 −.21 .57 .713 −.51 1.13 .651 Embarrassed to involve school .01 .03 .01 .96 1.14 .403 1.16 .95 .219 −.20 1.45 .888 Cheating .00 .00 .01 NE NE NE −.93 .84 .270 29.18 .84 .000 Anti-rewards belief .01 .03 .01 .94 1.41 .505 1.00 .80 .212 −.06 1.66 .971 Note. b=unstandardized beta, SE=standard error, NE=no estimate could be produced due to model instability, alpha=.05, marginal means are estimated at the mean of the covariates. Disclosure Statement: Dr. Sibley receives book royalties from Guilford Press and video royalties from Vimeo, Inc. for clinician informational material regarding the treatment evaluated herein. References Abba N , Chadwick P , & Stevenson C (2008). Responding mindfully to distressing psychosis: A grounded theory analysis. Psychotherapy research, 18 (1 ), 77–87. 10.1080/10503300701367992 18815958 Aggarwal NK , Pieh MC , Dixon L , Guarnaccia P , Alegría M , & Lewis-Fernández R (2015) Clinician descriptions of communication strategies to improve treatment engagement by racial/ethnic minorities in mental health services: A systematic review. Patient Education and Counseling, 99 (2 ), 198–209. 10.1016/j.pec.2015.09.002 26365436 American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). 10.1176/appi.books.9780890425596. Apodaca TR , Magill M , Longabaugh R , Jackson KM , & Monti PM (2013). Effect of a significant other on client change talk in motivational interviewing. Journal of Consulting and Clinical Psychology, 81 (1 ), 35–46. 10.1037/a0030881 23231575 Baker-Ericzén MJ , Jenkins MM , & Haine-Schlagel R (2013). Therapist, Parent, and Youth Perspectives of Treatment Barriers to Family-Focused Community Outpatient Mental Health Services. Journal of Child and Family Studies, 22 (6 ), 854–868. 10.1007/s10826-012-9644-7 24019737 Barkley RA , Edwards G , Laneri M , Fletcher K , & Metevia L (2001). The efficacy of problem-solving communication training alone, behavior management training alone, and their combination for parent–adolescent conflict in teenagers with ADHD and ODD. Journal of Consulting and Clinical Psychology, 69 (6 ), 926–941. 10.1037/0022-006X.69.6.926 11777120 Becker KD , Lee BR , Daleiden EL , Lindsey M , Brandt NE , & Chorpita BF (2015). The Common Elements of Engagement in Children’s Mental Health Services: Which Elements for Which Outcomes?. Journal of Clinical Child & Adolescent Psychology, 44 (1 ), 30–43. 10.1080/15374416.2013.814543 23879436 Becker KD , Boustani M , Resham G , & Chorpita BF (2018). Forty Years of Engagement Research in Children’s Mental Health Services: Multidimensional Measurement and Practice Elements. Journal of Clinical Child & Adolescent Psychology, 47 (1 ), 1–23. 10.1080/15374416.2017.1326121 28574780 Becker KD , Dickerson K , Boustani MM , & Chorpita BF (2021). Knowing What To Do and When To Do It: Mental Health Professionals and the Evidence Base for Treatment Engagement. Administration and Policy in Mental Health and Mental Health Services Research, 48 (2 ), 201–218. 10.1007/s10488-020-01067-6 32661787 Bird HR , Gould MS , & Staghezza B (1992). Aggregating data from multiple informants in child psychiatry epidemiological research. Journal of the American Academy of Child & Adolescent Psychiatry, 31 (1 ), 78–85. 10.1097/00004583-199201000-00012 1537785 Breaux RP , Langberg JM , Molitor SJ , Dvorsky MR , Bourchtein E , Smith ZR , & Green CD (2019). Predictors and trajectories of response to the Homework, Organization, and Planning Skills (HOPS) intervention for adolescents with ADHD. Behavior therapy, 50 (1 ), 140–154. 10.1016/j.beth.2018.04.001 30661554 Brinkman WB , Simon JO , & Epstein JN (2018). Reasons Why Children and Adolescents With Attention-Deficit/Hyperactivity Disorder Stop and Restart Taking Medicine. Academic Pediatrics, 18 (3 ), 273–280. 10.1016/j.acap.2017.09.005 28919571 Brown A , Rice S , Rickwood DJ , & Parker AG (2016) Systematic review of barriers and facilitators to accessing and engaging with mental health care among at-risk young people. Asia-Pacific Psychiatry, 8 , 3–22. 10.1111/appy.12199 26238088 Buckingham SL , Evangelista Brandt N , Becker KD , Gordon D , & Cammack N (2016). Collaboration, Empowerment, and Advocacy: Consumer Perspectives about Treatment Engagement. Journal of Child and Family Studies, 25 , 3702–3715. 10.1007/s10826-016-0507-5 Bussing R , Zima BT , Mason DM , Porter PC , & Garvan CW (2011). Receiving Treatment for Attention-Deficit Hyperactivity Disorder: Do the Perspectives of Adolescents Matter? Journal of Adolescent Health, 49 (1 ), 7–14. 10.1016/j.jadohealth.2010.08.014 Bussing R , Koro-Ljungberg M , Noguchi K , Mason D , Mayerson G , & Garvan CW (2012). Willingness to use ADHD treatments: A mixed methods study of perceptions by adolescents, parents, health professionals and teachers. Social Science & Medicine, 74 (1 ), 92–100. 10.1016/j.socscimed.2011.10.009 22133584 Chacko A , Wymbs BT , Rajwan E , Wymbs F , & Feirsen N (2017). Characteristics of Parents of Children with ADHD Who Never Attend, Drop Out, and Complete Behavioral Parent Training. Journal of Child and Family Studies, 26 , 950–960. 10.1007/s10826-016-0618-z Charmaz K (2014). Constructing grounded theory (2nd ed., Introducing qualitative methods). London; Thousand Oaks, California: SAGE Publications. Cicchetti DV (1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6 (4 ), 284–290. Clarke AT , Marshall SS , Mautone JA , Soffer SL , Jones HA , Costigan TE , Patterson A , Jawad AF , & Power TJ (2015). Parent Attendance and Homework Adherence Predict Response to a Family-School Intervention for Children with ADHD. Journal of Clinical Child and Adolescent Psychology, 44 (1 ), 58–67. 10.1080/15374416.2013.794697 23688140 Danielson ML , Bitsko RH , Ghandour RM , Holbrook JR , Kogan MD , & Blumberg SJ (2018). Prevalence of Parent-Reported ADHD Diagnosis and Associated Treatment Among U.S. Children and Adolescents, 2016. Journal of Clinical Child and Adolescent Psychology, 47 (2 ), 199–212. 10.1080/15374416.2017.1417860 29363986 Dean S , Britt E , Bell E , Stanley J , & Collings S (2016). Motivational interviewing to enhance adolescent mental health treatment engagement: A randomized clinical trial. Psychological Medicine, 46 (9 ), 1961–1969. 10.1017/S0033291716000568 27045520 Dvorsky MR , Friedman LM , Spiess M , & Pfiffner LJ (2021). Patterns of Parental Adherence and the Association to Child and Parenting Outcomes Following a Multicomponent School-Home Intervention for Youth With ADHD. Behavior Therapy. 10.1016/j.beth.2020.09.007 Evans SW , Schultz BK , White LC , Brady C , Sibley MH , & Van Eck K (2009). A school-based organization intervention for young adolescents with attention-deficit/hyperactivity disorder. School Mental Health, 1 (2 ), 78–88. 10.1007/s12310-009-9009-6 Evans SW , Owens JS , Wymb BT , & Ray AR (2017). Evidence-Based Psychosocial Treatments for Children and Adolescents with Attention Deficit/Hyperactivity Disorder. Journal of Clinical Child & Adolescent Psychology, 1–42. 10.1080/15374416.2017.1390757 Fabiano GA , Pelham WE Jr , Waschbusch DA , Gnagy EM , Lahey BB , Chronis AM , ... & Burrows-MacLean L (2006). A practical measure of impairment: Psychometric properties of the impairment rating scale in samples of children with attention deficit hyperactivity disorder and two school-based samples. Journal of Clinical Child and Adolescent Psychology, 35 (3 ), 369–385. 10.1207/s15374424jccp3503_3 16836475 Garland AF , Haine-Schlagel R , Accurso EC , Baker-Ericzén MJ , & Brookman-Frazee L (2012). Exploring the effect of therapists' treatment practices on client attendance in community-based care for children. Psychological services, 9 (1 ), 74–88. 10.1037/a0027098.22449089 Glynn LH , Hallgren KA , Houck JM , Moyers TB (2012) CACTI: Free, Open-Source Software for the Sequential Coding of Behavioral Interactions. PLoS ONE 7 (7 ): e39740. 10.1371/journal.pone.0039740 22815713 Haine-Schlagel R , & Walsh NE (2015). A Review of Parent Participation Engagement in Child and Family Mental Health Treatment. Clinical Child and Family Psychology Review, 18 (2 ), 133–150. 10.1007/s10567-015-0182-x 25726421 Houck JM , Moyers TB , Miller WR , Glynn LH , & Hallgren KA (2010). Motivational interviewing skill code (MISC) version 2.5. Unpublished Manual. https://casaa.unm.edu/download/misc25.pdf Ingoldsby EM (2010). Review of interventions to improve family engagement and retention in parent and child mental health programs. Journal of child and family studies, 19 (5 ), 629–645. 10.1007/s10826-009-9350-2 20823946 Johnson E , Mellor D , & Brann P (2008). Differences in Dropout between Diagnoses in Child and Adolescent Mental Health Services. Clinical Child Psychology and Psychiatry, 13 (4 ), 515–530. 10.1177/1359104508096767 18927138 Kazdin AE & Mazurick J (1994). Dropping Out of Psychotherapy: Distinguishing Early and Late Dropouts Over the Course of Treatment. Journal of Consulting and Clinical Psychology, 62 (5 ), 1069–1074. 10.1037//0022-006x.62.5.1069 7806717 Kazdin AE , & Wassell G (1999). Barriers to Treatment Participation and Therapeutic Change Among Children Referred for Conduct Disorder. Journal of Clinical Child Psychology, 28 (2 ), 160–172. 10.1207/s15374424jccp2802_4 10353076 Kazdin AE , & McWhinney E (2017). Therapeutic Alliance, Perceived Treatment Barriers, ad Therapeutic Change in the Treatment of Children with Conduct Problems. Journal of Child and Family Studies, 27 , 240–252. 10.1007/s10826-017-0869-3 Kerkorian D , McKay M , & Bannon WM Jr. (2006). Seeking Help a Second Time: Parents’/Caregivers’ Characterizations of Previous Experiences With Mental Health Services for Their Children and Perceptions of Barriers to Future Use. American Journal of Orthopsychiatry, 76 (2 ), 161–166. 10.1037/0002-9432.76.2.161 16719634 Langberg JM , Epstein JN , Becker SP , Girio-Herrera E , & Vaughn AJ (2012). Evaluation of the Homework, Organization, and Planning Skills (HOPS) Intervention for Middle School Students with ADHD as Implemented by School Mental Health Providers. School Psychology Review, 41 (3 ), 342–364.25355991 Langberg JM , Becker SP , Epstein JN , Vaughn AJ , & Girio-Herrera E (2013). Predictors of response and mechanisms of change in an organizational skills intervention for students with ADHD. Journal of child and family studies, 22 (7 ), 1000–1012. 10.1007/s10826-012-9662-5 Lindsey MA , Brandt NE , Becker KD , Lee BR , Barth RP , Daleiden EL , & Chorpita BF (2014). Identifying the common elements of treatment engagement interventions in children’s mental health services. Clinical child and family psychology review, 17 (3 ), 283–298.24379129 McKay M & Bannon WM Jr. (2004). Engaging families in child mental health services. Child and Adolescent Psychiatric Clinics of North America, 13 , 905–921. 10.1016/j.chc.2004.04.001 15380788 Miller WR , & Rollnick S (2013). Motivational interviewing. Help the person initiate the change, 2 . Molina BSG , Hinshaw SP , Swanson JM , Arnold LE ,Vitiello B , Jensen PS , Epstein JN , Hoza B , Hechtman L , Abikoff HB , Elliott GR , Greenhill LL , Newcorn JH , Wells KC , Wigal T , Gibbons RD , Hur K , & Houck PR (2009). The MTA at 8 Years: Prospective Follow-up of Children Treated for Combined-Type ADHD in a Multisite Study. Journal of the American Academy of Child and Adolescent Psychiatry, 48 (5 ), 484–500. 10.1097/CHI.0b013e31819c23d0 19318991 Moyers TB , & Martin T (2006). Therapist influence on client language during motivational interviewing sessions. Journal of Substance Abuse Treatment, 30 (3 ), 245–251. 10.1016/j.jsat.2005.12.003 16616169 Owens PL , Hoagwood K , Horwitz SM , Leaf PJ , Poduska JM , Kellam SG , & Ialongo NS (2002). Barriers to children's mental health services. Journal of American Academy of Child and Adolescent Psychiatry, 41 (6 ), 731–738. 10.1097/00004583-200206000-00013 Reardon T , Harvey K , Baranowska M , O’Brien D , Smith L , & Cresswell C (2017) What do parents perceive are the barriers and facilitators to accessing psychological treatment for mental health problems in children and adolescents? A systematic review of qualitative and quantitative studies. European Child Adolescent Psychiatry, 26 , 623–647. 10.1007/s00787-016-0930-6 28054223 Salloum A , Johnco C , Lewin AB , McBride NM , & Storch EA (2016). Barriers to access and participation in community mental health treatment for anxious children. Journal of Affective Disorders, 196 , 54–61. 10.1016/j.jad.2016.02.026 26901657 Sayal K , Prasad V , Daley D , Ford T , & Coghill D (2017). ADHD in children and young people: Prevalence, care pathways, and service provision. The Lancet. Psychiatry, 5 (2 ), 175–186. 10.1016/S2215-0366(17)30167-0 29033005 Schneider BW , Gerdes AC , Haack LM , & Lawton KE (2013). Predicting Treatment Dropout in Parent Training Interventions for Families of School-Aged Children with ADHD. Child and Family Behavior Therapy, 35 (2 ), 144–169. 10.1080/07317107.2013.789365 Schraw G , Wadkins T , & Olafson L (2007). Doing the things we do: A grounded theory of academic procrastination. Journal of Educational psychology, 99 (1 ), 12. 10.1037/0022-0663.99.1.12 Shaffer D , Fisher P , Lucas CP , Dulcan MK , & Schwab-Stone ME (2000). NIMH Diagnostic Interview Schedule for Children Version IV (NIMH DISC-IV): description, differences from previous versions, and reliability of some common diagnoses. Journal of the American Academy of Child & Adolescent Psychiatry, 39 (1 ), 28–38. 10.1097/00004583-200001000-00014 10638065 Sibley MH , (2016). Parent-Teen Therapy for Executive Function Deficits and ADHD: Building Skills and Motivation. New York: Guilford. Sibley MH , Pelham WE , Derefinko KD , Kuriyan AB , Sanchez F , & Graziano PA (2013). A Pilot Trial of Supporting Teens' Academic Needs Daily (STAND): A parent-adolescent collaborative intervention for ADHD. Journal of Psychopathology and Behavioral Assessment, 35 , 436–449. 10.1007/s10862-013-9353-6 Sibley MH , Olson S , Morley C , Campez M , & Pelham WE (2016). A school consultation intervention for adolescents with ADHD: barriers and implementation strategies. Child and Adolescent Mental Health, 21 (4 ), 183–191. 10.111/camh.12159 32680339 Sibley MH , Graziano PA , Kuriyan AB , Coxe S , Pelham WE , Rodriguez L , Sanchez F , Derefinko K , Helseth S , & Ward A (2016). Parent-Teen Behavior Therapy Motivational Interviewing for Adolescents With ADHD. Journal of Consulting and Clinical Psychology, 84 (8 ), 699–712. 10.1037/ccp0000106 27077693 Sibley MH , & Kuriyan AB (2016). DSM-5 changes enhance parent identification of symptoms in adolescents with ADHD. Psychiatry Research, 242 , 180–185. 10.1016/j.psychres.2016.05.036 27288736 Sibley MH , Rodriguez LM , Coxe SJ , Page T , & Espinal K (2020). Parent-Teen Group versus Dyadic Treatment for Adolescent ADHD: What Works for Whom? Journal of Clinical Child and Adolescent Psychology, 49 , 476–492. 10.1080/15374416.2019.1585257 30990088 Sibley MH , Graziano PA , Coxe S , Bickman L , Martin P (2021). Effectiveness of Motivational-Interviewing - Enhanced Behavior Therapy for Adolescents with Attention- Deficit/Hyperactivity Disorder: A Randomized Community-Based Trial. Journal of the American Academy of Child & Adolescent Psychiatry. Advance online publication. 10.1016/j.jaac.2020.07.907 Sibley MH , Morley C , Rodriguez L , Coxe SJ , Evans SW , Morsink S , & Torres F (2020). A Peer-Delivered Intervention for High School Students With Impairing ADHD Symptoms. School Psychology Review, 49 (3 ), 275–290. 10.1080/2372966X.2020.1720803 Sibley MH , Coxe SJ , Stein MA , & Meinzer MC , & Valente M (2021). Predictors of Treatment Engagement and Outcome among Adolescents with ADHD: An Integrative Data Analysis. Journal of the American Academy of Child and Adolescent Psychiatry. 10.1016/j.jaac.2021.03.017 Sibley MH , Graziano PA , Bickman L , Coxe SJ , Martin P , Rodriguez LM , Fallah N , & Ortiz M (2021). Implementing Parent-Teen Motivational Interviewing + Behavior Therapy for ADHD in Community Mental Health. Prevention Science, 22 , 701–711. 10.1007/s11121-020-01105-7 32103410 Sibley MH , Johansson M , Monroy JM , Hill DM , LaCount PA , Barney S , Molina N , & Delgado A (2021). Building a Theoretical Model for Supporting Teens’ Autonomy Daily (STAND): A Network Analysis of Family-Perceived Changes. Behavior Therapy. 10.1016/j.beth.2021.05.009 Sibley MH , Ortiz M , Rios-Davis A , Zulauf-McCurdy C , Graziano PA , & Bickman L (2021). Stakeholder-Generated Implementation Strategies to Promote Evidence-Based ADHD Treatment in Community Mental Health. Administration and Policy in Mental Health and Mental Health Services Research. 10.1007/s10488-021-01143-5 Smith E , Koerting J , Latter S , Knowles MM , McCann DC , Thompson M , & Sonuga-Barke EJ (2014). Overcoming barriers to effective early parenting interventions for attention-deficit hyperactivity disorder (ADHD): parent and practitioner views. Child: Care, Health and Development, 41 (1 ), 93–102. 10.1111/cch.12146 24814640 Sprich SE , Safren SA , Finkelstein D , Remmert JE , & Hammerness P (2016). A randomized controlled trial of cognitive behavioral therapy for ADHD in medication-treated adolescents. Journal of Child Psychology and Psychiatry, 57 (11 ), 1218–1226. 10.1111/jcpp.12549 26990084 Stevens J , Kelleher KJ , Ward-Estes J , & Hayes J (2006). Perceived Barriers to Treatment and Psychotherapy Attendance in Child Community Mental Health Centers. Community Mental Health Journal, 42 (5 ) 449–458. 10.1007/s10597-006-9048-5 16964565 Wright N , Moldavsky M , Schneider J , Chakrabarti I , Coates J , Daley D , Kochhar P , Mills J , Sorour W , & Sayal K (2015). Practitioner Review: Pathways to care for ADHD – a systematic review of barriers and facilitators. Journal of Child Psychology and Psychiatry, 56 (6 ), 598–617. 10.1111/jcpp.12398 25706049

PMC009xxxxxx/PMC9325916.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9209612 20367 Tob Control Tob Control Tobacco control 0964-4563 1468-3318 35086911 9325916 10.1136/tobaccocontrol-2021-056850 NIHMS1772657 Article COMPARISON OF NICOTINE EMISSIONS RATE, “NICOTINE FLUX”, FROM HEATED, ELECTRONIC, AND COMBUSTIBLE TOBACCO PRODUCTS: DATA, TRENDS, AND RECOMMENDATIONS FOR REGULATION El Hourani Mario ME 13 Shihadeh Alan ScD 13* Talih Soha PhD 13 Eissenberg Thomas PhD 23 the CSTP Nicotine Flux Work Group 1 Mechanical Engineering Department, Maroun Semaan Faculty of Engineering and Architecture, American University of Beirut, Bliss Street, PO. Box 11-0236, Beirut, Lebanon 2 Department of Psychology, Virginia Commonwealth University, 821 West Franklin Street, Richmond, Virginia 23284, United States 3 Center for the Study of Tobacco Products, Virginia Commonwealth University, 821 West Franklin Street, Richmond, Virginia 23284, United States * Corresponding Author: Alan Shihadeh, Tel: + 961 1 344444, as20@aub.edu.lb, Address: American University of Beirut, PO Box 11-0236 Beirut, Lebanon 21 1 2022 27 1 2022 27 7 2023 tobaccocontrol-2021-056850This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Tobacco smoking is a major cause of disease and premature death worldwide. While nicotine is recognized as the main addictive component in tobacco smoke, the total nicotine amount emitted (nicotine yield) and the rate of nicotine emission per second (“nicotine flux”) contribute to the abuse liability of a given product. These variables can be regulated for public health ends, and conveniently so for electronic cigarettes or electronic nicotine delivery systems (ENDS). In this study we computed nicotine flux from previously reported values of yield and puff topography for a wide range of tobacco products. We found that nicotine flux varied widely across tobacco products, from less than 0.1 μg/s to more than 100 μg/s, and that since 2015 the upper limit of the ENDS nicotine flux range has risen significantly and is now approaching that of combustible cigarettes. We also found that products that differ in nicotine flux may exhibit similar nicotine yields due to differences in user puffing behavior. Nicotine flux is a tool that can be used to regulate nicotine emissions of tobacco products, including ENDS. pmcINTRODUCTION Despite decades of tobacco control efforts, tobacco smoking remains one of the leading causes of premature death globally, estimated at 8 million deaths per year, and a major threat to public health.[1] The psychomotor stimulant nicotine is the main addictive agent in tobacco smoke and, without it, tobacco consumption would not be sustained.[2, 3] As with other abused drugs, the dose and the speed at which nicotine reaches brain are critical to producing the addictive character of tobacco smoking.[4] In principle, more rapid delivery and greater dose result in greater reinforcement and greater abuse liability.[5] One reason combustible cigarettes are addictive is that inhaled tobacco smoke delivers nicotine to the brain in seconds, more rapidly even than intravenous nicotine delivery.[6] Historically, nicotine yield has served as the metric for characterizing the amount of nicotine emitted by different combustible cigarette products.[7] Yield is defined as the mass of nicotine emitted through the mouth end of a tobacco product per unit of consumption (e.g., milligrams of nicotine per cigarette; mg/cig). The rate at which nicotine is delivered, the yield per unit time, is referred to as the “nicotine flux” (mg/s or μg/s).[8] Because combustible cigarettes a made in a standard size, and are consumed in roughly five minutes, nicotine yield and nicotine flux are closely coupled with combustible cigarettes – a cigarette with a high yield will also have a high flux. On the other hand, with electronic nicotine delivery systems (ENDS) and other products whose use patterns vary widely, the yield and the flux are not coupled closely. A product may have low yield and high flux (e.g., the one-puff “dokha”[9]) or vice-versa (e.g., nicotine patch). Typically, an ENDS product is consumed during multiple use sessions spanning a period of one to several days, depending on such factors as the size of the reservoir containing the nicotine solution and the electrical power of the device. Therefore, the nicotine yield of the product per unit sold may not be relevant to the yield obtained during a single use session. For example, a single JUUL pod emits roughly the same amount of nicotine as an entire pack of cigarettes but is unlikely to be consumed entirely in a single-use session.[10] Even the notion of a use session for an ENDS product may be difficult to define. Does taking a single puff just before entering an office building constitute a “session”? Nicotine patches, too, can deliver a dose of nicotine over a day that is comparable to a pack of cigarettes. Clearly, a comparison of the yield of a JUUL pod, a nicotine patch, and a cigarette stick has little value because the consumption patterns differ greatly; as a regulatory target, yield is not a useful construct. Nicotine flux, on the other hand, allows comparisons across products and product classes because it normalizes nicotine emission by time. In doing so, flux also highlights the key factor of speed of delivery: nicotine flux is the theoretical upper limit of the rate at which nicotine can reach the brain. As we have discussed elsewhere,[11] to be enforceable a flux standard implies that only closed systems will be allowed on the market. Figure 1 illustrates by analogy the relationship between nicotine flux, liquid nicotine concentration, device power, time, and nicotine yield for ENDS products. The large tank can be thought of as the liquid reservoir of an ENDS product, while the small container can be considered the mouth of the user. The nicotine concentration of the liquid in the tank was prepared by dissolving a given mass, m, of nicotine (mg) in a given volume, V, of liquid (mL), resulting in a liquid nicotine concentration C = m/V (mg/mL). When a puff is executed, the rate at which liquid is aerosolized by the ENDS device (i.e., in the form of an inhalable aerosol) and delivered to the mouth of the user is represented by opening the tap, allowing the flow to commence at some rate Q (mL/s). The nicotine flux is the product of the nicotine concentration C and the volume flow rate Q. To a close first approximation, Q is directly proportional to the power (P, Watts); greater power translates to a more open tap in Fig 1. As a result, nicotine flux is directly proportional to the product of C and P. Finally, the amount of nicotine collected from the tap while the valve was open is the yield, which is simply the product of the flux and time. In this study, we sought to estimate nicotine flux for a wide range of tobacco products to provide a base against which a potential ENDS product regulation could be considered. To date, extant EU and proposed US regulations have focused exclusively on limiting liquid nicotine concentration,[12, 13] an approach which, counter to the stated aims of those regulations, constrains neither yield nor flux, and therefore does not constrain exposure (i.e., nicotine dose inhaled by the user). METHODS Nicotine flux can be computed from published reports on tobacco product yields as the ratio of the yield to the cumulative puffing time of an inhaled tobacco product (e.g., a cigarette or pipe) or as the ratio of the yield to the cumulative time of use of a product that emits nicotine continuously (e.g., a nicotine patch). For inhaled products, we searched the Scopus database using the following Boolean expression: (“nicotine” OR “nicotine yield”) AND (“flow rate” OR “puff duration” OR “interpuff interval” OR “puff volume” OR “topography”). The search resulted in 651 documents, of which 39 reported values of nicotine yield, puff duration, and number of puffs; these 39 documents were retained for analysis. The nicotine flux was computed as: Nicotine flux(μgs)=1000*Nicotine yield(mgunit)Total puffnumber (puffsunit)*puff duration(spuff) For patch and gum products, we computed the flux as: Nicotine flux(μgs)=1000*Nicotine dose(mgunit)Total time of consumption(sunit) where the time of consumption was taken as 24 h for patches and 30 min for gum. The dose was taken as that provided by the manufacturer assuming complete release of nicotine during the time of product consumption. Average (standard deviations) of nicotine fluxes for each tobacco product were determined to compare different products. A simple linear regression was used to test the correlation between year of publication vs. nicotine flux. Statistical significance was taken as p < 0.05. RESULTS Published data were available to compute nicotine flux for approximately 90 products, spanning the categories of cigarettes, cigarillos, small cigars, waterpipes, ENDS, heated tobacco products, patches, and nicotine gum. Table S1 of the supplemental material lists the results obtained for tobacco products that were machine smoked by mimicking human puffing patterns or by standard machine smoking regimens (e.g., Canadian Intense, ISO). The nicotine flux across products ranged four orders of magnitude, from less than 0.1 μg/s to more than 100 μg/s, with the low end of the spectrum populated exclusively by nicotine patch and gum products and very low nicotine cigarettes and products above 100 μg/s consisting exclusively of conventional combustible cigarettes. The results are summarized in Table 1. We also found a significant increase in reported flux over time (Figure 2) for ENDS products (4.5μ g/s/year; p<0.001). Whereas prior to 2018, no publications reported products with a flux exceeding 40 ug/s, from 2019 onwards, nearly 40% of the tested products exceeded a flux of 60 μg/s. The upper quartile flux for ENDS products in a given year also increased significantly at a mean rate of 9.5 μg/s/year (p<0.001). DISCUSSION Nicotine flux is a performance metric that describes the acute nicotine throughput of a tobacco product, the net outcome of the interactions between numerous product design and operating variables. While nicotine flux represents the rate at which nicotine can enter the human body, and therefore the theoretical limit of the rate of delivery to the brain, the rate is mediated by other factors that influence the pharmacokinetics of nicotine delivery. For inhaled products, these factors include such things as the particle size distribution and freebase-to-protonated nicotine ratio of the aerosol. In this study, we sought to document nicotine flux from a range of tobacco products whose yields and puffing parameters had been reported in the literature. One limitation of this study is that products studied by previous researchers may not represent well the sales-weighted average of each category. A second limitation is that reported smoking machine studies may not have always used representative puffing parameters (e.g., puff velocity, duration, or interpuff interval), biasing nicotine yield, and, therefore, the computed flux. The most accurate analytical determinations of nicotine emissions are made using puffing conditions appropriate to the product in question; for example, users of large sub-Ohm ENDS devices typically draw up to an order of magnitude greater flow rate than a user of a small pod-based device. We found that for inhalable tobacco products, combustible cigarettes exhibited the greatest average nicotine flux, while waterpipes exhibited the greatest nicotine yield per session. Overall, nicotine patches had the greatest yields but also, owing to the long duration of use per unit, the lowest fluxes. These findings underscore the limitations of nicotine yield as a regulatory construct for tobacco products that vary widely; in these cases, greater yield was associated with lower abuse liability. We also found significant variability in flux within and across product categories, as illustrated in Figure 3. While ENDS generally exhibited nicotine fluxes lower than those of combustible cigarettes, reports from 2018 onwards began revealing ENDS products whose fluxes were equivalent to combustible cigarettes. Importantly, the 110 μg/s maximum flux reported to date for an ENDS product does not represent an intrinsic physical limit. With products available over the counter today, an ENDS user can readily access a liquid/device combination whose flux exceeds any value yet reported. For example, based on the mathematical model of Talih et al. [14], a device operating at 60 W with an EU-compliant 20 mg/ml nicotine concentration liquid can produce a flux of approximately 240 μg/s, roughly double the maximum reported for any combustible cigarette. The current regulatory environment therefore allows marketing ENDS products whose nicotine emission rate exceeds that of the high abuse liability combustible cigarette. Combined with emerging evidence that the convenience of ENDS use leads to far more frequent nicotine administration throughout the day than for combustible cigarettes,[15–17] the availability of high-flux ENDS products may portend greater population-wide nicotine dependence than was present prior to the advent of ENDS, if this outcome has not already been realized. Empirical data on the relationship between flux, acute delivery, and dependence is too thin to evaluate this hypothesis at present; such data is urgently needed. Of note, Do et al.[18] recently reported an association between nicotine flux and dependence scores in a pilot study of experienced users of pod-based devices. Given the approximate doubling of ENDS nicotine flux from 2015 to 2020, policy makers may not have the leeway to wait for a definitive evidence base to emerge, and may find it prudent to regulate flux in the interim. There is little reason to suspect that exceeding the nicotine flux of combustible cigarettes is necessary to improve public health. With this starting point, an upper limit on ENDS nicotine flux could be, at most, 140 μg/s (see Table 1). However, given the greater convenience and greater use frequency observed in ENDS users, this upper limit, if applied to ENDS, may still lead to greater population-level nicotine dependence. For this reason, one potential approach is to use the mean observed for combustible cigarette flux (i.e., approximately 80 mg/sec, see Table 1) as a temporary ceiling for over-the-counter ENDS products, with further adjustments informed by empirical investigations aimed at understanding the abuse liability of ENDS products across populations of particular interest (e.g., nicotine naïve individuals, former smokers at risk for relapse). Of course, if empirical work demonstrates that higher flux ENDS are safe and effective for smoking cessation, these products can be made available to cigarette smokers in a manner that does not risk the health of nicotine-naïve individuals (e.g., restricted access rather than over-the-counter availability). An additional concern is that ENDS aerosols contain varying concentrations of toxicants such as carbonyl species. Thus, minimizing the amount of inhaled aerosol may be desirable because it can reduce user exposure to harmful toxicants. From this perspective, to the extent that a user seeks to attain a given nicotine intake, too low a nicotine flux can, perversely, increase non-nicotine toxicant exposure because it may drive more prolonged puffing bouts. Policymakers interested in reducing nicotine dependence at the population level would do well to address nicotine flux as a regulatory target and avoid the mistake of using inappropriate proxies (e.g., liquid nicotine concentration) that cannot, by themselves, be used to control the nicotine dose inhaled by ENDS users. Supplementary Material Supp1 FUNDING SUPPORT This research is supported by grant number U54DA036105 from the National Institute on Drug Abuse of the National Institutes of Health and the Center for Tobacco Products of the US Food and Drug Administration. Figure 1. Relationship between liquid nicotine concentration, device power, flux (nicotine emission rate), and yield for an ENDS device by analogy to a reservoir emptying into a container through a valve and tap assembly. In this analogy, the electrical power of the ENDS device determines the degree to which the tap is open during a puff; greater power means a more open valve. Figure 2. Reported nicotine flux of ENDS products by year of manuscript publication. (p<0.001 for both regression lines) Figure 3. Nicotine flux ranges across tobacco products. The dashed line for ENDS represents the capacity of current over-the-counter products to exceed values reported to date. Table 1. Computed nicotine flux by tobacco product category. N indicates the number of products reported, while year span indicates years of publication for the studies included. SD = standard deviation. Flux (μg/s) Product N Year span Mean (SD) Range Combustible cigarettes 27 1988–2020 79(32) 29–140 ENDS 52 2015–2021 29(23) 3.7–110 Heated tobacco products 14 2018–2020 31(18) 5.8–58 Waterpipe 8 2003–2019 11(5.6) 5.8–20 Cigars/cigarillos 9 1976–2018 62(35) 12–110 Roll your own 3 1985–2014 69(28) 52–100 Bidi 2 1988–2003 60(6.8) 55–64 Kretek 3 2014 47(16) 29–60 Nicotine patch and gum products 4 2018–2019 0.4(0.48) 0.08–1.1 Very low nicotine cigarettes 1 2019 1.7(-) - WHAT THIS PAPER ADDS To date, tobacco product regulation aimed at nicotine abuse liability has been hampered by reliance on metrics which are not relevant to many tobacco products. Nicotine flux, the amount of nicotine emitted per unit time, is a metric of abuse liability that can be deployed across a diverse range of tobacco products, including those that are inhaled, chewed, or applied to the skin. Nicotine flux across products varies from less than 0.1 μg/s to more than 100 μg/s, and for electronic cigarettes has been rising at an average rate of 5 μg/s/year since 2015. DISCLAIMER The content is solely the responsibility of the authors and does not necessarily represent the views of the NIH or the FDA. DECLARATION OF INTERESTS The authors declare the following competing financial interest: Drs. Eissenberg and Shihadeh are paid consultants in litigation against the tobacco industry and also the electronic cigarette industry and are named on one patent for a device that measures the puffing behavior of electronic cigarette users. Dr. Eissenberg is also named on a patent for a smartphone app that determines electronic cigarette device and liquid characteristics. REFERENCES [1] WHO. Tobacco Fact Sheet. WHO Newsroom 2021;https://www.who.int/news-room/factsheets/detail/tobacco (accessed 3 December 2021). [2] Benowitz NL , Henningfield JE . Establishing a nicotine threshold for addiction. The implications for tobacco regulation. The New England journal of medicine 1994;331 (2 ):123–125.7818638 [3] Henningfield JE , Benowitz NL , Slade J , Reducing the addictiveness of cigarettes. Council on Scientific Affairs, American Medical Association. Tobacco control 1998;7 (3 ):281–293.9825424 [4] Carter LP , Stitzer ML , Henningfield JE , Abuse liability assessment of tobacco products including potential reduced exposure products. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology 2009;18 (12 ):3241–3262. [5] Allain F , Minogianis EA , Roberts DC , How fast and how often: The pharmacokinetics of drug use are decisive in addiction. Neuroscience and biobehavioral reviews 2015;56 :166–179.26116543 [6] Henningfield JE , Stapleton JM , Benowitz NL , Higher levels of nicotine in arterial than in venous blood after cigarette smoking. Drug and alcohol dependence 1993;33 (1 ):23–29.8370337 [7] Djordjevic MV , Stellman SD , Zang E . Doses of Nicotine and Lung Carcinogens Delivered to Cigarette Smokers. JNCI: Journal of the National Cancer Institute 2000;92 (2 ):106–111.10639511 [8] Shihadeh A , Eissenberg T . Electronic cigarette effectiveness and abuse liability: predicting and regulating nicotine flux. Nicotine & tobacco research : official journal of the Society for Research on Nicotine and Tobacco 2015;17 (2 ):158–162.25180079 [9] Samara F , Alam IA , ElSayed Y . Midwakh: Assessment of Levels of Carcinogenic Polycyclic Aromatic Hydrocarbons and Nicotine in Dokha Tobacco Smoke. Journal of Analytical Toxicology 2021. [10] Prochaska JJ , Vogel EA , Benowitz N . Nicotine delivery and cigarette equivalents from vaping a JUULpod. Tobacco control 2021:tobaccocontrol-2020–056367. [11] Eissenberg T , Soule E , Shihadeh A . ‘Open-System’ electronic cigarettes cannot be regulated effectively. Tobacco control 2021;30 (2 ):234–235.32184338 [12] Nedelman M . New bill aims to cap nicotine levels in e-cigarettes. CNN 2019; https://edition.cnn.com/2019/10/07/health/vaping-nicotine-cap-krishnamoorthi-bn/index.html (accessed 3 December 2021). [13] Directive 2014/40/EU of the European Parliament and of the Council of 3 April 2014 on the approximation of the laws, regulations and administrative provisions of the Member States concerning the manufacture, presentation and sale of tobacco and related products and repealing Directive 2001/37/EC. Official Journal of the European Union 2014;127 (1 ):1–38. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=OJ%3AJOL_2014_127_R_0001 (accessed 16 December 2021) [14] Talih S , Balhas Z , Eissenberg T , Effects of user puff topography, device voltage, and liquid nicotine concentration on electronic cigarette nicotine yield: measurements and model predictions. Nicotine & tobacco research : official journal of the Society for Research on Nicotine and Tobacco 2015;17 (2 ):150–157.25187061 [15] Cooper M , Harrell MB , Perry CL . A Qualitative Approach to Understanding Real-World Electronic Cigarette Use: Implications for Measurement and Regulation. Preventing chronic disease 2016;13 :E07.26766848 [16] Baweja R , Curci KM , Yingst J , Views of Experienced Electronic Cigarette Users. Addict Res Theory 2016;24 (1 ):80–88.29176939 [17] Paul K Breaking up with my Juul: why quitting vaping is harder than quitting cigarettes. The Guardian 2019; https://www.theguardian.com/society/2019/oct/10/breaking-up-with-my-juul-why-quitting-vaping-is-harder-than-quitting-cigarettes (accessed 3 December 2021). [18] Do E , O’Connor K , Perks S , E-cigarette Device and Liquid Characteristics and E-cigarette Dependence: A Pilot Study of Pod-Based and Disposable E-cigarette Users. Addictive Behaviors 2022;124 :107117.34555560

PMC009xxxxxx/PMC9325923.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8501885 25634 J Reprod Infant Psychol J Reprod Infant Psychol Journal of reproductive and infant psychology 0264-6838 1469-672X 35083966 9325923 10.1080/02646838.2022.2030052 NIHMS1776480 Article Measuring Post-Traumatic Stress After Childbirth: A Review and Critical Appraisal of Instruments Williams Meagan E. a Strobino Donna M. a Holliday Charvonne N. a a Department of Population, Family and Reproductive Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA Corresponding author: Meagan Williams, MSPH, Department of Population, Family and Reproductive Health, Johns Hopkins Bloomberg School of Public Health, Phone: 586-876-6233, meagan.williams@jhu.edu 10 2 2022 27 1 2022 27 7 2023 115 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Background: A traumatic birth can profoundly affect a person’s well-being, relationships, and attachment with their infant. Addressing traumatic births at a population level requires accurate measurement of the prevalence, risk factors, and outcomes of psychological trauma after childbirth using validated instruments that distinguish perceptions of traumatic birth, subclinical post-traumatic stress (PTS) symptoms, and symptoms meeting a diagnostic threshold. Objective: The purpose of this study was to review literature on psychological trauma following childbirth and appraise instruments that measure postpartum PTS. Methods: In January 2020, the authors searched for and evaluated peer-reviewed studies that quantitatively measured PTS following hospital-based live births in the United States, United Kingdom, Canada, Australia, Norway, Sweden, and Switzerland; 33 articles were selected and evaluated. Results: Levels of post-traumatic stress disorder (PTSD) were most commonly measured, followed by PTS symptoms. Diagnostic instruments suggested lower estimates of PTS prevalence than those screening for or assessing PTS symptoms. Community samples yielded lower prevalence estimates than samples recruited from the internet or from settings specifically addressing mental health challenges. Measurement soon after birth yielded higher estimates. Conclusion: Study design, sample characteristics, instruments, and timing of measurements likely impact postpartum PTS prevalence estimates. Variation in these characteristics make it difficult to draw conclusions on the prevalence of postpartum PTS. Understanding and addressing postpartum PTS requires further research and collaboration using rigorous study methodology. Future researchers should consider the appraisal of measurement tools presented in this review when exploring variability of traumatic birth experiences across populations and evaluating evidence-based interventions. posttraumatic stress childbirth traumatic birth postpartum birth trauma measurement pmcIntroduction Recent estimates suggest that between 20–48% of women describe childbirth as ‘traumatic’ (Simpson & Catling, 2016).1 While childbirth can trigger memories of past trauma among survivors of sexual violence or abuse, traumatic birth experience is not limited to people with prior trauma (Murphy & Strong, 2018). Psychological birth trauma – distinct from physical trauma to the body resulting from childbirth – is often perceived as being ‘in the eye of the beholder’ (Beck, 2004) and may go unpredicted, unnoticed, or dismissed by others attending the birth (Beck, 2004; Reed et al., 2017). There is no standard definition of a psychologically traumatic birth (Beck, 2004). The experience of psychological trauma following childbirth may be influenced by factors related to an individual’s personal history and health, the birth environment and experience, the newborn’s health and safety, and the clinical management of obstetric events and pain (Beck, 2004; Harris & Ayers, 2012; Simpson & Catling, 2016). While some people who perceive their birth as traumatic may not experience negative long-term consequences, others may develop post-traumatic stress (PTS) symptoms after childbirth; a subset of these individuals will meet clinical diagnostic criteria for post-traumatic stress disorder (PTSD) (Beck, 2004; 2013). The lack of a standard definition for psychologically traumatic birth makes it difficult to estimate the percentage of individuals who experience trauma as well as its potential long-term effects. In an early prospective study of postpartum PTSD, Ayers and Pickering (2001) found that 1.5% of women in their sample met diagnostic criteria for PTSD six months after birth (Ayers & Pickering, 2001). More recent studies estimate the prevalence of PTSD following childbirth to be 3–4% among the general population (Dikmen Yildiz et al., 2016; Grekin & O’Hara, 2014). Estimates for some at-risk populations, such as individuals with co-occurring mental illness, prior traumatic experiences, or obstetric complications during delivery, suggest a burden between 15 and 30% (see Grekin & O’Hara, 2014; Shaw et al., 2012; Wijma et al., 1997). We must rely on quantitative data to determine the prevalence and risk and protective factors for PTS following childbirth and to inform the development and evaluation of appropriate clinical and public health interventions. However, the quality of quantitative research on postpartum PTS is inconsistent. Alcorn et al. (2010) note several common methodological shortcomings in the literature on postpartum PTS such as inadequate sample sizes, varying inclusion criteria and retention rates, and failure to include potentially important confounders like mental health history (Alcorn et al., 2010). The literature also varies with respect to the distinction between ‘traumatic birth,’ PTS symptoms, and PTSD diagnosis. In obtaining data on prevalence, risk factors, and outcomes of PTS following childbirth, it is important to distinguish between individuals who perceive their birth as traumatic, those who experience subclinical PTS symptoms for some amount of time, and those who develop symptoms meeting a diagnostic threshold. Instruments used to measure postpartum PTS also vary in what they intend to measure. Given the gaps in existing research, it is difficult to draw conclusions about the prevalence and determinants of PTS symptoms and PTSD following a traumatic childbirth experience. The purpose of the current study was to systematically review literature on psychological trauma following childbirth to appraise the instruments used to measure postpartum PTS. The methods used to obtain articles for review are described first, then the commonly used quantitative instruments. The studies and the methods by which the instruments were applied are then evaluated, followed by a discussion of why understanding variability among the studies is a necessary requisite to drawing inferences from the literature. Methods Search strategy In January 2020, the primary author (MEW) searched PsycINFO, CINAHL, PubMed, and SCOPUS for relevant articles using iterations of the following search terms and platform-dependent search strategies, which were identified in consultation with a public health informationist: postpartum OR postnatal OR perinatal OR after delivery *birth* OR childbirth OR labor PTSD OR traumatic OR post traumatic OR psychological trauma The initial search resulted in 1,454 articles for further review. After removing 528 duplicates, the primary author examined the titles and abstracts of the 926 remaining articles to identify studies eligible for inclusion. Only studies published in English were included. Geographic locations were purposively selected to include studies from the United States, Canada, United Kingdom (England, Scotland, Wales, or Northern Ireland), Scandinavian countries (Sweden, Norway, or Denmark), Australia, and Switzerland. Additional countries besides the US were selected because research on this topic is quite limited in the US. The other locations are high-resource settings in which the research topic has been explored, with most studies taking place in the UK and Australia. Studies were included if they quantitatively measured PTS experienced by a birthing mother after delivery of a liveborn infant in a hospital setting and if data were collected from samples in the intended geographic locations. Qualitative studies, literature reviews, studies assessing or validating measurement tools, studies only using single-item measures, and articles that did not identify a quantitative tool or did not report results of the measures were excluded from the analysis. Quantitative articles were only included if the authors indicated that PTS resulted from a traumatic birthing experience; studies were excluded that measured PTS unrelated to childbirth. To eliminate potential confounding due to infant morbidity or mortality (see Andersen et al., 2012; Gold & Johnson, 2014), studies were excluded which focused on PTS after stillbirth or among mothers whose infants had health complications requiring admission to the neonatal intensive care unit. Some studies compared groups who did and did not meet inclusion criteria – for example, Silverstein et al. (2019) compared women with and without birth complications requiring an emergency response (Silverstein et al., 2019). In these cases, the authors included only studies that reported PTS rates for the included and excluded groups separately. If multiple articles used data from the same study, the most relevant study or the one that provided the most detailed explanation of the methods was selected. The authors deemed 145 articles eligible for full-text review. After all articles were reviewed by the primary author, 37 were selected for inclusion in this review. The process of selecting studies for inclusion in this review is summarised in Figure 1. The articles were reviewed and organised in Covidence, a systematic review management software (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia; available at www.covidence.org). Data extraction was performed by the primary author. Quantitative instruments used to measure postpartum post-traumatic stress Several instruments have been used to measure the experience of postpartum PTS. Some instruments screen for PTS symptoms, while others aim to measure PTSD as a clinical phenomenon. Clinical diagnosis of PTSD requires alignment with specific criteria set by the Diagnostic and Statistical Manual of Mental Disorders (DSM), the language of which has been critiqued and challenged since the introduction of PTSD into the DSM-III in 1980 (North et al., 2016). Early versions of the DSM defined a ‘stressor’ (i.e., the traumatic event that leads to PTSD) in a way that excluded childbirth as a potentially traumatic event, given its ‘physiologic and predictable’ nature (Modarres et al., 2012). However, this definition was modified in the DSM-IV in 1994, allowing for interpretation of this definition to include childbirth. Diagnostic criteria for PTSD have changed over time and, accordingly, their meaning and implications. Several instruments have been created to align with these changes, and the language used to describe these instruments reflects language used in the DSM-III through DSM-5 (e.g., reexperiencing, avoidance, arousal, intrusion). Changes made to the symptom criteria for a PTSD diagnosis have been explained and critiqued elsewhere (see North et al., 2016). These changes were considered when drawing conclusions across studies. The results of our analysis focus on three commonly used quantitative instruments summarized in Appendix A. The Impact of Event Scale (IES) (Horowitz, Wilner, & Alvarez, 1979) and Impact of Event Scale-Revised (IES-R) (Weiss & Marmar, 1997) measure PTS symptomatology. The IES, used in fifteen studies, measures symptoms of intrusions and avoidance after a distressing life event. The revised version of the IES, included in six studies, reflects the addition of hyperarousal symptoms to the DSM criteria for PTSD diagnosis (Weiss & Marmar, 1997). The Post-traumatic Stress Disorder Questionnaire (PTSD-Q) (Cross & McCanne, 2001), a self-report screening questionnaire corresponding to DSM-III-R criteria, was the most frequently used PTSD screening tool and was reported in five studies. Details of the development and psychometric properties including internal reliability, sensitivity and specificity, and recommended cut-off scores of these and other, less commonly used instruments are outlined in Appendix A. Appendix A also details whether each instrument is used to identify PTS symptoms or serves as a PTSD screening or diagnostic tool. Results The 37 studies identified in this review are shown in Appendix B and are numbered in alphabetical order; the Appendix includes the country, sample size, methods used to measure psychological trauma, and relevant results for each study. The numbered references in the section below refer to the number for each study in the first column of Appendix B. Additional details of each study’s methods are outlined in Appendix C. IES & IES-R Fifteen studies used the IES (2–5, 8, 9, 12–14, 18, 19, 21, 22, 24, 26) to measure PTS symptoms in the postpartum period. Samples and study designs varied widely (see Table 1). The sample size for nine studies was less than 100 participants (2–5, 8, 13, 22, 24, 26), and six studies (9, 12, 14, 18, 19, 21) had sample sizes over 100 participants, the highest of which was 1,893 (19). Most investigators recruited participants from hospital postpartum units or perinatal care settings; McDonald et al. (2011) recruited participants from a previous study on childbirth-related PTS symptoms (26). The timing of measurement after birth also varied, although most studies collected data at a single timepoint; two studies (21, 24) collected data two and three times after birth. Five studies measured symptoms within one week (3–5, 21, 24), and nine between one week and three months postpartum (8, 12–14, 18, 19, 21, 22, 24). Iles et al. (2011) measured symptoms at three months (21), Baxter (2019) at four to five months (9), Anderson and McGuinness (2008) at nine months (2), and McDonald et al. (2011) up to two years after birth (26). The reported prevalence and severity of postpartum PTS based on the IES varied depending on the cut-off point. Horowitz (1982) recommended scoring cut-offs of ‘low’ (0–8), ‘moderate’ (9–19), and ‘severe’ (20+) for the IES. Two studies followed these recommendations and reported rates of severe symptoms between 2–5% in their samples (8, 24). Cut-off values used in the remaining studies diverged from the recommendations. Baxter (2019) categorised trauma scores as ‘low’ (0–8) or ‘high’ (≥9) and reported that 37% of participants had high PTS symptoms (9). Czarnocka and Slade (2000), Davies et al. (2008), and McDonald et al. (2011) categorised cases as ‘clinically significant’ if their scores were 19 or above; rates ranged between 9.9% and 21.3% in their samples, with the highest from the sample recruited by McDonald et al. (2011) in a previous study of postpartum PTS (12, 14, 26). Furuta et al. (2014) used a higher clinically significant cut-off value of 20, the severe cut-off category suggested by Horowitz, and estimated clinically significant symptoms of intrusions among 6.4% of participants and avoidance among 8.4% (18). Garthus-Niegel et al. (2014) reported that 1.9% of women in their sample scored higher than 34 on the IES, indicating a ‘likely PTSD condition’ (19). Other investigators categorised cases as ‘mild’ (scores: 19–25) or ‘moderate to severe’ trauma or PTS (≥26) (2, 5). Using these values, Anderson & McGuinness (2008) indicated that 21.4% of their sample showed mild PTS and 7% showed moderate to severe PTS; Anderson (2010) reported rates of 11.8% mild and 41.2% moderate to severe (2, 5). Several studies did not report cut-off scores and instead reported group means: Dale-Hewitt et al. (2012) reported a mean score of 17.21 at 6 weeks (13); Iles et al. (2011) reported mean scores of 13.33, 10.55, and 6.81 at 7 days, 6 weeks, and 3 months, respectively (21); and Keogh et al. (2002) reported a group mean of 10.13 at 2 weeks (22). Garthus-Niegel et al. (2014) reported a group mean score of 6.3 at 8 weeks (19). While the IES measures symptoms of intrusions and avoidance, which are symptoms of PTSD gauged using the DSM, it does not map onto DSM criteria and is not used to determine whether PTSD is present. Rather, several studies used another common scale (i.e., PTSD-Q or PSS) in addition to the IES to present data on the presence of PTSD in their study sample (12–14, 21, 22). Six studies used the IES-R (16, 20, 27, 29, 30, 37); most investigators using this scale sought to determine clinical significance of PTS symptoms. Details of studies using the IES-R are outlined in Table 2. The sample size for three studies was between 100–300 (16, 27, 30), while the remaining studies had sample sizes of 502 (37) and over 1,000 participants (20, 29). Edworthy (2008) and Silverstein (2019) measured symptoms at 6 weeks post-birth (16, 30), Holt (2018) at 2–6 months (20), Williams (2016) within 1 year (37), and Mitchell (2018) within two years (27). Priest (2003) took measurements at 2 months, 6 months, and 1 year; however, comparisons were not made between measurements at these time points and results were only reported as a diagnosis within the first year (29). Three studies recruited participants from an antenatal clinic or postpartum unit (16, 29, 30); three recruited participants online (19, 27, 37). Holt et al. (2018), Mitchell et al. (2018), and Williams et al. (2016) used a cut-off score on the IES-R of 33 as ‘clinically significant’ (20, 27, 37), while Silverstein (2019) used a score of 24 (30). In the two studies using the recommended cut-off of 33, 8% of participants in the study by Mitchell et al. (2018) (27) showed clinically significant PTS symptoms and 18.9% did so in the study by Williams et al. (2016) (37). Edworthy et al. (2008) and Priest et al. (2003) did not report cut-off values (16, 29). Silverstein et al. (2019) reported a median score of 2 on the IES-R among participants not requiring an emergency team response (2018). Edworthy et al. (2008) (16) and Holt et al. (2018) (20) reported group means for the IES-R (7.51 and 14.98, respectively). The group mean of 14.98 should be interpreted with caution, as Holt et al. recruited women online and recruitment may have been biased toward women who experienced a particularly traumatic birth (20). PTSD-Q Five studies used the PTSD-Q; all used it along with the IES (12–14, 21, 26). These studies are summarized in Table 3. The studies by Dale-Hewitt et al. (2012) and McDonald et al. (2011) had sample sizes under 100 participants (13, 26), and the sample sizes for the remaining three were 211 (14), 264 (12), and 372 (21). Four of the five studies took measurements six weeks after birth (12–14, 21). Iles et al. (2011) also took measurements within 7 days and 3 months after birth (21). McDonald et al. (2011) measured symptoms within 2 years (26). Three studies recruited from hospital postpartum units (12, 14, 21), one from women attending a six-week health check (13), and one from a previous study of postpartum PTS (26). Cross and McCanne (2001) recommend an individual cut-off score of 60 on the PTSD-Q for diagnostic accuracy against the DSM-IV. Iles, Slade, and Spiby (2011) reported group mean scores of 29.46 at six weeks and 26.95 at three months (21), and Dale-Hewitt et al. (2012), a group mean score of 38.5 at six weeks (12). Information about the distribution of individual scores was unavailable in both studies (12, 21). Other investigators classified cases as “clinically significant” if scores on at least one item on the intrusion subscale, three on the avoidance subscale, and two on the hyperarousal subscale of the PTSD-Q were greater than or equal to four. Clinically significant levels of PTS symptoms ranged from 17.3% (at two years) (26) to 27.2% (at six weeks) (12). Rates of participants meeting PTSD diagnostic criteria ranged from 3–4% (Czarnocka & Slade, 2000; Davies et al., 2008) to 8% (Dale-Hewitt et al., 2012) in studies assessing this proportion (12–14). Discussion The aim of this study was to review and appraise the use of instruments that measure postpartum PTS. Prevalence estimates of PTS varied widely between studies based on the IES, IES-R, and PTSD-Q. Lower PTS estimates resulted from instruments using diagnostic criteria relative to those based on symptoms or screening instruments. Estimates of PTS prevalence were higher for symptomatology or screening instruments when cut-off values were applied in or below the range defined by their creators. We also found that several factors in addition to instrumentation and reporting may impact estimates of postpartum PTS prevalence, including the sampling frame and time of assessment. Lower estimates were found in community samples than among women recruited from the internet or from groups known to have experienced trauma. Timing also mattered; measurements closer to birth generally appeared to yield higher prevalence estimates. Longitudinal research, however, suggests that many factors are associated with the trajectory of PTSD over time (Dikmen-Yildiz, Ayers, & Phillips, 2018), and the relationship between timing and the trajectory of PTS symptoms should continue to be explored in future studies. It is challenging to draw conclusions about prevalence of PTS based on the literature due to variation in the reporting of results, sample characteristics, instrumentation, and timing of measurements. The very act of recruiting participants for a study focused on traumatic childbirth experiences – particularly online recruitment – likely yields a sample who experiences more symptoms and trauma than is seen in the general population. Results generally suggest that reports of postpartum PTS are lower when measured later after birth, suggesting that women likely have different needs for support and clinical care depending on the time of measurement. The results of studies that use internet-based recruiting, use a cut-off score lower than is recommended for adequate sensitivity and specificity, or do not report recruitment methods or timing of measurements should be interpreted with caution. Postpartum PTS was evaluated based on clinical criteria for PTSD in some studies and on PTS symptoms in other studies. It appears that far more women experience subclinical PTS symptoms following a traumatic childbirth than meet diagnostic criteria for PTSD. While the clinical diagnostic tools for PTSD can be used to assess postpartum PTSD, they were not developed for the purpose of evaluating the impact of traumatic birth experiences during the postpartum period. Two scales have been developed specifically to gauge PTS in the postpartum period: the Perinatal PTSD Questionnaire (PPQ) (Hynan, 1998), used by two studies in this review (Leeds & Hargreaves, 2008; MacKinnon et al., 2017), and the Post-Traumatic Childbirth Stress Inventory (PTCS) (Sorenson 2000, 2003), used by one (Sorenson & Tschetter, 2010). Both scales can be used to assess symptomatology, and the PPQ to screen for postpartum PTSD. They have been used, however, on a very limited basis; this review focused on more commonly used, widely established instruments. The results of the studies were reported using means, medians, or the percentage of participants with different values, each way with its benefits and disadvantages. While understanding a group mean score in relation to other factors is important, this number alone does not allow researchers to understand whether some women experience more severe symptomatology, who those women are, and factors that influence their experience. The use of a cut-off score for clinical significance allows for stratification based on symptomatology, and the placement of this cut-off has implications for analysis. While a higher cut-off may yield fewer cases, it allows researchers to understand who experiences the most severe symptomatology; a lower cut-off presents a wider range of PTS symptom severity in the study sample. Research on other types of screening instruments used during the postpartum period, such as the Edinburgh Postnatal Depression Scale, suggests that modifying the cut-off scores may have a substantial impact on estimates (Matthey, Henshaw, Elliott, & Barnett, 2006). Researchers comparing studies of postpartum PTS should consider whether variation in prevalence estimates presented in the literature is due to variation in the instruments used to measure PTS symptoms or other aspects of the studies, rather than true differences among the study samples. This review identified many studies involving quantitative measures of PTS following childbirth, indicating cultural and scientific recognition of postpartum PTS as an issue. Several studies used multiple instruments, which allows stronger conclusions to be drawn from their results. The wide range of years during which the studies occurred is a potential limitation, especially given changes to DSM criteria over time. Another limitation is that the inferences and conclusions derived from this review are limited to the geographic areas in which these studies took place; a review of studies from other settings (for example, low- and middle-income countries) may yield different findings. This review does not address the social-structural context in the selected study areas, although we recognize its potential impact on PTS measurement. The initial search and full-text review were conducted by only the primary author, which may have introduced some bias in the selection of studies. Finally, the authors did not conduct a standardised quality assessment of the studies included in this review; the purpose of our initial screening was to identify studies which used the various measures, not to evaluate the overall quality of the studies selected. Using a validated, appropriate scale is only one factor in ensuring scientific validity of inferences about the impact of birth experiences on postpartum stress response. Research suggests that postpartum PTS symptomatology may be influenced by an interplay of participants’ characteristics, such as negative past childbirth experiences; feelings of powerlessness, lack of control, or extreme pain during labor; history of sexual or domestic violence, emotional trauma, or physical assault; perceived or actual health of the infant; and prior mental health conditions (see Ayers, 2004). Each of these factors may influence the emotional response to childbirth. Some investigators in this review excluded women with certain characteristics that may predispose them to experiencing PTS (e.g., Davies et al., 2008; Iles et al., 2011); others took these potential risk factors into account during their analyses (e.g., Alcorn et al., 2010). These risk factors need to be considered in future research in interpreting PTS prevalence estimates as well as adjustment for participant characteristics such as age, parity, race, ethnicity, partnership or marital status, and physical health outcomes of the parent and newborn. Postpartum PTS can have profound effects on a person’s health, well-being, relationships, and attachment to their infant (Ayers & Pickering, 2001). Addressing these issues with effective clinical and public health interventions requires additional research and collaboration to determine the prevalence and risk and protective factors of PTS after childbirth. Evaluation of such interventions must also involve appropriate instruments to measure PTS. Future researchers should consider the appraisal of measurement tools illustrated throughout this review and continue to explore variability of traumatic birth experiences and outcomes among different populations using rigorous study design in order to yield reliable estimates of prevalence of postpartum PTS and PTSD. Acknowledgements The authors thank public health informationist Donna Hesson, MLS, for her help in developing the search strategy for this paper. The corresponding author, Meagan Williams, would also like to express sincere gratitude toward Dr. Julia Seng at the University of Michigan for inspiring her to pursue this topic. Funding details This work was supported in part by funding from the Maternal and Child Health Bureau (HHS) under Grant T76MC00003 and the National Institute on Minority Health and Health Disparities (1L60MD012089, Holliday). Appendix A: Commonly used measurement tools Measurement tool Development; initial sample About PTS/PTSD Internal reliability (Cronbach’s alpha) Sensitivity, specificity Recommended cutoffs Impact of Event Scale (IES) (Horowitz, Wilner, & Alvarez, 1979) Developed using in-depth evaluation and psychotherapy interviews 16 men and 50 women between 20–75 years of age, lower-middle to middle class, in outpatient clinic for treatment of PTSD symptoms after a serious life event 4-point scale (scored as 0, 1, 3, 5) 15 items: 7 intrusion questions (subscale score 0–35) and 8 avoidance questions (subscale score 0–35) PTS symptoms 0.86 overall 0.78 for intrusion subscale and 0.82 avoidance Sensitivity 0.89, specificity 0.88 (using cut-off of 35) against SCID* 0–8 = low symptoms, 9–19 = moderate, 20+ = severe Varies; Neal et al. (1994) recommends cut-off of 35 for diagnostic accuracy against DSM-III-R Impact of Event Scale – Revised (IES-R) (Weiss & Marmar, 1997) Revised the IES (Horowitz 1979) to include arousal items 22 items (original questions plus 7 arousal) PTS symptoms 0.89 overall Between 0.87–0.94 for intrusion subscale, 0.84–0.87 avoidance, and 0.79–0.91 hyperarousal Sensitivity 0.91, specificity 0.82 (using cut-off 33) against PCL Varies; Creamer et al. (2003) recommends cut-off of 33 for diagnostic accuracy against PCL Post-traumatic Stress Disorder Interview (PTSD-I) (Watson, Juba, Manifold, Kucala, & Anderson, 1991) Corresponds to DSM-III-R criteria 31 male Vietnam combat veteran inpatients in Minnesota Likert scale ranged 1 (no/never) to 7 (extremely/always) Trauma screening; 17 items about reexperiencing, avoidance, arousal; questions to assess whether symptoms have been present for at least 1 month PTSD screening 0.92 Sensitivity 0.89, specificity 0.94 against DIS† diagnosis 4 on each item for each subscale considered to meet DSM criteria for each subscale Post-traumatic Stress Disorder Questionnaire (PTSD-Q) (Cross & McCanne, 2001) Identical to PTSD-I, but a self-report questionnaire 822 undergraduate women, 77.8% Caucasian Identical to PTSD-I PTSD screening 0.97 overall Sensitivity 0.81, specificity 0.82 against SCID Recommended cut-off of 60 for diagnostic accuracy against DSM-IV Posttraumatic Stress Disorder Symptom Scale – Interview (PSS-I) (Foa, Riggs, Dancu, & Rothbaum, 1993) Corresponds to DSM-III-R criteria 118 women who experienced rape (46) or non-sexual assault (72) within the past 2 weeks, mean age 31 years, 70% African American, 29% white, no previous or current diagnosis of other mental disorder 4-point scale ranged 0–3 17 items: 4 reexperiencing, 7 avoidance, 6 arousal PTS symptoms; PTSD diagnosis 0.85 overall 0.69 for reexperiencing, 0.65 avoidance, 0.71 arousal Sensitivity 0.88, specificity 0.96 against SCID ≥12: elevated post-traumatic stress symptoms Positive PTSD diagnosis if rated 1 or higher on 1 intrusion, 3 avoidance, 2 arousal symptoms Posttraumatic Stress Disorder Symptom Scale – Self Report (PSS-SR) (Foa, Riggs, Dancu, & Rothbaum, 1993) Identical to PSS-I, but self-report version. Six items rewritten to increase clarity. Adapted by Ayers and Pickering (2011) to be specific to childbirth event Identical to PSS-I PTS symptoms; PTSD diagnosis 0.91 overall 0.78 reexperiencing, 0.80 avoidance, 0.82 arousal Childbirth-specific: 0.9 overall, 0.91 intrusion, 0.81 avoidance, 0.84 arousal Sensitivity 0.62, specificity 1.0 against SCID Identical to PSS-I Posttraumatic Diagnostic Scale (PDS) (Foa, Cashman, Jaycox, & Perry, 1997) Corresponds to DSM-IV criteria 248 volunteers who had experienced a wide variety of traumas (e.g., accident, fire, natural disaster, assault, combat) Likert scale 0 to 4 49 items: 12 trauma screening, 4 injury, 5 reexperiencing, 7 avoidance, 5 arousal, 9 impairment PTSD symptoms; PTSD diagnosis 0.92 overall 0.78 reexperiencing, 0.84 avoidance, 0.84 arousal Sensitivity 0.89, specificity 0.75 against SCID 0 = no rating, 1–10 = mild, 11–20 = moderate, 21–35 = moderate to severe, >36 = severe Positive PTSD diagnosis if presence of injury, sense of helplessness, or fear, 1 reexperiencing, 3 avoidance, 2 arousal, for 1 month, impaired functioning in at least one area Posttraumatic Diagnostic Scale for DSM-5 (PDS-5) (Foa et al., 2016a) Corresponds to DSM-5 criteria 242 veterans, college students, and community members who experienced a traumatic event Likert scale 0 to 4 Trauma screening; 24-item self-report measure: 5 intrusion, 2 avoidance, 7 changes in mood and cognition, 6 arousal / hyperactivity; additional questions for distress, interference, onset, and duration PTSD symptoms; PTSD diagnosis 0.95 overall sensitivity 0.84, specificity 0.73 against PSSI-5‡ ≥28 = probable PTSD diagnosis PTSD Checklist (PCL) (Weathers, Litz, Herman, Huska, & Keane, 1993) Corresponds to DSM-III-R criteria Two initial studies: - 123 male Vietnam veterans, receiving services or participating in research at the National Center for PTSD - 1006 male and female veterans, surveyed nationally Likert scale ranging 1–5 17 items PTSD screening 0.94 overall 0.93 for reexperiencing, 0.92 avoidance, 0.92 hyperarousal Sensitivity 0.82, specificity 0.83 (further information unavailable) ≥50 for sensitive PTSD diagnostic utility * SCID = Structured Clinical Interview for DSM-III-R (Spitzer et al., 1990) † DIS = Diagnostic Interview Schedule (Robins & Helzer, 1985) ‡ PSSI-5 – Posttraumatic Stress Disorder Symptom Scale Interview for DSM-5 (Foa et al., 2016b) Appendix B: Overview of selected studies (n=37) # Authors (Year) Country n* Instruments used Timing of measurements (after birth) Results specific to measurement of post-traumatic stress (PTS) or post-traumatic stress disorder (PTSD) 1 Alcorn, O’Donovan, Patrick, Creedy, & Devilly (2010) Australia 933 PDS 4–6 weeks, 12 weeks, 24 weeks 8.9% of women met PTSD criteria at least once over the three assessments. Controlling for anxiety, depression, and previous PTSD/trauma, PTSD rates were less at 1.2% at 4–6 weeks, 3.1% at 12 weeks and 3.1% at 24 weeks postpartum. 2 Anderson & McGuinness (2008) US 28 IES 9 months IES scores ranged from 1 to 34 (mean = 13.96, SD = 10.75). Six participants had mild PTS; 2 had moderate to severe PTS. 3 Anderson & Perez (2015) US 44 IES Other: self-appraisal Within 72 hours IES scores ranged from 0 to 52. 33.3% of participants reported subclinical PTS symptoms. 50% of participants had moderate to severe symptoms of acute traumatic stress. 22.7% appraised the birth experience as ‘awful’/ traumatic 4 Anderson & Strickland (2017) US 66 IES Other: self-appraisal Within 72 hours 4.5% of participants of reported IES scores over 34 (high risk of PTSD). 48.6% of participants reported mild to moderate symptoms of subjective distress. 3% had severe symptoms of subjective distress. 5 Anderson (2010) US 85 IES Other: self-appraisal Within 72 hours 53% of participants were mildly to severely impacted by the childbirth experience. 11.8% indicated mild trauma. 41.2% indicated moderate to severe trauma. On a scale of 0 (no trauma) to 10 (traumatic), 35% appraised their childbirth experience at 7 or greater. 6 Ayers & Pickering (2001) UK 289 PSS-SR 6 weeks, 6 months 2.8% of women met PTSD diagnostic criteria at 6 weeks postpartum and 1.5% at 6 months postpartum. 7 Ayers, Harris, Sawyer, Parfitt, & Ford (2009) UK 1423 PDS Within 1 year 2.5% of women from the community and 21% on the internet met PTSD diagnostic criteria. Many others endorsed individual PTSD symptom criteria. 8 Ayers, Wright, & Wells (2007) UK 64 couples (men and women) IES 9 weeks Three men and three women (5% of sample) had severe PTS symptoms. 9 Baxter (2019) UK 170 IES 4–5 months 37% of participants had high PTS symptoms. 10 Beck, Gable, Sakala, & Declercq (2011) US n/a PSS-SR Not specified 9% of sample met PTSD diagnostic criteria. 18% of women scored above the cutoff score on the PSS-SR, indicating elevated levels of PTS. 11 Creedy, Shochet, & Horsfall (2000) Australia 499 Other: self-reflection PSS-I 4–6 weeks 33% of participants identified a traumatic birthing event and reported at least 3 trauma symptoms. 5.6% of participants met PTSD diagnostic criteria. 22.6% reported some PTS symptoms but did not meet clinical diagnostic criteria. 12 Czarnocka & Slade (2000) UK 264 IES PTSD-Q 6 weeks 9.9% of participants reported clinically significant levels of intrusions or avoidance. 3% of participants met PTSD diagnostic criteria. 27.2% reported some clinically significant PTS symptoms but did not meet clinical diagnostic criteria. 13 Dale-Hewitt, Slade, Wright, Cree, & Tully (2012) UK 50 IES PTSD-Q 6 weeks 8% of participants met PTSD diagnostic criteria. The total mean score for the IES was 17.28 (SD: 16.56). The total mean score for the PTSD-Q was 38.50 (SD: 15.71). 14 Davies, Slade, Wright, & Stewart (2008) UK 211 IES PTSD-Q 6 weeks 3.8% of participants met PTSD diagnostic criteria. 21.3% of participants reported some clinically significant PTS symptoms but did not meet clinical diagnostic criteria. 15 Dekel, Ein-Dor, Dishy, & Mayopoulos (2019) US 685 PCL Other: PDI, PDEQ 3 months (on average) 18% of participants were classified as having probable childbirth-related PTSD. 16 Edworthy, Chasey, & Williams (2008) UK 121 IES-R 6 weeks The group mean for the IES-R was 7.51 (SD: 8.91). The mean item score for the intrusion subscale was 3.52 (SD 4.11), avoidance 2.44 (SD 3.79) and hyperarousal 1.45 (SD 2.36). 1% of participants met PTSD diagnostic criteria. 8.3% showed medium distress across all subscales. 9.1% of participants met clinically significant PTS symptoms on the hyperarousal subscale and 1.7 on the intrusion subscale. 17 Ford & Ayers (2011) UK 138 PDS 3 weeks, 3 months At 3 weeks: 0.8% of participants met all PTSD diagnostic criteria except the duration criterion. The group mean symptom score was 4.2 (SD: 6.0). At 3 months, 0.9% of participants met PTSD diagnostic criteria. The group mean symptom score was 4.2 (SD: 5.6). 18 Furuta, Sandall, Cooper, & Bick (2014) UK 1824 IES 6–8 weeks (most) 6.4% of participants reported clinically significant levels of intrusions and 8.4% avoidance. 19 Garthus-Niegel, Knoph, von Soest, Nielsen, & Eberhard-Gran (2014) Norway 1893 IES 8 weeks 1.9% of women were reported to “likely” have PTSD. The group mean for the IES was 6.73 (SD: 8.10). 20 Holt, Sellwood, & Slade (2018) UK 1393 IES-R Other: self-appraisal 2–6 months The group mean for the IES-R was 14.98 (SD: 15.1). The group mean for the trauma appraisal was 0.7 (SD: 0.85). 21 Iles, Slade, & Spiby (2011) UK 372 IES PTSD-Q Within 7 days, 6 weeks, 3 months The group mean for the IES was 13.33 (SD: 11.64) at T1, 10.55 (SD: 10.78) at T2, and 6.81 (SD: 8.61) at T3. The group mean for the PTSD-Q was 29.46 (SD: 11.96) at T2 and 26.95 (SD: 9.73) at T3. 22 Keogh, Ayers, & Francis (2002) UK 40 IES PSS-SR 2 weeks The group mean for the IES was 10.13 (SD: 12.23). The mean item score for the intrusion subscale was 5.30 (SD: 7.19) and avoidance 4.83 (SD: 6.36). The group mean for the PSS-SR was 9.88 (SD: 8.96). The mean item score for the intrusion subscale was 2.23 (SD: 3.05), numbing 3.75 (SD: 3.84), and arousal 3.90 (SD: 3.19). 23 Leeds & Hargreaves (2008) UK 102 PPQ 6–12 months 3.9% of participants had clinically significant levels of PTSD. 19.6% reported some PTS symptoms but did not meet clinical diagnostic criteria. 24 Lyons (1998) UK 42 IES 4 days, 1 month Three mothers (7.1%) had IES scores in the medium distress range. One mother (2.4%) had an IES score of 56, in the high distress range. 25 MacKinnon, Yang, Feeley, Gold, Hayton, & Zelkowitz (2017) Canada 341 PPQ 7–9 weeks Mean PPQ score of 4.60 for hospital births and 4.43 for birthing center births. Prevalence estimates for births without complications not presented separately from births with complications (i.e., those requiring transfer to hospital). 26 McDonald, Slade, Spiby, & Iles (2011) UK 81 IES PTSD-Q 2 years 17.3% of participants had clinically significant levels of some PTS symptoms. 27 Mitchell, Whittingham, Steindl, & Kirby (2018) Australia 262 IES-R Within 2 years 8.0% of participants had clinically significant levels of PTS symptoms pre-intervention. 61.9% of those participants were in the non-clinical range by 1 month post-intervention 28 Parfitt & Ayers (2009) UK 152 parents (126 women) PDS Within 2 years 5.6% of women met PTSD diagnostic criteria. 29 Priest, Henderson, Evans, & Hagan (2003) Australia 1745 IES-R 2 months, 6 months, 1 year 0.8% of participants in the control group and 0.6% in the intervention group were diagnosed with a stress disorder in the year after giving birth 30 Silverstein, Centore, Pollack, Barrieau, Gopalan, & Lim (2019) US 249 IES-R PCL 6 weeks The median score for the IES-R was 16 among participants requiring an emergency team response (ETR) and 2 among participants not requiring an ETR. The median score for the PCL was 22.5 among participants requiring an ETR and 20 among participants not requiring an ETR. 31 Söderquist, Wijma, & Wijma (2006) Sweden 1224 TES 1 month, 4 months, 7 months, 11 months 1.3% of participants met criteria B, C, and D 1 month postpartum. At 4 months postpartum, 1.7% met criteria; at 7 months, 1.7%; and at 11 months, 0.9%. 32 Soet, Brack, & Dilorio (2003) US 103 TES 4 weeks (approx.) 34% of participants reported their childbirth experience as traumatic. 1.9% met PTSD diagnostic criteria. 30.1% reported some PTS symptoms but did not meet clinical diagnostic criteria. 33 Sorenson & Tschetter (2010) US 71 Other: PTCS 6–7 months 44.9% of participants reported PTS symptoms in the ‘low trauma’ range, 44.9% ‘low-middle’, 7.3% ‘high-middle’, and 2.9% ‘high trauma’. 34 Thompson, Roberts, & Ellwood (2011) Australia 206 PCL 2 months, 4 months 5% of participants showed evidence of PTSD at two months and 3% at four months 35 Verreault et al. (2012) Canada 308 Modified PSS-SR Other: SCID-I 1 month, 3 months, 6 months MPSS-SR: 7.6%, 6.1%, and 4.9% of participants showed ‘full PTSD’ and 16.6%, 5.3%, 4.3% ‘partial PTSD’ at 1, 3, and 6 months, respectively. SCID-I: 1.1%, 0.8%, and 0.0% of participants showed ‘full PTSD’ and 3.2%, 1.2%, 2.9% ‘partial PTSD’ at 1, 3, and 6 months, respectively. 36 White, Matthey, Boyd, & Barnett (2006) Australia 400 DSM criteria PSS-S 6 weeks, 6 months, 12 months 2% of participants met PTSD diagnostic criteria at 6 weeks postpartum. 10.5% reported some PTS symptoms but did not meet clinical diagnostic criteria. 2.6% of participants met PTSD diagnostic criteria at 6 months, and 2.4% at 12 months. 37 Williams, Taylor, & Schwannauer (2016) UK 502 IES-R Within 1 year 18.9% of participants had clinically significant levels of PTS symptoms. * Sample size (n) refers to the number of birthing mothers included in the study, unless otherwise noted Appendix C: Studies and methods Authors (Year) Years of data collection Country Recruitment details; inclusion criteria Instruments used Timing of measurements (after birth) Measurement details (e.g., mean, cut-offs) Results specific to measurement of post-traumatic stress (PTS) or post-traumatic stress disorder (PTSD) Alcorn, O’Donovan, Patrick, Creedy, & Devilly (2010) n/a Australia Recruited 933 women from antenatal clinics during pregnancy; >18 years old, in 3rd trimester, able to read/write English, contactable by phone PDS 4–6 weeks, 12 weeks, 24 weeks ‘PTSD’ if all criteria met “Partial PTSD’ if met all criteria except for 1 or 2 of the avoidance and/or hyperarousal symptoms; 8.9% of women met PTSD criteria at least once over the three assessments. Controlling for anxiety, depression, and previous PTSD/trauma, PTSD rates were less at 1.2% at 4–6 weeks, 3.1% at 12 weeks and 3.1% at 24 weeks postpartum. Anderson & McGuinness (2008) n/a USA Recruited 28 adolescent women from hospital postpartum units; 13–19 years old, spoke English or Spanish, accessible for follow-up IES 9 months 19–25 = ‘mild PTS’, >26 = ‘moderate to severe PTS’ IES scores ranged from 1 to 34 (mean = 13.96, SD = 10.75). Six participants had mild PTS; 2 had moderate to severe PTS. Anderson & Perez (2015) n/a USA Recruited 44 adolescent women from hospital postpartum units; 13–19 years old, experienced live cesarean birth, spoke English or Spanish IES Other: self-appraisal Within 72 hours IES: 0–8 = ‘no symptoms’ or ‘subclinical’, 9–25 = ‘mild symptoms’, 26–43 = ‘moderate symptoms, 44–75 = ‘severe symptoms’ Subjective appraisal of birth experience: rated between 1 (‘great/nontraumatic’) and 10 (‘awful/traumatic’). 1–3 = ‘nontraumatic’, 8–10 = ‘traumatic’ IES scores ranged from 0 to 52. 33.3% of participants reported subclinical PTS symptoms. 50% of participants had moderate to severe symptoms of acute traumatic stress. 22.7% appraised the birth experience as ‘awful’/ traumatic. Anderson & Strickland (2017) n/a USA Recruited 66 adolescent women from hospital postpartum unit; 13–19 years old, Hispanic, spoke English or Spanish IES Other: self-appraisal Within 72 hours IES: 0–8 = ‘subclinical’, 9–25 = ‘mild’, 26–43 = ‘moderate’, ≤44 = ‘severe’; scores >34 = ‘high risk of PTSD’ Subjective appraisal: rated between 1 ‘great’ and 10 ‘awful’. >6 rated ‘high negative appraisal’ 4.5% of participants of reported IES scores over 34 (high risk of PTSD). 48.6% of participants reported mild to moderate symptoms of subjective distress. 3% had severe symptoms of subjective distress. Anderson (2010) n/a USA Recruited 85 adolescent women from hospital postpartum units; 13–19 years old, spoke English or Spanish, accessible for follow-up IES Other: self-appraisal Within 72 hours 19–25 = ‘mild trauma’, >26 = ‘moderate to severe trauma’ Subjective appraisal: rated birth experience from 0 (no trauma) to 10 (very traumatic). No cutoffs provided. 53% of participants were mildly to severely impacted by the childbirth experience. 11.8% indicated mild trauma. 41.2% indicated moderate to severe trauma. On a scale of 0 (no trauma) to 10 (traumatic), 35% appraised their childbirth experience at 7 or greater. Ayers & Pickering (2001) 1996–1998 UK Recruited 289 women from antenatal clinics during pregnancy; between 16–36 weeks’ gestation, planning vaginal delivery, spoke English well PSS-SR 6 weeks, 6 months ‘PTSD’ if symptom frequency ≥18 and disability ≥2 2.8% of women met PTSD diagnostic criteria at 6 weeks postpartum and 1.5% at 6 months postpartum. Ayers, Harris, Sawyer, Parfitt, & Ford (2009) n/a UK n=1,423 women; recruited 502 women from hospital and antenatal clinics, 921 women online; inclusion criteria not specified PDS Within 1 year ‘PTSD’ if all criteria met 2.5% of women from the community and 21% on the internet met PTSD diagnostic criteria. Many others endorsed individual PTSD symptom criteria. Ayers, Wright, & Wells (2007) n/a UK Identified 64 couples (64 men and 64 women) from maternity ward registers at hospital; married or long-term partnered couples, healthy live birth attended by male partner IES 9 weeks 0–8 ‘low symptoms’, 919 ‘moderate’, 20+ ‘severe’ Three men and three women (5% of sample) had severe PTS symptoms. Baxter (2019) 2013 UK n=170; all women who gave birth at study hospital during June 2013 IES 4–5 months 0–8 ‘low’, ≥9 ‘high’ 37% of participants had high PTS symptoms. Beck, Gable, Sakala, & Declercq (2011) 2006 USA 1,373 women completed survey online and 200 by telephone; recruitment details and inclusion criteria not specified PSS-SR Not specified Positive screen indicating high likelihood of PTSD made in presence of at least 1 intrusion, 3 avoidance, and 3 arousal symptoms. Score of 12 or higher reflects suffering from some PTS symptoms. 9% of sample met PTSD diagnostic criteria. 18% of women scored above the cutoff score on the PSS-SR, indicating elevated levels of PTS. Creedy, Shochet, & Horsfall (2000) 1997–1998 Australia Recruited 499 women from antenatal clinics; >18 years old, in 3rd trimester, low obstetrical risk, able to understand English Other: self-reflection PSS-I 4–6 weeks PTSD diagnosis made in presence of 6 symptoms, with at least 1 reexperiencing, 3 avoidance, 2 arousal and event occurred at least 1 month ago 33% of participants identified a traumatic birthing event and reported at least 3 trauma symptoms. 5.6% of participants met PTSD diagnostic criteria. 22.6% reported some PTS symptoms but did not meet clinical diagnostic criteria. Czarnocka & Slade (2000) n/a UK Recruited 264 women from postpartum unit in hospital; >18 years old, healthy live birth, spoke and understood English, no plans to move out of area in near future IES PTSD-Q 6 weeks IES: clinically significant if score ≥19 PTSD-Q: scale from 1–7, clinically significant if ≥4 9.9% of participants reported clinically significant levels of intrusions or avoidance. 3% of participants met PTSD diagnostic criteria. 27.2% reported some clinically significant PTS symptoms but did not meet clinical diagnostic criteria. Dale-Hewitt, Slade, Wright, Cree, & Tully (2012) n/a UK Recruited 50 women during 6-week health check; >16 years old, birth between 6 weeks and 6 months prior to assessment, child did not receive special care for health problems, birthing parent perceived birth as ‘traumatic’. Excluded: colorblind participants, elective cesarean IES PTSD-Q 6 weeks Cut-offs not specified in article; means reported in Appendix 8% of participants met PTSD diagnostic criteria. The total mean score for the IES was 17.28 (SD: 16.56). The total mean score for the PTSD-Q was 38.50 (SD: 15.71). Davies, Slade, Wright, & Stewart (2008) n/a UK Recruited 211 women from postpartum unit in hospital; >16 years old, healthy infant with no adverse clinical event during labor and delivery, no maternal mental health or social problems, spoke and understood English IES PTSD-Q 6 weeks IES: ‘clinically significant’ if ≥19 PTSD-Q: significant if ≥4 3.8% of participants met PTSD diagnostic criteria. 21.3% of participants reported some clinically significant PTS symptoms but did not meet clinical diagnostic criteria. Dekel, Ein-Dor, Dishy, & Mayopoulos (2019) 2016–2017 USA Recruited 685 women online; ≥18 years old, healthy live birth within past 6 months PCL Other: PDI, PDEQ 3 months (on average) Classified as having ‘probable’ childbirth-related PTSD in presence of at least 1 intrusion item, 1 avoidance, 2 alterations in cognition and mood, and 2 reactivity and hyperarousal, with at least moderate severity 18% of participants were classified as having probable childbirth-related PTSD. Edworthy, Chasey, & Williams (2008) n/a UK Recruited 121 women from antenatal clinics and parenting classes; expecting first baby, no medical concerns about baby’s well-being, literate in English IES-R 6 weeks IES-R scores converted into percentages of maximum possible score and compared to IES categories of ‘low’, ‘medium’, and ‘high’ distress The group mean for the IES-R was 7.51 (SD: 8.91). The mean item score for the intrusion subscale was 3.52 (SD 4.11), avoidance 2.44 (SD 3.79) and hyperarousal 1.45 (SD 2.36). 1% of participants met PTSD diagnostic criteria. 8.3% showed medium distress across all subscales. 9.1% of participants met clinically significant PTS symptoms on the hyperarousal subscale and 1.7 on the intrusion subscale. Ford & Ayers (2011) n/a UK Recruited 138 women from hospital clinics; >18 years old, between 33rd-37th week of pregnancy, spoke English. Excluded if midwife felt it would be inappropriate to contact them after delivery PDS 3 weeks, 3 months ‘PTSD’ if all criteria met; means reported At 3 weeks: 0.8% of participants met all PTSD diagnostic criteria except the duration criterion. The group mean symptom score was 4.2 (SD: 6.0). At 3 months, 0.9% of participants met PTSD diagnostic criteria. The group mean symptom score was 4.2 (SD: 5.6). Furuta, Sandall, Cooper, & Bick (2014) 2010–2011 UK Recruited 1824 women from postnatal unit; >16 years old, live birth in hospital between June-December 2010 over 24 weeks’ gestation, read/understood English IES 6–8 weeks (most) Subscale score ≥20 defined as ‘clinically significant’ level of distress 6.4% of participants reported clinically significant levels of intrusions and 8.4% avoidance. Garthus-Niegel, Knoph, von Soest, Nielsen, & Eberhard-Gran (2014) 2008–2010 Norway Recruited 1,893 women from clinic during pregnancy after first ultrasound; able to complete a questionnaire in Norwegian, did not have Cesarean or have severe complications IES 8 weeks IES: ‘Likely PTSD condition’ if score >34; mean reported 1.9% of women were reported to ‘likely’ have PTSD. The group mean for the IES was 6.73 (SD: 8.10). Holt, Sellwood, & Slade (2018) n/a UK Recruited 1393 women online; 16–50 years old, gave birth to first baby 2–6 months before recruitment, proficient in English IES-R Other: self-appraisal 2–6 months IES-R: Clinical cutoff ≥33; mean reported Self-appraisal: binary responses, mean reported The group mean for the IES-R was 14.98 (SD: 15.1). The group mean for the trauma appraisal was 0.77 (SD: 0.85). Iles, Slade, & Spiby (2011) 2005–2006 UK Recruited 372 women from hospital postnatal ward; >16 years old, married or cohabitating, partner present for some or all of labor and delivery, no known domestic violence, child did not receive special medical attention for more than 24 hours, sufficient English IES PTSD-Q Within 7 days, 6 weeks, 3 months IES: cut-offs not specified PTSD-Q: ≥4 on at least one intrusion, 3 avoidance, 2 hyperarousal; item 12 removed The group mean for the IES was 13.33 (SD: 11.64) at T1, 10.55 (SD: 10.78) at T2, and 6.81 (SD: 8.61) at T3. The group mean for the PTSD-Q was 29.46 (SD: 11.96) at T2 and 26.95 (SD: 9.73) at T3. Keogh, Ayers, & Francis (2002) n/a UK Recruited 40 women from antenatal clinic during pregnancy; 36 weeks’ gestation, no major obstetric complications IES PSS-SR 2 weeks Calculated total scores, no cut-offs; means reported The group mean for the IES was 10.13 (SD: 12.23). The mean item score for the intrusion subscale was 5.30 (SD: 7.19) and avoidance 4.83 (SD: 6.36). The group mean for the PSS-SR was 9.88 (SD: 8.96). The mean item score for the intrusion subscale was 2.23 (SD: 3.05), numbing 3.75 (SD: 3.84), and arousal 3.90 (SD: 3.19). Leeds & Hargreaves (2008) n/a UK Recruited 102 women postnatally from hospital patient record; live births between October 2003-March 2004, no severe maternal mental health problems. Excluded if practitioner felt they would be distressed by participation PPQ 6–12 months Classified as ‘fully symptomatic’ if 1 reexperiencing, 3 avoidance, 2 increased arousal; ‘partially symptomatic’ if clinically significant in one area; ‘not significant’ if no symptoms 3.9% of participants had clinically significant levels of PTSD. 19.6% reported some PTS symptoms but did not meet clinical diagnostic criteria. Lyons (1998) n/a UK Recruited 42 women from postnatal unit of hospital; all available first-time mothers who had given birth in previous 4 days IES 4 days, 1 month 0–8 = ‘low distress’, 9–19 = ‘medium distress’, 20+ = ‘high distress’ for each subscale Three mothers (7.1%) had IES scores in the medium distress range. One mother (2.4%) had an IES score of 56, in the high distress range. MacKinnon, Yang, Feeley, Gold, Hayton, & Zelkowitz (2017) 2009–2011 Canada Recruited 341 women from prenatal appointments and birthing center information sessions; ≥18 years old, able to read and speak English or French; later excluded if medical complications, poor reading comprehension, or change in care site PPQ 7–9 weeks Clinically significant if ≥19; means reported Mean PPQ score of 4.60 for hospital births and 4.43 for birthing center births. Prevalence estimates for births without complications not presented separately from births with complications (i.e., those requiring transfer to hospital). McDonald, Slade, Spiby, & Iles (2011) n/a UK Recruited 81 women from previous study on childbirth-related post-traumatic stress symptoms; inclusion criteria not specified IES PTSD-Q 2 years Clinically significant if ≥19 17.3% of participants had clinically significant levels of some PTS symptoms. Mitchell, Whittingham, Steindl, & Kirby (2018) 2016 Australia Recruited 262 women online; ≥18 years old, healthy live birth within past 24 months, currently residing in Australia or New Zealand IES-R Within 2 years Clinically significant if ≥33 8.0% of participants had clinically significant levels of PTS symptoms pre-intervention. 61.9% of those participants were in the non-clinical range by 1 month post-intervention Parfitt & Ayers (2009) UK Recruited 152 parents (126 women) online; >18 years old, infants between 1 and 24 months old, spoke English PDS Within 2 years ‘PTSD’ if all criteria met 5.6% of women met PTSD diagnostic criteria. Priest, Henderson, Evans, & Hagan (2003) 1996–1997 Australia Recruited 1745 women by midwife within 24–72 hours of birth; ≥18 years old, delivered at over 35 weeks’ gestation, infant did not require intensive care, mother not under psychological care IES-R 2 months, 6 months, 1 year No cut-offs reported 0.8% of participants in the control group and 0.6% in the intervention group were diagnosed with a stress disorder in the year after giving birth Silverstein, Centore, Pollack, Barrieau, Gopalan, & Lim (2019) 2016 USA Recruited 249 women from postnatal unit; admitted to birth center between June 6 – July 1, 2016, viable fetus, contactable over email; cases had emergency team response called during labor and delivery, controls delivered within same period but did not have emergency team response IES-R PCL 6 weeks IES: positive screen at cutoff ≥24 PCL: scores range 17–85, positive screen at cutoffs 30–35 The median score for the IES was 16 among participants requiring an emergency team response (ETR) and 2 among participants not requiring an ETR. The median score for the PCL was 22.5 among participants requiring an ETR and 20 among participants not requiring an ETR. Söderquist, Wijma, & Wijma (2006) 1997–1998 Sweden Recruited 1224 women from OBGYN clinic during pregnancy after first ultrasound; no plans for legal abortion, no obstetric complications, spoke Swedish TES 1 month, 4 months, 7 months, 11 months Reported on dichotomous scale (met criteria for PTS - including intrusion (B), avoidance (C), and arousal (D) items - or not). Symptoms for each item ‘present’ if rated 3 or 4. Cut-offs for meeting criteria not reported. 1.3% of participants met criteria B, C, and D 1 month postpartum. At 4 months postpartum, 1.7% met criteria; at 7 months, 1.7%; and at 11 months, 0.9%. Soet, Brack, & Dilorio (2003) 2000–2001 USA Recruited 103 women from childbirth classes in Atlanta, GA; inclusion criteria not specified TES 4 weeks (approx.) Continuous measure to assess total symptoms; PTSD criteria met if positive for 1 intrusion item, 3 avoidance items, and 2 arousal items; classified as ‘symptomatic’, ‘partially symptomatic’, or ‘fully symptomatic’, 34% of participants reported their childbirth experience as traumatic. 1.9% met PTSD diagnostic criteria. 30.1% reported some PTS symptoms but did not meet clinical diagnostic criteria. Sorenson & Tschetter (2010) n/a USA Sample of 71 women drawn from archived birth announcements published in daily newspaper over 59-day period; spoke English Other: PTCS 6–7 months Possible scores range 15–75. Classified as ‘low trauma’ if scores between 15–30, ‘low-middle’ between 31–45, ‘high-middle’ between 46–60, and ‘high trauma’ between 61–75 44.9% of participants reported PTS symptoms in the ‘low trauma’ range, 44.9% ‘low-middle’, 7.3% ‘high-middle’, and 2.9% ‘high trauma’. Thompson, Roberts, & Ellwood (2011) 2006–2007 Australia Recruited 206 women from hospital postpartum units; recently delivered with estimated blood loss of 1500 mL or more in the first 24 hours, and/or a peripartum fall in hemoglobin (Hb) concentration to 7 gdL or less, and/or a peripartum fall in Hb concentration of 4 gdL or more PCL 2 months, 4 months Categorised as ‘cases’ if score >44 5% of participants showed symptoms of PTSD at two months and 3% at four months Verreault et al. (2012) 2005–2009 Canada Recruited 308 women during pregnancy from OBGYN office or ultrasound clinic; >18 years old, spoke French or English Modified PSS-SR Other: SCID-I 1 month, 3 months, 6 months MPSS-SR: symptom ‘present’ if scored 1 or greater SCID-I: diagnostic criteria; ‘full PTSD’ if clusters A-F endorsed ‘Partial PTSD’ if 1 or more re-experiencing symptoms and 1 or more avoidance or arousal symptoms MPSS-SR: 7.6%, 6.1%, and 4.9% of participants showed ‘full PTSD’ and 16.6%, 5.3%, 4.3% ‘partial PTSD’ at 1, 3, and 6 months, respectively. SCID-I: 1.1%, 0.8%, and 0.0% of participants showed ‘full PTSD’ and 3.2%, 1.2%, 2.9% ‘partial PTSD’ at 1, 3, and 6 months, respectively. White, Matthey, Boyd, & Barnett (2006) n/a Australia Recruited 400 participants from postpartum unit of hospital 2–3 days per week over a 10-month period; spoke English DSM criteria PSS-SR 6 weeks, 6 months, 12 months DSM criteria: questions to assess criteria A, E, and F PSS-SR: ‘PTSD profile’ if met all criteria on DSM questions and total score ≥18 on PSS 2% of participants met PTSD diagnostic criteria at 6 weeks postpartum. 10.5% reported some PTS symptoms but did not meet clinical diagnostic criteria. 2.6% of participants met PTSD diagnostic criteria at 6 months, and 2.4% at 12 months. Williams, Taylor, & Schwannauer (2016) 2012 UK Recruited 502 women from websites associated with birth organizations; >18 years old, gave birth in past 12 months IES-R Within 1 year Clinically significant if ≥33 18.9% of participants had clinically significant levels of PTS symptoms. Figure 1 Flow diagram of systematic literature review Authors’ note: This flow diagram was adapted from the PRISMA 2020 flow diagram for new systematic reviews (Page et al., 2021). Table 1: Overview of all studies (15 total) using the IES* # Authors (Year) Country n† Instruments used Timing of measurements (after birth) Results specific to measurement of post-traumatic stress (PTS) or post-traumatic stress disorder (PTSD) 2 Anderson & McGuinness (2008) USA 28 IES 9 months IES scores ranged from 1 to 34 (mean = 13.96, SD = 10.75). Six participants had mild PTS; 2 had moderate to severe PTS. 3 Anderson & Perez (2015) USA 44 IES Other: self-appraisal Within 72 hours IES scores ranged from 0 to 52. 33.3% of participants reported subclinical PTS symptoms. 50% of participants had moderate to severe symptoms of acute traumatic stress. 22.7% appraised the birth experience as ‘awful’/ traumatic 4 Anderson & Strickland (2017) USA 66 IES Other: self-appraisal Within 72 hours 4.5% of participants of reported IES scores over 34 (high risk of PTSD). 48.6% of participants reported mild to moderate symptoms of subjective distress. 3% had severe symptoms of subjective distress. 5 Anderson (2010) USA 85 IES Other: self-appraisal Within 72 hours 53% of participants were mildly to severely impacted by the childbirth experience. 11.8% indicated mild trauma. 41.2% indicated moderate to severe trauma. On a scale of 0 (no trauma) to 10 (traumatic), 35% appraised their childbirth experience at 7 or greater. 8 Ayers, Wright, & Wells (2007) UK 64 couples (men and women) IES 9 weeks Three men and three women (5% of sample) had severe PTS symptoms. 9 Baxter (2019) UK 170 IES 4–5 months 37% of participants had high PTS symptoms. 12 Czarnocka & Slade (2000) UK 264 IES PTSD-Q 6 weeks 9.9% of participants reported clinically significant levels of intrusions or avoidance. 3% of participants met PTSD diagnostic criteria. 27.2% reported some clinically significant PTS symptoms but did not meet clinical diagnostic criteria. 13 Dale-Hewitt, Slade, Wright, Cree, & Tully (2012) UK 50 IES PTSD-Q 6 weeks 8% of participants met PTSD diagnostic criteria. The total mean score for the IES was 17.28 (SD: 16.56). The total mean score for the PTSD-Q was 38.50 (SD: 15.71). 14 Davies, Slade, Wright, & Stewart (2008) UK 211 IES PTSD-Q 6 weeks 3.8% of participants met PTSD diagnostic criteria. 21.3% of participants reported some clinically significant PTS symptoms but did not meet clinical diagnostic criteria. 18 Furuta, Sandall, Cooper, & Bick (2014) UK 1824 IES 6–8 weeks (most) 6.4% of participants reported clinically significant levels of intrusions and 8.4% avoidance. 19 Garthus-Niegel, Knoph, von Soest, Nielsen, & Eberhard-Gran (2014) Norway 1893 IES 8 weeks 1.9% of women were reported to “likely” have PTSD. The group mean for the IES was 6.73 (SD: 8.10). 21 Iles, Slade, & Spiby (2011) UK 372 IES PTSD-Q Within 7 days, 6 weeks, 3 months The group mean for the IES was 13.33 (SD: 11.64) at T1, 10.55 (SD: 10.78) at T2, and 6.81 (SD: 8.61) at T3. The group mean for the PTSD-Q was 29.46 (SD: 11.96) at T2 and 26.95 (SD: 9.73) at T3. 22 Keogh, Ayers, & Francis (2002) UK 40 IES PSS-SR 2 weeks The group mean for the IES was 10.13 (SD: 12.23). The mean item score for the intrusion subscale was 5.30 (SD: 7.19) and avoidance 4.83 (SD: 6.36). The group mean for the PSS-SR was 9.88 (SD: 8.96). The mean item score for the intrusion subscale was 2.23 (SD: 3.05), numbing 3.75 (SD: 3.84), and arousal 3.90 (SD: 3.19). 24 Lyons (1998) UK 42 IES 4 days, 1 month Three mothers (7.1%) had IES scores in the medium distress range. One mother (2.4%) had an IES score of 56, in the high distress range. 26 McDonald, Slade, Spiby, & Iles (2011) UK 81 IES PTSD-Q 2 years 17.3% of participants had clinically significant levels of some PTS symptoms. * See Appendix B for full list of all studies included in this review † Sample size (n) refers to the number of birthing mothers included in the study, unless otherwise noted Table 2: Overview of all studies (6 total) using the IES-R* # Authors (Year) Country n† Instruments used Timing of measurements (after birth) Results specific to measurement of post-traumatic stress (PTS) or post-traumatic stress disorder (PTSD) 16 Edworthy, Chasey, & Williams (2008) UK 121 IES-R 6 weeks The group mean for the IES-R was 7.51 (SD: 8.91). The mean item score for the intrusion subscale was 3.52 (SD 4.11), avoidance 2.44 (SD 3.79) and hyperarousal 1.45 (SD 2.36). 1% of participants met PTSD diagnostic criteria. 8.3% showed medium distress across all subscales. 9.1% of participants met clinically significant PTS symptoms on the hyperarousal subscale and 1.7 on the intrusion subscale. 20 Holt, Sellwood, & Slade (2018) UK 1393 IES-R Other: self-appraisal 2–6 months The group mean for the IES-R was 14.98 (SD: 15.1). The group mean for the trauma appraisal was 0.7 (SD: 0.85). 27 Mitchell, Whittingham, Steindl, & Kirby (2018) Australia 262 IES-R Within 2 years 8.0% of participants had clinically significant levels of PTS symptoms pre-intervention. 61.9% of those participants were in the non-clinical range by 1 month post-intervention 29 Priest, Henderson, Evans, & Hagan (2003) Australia 1745 IES-R 2 months, 6 months, 1 year 0.8% of participants in the control group and 0.6% in the intervention group were diagnosed with a stress disorder in the year after giving birth 30 Silverstein, Centore, Pollack, Barrieau, Gopalan, & Lim (2019) USA 249 IES-R PCL 6 weeks The median score for the IES-R was 16 among participants requiring an emergency team response (ETR) and 2 among participants not requiring an ETR. The median score for the PCL was 22.5 among participants requiring an ETR and 20 among participants not requiring an ETR. 37 Williams, Taylor, & Schwannauer (2016) UK 502 IES-R Within 1 year 18.9% of participants had clinically significant levels of PTS symptoms. * See Appendix B for full list of all studies included in this review † Sample size (n) refers to the number of birthing mothers included in the study, unless otherwise noted Table 3: Overview of all studies (5 total) using the PTSD-Q* # Authors (Year) Country n† Instruments used Timing of measurements (after birth) Results specific to measurement of post-traumatic stress (PTS) or post-traumatic stress disorder (PTSD) 12 Czarnocka & Slade (2000) UK 264 IES PTSD-Q 6 weeks 9.9% of participants reported clinically significant levels of intrusions or avoidance. 3% of participants met PTSD diagnostic criteria. 27.2% reported some clinically significant PTS symptoms but did not meet clinical diagnostic criteria. 13 Dale-Hewitt, Slade, Wright, Cree, & Tully (2012) UK 50 IES PTSD-Q 6 weeks 8% of participants met PTSD diagnostic criteria. The total mean score for the IES was 17.28 (SD: 16.56). The total mean score for the PTSD-Q was 38.50 (SD: 15.71). 14 Davies, Slade, Wright, & Stewart (2008) UK 211 IES PTSD-Q 6 weeks 3.8% of participants met PTSD diagnostic criteria. 21.3% of participants reported some clinically significant PTS symptoms but did not meet clinical diagnostic criteria. 21 Iles, Slade, & Spiby (2011) UK 372 IES PTSD-Q Within 7 days, 6 weeks, 3 months The group mean for the IES was 13.33 (SD: 11.64) at T1, 10.55 (SD: 10.78) at T2, and 6.81 (SD: 8.61) at T3. The group mean for the PTSD-Q was 29.46 (SD: 11.96) at T2 and 26.95 (SD: 9.73) at T3. 26 McDonald, Slade, Spiby, & Iles (2011) UK 81 IES PTSD-Q 2 years 17.3% of participants had clinically significant levels of some PTS symptoms. * See Appendix B for full list of all studies included in this review † Sample size (n) refers to the number of birthing mothers included in the study, unless otherwise noted Disclosure of interest The authors report no conflict of interest. 1 Authors’ note: The language used in this review will reflect the language and participants of the studies presented; the word ‘women’ used throughout the paper will refer to the birthing parent whose psychological state is being measured. The authors recognise and celebrate that not all birthing people identify as women or mothers. References Alcorn KL , O’Donovan A , Patrick JC , Creedy D , & Devilly GJ (2010). A prospective longitudinal study of the prevalence of post-traumatic stress disorder resulting from childbirth events. Psychol Med, 40 (11 ), 1849–1859. 10.1017/S0033291709992224 20059799 Andersen LB , Melvaer LB , Videbech P , Lamont RF , & Joergensen JS (2012). Risk factors for developing post-traumatic stress disorder following childbirth: a systematic review. Acta Obstetricia et Gynecologica Scandinavica, 91 (11 ), 1261–1272. 10.1111/j.1600-0412.2012.01476.x 22670573 Anderson C (2010). Impact of traumatic birth experience on Latina adolescent mothers. Issues in Mental Health Nursing, 31 (11 ), 700–707. 10.3109/01612840.2010.518784 20936891 Anderson C & McGuinness TM (2008). Do teenage mothers experience childbirth as traumatic? Journal of Psychosocial Nursing, 46 (4 ), 21–24. 10.3928/02793695-20080401-01 Anderson C & Perez C (2015). Adolescent psychological birth trauma following cesarean birth. Pediatric Nursing, 41 (2 ), 78–83.26292455 Anderson C & Strickland S (2017). The influence of acculturation on traumatic stress following childbirth among Hispanic adolescents. Hispanic Health Care International, 15 (3 ), 99–106. 10.1177/1540415317717410 29164933 Ayers S (2004). Delivery as a traumatic event: Prevalence, risk factors, and treatment for postnatal posttraumatic stress disorder. Clinical Obstetrics and Gynecology, 47 , 552–567. 10.1097/01.grf.0000129919.00756.9c 15326418 Ayers S , Harris R , Sawyer A , Parfitt Y , & Ford E (2009). Posttraumatic stress disorder after childbirth: analysis of symptom presentation and sampling. J Affect Disord, 119 (1–3 ), 200–204. 10.1016/j.jad.2009.02.029 19368975 Ayers S & Pickering AD (2001). Do women get posttraumatic stress disorder as a result of childbirth? A prospective study of incidence. Birth, 28 (2 ), 111–118. 10.1046/j.1523-536x.2001.00111.x 11380382 Ayers S , Wright DB , & Wells N (2007). Symptoms of post-traumatic stress disorder in couples after birth: association with the couple’s relationship and parent-baby bond. Journal of Reproductive and Infant Psychology, 25 (1 ), 40–50. 10.1080/02646830601117175 Baxter J (2019). Postnatal debriefing: women’s need to talk after birth. British Journal of Midwifery, 27 (9 ), 563–571. 10.12968/bjom.2019.27.9.563 Beck CT (2004). Birth trauma: in the eye of the beholder. Nursing Research, 53 (1 ), 28–35. 10.1097/00006199-200401000-00005 14726774 Beck CT (2013). Traumatic childbirth. Routledge; 1st Edition. Beck CT , Gable RK , Sakala C , & Declercq ER (2011). Posttraumatic stress disorder in new mothers: Results from a two-stage U.S. national survey. Birth: Issues in Prenatal Care. 38 (3 ), 216–227. 10.1111/j.1523-536X.2011.00475.x Beck JG , Grant DM , Read JP , Clapp JD , Coffey SF , Miller LM , & Palyo SA (2009). The Impact of Event Scale – Revised: Psychometric properties in a sample of motor vehicle accident survivors. J Anxiety Disord, 22 (2 ), 187–198. 10.1016/j.janxdis.2007.02.007 Brunet A , Weiss DS , Metzler TJ , Best SR , Neylan TC , Rogers C , … Marmar CR (2001). The Peritraumatic Distress Inventory: a proposed measure of PTSD criterion A2. Am J Psychiatry, 158 (9 ), 1480–1485. 10.1176/appi.ajp.158.9.1480 11532735 Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia. Available at www.covidence.org Creamer M , Bell R , & Failla S (2003). Psychometric properties of the Impact of Event Scale – Revised. Behavior Research and Therapy, 41 (12 ), 1489–1496. 10.1016/j.brat.2003.07.010 Creedy DK , Shochet IM , & Horsfall J (2000). Childbirth and the development of acute trauma symptoms: Incidence and contributing factors. Birth, 27 (2 ), 104–111. 10.1046/j.1523-536x.2000.00104.x 11251488 Cross MR & McCanne TR (2001). Validation of a self-report measure of posttraumatic stress disorder in a sample of college-age women. Journal of Traumatic Stress, 14 (1 ), 135–147. 10.1023/A:1007843800664 Czarnocka J & Slade P (2000). Prevalence and predictors of post-traumatic stress symptoms following childbirth. Br J Clin Psychol, 39 (1 ), 35–51. 10.1348/014466500163095 10789027 Dale-Hewitt V , Slade P , Wright I , Cree M , & Tully C (2012). Patterns of attention and experiences of post-traumatic stress symptoms following childbirth: an experimental study. Arch Womens Ment Health, 15 (4 ), 289–96. 10.1007/s00737-012-0290-2 22688919 Davies J , Slade P , Wright I , & Stewart P (2008). Posttraumatic stress symptoms following childbirth and mothers’ perceptions of their infants. Infant Mental Health Journal, 29 (6 ), 537–554. 10.1002/imhj.20197 28636249 Dekel S , Ein-Dor T , Dishy GA , & Mayopoulos PA (2019). Beyond postpartum depression: posttraumatic stress-depressive response following childbirth. Archives of Women’s Mental Health. 10.1007/s00737-019-01006-x Dikmen Yildiz P , Ayers S , & Phillips L (2016). The prevalence of posttraumatic stress disorder in pregnancy and after birth: A systematic review and meta-analysis. Journal of Affective Disorders, 208 , 634–645. 10.1016/j.jad.2016.10.009 27865585 Dikmen-Yildiz P , Ayers S , & Phillips L (2018). Longitudinal trajectories of post-traumatic stress disorder (PTSD) after birth and associated risk factors. Journal of Affective Disorders, 229 , 377–385. 10.1016/j.jad.2017.12.074 29331697 Edworthy Z , Chasey R , & Williams H (2008). The role of schema and appraisals in the development of post-traumatic stress symptoms following birth. Journal of Reproductive and Infant Psychology, 26 (2 ), 123–128. 10.1080/02646830801918422 Foa EB , Cashman L , Jaycox L , & Perry K (1997). The validation of a self-report measure of posttraumatic stress disorder: The Posttraumatic Diagnostic Scale. Psychological Assessment, 9 (4 ), 445–451. 10.1037/1040-3590.9.4.445 Foa EB , McLean CP , Zang Y , Zhong J , Powers MB , Kauffman BY , … Knowles K (2016a). Psychometric properties of the Posttraumatic Diagnostic Scale for DSM-5 (PDS-5). Psychological Assessment, 28 (10 ), 1166–1171. 10.1037/pas0000258 26691504 Foa EB , McLean CP Zang Y , Zhong J , Rauch S , Porter K , … Kauffman BY (2016b). Psychometric properties of the Posttraumatic Stress Disorder Symptom Scale Interview for DSM-5 (PSSI-5). Psychological Assessment, 28 (1 ), 1159–1165. 10.1037/pas0000259 26691507 Foa EB , Riggs DS , Dancu CV , & Rothbaum BO (1993). Reliability and validity of a brief instrument for assessing post-traumatic stress disorder. Journal of Traumatic Stress, 6 (4 ), 459–473. 10.1002/jts.2490060405 Ford E & Ayers S (2011). Support during birth interacts with prior trauma and birth intervention to predict postnatal post-traumatic stress symptoms. Psychology & Health, 26 (12 ), 1553–1570. 10.1080/08870446.2010.533770 21598181 Furuta M , Sandall J , Cooper D , & Bick D (2014). The relationship between severe maternal morbidity and psychological health symptoms at 6–8 weeks postpartum: a prospective cohort study in one English maternity unit. BMC Pregnancy and Childbirth, 14 (133 ). http://www.biomedcentral.com/1471-2393/14/133 Garthus-Niegel S , Knoph C , von Soest T , Nielsen CS , & Eberhard-Gran M (2014). The role of labor pain and overall birth experience in the development of posttraumatic stress symptoms: A longitudinal cohort study. Birth, 41 (1 ), 108–115. 10.1111/birt.12093 24654643 Gold KJ & Johnson TRB (2014). Mothers at risk: Maternal mental health outcomes after perinatal death. Obstetrics & Gynecology, 123 ,6S. 10.1097/AOG.0000000000000204 Grekin R , & O’Hara MW (2014). Prevalence and risk factors of postpartum posttraumatic stress disorder: A meta-analysis. Clinical Psychology Review, 34 (5 ), 389–401. 10.1016/j.cpr.2014.05.003 24952134 Harris R & Ayers S (2012). What makes labour and birth traumatic? A survey of intrapartum ‘hotspots’. Psychology & Health, 27 (10 ), 1166–1177. 10.1080/08870446.2011.649755 22292475 Holt L , Sellwood W , & Slade P (2018). Birth experiences, trauma responses and self-concept in postpartum psychotic-like experiences. Schizophrenia Research, 197 , 531–538. 10.1016/j.schres.2017.12.015 29402582 Horowitz M (1982). Stress response syndromes and their treatment. In: Goldberger L , Breznitz S (eds.) Handbook of Stress: Theoretical and Clinical Aspects. New York: The free Press; 711–732. Horowitz M , Wilner N , & Alvarez W (1979). Impact of Event Scale: a measure of subjective stress. Psychosom Med, 41 (3 ), 209–218. 10.1097/00006842-197905000-00004 472086 Hynan MT (1998). The Perinatal Posttraumatic Stress Disorder (PTSD) Questionnaire (PPQ). In Wood R & Zalaquette CP (Eds.). Evaluating stress: A book of resources (Vol. 2 ; pp. 199–220) Scarecrow Press: Lanham, MD. Iles J , Slade P , & Spiby H (2011). Posttraumatic stress symptoms and postpartum depression in couples after childbirth: the role of partner support and attachment. J Anxiety Discord, 25 (4 ), 520–530. 10.1016/j.janxdis.2010.12.006 Keogh E , Ayers S , & Francis H (2002). Does anxiety sensitivity predict post-traumatic stress symptoms following childbirth? A preliminary report. Cognitive Behaviour Therapy, 31 (4 ), 145–155. 10.1080/165060702321138546 Leeds L & Hargreaves I (2008). The psychological consequences of childbirth. Journal of Reproductive and Infant Psychology, 26 (2 ), 108–122. 10.1080/02646830701688299 Lyons S (1998). A prospective study of post traumatic stress symptoms 1 month following childbirth in a group of 42 first-time mothers. Journal of Reproductive and Infant Psychology, 16 (2–3 ), 91–105. 10.1080/02646839808404562 MacKinnon AL , Yang L , Feeley N , Gold I , Hayton B , & Zelkowitz P (2017). Birth setting, labour experience, and postpartum psychological distress. Midwifery, 50 , 110–116. 10.1016/j.midw.2017.03.023 28412526 Marmar CR , Weiss DS , & Metzler TJ (1997). The Peritraumatic Dissociative Experiences Questionnaire. In Wilson JP & Keane TM (Eds.), Assessing psychological trauma and PTSD (p. 412–428). Guilford Press. Matthey S , Henshaw C , Elliott S , Barnett B . Variability in use of cut-off scores and formats on the Edinburgh Postnatal Depression Scale: Implications for clinical and research practice. Arch Womens Ment Health. 2006;9 (6 ):309–315. doi:10.1007/s00737-006-0152-x 17013761 McDonald S , Slade P , Spiby H , & Iles J (2011). Post-traumatic stress symptoms, parenting stress and mother-child relationships following childbirth and at 2 years postpartum. J Psychosom Obstet Gynaecol, 32 (2 ), 141–146. 10.3109/0167482X.2011.596962 21824044 Mitchell AE , Whittingham K , Steindl S , & Kirby J (2018). Feasibility and acceptability of a brief online self-compassion intervention for mothers of infants. Archives of Women’s Mental Health, 21 , 533–561. 10.1007/s00737-018-0829-y Modarres M , Afrasiabi S , Rahnama P , & Montazeri A (2012). Prevalence and risk factors of childbirth-related post-traumatic stress symptoms. BMC Pregnancy and Childbirth, 12 (88 ). 10.1186/1471-2393-12-88 Murphy H & Strong J (2018). Just another ordinary bad birth? A narrative analysis of first time mothers’ traumatic birth experiences. Health Care for Women International, 39 (6 ), 619–643. 10.1080/07399332.2018.1442838 29474791 Neal LA , Busuttil W , Rollins J , Herepath R , Strike P , & Turnbull G (1994). Convergent validity of measures of post-traumatic stress disorder in a mixed military and civilian population. Journal of Traumatic Stress, 7 (3 ), 447–455. 10.1007/bf02102789 8087405 North C , Suris A , Pringle Smith R , & King RV (2016). The evolution of PTSD criteria across editions of DSM. Annals of Clinical Psychiatry, 28 (3 ), 197–208.27490836 Olde E , Kleber RJ , van der Hart O , & Pop VJM (2006). Childbirth and posttraumatic stress responses: A validation study of the Dutch Impact of Event Scale-Revised. European Journal of Psychological Assessment, 22 (4 ), 259–267. 10.1027/1015-5759.22.4.259 Page MJ , McKenzie JE , Bossuyt PM , Boutron I , Hoffmann TC , Mulrow CD , The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372 :n71. doi: 10.1136/bmj.n71 33782057 Parfitt Y & Ayers S (2009). The effect of postnatal symptoms of post-traumatic stress and depression on the couple’s relationship and parent-baby bond. Journal of Reproductive and Infant Psychology, 27 (2 ), 127–142. 10.1080/02646830802350831 Priest SR , Henderson J , Evans SF , & Hagan R (2003). Stress debriefing after childbirth: a randomised controlled trial. MJA, 178 , 542–545.12765500 Reed R , Sharman R , & Inglis C (2017). Women’s descriptions of childbirth trauma relating to care provider actions and interactions. BMC Pregnancy and Childbirth, 17 (1 ), 21. 10.1186/s12884-016-1197-0 28068932 Robins LN & Helzer JE (1985). Diagnostic Interview Schedule (DIS Version III-A). St. Louis, MO: Washington University, Department of Psychiatry. Shaw RJ , Bernard RS , Storfer-Isser A , Whine W , & Horwitz SM (2012). Parental coping in the neonatal intensive care unit. Journal of Clinical Psychology in Medical Settings, 20 , 135–142. 10.1007/s10880-012-9328-x Silverstein GR , Centore M , Pollack A , Barrieau G , Gopalan P , & Lim G (2019). Postpartum psychological distress after emergency team response during childbirth. Journal of Psychosomatic Obstetrics & Gynecology, 40 (4 ), 304–310. 10.1080/0167482X.2018.1512095 30204531 Simpson M & Catling C (2016). Understanding psychological traumatic birth experiences: A literature review. Women and Birth, 29 (3 ), 203–207. 10.1016/j.wombi.2015.10.009 26563636 Soet JE , Brack GA , & Dilorio C (2003). Prevalence and predictors of women’s experience of psychological trauma during childbirth. Birth, 30 (1 ), 36–46. 10.1046/j.1523-536x.2003.00215.x 12581038 Sorenson DS (2000). Post-traumatic stress disorder inventory. Brookings, SD: South Dakota State University. Sorenson DS (2003). Healing traumatizing provider interactions among women through short-term group therapy. Archives in Psychiatric Nursing, 17 (6 ), 259–269. 10.1053/j.apnu.2003.10.002 Sorenson DS , & Tschetter L (2010). Prevalence of negative birth perception, disaffirmation, perinatal trauma symptoms, and depression among postpartum women. Perspect Psychiatr Care, 46 (1 ), 14–25. 10.1111/j.1744-6163.2009.00234.x.20051075 Spitzer RL , Williams JBW , Gibbon M , & First MB (1990). Structured Clinical Interview for DSM-W-R-Patient ed., (With Psychotic Screen; SCID-P). Washington, DC: American Psychiatric Press. Söderquist J , Wijma B , & Wijma K (2006). The longitudinal course of post-traumatic stress after childbirth. Journal of Psychosomatic Obstetrics & Gynecology, 27 (2 ), 113–119. 10.1080/01674820600712172 16808086 Thompson JF , Roberts CL , & Ellwood DA (2011). Emotional and physical health outcomes after significant primary post-partum haemorrhage (PPH): A multicentre cohort study. Australian and New Zealand Journal of Obstetrics and Gynaecology, 51 , 365–371. 10.1111/j.1479-828X.2011.01317.x 21806578 Verreault N , Da Costa D , Marchand A , Ireland K , Banack H , Dritsa M , and Khalifé S (2012). PTSD following childbirth: A prospective study of incidence and risk factors in Canadian women. Journal of Psychosomatic Research, 73 (4 ), 257–263. 10.1016/j.jpsychores.2012.07.010 22980529 Watson CG , Juba MP , Manifold V , Kucala T , & Anderson PE (1991). The PTSD interview: rationale, description, reliability, and concurrent validity of a DSM-III-based technique. J Clin Psychol, 47 (2 ), 179–188.2030122 Weathers FW , Litz BT , Herman DS , Huska JA , & Keane TM (1993). The PTSD checklist: reliability, validity, & diagnostic utility. Paper presented at the Annual Convention of the International Society for Traumatic Stress Studies, San Antonio, TX. Weiss DS & Marmar CR (1997). The Impact of Event Scale – Revised. In Wilson JP & Keane TM (Eds.), Assessing psychological trauma and PTSD, 399–411. Guilford Press. White T , Matthey S , Boyd K , & Barnett B (2006). Postnatal depression and post-traumatic stress after childbirth: Prevalence, course and co-occurrence. Journal of Reproductive and Infant Psychology, 24 (2 ), 107–120. 10.1080/02646830600643874 Wijma K , Söderquist J , & Wijma B (1997). Posttraumatic stress disorder after childbirth: a cross sectional study. J Anxiety Disord, 11 (6 ), 587–597. 10.1016/s0887-6185(97)00041-8 9455721 Williams C , Taylor EP , & Schwanneur M (2016). A web-based survey of mother-infant bond, attachment experiences, and metacognition in posttraumatic stress following childbirth. Infant Ment Health J, 37 (3 ), 259–273. 10.1002/imhj.21564 27090507

PMC009xxxxxx/PMC9335864.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7503056 4435 J Am Chem Soc J Am Chem Soc Journal of the American Chemical Society 0002-7863 1520-5126 35819848 9335864 10.1021/jacs.2c03665 NIHMS1823201 Article Adaptor-specific antibody fragment inhibitors for the intracellular modulation of p97 (VCP) protein-protein interactions Jiang Ziwen †‡§ Kuo Yu-Hsuan †‡§ Zhong Mengqi †‡§ Zhang Jianchao ¶§ Zhou Xin X. †#∥ Xing Lijuan ¶ Wells James A. † Wang Yanzhuang ¶ Arkin Michelle R. *†‡ † Department of Pharmaceutical Chemistry, University of California, San Francisco, CA 94158, USA ‡ Small Molecule Discovery Center, University of California, San Francisco, CA 94158, USA ¶ Department of Molecular, Cellular and Developmental Biology, University of Michigan, Ann Arbor, MI 48109-1085, USA # Department of Cancer Biology, Dana-Farber Cancer Institute, Boston, MA 02215 USA ∥ Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA 02115 USA. * Corresponding Author Michelle R. Arkin, michelle.arkin@ucsf.edu § Author Contributions These authors equally contributed to this work. All authors reviewed and edited the manuscript. 19 7 2022 27 7 2022 12 7 2022 27 7 2023 144 29 1321813225 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Protein-protein interactions (PPIs) form complex networks to drive cellular signaling and cellular functions. Precise modulation on target PPI helps explain the role of the PPI in cellular events and possesses therapeutic potential. For example, valosin-containing protein (VCP/p97) is a hub protein that interacts with more than 30 adaptor proteins involved in various cellular functions. However, the role of each p97 PPI during the relevant cellular event is underexplored. The development of small-molecule PPI modulators remains challenging due to a lack of grooves and pockets in the relatively large PPI interface, and the fact that a common binding groove in p97 binds to multiple adaptors. Here, we report an antibody fragment-based modulator for the PPI between p97 and its adaptor protein NSFL1C (p47). We engineered these antibody modulators by phage display against the p97-interacting domain of p47 and minimizing binding to other p97 adaptors. The selected antibody fragment modulators specifically disrupt the intracellular p97/p47 interaction. The potential of this antibody platform to develop PPI inhibitors in therapeutic applications was demonstrated through the inhibition of Golgi reassembly, which requires the p97/p47 interaction. This study presents a unique approach to modulate specific intracellular PPI using engineered antibody fragments, demonstrating a method to dissect the function of a PPI within a convoluted PPI network. Graphical Abstract Protein-protein interaction phage display antibody fragment inhibitor valosin-containing protein Golgi reassembly pmcIntroduction Protein-protein interactions (PPIs) are essential for intracellular signal transduction and transcriptional regulation.1-2 These cellular events are generally controlled by networks that are composed of several interconnected PPIs. Misregulation of these cooperative PPIs have been shown to cause diseases such as cancer and neurodegeneration.3-4 Aberrant PPIs include either the loss of a crucial interaction or the gain of a spatiotemporally incorrect interaction.5 The advancement of proteomics has facilitated the understanding of PPI networks, specifically in elucidating the interacting protein partners.6-7 Precisely mapped PPI networks provide fundamental information to explore the possibilities in controlling cellular functions through specific modulation of protein complexes. Therefore, we seek systematic methodologies for PPI inhibition or stabilization to understand the function of PPI and select those targets with relevant therapeutic avenues for disease treatment.8-9 An ideal PPI-specific modulator tool should focus on the interaction between the protein partners of interest, leaving other functions and PPIs of the targeted protein partners unaltered. Efficient genetic modulation methods such as knockdown10 or knockout assays11 do not offer such PPI-specific regulation, as the depletion of one protein will remove all of its PPIs simultaneously. Likewise, overexpression of the protein of interest can lead to multiple enhanced PPIs, cumulatively affecting the network. Therefore, the development of PPI-specific modulation can benefit from selective blockade of the PPI interface. Extracellular proteins are readily orthosterically inhibited by antibodies, whereas small-molecules and peptides are most often targeted to intracellular PPIs. However, the lack of pockets or grooves and the relatively large area of the PPI interfaces pose challenges for discovering small-molecule modulators, making it difficult to investigate multiple related PPIs rapidly and systematically.1, 12-14 Antibodies possess unique properties as potential intracellular PPI regulators when compared to small molecules and peptides. The protein nature of antibodies allows convenient cloning modifications to install subcellular localization signals,15-16 precisely refining the intracellular function of these antibodies to the targeted cellular milieu. The variable platforms of antibodies are available from nanobodies (~15 kDa),17 single-chain variable fragments (scFvs, ~27 kDa),18 antigen-binding fragments (Fabs, ~50 kDa)19 to full length IgG (~150 kDa), offering a tunable size range to tackle different PPI interfaces. Moreover, the constant chain within these antibody platforms enhances their stability against hydrolysis.20-21 Nevertheless, antibody-based modulators so far are mainly applied to secreted proteins or cell membrane targets.22-23 As an intracellular hub protein, valosin-containing protein (VCP/p97) interacts with more than 30 adaptor proteins to regulate multiple cellular functions, including the maintenance of protein homeostasis and facilitating protein degradation.24-26 Dissecting the particular function of an individual p97/adaptor protein interaction is therefore important but complicated within the p97 PPI network. In this work, we engineered antibody fragment inhibitors via phage display to modulate the interaction between p97 and its adaptor protein, NSFL1C (p47) (Figure 1). We chose the p97/p47 interaction as our model target for the development of PPI-specific antibody-based modulators because it is involved in membrane fusion processes,27 particularly during Golgi fragmentation and reassembly that are distinctive and readily measured.28 These engineered anti-p47 antibody fragments with nanomolar binding affinities successfully disrupted the intracellular interaction between p97 and p47. Expressing variations of the antibody fragments and nuclear localization signal sequences resulted in different phenotypic responses for Golgi fragmentation, further elucidating the role of p97/p47 interaction during Golgi dynamics. The study highlights a unique antibody-based approach for intracellular modulation of the p97/p47 interaction, providing a new tool to untangle the convoluted PPI networks during a cellular process. Results and Discussion Engineered antibody fragments for p47-UBX domain demonstrated nanomolar binding affinities and high selectivity We chose the UBX (ubiquitin regulatory X) domain of p47 as the antigen to discover inhibitors of p47/p97 by Fab-phage display. Previous studies have shown that the p97-N terminal domain directly interacted with the p47-UBX domain, and we hypothesized that some anti-p47-UBX antibodies would inhibit the p47/p97 interaction.29 Furthermore, antibodies against p97 would likely result in the inhibition of multiple PPIs because most adapters bind to a common site on p97-N domain; thirteen different p97 adaptors contain a UBX domain.30-31 Thus, inhibition from the p47 side should be more specific. To engineer antibodies for the p47-UBX domain, we carried out phage display using a previously developed Fab-phage library.32 The Fab scaffold derived from the 4D5 anti-HER2 antibody is highly stable (Tm 80 °C) and designed with amino acid variations in four of the six complementarity-determining regions (CDRs).32-33 After four rounds of selection against p47-UBX domain, more than five clones were identified through phage-ELISA (Figure 2a, S1). These selected Fab hits (Figure 2b) for p47-UBX were sequenced and cloned from phagemid into scFv platforms for monoclonal antibody production. Next, we assessed the binding affinity between the anti-p47-UBX scFv and p47-UBX domain using biolayer interferometry (BLI). After anchoring biotinylated p47-UBX on the BLI sensor tip, an increasing amount of each scFv was separately incubated with the p47-UBX-decorated sensor. From the dose response profiles, scFv-A06 and scFv-E04 demonstrated nanomolar binding affinities (~3 nM and ~7 nM, respectively) to the p47-UBX domain (Figure 2c,d, S2). We also evaluated the selectivity of these scFvs against the p97-binding domain of other adaptor proteins [e.g., NPL4 (nuclear protein localization homolog 4), FAF1 (FAS-Associated Factor 1), and p37 (UBX domain protein 2B)] using BLI. All five scFv clones were highly selective towards p47-UBX domain when compared to the NPL4-UBXL and FAF1-UBX domain (Figure 2e, S3). Moreover, as p47 and p37 have overlapping functions and are highly homologous (65% sequence identity; Figure S4),34 it was critical to ensure that the selected anti-p47-UBX scFv did not bind to p37-UBX domain. Upon evaluating the two nanomolar anti-p47-UBX scFv binders with BLI, scFv-E04 did bind to p37-UBX with a dissociation constant (KD) of ~4 nM, while scFv-A06 did not bind to p37-UBX (Figure S5). Overall, scFv-A06 is the most selective clone for p47-UBX domain with a KD at ~3 nM. Anti-p47-UBX antibody fragments disrupt the p97/p47 PPI in vitro Though selected based on binding to the p47-UBX domain, an essential requirement for these anti-p47-UBX antibody fragments is that they can disrupt the PPI of interest. We employed surface plasmon resonance (SPR) to test if these scFvs interfered with the p97/p47 interaction. As a flow-based approach, SPR offers a dynamic monitoring of the PPI status.35-36 We anchored biotinylated full-length p97 on a streptavidin-functionalized SPR sensor chip, followed by flowing a mixture of increasing concentrations of scFvs and a fixed concentration of p47-UBX. Considering the conformational changes of full-length p97 in the presence of either ATP or ADP,37 we conducted the SPR evaluation with both nucleotides. By SPR, we observed that four selected scFvs (A06, B01, E04, and G08) disrupted the p97/p47-UBX interaction regardless of the presence of nucleotide (Figure 2f~h, S6). In the presence of an equal molar amount of p47-UBX and scFv-A06 (both at 50 nM), the binding of p47-UBX to the p97-anchored sensor was reduced ~50% when compared with the control group where scFv-A06 was not present. Recent data indicated that there are three p97-binding modules on p47.38 Our construct for phage display included the C-terminal SHP motif and UBX domain but lacked the N-terminal SHP motif of p47 (Figure S7a). We therefore evaluated the full-length p47 and found that the same scFvs that inhibited the p97/p47-UBX interaction also blocked the p97/full-length-p47 interaction (Figure S7), demonstrating that our p47 construct for phage display was adequate for selecting inhibitors of the full-length p47. Translating these antibodies into functional PPI modulators requires their intracellular presence. Therefore, we cloned the encoding sequences of these antibody fragments into a mammalian expression vector. In the mammalian expression vector design, we included both the scFv and scFab (single chain Fab fragment) formats, as well as an N-terminal nuclear localization signal (NLS)-tagged39 scFab (scFab-NLS) format in order to vary their cellular localization. Each clone contained a C-terminal HA (human influenza hemagglutinin) epitope tag for detection purposes. After transfecting these plasmids with Xfect transfection reagents for 24 hours in human bone osteosarcoma epithelial (U2OS) cells (Figure S8), we visualized the antibody fragments and p47 by immunofluorescence (IF) (Figure 3a). P47 primarily localized in the nucleus, agreeing with previous reports.28 ScFv-A06 showed some cytoplasmic distribution, but primarily colocalized with p47 in the nucleus, suggesting the interaction between scFv-A06 and p47. Interestingly, due to the passive diffusion limit for the nuclear pore (~60 kDa),40-41 scFab (~60 kDa) primarily localized in the cytosol, whereas the scFab-NLS was mostly confined to the nucleus and overlapped with p47. Due to the low intensity of cytoplasmic p47, it was difficult to confidently state that scFab-A06 colocalized with p47 in the cytosol. Taken together, IF results indicated that the molecular weight of the antibody fragments and the nuclear localization tag determined the localization of the antibody fragment. Western blotting of the scFab and scFv antibody fragments indicated that E04 was expressed less than A06 antibody fragments in U2OS cells, possibly due to a lack of stability (Figure S9). We also confirmed the p47/antibody-fragment interaction through co-immunoprecipitation (co-IP) of p47. All A06 and E04-based antibody fragments (scFv, scFab, and scFab-NLS) bound to p47 (Figure 3b). Binding of scFab in the IP lysates demonstrated that this format was capable of binding even though it was not colocalized in intact cells. Interestingly, we observed that p47 was not completely detached from p97 upon the binding of antibody fragments. We next evaluated the co-IP of p97 and confirmed the formation of the trimeric complex containing p97, p47, and scFv-A06 (Figure S10). This result contrasted with our biophysical experiments, where these antibody fragments could completely compete full-length p97 (Figure 2g,h, Figure S6 and S7). The cellular result was understandable for a few reasons. First, endogenous levels of protein expression were not precisely controlled, in contrast to our biophysical assays using recombinant proteins; hence, the expression level of scFv-A06 when compared to p47 might have been insufficient to completely displace p47 from p97. Second, as previous studies have shown that p47 has an oligomeric form,42 one p47 “arm” might have been inhibited by antibody fragments while the other(s) remained bound. Third, a recent report demonstrated that the UBX domain was not the only module of p47 that binds to p97.38 Our antibody fragments might have blocked only the UBX domain and/or the C-terminal SHP motif of p47 that were present in the selection construct (Figure S7a) during phage display, while the N-terminal SHP motif of p47 retained its interaction with the adjacent p97 protomer. We next utilized the NanoBRET (bioluminescence resonance energy transfer) assay43 to quantitatively probe the effect of engineered antibody fragments on the p97/p47 interaction. The optimized NanoBRET assay for p97/p47 PPI contained an N-terminal Nanoluciferase-tagged p97-N domain (pNLF1N-p97-N) as the donor and a C-terminal Halo-tagged p47-UBX domain (pHTC-p47-UBX) as the acceptor (Figure 3c). To rule out the potential competition with the NanoBRET pair by endogenous p47, we generated p47-knockout (p47-KO) U2OS cells by CRISPR/Cas9 genome editing (Figure S11). During the NanoBRET evaluation, the donor, acceptor, and antibody-fragment (scFv and scFab)-expressing plasmids were co-transfected into the p47-KO U2OS cells to evaluate the proximity change between the NanoBRET pair (Figure S12). In parallel, we employed the scFv and scFab platforms of anti-RNase (ribonuclease A) antibody44 as negative controls and p47-expressing vector (p47-FLAG) as the positive control. When compared to the BRET ratio in the scFv-RNase and scFab-RNase control groups, the co-transfection of either scFv-A06 or p47-FLAG similarly reduced the BRET ratio by ~15%, indicating a reduced interaction between the p97/p47 NanoBRET pair (Figure 3d, S12c). In comparison, the lack of effect on the BRET signal from scFv-E04 may be attributed to its low expression level (Figure 3b). The lack of effect on the BRET signal by the scFab-A06 may be attributed to the localization of the antibody fragment in the cytosol (Figure 3a). Based on the colocalization, co-IP, and NanoBRET results, scFv-A06 was the most potent inhibitor of the p97/p47 PPI among the anti-p47-UBX antibody fragments tested. Antibody fragments inhibit reassembly of the Golgi apparatus Antibody fragment-based PPI inhibitors are expected to affect PPI-related phenotypes and modulate cellular functions. The p97/p47 complex facilitates the fusion of Golgi membrane fragments to reassemble the Golgi apparatus during the cell cycle at late mitosis and early interphase.28 Therefore, we evaluated the effect of A06 and E04 anti-p47-UBX antibody fragments on Golgi reassembly processes using the morphological distribution of GRASP55, a Golgi reassembly stacking protein that reflects the Golgi structure.45 After transfecting the antibody fragment-expressing plasmids into HeLa cells, the two scFv and scFab clones significantly increased Golgi fragmentation, with the scFv groups showing more Golgi fragments per cell than the scFab groups (Figure 4a~c, S13). Since the Golgi area per cell remained constant (Figure 4d), an increased number of Golgi fragments indicated an attenuated reassembly process of Golgi membranes. Both p47 and the close homolog p37 have been associated with Golgi reassembly.34 It is noteworthy that A06 was selective for p47, whereas E04 bound similarly to both p47 and p37 (Figure S4). The dual inhibition mechanism of E04 may have accounted for the effective inhibition of Golgi reassembly, even given the lower expression level of E04 compared to A06. The tunability of the antibody fragment platform contributed to the understanding of p97/p47 PPI in Golgi reassembly. Previous mechanistic studies revealed that the p97/p47 complexes participated in membrane fusion during mitosis, whereas p47 primarily resided in the nucleus during interphase in HeLa cells.28 It was therefore curious that expression of both the scFvs and scFabs without an NLS yielded fragmented Golgi, while the scFabs with NLS tags showed minimal interference on the Golgi morphology. Cytoplasmic distribution of antibody fragments was therefore correlated with activity, suggesting the importance of p47 during phases of the cell cycle where the nuclear membrane was intact. To further validate the effect of these scFvs on Golgi reassembly, we carried out a cell-free Golgi reassembly assay29, 46-47 using purified rat liver Golgi membranes. Our engineered scFvs for human p47 also bound to rat p47 (Figure S14). In the ex vivo assay, rat liver Golgi membranes were treated with mitotic cytosolic lysates of HeLa cells to form mitotic Golgi fragments (MGFs), i.e., disassembling the Golgi cisternae into smaller vesicles (Figure 4e). Next, the MGFs were treated with interphase cytosolic lysates from HeLa cells that contained endogenously expressed p97/p47 complexes, inducing reassembly of the MGFs to large Golgi cisternae.46 When scFv-A06 or scFv-E04 was spiked into the interphase cytosolic lysates, the cisternal regrowth was slowed down to less than 50% (Figure 4f), confirming the inhibitory effect of these engineered antibody fragments as PPI modulators during Golgi reassembly (Figure 4g). Conclusions We have described the engineering of antibody fragment inhibitors for a specific PPI of the multifunctional protein p97 to modulate a cellular function of interest. These antibody fragments were selected by Fab-phage display for binding to the p97-interacting domain of the adaptor protein p47. Compared to genetic knockdown and knockout approaches, the well-defined steric blockade by these antibody inhibitors potentially reduces the interference to the overall p97-PPI network. Such antibody fragments are available in tunable molecular weights and subcellular localization tags through protein engineering, making it an attractive approach to study the PPI of interest in different cellular locations and providing toolkits to tackle biological questions. For example, this work demonstrates the first reported antibody fragment that modulates Golgi structure. As Golgi defects are observed in an increasing number of human diseases,48-49 this antibody-fragment inhibitor could potentially be useful in aiding the development of disease treatment. While we and others have worked to develop small-molecule inhibitors of PPI,50-51 it is still challenging to develop inhibitors for highly related targets systematically. By contrast, phage-selected antibody fragment can readily be developed for a panel of PPIs with a common hub protein. Furthermore, antibody discovery by phage display can also be applied to protein targets with post-translational modifications, protein isoforms, splice variants, or conformational variants, tremendously improving the applicability of our antibody inhibitor platform. Currently, the cell membrane and endosomal entrapment of proteins limit the ability to deliver antibodies intracellularly.52 Intracellular antibodies have been used as PPI modulators in a few cases, such as the signaling pathway of RAS proteins, and we now show their utility for selectively blocking one PPI of a hub protein that binds to many proteins at the same site.53-55 Development of efficient and robust methods for delivering recombinant antibodies56 to intracellular PPI targets will further extend the applicability of the technology for biomedical applications, ideally by perturbing the formation of malfunctional protein complexes that are associated with diseases and disorders. Supplementary Material Supporting Information ACKNOWLEDGMENT M.R.A. acknowledges the support from the NIH/NIGMS (R01GM130145). Y.W. is supported by NIH/NIGMS (R35GM130331), NIH/NINDS (R01NS102279), and the Fast Forward Protein Folding Disease Initiative of the University of Michigan. J.A.W. acknowledges the support from the NIH/NIGMS (R35GM122451). Z.J. acknowledges the support from the NIH/NIGMS (F32GM139242). X.X.Z. acknowledges the support from a Merck Fellowship of the Damon Runyon Cancer Research Foundation (DRG-2297-17) and an NIH K99/R00 award (1K99EB030587). Figure 1. Workflow of the antibody modulator for specific VCP protein-protein interaction through phage display. MB, magnetic beads. Figure 2. Selection and characterization of anti-p47-UBX antibody fragments. (a) Plot of phage ELISA to select the binders for p47-UBX domain. The ratio of OD450 nm (p47-UBX)/OD450 nm (BSA) represents the selectivity of the binder, where a higher ratio represents a more specific binder. The ratio of OD450 nm (Competition)/OD450 nm (p47-UBX) represents the capability of the soluble p47-UBX to compete with the p47-UBX coated on plate for Fab-phage binding. A lower ratio value generally indicates a tighter binder. The pink quadrant represents the hit-set. (b) Sequence of the complementarity-determining regions for selected antibody binders to p47-UBX. (c) Biolayer interferometry (BLI) dose-response profiles of scFv-A06 to p47-UBX domain. (d) Binding affinities of selected scFvs for p47-UBX based on BLI results. Standard deviations represent N = 2 independent experiments. (e) BLI results of scFv-A06 binding to the interacting domains of different p97 adaptor proteins, showing selectivity for p47 over NPL4, FAF1, and p37. (f) Schematic illustration of the surface plasmon resonance (SPR) assessment of the competition between selected scFv and p47-UBX for p97 binding. (g, h) SPR sensorgrams for the scFv-A06/p47-UBX mixture binding to full-length human p97 in the presence of either 100 μM ATP (g) or 100 μM ADP (h) at the listed concentrations of scFv. Data are representative N = 2 independent experiments. Figure 3. Anti-p47-UBX antibody fragments interact with p47 and inhibit p97 binding in U2OS cells. (a) Representative images for the cellular localization of p47 and selected antibody fragments in U2OS cells. Plasmids that encode antibody fragments were transfected in U2OS cells for 24 hours. Scale bar, 20 μm. (b) Co-immunoprecipitation (co-IP) of p47 from U2OS cells after transfection of plasmids that encode antibody fragments. P47-containing protein complexes were captured from the lysates and blotted for the co-IP analysis. Data represent N = 2 independent experiments. (c) Schematic illustration of the NanoBRET assay for p97/p47 interaction in the presence of antibody fragment inhibitors (Ab-inhibitor). (d) NanoBRET assay in p47-knockout (p47-KO) U2OS cells. ScFv-A06 reduces the p97-N/p47-UBX interaction signal as well as the p47-overexpression group. Dashed line represents the normalized BRET ratio of the average between scFv-RNase and scFab-RNase groups. Error bars represent standard deviations of N = 4. Statistical analyses are performed using two-tailed Student’s t test. **, p < 0.01; ***, p < 0.001; n.s., no significance. Figure 4. Anti-p47-UBX antibody fragments disrupt Golgi structure by inhibiting its post-mitotic reassembly process. (a) Representative immunofluorescence images of HeLa cells transfected with HA-tagged anti-p47-UBX antibody fragments for 24 h and stained with an antibody to Golgi marker GRASP55. Scale bar, 10 μm. (b~d) (b-d) Quantification of GRASP55 for the percentage of cells with fragmented Golgi (b), number of Golgi items per cell (c), and the Golgi area per cell (d). Data are shown as mean ± SEM from N = 3 independent experiments. (e) Representative transmission electron microscopy images of rat liver Golgi (RLG), mitotic Golgi fragments (MGF, RLG treated with mitotic cytosol), and reassembled samples (MGF treated with interphase cytosol). In brief, rat liver Golgi membranes were fragmented by treatment with mitotic HeLa cytosol, and MGFs were reisolated and incubated with interphase cytosol alone or in the presence of recombinant anti-p47-UBX scFvs. Scale bar, 500 nm. (f) Quantification of the cisternal regrowth in (e). Results are shown as the mean percentage of membrane in cisternae ± SEM, where 0% represents cisternal regrowth in MGF (10.8 ± 1.7% of membrane in cisternae), and 100% represents cisternal regrowth of MGFs incubated with interphase cytosol alone (56.7 ± 1.1% of membrane in cisternae). Statistical analyses were performed using two-tailed Student’s t test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; n.s., no significance. (g) Scheme showing antibody-fragment inhibitors of p97/p47 protein-protein interaction inhibiting Golgi reassembly. Supporting Information Supplementary figures, methods, sequences, spectral, and characterization data. This material is available free of charge via the Internet at http://pubs.acs.org. The authors declare no competing interest. REFERENCES 1. Wells JA ; McClendon CL Reaching for high-hanging fruit in drug discovery at protein-protein interfaces. Nature 2007, 450 , 1001–1009.18075579 2. Keskin O ; Tuncbag N ; Gursoy A Predicting Protein-Protein Interactions from the Molecular to the Proteome Level. Chem. Rev 2016, 116 , 4884–4909.27074302 3. Ryan DP ; Matthews JM Protein-protein interactions in human disease. Curr. Opin. Struct. Biol 2005, 15 , 441–446.15993577 4. Zhong MQ ; Lee GM ; Sijbesma E ; Ottmann C ; Arkin MR Modulating protein-protein interaction networks in protein homeostasis. Curr. Opin. Chem. Biol 2019, 50 , 55–65.30913483 5. Zinzalla G ; Thurston DE Targeting protein-protein interactions for therapeutic intervention: a challenge for the future. Future Med. Chem 2009, 1 , 65–93.21426071 6. Berggard T ; Linse S ; James P Methods for the detection and analysis of protein-protein interactions. Proteomics 2007, 7 , 2833–2842.17640003 7. Tate S ; Larsen B ; Bonner R ; Gingras AC Label-free quantitative proteomics trends for protein-protein interactions. J. Proteom 2013, 81 , 91–101. 8. Jin LY ; Wang WR ; Fang GW Targeting Protein-Protein Interaction by Small Molecules. Annu. Rev. Pharmacol. Toxicol 2014, 54 , 435–456.24160698 9. Rognan D Rational design of protein-protein interaction inhibitors. MedChemComm 2015, 6 , 51–60. 10. Fire A ; Xu SQ ; Montgomery MK ; Kostas SA ; Driver SE ; Mello CC Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 1998, 391 , 806–811.9486653 11. Hsu PD ; Lander ES ; Zhang F Development and Applications of CRISPR-Cas9 for Genome Engineering. Cell 2014, 157 , 1262–1278.24906146 12. Smith MC ; Gestwicki JE Features of protein-protein interactions that translate into potent inhibitors: topology, surface area and affinity. Expert Rev. Mol. Med 2012, 14 , e16.22831787 13. Nevola L ; Giralt E Modulating protein-protein interactions: the potential of peptides. Chem. Commun 2015, 51 , 3302–3315. 14. Lu HY ; Zhou QD ; He J ; Jiang ZL ; Peng C ; Tong RS ; Shi JY Recent advances in the development of protein-protein interactions modulators: mechanisms and clinical trials. Signal Transduct. Target. Ther 2020, 5 , 213.32968059 15. Biocca S ; Cattaneo A Intracellular Immunization - Antibody Targeting to Subcellular Compartments. Trends Cell Biol. 1995, 5 , 248–252.14732130 16. Tanaka T ; Williams RL ; Rabbitts TH Tumour prevention by a single antibody domain targeting the interaction of signal transduction proteins with RAS. EMBO J. 2007, 26 , 3250–3259.17568777 17. Muyldermans S Nanobodies: Natural Single-Domain Antibodies. Annu. Rev. Biochem 2013, 82 , 775–797.23495938 18. Holliger P ; Hudson PJ Engineered antibody fragments and the rise of single domains. Nat. Biotechnol 2005, 23 , 1126–1136.16151406 19. Hust M ; Jostock T ; Menzel C ; Voedisch B ; Mohr A ; Brenneis M ; Kirsch MI ; Meier D ; Dubel S Single chain Fab (scFab) fragment. BMC Biotechnol. 2007, 7 , 14.17346344 20. Kramer K ; Fiedler M ; Skerra A ; Hock B A generic strategy for subcloning antibody variable regions from the scFv phage display vector pCANTAB 5 E into pASK85 permits the economical production of F-ab fragments and leads to improved recombinant immunoglobulin stability. Biosens. Bioelectron 2002, 17 , 305–313.11849927 21. Ma H ; O'Fagain C ; O'Kennedy R Antibody stability: A key to performance - Analysis, influences and improvement. Biochimie 2020, 177 , 213–225.32891698 22. Lobato MN ; Rabbitts TH Intracellular antibodies and challenges facing their use as therapeutic agents. Trends Mol. Med 2003, 9 , 390–396.13129705 23. Slastnikova TA ; Ulasov AV ; Rosenkranz AA ; Sobolev AS Targeted Intracellular Delivery of Antibodies: The State of the Art. Front. Pharmacol 2018, 9 , 1208.30405420 24. Meyer H ; Bug M ; Bremer S Emerging functions of the VCP/p97 AAA-ATPase in the ubiquitin system. Nat. Cell Biol 2012, 14 , 117–123.22298039 25. Buchberger A ; Schindelin H ; Hanzelmann P Control of p97 function by cofactor binding. FEBS Lett. 2015, 589 , 2578–2589.26320413 26. Stach L ; Freemont PS The AAA+ ATPase p97, a cellular multitool. Biochem. J 2017, 474 , 2953–2976.28819009 27. Kondo H ; Rabouille C ; Newman R ; Levine TP ; Pappin D ; Freemont P ; Warren G p47 is a cofactor for p97-mediated membrane fusion. Nature 1997, 388 , 75–78.9214505 28. Uchiyama K ; Jokitalo E ; Lindman M ; Jackman M ; Kano F ; Murata M ; Zhang XD ; Kondo H The localization and phosphorylation of p47 are important for Golgi disassembly-assembly during the cell cycle. J. Cell Biol 2003, 161 , 1067–1079.12810701 29. Meyer HH ; Wang YZ ; Warren G Direct binding of ubiquitin conjugates by the mammalian p97 adaptor complexes, p47 and Ufd1-Npl4. EMBO J. 2002, 21 , 5645–5652.12411482 30. Xue L ; Blythe EE ; Freiberger EC ; Mamrosh JL ; Hebert AS ; Reitsma JM ; Hess S ; Coon JJ ; Deshaies RJ Valosin-containing protein (VCP)-Adaptor Interactions are Exceptionally Dynamic and Subject to Differential Modulation by a VCP Inhibitor. Mol. Cell. Proteomics 2016, 15 , 2970–2986.27406709 31. Meyer H ; Weihl CC The VCP/p97 system at a glance: connecting cellular function to disease pathogenesis. J. Cell Sci 2014, 127 , 3877–3883.25146396 32. Hornsby M ; Paduch M ; Miersch S ; Saaf A ; Matsuguchi T ; Lee B ; Wypisniak K ; Doak A ; King D ; Usatyuk S ; Perry K ; Lu V ; Thomas W ; Luke J ; Goodman J ; Hoey RJ ; Lai D ; Griffin C ; Li ZJ ; Vizeacoumar FJ ; Dong D ; Campbell E ; Anderson S ; Zhong N ; Graslund S ; Koide S ; Moffat J ; Sidhu S ; Kossiakoff A ; Wells J A High Through-put Platform for Recombinant Antibodies to Folded Proteins. Mol. Cell. Proteomics 2015, 14 , 2833–2847.26290498 33. Persson H ; Ye W ; Wernimont A ; Adams JJ ; Koide A ; Koide S ; Lam R ; Sidhu SS CDR-H3 Diversity Is Not Required for Antigen Recognition by Synthetic Antibodies. J. Mol. Biol 2013, 425 , 803–811.23219464 34. Uchiyama K ; Totsukawa G ; Puhka M ; Kaneko Y ; Jokitalo E ; Dreveny I ; Beuron F ; Zhang XD ; Freemont P ; Kondo H P37 is a p97 adaptor required for Golgi and ER biogenesis in interphase and at the end of mitosis. Dev. Cell 2006, 11 , 803–816.17141156 35. Schuck P Use of surface plasmon resonance to probe the equilibrium and dynamic aspects of interactions between biological macromolecules. Annu. Rev. Biophys. Biomol. Struct 1997, 26 , 541–566.9241429 36. Green RJ ; Frazier RA ; Shakesheff KM ; Davies MC ; Roberts CJ ; Tendler SJB Surface plasmon resonance analysis of dynamic biological interactions with biomaterials. Biomaterials 2000, 21 , 1823–1835.10919686 37. Banerjee S ; Bartesaghi A ; Merk A ; Rao P ; Bulfer SL ; Yan Y ; Green N ; Mroczkowski B ; Neitz RJ ; Wipf P ; Falconieri V ; Deshaies RJ ; Milne JLS ; Huryn D ; Arkin M ; Subramaniam S 2.3 angstrom resolution cryo-EM structure of human p97 and mechanism of allosteric inhibition. Science 2016, 351 , 871–875.26822609 38. Conicella AE ; Huang R ; Ripstein ZA ; Nguyen A ; Wang E ; Löhr T ; Schuck P ; Vendruscolo M ; Rubinstein JL ; Kay LE An intrinsically disordered motif regulates the interaction between the p47 adaptor and the p97 AAA+ ATPase. Proc. Natl. Acad. Sci. U.S.A 2020, 117 , 26226–26236.33028677 39. Ray M ; Tang R ; Jiang ZW ; Rotello VM Quantitative Tracking of Protein Trafficking to the Nucleus Using Cytosolic Protein Delivery by Nanoparticle-Stabilized Nanocapsules. Bioconjugate Chem. 2015, 26 , 1004–1007. 40. Mattaj IW ; Englmeier L Nucleocytoplasmic transport: the soluble phase. Annu. Rev. Biochem 1998, 67 , 265–306.9759490 41. Silver PA How proteins enter the nucleus. Cell 1991, 64 , 489–97.1991319 42. Beuron F ; Dreveny I ; Yuan X ; Pye VE ; Mckeown C ; Briggs LC ; Cliff MJ ; Kaneko Y ; Wallis R ; Isaacson RL ; Ladbury JE ; Matthews SJ ; Kondo H ; Zhang X ; Freemont PS Conformational changes in the AAA ATPase p97–p47 adaptor complex. EMBO J. 2006, 25 , 1967–1976.16601695 43. Dale NC ; Johnstone EKM ; White CW ; Pfleger KDG NanoBRET: The Bright Future of Proximity-Based Assays. Front. Bioeng. Biotech 2019, 7 , 56. 44. Katakura Y ; Kobayashi E ; Kurokawa Y ; Omasa T ; Fujiyama K ; Suga K-I Cloning of cDNA and characterization of anti-RNase A monoclonal antibody 3A21. J. Ferment. Bioeng 1996, 82 , 312–314. 45. Zhang XY ; Wang YZ Nonredundant Roles of GRASP55 and GRASP65 in the Golgi Apparatus and Beyond. Trends Biochem. Sci 2020, 45 , 1065–1079.32893104 46. Tang DM ; Xiang Y ; Wang YZ Reconstitution of the cell cycle-regulated Golgi disassembly and reassembly in a cell-free system. Nat. Protoc 2010, 5 , 758–772.20360770 47. Huang SJ ; Tang DM ; Wang YZ Monoubiquitination of Syntaxin 5 Regulates Golgi Membrane Dynamics during the Cell Cycle. Dev. Cell 2016, 38 , 73–85.27404360 48. Joshi G ; Bekier ME ; Wang YZ Golgi fragmentation in Alzheimer's disease. Front. Neurosci 2015, 9 , 340.26441511 49. Liu JY ; Huang Y ; Li T ; Jiang Z ; Zeng LW ; Hu ZP The role of the Golgi apparatus in disease. Int. J. Mol. Med 2021, 47 . 50. Arkin MR ; Tang YY ; Wells JA Small-Molecule Inhibitors of Protein-Protein Interactions: Progressing toward the Reality. Chem. Biol 2014, 21 , 1102–1114.25237857 51. Scott DE ; Bayly AR ; Abell C ; Skidmore J Small molecules, big targets: drug discovery faces the protein-protein interaction challenge. Nat. Rev. Drug Discov 2016, 15 , 533–550.27050677 52. Pei DH ; Buyanova M Overcoming Endosomal Entrapment in Drug Delivery. Bioconjugate Chem. 2019, 30 , 273–283. 53. Zeng J ; Zhang J ; Tanaka T ; Rabbitts TH Single Domain Antibody Fragments as Drug Surrogates Targeting Protein–Protein Interactions inside Cells. Antibodies 2013, 2 , 306–320. 54. Teng KW ; Tsai ST ; Hattori T ; Fedele C ; Koide A ; Yang C ; Hou X ; Zhang Y ; Neel BG ; O’Bryan JP ; Koide S Selective and noncovalent targeting of RAS mutants for inhibition and degradation. Nat. Commun 2021, 12 , 2656.33976200 55. Akkapeddi P ; Teng KW ; Koide S Monobodies as tool biologics for accelerating target validation and druggable site discovery. RSC Med. Chem 2021, 12 , 1839–1853.34820623 56. Schneider AFL ; Kithil M ; Cardoso MC ; Lehmann M ; Hackenberger CPR Cellular uptake of large biomolecules enabled by cell-surface-reactive cell-penetrating peptide additives. Nat. Chem 2021, 13 , 530–539.33859390

PMC009xxxxxx/PMC9339253.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0413066 2830 Cell Cell Cell 0092-8674 1097-4172 35809571 9339253 10.1016/j.cell.2022.06.016 NIHMS1815634 Article Tools for Mammalian Glycoscience Research Griffin Matthew E. 1 Hsieh-Wilson Linda C. 2* 1 Department of Chemistry, University of California Irvine, Irvine, CA 92697, USA 2 Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 92115, USA * Correspondence: lhw@caltech.edu (L.C.H.W.) 5 7 2022 21 7 2022 08 7 2022 21 7 2023 185 15 26572677 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Cellular carbohydrates or glycans are critical mediators of biological function. Their remarkably diverse structures and varied activities present exciting opportunities for understanding many areas of biology. In this Primer, we discuss key methods and recent breakthrough technologies for identifying, monitoring, and manipulating glycans in mammalian systems. pmcIntroduction Glycans are ubiquitous and play important roles in fundamental processes ranging from neural development and cell signaling to immune regulation and host-pathogen interactions (Varki, 2017). All living cells are coated with a complex array of glycans that adorn the proteins, lipids, and even RNAs in cell membranes and cell walls. Glycosylation is one of the most common post-translational modifications. The majority of all human proteins are glycosylated, and many glycans become altered and contribute to the pathophysiology of major diseases, including cancer (Pinho and Reis, 2015), diabetes (Reily et al., 2019), Alzheimer’s disease (Haukedal and Freude, 2020), and infectious diseases such as COVID-19 (Clausen et al., 2020; Zhou and Cobb, 2021). With the advent of new technologies for studying glycans, there is a growing ability to understand glycosylation and its myriad of functions. Glycans have historically been challenging to study compared to other major classes of biomolecules. Unlike DNA, RNA, and proteins, glycans are not produced using a specific template. Rather, their biosynthesis relies on the intrinsic specificities and spatiotemporal expression of a series of glycosyltransferases (GTs), which ultimately leads to families of related but non-equivalent glycan structures. Glycans can also be assembled into both linear and branched structures, wherein the monosaccharide building blocks are joined at one of several positions around the sugar ring with α- or β- stereochemistry at the anomeric carbon (Figure 1A). Thus, the potential structural complexity of glycans vastly exceeds that of DNA, RNA, and proteins. Due to the decentralized biosynthetic machinery that may have purposefully evolved to generate microheterogeneity and the resulting structural diversity of glycans, standardized methods used for other biopolymers like polymerase chain reaction (PCR) amplification and sequencing are not directly applicable to glycans. Moreover, the linear, stepwise nature of glycan biosynthesis limits the power of genetic methods, which disrupt not only the target glycan on multiple glycoconjugates but also other glycan structures produced in subsequent biosynthetic steps. Glycoscientists have addressed many of these challenges through the development of new technologies for understanding glycan function that often embrace and even exploit the structure-specific differences across glycan classes. This approach has yielded a wide-ranging set of customized tools, which have been accompanied by an enthusiasm from glycoscientists to collaborate with the broader scientific community. In this Primer, we aim to provide a critical overview and interpretation of current methods in mammalian glycobiology. Because of the sheer breadth of glycans and approaches, this Primer is by no means comprehensive. Instead, we have selected important techniques that can be used to address major questions in glycoscience research. Throughout the Primer, we will draw examples from four major classes of mammalian glycans: O-linked/N-acetylglucosamine (O-GlcNAc), O- and N-glycans, and glycosaminoglycans (GAGs). These classes exemplify both the structural diversity of carbohydrates as well as the range of approaches that can be used to study them. When possible, we also highlight other notable reviews for further reading. Glycoscience is a central field with links to all areas of biology. We hope that this Primer on the current glycoscience toolkit will inspire the broader scientific community to take on the exciting challenge of understanding glycan function and help define new roles for these exquisitely varied and complex molecules. A Brief Guide to Glycans and Glycosylation Glycans are a wide-ranging group of biomolecules that vary significantly in size, composition, localization, and attachment. Their structures span from single monosaccharides to elaborately branched oligosaccharides as well as long polysaccharide polymers with molecular masses >1000 kDa. In vertebrates, glycans are composed of nine main monosaccharide building blocks: glucose (Glc), galactose (Gal), xylose (Xyl), mannose (Man), fucose (Fuc), N-acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc), glucuronic acid (GlcA), and N-acetylneuraminic acid (Neu5Ac) (Figure 1B) and are synthesized by GT enzymes using activated nucleotide sugar donors (Figure 1C). Unlike nucleic acids and proteins, glycans are not linear polymers with a conserved backbone and functional side groups. Rather, monosaccharides form the backbone of glycan structures and are joined by various regioisomeric and stereoisomeric linkages. Further diversification of glycan structure arises through the post-glycosylational modification of certain glycans by sulfation, acetylation, methylation, phosphorylation, and epimerization. To streamline and standardize the depiction of glycans, the glycoscience community has adopted the modern Symbol Nomenclature for Glycans (SNFG) (Neelamegham et al., 2019), which represents monosaccharides as color-coded shapes with text abbreviations to indicate glycosidic linkages and post-glycosylational modifications. Glycan diagrams are conventionally drawn from their non-reducing end on the left or top to their reducing end on the right or bottom to rapidly convey and compare complex structural information across related families. We will use the SNFG standard throughout the Primer and strongly encourage its general use by all scientists to facilitate accurate communication regarding glycans. Many mammalian glycans can be categorized into one of four structurally diverse classes: O-GlcNAc, O-and N-glycans, and GAGs. O-linked N-acetylglucosamine or O-GlcNAc, is a single GlcNAc monosaccharide that is attached to serine or threonine residues of proteins (Figure 2A). Unlike nearly all other forms of glycosylation, the O-GlcNAc sugar is not further elaborated, and O-GlcNAc glycosylation (also known as O-GlcNAcylation) is a dynamic, inducible modification that occurs primarily on intracellular proteins. O-GlcNAcylation has been identified on thousands of proteins yet is mediated by only a single pair of enzymes: O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA). OGT, like other GTs, uses a nucleotide sugar donor, uridine diphosphate N-acetylglucosamine (UDP-GlcNAc), to modify substrate proteins. As UDP-GlcNAc biosynthesis incorporates metabolites from central carbon, lipid, and nucleotide production, O-GlcNAcylation often serves as a nutrient and stress sensor that links the global metabolic status of the cell to the regulation of fundamental processes such as transcription, translation, and signal transduction (Yang and Qian, 2017), with aberrant O-GlcNAcylation events associated with metabolic and aging disorders (Bond and Hanover, 2013). Glycans are also often attached to cell-surface or secreted extracellular proteins. These O- and N-glycans are named based on their site of attachment: serine or threonine for O-glycans and asparagine for N-glycans. Although similar to O-GlcNAc in terms of attachment, O-glycans contain a much broader range of sugars, encompassing O-linked GalNAc, GlcNAc, Fuc, Glc, Xyl, and Man mono- and oligosaccharides, and typically range in size from one to more than twelve sugar residues. Mucin-type O-glycans are the most abundant class of these structures and are characterized by an O-linked GalNAc residue, also known as the Tn antigen (Figure 2B) (Rangel-Angarita and Malaker, 2021). The first GalNAc residue can be extended with multiple monosaccharides to produce eight core structures, which in turn are further elaborated through the activities of other GTs to contain other motifs such as fucosylation and sialylation. Hundreds of these modifications are found on the mucin proteins that line mucosal surfaces and form a physical barrier between host and the environment. Moreover, mucin and its glycans are highly dysregulated across many cancers and have been explored extensively for the development of cancer vaccines (Pinzon Martin et al., 2019). N-glycans are also built on a conserved core (the branched pentasaccharide Man3GlcNAc2), which is modified with glycan “antennae,” generating structures that are classified into three broad groups: high mannose, complex, and hybrid (Figure 2C). During N-glycan synthesis, a large N-glycan precursor is attached to a dolichol phosphate lipid anchor and then transferred by oligosaccharyltransferase (OST) onto targeted Asn residues. The glycan undergoes trimming by glycosidases and further elaboration by GTs. Under normal conditions, N-glycans are first produced in the endoplasmic reticulum and O-glycans in the Golgi apparatus. As protein substrates transit the Golgi, nascent O- and N-glycans encounter a network of GTs that extend and cap the growing oligosaccharide in an incompletely understood process dictated by the expression levels, localization, and activation of each enzyme, as well as nucleotide sugar donor levels. Unlike other biosynthetic processes like transcription and translation where low fidelity can be detrimental, the imprecise process of O- and N-glycan biosynthesis is likely an evolutionary benefit. The resulting divergent glycans may allow cells to finely tune protein structure, folding, and function by sampling multiple combinations of glycan structures or “glycoforms” on individual proteins. Mammals can also produce polysaccharides, which are represented by the ubiquitous, abundant linear polymers known collectively as glycosaminoglycans (GAGs) (Figure 2D). GAGs such as heparin/heparan sulfate (HS), chondroitin sulfate (CS), and dermatan sulfate (DS) are assembled from disaccharide units consisting of a glucuronic acid (GlcA) and a hexosamine sugar (GlcNAc or GalNAc). The iduronic acid (IdoA) residues found at irregular intervals in both HS and DS are generated through modification of GlcA by epimerase enzymes. Each disaccharide is also differentially sulfated along the polysaccharide chain by sulfotransferases, resulting in diverse patterns of sulfation. Consequently, GAGs have remarkable structural complexity in the form of many sulfation sequences, as well as domains of high and low sulfation density, which serve as docking sites for more than 800 proteins (Vallet et al., 2021; Xu and Esko, 2014). Accordingly, GAGs regulate a wide array of biological processes, ranging from animal development and immune regulation to infection and neuroregeneration (Mikami and Kitagawa, 2013; Xu and Esko, 2014). In the case of HS, CS, and DS, the polysaccharide chains are attached to serine residues of proteins through a xylose-containing tetrasaccharide linker. Defects in the enzymes that generate this linker region are associated with severe congenital disorders of glycosylation (CDGs) (Ng and Freeze, 2018). Proteoglycans, the cell-surface or secreted proteins to which GAGs are attached, generally possess between one and five polysaccharide chains from one or more GAG classes. Thus, the multiple levels of structural diversity found in GAGs, from sulfation motif to charge density to protein anchor, provide ample means to engage and direct proteins and cellular activity. Beyond the glycans described here, many other carbohydrate structures exist in mammalian systems. These include unique modifications to other protein residues including C-mannosylation of tryptophans within thrombospondin type I repeats and galactosylation of hydroxyproline and hydroxylysine residues within collagen. In addition to protein anchors, glycans are attached to lipids on the cell surface such as sphingo- or glycerolipids. Specialized glycosylphosphatidylinositol (GPI) lipids are used to anchor over 100 different proteins to the cell surface. Very recently, small RNAs on the cell surface were also found to be decorated with N-glycans (Flynn et al., 2021), extending the scope of glycan modifications to nucleic acids. More detailed information about glycan structures and their biosynthesis can be found in Essentials of Glycobiology (https://www.ncbi.nlm.nih.gov/books/NBK579918/), an authoritative and freely available textbook written by leading experts in the field. The functions and mechanisms of mammalian glycans mirror their wide range of structures. Small modifications such as O-GlcNAcylation can dynamically modulate proteins similar to phosphorylation and other post-translational modifications, altering tertiary structure, blocking ligand interactions, competing with other post-translational modifications, and/or controlling enzyme activity (Yang and Qian, 2017). As the glycan size increases, modifications like O- and N-glycans can alter the physicochemical properties of proteins such as solubility and folding (Xu and Ng, 2015), while also directly serving as binding partners for cell-surface receptors on nearby cells or for microorganisms such as viruses or bacteria (Raman et al., 2016). For larger polymeric glycans like GAGs, the polysaccharides can act independently or in concert with their protein anchors, recruiting soluble ligands to the cell surface to control protein diffusion and establish protein gradients or engaging cell-surface receptors directly to initiate signaling independently of canonical protein ligands (Kjellen and Lindahl, 2018). Considering the many diverse roles of glycans, key questions arise when aiming to connect glycan structure or “glycotype” to phenotype. For example, which glycan structures are present on which cell types, and how do these glycan populations and their interactions change during development, normal physiology, and disease? With the advent of single-cell technologies, the ability to identify and quantify levels of glycans and glycoconjugates with increasing cellular resolution will be critical for advancing an understanding of glycan function. While some glycans correlate well with a given cell type, state, or disease and may serve as good biomarkers, others may directly influence function. How can one establish causation for specific glycan-associated phenomena? Questions such as these require robust methods to detect, quantify, and manipulate individual glycans or glycosylation events both in vitro and in vivo. Here, we will provide a broad overview of state-of-the-art methods in glycoscience, as well as a guide to interpreting results and the potential limitations of each approach. Our goal is to inspire new testable hypotheses for glycan function, provide practical guidance, and connect the broader scientific community with glycoscientists to address these central questions across various biological fields. Tools of the Trade Identifying relevant glycan structures or “glycotypes” The diversity of glycan structures may initially seem daunting when investigating their biological functions. A crucial first step is to identify the relevant glycan class, the specific glycan structure if possible, and its mode of attachment to proteins or lipids. Bioinformatic resources are often helpful for determining glycotypes and guiding experimental designs. Three large-scale web-based glycoinformatics resources include: (1) GlyGen (https://glygen.org/) (York et al., 2020), (2) Glycomics@ExPASy (https://glycoproteome.expasy.org/) (Mariethoz et al., 2018), and (3) GlyCosmos (https://glycosmos.org) (Yamada et al., 2020). Coordinated by the GlySpace Alliance and supported by national scientific funding agencies, these three bioinformatic organizations integrate data regarding glycan structure, biosynthesis, gene, organism, and disease. Although the databases are relatively new and still undergoing expansion, these resources can help clarify glycan-related hits from genetic screens or proteomics experiments, and they provide key infrastructure for the collection and dissemination of data in the field. A variety of experimental reagents and approaches can be used to identify the relevance of individual glycotypes in specific biological contexts. Pharmacological inhibition can be a good starting point to link glycans to a particular function in systems with well-defined phenotypes. Commonly used inhibitors of glycosylation target broad classes of glycans by impeding their biosynthesis or preventing glycan attachment to proteins. For example, OSMI-1 acts directly on OGT to prevent transfer of the GlcNAc sugar to O-GlcNAcylated proteins (Figure 3A) (Ortiz-Meoz et al., 2015). Modified sugar donors have also been employed as mechanism-based GT inhibitors. In the case of O-GlcNAc, the sugar analog 5SGlcNAc, in which the endocyclic ring oxygen of GlcNAc is replaced with a sulfur atom, is converted by biosynthetic pathways to the corresponding nucleotide 5-thiosugar donor and transferred by OGT less efficiently than the natural UDP-GlcNAc donor, resulting in OGT inhibition and decreasing cellular O-GlcNAc levels (Gloster et al., 2011). Conversely, global O-GlcNAc levels can be increased by OGA inhibitors such as the widely used Thiamet-G (Yuzwa et al., 2008). Despite these successes, cell-permeable inhibitors selective for specific GTs or glycosyl hydrolases have been generally difficult to obtain, due partly to similar substrate binding sites across the protein families. Thus, natural toxins are commonly used to alter glycosylation levels more globally. Originally isolated as a class of antibiotics from Streptomyces species, the natural product tunicamycin blocks N-glycosylation by inhibiting transfer of the initial GlcNAc residue onto dolichol phosphate (Figure 3B) (Duksin and Mahoney, 1982). Azaserine and DON (6-diazo-5-oxo-L-norleucine), other Streptomyces products, inhibit hexosamine biosynthesis by blocking glutamine fructose-6-phosphate amidotransferase (GFAT), thereby reducing O-GlcNAcylation and other forms of glycosylation (Figure 3C) (Brimble et al., 2010). However, these molecules act as glutamine mimics that broadly inhibit amidotransferases, causing pleiotropic effects. Synthetic “look-alike” monosaccharides such as fluorinated and deoxy sugar analogues antagonize glycosylation through various mechanisms (Figure 3D). For example, fluorinated Fuc and Neu5Ac analogues have been shown to prevent fucosylation and sialylation of glycans, respectively, through competitive inhibition of the respective GTs (Rillahan et al., 2012). Although the mechanisms underlying their inhibitory activity remain relatively unclear and likely involve multiple pathways, deoxysugars like 2-deoxyglucose have been used to prevent N-glycosylation (Kurtoglu et al., 2007). For Fucα(1,2)Galcontaining glycans, 2-deoxygalactose has been employed to prevent fucosylation due to lack of the 2-hydroxyl group on galactose required for Fuc attachment (Bullock et al., 1990). Hydrophobic xylosides can partially prevent GAG assembly by acting as competitive “decoy” substrates for GTs and diverting enzymatic activity away from natural proteoglycan substrates (Chua and Kuberan, 2017). GAG sulfation can also be inhibited by using sodium chlorate, a sulfate analogue that reduces global production of the sulfate donor phosphoadenosine 5’-phosphosulfate (PAPS) and affects all forms of sulfation including protein sulfation (Greve et al., 1988), or by using specific inhibitors of GAG sulfotransferases (Cheung et al., 2017). Nevertheless, results obtained using pharmacological inhibitors should be interpreted with caution and supported by other independent methods as they typically modulate large classes of glycans across the entire proteome and can result in pleiotropic effects such as ER stress or cytotoxicity that may complicate observable phenotypes. A complementary approach to chemical inhibitors is the use of recombinant enzymes to modify or cleave glycans of interest. Removal of N-glycans can be accomplished with PNGase F or Endo F1/2/3 (Figure 4A) (Tarentino and Plummer, 1994). PNGase F cleaves at the Asn-GlcNAc bond that attaches N-glycans to the protein, whereas Endo F enzymes hydrolyze N-glycans between the first and second GlcNAc residues. Each enzyme shows a distinct specificity toward N-glycan structures (e.g., number and composition of antennae or core fucosylation). Therefore, the selective use of endoglycosidases (ENGases) individually or in combination can also help to determine the relevant N-glycan structures. In addition to ENGases, exoglycosidases (EXGases) can be employed to cleave terminal glycans. For example, recombinant sialidases and fucosidases were used to study the roles of Neu5Ac and Fuc in cancer cell progression and clearance (Hudak et al., 2014; Yuan et al., 2008). Although deglycosylating enzymes are commonly applied to purified proteins, many of these enzymes have been successfully employed with live cells. For GAGs, chondroitinase ABC and heparinase I/II/III can be used to digest CS and HS polysaccharides, respectively, and have been applied to purified proteins, cultured cells, and organisms in vivo (Bradbury et al., 2002; Brown et al., 2012; Griffin et al., 2021). The presence or absence of specific glycan structures can often be confirmed using protein-based probes such as lectins and antibodies. Lectins are naturally occurring glycan-binding proteins (GBPs) that are often isolated from plants. Many lectins are commercially available and have moderate affinities for defined glycan motifs. Commonly used lectins include wheat germ agglutinin (WGA) for terminal GlcNAc residues, concanavalin A (ConA) for branching oligomannose residues of N-glycans, Sambucus nigra agglutinin (SNA) for α(2,6)-sialic acids, and Ulex europaeus agglutinin-I (UEA-I)-for α(1,2)-linked Fuc. Potential cross-reactive binding of lectins towards multiple glycan structures must be considered. The Consortium for Functional Glycomics has made microarray data on the binding of lectins to hundreds of defined glycan structures publicly available (https://ncfg.hms.harvard.edu/ncfg-data/microarray-data/lectin-quality-assurancequality-control). Moreover, machine-learning methods were recently applied to these data to annotate the complex binding specificities of 57 commercially available lectins, providing a critical guide to these important reagents (Bojar et al., 2022). As a complement to lectins, antibodies have been generated against all major classes of mammalian glycans, with over 1,000 monoclonal antibodies previously described (Sterner et al., 2016). The Database for Glycan Reagents (DAGR) was established through the Center for Cancer Research at the National Cancer Institute (https://dagr.ccr.cancer.gov) to facilitate the use of anti-carbohydrate antibodies and lectins. Glycan-binding lectins and antibodies have been leveraged for glycan detection using standard methods, including histology, protein or western blotting, and enzyme-linked lectin/immunosorbent assays. Microarrays of lectins and carbohydrate antibodies have also been constructed for glycomics applications (Dang et al., 2020) and to identify glycans involved in processes such as melanoma metastasis (Agrawal et al., 2017) and viral host response (Heindel et al., 2020). As mentioned above, care must be taken when interpreting data using such reagents as they often bind multiple glycans containing related (and sometimes unrelated) structural motifs. Glycan structure cannot be definitively determined based solely on lectin or antibody binding. The comparative use of multiple glycan-binding reagents should be considered to provide further evidence of glycan identity, along with mass spectrometry (MS) and other methods such as metabolic or chemoenzymatic labeling (discussed in “Detecting and monitoring glycans”). By far, MS analysis is the most definitive and comprehensive method to identify glycan structures (Ruhaak et al., 2018). Unlike other methods described above, liquid chromatography in conjunction with MS (LC-MS) provides a direct readout of the glycan structure through intact molecular mass determination and sequencing by MS/MS fragmentation (Figure 4A). Although MS cannot easily discriminate between isobaric diastereomers (e.g., GlcNAc vs. GalNAc) without further fragmentation of the monosaccharide, glycomic approaches have now become more standardized, with commercialized kits for sample preparation and internal standards for spectral matching. Kits frequently employ chemical or enzymatic methods to release the targeted glycans, which are then derivatized with UV-active compounds like 2-aminobenzamide (2-AB) to facilitate detection during LC separation. Alternatively, benzyl GalNAc glycosides have also been employed as decoys to produce secreted O-glycans for cell state-dependent O-glycome profiling (Kudelka et al., 2016). For larger polysaccharides like GAGs, the full determination of individual linear sequences remains challenging due to their overall size and multiple levels of structural heterogeneity. While a few examples of GAG sequencing have been reported (Ly et al., 2011; van Kuppevelt et al., 2017), GAGs are typically enzymatically digested and subjected to disaccharide compositional analysis by LC-MS, providing a relative quantification of individual disaccharide motifs at the expense of linear sequence data. Although glycomics analyses can often determine the glycan structure, critical information regarding their attachment sites to proteins is often lost. O-glycans, including O-GlcNAc, generally lack consensus motifs, and knowledge of the glycosylation sites can be critical for generating and testing hypotheses regarding glycan function. Therefore, the goal of glycoproteomics approaches is not only to identify the glycan structure but also to sequence the underlying peptide (Figure 4B). The mass spectra of complex O- and N-glycans can be severely complicated by partial fragmentation of the glycan. Next-generation bioinformatic approaches and various MS fragmentation methods have been developed to address this complexity and increase the number of identified glycosylated proteins (reviewed in (Chernykh et al., 2021; Oliveira et al., 2021)). Glycoproteomics has also greatly benefited from new methods to enrich for glycosylated proteins and peptides (Riley et al., 2021). The presence of abundant, nonglycosylated peptides often obscures the detection of rarer, glycosylated peptides. Glycopeptide enrichment can be achieved by lectin affinity chromatography, hydrophilic interaction liquid chromatography, and metabolic or chemoenzymatic labeling of glycans with affinity tags (described in ‘Detecting and monitoring glycosylation’). Moreover, specialized methods for enriching and mapping O-GalNAc sites have also been developed using bacterial proteases that cleave specific peptide sequences proximal to mucin-type O-glycans (Rangel-Angarita and Malaker, 2021). When used successfully, these methods have enabled large-scale, proteome-wide profiling of glycoproteins in multiple contexts, including the identification of over 1,750 O-GlcNAcylation sites in neuronal synaptosomes (Trinidad et al., 2012), 2,200+ O-glycopeptides in activated T cells (Woo et al., 2018), 600+ N-glycopeptides from human serum (Li et al., 2019a), and full profiling of GAG composition in 20 human cell lines (Li et al., 2015). Generating chemically defined glycans and glycoconjugates Just as the automated synthesis of oligonucleotides and peptides revolutionized our understanding of these biomolecules and ushered in a new era of modern molecular biology, access to a broad range of glycan structures will be critical for advancing glycoscience and constitutes an essential part of the glycoscientist’s toolkit. Glycans and glycoconjugates of unique, defined structure serve as invaluable tools for many applications, acting as standards for structure identification, as ligands for studying GBP interactions, and as simplified glycan or glycoprotein mimetics for probing function. Significant advances in carbohydrate chemistry have recently been made toward producing complex oligosaccharides and are reviewed elsewhere (Boltje et al., 2009; Mende et al., 2016). Here, we will highlight the emergence of innovative technologies that facilitate the production of a wide range of glycans and increase availability of these chemical probes for broad distribution. A major focus of modern carbohydrate chemistry has been the development of methodologies for the automated assembly of oligo- and polysaccharides (Figure 5A) (Panza et al., 2018; Wen et al., 2018). Glycan synthesis is challenging due to the need to control the regioselectivity (which hydroxyl position around the sugar ring) and the stereoselectivity (α or β anomer) of each newly formed glycosidic bond. To achieve this, monosaccharide building blocks with chemical protecting groups at various hydroxyl positions must first be prepared. These protecting groups must be removable to unmask a desired position for coupling with another monosaccharide yet remain inert under other reaction conditions. As the optimal set of protecting groups depends on the glycan structure, there are no universal, “one-size-fits-all” building blocks for glycan synthesis. Thus, the synthesis of a glycan target can take several months and in some cases years to complete. Nonetheless, multiple methods have been developed to expedite glycan assembly. For example, elongation of the growing oligosaccharide chain while attached to Merrifield and other resins used for solid-phase peptide synthesis has enabled the production of various complex glycans, including α-glucan polysaccharides, mycobacterial oligoarabinofuranosides, blood group antigens, and GAGs (Guberman and Seeberger, 2019). Alternatively, installation of a fluorinated tag onto the reducing end of the growing oligosaccharide chain can facilitate automated solution-phase synthesis and purification of reaction intermediates by fluorous solid-phase extraction (Tang and Pohl, 2016). These methods have been incorporated into automation platforms such as the commercially available synthesizer Glyconeer 2.1 (Hahm et al., 2017), HPLC-based systems (Panza et al., 2020), and microwave-assisted peptide synthesizers (Danglad-Flores et al., 2021). Recent technologies have also advanced the large-scale purification of complex glycans from natural sources (Zhang et al., 2020). A key step in the purification is the release of natural glycans from their pendant protein or lipid conjugates. A mild method employing dilute bleach (NaClO) was developed to oxidatively release O- and N-glycans from glycoproteins and glycan nitriles from glycosphingolipids (Song et al., 2016; Zhu et al., 2018) (Figure 4A). Notably, these chemical methods could be scaled to kilogram quantities of protein or tissue and eliminated the need for expensive enzymes like PNGase F. Although chemical synthesis is still the better choice for lower abundance glycans, the harvesting of oligosaccharides from plant and animal tissue now provides a practical alternative for certain N- and O-glycans. Enzymes have also been exploited to facilitate glycan production (Figure 5B). These methods often employ purified natural or engineered bacterial GTs along with their nucleotide sugar donor substrates. Powerful multienzyme systems have been developed that combine GTs with upstream biosynthetic enzymes to produce the required nucleotide sugars, providing one-pot syntheses of oligosaccharides (Yu and Chen, 2016). The scope of enzymatic glycan synthesis can be expanded by combining enzymes with chemically modified substrates. Such chemoenzymatic approaches have enabled the production of many bioactive carbohydrates and continue to transform the field. For example, the anticoagulant drug Arixtra is a synthetic heparin-based pentasaccharide used for the treatment of deep vein thrombosis. While Arixtra requires 50 chemical steps to synthesize (~0.1% yield), a biosimilar heptasaccharide containing the Arixtra pentasaccharide motif was obtained chemoenzymatically using GT and sulfotransferase enzymes in 12 steps and 37% overall yield (Xu et al., 2011). In another example, the challenge of producing asymmetrically branched N-glycans was overcome using a chemoenzymatic approach (Wang et al., 2013b). Automation platforms for enzymatic and chemoenzymatic glycan assembly have also been developed (Li et al., 2019c; Wen et al., 2018), laying the foundation for the production and broad distribution of large collections of complex oligosaccharides. Glycans and GBPs are often presented in multivalent forms in vivo, which enhances the affinity of glycan-protein interactions through avidity. Glycopolymers have been synthesized to mimic these high avidity interactions by presenting multiple copies of individual glycan motifs (Kiessling and Grim, 2013) and are useful tools for modulating biological function. For example, linear polymers containing pendant sulfated Lewis X epitopes or Neu5Ac glycans have been exploited to characterize leukocyte rolling (Sanders et al., 1999) and suppress B cell activation (Courtney et al., 2009), respectively. As GAGs are naturally multivalent, polymers with pendant HS or CS disaccharides are excellent simplified mimetics that can elicit diverse phenotypes, including neurite outgrowth (Rawat et al., 2008), chemokine signaling activation (Sheng et al., 2013), and blood anticoagulation (Oh et al., 2013) in a sulfation motif-dependent manner. Multivalent glycopolymers have also been functionalized with lipid tails to remodel cell surfaces with glycans (described in ‘Modulating glycans to probe function: connecting ‘glycotype’ to phenotype’) or with photocrosslinking functionalities to identify GBPs (described in ‘Discovering and characterizing glycan-protein interactions’). Glycans often modulate the proteins to which they are attached, affecting physical properties such as folding, stability, and solubility, as well as their biological activities. Thus, the production of homogeneous glycoproteins has been crucial for defining the functions of specific glycans and advancing the development of biologic drugs. Semi-synthesis methods such as native chemical ligation and expressed protein ligation, in which synthetic glycopeptides are ligated to peptide or protein fragments, are highly effective at producing homogeneously glycosylated proteins (Figure 5C). These approaches have been applied to glycoproteins such as the drug erythropoietin (Wang et al., 2013a), a 166-amino acid protein with four glycosylation sites that stimulates erythrocyte production, as well as α-synuclein (Marotta et al., 2015), a 140-amino acid O-GlcNAcylated protein whose aggregation is associated with the pathology of Parkinson’s disease. Alternatively, the glycan profiles of proteins can be tailored by engineering glycosylation pathways in cells (similar to Figure 6A). Genetic engineering to manipulate the glycosylation patterns on purified proteins is commonly performed, and although the resulting proteins are still glycoform mixtures, these facile and scalable production systems reduce structural complexity and remain an industry standard. Chemical biology approaches provide more precise control over glycoform production in vitro. For example, the N-glycans on therapeutic antibodies can be modified by ENGases, which first trim the conserved N-glycans to a single GlcNAc residue (similar to Figure 6B) (Giddens et al., 2018; Huang et al., 2012). Mutant ENGases then transfer a chemically modified intact N-glycan oxazoline donor onto the glycan stub, producing a homogeneously N-glycosylated antibody. Additional modifications using other enzymes like fucosidases can be added to the process, further defining the specific glycoform of the antibody. Another complementary approach uses genetic code expansion to make defined glycosylated proteins by installing chemically reactive, nonnatural amino acids at specific sites for click chemistry-based functionalization (Machida et al., 2015). A nonnatural dehydroalanine amino acid was employed to produce O-GlcNAc mimics via radical-mediated ligation of alkyl halides, and N-glycan mimics were installed through ENGase-mediated extension of the GlcNAc residue, providing a general method to produce glycoprotein mimetics (Wright et al., 2016). Many of these defined synthetic glycans and new technologies are currently available through individual laboratories or companies. Through the support of the NIH Common Fund Glycoscience Program, great progress has been made in creating new resources, tools, and methods to render the study of glycans more accessible to the larger research community. A partial listing of these resources can be found at https://commonfund.nih.gov/Glycoscience/programresources. Discovering and characterizing glycan-protein interactions Glycans often exert their activities through direct interactions with proteins. Glycoproteins, glycolipids, and polysaccharides form an extramembrane compartment, termed the glycocalyx, which is found on nearly all eukaryotic cells. The glycocalyx is the first site of cellular contact with the environment, and thus glycans play key roles in cell-cell and cell-matrix interactions critical for processes such as immune cell trafficking, embryonic development, and cancer metastasis. GBPs include lectins but also extend to many other proteins not classically defined as lectins. For example, a wide variety of proteins bind to GAGs, including soluble ligands like growth factors and cytokines, transmembrane proteins such as receptor tyrosine kinases and phosphatases, as well as proteins from microbial pathogens like the SARS-CoV-2 spike glycoprotein (Vallet et al., 2021). Therefore, the study of carbohydrate-protein interactions is critical to understanding glycan function. Glycan molecules of defined structure have greatly facilitated the discovery of novel protein receptors. For example, affinity purification using immobilized glycans followed by mass spectrometry (AP-MS) is a powerful method to enrich and identify GBPs. Because carbohydrate-protein interactions can be low to moderate affinity, methods that capitalize on multivalent interactions to strengthen the interaction or covalently capture proteins using chemical crosslinking agents are highly effective. For example, GBPs have been identified using gold nanoparticles (Sakurai et al., 2016) and synthetic glycopolymers (Wibowo et al., 2014) functionalized with multiple copies of sugar epitopes, along with photocrosslinkers such as benzophenone or nitrophenylazide. Direct conjugation of a bifunctional probe containing a photocrosslinker and an alkyne group for appending a biotin tag to commercially available, natural GAG polysaccharides enabled sulfation motif-specific CS binding proteins to be enriched and identified from neurons (Joffrin and Hsieh-Wilson, 2020). In some cases, binding may be mediated by interactions not only with the glycan itself but also with the associated glycoprotein. To enable detection of such glycoprotein-protein interactions, metabolic labeling (ML) can be used to install photoaffinity labels onto cellular glycoproteins. ML, also known as metabolic oligosaccharide engineering, exploits the promiscuity of mammalian salvage pathways to convert nonnatural monosaccharides into nucleotide sugar donors, which are then incorporated into cellular glycans by GTs (Figure 5D). Sugar analogs of GlcNAc, ManNAc, and Neu5Ac have been synthesized containing aryl azide and diazirine groups for photocrosslinking (Han et al., 2005; Tanaka and Kohler, 2008; Yu et al., 2012). Cells are treated with membrane-permeable versions of these analogs, which are deacetylated by intracellular esterases and subsequently incorporated into newly formed glycoproteins. Following light-induced crosslinking, immunoprecipitation of the glycoprotein, coupled with MS analysis, can be used to identify putative glycoprotein interactors. Specific glycoprotein-protein interactions can also be identified by proximity labeling methods. For example, ML was employed to install a non-natural azide group into Neu5Ac glycans on cell-surface glycoproteins (Li et al., 2019b). This modified glycan was then reacted using bioorthogonal chemistry with a cyclooctyne probe containing a coordinated Fe(III) ion. When treated with hydrogen peroxide, Fe(III) generated hydroxyl radicals, which in turn oxidized nearby amino acid residues that could be detected by MS. In another approach, tyramide radicalization was used to identify glycoprotein ligands for Siglecs, a family of Neu5Ac recognition proteins essential for self-nonself discrimination by the immune system (Chang et al., 2017). Siglec-horseradish peroxidase (HRP) complexes were formed using recombinant FLAG-tagged Siglec proteins and anti-FLAG antibodies conjugated to HRP. Incubation of the Siglec-HRP complexes with cells, followed by addition of biotin tyramide and hydrogen peroxide generated short-lived tyramide radicals that biotinylated nearby proteins, allowing for identification of both known and new Siglec ligands. Similarly, glycoprotein ligands for galectin-1 and galectin-3 were identified by conjugating these GBPs to the ascorbate peroxidase APEX2, which enabled labeling of proximal proteins using biotin tyramide and hydrogen peroxide (Joeh et al., 2020; Vilen et al., 2021). As the HRP and APEX methods use a commercially available biotin probe, application to other GBPs should be readily feasible. Upon identifying a glycoprotein-protein interaction, establishing the glycan structural motif(s) responsible for mediating the interaction can provide crucial insights into its specificity and function. The selectivity of glycan-protein interactions can range from promiscuous affinity for multiple, related carbohydrate structures to high selectivity for individual structures. Glycan microarrays have emerged as a high-throughput technology to determine the binding specificities of GBPs (Rillahan and Paulson, 2011). Similar to traditional DNA microarrays, glycan microarrays containing a wide range of both synthetic and natural glycans have been constructed using robotic printing technologies. Protein binding to specific glycans spotted on the microarray is typically detected using biotinylated or fluorescently labeled antibodies. This miniaturized format permits rapid interrogation of many glycan-protein interactions in parallel, while requiring minimal glycan and protein material. As mentioned above, glycan microarrays have provided vital information regarding lectin specificity and have been applied to a wide range of GBPs (Gao et al., 2019), serum antibodies (Xia and Gildersleeve, 2015), and even intact viruses (Smith and Cummings, 2014). Despite their ease, glycan microarrays are inherently limited by the diversity of glycans on the array. Current microarrays contain only a fraction of the mammalian glycome and an even smaller proportion of the microbial glycome, highlighting a critical need to expand access to pure, well-defined glycan molecules. As discussed above, the synthesis and biochemical isolation of large panels of glycans is technically challenging. To combat these issues and facilitate the use of glycan microarrays in the broader community, the NIH has supported large-scale efforts to produce glycans and glycan microarrays. Numerous glycan microarrays are currently available upon request from the Consortium for Functional Glycomics (CFG) and National Center for Functional Glycomics (https://ncfg.hms.harvard.edu/microarrays). The widely used CFG array (version 5.2) has approximately 600 mammalian glycans, while the microbial glycan array has over 300 glycans. In addition, the Glycosciences Laboratory at Imperial College London has a large collection of glycans and offers microarray analyses (http://www.imperial.ac.uk/glycosciences/). Several companies also provide glycan microarrays and analysis services for targeted subsets of glycans, including GAGs and common N- and O-glycans found in serum, plasma, and other tissues. Newer approaches with chemically released natural glycans, multiplexing capabilities, and alternative solution-binding assays will continue to advance this cornerstone technology. As glycan density, spatial arrangement, and presentation on lipids or proteins, as well as the presence of competing glycan structures, can significantly influence glycan recognition (Kiessling and Grim, 2013; Mende et al., 2019), cell-based platforms have also been explored as a complementary approach to glycan microarrays. Knockout cell lines generated by chemical mutagenesis and selected for altered glycosylation have historically been used to characterize glycosylation pathways, identify relevant genes, and elucidate the functional roles of glycans (Patnaik and Stanley, 2006). Modern genetic engineering techniques like zinc finger nuclease (ZFN) and CRISPR/Cas9-directed editing have recently been employed to generate panels of isogenic HEK293 cells with predictable structural changes to O- and N-glycans, as well as GAGs (Briard et al., 2018; Narimatsu et al., 2019). This “glycotopiary” approach to prune cell-surface glycans provides a valuable cell-based array for investigating binding specificities to glycans. Such cell lines have been exploited in conjunction with flow cytometry to determine the binding selectivity of Neu5Ac-binding Siglec proteins and hemagglutinin (HA) proteins from individual strains of influenza. It is worth noting, however, that the cells generated by genetic and chemoenzymatic approaches simultaneously present multiple different glycoforms, necessitating comparative analyses across various cell lines to determine specificity for individual glycans. Nevertheless, cell-based platforms are a powerful complement to traditional glycan microarrays and are being made broadly available to the scientific community. Another recently developed approach that expands the capabilities of traditional glycan microarrays is the “liquid” glycan array platform. In this approach, defined glycans are covalently attached to DNA-barcoded M13 bacteriophages using click chemistry (Sojitra et al., 2021). The bacteriophage mixture is applied to GBPs in vitro or even in vivo, and bound bacteriophages are then sequenced to identify the structures and valencies of potential interacting glycans. Notably, this liquid glycan array method enables not only the study of GBP specificities, but also other important facets of glycan recognition, such as multivalency, avidity, as well as the potential for crosstalk and dynamic competition between glycans in vivo. Here again, access to defined, azide-functionalized glycans is required, which may limit the diversity of structures that can be used. Nonetheless, this DNA-encoded approach has unique advantages and may become widely utilized as the library of phage-displayed glycans grows. Glycan microarrays are valuable as a hypothesis-generating discovery tool to rapidly screen putative GBPs and identify potential glycan structures important for recognition. These initial screens should be followed up with conventional binding assays such as enzyme-linked immunosorbent assays, surface plasmon resonance, biolayer interferometry, isothermal titration calorimetry, fluorescence polarization, and frontal affinity chromatography (Nagae and Yamaguchi, 2018). New technologies such as mass photometry are also likely amenable to measuring protein complexes mediated by glycans (Young et al., 2018). Importantly, these assays provide independent validation of structure-dependent binding and can be used to derive kinetic and thermodynamic parameters such as enthalpy and entropy (ΔH, ΔS), dissociation constants (Kd), and Kon and Koff rate constants. Structural approaches and site-directed mutagenesis help to define further the molecular basis of glycan-protein recognition, including the binding interface, intermolecular forces, and specificity of the interaction. The most widely used techniques, nuclear magnetic resonance (NMR) and X-ray crystallography, require pure, structurally defined glycans, which limits their use to certain accessible carbohydrate classes. Another major challenge to structural glycobiology and high-resolution structure determination is the intrinsic flexibility of glycans, which leads to heterogeneous ensembles of defined conformational states. NMR methods are particularly well suited to studying glycan conformation and dynamics, and various NMR techniques have been utilized to gain insights into glycan-protein interactions, most notably nuclear Overhauser effect spectroscopy (NOESY), saturation transfer difference NMR (STD-NMR), and WaterLOGSY (reviewed in (Gimeno et al., 2020; Nieto, 2018)). X-ray crystallography studies have provided crucial data on conformational features of carbohydrates and their interactions within protein binding clefts. For example, seminal structures of the ternary HS-fibroblast growth factor (FGF)-fibroblast growth factor receptor (FGFR) complex highlighted how GAGs can engage both ligands and receptors in a single complex, likely aiding in receptor activation (Pellegrini et al., 2000; Schlessinger et al., 2000). Computational modeling methods are often used in combination with experimental structures or protein homology models when structures are unobtainable. For example, a combined computational-experimental approach was used to investigate the specificity of an antibody against the tumor-associated carbohydrate antigen sialyl-Tn (Amon et al., 2018). Antibody specificity was first investigated by glycan microarray, alanine-scanning mutagenesis, and STD-NMR to define the antibody-glycan contact surface. Computational grafting of various sialyl-Tn-related carbohydrates onto the modeled antibody structure using Gly-Spec (www.glycam.org) led to a 3D model of the antibody-glycan complex that was consistent with the experimental data, revealing features important for the high selectivity of the antibody. For GAGs, various computational programs such as GAG-Dock (Griffith et al., 2017), VinaCarb (Nivedha et al., 2016), and GlycoTorch Vina (Boittier et al., 2020) have been developed for docking of GAG oligosaccharides to their binding proteins, providing structural models that closely mimic known crystal structures and guiding investigations into the functional consequences of GAG binding. Detecting and monitoring glycosylation in vitro and in vivo The glycome responds dynamically within minutes in response to cellular stimuli or over long periods of time, such as during development or disease progression. Thus, robust methods to detect and monitor specific glycans are critical for establishing the functions of glycans in physiology and disease. As mentioned above, lectins or antibodies are often used to detect carbohydrates on glycoconjugates by western blotting or immunocytochemistry, and because many recognize the carbohydrate epitope independent of the pendant protein or lipid, they can be used to compare the relative glycosylation levels of a given glycoconjugate across different conditions. Antibodies that selectively recognize glycans at specific sites in proteins have been difficult to generate, although site-specific O-GlcNAc antibodies have been produced against a small number of O-GlcNAcylated targets (Gorelik and van Aalten, 2020). As lectins and antibodies may recognize multiple related glycans with varying affinities, care should be taken in interpreting results using these reagents, and the glycan structure and glycosylation site(s) on a protein should be confirmed by mass spectrometry and/or site-directed mutagenesis in all cases. Glycans can also be detected by modification of specific glycans with chemical reporters such as fluorescent dyes or biotin. For example, mild periodate oxidation of Neu5Ac-containing glycans generates a terminal aldehyde that can be functionalized with reporter groups using oxime chemistry (Zeng et al., 2009). Metabolic labeling (ML) and chemoenzymatic labeling (CL) methods allow multiple other classes of carbohydrates to be targeted. As described above (‘Discovering and characterizing glycan-protein interactions’ section), ML exploits the substrate promiscuity of biosynthetic enzymes to introduce nonnatural glycans bearing small chemical functionalities into cellular glycoconjugates (Figure 5D) (Wang and Mooney, 2020). In CL, an exogenous GT is used to tag existing glycans on cellular glycoconjugates with a nonnatural sugar modified with a small chemical functionality (Figure 5E) (Lopez Aguilar et al., 2017). Covalent attachment of chemical reporters to these nonnatural sugars using biorthogonal chemistry enables rapid labeling of glycans in cells, tissues, and even whole organisms. For live-cell imaging, the biorthogonal copper-catalyzed azide-alkyne cycloaddition (CuAAC) reaction can be cytotoxic, likely due to reactive oxygen species generated by the catalyst (Kennedy et al., 2011), but cytotoxicity can sometimes be avoided with improved copper ligands (Parker and Pratt, 2020) or catalyst-free reactions such as strain-promoted azide-alkyne cycloaddition (SPAAC) or tetrazine ligation (Nguyen and Prescher, 2020). CL was originally developed as a sensitive method for detecting O-GlcNAcylated proteins. A mutant β-1,4-galactosyltransferase (Y289L GalT) can be used to append a nonnatural azido- or keto-GalNAc sugar onto O-GlcNAc moieties (Clark et al., 2008; Khidekel et al., 2003). The nonnatural sugar can then be reacted to label the glycoproteins with a biotin tag, which allows for both sensitive detection and affinity enrichment for MS analysis. CL has facilitated the identification of O-GlcNAc-modified proteins, MS determination of their glycosylation sites, and quantitative proteomic profiling of O-GlcNAcylated proteins across cells and tissues. In addition to biotin, other reporters have been employed with CL, including polyethylene glycol polymers of defined mass (Rexach et al., 2010). These polymer mass tags shift the molecular weight of the glycoprotein and are visualized by western blotting to determine the stoichiometry of glycosylation. Using this approach, a significant increase in O-GlcNAcylation on phosphofructokinase-1 (PFK1) was quantified in human breast tumor tissue compared to normal tissue (Yi et al., 2012). CL with polymer mass tags has also been used to study the interplay between post-translational modifications, demonstrating that O-GlcNAcylation was induced specifically on the phosphorylated subpopulation of the transcription factor CREB in response to neuronal depolarization (Rexach et al., 2012). Notably, CL has been expanded to the detection of many other glycan motifs, including the O- and N-glycan disaccharide N-acetyllactosamine (LacNAc) (Zheng et al., 2011), the Tn antigen (O-linked GalNAc) (Wu et al., 2016), the TF antigen (O-linked Gal-GalNAc) (Li et al., 2014), Fucα(1,2)Gal (Chaubard et al., 2012), Neu5Acα(2,3)Gal (Wen et al., 2016), and terminal HS residues (Wu et al., 2018). CL has also been developed as a broad labeling method for O- and N-glycans by exploiting the substrate specificities of different sialyltransferases (Mbua et al., 2013; Yu et al., 2016). Labeling kits for O-GlcNAc, as well as many of the enzymes and nonnatural sugar donors, are commercially available or accessible as shared reagents from researchers. ML has been used in conjunction with fluorescent reporters to image cellular glycans in a variety of contexts. For example, super-resolution imaging of the glycocalyx was achieved through ML of GalNAc-containing glycans and periodate labeling of Neu5Ac (Möckl et al., 2019). The results showed nanoscale organization of glycans in the glycocalyx, along with changes in glycocalyx thickness upon oncogenic KRAS activation. ML has also been applied to image glycans in mammalian organs, including the heart (Rong et al., 2014) and brain (Xie et al., 2016), and in whole organisms such as zebrafish (Laughlin et al., 2008). To improve the selectivity of ML in vivo, “caged” sugars can be employed, in which chemical groups on the nonnatural sugar are cleaved by enzymes in the target tissue, allowing the nonnatural sugar to be incorporated (Chang et al., 2010). “Decaging” of sugars by histone deacetylase and cathepsin L enabled selective labeling of glycoproteins in ectopic tumors in mice (Wang et al., 2017). To enhance the efficiency of ML in vivo, liposomes have been used to encapsulate and deliver the nonnatural sugar analog (Xie et al., 2016). This approach allowed an azide-functionalized Neu5Ac derivative to cross the blood-brain barrier and become incorporated into sialylated glycans in mouse brains. Glycan labeling via ML and CL can be combined with other protein-specific labeling procedures to monitor the glycosylation status of specific glycoproteins directly. For example, a protein-specific fluorescent donor can be combined with a glycan-specific acceptor dye, appended via ML, for fluorescence resonance energy transfer (FRET) to measure the glycosylation status, occupancy, and cell-surface localization of glycoproteins. A modified in situ proximity ligation assay (PLA) can also be used to detect glycosylation levels on specific glycoproteins of interest (Robinson et al., 2016). In this assay, CL is used to install a biotin tag on the glycan of interest. An antibody that recognizes the target protein and an anti-biotin antibody are then incubated with the samples. Each antibody contains a complementary single-stranded DNA oligonucleotide, which hybridize with one another when in proximity, such as on the same glycoprotein. Hybridization can be detected after DNA ligation using fluorescently tagged complementary oligonucleotides or by quantitative polymerase chain reaction (qPCR). The selection of reagents for ML is critical to target specific glycans (Parker and Pratt, 2020). ManNAc-based probes (Ac4ManNAz and Ac4ManNAlk) are the preferred method to label Neu5Ac-containing glycans in mammalian systems, in part due to their cell permeability. Fuc-specific probes have been generated by additions to the C-6 position (Ac46AzFuc and Ac46AlkFuc). As GalNAc-based probes have been shown to undergo metabolic crosstalk through UDP-GlcNAc/GalNAc-4-epimerase (GALE), leading to promiscuous labeling of various glycans (Boyce et al., 2011), significant efforts have been undertaken to develop selective probes for both GlcNAc- and GalNAc-containing glycans. The development of C-6 modified GlcNAc metabolic probes (Ac36AzGlcNAc and Ac36AlkGlcAc) has enabled the specific profiling of O-GlcNAcylation (Chuh et al., 2017; Chuh et al., 2014), whereas probes with acetamide modifications (Ac4GlcNAz and Ac4GlcNAlk) will label both O-GlcNAc and GlcNAc-containing cell-surface glycoproteins. Work exploring sterically bulky acetamide derivatives of GalNAc has led to the generation of a caged derivative of N-(2-azidopropanoyl)-GalNAc-1-phosphate that can be incorporated specifically into O-GalNAc glycans in cells expressing a mutant form of the nucleotide sugar donor biosynthetic enzyme AGX1 (Debets et al., 2020). As some ML probes can nonspecifically label cysteine residues (Qin et al., 2018), newly discovered glycoproteins should be further verified by mass spectrometry or other methods. Ultimately, the choice of ML or CL depends on the specific glycan of interest, the extent of labeling and specificity required, and the experimental question being addressed. As described above, ML works particularly well for imaging glycans in vivo, whereas CL is well suited for analyzing human tissue. ML is typically substoichiometric due to competition with natural substrates and therefore should be used with caution when quantifying glycosylation levels. As CL can allow for stoichiometric tagging of endogenous glycans, CL may be better suited for quantification or when high detection sensitivity is required. In terms of specificity, ML generally labels multiple glycan classes containing the monosaccharide of interest, whereas the selectivity of CL depends on the substrate specificity of the GT employed. Thus, CL can be used to detect specific di- or trisaccharide glycan motifs unlike ML. Several detailed reviews are available that outline the best practices for these methods (Cheng et al., 2021; Lopez Aguilar et al., 2017; Parker and Pratt, 2020). In the future, advancing an understanding of the structure and function of glycans will require new tools for monitoring the glycome with increasing cellular resolution across different cell types, organs, physiological stimuli, and disease states. Single-cell techniques such as flow cytometry and single-cell RNA-sequencing (scRNA-seq) have provided key insights into cellular diversity and heterogeneity, as well as molecular and cellular states important for health and disease. Methods that employ oligonucleotide-conjugated antibodies such as CITE-seq have been developed to translate protein detection into “sequenceable” readouts (Stoeckius et al., 2017). Recently, glycan detection using DNA-barcoded lectins has been combined with this technology to quantify N-glycans on individual cells (Kearney et al., 2021; Minoshima et al., 2021). Although limited by the cross-reactive nature of lectin binding, this proof-of-principle method suggests that new technologies can be developed to profile glycans more precisely at the single-cell level in the future. As a complement to single-cell barcoding technologies, MS methods to monitor glycan structures have greatly improved and are becoming more routine (described in ‘Identifying relevant glycan structures or “glycotypes”‘). These methods have elucidated important differences in O- and N-glycan structures between mouse brain regions and between sexes, suggesting that distinct glycan repertoires are expressed in different tissues and are subject to tight regulation (Williams et al., 2022). The localization of glycans in tissues from cancer biopsies has also been studied using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) MS imaging (MSI) (Powers et al., 2013). This approach has been most extensively applied to N-glycans, which are first released from the tissue using PNGase F and then identified in specific regions targeted by the ionizing laser. Such methods have also been applied to formalin-fixed paraffin embedded tissue samples, allowing new analyses for archival tissue samples (Powers et al., 2014). The continued development of these and other related technologies, such as single-cell proteomics and mass cytometry imaging, should enable glycans to be added to multi-omics analyses, providing an essential, missing element for understanding cellular complexity and function. Modulating glycans to probe function: connecting ‘glycotype’ to phenotype The ability to selectively modulate glycans on glycoconjugates is critical for relating specific glycan structures and glycosylation events to particular cellular phenotypes. Both loss-of-function and gain-of-function approaches are available for manipulating glycans. For example, pharmacological inhibitors, glycosidase treatment, or genetic knockdown/knockout approaches often lead to loss of function by removing specific cellular glycans. As these methods typically deplete entire families of glycan structures, the results can sometimes be difficult to interpret. Comparative studies in which multiple genes are systematically deleted within a single biosynthetic pathway can provide better control of glycan populations and be more informative. With the advent of facile gene knockout and editing by ZFNs, TALENs, and CRISPR/Cas9, libraries of defined mutants have been developed for O- and N-glycans (Narimatsu et al., 2019), GPI anchors (Liu et al., 2021), and GAGs (Chen et al., 2018; Qiu et al., 2018), significantly expanding the available toolkit of cell lines with simplified glycomes (Figure 6A). In addition to cell lines, this loss-of-function approach has been applied to more complex systems such as human organotypic skin models to understand the role of glycans in tissue formation (Dabelsteen et al., 2020). Genetic disruption approaches have also been used to generate secreted mucins with defined O-glycan composition as protein-based probes (Nason et al., 2021). Beyond targeting glycan biosynthetic pathways, genetic manipulation enables the interrogation of specific glycosylation events in various biological settings. For example, viral-mediated expression of a site-specific O-GlcNAc-deficient mutant of CREB in cultured neurons and murine brains in vivo revealed that activity-induced O-GlcNAcylation at Ser-40 modulates dendritic and axonal growth, as well as long-term memory consolidation (Rexach et al., 2012). In addition to alanine mutagenesis, specific O-GlcNAcylation sites on proteins have been mutated to Cys using CRISPR-Cas9 technologies. OGT-mediated GlcNAcylation still occurred at the mutated site, producing a hydrolytically resistant, structural mimic of O-GlcNAc (Gorelik et al., 2019). For example, S405C mutation of OGA led to hyper-S-GlcNAcylation and substantially reduced its cellular half-life, suggesting a role for this site-specific modification in regulating OGA stability. Knockout animal models have also been extensively used to study the roles of glycans and are particularly helpful for examining biological phenotypes that cannot be recapitulated in vitro. As reviewed elsewhere (Stanley, 2016), many constitutive GT knockouts in mice exhibit embryonic lethality, underscoring the importance of glycans for development and health and prompting the development of conditional knockout models. However, interpretation of knockout animal phenotypes can be complicated by the pleiotropic functions of glycans in vivo, whereby similar but nonequivalent animal models may affect only some of the pleiotropic roles of glycans (Häcker et al., 2005). In addition to disrupting glycan biosynthesis, more precise targeting of specific glycan-binding events in model organisms is possible using gene editing technologies. For example, HS GAGs were found to engage TIE1, an orphan receptor critical for vascular development and homeostasis (Griffin et al., 2021). However, knockout of either the glycan or receptor leads to embryonic lethality, preventing the study of HS-TIE1 interactions in the maturing vasculature. To address this, Cas9-targeted mutations were generated in the HS binding cleft of TIE1, allowing selective ablation of the interaction without loss of the protein or polysaccharide. These mutant animals showed aberrant vasculature and altered vascular survival signaling, highlighting how the functional roles of individual glycan-protein interactions can be teased apart and linked to complex in vivo processes. Gain-of-function approaches have been developed to install a variety of O- and N-glycans onto proteins and cell surfaces. For example, N-glycans can be remodeled on cells using a two-step enzymatic approach like the remodeling of N-glycans on therapeutic antibodies (described in ‘Generating chemically defined glycans and glycoconjugates’) (Figure 6B). This approach greatly simplifies the overall structural complexity of N-glycosylation on the cell. N-glycan engineering has also been applied to impart new functions onto purified proteins. For example, N-glycan engineering of antibodies using CL methods has been used to attach cytotoxic molecules for antibody-drug conjugates (Tang et al., 2017; van Geel et al., 2015). O-glycan engineering approaches have exploited artificial scaffolds with defined O-glycan structures (Figure 6C). For example, synthetic glycopolymers displaying Neu5Ac-containing motifs and a terminal phospholipid anchor were incorporated into cell surfaces and shown to engage immune inhibitory Siglec receptors and inhibit NK cell-based cytotoxicity (Hudak et al., 2014). In another study, glycopolymers of varying lengths were exploited to modulate the thickness of the glycocalyx surrounding cancer cells (Paszek et al., 2014). These studies demonstrated that the physical properties of the glycocalyx can drive integrin clustering in cancer cells. A variety of cell-surface glycan engineering methods have also been developed for GAGs (Figure 6C). Using a similar lipid-anchored approach, the presentation of glycopolymers decorated with specific sulfated HS disaccharides was found to accelerate the differentiation of embryonic stem cells into neuronal precursors (Huang et al., 2014) and facilitate agrin-induced clustering of neurotransmitter receptors at the neuromuscular junction (Huang et al., 2018). In a complementary approach, commercially available CS polysaccharides were conjugated to a simple lipid anchor and incorporated into liposomes (Pulsipher et al., 2014). Addition of these functionalized GAG liposomes to primary neurons remodeled the cell surface and promoted signaling and outgrowth in a sulfation-dependent manner. Lipid anchoring of GAGs is often short-lived, with membrane-inserted probes showing half-lives on the order of several hours, but anchoring GAGs to a transmembrane HaloTag protein can produce longer-lived GAGs that are stably detected on the cell surface for over one week (Pulsipher et al., 2015). HS GAG engineering was shown to potentiate neural differentiation of embryonic stem cells (Pulsipher et al., 2015) and angiopoietin signaling (Griffin et al., 2021) based on the sulfation pattern of the displayed GAG. New methods enable the creation of semi-synthetic proteoglycans, wherein both the core protein and GAG composition can be systematically engineered and displayed on cell surfaces (O'Leary et al., 2022). Such approaches may shed light on the importance of the proteoglycan core protein and architecture in the biology of GAGs. Finally, intracellular O-GlcNAcylation can be modulated by targeting nanobody-fused OGT or OGA enzymes to O-GlcNAcylated substrates tagged with a nanobody recognition epitope (Figure 6D) (Ge et al., 2021; Ramirez et al., 2020). This approach allows for the control of O-GlcNAcylation stoichiometry on specific overexpressed proteins in cells and will likely be extendable to endogenous substrates using CRISPR/Cas9-mediated gene editing. Analysis of proteins containing multiple modification sites may be complicated as the exact sites and their relative glycosylation stoichiometries may differ using this strategy compared to native O-GlcNAcylation. Nevertheless, this emerging approach has been successfully used to show that O-GlcNAcylation alters the kinase activity of casein kinase 2α (Schwein et al., 2022), enabling new studies to understand O-GlcNAc function and its crosstalk with other post-translational modifications. Conclusions and Outlook Glycans are intricately involved in all facets of biology, and new breakthrough technologies have begun to reveal mechanisms by which these biomolecules regulate critical functions. In this Primer, we have outlined current methods to identify, characterize, monitor, and modulate glycans for a variety of applications. These approaches overcome the difficulties of glycan complexity and heterogeneity that have historically challenged glycoscience research, providing a clearer picture of the central features that underlie glycan activity. The glycoscience field is still in its exponential phase of growth, and future work will continue to expand the modern toolkit and improve our ability to decipher glycan function. A key remaining challenge for the future will be to rapidly produce larger collections of diverse, chemically pure glycan and glycoconjugate structures. As methods toward streamlined and automated syntheses advance, we envision that these molecules will one day be as accessible as tailored oligonucleotides and peptides are today. Progress in other fields such as protein structure prediction and directed evolution may help to develop new probes and enzymatic reagents with greater selectivity for specific glycan structures. New techniques to analyze data such as machine learning approaches combined with glycan array and other screening methods could propel novel discoveries by parsing information-rich experiments into actionable structure-function hypotheses. The application of new imaging and structural techniques such as super-resolution microscopy, micro-crystal electron diffraction, as well as cryo-electron microscopy and tomography should reveal key insights into the relationships between glycans and the structural organization of multiprotein complexes and subcellular compartments. Looking forward, the development of single-cell, spatial, and temporal technologies for quantifying glycans will greatly expand an understanding of their functions across multicellular systems. This information, combined with glycomics and glycoproteomics data, will add another crucial dimension to multi-omics experiments, providing important insights into cellular complexity and disease, as well as novel disease biomarkers. Although this Primer has focused specifically on tools to study mammalian glycans, a great diversity of carbohydrates is found across other kingdoms of life, including bacteria, fungi, and archaea. These microbial glycans can act as crucial regulators of host function and thus are key to understanding the human holobiont. The broad integration of glycoscience across other fields of biology will require access to readily usable tools and expertise. The focus on methods development and dissemination, epitomized by the NIH Common Fund Glycoscience Program and similar international efforts, will enable the democratization of these powerful technologies. When coupled with the proper selection of methods, controls, and data analysis, these tools can uncover new biological mechanisms and therapeutic strategies. The glycoscience community at large has a rich history of collaboration. With a broad repertoire of tools in hand along with the enthusiasm of the community, glycoscience research will continue to expand our understanding of the intricacies of mammalian biology. Acknowledgments This work is supported the Hope Funds for Cancer Research (HCFR-19-03-02, M.E.G.), the Melanoma Research Foundation (career development award, M.E.G.), and the National Institutes of Health (U01 GM116262, RF1-AG060540, and RF1-AG062324, L.C.H.W.). Figure 1. Chemical diversity of mammalian glycans. (A) The native assembly of two monosaccharides leads to eight potential disaccharides, depending on the regioselectivity (e.g., 1,2 versus 1,3) and stereoselectivity (α versus β) of the newly formed glycosidic bond. In contrast, only a single structure is produced from the native assembly of dinucleotides or dipeptides. (B) Mammalian glycans are composed of nine monosaccharides, which are pictorially represented by specific symbols from the Symbol Nomenclature for Glycans. (C) Glycosyltransferases utilize activated nucleotide sugar donors to transfer monosaccharide units onto growing glycan chains. Figure 2. Major glycan classes in mammals. (A) O-GlcNAcylation is the dynamic and reversible addition of N-acetylglucosamine to Ser and Thr residues on thousands of intracellular proteins. (B) O-GalNAc or mucin-like glycans are a broad class of O-linked extracellular glycans categorized by one of eight core structures that can be elaborated with a number of glycan antigens. (C) N-glycans are branched glycans attached to Asn residues of extracellular proteins and are categorized by the number and composition of their antennae branching from a conserved core structure. (D) Glycosaminoglycans (GAGs) are linear extracellular polysaccharides that can be sulfated at different hydroxyl and amine positions along the length of the glycan chain. Figure 3. Pharmacological inhibitors of glycosylation. (A) Protein O-GlcNAcylation can be blocked through the OGT inhibitors OSMI-1 and 5SGlcNAc, and the removal of O-GlcNAc can be targeted through the OGA inhibitor Thiamet G. (B) N-glycosylation can be broadly inhibited using tunicamycin, which inhibits the attachment of GlcNAc-1-phosphate to dolichol phosphate (Dol-P) by GlcNAc-1-phosphate transferase (GPT). (C) Glycans containing GlcNAc and GalNAc can be targeted using the glutamine mimics azaserine and DON, which target GFAT activity in hexosamine biosynthesis. (D) “Look-alike” mimics of monosaccharides can inhibit specific modifications such as fucosylation (2F-Fuc and 2-deoxyGal) and sialylation (3F-Neu5Ac) or act as decoys for GTs in GAG biosynthesis (Naph-Xyl). Figure 4. Glycomics and glycoproteomics workflows. (A) Glycomics, or the analysis of glycan composition, is accomplished through the enzymatic or chemical release of glycans, followed by chemical derivatization, purification, separation, and mass spectrometric characterization of glycan structures. (B) Glycoproteomics, or the analysis of protein glycosylation, is accomplished through the digestion of glycoproteins and enrichment of glycopeptides, followed by separation and mass spectrometric identification of glycosylated peptides. Figure 5. Methods to generate and modify glycans and glycoconjugates. (A) Automated methods for the chemical synthesis of oligosaccharides use a similar approach as solid-phase peptide or oligonucleotide synthesis. The glycan is elongated through a series of deprotection and coupling steps, followed by release from the resin. (B) The chemoenzymatic synthesis of an Arixtra biosimilar oligosaccharide employs multiple enzymatic and chemical steps to assemble and functionalize an HS heptasaccharide. (C) Semi-synthesis of O-GlcNAcylated α-synuclein utilizes synthetic glycopeptide fragments for expressed protein ligation to generate a library of specific protein glycoforms. (D) Metabolic labeling (ML) utilizes peracetylated (indicated by OAc) nonnatural monosaccharides that cross cell membranes, are deprotected (indicated by OH) by endogenous esterases, converted into nucleotide sugar donors, and then incorporated by GTs into glyconjugates. Nonnatural glycans with diazirine or alkyne functionalities are shown as representative examples. (E) Chemoenzymatic labeling (CL) employs exogenous GTs and nonnatural nucleotide sugar donors to modify specific glycan structures recognized by the GT. Nonnatural glycans with azide functionalities are shown. Figure 6. Methods to modulate glycans. (A) Genetic knockout of individual glycosyltransferases (GTs) and other glycan biosynthetic enzymes globally affects cellular glycan populations, leading to truncated glycan structures. (B) Chemoenzymatic remodeling using endoglycosidases (ENGase) simplifies N-glycan heterogeneity on the cell surface or on proteins. Installation of specific structures is accomplished using mutant ENGases or GTs that can install specific structures functionalized for further biorthogonal reactions. (C) Cell-surface engineering has been accomplished with both synthetic and naturally occurring glycopolymers anchored to the cell surface by lipids or proteins. Techniques can modulate the glycocalyx by changing its composition or thickness. (D) Targeted O-GlcNAc modification utilizes nanobody (nb)-fused OGT or OGA, which directs the enzymes to tagged target proteins. In the case of OGA, this approach is accomplished through a split OGA construct. Glycans are complex, ubiquitous, heterogeneous and play important roles in fundamental processes broadly across biology. Griffin and Hsieh-Wilson provide a critical overview and discuss current methods to identify, characterize, monitor and modulate mammalian glycans and how technological advances have overcome the difficulties of glycan complexity and heterogeneity to provide a clearer picture of central features of that underlie glycan activity. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Declaration of Interests The authors declare no competing interests. References Agrawal P , Fontanals-Cirera B , Sokolova E , Jacob S , Vaiana CA , Argibay D , Davalos V , McDermott M , Nayak S , Darvishian F , (2017). A systems biology approach identifies FUT8 as a driver of melanoma metastasis. Cancer Cell 31 , 804–819.e807. 10.1016/j.ccell.2017.05.007.28609658 Amon R , Grant OC , Leviatan Ben-Arye S , Makeneni S , Nivedha AK , Marshanski T , Norn C , Yu H , Glushka JN , Fleishman SJ , (2018). A combined computational-experimental approach to define the structural origin of antibody recognition of sialyl-Tn, a tumor-associated carbohydrate antigen. Sci Rep 8 ,10786. 10.1038/s41598-018-29209-9.30018351 Boittier ED , Burns JM , Gandhi NS , and Ferro V (2020). GlycoTorch Vina: docking designed and tested for glycosaminoglycans. J Chem Inf Model 60 , 6328–6343. 10.1021/acs.jcim.0c00373.33152249 Bojar D , Meche L , Meng G , Eng W , Smith DF , Cummings RD , and Mahal LK (2022). A useful guide to lectin binding: machine-learning directed annotation of 57 unique lectin specificities. ACS Chem Biol. 10.1021/acschembio.1c00689. Boltje TJ , Buskas T , and Boons GJ (2009). Opportunities and challenges in synthetic oligosaccharide and glycoconjugate research. Nat Chem 1 , 611–622. 10.1038/nchem.399.20161474 Bond MR , and Hanover JA (2013). O-GlcNAc cycling: a link between metabolism and chronic disease. Annu Rev Nutr 33 , 205–229. 10.1146/annurev-nutr-071812-161240.23642195 Boyce M , Carrico IS , Ganguli AS , Yu SH , Hangauer MJ , Hubbard SC , Kohler JJ , and Bertozzi CR (2011). Metabolic cross-talk allows labeling of O-linked beta-N-acetylglucosamine-modified proteins via the N-acetylgalactosamine salvage pathway. Proc Natl Acad Sci U S A 108 , 3141–3146. 10.1073/pnas.1010045108.21300897 Bradbury EJ , Moon LD , Popat RJ , King VR , Bennett GS , Patel PN , Fawcett JW , and McMahon SB (2002). Chondroitinase ABC promotes functional recovery after spinal cord injury. Nature 416 , 636–640. 10.1038/416636a.11948352 Briard JG , Jiang H , Moremen KW , Macauley MS , and Wu P (2018). Cell-based glycan arrays for probing glycan-glycan binding protein interactions. Nat Commun 9 , 880. 10.1038/s41467-018-03245-5.29491407 Brimble S , Wollaston-Hayden EE , Teo CF , Morris AC , and Wells L (2010). The Role of the O-GlcNAc modification in regulating eukaryotic gene expression. Curr Signal Transduct Ther 5 , 12–24. 10.2174/157436210790226465.25484640 Brown JM , Xia J , Zhuang B , Cho KS , Rogers CJ , Gama CI , Rawat M , Tully SE , Uetani N , Mason DE , (2012). A sulfated carbohydrate epitope inhibits axon regeneration after injury. Proc Natl Acad Sci U S A 109 , 4768–4773.10.1073/pnas.1121318109.22411830 Bullock S , Potter J , and Rose SP (1990). Effects of the amnesic agent 2-deoxygalactose on incorporation of fucose into chick brain glycoproteins. J Neurochem 54 , 135–142. 10.1111/j.1471-4159.1990.tb13293.x.2403432 Chang L , Chen YJ , Fan CY , Tang CJ , Chen YH , Low PY , Ventura A , Lin CC , Chen YJ , and Angata T (2017). Identification of Siglec ligands using a proximity labeling method. J Proteome Res 16 , 3929–3941.10.1021/acs.jproteome.7b00625.28899088 Chang PV , Dube DH , Sletten EM , and Bertozzi CR (2010). A strategy for the selective imaging of glycans using caged metabolic precursors. J Am Chem Soc 132 , 9516–9518. 10.1021/ja101080y.20568764 Chaubard JL , Krishnamurthy C , Yi W , Smith DF , and Hsieh-Wilson LC (2012). Chemoenzymatic probes for detecting and imaging fucose-alpha(1-2)-galactose glycan biomarkers. J Am Chem Soc 134 , 4489–4492. 10.1021/ja211312u.22339094 Chen YH , Narimatsu Y , Clausen TM , Gomes C , Karlsson R , Steentoft C , Spliid CB , Gustavsson T , Salanti A , Persson A , (2018). The GAGOme: a cell-based library of displayed glycosaminoglycans. Nat Methods 15 , 881–888.10.1038/s41592-018-0086-z.30104636 Cheng B , Tang Q , Zhang C , and Chen X (2021). Glycan labeling and analysis in cells and in vivo. Annu Rev Anal Chem 14 , 363–387. 10.1146/annurev-anchem-091620-091314. Chernykh A , Kawahara R , and Thaysen-Andersen M (2021). Towards structure-focused glycoproteomics. Biochem Soc Trans 49 , 161–186. 10.1042/BST20200222.33439247 Cheung ST , Miller MS , Pacoma R , Roland J , Liu J , Schumacher AM , and Hsieh-Wilson LC (2017). Discovery of a small-molecule modulator of glycosaminoglycan sulfation. ACS Chem Biol 12 , 3126–3133.10.1021/acschembio.7b00885.29099173 Chua JS , and Kuberan B (2017). Synthetic xylosides: probing the glycosaminoglycan biosynthetic machinery for biomedical applications. Acc Chem Res 50 , 2693–2705.10.1021/acs.accounts.7b00289.29058876 Chuh KN , Batt AR , Zaro BW , Darabedian N , Marotta NP , Brennan CK , Amirhekmat A , and Pratt MR (2017). The new chemical reporter 6-alkynyl-6-deoxy-GlcNAc reveals O-GlcNAc modification of the apoptotic caspases that can block the cleavage/activation of caspase-8. J Am Chem Soc 139 , 7872–7885.10.1021/jacs.7b02213.28528544 Chuh KN , Zaro BW , Piller F , Piller V , and Pratt MR (2014). Changes in metabolic chemical reporter structure yield a selective probe of O-GlcNAc modification. J Am Chem Soc 136 , 12283–12295. 10.1021/ja504063c.25153642 Clark PM , Dweck JF , Mason DE , Hart CR , Buck SB , Peters EC , Agnew J , and Hsieh-Wilson LC (2008). Direct in-gel fluorescence detection and cellular imaging of O-GlcNAc-modified proteins. J Am Chem Soc 130 , 11576–11577.10.1021/ja8030467.18683930 Clausen TM , Sandoval DR , Spliid CB , Pihl J , Perrett HR , Painter CD , Narayanan A , Majowicz SA , Kwong EM , McVicar RN , (2020). SARS-CoV-2 infection depends on cellular heparan sulfate and ACE2. Cell 183 , 1043–1057 e1015. 10.1016/j.cell.2020.09.033.32970989 Courtney AH , Puffer EB , Pontrello JK , Yang ZQ , and Kiessling LL (2009). Sialylated multivalent antigens engage CD22 in trans and inhibit B cell activation. Proc Natl Acad Sci U S A 106 , 2500–2505. 10.1073/pnas.0807207106.19202057 Dabelsteen S , Pallesen EMH , Marinova IN , Nielsen MI , Adamopoulou M , Romer TB , Levann A , Andersen MM , Ye Z , Thein D , (2020). Essential functions of glycans in human epithelia dissected by a CRISPR-Cas9-engineered human organotypic skin model. Dev Cell 54 , 669–684 e667.10.1016/j.devcel.2020.06.039.32710848 Dang K , Zhang W , Jiang S , Lin X , and Qian A (2020). Application of lectin microarrays for biomarker discovery. ChemistryOpen 9 , 285–300.10.1002/open.201900326.32154049 Danglad-Flores J , Leichnitz S , Sletten ET , Abragam Joseph A , Bienert K , Le Mai Hoang K , and Seeberger PH (2021). Microwave-assisted automated glycan assembly. J Am Chem Soc 143 , 8893–8901. 10.1021/jacs.1c03851.34060822 Debets MF , Tastan OY , Wisnovsky SP , Malaker SA , Angelis N , Moeckl LKR , Choi J , Flynn H , Wagner LJS , Bineva-Todd G , (2020). Metabolic precision labeling enables selective probing of O-linked N-acetylgalactosamine glycosylation. Proc Natl Acad Sci U S A 117 , 25293–25301. 10.1073/pnas.2007297117.32989128 Duksin D , and Mahoney WC (1982). Relationship of the structure and biological activity of the natural homologues of tunicamycin. J Biol Chem 257 , 3105–3109.7061468 Flynn RA , Pedram K , Malaker SA , Batista PJ , Smith BAH , Johnson AG , George BM , Majzoub K , Villalta PW , Carette JE , and Bertozzi CR (2021).Small RNAs are modified with N-glycans and displayed on the surface of living cells. Cell 184 , 3109–3124.e3122. 10.1016/j.cell.2021.04.023.34004145 Gao C , Wei M , McKitrick TR , McQuillan AM , Heimburg-Molinaro J , and Cummings RD (2019). Glycan microarrays as chemical tools for identifying glycan recognition by immune proteins. Front Chem 7 , 833. 10.3389/fchem.2019.00833.31921763 Ge Y , Ramirez DH , Yang B , D'Souza AK , Aonbangkhen C , Wong S , and Woo M (2021). Target protein deglycosylation in living cells by a nanobody-fused split O-GlcNAcase. Nat Chem Biol 17 , 593–600. 10.1038/s41589-021-00757-y.33686291 Giddens JP , Lomino JV , DiLillo DJ , Ravetch JV , and Wang LX (2018). Site-selective chemoenzymatic glycoengineering of Fab and Fc glycans of a therapeutic antibody. Proc Natl Acad Sci U S A 115 , 12023–12027. 10.1073/pnas.1812833115.30397147 Gimeno A , Valverde P , Arda A , and Jimenez-Barbero J (2020). Glycan structures and their interactions with proteins. A NMR view. Curr Opin Struct Biol 62 , 22–30.10.1016/j.sbi.2019.11.004.31835069 Gloster TM , Zandberg WF , Heinonen JE , Shen DL , Deng L , and Vocadlo DJ (2011). Hijacking a biosynthetic pathway yields a glycosyltransferase inhibitor within cells. Nat Chem Biol 7 , 174–181. 10.1038/nchembio.520.21258330 Gorelik A , Bartual SG , Borodkin VS , Varghese J , Ferenbach AT , and van Aalten DMF (2019). Genetic recoding to dissect the roles of site-specific protein O-GlcNAcylation. Nat Struct Mol Biol 26 , 1071–1077. 10.1038/s41594-019-0325-8.31695185 Gorelik A , and van Aalten DMF (2020). Tools for functional dissection of site-specific O-GlcNAcylation. RSC Chem Biol 1 , 98–109. 10.1039/d0cb00052c.34458751 Greve H , Cully Z , Blumberg P , and Kresse H (1988). Influence of chlorate on proteoglycan biosynthesis by cultured human fibroblasts. J Biol Chem 263 , 12886–12892.3417641 Griffin ME , Sorum AW , Miller GM , Goddard WA 3rd , and Hsieh-Wilson LC (2021). Sulfated glycans engage the Ang-Tie pathway to regulate vascular development. Nat Chem Biol 17 , 178–186. 10.1038/s41589-020-00657-7.33020664 Griffith AR , Rogers CJ , Miller GM , Abrol R , Hsieh-Wilson LC , and Goddard WA 3rd (2017). Predicting glycosaminoglycan surface protein interactions and implications for studying axonal growth. Proc Natl Acad Sci U S A 114 , 13697–13702.10.1073/pnas.1715093115.29229841 Guberman M , and Seeberger PH (2019). Automated glycan assembly: a perspective. J Am Chem Soc 141 , 5581–5592. 10.1021/jacs.9b00638.30888803 Häcker U , Nybakken K , and Perrimon N (2005). Heparan sulphate proteoglycans: the sweet side of development. Nat Rev Mol Cell Biol 6 , 530–541. 10.1038/nrm1681.16072037 Hahm HS , Schlegel MK , Hurevich M , Eller S , Schuhmacher F , Hofmann J , Pagel K , and Seeberger PH (2017). Automated glycan assembly using the Glyconeer 2.1 synthesizer. Proc Natl Acad Sci U S A 114 , E3385–E3389. 10.1073/pnas.1700141114.28396442 Han S , Collins BE , Bengtson P , and Paulson JC (2005). Homomultimeric complexes of CD22 in B cells revealed by protein-glycan cross-linking. Nat Chem Biol 1 , 93–97. 10.1038/nchembio713.16408005 Haukedal H , and Freude KK (2020). Implications of glycosylation in Alzheimer's disease. Front Neurosci 14 , 625348. 10.3389/fnins.2020.625348.33519371 Heindel DW , Koppolu S , Zhang Y , Kasper B , Meche L , Vaiana CA , Bissel SJ , Carter CE , Kelvin AA , Elaish M , (2020). Glycomic analysis of host response reveals high mannose as a key mediator of influenza severity. Proc Natl Acad Sci U S A 117 , 26926–26935. 10.1073/pnas.2008203117.33046650 Huang ML , Smith RA , Trieger GW , and Godula K (2014). Glycocalyx remodeling with proteoglycan mimetics promotes neural specification in embryonic stem cells. J Am Chem Soc 136 , 10565–10568. 10.1021/ja505012a.25019314 Huang ML , Tota EM , Lucas TM , and Godula K (2018). Influencing early stages of neuromuscular junction formation through glycocalyx engineering. ACS Chem Neurosci 9 , 3086–3093. 10.1021/acschemneuro.8b00295.30095249 Huang W , Giddens J , Fan SQ , Toonstra C , and Wang LX (2012). Chemoenzymatic glycoengineering of intact IgG antibodies for gain of functions. J Am Chem Soc 134 , 12308–12318. 10.1021/ja3051266.22747414 Hudak JE , Canham SM , and Bertozzi CR (2014). Glycocalyx engineering reveals a Siglec-based mechanism for NK cell immunoevasion. Nat Chem Biol 10 , 69–75. 10.1038/nchembio.1388.24292068 Joeh E , O'Leary T , Li W , Hawkins R , Hung JR , Parker CG , and Huang ML (2020). Mapping glycan-mediated galectin-3 interactions by live cell proximity labeling. Proc Natl Acad Sci U S A 117 , 27329–27338. 10.1073/pnas.2009206117.33067390 Joffrin AM , and Hsieh-Wilson LC (2020). Photoaffinity probes for the identification of sequence-specific glycosaminoglycan-binding proteins. J Am Chem Soc 142 , 13672–13676. 10.1021/jacs.0c06046.32786811 Kearney CJ , Vervoort SJ , Ramsbottom KM , Todorovski I , Lelliott EJ , Zethoven M , Pijpers L , Martin BP , Semple T , Martelotto L , (2021). SUGAR-seq enables simultaneous detection of glycans, epitopes, and the transcriptome in single cells. Sci Adv 7 . 10.1126/sciadv.abe3610. Kennedy DC , McKay CS , Legault MC , Danielson DC , Blake JA , Pegoraro AF , Stolow A , Mester Z , and Pezacki JP (2011). Cellular consequences of copper complexes used to catalyze bioorthogonal click reactions. J Am Chem Soc 133 , 17993–18001. 10.1021/ja2083027.21970470 Khidekel N , Arndt S , Lamarre-Vincent N , Lippert A , Poulin-Kerstien KG , Ramakrishnan B , Qasba PK , and Hsieh-Wilson LC (2003). A chemoenzymatic approach toward the rapid and sensitive detection of O-GlcNAc posttranslational modifications. J Am Chem Soc 125 , 16162–16163. 10.1021/ja038545r.14692737 Kiessling LL , and Grim JC (2013). Glycopolymer probes of signal transduction. Chem Soc Rev 42 , 4476–4491. 10.1039/c3cs60097a.23595539 Kjellen L , and Lindahl U (2018). Specificity of glycosaminoglycan-protein interactions. Curr Opin Struct Biol 50 , 101–108. 10.1016/j.sbi.2017.12.011.29455055 Kudelka MR , Antonopoulos A , Wang Y , Duong DM , Song X , Seyfried NT , Dell A , Haslam SM , Cummings RD , and Ju T (2016). Cellular O-glycome reporter/amplification to explore O-glycans of living cells. Nat Methods 13 , 81–86.10.1038/nmeth.3675.26619014 Kurtoglu M , Maher JC , and Lampidis TJ (2007). Differential toxic mechanisms of 2-deoxy-D-glucose versus 2-fluorodeoxy-D-glucose in hypoxic and normoxic tumor cells. Antioxid Redox Signal 9 , 1383–1390. 10.1089/ars.2007.1714.17627467 Laughlin ST , Baskin JM , Amacher SL , and Bertozzi CR (2008). In vivo imaging of membrane-associated glycans in developing zebrafish. Science 320 , 664–667. 10.1126/science.1155106.18451302 Li G , Li L , Tian F , Zhang L , Xue C , and Linhardt RJ (2015). Glycosaminoglycanomics of cultured cells using a rapid and sensitive LC-MS/MS approach. ACS Chem Biol 10 , 1303–1310. 10.1021/acschembio.5b00011.25680304 Li Q , Kailemia MJ , Merleev AA , Xu G , Serie D , Danan LM , Haj FG , Maverakis E , and Lebrilla CB (2019a). Site-specific glycosylation quantitation of 50 serum glycoproteins enhanced by predictive glycopeptidomics for improved disease biomarker discovery. Anal Chem 91 , 5433–5445. 10.1021/acs.analchem.9b00776.30882205 Li Q , Li Z , Duan X , and Yi W (2014). A tandem enzymatic approach for detecting and imaging tumor-associated Thomsen-Friedenreich antigen disaccharide. J Am Chem Soc 136 , 12536–12539. 10.1021/ja5054225.25157422 Li Q , Xie Y , Xu G , and Lebrilla CB (2019b). Identification of potential sialic acid binding proteins on cell membranes by proximity chemical labeling. Chem Sci 10 , 6199–6209. 10.1039/c9sc01360a.31360427 Li T , Liu L , Wei N , Yang JY , Chapla DG , Moremen KW , and Boons GJ (2019c). An automated platform for the enzyme-mediated assembly of complex oligosaccharides. Nat Chem 11 , 229–236. 10.1038/s41557-019-0219-8.30792508 Liu SS , Liu YS , Guo XY , Murakami Y , Yang G , Gao XD , Kinoshita T , and Fujita M (2021). A knockout cell library of GPI biosynthetic genes for functional studies of GPI-anchored proteins. Commun Biol 4 , 777. 10.1038/s42003-021-02337-1.34162996 Lopez Aguilar A , Briard JG , Yang L , Ovryn B , Macauley MS , and Wu P (2017). Tools for studying glycans: recent advances in chemoenzymatic glycan labeling. ACS Chem Biol 12 , 611–621. 10.1021/acschembio.6b01089.28301937 Ly M , Leach FE 3rd , Laremore TN , Toida T , Amster IJ , and Linhardt RJ (2011). The proteoglycan bikunin has a defined sequence. Nat Chem Biol 7 , 827–833.10.1038/nchembio.673.21983600 Machida T , Lang K , Xue L , Chin JW , and Winssinger N (2015). Site-Specific glycoconjugation of protein via bioorthogonal tetrazine cycloaddition with a genetically encoded trans-cyclooctene or bicyclononyne. Bioconjug Chem 26 , 802–806. 10.1021/acs.bioconjchem.5b00101.25897481 Mariethoz J , Alocci D , Gastaldello A , Horlacher O , Gasteiger E , Rojas-Macias M , Karlsson NG , Packer NH , and Lisacek F (2018). Glycomics@ExPASy: bridging the gap. Mol Cell Proteomics 17 , 2164–2176. 10.1074/mcp.RA118.000799.30097532 Marotta NP , Lin YH , Lewis YE , Ambroso MR , Zaro BW , Roth MT , Arnold B , Langen R , and Pratt MR (2015). O-GlcNAc modification blocks the aggregation and toxicity of the protein alpha-synuclein associated with Parkinson's disease. Nat Chem 7 , 913–920. 10.1038/nchem.2361.26492012 Mbua NE , Li X , Flanagan-Steet HR , Meng L , Aoki K , Moremen KW , Wolfert MA , Steet R , and Boons GJ (2013). Selective exo-enzymatic labeling of N-glycans on the surface of living cells by recombinant ST6Gal I. Angew Chem Int Ed Engl 52 , 13012–13015. 10.1002/anie.201307095.24129959 Mende M , Bednarek C , Wawryszyn M , Sauter P , Biskup MB , Schepers U , and Brase S (2016). Chemical synthesis of glycosaminoglycans. Chem Rev 116 , 8193–8255. 10.1021/acs.chemrev.6b00010.27410264 Mende M , Bordoni V , Tsouka A , Loeffler FF , Delbianco M , and Seeberger PH (2019). Multivalent glycan arrays. Faraday Discuss 219 , 9–32. 10.1039/c9fd00080a.31298252 Mikami T , and Kitagawa H (2013). Biosynthesis and function of chondroitin sulfate. Biochim Biophys Acta 1830 , 4719–4733. 10.1016/j.bbagen.2013.06.006.23774590 Minoshima F , Ozaki H , Odaka H , and Tateno H (2021). Integrated analysis of glycan and RNA in single cells. iScience 24 , 102882. 10.1016/j.isci.2021.102882.34401666 Möckl L , Pedram K , Roy AR , Krishnan V , Gustavsson AK , Dorigo O , Bertozzi CR , and Moerner WE (2019). Quantitative super-resolution microscopy of the mammalian glycocalyx. Dev Cell 50 , 57–72.e56. 10.1016/j.devcel.2019.04.035.31105009 Nagae M , and Yamaguchi Y (2018). Biophysical analyses for probing glycan-protein interactions. Adv Exp Med Biol 1104 , 119–147. 10.1007/978-981-13-2158-0_7.30484247 Narimatsu Y , Joshi HJ , Nason R , Van Coillie J , Karlsson R , Sun L , Ye Z , Chen YH , Schjoldager KT , Steentoft C , (2019). An atlas of human glycosylation pathways enables display of the human glycome by gene engineered cells. Mol Cell 75 , 394–407 e395. 10.1016/j.molcel.2019.05.017.31227230 Nason R , Bull C , Konstantinidi A , Sun L , Ye Z , Halim A , Du W , Sorensen DM , Durbesson F , Furukawa S , (2021). Display of the human mucinome with defined O-glycans by gene engineered cells. Nat Commun 12 , 4070. 10.1038/s41467-021-24366-4.34210959 Neelamegham S , Aoki-Kinoshita K , Bolton E , Frank M , Lisacek F , Lutteke T , O'Boyle N , Packer NH , Stanley P , Toukach P , (2019). Updates to the Symbol Nomenclature for Glycans guidelines. Glycobiology 29 , 620–624. 10.1093/glycob/cwz045.31184695 Ng BG , and Freeze HH (2018). Perspectives on glycosylation and its congenital disorders. Trends Genet 34 , 466–476. 10.1016/j.tig.2018.03.002.29606283 Nguyen SS , and Prescher JA (2020). Developing bioorthogonal probes to span a spectrum of reactivities. Nat Rev Chem 4 , 476–489. 10.1038/s41570-020-0205-0.34291176 Nieto PM (2018). The Use of NMR to Study Transient Carbohydrate-Protein Interactions. Front Mol Biosci 5 , 33. 10.3389/fmolb.2018.00033.29696146 Nivedha AK , Thieker DF , Makeneni S , Hu H , and Woods RJ (2016). Vina-Carb: improving glycosidic angles during carbohydrate docking. J Chem Theory Comput 12 , 892–901. 10.1021/acs.jctc.5b00834.26744922 O'Leary TR , Critcher M , Stephenson TN , Yang X , Hassan AA , Bartfield NM , Hawkins R , and Huang ML (2022). Chemical editing of proteoglycan architecture. Nat Chem Biol 18 , 634–642. 10.1038/s41589-022-01023-5.35551261 Oh YI , Sheng GJ , Chang SK , and Hsieh-Wilson LC (2013). Tailored glycopolymers as anticoagulant heparin mimetics. Angew Chem Int Ed Engl 52 , 11796–11799. 10.1002/anie.201306968.24123787 Oliveira T , Thaysen-Andersen M , Packer NH , and Kolarich D (2021). The hitchhiker's guide to glycoproteomics. Biochem Soc Trans 49 , 1643–1662.10.1042/BST20200879.34282822 Ortiz-Meoz RF , Jiang J , Lazarus MB , Orman M , Janetzko J , Fan C , Duveau DY , Tan ZW , Thomas CJ , and Walker S (2015). A small molecule that inhibits OGT activity in cells. ACS Chem Biol 10 , 1392–1397. 10.1021/acschembio.5b00004.25751766 Panza M , Pistorio SG , Stine KJ , and Demchenko AV (2018). Automated chemical oligosaccharide synthesis: novel approach to traditional challenges. Chem Rev 118 , 8105–8150. 10.1021/acs.chemrev.8b00051.29953217 Panza M , Stine KJ , and Demchenko AV (2020). HPLC-assisted automated oligosaccharide synthesis: the implementation of the two-way split valve as a mode of complete automation. Chem Commun 56 , 1333–1336. 10.1039/c9cc08876h. Parker CG , and Pratt MR (2020). Click chemistry in proteomic investigations. Cell 180 , 605–632. 10.1016/j.cell.2020.01.025.32059777 Paszek MJ , DuFort CC , Rossier O , Bainer R , Mouw JK , Godula K , Hudak JE , Lakins JN , Wijekoon AC , Cassereau L , (2014). The cancer glycocalyx mechanically primes integrin-mediated growth and survival. Nature 511 , 319–325. 10.1038/nature13535.25030168 Patnaik SK , and Stanley P (2006). Lectin-resistant CHO glycosylation mutants. Methods Enzymol 416 , 159–182. 10.1016/S0076-6879(06)16011-5.17113866 Pellegrini L , Burke DF , von Delft F , Mulloy B , and Blundell TL (2000). Crystal structure of fibroblast growth factor receptor ectodomain bound to ligand and heparin. Nature 407 , 1029–1034. 10.1038/35039551.11069186 Pinho SS , and Reis CA (2015). Glycosylation in cancer: mechanisms and clinical implications. Nat Rev Cancer 15 , 540–555. 10.1038/nrc3982.26289314 Pinzon Martin S , Seeberger PH , and Varon Silva D (2019). Mucins and pathogenic mucin-like molecules are immunomodulators during infection and targets for diagnostics and vaccines. Front Chem 7 , 710. 10.3389/fchem.2019.00710.31696111 Powers TW , Jones EE , Betesh LR , Romano PR , Gao P , Copland JA , Mehta AS , and Drake RR (2013). Matrix assisted laser desorption ionization imaging mass spectrometry workflow for spatial profiling analysis of N-linked glycan expression in tissues. Anal Chem 85 , 9799–9806. 10.1021/ac402108x.24050758 Powers TW , Neely BA , Shao Y , Tang H , Troyer DA , Mehta AS , Haab BB , and Drake RR (2014). MALDI imaging mass spectrometry profiling of N-glycans in formalin-fixed paraffin embedded clinical tissue blocks and tissue microarrays. PLoS One 9 , e106255. 10.1371/journal.pone.0106255.25184632 Pulsipher A , Griffin ME , Stone SE , Brown JM , and Hsieh-Wilson LC (2014). Directing neuronal signaling through cell-surface glycan engineering. J Am Chem Soc 136 , 6794–6797. 10.1021/ja5005174.24746277 Pulsipher A , Griffin ME , Stone SE , and Hsieh-Wilson LC (2015). Long-lived engineering of glycans to direct stem cell fate. Angew Chem Int Ed Engl 54 , 1466–1470. 10.1002/anie.201409258.25476911 Qin W , Qin K , Fan X , Peng L , Hong W , Zhu Y , Lv P , Du Y , Huang R , Han M , (2018). Artificial cysteine S-glycosylation induced by per-O-acetylated unnatural monosaccharides during metabolic glycan labeling. Angew Chem Int Ed Engl 57 , 1817–1820. 10.1002/anie.201711710.29237092 Qiu H , Shi S , Yue J , Xin M , Nairn AV , Lin L , Liu X , Li G , Archer-Hartmann SA , Dela Rosa M , (2018). A mutant-cell library for systematic analysis of heparan sulfate structure-function relationships. Nat Methods 15 , 889–899. 10.1038/s41592-018-0189-6.30377379 Raman R , Tharakaraman K , Sasisekharan V , and Sasisekharan R (2016). Glycan-protein interactions in viral pathogenesis. Curr Opin Struct Biol 40 , 153–162. 10.1016/j.sbi.2016.10.003.27792989 Ramirez DH , Aonbangkhen C , Wu HY , Naftaly JA , Tang S , O'Meara TR , and Woo CM (2020). Engineering a proximity-directed O-GlcNAc transferase for selective protein O-GlcNAcylation in cells. ACS Chem Biol 15 , 1059–1066. 10.1021/acschembio.0c00074.32119511 Rangel-Angarita V , and Malaker SA (2021). Mucinomics as the next frontier of mass spectrometry. ACS Chem Biol 16 , 1866–1883. 10.1021/acschembio.1c00384.34319686 Rawat M , Gama CI , Matson JB , and Hsieh-Wilson LC (2008). Neuroactive chondroitin sulfate glycomimetics. J Am Chem Soc 130 , 2959–2961. 10.1021/ja709993p.18275195 Reily C , Stewart TJ , Renfrow MB , and Novak J (2019). Glycosylation in health and disease. Nat Rev Nephrol 15 , 346–366. 10.1038/s41581-019-0129-4.30858582 Rexach JE , Clark PM , Mason DE , Neve RL , Peters EC , and Hsieh-Wilson LC (2012). Dynamic O-GlcNAc modification regulates CREB-mediated gene expression and memory formation. Nat Chem Biol 8 , 253–261. 10.1038/nchembio.770.22267118 Rexach JE , Rogers CJ , Yu SH , Tao J , Sun YE , and Hsieh-Wilson LC (2010). Quantification of O-glycosylation stoichiometry and dynamics using resolvable mass tags. Nat Chem Biol 6 , 645–651. 10.1038/nchembio.412.20657584 Riley NM , Bertozzi CR , and Pitteri SJ (2021). A pragmatic guide to enrichment strategies for mass spectrometry-based glycoproteomics. Mol Cell Proteomics 20 , 100029. 10.1074/mcp.R120.002277.33583771 Rillahan CD , Antonopoulos A , Lefort CT , Sonon R , Azadi P , Ley K , Dell A , Haslam SM , and Paulson JC (2012). Global metabolic inhibitors of sialyl- and fucosyltransferases remodel the glycome. Nat Chem Biol 8 , 661–668.10.1038/nchembio.999.22683610 Rillahan CD , and Paulson JC (2011). Glycan microarrays for decoding the glycome. Annu Rev Biochem 80 , 797–823. 10.1146/annurev-biochem-061809-152236.21469953 Robinson PV , Tsai CT , de Groot AE , McKechnie JL , and Bertozzi CR (2016). Glyco-seek: ultrasensitive detection of protein-specific glycosylation by proximity ligation polymerase chain reaction. J Am Chem Soc 138 , 10722–10725. 10.1021/jacs.6b03861.27454154 Rong J , Han J , Dong L , Tan Y , Yang H , Feng L , Wang QW , Meng R , Zhao J , Wang SQ , and Chen X (2014). Glycan imaging in intact rat hearts and glycoproteomic analysis reveal the upregulation of sialylation during cardiac hypertrophy. J Am Chem Soc 136 , 17468–17476. 10.1021/ja508484c.25313651 Ruhaak LR , Xu G , Li Q , Goonatilleke E , and Lebrilla CB (2018). Mass spectrometry approaches to glycomic and glycoproteomic analyses. Chem Rev 118 , 7886–7930. 10.1021/acs.chemrev.7b00732.29553244 Sakurai K , Hatai Y , and Okada A (2016). Gold nanoparticle-based multivalent carbohydrate probes: selective photoaffinity labeling of carbohydrate-binding proteins. Chem Sci 7 , 702–706. 10.1039/c5sc03275j.28791113 Sanders WJ , Gordon EJ , Dwir O , Beck PJ , Alon R , and Kiessling LL (1999). Inhibition of L-selectin-mediated leukocyte rolling by synthetic glycoprotein mimics. J Biol Chem 274 , 5271–5278. 10.1074/jbc.274.9.5271.10026133 Schlessinger J , Plotnikov AN , Ibrahimi OA , Eliseenkova AV , Yeh BK , Yayon A , Linhardt RJ , and Mohammadi M (2000). Crystal structure of a ternary FGF-FGFR-heparin complex reveals a dual role for heparin in FGFR binding and dimerization. Mol Cell 6 , 743–750. 10.1016/s1097-2765(00)00073-3.11030354 Schwein PA , Ge Y , Yang B , D'Souza A , Mody A , Shen D , and Woo CM (2022). Writing and erasing O-GlcNAc on casein kinase 2 alpha alters the phosphoproteome. ACS Chem Biol 17 , 1111–1121. 10.1021/acschembio.1c00987.35467332 Sheng GJ , Oh YI , Chang SK , and Hsieh-Wilson LC (2013). Tunable heparan sulfate mimetics for modulating chemokine activity. J Am Chem Soc 135 , 10898–10901. 10.1021/ja4027727.23879859 Smith DF , and Cummings RD (2014). Investigating virus-glycan interactions using glycan microarrays. Curr Opin Virol 7 , 79–87. 10.1016/j.coviro.2014.05.005.24995558 Sojitra M , Sarkar S , Maghera J , Rodrigues E , Carpenter EJ , Seth S , Ferrer Vinals D , Bennett NJ , Reddy R , Khalil A , (2021). Genetically encoded multivalent liquid glycan array displayed on M13 bacteriophage. Nat Chem Biol 17 , 806–816. 10.1038/s41589-021-00788-5.33958792 Song X , Ju H , Lasanajak Y , Kudelka MR , Smith DF , and Cummings RD (2016). Oxidative release of natural glycans for functional glycomics. Nat Methods 13 , 528–534. 10.1038/nmeth.3861.27135973 Stanley P (2016). What have we learned from glycosyltransferase knockouts in mice? J Mol Biol 428 , 3166–3182. 10.1016/j.jmb.2016.03.025.27040397 Sterner E , Flanagan N , and Gildersleeve JC (2016). Perspectives on anti-glycan antibodies gleaned from development of a community resource database. ACS Chem Biol 11 , 1773–1783. 10.1021/acschembio.6b00244.27220698 Stoeckius M , Hafemeister C , Stephenson W , Houck-Loomis B , Chattopadhyay PK , Swerdlow H , Satija R , and Smibert P (2017). Simultaneous epitope and transcriptome measurement in single cells. Nat Methods 14 , 865–868. 10.1038/nmeth.4380.28759029 Tanaka Y , and Kohler JJ (2008). Photoactivatable crosslinking sugars for capturing glycoprotein interactions. J Am Chem Soc 130 , 3278–3279. 10.1021/ja7109772.18293988 Tang F , Wang LX , and Huang W (2017). Chemoenzymatic synthesis of glycoengineered IgG antibodies and glycosite-specific antibody-drug conjugates. Nat Protoc 12 , 1702–1721. 10.1038/nprot.2017.058.28749929 Tang SL , and Pohl NLB (2016). Automated fluorous-assisted solution-phase synthesis of beta-1,2-, 1,3-, and 1,6-mannan oligomers. Carbohydr Res 430 , 8–15. 10.1016/j.carres.2016.03.025.27155895 Tarentino AL , and Plummer TH Jr. (1994). Enzymatic deglycosylation of asparagine-linked glycans: purification, properties, and specificity of oligosaccharidecleaving enzymes from Flavobacterium meningosepticum. Methods Enzymol 230 , 44–57. 10.1016/0076-6879(94)30006-2.8139511 Trinidad JC , Barkan DT , Gulledge BF , Thalhammer A , Sali A , Schoepfer R , and Burlingame AL (2012). Global identification and characterization of both O-GlcNAcylation and phosphorylation at the murine synapse. Mol Cell Proteomics 11 , 215–229. 10.1074/mcp.O112.018366.22645316 Vallet SD , Clerc O , and Ricard-Blum S (2021). Glycosaminoglycan-protein interactions: the first draft of the glycosaminoglycan interactome. J Histochem Cytochem 69 , 93–104. 10.1369/0022155420946403.32757871 van Geel R , Wijdeven MA , Heesbeen R , Verkade JM , Wasiel AA , van Berkel SS , and van Delft FL (2015). Chemoenzymatic conjugation of toxic payloads to the globally conserved N-glycan of native mAbs provides homogeneous and highly efficacious antibody-drug conjugates. Bioconjug Chem 26 , 2233–2242. 10.1021/acs.bioconjchem.5b00224.26061183 van Kuppevelt TH , Oosterhof A , Versteeg EMM , Podhumljak E , van de Westerlo EMA , and Daamen WF (2017). Sequencing of glycosaminoglycans with potential to interrogate sequence-specific interactions. Sci Rep 7 , 14785. 10.1038/s41598-017-15009-0.29093576 Varki A (2017). Biological roles of glycans. Glycobiology 27 , 3–49.10.1093/glycob/cww086.27558841 Vilen Z , Joeh E , Critcher M , Parker CG , and Huang ML (2021). Proximity tagging identifies the glycan-mediated glycoprotein interactors of galectin-1 in muscle stem cells. ACS Chem Biol 16 , 1994–2003. 10.1021/acschembio.1c00313.34181849 Wang H , and Mooney DJ (2020). Metabolic glycan labelling for cancer-targeted therapy. Nat Chem 12 , 1102–1114. 10.1038/s41557-020-00587-w.33219365 Wang H , Wang R , Cai K , He H , Liu Y , Yen J , Wang Z , Xu M , Sun Y , Zhou X , (2017). Selective in vivo metabolic cell-labeling-mediated cancer targeting. Nat Chem Biol 13 , 415–424. 10.1038/nchembio.2297.28192414 Wang P , Dong S , Shieh JH , Peguero E , Hendrickson R , Moore MAS , and Danishefsky SJ (2013a). Erythropoietin derived by chemical synthesis. Science 342 , 1357–1360. 10.1126/science.1245095.24337294 Wang Z , Chinoy ZS , Ambre SG , Peng W , McBride R , de Vries RP , Glushka J , Paulson JC , and Boons GJ (2013b). A general strategy for the chemoenzymatic synthesis of asymmetrically branched N-glycans. Science 341 , 379–383. 10.1126/science.1236231.23888036 Wen L , Edmunds G , Gibbons C , Zhang J , Gadi MR , Zhu H , Fang J , Liu X , Kong Y , and Wang PG (2018). Toward automated enzymatic synthesis of oligosaccharides. Chem Rev 118 , 8151–8187. 10.1021/acs.chemrev.8b00066.30011195 Wen L , Zheng Y , Jiang K , Zhang M , Kondengaden SM , Li S , Huang K , Li J , Song J , and Wang PG (2016). Two-Step chemoenzymatic detection of N-acetylneuraminic acid-alpha(2-3)-galactose glycans. J Am Chem Soc 138 , 11473–11476. 10.1021/jacs.6b07132.27554522 Wibowo A , Peters EC , and Hsieh-Wilson LC (2014). Photoactivatable glycopolymers for the proteome-wide identification of fucose-alpha(1-2)-galactose binding proteins. J Am Chem Soc 136 , 9528–9531. 10.1021/ja502482a.24937314 Williams SE , Noel M , Lehoux S , Cetinbas M , Xavier RJ , Sadreyev RI , Scolnick EM , Smoller JW , Cummings RD , and Mealer RG (2022). Mammalian brain glycoproteins exhibit diminished glycan complexity compared to other tissues. Nat Commun 13 , 275. 10.1038/s41467-021-27781-9.35022400 Woo CM , Lund PJ , Huang AC , Davis MM , Bertozzi CR , and Pitteri SJ (2018). Mapping and quantification of over 2000 O-linked glycopeptides in activated human T cells with isotope-targeted glycoproteomics (Isotag). Mol Cell Proteomics 17 , 764–775. 10.1074/mcp.RA117.000261.29351928 Wright TH , Bower BJ , Chalker JM , Bernardes GJ , Wiewiora R , Ng WL , Raj R , Faulkner S , Vallee MR , Phanumartwiwath A , (2016). Posttranslational mutagenesis: A chemical strategy for exploring protein side-chain diversity. Science 354 . 10.1126/science.aag1465. Wu ZL , Huang X , Burton AJ , and Swift KA (2016). Probing sialoglycans on fetal bovine fetuin with azido-sugars using glycosyltransferases. Glycobiology 26 , 329–334. 10.1093/glycob/cwv109.26589574 Wu ZL , Person AD , Anderson M , Burroughs B , Tatge T , Khatri K , Zou Y , Wang L , Geders T , Zaia J , and Sackstein R (2018). Imaging specific cellular glycan structures using glycosyltransferases via click chemistry. Glycobiology 28 , 69–79. 10.1093/glycob/cwx095.29186441 Xia L , and Gildersleeve JC (2015). The glycan array platform as a tool to identify carbohydrate antigens. Methods Mol Biol 133l , 27–40. 10.1007/978-1-4939-2874-3_3. Xie R , Dong L , Du Y , Zhu Y , Hua R , Zhang C , and Chen X (2016). In vivo metabolic labeling of sialoglycans in the mouse brain by using a liposome-assisted bioorthogonal reporter strategy. Proc Natl Acad Sci U S A 113 , 5173–5178.10.1073/pnas.1516524113.27125855 Xu C , and Ng DT (2015). Glycosylation-directed quality control of protein folding. Nat Rev Mol Cell Biol 16 , 742–752. 10.1038/nrm4073.26465718 Xu D , and Esko JD (2014). Demystifying heparan sulfate-protein interactions. Annu Rev Biochem 83 , 129–157. 10.1146/annurev-biochem-060713-035314.24606135 Xu Y , Masuko S , Takieddin M , Xu H , Liu R , Jing J , Mousa SA , Linhardt RJ , and Liu J (2011). Chemoenzymatic synthesis of homogeneous ultralow molecular weight heparins. Science 334 , 498–501. 10.1126/science.1207478.22034431 Yamada I , Shiota M , Shinmachi D , Ono T , Tsuchiya S , Hosoda M , Fujita A , Aoki NP , Watanabe Y , Fujita N , (2020). The GlyCosmos Portal: a unified and comprehensive web resource for the glycosciences. Nat Methods 17 , 649–650. 10.1038/s41592-020-0879-8.32572234 Yang X , and Qian K (2017). Protein O-GlcNAcylation: emerging mechanisms and functions. Nat Rev Mol Cell Biol 18 , 452–465. 10.1038/nrm.2017.22.28488703 Yi W , Clark PM , Mason DE , Keenan MC , Hill C , Goddard WA 3rd , Peters EC , Driggers EM , and Hsieh-Wilson LC (2012). Phosphofructokinase 1 glycosylation regulates cell growth and metabolism. Science 337 , 975–980.10.1126/science.1222278.22923583 York WS , Mazumder R , Ranzinger R , Edwards N , Kahsay R , Aoki-Kinoshita KF , Campbell MP , Cummings RD , Feizi T , Martin M , (2020). GlyGen: computational and informatics resources for glycoscience. Glycobiology 30 , 72–73. 10.1093/glycob/cwz080.31616925 Young G , Hundt N , Cole D , Fineberg A , Andrecka J , Tyler A , Olerinyova A , Ansari A , Marklund EG , Collier MP , (2018). Quantitative mass imaging of single biological macromolecules. Science 360 , 423–427. 10.1126/science.aar5839.29700264 Yu H , and Chen X (2016). One-pot multienzyme (OPME) systems for chemoenzymatic synthesis of carbohydrates. Org Biomol Chem 14 , 2809–2818. 10.1039/c6ob00058d.26881499 Yu SH , Boyce M , Wands AM , Bond MR , Bertozzi CR , and Kohler JJ (2012). Metabolic labeling enables selective photocrosslinking of O-GlcNAc-modified proteins to their binding partners. Proc Natl Acad Sci U S A 109 , 4834–4839. 10.1073/pnas.1114356109.22411826 Yu SH , Zhao P , Sun T , Gao Z , Moremen KW , Boons GJ , Wells L , and Steet R (2016). Selective exo-enzymatic labeling detects increased cell surface sialoglycoprotein expression upon megakaryocytic differentiation. J Biol Chem 291 , 3982–3989. 10.1074/jbc.M115.700369.26733198 Yuan K , Listinsky CM , Singh RK , Listinsky JJ , and Siegal GP (2008). Cell surface associated alpha-L-fucose moieties modulate human breast cancer neoplastic progression. Pathol Oncol Res 14 , 145–156. 10.1007/s12253-008-9036-x.18553163 Yuzwa SA , Macauley MS , Heinonen JE , Shan X , Dennis RJ , He Y , Whitworth GE , Stubbs KA , McEachern EJ , Davies GJ , and Vocadlo DJ (2008). A potent mechanism-inspired O-GlcNAcase inhibitor that blocks phosphorylation of tau in vivo. Nat Chem Biol 4 , 483–490. 10.1038/nchembio.96.18587388 Zeng Y , Ramya TN , Dirksen A , Dawson PE , and Paulson JC (2009). High-efficiency labeling of sialylated glycoproteins on living cells. Nat Methods 6 , 207–209.10.1038/nmeth.1305.19234450 Zhang Q , Li Z , and Song X (2020). Preparation of complex glycans from natural sources for functional study. Front Chem 8 , 508. 10.3389/fchem.2020.00508.32719769 Zheng T , Jiang H , Gros M , del Amo DS , Sundaram S , Lauvau G , Marlow F , Liu Y , Stanley P , and Wu P (2011). Tracking N-acetyllactosamine on cell-surface glycans in vivo. Angew Chem Int Ed Engl 50 , 4113–4118. 10.1002/anie.201100265.21472942 Zhou JY , and Cobb BA (2021). Glycans in immunologic health and disease. Annu Rev Immunol 39 , 511–536. 10.1146/annurev-immunol-101819-074237.33577348 Zhu Y , Yan M , Lasanajak Y , Smith DF , and Song X (2018). Large scale preparation of high mannose and paucimannose N-glycans from soybean proteins by oxidative release of natural glycans (ORNG). Carbohydr Res 464 , 19–27. 10.1016/j.carres.2018.05.002.29803109

PMC009xxxxxx/PMC9339482.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7503056 4435 J Am Chem Soc J Am Chem Soc Journal of the American Chemical Society 0002-7863 1520-5126 35849554 9339482 10.1021/jacs.2c05491 NIHMS1823700 Article Enzyme Responsive Rigid-Rod Aromatics Target “Undruggable” Phosphatases to Kill Cancer Cells in Mimetic Bone Microenvironment Yi Meihui http://orcid.org/0000-0002-1912-1043 1 Wang Fengbin http://orcid.org/0000-0003-1008-663X 2 Tan Weiyi http://orcid.org/0000-0001-5316-8278 1 Hsieh Jer-Tsong 3 Egelman Edward H. http://orcid.org/0000-0003-4844-5212 2 Xu Bing http://orcid.org/0000-0002-4639-387X 1 1 Department of Chemistry, Brandeis University, 415 South Street, Waltham, MA 02453, USA 2 Department of Biochemistry and Molecular Genetics, University of Virginia, Charlottesville, VA 22908, USA 3 Department of Urology, Southwestern Medical Center, University of Texas, Dallas, TX 75235, USA Corresponding Author: Bing Xu − Department of Chemistry, Brandeis University, 415 South Street, Waltham, MA 02453, USA; bxu@brandeis.edu, Edward Egelman- Department of Biochemistry and Molecular Genetics, University of Virginia, Charlottesville, VA 22908, USA; egelman@virginia.edu. Jer-Tsong Hsieh- Department of Urology, Southwestern Medical Center, University of Texas, Dallas, TX 75235, USA; JT.Hsieh@utsouthwestern.edu 14 7 2022 27 7 2022 18 7 2022 27 7 2023 144 29 1305513059 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Bone metastasis remains a challenge in cancer treatment. Here we show that enzymatic responsive rigid-rod aromatics, acting as the substrates of “undruggable” phosphatases, to kill cancer cells in mimetic bone microenvironment. By phosphorylation and conjugating nitrobenzoxadiazole (NBD) to hydroxybiphenylcarboxylate (BP), we obtained pBP-NBD (1P) as a substrate of both acid and alkaline phosphatases. 1P effectively kill both metastatic castration-resistant prostate cancer cells (mCRPCs) and osteoblast mimic cells in their co-culture. 1P enter Saos2 almost instantly to target the endoplasmic reticulum (ER) of the cells. Co-culturing with Saos2 cells boosts the cellular uptake of 1P by mCRPCs. Cryo-EM reveals the nanotube structures of both 1P (2.4 Å resolution, pH 5.6) and 1 (2.2 Å resolution, pH 7.4). The helical packing of both nanotubes is identical, held together by strong pi-stackings interactions. Besides reporting the atomistic structure of nanotubes formed by the assembly of rigid-rod aromatics, this work expands the pool of molecules for designing EISA substrates that selectively target TME. Graphical Abstract pmcThis communication reports the first example of enzyme-instructed self-assembly (EISA) of rigid-rod aromatics for killing cancer cells in mimetic bone microenvironment. Despite considerable progress in cancer therapy, tumor metastasis still causes most of cancer-related death.1 For example, approximately 90% of patients who die of prostate cancer have bone metastases.2 Bone metastases are a vicious self-reinforcing cycle in tumor microenvironment (TME), where cancer cells promote the differentiation of osteoclasts and osteoblasts, and the increased bone turnover releases cytokines to benefit metastatic cancer cells. Moreover, bone is also a site for further spread of other metastasis.3 The current treatment of bone metastasis, Radium-223 (Ra-223)4,5 is only palliative because the acceptable radiation dosage of Radium-223 is limited. Thus, there is an urgent need to develop novel approaches for killing cancer cells in bone metastasis sites or TME with minimal side effects. Inspired by the dual targeting mode-of-action of Radium-223, we are aiming to develop non-radioactive small molecules as a single agent that self-assemble in-situ or on-site to kill both metastatic castration-resistant prostate cancer (mCRPC) and osteoblast cells in TME. A prominent feature of osteoblastic mCRPC at the TME is that mCRPC and osteoblasts overexpress prostatic acid phosphatase (PAP)6 and alkaline phosphatase (ALP)7, respectively. However, developing inhibitors of PAP and ALP to treat mCRPC has been unsuccessful for several reasons: (i) PAP acts as tumor suppressor,8 (ii) inhibitor of ALP is unable to inhibit cancer cells,9 and (iii) most phosphatases previously are considered as undruggable targets.10–11 Thus, we decide to explore EISA, which kills cancer cells by enzymatic reaction and self-assembly,12 for developing therapeutics for osteoblastic mCRPC. EISA is particularly attractive because EISA kills cancer cells without inhibiting the targeted enzymes. Moreover, ALP is metabolically inert in serum,13 a subtlety allowing EISA to occur at TME. Recently, a variety of small molecule substrates, including peptides,14–18 carbohydrates,19 and lipids,20–21 of ALP-based EISA are able to induce death of the cancer cells that overexpressing ALP because EISA allows “on-site” generating nanostructures from small molecules.22 Although peptide-based substrates are the most explored among these building blocks up-to-date, their efficacy for killing mCRPC still remains to be improved.23 Thus, we choose to examine other nonpeptidic molecules as the self-assembling building blocks for EISA against mCRPC. Based on aggregate advisors24 and rigid-rod molecules for self-assembly, we choose a rigid-rod aromatic molecule, biphenyl, as the core motif for developing EISA substrates of PAP and ALP. Biphenyl has found various applications in liquid crystals,25–28 dendrimers,29 long-range self-assembly,30 amphiphiles,31 peptides.32 We recently found that biphenyl enable rapid enzymatic self-assembly and hydrogelation of peptides.33 But biphenyl has yet to be explored for EISA without involving peptides. Based on the above rationale, we phosphorylated the hydroxyl and conjugated a fluorophore (nitrobenzoxadiazole (NBD)) at the carboxylic end of hydroxybiphenylcarboxylate (BP) or hydroxyterphenylcarboxylate (TP) to produce pBP-NBD (1P) or pTP-NBD (2P), respectively. As a substrate of both PAP and ALP, 1P effectively kills both metastatic castration-resistant prostate cancer cells (mCRPCs) (VCaP or PC3) and osteoblast mimic cells (Saos2) in their co-culture. Fluorescent imaging shows that 1P enters Saos2 almost instantly to target the endoplasmic reticulum (ER) of the cells and that Saos2 in the co-culture boosts the cellular uptake of 1P or 2P by VCaP or PC3 cells. Cryo-EM reveals that the 1 (from dephosphorylating 1P at physiological pH) or 1P (at acidic pH) self-assembles to form cones with C7 symmetry, and the supramolecular cones stack helically with a 3.5 Å rise to form nanotubes. Besides reporting the atomistic structure of nanotubes formed by the assembly of rigid-rod aromatics, this work, using enzymatic reactions to form intracellular nanotubes of small rigid-rod molecules, expands the pool of molecules for designing EISA substrates that selectively target TME. Scheme 1 shows the structures and the key synthetic step for making, 1P, 2P, and 3P, which have biphenyl, terphenyl, and phenyl motifs, respectively. While BP and hydroxylphenylcarboxylate are commercially available, TP is produced in almost quantitative yield from 4-bromo-4’-hydroxylbiphenyl and 4-carboxyphenylboronic acid by palladium-catalyzed cross coupling.34 Mixing BP or hydroxylphenylcarboxylate with phosphorus pentachloride directly under heat or reacting TP with phosphorous oxychloride in pyridine produces the phosphorylated aromatic carboxylic acids. After getting all three phosphorylated compounds, we conjugated them with NBD-ethylene diamine to generate 1P, 2P, and 3P. As shown by transmission electron microscopy (TEM), adding ALP to solution of 1P (500 μM and pH7.4) transforms the amorphous aggregates of 1P to ~7 nm diameter nanotubes of 1 (Figure 1). At pH 5.6, 1P (50 μM) forms nanotubes like those of 1. These results indicate 1P self-assembling more readily at lower pH because protonation decreases phosphate repulsion. Adding PAP in 1P solution (50 μM and at pH 5.6) results in more nanotubes (Figure 1). Adding ALP to the solution of 2P also switches the aggregates of 2P into nanotubes with similar diameters (Figure S2). At pH 5.6, 2P (50 μM) transforms from aggregates into twisted nanoribbons (Figure S1). 3P exists as amorphous aggregates both at pH 7.4 or pH 5.6 and rarely forms nanotubes after adding ALP or PAP (Figure S1), indicating that noncovalent interactions from phenyl are insufficient for 3P or 3 to self-assemble into nanotubes. We tested 1P for its activity against six different cell lines, Saos2, SJSA1, VCaP, PC3, HepG2 and PNT1A (Figures 2A and S4). 1P shows the first day IC50 around 20.9 μM against Saos2 cells. The IC50 values of 1P against SJSA1 and Saos2 are comparable, except day 1. While 1P significantly inhibit VCaP and PC3 cells from day 2 and day 3, respectively, 1P is rather compatible with HepG2 and PNT1A even at day 3, with IC50 around 108 μM and 89 μM. Even though 2P showed potent activity against Saos2 and SJSA1, it hardly inhibits VCaP, PC3, HepG2 and PNT1A (Figures S3 and S4). Compared to 1P and 2P, 1 and 2 is more toxic to the normal prostate cells, PNT1A, indicating that phosphorylation enhances the selectivity for targeting cancer cells. ALP inhibitor (DQB) reduces the cytotoxicity of 1P and 2P against Saos2 or SJSA1 cells (Figure S5), indicating that ALP-catalyzed EISA of 1P and 2P contributes to their cytotoxicity. The low cytotoxicity of 1P and 2P towards hepatocyte (HepG2) indicates low toxicity of 1P and 2P to liver. 3P hardly inhibits these cells (Figure S2B), consistent with the poor self-assembling ability of 3. M-β-CD, an inhibitor for caveolae mediated endocytosis, rescues Saso2 and SJSA1 treated by 1P or 2P, suggesting that 1P or 2P enters the cells via endocytosis (Figure S6). As shown in Figure 2B, 1P enters Saos2 or SJSA1 within 1 minute and accumulates inside the cells. On the other hand, 1P accumulates in VCaP or PC3 cells slowly that visible fluorescence from 1 emerges in VCaP and PC3 cells after 30 and 60 minutes, respectively. The inhibitory activity of 1P correlates well with the rate of cellular uptake, except in the case of HepG2. Although 1P enters HepG2 faster (Figure S7) than entering PC3, 1P hardly inhibits HepG2, agreeing with detoxification function of hepatocytes.35 2P exhibits a similar trend as that of 1P to enter Saos2, SJSA1, VCaP, and PC3 cells (Figure S9). Agreeing with its cell compatibility, 3P enters Saos2 much slower than 1P or 2P does (Figure S8). For Saos2 and SJSA1 cells, the distribution of 1P in cytosol resembles to that of ER (Figures 2C and S10). Co-staining 1P with ER-tracker, Lyso-tracker, or Mito-tracker in Saos2 or SJSA1 cells shows that the fluorescence from 1P mostly overlaps with the fluorescence of ER-tracker, indicating that 1P, after entering cells and being dephosphorylated, specifically localizes in ER of Saos2 and SJSA1. The ER accumulation agrees with that phosphobipenyl is a substrate of PTP1B36, which mainly locates at ER37. 2P localizes more in ER of Saos2 and SJSA1 cells than in lysosomes and mitochondria, and forms denser fluorescent puncta in Saos2 and SJSA1 than 1P does (Figures S11 and S12). The fluorescence of 1P in VCaP and PC3 cells overlap more with that of Lysotracker, indicating that 1P mainly undergoes dephosphorylation in the lysosomes. This result agrees with that PAP localized in lysosome.38 We cocultured the mCRPC cells (e.g., VCaP or PC3) and the osteoblast mimic cells (Saos2) in a 1:1 ratio to create a mimetic bone microenvironment and tested the efficacy of 1P. As shown in Figure 3A, being incubated with 1P for 24 h, the viability of Saos2 or VCaP cells is 1.6% and 41.3%, respectively. The cell viability of the coculture of Saos2 and VCaP is 1.9% after adding 1P in the coculture. Similarly, 2P inhibits VCaP more effectively in the coculture. Moreover, Figure 3B shows the viability of PC3 cells, being incubated with 1P or 2P in the co-culture of PC3 and Saos2, is significantly lower than that of PC3 cells alone. These results suggest that Saos2 in the coculture significantly enhances the inhibitory activity of 1P or 2P against VCaP or PC3 cells. Fluorescence imaging reveals the cellular uptake of 1P by VCaP or PC3 cells (Figures 3C, D, S13–15). The fluorescence in the VCaP or PC3 cells increases faster in the coculture than being cultured alone. These results indicate that Saos2 cells, in the co-culture, boost the cellular uptake of 1P or 2P by VCaP and PC3 cells, thus leading to more effective inhibition of VCaP or PC3 cells. We used cryo-EM and determined the high-resolution nanotube structures of both 1 at pH 7.4 and 1P at pH 5.6 (Figures S16A and S17A). Possible helical symmetries were calculated from the averaged power spectrum of boxed filaments (Figures S16B and S17B) and the correct ones were found39. The nanotubes of 1 and 1P have almost identical diameters (Figure 4A–B) and helical symmetries (Figure 4C–D, Table S1), with aromatic rings held together by extensive pi-stacking interactions (Figure 4E). The NBD motifs points to the center and the biphenyl groups constitute the periphery of the nanotubes. The ability of 1P to form nanotubes at pH 5.6 is consistent with the lysosomal accumulation of 1P (Figure 2A, C) and its increased activities against VCaP and PC3 cells after 48 h. In summary, this work reports a new class of EISA substrates to form nanotubes for effectively killing osteoblast model cells and mCRPCs in mimetic bone TME. The toxicity of 1 and 2 to PNT1A imply that nanotubes of 1 or 2 likely results in cell death. It illustrates the dual targeting mode-of-action of EISA substrates, which promising to kill both cancer and osteoblast cells at bone TME. Although the detailed mechanism of action of EISA of 1P or 2P in the co-culture remains to be elucidated, using the enzymatic feature of TME to boost cellular uptake of EISA substrates in mCRPCs promising a new way to target TME of mCRPCs. The approach reported here should be applicable to other aggregation-prone small molecules or drug candidates,24, 40 which may lead to more effective EISA substrates for targeting TME of other metastatic cancers. Supplementary Material Supporting info ACKNOWLEDGMENT This work is partially supported by NIH CA142746 (B.X.) and CA252364 (B.X.), GM122510 (E.H.E.), K99GM138756 (F.W.) and NSF DMR-2011846 (B.X.). Figure 1. TEM of 1P (500 μM, pH 7.4) before and after adding ALP (0.1 U/mL) for 24 h and of 1P (50 μM, pH 5.6) before and after adding PAP (0.1 U/mL) for 24 h. Scale bar: 100 nm. Figure 2. (A) IC50 of 1P incubated with Saos2, SJSA1, VCaP, PC3, and HepG2 cells for 24 h. (B) Instant cell entry of 1P (100 μM). (C) Intracellular distribution of 1P (50 μM) in Saos2, SJSA1, VCaP and PC3 after 4 h incubation. Red arrows indicate the marker, green arrow indicates 1. Yellow indicates their colocalization. Figure 3. Viabilities of cells treated by 1P (100 μM) or 2P (50 μM) in (A) Saos2, VCaP, and the coculture of Saos2 and VCaP and in (B) Saos2, PC3, and the coculture of Saos2 and PC3. The mean fluorescence intensities of 1P (50 μM) in (C) VCaP and VCaP and the VCaP cocultured with Saos2-DsRed and (D) PC3-DsRed cells and the PC3-DsRed cocultured with Saos2 (each line representing one cell). Figure 4. A cyro-EM images of the filaments of (A) 1P and (B) 1. Scale bar 50 nm. (C) 3D reconstruction of tubes of 1P and 1 from cryo-EM. (D) Top view of tubes of 1P and 1, phosphate of 1P can be seen at the periphery. (E) Side view of tubes of 1. Scheme 1. EISA for simultaneously killing mCRPC and osteoblast cells and the relevant synthesis. Supporting Information Materials and detailed experimental procedures, TEM and CLSM images, cell viabilities, chemical structures of the compounds (PDF), and coordinates of the cryo-EM structures (PDB). The authors declare no competing financial interests. REFERENCES 1. American Cancer Society. Cancer Facts & Figures 2022. Atlanta: American Cancer Society; 2022. 2. Pego ER ; Fernández I ; Núñez MJ , Molecular basis of the effect of MMP-9 on the prostate bone metastasis: A review. Urol. Oncol. Semin. Orig. Invest 2018, 36 (6 ), 272–282. 3. Zhang W ; Bado IL ; Hu J ; Wan YW ; Wu L ; Wang H ; Gao Y ; Jeong HH ; Xu Z ; Hao X ; Lege BM ; Al-Ouran R ; Li L ; Li J ; Yu L ; Singh S ; Lo HC ; Niu M ; Liu J ; Jiang W ; Li Y ; Wong STC ; Cheng C ; Liu Z ; Zhang XH , The bone microenvironment invigorates metastatic seeds for further dissemination. Cell 2021, 184 (9 ), 2471–2486 e20.33878291 4. Cha TL ; Wu TTL ; Vogelzang NJ ; Huang CY ; Huang SP ; Lin CC ; Ou YC ; Pang ST ; Shen DHY ; Wu WJ ; Chang WYH , Optimal usage of radium-223 in metastatic castration-resistant prostate cancer. J. Formos. Med. Assoc 2017, 116 (11 ), 825–836.29046247 5. Suominen MI ; Fagerlund KM ; Rissanen JP ; Konkol YM ; Morko JP ; Peng Z ; Alhoniemi EJ ; Laine SK ; Corey E ; Mumberg D ; Ziegelbauer K ; Käkönen SM ; Halleen JM ; Vessella RL ; Scholz A , Radium-223 inhibits osseous prostate cancer growth by dual targeting of cancer cells and bone microenvironment in mouse models. Clin. Cancer Res 2017, 23 (15 ), 4335–4346.28364014 6. Kirschenbaum A ; Liu XH ; Yao S ; Leiter A ; Levine AC , Prostatic acid phosphatase is expressed in human prostate cancer bone metastases and promotes osteoblast differentiation. In Ann Ny Acad Sci, Blackwell Publishing Inc.: 2011; Vol. 1237 , pp 64–70.22082367 7. Heinrich D ; Bruland O ; Guise TA ; Suzuki H ; Sartor O , Alkaline phosphatase in metastatic castration-resistant prostate cancer: Reassessment of an older biomarker. Future Oncology 2018, 14 (24 ), 2543–2556.29925281 8. Lin M-F ; Lee M-S ; Zhou X-W ; Andressen JC ; Meng T-C ; Johansson SL ; West WW ; Taylor RJ ; Anderson JR ; Lin F-F , Decreased Expression of Cellular Prostatic Acid Phosphatase Increases Tumorigenicity of Human Prostate Cancer Cells. Journal of Urology 2001, 166 (5 ), 1943–1950.11586265 9. Wang H ; Feng Z ; Wang Y ; Zhou R ; Yang Z ; Xu B , Integrating Enzymatic Self-Assembly and Mitochondria Targeting for Selectively Killing Cancer Cells without Acquired Drug Resistance. J Am Chem Soc 2016, 138 (49 ), 16046–16055.27960313 10. Stanford SM ; Bottini N , Targeting Tyrosine Phosphatases: Time to End the Stigma. Trends Pharmacol Sci 2017, 38 (6 ), 524–540.28412041 11. Vainonen JP ; Momeny M ; Westermarck J , Druggable cancer phosphatases. Sci. Trans. Med 2021, 13 (588 ), eabe2967. 12. He H ; Tan W ; Guo J ; Yi M ; Shy AN ; Xu B , Enzymatic Noncovalent Synthesis. Chem Rev 2020, 120 (18 ), 9994–10078.32812754 13. Kaplan MM , Alkaline phosphatase. N Engl J Med 1972, 286 (4 ), 200–2.4550137 14. Yang ZM ; Xu KM ; Guo ZF ; Guo ZH ; Xu B , Intracellular Enzymatic Formation of Nanofibers Results in Hydrogelation and Regulated Cell Death. Adv. Mater 2007, 19 (20 ), 3152–3156. 15. Kuang Y ; Shi J ; Li J ; Yuan D ; Alberti KA ; Xu Q ; Xu B , Pericellular hydrogel/nanonets inhibit cancer cells. Angew Chem Int Ed Engl 2014, 53 (31 ), 8104–7.24820524 16. Tanaka A ; Fukuoka Y ; Morimoto Y ; Honjo T ; Koda D ; Goto M ; Maruyama T , Cancer cell death induced by the intracellular self-assembly of an enzyme-responsive supramolecular gelator. J Am Chem Soc 2015, 137 (2 ), 770–5.25521540 17. Zhou J ; Du X ; Yamagata N ; Xu B , Enzyme-Instructed Self-Assembly of Small D-Peptides as a Multiple-Step Process for Selectively Killing Cancer Cells. J Am Chem Soc 2016, 138 (11 ), 3813–23.26966844 18. Gao Y ; Kuang Y ; Guo ZF ; Guo Z ; Krauss IJ ; Xu B , Enzyme-instructed molecular self-assembly confers nanofibers and a supramolecular hydrogel of taxol derivative. J Am Chem Soc 2009, 131 (38 ), 13576–7.19731909 19. Pires RA ; Abul-Haija YM ; Costa DS ; Novoa-Carballal R ; Reis RL ; Ulijn RV ; Pashkuleva I , Controlling cancer cell fate using localized biocatalytic self-assembly of an aromatic carbohydrate amphiphile. J Am Chem Soc 2015, 137 (2 ), 576–9.25539667 20. Wang H ; Feng Z ; Wu D ; Fritzsching KJ ; Rigney M ; Zhou J ; Jiang Y ; Schmidt-Rohr K ; Xu B , Enzyme-Regulated Supramolecular Assemblies of Cholesterol Conjugates against Drug-Resistant Ovarian Cancer Cells. J Am Chem Soc 2016, 138 (34 ), 10758–61.27529637 21. Wang J ; Tan W ; Li G ; Wu D ; He H ; Xu J ; Yi M ; Zhang Y ; Aghvami SA ; Fraden S ; Xu B , Enzymatic Insertion of Lipids Increases Membrane Tension for Inhibiting Drug Resistant Cancer Cells. Chemistry 2020, 26 (66 ), 15116–15120.32579262 22. Parak WJ ; Weil T ; Weiss PS , A Virtual Issue on Nanomedicine. ACS Nano 2021, 15 (10 ), 15397–15401. 23. Feng Z ; Wang H ; Yi M ; Lo CY ; Sallee A ; Hsieh JT ; Xu B , Instructed-Assembly of Small Peptides Inhibits Drug-Resistant Prostate Cancer Cells. Pept Sci (Hoboken) 2020, 112 (1 ), e24123. 24. Irwin JJ ; Duan D ; Torosyan H ; Doak AK ; Ziebart KT ; Sterling T ; Tumanian G ; Shoichet BK , An Aggregation Advisor for Ligand Discovery. J Med Chem 2015, 58 (17 ), 7076–87.26295373 25. Durbin SD ; Arakelian SM ; Shen YR , Optical-field-induced birefringence and Freedericksz transition in a nematic liquid crystal. Physical Review Letters 1981, 47 (19 ), 1411–1414. 26. Kouwer PHJ ; Swager TM , Synthesis and mesomorphic properties of rigid-core ionic liquid crystals. J. Am. Chem. Soc 2007, 129 (45 ), 14042–14052.17949091 27. Kato T ; Hirota N ; Fujishima A ; Fréchet JMJ , Supramolecular hydrogen-bonded liquid-crystalline polymer complexes. Design of side-chain polymers and a host-guest system by noncovalent interaction. J Polym Sci Part A 1996, 34 (1 ), 57–62. 28. Cao Y ; Alaasar M ; Zhang L ; Zhu C ; Tschierske C ; Liu F , Supramolecular meso-Trick: Ambidextrous Mirror Symmetry Breaking in a Liquid Crystalline Network with Tetragonal Symmetry. J Am Chem Soc 2022, 144 (15 ), 6936–6945.35394276 29. Peterca M ; Imam MR ; Leowanawat P ; Rosen BM ; Wilson DA ; Wilson CJ ; Zeng X ; Ungar G ; Heiney PA ; Percec V , Self-assembly of hybrid dendrons into doubly segregated supramolecular polyhedral columns and vesicles. J. Am. Chem. Soc 2010, 132 (32 ), 11288–11305.20698696 30. Van Esch J ; De Feyter S ; Kellogg RM ; De Schryver F ; Feringa BL , Self-assembly of bisurea compounds in organic solvents and on solid substrates. CHEM. EUR. J 1997, 3 (8 ), 1238–1243. 31. Kunitake T ; Okahata Y ; Shimomura M ; Yasunami SI ; Takarabe K , Formation of Stable Bilayer Assemblies in Water from Single-Chain Amphiphiles. Relationship between the Amphiphile Structure and the Aggregate Morphology. J. Am. Chem. Soc 1981, 103 (18 ), 5401–5413. 32. Nesloney CL ; Kelly JW , A 2,3’-substituted biphenyl-based amino acid facilitates the formation of a monomeric β-hairpin-like structure in aqueous solution at elevated temperature. J. Am. Chem. Soc 1996, 118 (25 ), 5836–5845. 33. Yi M ; Guo J ; He H ; Tan W ; Harmon N ; Ghebreyessus K ; Xu B , Phosphobisaromatic motifs enable rapid enzymatic self-assembly and hydrogelation of short peptides. Soft Matter 2021, 17 , 8590–8594.34545895 34. Jansa J ; Řezníček T ; Jambor R ; Bureš F ; Lyčka A , Synthesis of Hydroxy-Substituted p-Terphenyls and some Larger Oligophenylenes via Palladium on Charcoal Catalyzed Suzuki-Miyaura Reaction. Advanced Synthesis & Catalysis 2017, 359 (2 ), 339–350. 35. Miners JO ; Knights KM ; Houston JB ; Mackenzie PI , In vitro–in vivo correlation for drugs and other compounds eliminated by glucuronidation in humans: Pitfalls and promises. Biochemical Pharmacology 2006, 71 (11 ), 1531–1539.16455060 36. Montserat J ; Chen L ; Lawrence DS ; Zhang ZY , Potent low molecular weight substrates for protein-tyrosine phosphatase. J Biol Chem 1996, 271 (13 ), 7868–7872.8631832 37. Haj FG ; Verveer PJ ; Squire A ; Neel BG ; Bastiaens PIH , Imaging Sites of Receptor Dephosphorylation by PTP1B on the Surface of the Endoplasmic Reticulum. Science 2002, 295 (5560 ), 1708–1711.11872838 38. Tappel A , Lysosomal and prostasomal hydrolytic enzymes and redox processes and initiation of prostate cancer. Medical Hypotheses 2005, 64 (6 ), 1170–1172.15823710 39. Wang F ; Gnewou O ; Solemanifar A ; Conticello VP ; Egelman EH , Cryo-EM of Helical Polymers. Chem Rev 2022. DOI: 10.1021/acs.chemrev.1c00753 40. McGovern SL ; Caselli E ; Grigorieff N ; Shoichet BK , A common mechanism underlying promiscuous inhibitors from virtual and high-throughput screening. J Med Chem 2002, 45 (8 ), 1712–22.11931626

PMC009xxxxxx/PMC9339536.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7503056 4435 J Am Chem Soc J Am Chem Soc Journal of the American Chemical Society 0002-7863 1520-5126 35830682 9339536 10.1021/jacs.2c04937 NIHMS1823669 Article Engineered P450 Atom Transfer Radical Cyclases Are Bifunctional Biocatalysts: Reaction Mechanism and Origin of Enantioselectivity Fu Yue http://orcid.org/0000-0003-3085-7817 a Chen Heyu http://orcid.org/0000-0002-6068-9386 b Fu Wenzhen http://orcid.org/0000-0002-8454-4720 b Garcia-Borràs Marc http://orcid.org/0000-0001-9458-1114 c Yang Yang http://orcid.org/0000-0002-4956-2034 bd Liu Peng http://orcid.org/0000-0002-8188-632X a a Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, United States b Department of Chemistry and Biochemistry, University of California, Santa Barbara, California 93106, United States c Institut de Química Computacional i Catalisi (IQCC) and Departament de Química, Universitat de Girona, Girona 17003, Spain d Biomolecular Science and Engineering (BMSE) Program, University of California, Santa Barbara, California 93106, United States Corresponding Author: Marc Garcia-Borràs –Institut de Química Computacional i Catalisi (IQCC) and Departament de Química, Universitat de Girona, Girona 17003, Spain; marc.garcia@udg.edu, Yang Yang – Department of Chemistry and Biochemistry and Biomolecular Science and Engineering (BMSE) Program, University of California, Santa Barbara, California 93106, United States; yang@chem.ucsb.edu, Peng Liu – Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, United States; pengliu@pitt.edu Yue Fu – Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, United States; Heyu Chen – Department of Chemistry and Biochemistry, University of California, Santa Barbara, California 93106, United States; Wenzhen Fu – Department of Chemistry and Biochemistry, University of California, Santa Barbara, California 93106, United States; 16 7 2022 27 7 2022 13 7 2022 27 7 2023 144 29 1334413355 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. New-to-nature radical biocatalysis has recently emerged as a powerful strategy to tame fleeting open-shell intermediates for stereoselective transformations. In 2021, we introduced a novel metalloredox biocatalysis strategy that leverages the innate redox properties of the heme cofactor of P450 enzymes, furnishing new-to-nature atom transfer radical cyclases (ATRCases) with excellent activity and stereoselectivity. Herein, we report a combined computational and experimental study to shed light on the mechanism and the origins of enantioselectivity for this system. Molecular dynamics and QM/MM calculations revealed an unexpected role of the key beneficial mutation I263Q. The glutamine residue serves as an essential hydrogen bond donor that engages with the carbonyl moiety of the substrate to promote bromine atom abstraction and enhance the enantioselectivity of radical cyclization. Therefore, the evolved ATRCase is a bifunctional biocatalyst, wherein the heme cofactor enables atom transfer radical biocatalysis while the hydrogen bond donor residue further enhances the activity and enantioselectivity. Unlike many enzymatic stereocontrol rationales based on a rigid substrate binding model, our computations demonstrate a high degree of rotational flexibility of the allyl moiety in enzyme–substrate complex and succeeding intermediates. Therefore, the enantioselectivity is controlled by the radical cyclization transition states rather than substrate orientation in ground state complexes in the preceding steps. During radical cyclization, anchoring effects of the Q263 residue and steric interactions with the heme cofactor concurrently control the π-facial selectivity, allowing for highly enantioselective C–C bond formation. Our computational findings are corroborated by experiments with ATRCase mutants generated from site-directed mutagenesis. Graphical Abstract: pmcINTRODUCTION Due to their ability to exert exquisite stereocontrol over challenging chemical reactions, enzymes are excellent catalysts for asymmetric synthesis in applications that range from small-scale synthesis to industrial manufacturing.1 Traditional biocatalysis research focuses on the discovery, engineering, and application of naturally existing enzyme functions of outstanding synthetic value. However, compared to the immensely diverse range of organic reactions discovered and optimized by synthetic chemists, only a small subset of these reactivity patterns is found in natural enzymology and are currently being utilized by biocatalysis practitioners, thus imposing a major limitation on the utility of contemporary enzyme technologies. The implementation of unnatural chemistries by repurposing naturally existing enzymatic machineries promises to expand the reaction space of biocatalysis, thereby significantly augmenting the synthetic chemist’s toolbox.2 Recently, we commenced a research program to repurpose naturally occurring metalloenzymes to catalyze unnatural stereoselective radical reactions using a metalloredox mechanism.3 Almost 50% of naturally occurring proteins are metalloproteins,4 among which redox-active first-row transition-metal cofactors such as Fe(II)/Fe(III),5 Co(I)/Co(II),6 and Cu(I)/Cu(II)7 are ubiquitous. Cognizant of the tremendous synthetic potential of these metalloproteins in facilitating redox-mediated radical reactions, we recently repurposed cytochromes P450, a class of promiscuous metalloenzymes with numerous applications,5a-b,8–9 to catalyze new-to-nature atom transfer radical cyclization (ATRC) in an enantio- and diastereodivergent fashion (Figure 1a).3 Due in large part to the difficulties in maintaining a tight association with the free radical intermediate and/or the unfunctionalized olefin, inducing high levels of stereocontrol for free radical-mediated olefin functionalization reactions continues to pose a formidable challenge for chiral small-molecule catalysts.10 In particular, catalytic asymmetric ATRC reactions remain rare.11 Thus, our evolved P450 atom transfer radical cyclases provide a new means of taming radical intermediates for a synthetically valuable but underdeveloped class of asymmetric transformations. This metalloredox strategy is complementary to the elegant work of Hyster12 and Zhao13 on reductive C–C bond forming photoredox transformations using flavoenzymes, as the metallocofactor in our work allows redox-neutral atom transfer reactions to proceed with excellent stereocontrol. To further advance this recently developed mode of metalloredox radical biocatalysis, it is imperative to gain further understanding of reaction mechanism and origin of enzyme-controlled stereoselectivity. The proposed mechanism of this enzymatic ATRC reaction involves radical initiation via bromine atom transfer from the substrate to the heme cofactor, enantioselective radical cyclization, and bromine atom rebound to form the product (Figure 1c). However, several key mechanistic questions remain unaddressed. First, our previous work showed that evolved P450 radical cyclases displayed substantially faster kinetics and higher total turnover numbers relative to free cofactor in promoting this ATRC process,3 but the origin of this enhanced activity is unclear. Second, the mode of enantioinduction for this radical-mediated olefin functionalization is potentially distinct from those of other types of well-established natural and unnatural enzymatic reactions and remains to be uncovered. How these enhancements in reactivity and stereoselectivity relate to key evolved structural elements of the ATRCase needs to be elucidated. Stereocontrol of many enzymatic olefin functionalization reactions has been rationalized through π-facial selectivity models based on ground-state structure of enzyme–substrate complexes,13a,14 where the rotational freedom of the olefin is greatly reduced and the two prochiral π-faces are easily differentiated. Such intuitive substrate binding models obtained from experimental X-ray structures and computational substrate docking and/or classical molecular dynamics (MD) simulations have been widely used in biocatalysis and protein engineering. Nevertheless, an increasing number of studies underscored the importance of interrogating transition-state models to gain an accurate understanding of enzymatic stereoselectivities,15–16 especially when the reactive functional group of the substrate (e.g., an olefin) does not strongly interact with the protein scaffold and is flexible in the enzyme–substrate complex (Figure 1d). In this situation, substrate binding models become ineffective, and computational models based on transition-state analysis are critical to describe the origin of enzymatic stereocontrol. In the recently developed biocatalytic enantioselective ATRC, it is not clear which enantioinduction scenario is operative. Depending on the conformational flexibility of the olefin moiety and the carbon-centered radical in the enzyme active site, the enantioselectivity may be rationalized by substrate binding conformation or by the π-facial selectivity of the radical cyclization transition state (Figure 1d). Herein, we performed computational studies to investigate the reaction mechanism and key factors promoting this new-to-nature atom transfer radical cyclization and to explore the origin of enantioselectivity. We studied how interactions with active site residues facilitate the substrate activation step, leading to faster radical initiation. To compare the two enantioinduction scenarios (Figure 1d), we examined substrate binding modes and conformational flexibility of the olefin in the enzyme–substrate complex and the radical intermediate via classical MD simulations and hybrid quantum mechanics/molecular mechanics (QM/MM) metadynamics simulations. These ground state behaviors are compared with transition state enantiocontrol by computing the selectivity-determining radical cyclization transition states via QM/MM-optimizations and QM/MM metadynamics17 simulations. Our work revealed the highly flexible nature of the olefin in the enzyme–substrate complex, clearly demonstrating that enantiocontrol is governed by transition-state stability and not substrate conformational control upon binding. This study unveiled the unexpected role of a glutamine residue (Q263) acting as the hydrogen bond donor13a,18 to activate the substrate toward radical initiation and enhance the enantioselectivity in radical cyclization (Figure 1b). The importance of this key residue in promoting reactivity and selectivity was then validated experimentally using enzyme variants derived from site-directed mutagenesis. Together, these studies showed that our directed evolution efforts led to the serendipitous discovery of a bifunctional biocatalyst, wherein the heme cofactor enables atom transfer radical biocatalysis and the hydrogen bond donor residue further activates the substrate and enhances the enantioselectivity. COMPUTATIONAL AND EXPERIMENTAL METHODS Classical MD Simulations. In this study, we focused on the enzymatic reaction catalyzed by P450ATRCase1, an (R)-product forming enzyme. The initial geometry of P450ATRCase1 used in the modeling was generated by modifying the available X-ray crystal structure of a closely related P450 variant (PDB ID: 4H23).19 Six mutations (A82T, L181F, I263Q, H266T, T327I, and T438S) were introduced into 4H23 using the mutagenesis tool in PyMOL20 to prepare P450ATRCase1. Classical molecular dynamics (MD) simulations were carried out using the pmemd module of the GPU-accelerated Amber 20 package.21 Force field parameters for the iron–porphyrin complex were generated using the MCPB.py module22 with the general Amber force field (gaff).23 Parameters for substrate 1 were generated using the gaff force field, whereas the Amber ff14SB force field24 was used for standard residues and TIP3P for solvent water molecules. First, three replicas25 of independent 500 ns MD simulations were performed in the holo state of P450ATRCase1 in the absence of substrate 1. Clustering analysis based on the root-mean-square deviation (RMSD) of backbone was carried out using the cpptraj module26 to identify the most populated protein conformation in the MD simulations of all three replicas. A representative snapshot of the most visited structure was used for docking calculations with substrate 1 using the AutoDock package.27 Then, MD simulations of substrate-bound P450ATRCase1 were performed with and without restraints to study the preferred substrate binding pose and the possible interaction modes between activate site residues and the substrate. In the unrestrained MD simulations, three replicas of 500 ns simulations were performed without including external forces. In the restrained MD simulations, three replicas of 500 ns MD simulations were performed by restraining the Fe–Br distance (2.7–4.0 Å) by applying a harmonic potential of 500 kcal·mol−1·Å−2. These restraints were applied to simulate substrate near attack conformation (NAC) in the inner-sphere bromine atom transfer pathway. This strategy is similar to those applied in previous studies.14f,28 The restrained distance range used (2.7–4.0 Å) was determined based on the Fe–Br distance observed in a DFT-optimized dative complex using Fe–porphine as a model, which has a Fe–Br distance of 3.80 Å (see Figure S1 of the SI for details). Additional restrained classical MD simulations where both the Fe–Br distance and the hydrogen bond distance between the carbonyl group of the substrate and the amide of the Q263 residue were restrained (the HQ263⋯Osub distance was restrained in the range of 1–3 Å with a harmonic potential of 200 kcal·mol−1·Å−2). The most representative snapshots from the restrained MD simulations, based on protein backbone RMSD analysis, were used as the initial geometries for QM/MM calculations and QM/MM metadynamics simulations. QM/MM Calculations of Reaction Energy Profiles. The ONIOM algorithm29 implemented in Gaussian 1630 was used in QM/MM calculations to characterize the stationary points (intermediates and transition states). Water molecules and counterions within 5 Å from the enzyme were included in the QM/MM calculations. Several conformers of the substrate were considered for each intermediate and transition state (see Figures S2–S3 of the SI for higher energy conformers). The QM region includes the heme cofactor, the side chain of the Fe-binding serine residue (S400), the substrate, and boundary hydrogen atoms. This includes a total of 77 atoms in the QM region. For the QM region, the dispersion-corrected B3LYP31-D332/6–31G(d)–LANL2DZ(Fe) level of theory was used in geometry optimization and vibrational frequency calculations, and the B3LYP-D3/6–311+G(d,p)–LANL2TZ(f)(Fe) level of theory was used in single-point energy calculations. This level of theory has been shown to provide good agreement with PNO-LCCSD(T)-F12 benchmark results.3 For the MM region, the same force field parameters from the classical MD simulations discussed above were used. The quadratic coupled algorithm33 and the mechanical embedding scheme were used in geometry optimization. Residues greater than 6 Å away from the QM region were kept fixed during geometry optimization. Single-point energy calculations were performed with the electronic embedding scheme, which better describes electrostatic interactions between QM and MM regions.34 Open-shell singlet, triplet, quintet, and septet spin35 states for each structure were considered. Wavefunction stability of all structures was confirmed by using the “stable=opt” keyword. QM/MM Metadynamics Simulations. All QM/MM Born Oppenheimer MD metadynamics simulations were performed with the CP2K 7.1 package,36 combining the QM program QUICKSTEP37 and the MM driver FIST. In this program, a real-space multigrid technique is used to compute the electrostatic coupling between the QM and MM regions.38 The heme cofactor, the side chains of F181, Q263 (two key active site residues identified by protein engineering), and the Fe-binding S400, the substrate, and boundary hydrogen atoms were included in the QM region. This leads to 137 atoms in the QM region. The remaining part of the system was modeled at the MM level using the same parameters as in the classical MD simulations. The QM region was treated at the DFT (BLYP-D3) level,39 employing the Gaussian and plane waves method (GPW) that combines Gaussian-type basis functions and plane-waves as an auxiliary basis. The DZVP basis set40 and Goedecker-Teter-Hutter pseudopotentials41 were employed. The auxiliary plane-wave basis set was expanded up to a 280 Ry cutoff. Trajectories starting from different initial geometries, obtained from snapshots of the restrained classical MD simulations, were simulated in the QM/MM metadynamics calculations. All QM/MM metadynamics simulations were performed in the NVT (constant number of atoms, volume, and temperature) ensemble using an integration time step of 0.5 fs. First, the system was equilibrated without any restraint for 2.0 ps. Then, the metadynamics method17 was used to compute the free energy profiles. In the simulations of the radical cyclization pathways, one collective variable was defined as the distance of the forming C–C bond between the radical center and the alkenyl carbon of the substrate. In the simulations to study the flexibility of the N-allyl group in the radical intermediate, two collective variables were defined as dihedral angles about the allylic C–C (ϕ) and N–C(allyl) (θ) bonds. Repulsive Gaussian-shaped potential hills with a height of 0.3 kcal/mol and a width of 0.1 bohr for distance and 0.1 rad for dihedral angle were added to the potential every 20 molecular dynamics steps. Expression of P450ATRCase1 variants. E. coli (E. cloni BL21(DE3)) cells carrying plasmid encoding the indicated P450ATRCase1 variant were grown overnight (12–14 h) in Luria broth with ampicillin (LBamp, 2.5 mL) in a culture tube. Preculture (1.5 mL, 5% v/v) was used to inoculate 30 mL of HBamp in a 125 mL Erlenmeyer flask. This culture was incubated at 37 °C, 230 rpm for 2 h in a New Brunswick Innova 44R shaker. The culture was then cooled on ice for 20 min and induced with 0.5 mM IPTG and 1.0 mM 5-aminolevulinic acid (final concentrations). Protein expression was conducted at 22 °C, 150 rpm, for 20–22 h. E. coli cells were then transferred to a 50 mL conical tube and pelleted by centrifugation (3000 rpm, 5 min, 4 °C) using an Eppendorf 5910R tabletop centrifuge. The supernatant was removed and the resulting cell pellet was resuspended in M9-N buffer to OD600 = 30. An aliquot of this cell suspension (2 mL) was taken to determine protein concentration by hemochrome assay after cell lysis by sonication. Biotransformations using whole E. coli cells. Suspensions of E. coli cells expressing the P450ATRCase1 variant in M9-N buffer (OD600 = 30, pH = 7.40) were kept on ice. In another conical tube, a stock solution of D-glucose (500 mM in M9-N) was prepared. To a 2 mL vial were added the suspension of E. coli cells (typically OD600 = 30, 345 μL) and D-glucose (40 μL of 500 mM stock solution in M9-N buffer). This 2 mL vial was then transferred into an anaerobic chamber, where the ATRC substrate (15 μL of 270 mM stock solution in EtOH) was added. The final reaction volume was 400 μL; the final concentrations of substrate and D-glucose were 10 mM and 50 mM, respectively. The vials were sealed and shaken in a Corning digital microplate shaker at room temperature and 680 rpm for 12 h. The reaction mixture was then extracted with 1:1 EtOAc/hexanes and analyzed by chiral HPLC using mesitylene as the internal standard. For each P450ATRCase1 variant, whole-cell reactions were performed in triplicate. Averaged yields and total turnover numbers (TTNs) were reported. RESULTS AND DISCUSSION Preferred Substrate Binding Pose and Unexpected Hydrogen Bonding Interaction with Key Residue Q263. To explore the preferred substrate binding pose and interaction modes between the substrate and active site residues, we performed classical MD simulations of the enzyme–substrate complex. After docking substrate 1 into the active site of P450ATRCase1, we performed three replicas of 500 ns MD simulation without any restraint (unrestrained MD). We also performed another three replicas of 500 ns MD simulations by restraining the Fe–Br distance within 2.7–4.0 Å to mimic the near attack conformation (NAC)14f,28 for bromine atom abstraction (restrained MD). Both MD simulations revealed the existence of two dominant interaction modes with Q263 (Figure 2a), where the carbonyl group of the substrate forms a hydrogen bond with the NH2 group of the side chain of Q263 (interaction mode A) or with a water molecule bridging Q263 and the substrate (interaction mode B). The unrestrained MD simulations revealed that in most of the simulation time (63.9%), the N–H∙∙∙O distance between the side chain NH2 group in Q263 and the amide carbonyl oxygen of 1 is shorter than 3 Å. In the restrained MD simulations, this direct Q263–substrate hydrogen bond was observed in a smaller percentage of the simulation time (21.2 %), because the distance restriction between Fe and Br induces a less favorable spatial arrangement for the hydrogen bond. Nonetheless, most snapshots maintain a relatively short distance between Q263 and the substrate (< 5 Å), with either a direct hydrogen bond with Q263’s NH2 group or a water-bridged hydrogen bond between these two groups (Figure 2a). These MD simulations suggest that hydrogen bonding interactions with Q263 are important for substrate binding and may be involved in subsequent steps of the catalytic cycle. This will be examined using QM/MM calculations in the next section. Both unrestrained and restrained MD simulations describe a preferred binding pose of the substrate in which the N-benzyl group of 1 is placed in proximity to L437, establishing hydrophobic C–H∙∙∙π interactions (Figure 2b). Due to this stabilizing interaction, the s-cis conformer of the amide is strongly favored within the active site, as seen in greater than 93% of the simulation time (see Figures S5 and S9 of the SI for details). In the favored s-cis conformer, the N-allyl group is cis to the bromoalkyl group, a conformation required in the subsequent radical cyclization step. In the absence of enzyme scaffold, rotation along the amide bond led to less efficient ATRC of N-allyl α-haloamides,42 demonstrating the templating effect of the protein scaffold in facilitating radical catalysis. Overall, the preferred binding pose of 1 involves both hydrogen bonding interaction with the amide carbonyl and C–H∙∙∙π interactions with the N-benzyl group. These interactions not only promote substrate binding but also stabilize the s-cis conformer of the amide poised to undergo radical cyclization. MM-GBSA substrate–residue pair interaction calculations43 (Figure 2c) revealed that Q263 and L437 are among residues establishing the most stabilizing interactions with the substrate, further highlighting their importance for the substrate binding via hydrogen bonding and C−H∙∙∙π interactions with these residues, respectively. Reaction Energy Profiles from QM/MM Calculations and the Roles of Q263 on Reactivity of Substrate Activation. We next used QM/MM methods to compute the free energy profile of this biocatalytic ATRC process. QM/MM calculations were performed starting from the preferred substrate binding pose characterized by MD simulations, and considering interaction mode A with Q263 residue (Figure 2b), where the amide side chain of Q263 engages the substrate in hydrogen bonding interactions. Open-shell singlet, triplet, quintet, and septet spin states of each intermediate and transition state structure were optimized using QM/MM (Figure S14). Gibbs free energy profiles involving the two most favorable spin states, quintet and septet, affording the major enantiomeric product (R)-2 via radical addition to the (Si)-face of the alkene (TS2-(Si)) are shown in Figure 3. The quintet spin state was found to be the most favorable spin state for the enzyme–substrate and enzyme–product complexes and bromine atom abstraction and bromine atom rebound transition states (TS1 and (R)-TS3), whereas the septet spin state was found to be the most stable in α-carbonyl radical 4, radical cyclization transition state TS2-(Si), and the succeeding cyclized primary radical (R)-5 (see Figure S15 for spin densities of QM/MM-optimized structures). The QM/MM-computed energy profiles revealed several key mechanistic features critical for the reactivity and enantioselectivity of this enzymatic ATRC. First, the Fe(II)/Fe(III) metalloredox processes (TS1 and (R)-TS3) are both kinetically facile. Although the radical initiation via bromine atom abstraction (TS1) is endergonic by 6.4 kcal/mol, it requires a relatively low activation free energy of 17.3 kcal/mol. The endergonicity of this step is comparable to the bromine atom abstraction step in Cu-catalyzed atom transfer radical polymerization (Cu-ATRP), which has an equilibrium constant of KATRP = 10−9 ~ 10−4 in most common Cu-ATRP systems.44 The relatively high HOMO energy of the heme cofactor (−3.3 eV, compared with −5.6 eV for Cu(TPMA)+, a representative Cu-ATRP catalyst)45 suggests that this Fe-mediated bromine atom abstraction is kinetically promoted due to effective metal-to-substrate charge transfer in the bromine atom abstraction transition state.46 Because bromine atom abstraction is the rate-determining step in the QM/MM-computed catalytic cycle, a low kinetic barrier is essential for the reactivity of the ATRC. On the other hand, the exergonicity of the bromine atom rebound step enables rapid trapping of the enantioenriched cyclized primary radical intermediate (R)-5 via (R)-TS3. Because the Gibbs free energy of (R)-TS3 is lower than that of TS2-(Si), the radical cyclization (TS2-(Si)) is irreversible, and thus determines the enantioselectivity. The reactivity of bromine atom abstraction is promoted by hydrogen bonding interaction between the amide side chain in Q263 and the carbonyl group of substrate 1. This hydrogen bond persists throughout catalysis among all the QM/MM-optimized intermediate and transition state structures (Figure 4). Furthermore, our QM/MM calculations showed slightly shorter N–H∙∙∙O distances in bromine atom abstraction transition state TS1 and α-carbonyl radical intermediate 4 compared to that in the enzyme–substrate complex 3 (Figure 4a). These results indicate that this hydrogen bond not only promotes the substrate binding but also more substantially stabilizes bromine atom abstraction TS and the radical being formed, promoting this rate-determining substrate activation step.47 Further calculations using truncated model systems showed that this hydrogen bonding interaction lowers the energy of the LUMO orbital of the α-bromoamide moiety, thereby weakening the α-C–Br bond (see Figure S17 for the effects of hydrogen bonding interactions on LUMO energies and C–Br BDE). The I263Q mutation represents one of the most important beneficial mutations in our previously reported directed evolution effort, as it led to dramatically enhanced activity and enantioselectivity of P450ATRCase1. Compared to its parent, the I263Q mutant increased the total turnover number (TTN) from 1810 to 5370 and enantiomeric ratio (e.r.) from 67:33 to 89:11.3 Despite these results, the role of this I263Q mutation was not known at the time P450ATRCase1 was engineered. The computational results disclosed herein rationalized the role of Q263 on the experimentally observed reactivity. The higher e.r. with the I263Q variant suggests that this residue also plays a key role in the enantioselectivity-determining step. This effect is discussed in the next section. Origin of Enantioselectivity and the Cooperative Effects of Q263 and Heme Cofactor on Enantioinduction. To understand the origin of enantioselectivity, we performed QM/MM calculations to study the enantioselectivity-determining radical cyclization transition states (Figure 4b). The transition state of radical addition to the (Si)-face of the alkenyl group TS2-(Si) leading to the experimentally observed major enantiomeric product (R)-2 is 2.5 kcal/mol lower in energy than TS2-(Re) leading to the opposite enantiomeric product, (S)-2. Hydrogen bonding interactions between Q263 and the carbonyl group of substrate 1 and C–H∙∙∙π interactions between L437 and the N-benzyl group on 1 are observed in both transition states TS2-(Si) and TS2-(Re) (Figure 4b). These interactions restrained the positioning of the substrate in the active site, placing the α-carbonyl radical center relatively close to the heme cofactor. When approaching the α-carbonyl radical during the radical cyclization, the alkenyl group is placed closer to the heme cofactor. In the favored radical cyclization transition state TS2-(Si), the alkenyl group points away from the heme, whereas in the disfavored transition state TS2-(Re), the alkenyl group points towards the heme, leading to unfavorable steric repulsions. This unfavorable steric effect is evidenced by the short distance between the terminal olefinic carbon and the bromine atom on heme (3.40 Å) in TS2-(Re). Next, we performed QM/MM metadynamics simulations to study the structural features along the radical cyclization reaction coordinate. The radical cyclization transition state geometries and activation free energies from QM/MM metadynamics are similar to those obtained from QM/MM geometry optimizations (see Figure S19 for details). The QM/MM metadynamics trajectories indicate that the Q263–substrate hydrogen bond along the radical cyclization pathway to form (R)-2 via TS2-(Si) remains relatively strong with an average HQ263⋯Osub distance smaller than 2.5 Å (Figure 5).48 On the other hand, the hydrogen bonding interaction with Q263 is weaker in the region near the disfavored transition state TS2-(Re), evidenced by slightly longer HQ263⋯Osub distances explored along the disfavored radical cyclization pathway. The steric repulsions with heme lead to unfavorable distortion of TS2-(Re), weakening the hydrogen bond with Q263, a key enzyme–substrate interaction. Overall, both the QM/MM and the metadynamics simulations highlighted the cooperative effects of the Q263 residue, hydrophobic active site residues, such as L437, and the heme cofactor in anchoring the substrate and exerting steric interactions to affect the enantioinduction in radical cyclization transition states. Classical MD and QM/MM Metadynamics Simulations on the Conformational Flexibility of the N-Allyl Group in Ground State Complexes. We performed molecular dynamics simulations using both classical MD and QM/MM metadynamics to explore the conformational flexibility of the N-allyl group in the enzyme–substrate complex 3 and the α-carbonyl radical intermediate 4 (Figure 6). We surmised that these simulations, in conjunction with the transition state modeling discussed above, would reveal which of the two enantioinduction scenarios shown in Figure 1d is operative in this enzymatic ATRC. In particular, these ground-state simulations could reveal whether the allyl group rotation is restricted prior to the radical cyclization transition state, therefore offering a binding-based enantioinduction model for π-facial discrimination. The conformations of the N-allyl group in the enzyme–substrate complex 3 observed along the unrestrained and restrained classical MD simulations are described in Figure 6a. These MD simulations showed four clusters of conformers (3a-d) with almost equal distributions, resulting from rotations about the N–C(allyl) (θ) and the allylic C–C (ϕ) bond. In the centroids of each cluster, the allyl group and the carbonyl are anticlinal (θ is within 90~150° or −90~−150°) rather than having the synperiplanar conformation (θ = 30~−30°) in the radical cyclization transition states (see Figures S20 and S21 of the SI for representative snapshots of these conformers). The lack of sterically bulky residues around the N-allyl group allows for the facile conformational change in the enzyme–substrate complex. Due to this conformational flexibility of the N-allyl group, there is no clear preference for the (Re)- or the (Si)-face of the C=C double bond to be exposed to the α-bromoamide moiety. Next, we performed QM/MM metadynamics simulations on the α-carbonyl radical intermediate 4 to investigate the rate of N-allyl group rotation once the radical is formed (Figure 6b). In these simulations, we used the dihedral angles about the allylic C–C (ϕ) and N–C(allyl) (θ) bonds as the collective variables. Similar to conformers 3a-d, the allyl group and the carbonyl are anti- or synclinal in all of the low-energy conformers of 4 (Figure S21). These conformers isomerize to synperiplanar conformation, such as in 4’-(Si) and 4’-(Re), via rotation about the N–C(allyl) (θ) bond prior to the radical cyclization transition state. Although 4’-(Si) and 4’-(Re) are not minima on the free energy surface, the conformational change to these synperiplanar structures is kinetically facile (see Figure S21 in the SI for the complete rotational free energy surface of the N-allyl group in 4). The QM/MM metadynamics calculations indicate conformer 4-(Si), which leads to the favored (Si)-face radical cyclization after N–C(allyl) (θ) bond rotation and radical addition, is 3.6 kcal/mol more stable than conformer 4-(Re), which leads to the less favorable radical cyclization with the (Re)-face of the olefin. Here, 4-(Re) is destabilized by steric repulsions between the terminal alkenyl group and heme cofactor, similar to the steric effect that destabilizes TS2-(Re). The low barrier to the interconversion between 4-(Si) and 4-(Re) via allylic C–C bond (ϕ) rotation (ΔG‡rot = 5.2 kcal/mol) indicates that the N-allyl conformational change is much faster than the radical cyclization (ΔG‡ = 8.1 kcal/mol via TS2-(Si)). The interconversion barrier between 4-(Si) and 4-(Re) is comparable to that of N-allylamide in the absence of the enzyme (see Figure S22 of the SI for details), indicating minimal interactions between the allyl group and active site residues in the α-carbonyl radical intermediate. Overall, these simulations indicated a highly flexible N-allyl group in both the enzyme-bound substrate and the enzyme-bound α-carbonyl radical intermediate. Due to the rapid conformational interconversion of the N-allyl group in these ground state complexes, the enantioselectivity of this new-to-nature enzymatic ATRC process is solely determined by the radical cyclization transition state and not by the initial substrate conformation. Experimental Investigations on the Importance of Residue 263 on Reactivity and Enantioselectivity. In light of the key role of residue Q263 uncovered by the computational studies, we generated P450ATRCase1 Q263X mutants (X = R, K, N, S, A, I, and E) by site-directed mutagenesis and examined their catalytic activity and enantioselectivity in the radical cyclization of 1 (Table 1). In this study, other potential hydrogen bond donors, including arginine, lysine, asparagine, and serine, were evaluated in addition to residues lacking a hydrogen bond donor, including alanine, isoleucine, and glutamate. Consistent with our computational insights, when Q263 was replaced by an appropriate alternative hydrogen bond donor residue, similar enzyme activity and enantioselectivity were observed. The second-best residue at 263 was found to be arginine (R263, Table 1, entry 2), which bears a guanidine functional group that can potentially serve as a hydrogen bond donor. With this Q263R mutant, yield, total turnover number (TTN), and enantioselectivity very similar to the Q263 parent were observed. The Q263K mutant provided slightly further reduced enantioselectivity (entry 3). Interestingly, a further drop in e.r. was observed when this glutamine was replaced by an asparagine (entry 4), highlighting the importance of the tethering unit length for this hydrogen bond donor to engage the amide substrate. A263 lacking a hydrogen bond donor side chain and S263 with a much shorter hydrogen bond donor hydroxymethyl side chain provided greatly reduced enzyme activity and enantioselectivity (entries 5–6). Similar to the Q263A mutant, reverting this Q263 to I263 in native P450BM3 led to inferior enzyme performance (entry 7). The E263 mutant bearing a presumably deprotonated glutamate at residue 263 also provided low activity and enantioselectivity (entry 8). Together, these studies provided further evidence to support the essential role of residue Q263 of P450ATRCase1, underscoring the importance of a hydrogen bond donor residue to both the enzyme activity and enantioselectivity. CONCLUSION Using a combined computational and experimental approach, we elucidated the mechanism and the origin of enantioselectivity of our recently developed biocatalytic atom transfer radical cyclization using a laboratory-evolved P450 cyclase. QM/MM and classical MD simulations showed that the substrate binds to the enzyme active site, establishing a stabilizing hydrogen bonding interaction with Q263 and C–H∙∙∙π interactions with L437. While these stabilizing interactions are maintained throughout the catalytic process, leading to a relatively rigid positioning of the substrate carbonyl within the enzyme active site, the N-allyl group of the substrate is highly flexible and undergoes rapid conformational change in enzyme-bound forms. The facile conformational change of the N-allyl group in ground state complexes makes the enantioselectivity entirely determined in the radical cyclization transition state. Notwithstanding the lack of conformational preference at the stage of various ground-state intermediates, high levels of enantioselectivity are achieved in the radical cyclization transition state where the olefin approaches the radical center, leading to further accentuated steric interactions with the heme cofactor. This study revealed the critical role of Q263 in promoting both reactivity and enantioselectivity, as it stabilizes substrate binding, promotes the rate-determining bromine atom abstraction, and controls the substrate orientation in the enantioselectivity-determining radical cyclization step. The multiple functions of Q263 were further corroborated by experiments evaluating the activity and enantioselectivity of enzyme variants generated by site-directed mutagenesis. Together, this study highlights the synergy between computations and experiments in providing insights into the mechanism of enantioinduction in radical-mediated enzymatic reactions. We expect that these insights will guide the further engineering of stereoselective ATRCases and development of other asymmetric new-to-nature radical-mediated enzymatic reactions. Supplementary Material Supporting Information ACKNOWLEDGMENT We thank the NIH (R35GM128779 for P.L. and R35GM147387 for Y.Y.), the University of California, Santa Barbara (startup funds to Y.Y.), and the Spanish MICINN (PID2019-111300GA-I00 project and RYC2020-028628-I fellowship to M.G.B.) for financial support. DFT calculations were performed at the Center for Research Computing of the University of Pittsburgh and the Extreme Science and Engineering Discovery Environment (XSEDE) supported by the National Science Foundation grant number ACI-1548562. Y.F. thanks the Andrew W. Mellon Predoctoral Fellowship. We acknowledge the BioPACIFIC MIP at UCSB (NSF Materials Innovation Platform, DMR-1933487) and the NSF MRSEC program (DMR-1720256) for access to instrumentation. We thank Prof. Yiming Wang (University of Pittsburgh) for critical reading of this manuscript. Figure 1. P450-catalyzed enantioselective atom transfer radical cyclization (ATRC). Figure 2. Classical MD simulations of the enzyme–substrate complex and analysis of substrate--protein interactions in the active site of the P450ATRCase1. Figure 3. Computed Gibbs free energy profiles of the P450ATRCase1-catalyzed ATRC from QM/MM calculations. The Gibbs free energies and enthalpies are with respect to a substrate–heme complex 3 where the bromine atom of the α-bromoamide substrate binds to the Fe center of the heme cofactor. Figure 4. QM/MM-optimized structures of select intermediates and transition states in the P450ATRCase1-catalyzed ATRC of 1. Gibbs free energies of all structures are with respect to 3. Figure 5. Q263 hydrogen bonding interactions along the radical cyclization pathways from QM/MM metadynamics simulations. The moving averages are shown in dark green lines. Figure 6. Conformational change of the N-allyl group in enzyme–substrate complex 3 and α-carbonyl radical intermediate 4 from (a) classical MD and (b) QM/MM metadynamics simulations. The black dots in (a) indicate the centroids of each cluster representing the rotamers about the N–C(allyl) (θ) and allylic C–C (ϕ) bonds. Table 1. Experimental validation. entry mutation of P450ATRCase1 yield (%)a TTN e.r.a 1 None 89 ± 2 4400 ± 100 96:4 2 Q263R 82 ± 3 3700 ± 100 95:5 3 Q263K 76 ± 2 3730 ± 90 91:9 4 Q263N 75 ± 0 4340 ± 20 84:16 5 Q263S 36 ± 1 2310 ± 50 78:22 6 Q263A 37 ± 1 2490 ± 70 79:21 7 Q263I 52 ± 8 1300 ± 200 87:13 8 Q263E 25 ± 0 1300 ± 10 66:34 a Yields and e.r.’s were determined by HPLC analysis. Reactions were carried out using whole E. coli cells harboring P450ATRCase1 mutants. Supporting Information. The Supporting Information is available free of charge at http://pubs.acs.org. Additional computational results and experimental details (PDF) REFERENCES 1 For reviews: (a) Reetz MT Laboratory Evolution of Stereoselective Enzymes: A Prolific Source of Catalysts for Asymmetric Reactions. Angew. Chemie Int. Ed 2011, 50 , 138–174. (b) Bornscheuer UT ; Huisman GW ; Kazlauskas RJ ; Lutz S ; Moore JC ; Robins K Engineering the Third Wave of Biocatalysis. Nature 2012, 485 , 185–194.22575958 (c) Devine PN ; Howard RM ; Kumar R ; Thompson MP ; Truppo MD ; Turner NJ Extending the Application of Biocatalysis to Meet the Challenges of Drug Development. Nat. Rev. Chem 2018, 2 , 409–421. (d) Bornscheuer UT ; Hauer B ; Jaeger KE ; Schwaneberg U Directed Evolution Empowered Redesign of Natural Proteins for the Sustainable Production of Chemicals and Pharmaceuticals. Angew. Chem. Int. Ed 2019, 58 , 36–40. (e) Zetzsche LE ; Chakrabarty S ; Narayan ARH The Transformative Power of Biocatalysis in Convergent Synthesis. J. Am. Chem. Soc 2022, 144 , 5214–5225.35290055 2 For reviews: (a) Renata H ; Wang ZJ ; Arnold FH Expanding the Enzyme Universe: Accessing Non-Natural Reactions by Mechanism-Guided Directed Evolution. Angew. Chemie Int. Ed 2015, 54 , 3351–3367. (b) Leveson-Gower RB ; Mayer C ; Roelfes G The Importance of Catalytic Promiscuity for Enzyme Design and Evolution. Nat. Rev. Chem 2019, 3 , 687–705. (c) Chen K ; Arnold FH Engineering New Catalytic Activities in Enzymes. Nat. Catal 2020, 3 , 203–213. (d) Miller DC ; Athavale SV ; Arnold FH Combining Chemistry and Protein Engineering for New-to-Nature Biocatalysis. Nat. Synth 2022, 1 , 18–23.35415721 (e) Klaus C ; Hammer SC New Catalytic Reactions by Enzyme Engineering. Trends Chem. 2022, 4 , 363–366. 3 Zhou Q ; Chin M ; Fu Y ; Liu P ; Yang Y Stereodivergent Atom-Transfer Radical Cyclization by Engineered Cytochromes P450. Science 2021, 374 , 1612–1616.34941416 4 Lu Y ; Yeung N ; Sieracki N ; Marshall NM Design of Functional Metalloproteins. Nature 2009, 460 , 855–862.19675646 5 (a) Denisov IG ; Makris TM ; Sligar SG ; Schlichting I Structure and Chemistry of Cytochrome P450. Chem. Rev 2005, 105 , 2253–2278.15941214 (b) Shaik S ; Kumar D ; de Visser SP ; Altun A ; Thiel W Theoretical Perspective on the Structure and Mechanism of Cytochrome P450 Enzymes. Chem. Rev 2005, 105 , 2279–2328.15941215 (c) Kovaleva EG ; Lipscomb JD Versatility of Biological Non-Heme Fe(II) Centers in Oxygen Activation Reactions. Nat. Chem. Biol 2008, 4 , 186–193.18277980 (d) Poulos TL Heme Enzyme Structure and Function. Chem. Rev 2014, 114 , 3919–3962.24400737 6 (a) Banerjee R Radical Carbon Skeleton Rearrangements: Catalysis by Coenzyme B12-Dependent Mutases. Chem. Rev 2003, 103 , 2083–2094.12797824 (b) Brown KL Chemistry and Enzymology of Vitamin B12. Chem. Rev 2005, 105 , 2075–2150.15941210 7 Solomon EI ; Heppner DE ; Johnston EM ; Ginsbach JW ; Cirera J ; Qayyum M ; Kieber-Emmons MT ; Kjaergaard CH ; Hadt RG ; Tian L Copper Active Sites in Biology. Chem. Rev 2014, 114 , 3659–3853.24588098 8 Phillips IR ; Shephard EA ; Ortiz de Montellano PR : Cytochrome P450 Protocols; Humana; Springer: Totowa, N.J.; London, 2013. 9 For reviews on new-to-nature carbene and nitrene transfer: (a) Brandenberg OF ; Fasan R ; Arnold FH Exploiting and Engineering Hemoproteins for Abiological Carbene and Nitrene Transfer Reactions. Curr. Opin. Biotechnol 2017, 47 , 102–111.28711855 (b) Yang Y ; Arnold FH Navigating the Unnatural Reaction Space: Directed Evolution of Heme Proteins for Selective Carbene and Nitrene Transfer. Acc. Chem. Res 2021, 54 , 1209–1225.33491448 10 For reviews: (a) Sibi MP ; Manyem S ; Zimmerman J Enantioselective Radical Processes. Chem. Rev 2003, 103 , 3263–3296.12914498 (b) Proctor RSJ ; Colgan AC ; Phipps RJ Exploiting Attractive Non-Covalent Interactions for the Enantioselective Catalysis of Reactions Involving Radical Intermediates. Nat. Chem 2020, 12 , 990–1004.33077927 (c) Mondal S ; Dumur F ; Gigmes D ; Sibi MP ; Bertrand MP ; Nechab M Enantioselective Radical Reactions Using Chiral Catalysts. Chem. Rev 2022, 122 , 5842–5976.35073048 11 (a) Yang D ; Gu S ; Yan Y-L ; Zhu N-Y ; Cheung K-K Highly Enantioselective Atom-Transfer Radical Cyclization Reactions Catalyzed by Chiral Lewis Acids. J. Am. Chem. Soc 2001, 123 , 8612–8613.11525676 (b) Chen B ; Fang C ; Liu P ; Ready JM Rhodium-Catalyzed Enantioselective Radical Addition of CX4 Reagents to Olefins. Angew. Chem. Int. Ed 2017, 56 , 8780–8784. 12 (a) Biegasiewicz KF ; Cooper SJ ; Gao X ; Oblinsky DG ; Kim JH ; Garfinkle SE ; Joyce LA ; Sandoval BA ; Scholes GD ; Hyster TK Photoexcitation of Flavoenzymes Enables a Stereoselective Radical Cyclization. Science 2019, 364 , 1166–1169.31221855 (b) Page CG ; Cooper SJ ; Dehovitz JS ; Oblinsky DG ; Biegasiewicz KF ; Antropow AH ; Armbrust KW ; Ellis JM ; Hamann LG ; Horn EJ ; Oberg KM ; Scholes GD ; Hyster TK Quaternary Charge-Transfer Complex Enables Photoenzymatic Intermolecular Hydroalkylation of Olefins. J. Am. Chem. Soc 2021, 143 , 97–102.33369395 (c) Gao X ; Turek-Herman JR ; Choi YJ ; Cohen RD ; Hyster TK Photoenzymatic Synthesis of α-Tertiary Amines by Engineered Flavin-Dependent “Ene”-Reductases. J. Am. Chem. Soc 2021, 143 , 19643–19647.34784482 (d) Nicholls BT ; Qiao T ; Hyster TK A Photoenzyme for Challenging Lactam Radical Cyclizations. Synlett 2022, 33 , A–E. (e) Nicholls BT ; Oblinsky DG ; Kurtoic SI ; Grosheva D ; Ye Y ; Scholes GD ; Hyster TK Engineering a Non-Natural Photoenzyme for Improved Photon Efficiency**. Angew. Chemie Int. Ed 2022, 61 , e202113842. For a review: (f) Ye Y ; Fu H ; Hyster TK Activation Modes in Biocatalytic Radical Cyclization Reactions. J. Ind. Microbiol. Biotechnol 2021, 48 , kuab021.33674826 13 (a) Huang X ; Wang B ; Wang Y ; Jiang G ; Feng J ; Zhao H Photoenzymatic Enantioselective Intermolecular Radical Hydroalkylation. Nature 2020, 584 , 69–74.32512577 For a review: (b) Harrison W ; Huang X ; Zhao H Photobiocatalysis for Abiological Transformations. Acc. Chem. Res 2022, 55 , 1087–1096.35353478 14 (a) Su H ; Ma G ; Liu Y Theoretical Insights into the Mechanism and Stereoselectivity of Olefin Cyclopropanation Catalyzed by Two Engineered Cytochrome P450 Enzymes. Inorg. Chem 2018, 57 , 11738–11745.30156099 (b) Biler M ; Crean RM ; Schweiger AK ; Kourist R ; Kamerlin SCL Ground-State Destabilization by Active-Site Hydrophobicity Controls the Selectivity of a Cofactor-Free Decarboxylase. J. Am. Chem. Soc 2020, 142 , 20216–20231.33180505 (c) Nakano Y ; Black MJ ; Meichan AJ ; Sandoval BA ; Chung MM ; Biegasiewicz KF ; Zhu T ; Hyster TK Photoenzymatic Hydrogenation of Heteroaromatic Olefins Using ‘Ene’-Reductases with Photoredox Catalysts. Angew. Chemie Int. Ed 2020, 59 , 10484–10488. (d) Hall M Enzymatic Strategies for Asymmetric Synthesis. RSC Chem. Biol 2021, 2 , 958–989.34458820 (e) Coleman T ; Kirk AM ; Chao RR ; Podgorski MN ; Harbort JS ; Churchman LR ; Bruning JB ; Bernhardt PV ; Harmer JR ; Krenske EH ; De Voss JJ ; Bell SG Understanding the Mechanistic Requirements for Efficient and Stereoselective Alkene Epoxidation by a Cytochrome P450 Enzyme. ACS Catal. 2021, 11 , 1995–2010. (f) Gergel S ; Soler J ; Klein A ; Schülke KH ; Hauer B ; Garcia-Borràs M ; Hammer SC Directed Evolution of a Ketone Synthase for Efficient and Highly Selective Functionalization of Internal Alkenes by Accessing Reactive Carbocation Intermediates. Summitted 2022-01-05. ChemRxiv DOI: 10.26434/chemrxiv-2022-dp94p (accessed 2022-06-27). 15 For reviews: (a) Sheng X ; Kazemi M ; Planas F ; Himo F Modeling Enzymatic Enantioselectivity Using Quantum Chemical Methodology. ACS Catal. 2020, 10 , 6430–6449. (b) Reetz MT ; Garcia-Borràs M The Unexplored Importance of Fleeting Chiral Intermediates in Enzyme-Catalyzed Reactions. J. Am. Chem. Soc 2021, 143 , 14939–14950.34491742 16 (a) Tang W ; Jiménez-Osés G ; Houk KN ; van der Donk WA Substrate Control in Stereoselective Lanthionine Biosynthesis. Nat. Chem 2015, 7 , 57–64.25515891 (b) Bauer P ; Carlsson ÅJ ; Amrein BA ; Dobritzsch D ; Widersten M ; Kamerlin SCL Conformational Diversity and Enantioconvergence in Potato Epoxide Hydrolase 1. Org. Biomol. Chem 2016, 14 , 5639–5651.27049844 (c) Sun Z ; Wu L ; Bocola M ; Chan HCS ; Lonsdale R ; Kong X-D ; Yuan S ; Zhou J ; Reetz MT Structural and Computational Insight into the Catalytic Mechanism of Limonene Epoxide Hydrolase Mutants in Stereoselective Transformations. J. Am. Chem. Soc 2018, 140 , 310–318.29232125 (d) Sato M ; Kishimoto S ; Yokoyama M ; Jamieson CS ; Narita K ; Maeda N ; Hara K ; Hashimoto H ; Tsunematsu Y ; Houk KN ; Tang Y ; Watanabe K Catalytic Mechanism and Endo-to-Exo Selectivity Reversion of an Octalin-Forming Natural Diels–Alderase. Nat. Catal 2021, 4 , 223–232.35873532 (e) Dong Y ; Li T ; Zhang S ; Sanchis J ; Yin H ; Ren J ; Sheng X ; Li G ; Reetz MT Biocatalytic Baeyer–Villiger Reactions: Uncovering the Source of Regioselectivity at Each Evolutionary Stage of a Mutant with Scrutiny of Fleeting Chiral Intermediates. ACS Catal. 2022, 12 , 3669–3680. 17 (a) Alessandro L ; Michele P Escaping Free-Energy Minima. Proc. Natl. Acad. Sci 2002, 99 , 12562–12566.12271136 (b) Ensing B ; De Vivo M ; Liu Z ; Moore P ; Klein ML Metadynamics as a Tool for Exploring Free Energy Landscapes of Chemical Reactions. Acc. Chem. Res 2006, 39 , 73–81.16489726 (c) Hutter J ; Iannuzzi M ; Schiffmann F ; VandeVondele J Cp2k: Atomistic Simulations of Condensed Matter Systems. WIREs Comput. Mol. Sci 2014, 4 , 15–25. (d) Valsson O ; Tiwary P ; Parrinello M Enhancing Important Fluctuations: Rare Events and Metadynamics from a Conceptual Viewpoint. Annu. Rev. Phys. Chem 2016, 67 , 159–184.26980304 (e) Bussi G ; Laio A ; Tiwary P Metadynamics: A Unified Framework for Accelerating Rare Events and Sampling Thermodynamics and Kinetics. In Handbook of Materials Modeling: Methods: Theory and Modeling, Andreoni W ; Yip S , Eds. Springer International Publishing: Cham, 2020; pp 565–595. 18 For previous computational studies on the roles of active site hydrogen bonding interactions, see: (a) Herschlag D ; Eckstein F ; Cech TR The Importance of Being Ribose at the Cleavage Site in the Tetrahymena Ribozyme Reaction. Biochemistry 1993, 32 , 8312–8321.7688573 (b) Adalsteinsson H ; Maulitz AH ; Bruice TC Calculation of the Potential Energy Surface for Intermolecular Amide Hydrogen Bonds Using Semiempirical and Ab Initio Methods. J. Am. Chem. Soc 1996, 118 , 7689–7693. (c) Shan S ; Herschlag D The Change in Hydrogen Bond Strength Accompanying Charge Rearrangement: Implications for Enzymatic Catalysis. Proc. Natl. Acad. Sci 1996, 93 , 14474–14479.8962076 (d) Khade RL ; Zhang Y Catalytic and Biocatalytic Iron Porphyrin Carbene Formation: Effects of Binding Mode, Carbene Substituent, Porphyrin Substituent, and Protein Axial Ligand. J. Am. Chem. Soc 2015, 137 , 7560–7563.26067900 (e) Yang Z ; Doubleday C ; Houk KN QM/MM Protocol for Direct Molecular Dynamics of Chemical Reactions in Solution: The Water-Accelerated Diels–Alder Reaction. J. Chem. Theory Comput 2015, 11 , 5606–5612.26588803 (f) Mehmood R ; Qi HW ; Steeves AH ; Kulik HJ The Protein’s Role in Substrate Positioning and Reactivity for Biosynthetic Enzyme Complexes: The Case of SyrB2/SyrB1. ACS Catal. 2019, 9 , 4930–4943. (g) Vennelakanti V ; Qi HW ; Mehmood R ; Kulik HJ When Are Two Hydrogen Bonds Better than One? Accurate First-Principles Models Explain the Balance of Hydrogen Bond Donors and Acceptors Found in Proteins. Chem. Sci 2021, 12 , 1147–1162.35382134 (h) Liu Z ; Calvó-Tusell C ; Zhou AZ ; Chen K ; Garcia-Borràs M ; Arnold FH Dual-Function Enzyme Catalysis for Enantioselective Carbon–Nitrogen Bond Formation. Nat. Chem 2021, 13 , 1166–1172.34663919 19 Coelho PS ; Wang ZJ ; Ener ME ; Baril SA ; Kannan A ; Arnold FH ; Brustad EM A Serine-Substituted P450 Catalyzes Highly Efficient Carbene Transfer to Olefins in Vivo. Nat. Chem. Biol 2013, 9 , 485–487.23792734 20 PyMOL Molecular Graphics System, version 2.0; Schrödinger, LLC, 2017. 21 Case DA ; Aktulga HM ; Belfon K ; Ben-Shalom IY ; Brozell SR ; Cerutti DS ; Cheatham TE ; Cisneros GA ; Cruzeiro VWD ; Darden TA ; Duke RE ; Giambasu G ; Gilson MK ; Gohlke H ; Goetz AW ; Harris R ; Izadi S ; Izmailov SA ; Jin C ; Kasavajhala K ; Kaymak MC ; King E ; Kovalenko A ; Kurtzman T ; Lee TS ; LeGrand S ; Li P ; Lin C ; Liu J ; Luchko T ; Luo R ; Machado M ; Man V ; Manathunga M ; Merz KM ; Miao Y ; Mikhailovskii O ; Monard G ; Nguyen H ; O’Hearn KA ; Onufriev A ; Pan F ; Pantano S ; Qi R ; Rahnamoun A ; Roe DR ; Roitberg A ; Sagui C ; Schott-Verdugo S ; Shen J ; Simmerling CL ; Skrynnikov NR ; Smith J ; Swails J ; Walker RC ; Wang J ; Wei H ; Wolf RM ; Wu X ; Xue Y ; York DM ; Zhao S ; Kollman PA AMBER 2020; University of California: San Francisco, 2020. 22 Li P ; Merz KM MCPB.Py: A Python Based Metal Center Parameter Builder. J. Chem. Inf. Model 2016, 56 , 599–604.26913476 23 Wang J ; Wolf RM ; Caldwell JW ; Kollman PA ; Case DA Development and Testing of a General Amber Force Field. J. Comput. Chem 2004, 25 , 1157–1174.15116359 24 Maier JA ; Martinez C ; Kasavajhala K ; Wickstrom L ; Hauser KE ; Simmerling C ff14SB: Improving the Accuracy of Protein Side Chain and Backbone Parameters from Ff99SB. J. Chem. Theory Comput 2015, 11 , 3696–3713.26574453 25 Knapp B ; Ospina L ; Deane CM Avoiding False Positive Conclusions in Molecular Simulation: The Importance of Replicas. J. Chem. Theory Comput 2018, 14 , 6127–6138.30354113 26 Roe DR ; Cheatham TE PTRAJ and CPPTRAJ: Software for Processing and Analysis of Molecular Dynamics Trajectory Data. J. Chem. Theory Comput 2013, 9 , 3084–3095.26583988 27 Morris GM ; Huey R ; Lindstrom W ; Sanner MF ; Belew RK ; Goodsell DS ; Olson AJ AutoDock4 and AutoDockTools4: Automated Docking with Selective Receptor Flexibility. J. Comput. Chem 2009, 30 , 2785–2791.19399780 28 (a) Knorrscheidt A ; Soler J ; Hünecke N ; Püllmann P ; Garcia-Borràs M ; Weissenborn MJ Accessing Chemo- and Regioselective Benzylic and Aromatic Oxidations by Protein Engineering of an Unspecific Peroxygenase. ACS Catal. 2021, 11 , 7327–7338.34631225 (b) Garcia-Borràs M ; Kan SBJ ; Lewis RD ; Tang A ; Jimenez-Osés G ; Arnold FH ; Houk KN Origin and Control of Chemoselectivity in Cytochrome c Catalyzed Carbene Transfer into Si–H and N–H Bonds. J. Am. Chem. Soc 2021, 143 , 7114–7123.33909977 (c) Soler J ; Gergel S ; Hammer S ; Garcia-Borràs M Enzymatic Control Over Reactive Intermediates Enables Direct Oxidation of Alkenes to Carbonyls by a P450 Iron-Oxo Species. Summitted 2022-01-27. ChemRxiv DOI: 10.26434/chemrxiv-2022-6lpjf (accessed 2022-06-27). 29 Chung LW ; Sameera WMC ; Ramozzi R ; Page AJ ; Hatanaka M ; Petrova GP ; Harris TV ; Li X ; Ke Z ; Liu F ; Li H-B ; Ding L ; Morokuma K The ONIOM Method and Its Applications. Chem. Rev 2015, 115 , 5678–5796.25853797 30 Frisch MJ ; Trucks GW ; Schlegel HB ; Scuseria GE ; Robb MA ; Cheeseman JR ; Scalmani G ; Barone V ; Petersson GA ; Nakatsuji H ; Li X ; Caricato M ; Marenich AV ; Bloino J ; Janesko BG ; Gomperts R ; Mennucci B ; Hratchian HP ; Ortiz JV ; Izmaylov AF ; Sonnenberg JL ; Williams-Young D ; Ding F ; Lipparini F ; Egidi F ; Goings J ; Peng B ; Petrone A ; Henderson T ; Ranasinghe D ; Zakrzewski VG ; Gao J ; Rega N ; Zheng G ; Liang W ; Hada M ; Ehara M ; Toyota K ; Fukuda R ; Hasegawa J ; Ishida M ; Nakajima T ; Honda Y ; Kitao O ; Nakai H ; Vreven T ; Throssell K ; Montgomery JA ; Peralta JE ; Ogliaro F ; Bearpark MJ ; Heyd JJ ; Brothers EN ; Kudin KN ; Staroverov VN ; Keith TA ; Kobayashi R ; Normand J ; Raghavachari K ; Rendell AP ; Burant JC ; Iyengar SS ; Tomasi J ; Cossi M ; Millam JM ; Klene M ; Adamo C ; Cammi R ; Ochterski JW ; Martin RL ; Morokuma K ; Farkas O ; Foresman JB ; Fox DJ , Gaussian 16 Rev. C.01, Wallingford, CT, 2016. 31 (a) Lee C ; Yang W ; Parr RG Development of the Colle-Salvetti Correlation-Energy Formula into a Functional of the Electron Density. Phys. Rev. B 1988, 37 , 785–789. (b) Becke AD Density-Functional Thermochemistry. III. The Role of Exact Exchange. J. Chem. Phys 1993, 98 , 5648–5652. 32 Grimme S Semiempirical GGA-Type Density Functional Constructed with a Long-Range Dispersion Correction. J. Comput. Chem 2006, 27 , 1787–1799.16955487 33 Vreven T ; Frisch MJ ; Kudin KN ; Schlegel HB ; Morokuma K Geometry Optimization with QM/MM Methods II: Explicit Quadratic Coupling. Mol. Phys 2006, 104 , 701–714. 34 Lin H ; Truhlar DG QM/MM: What Have We Learned, Where Are We, and Where Do We Go from Here? Theor. Chem. Acc 2006, 117 , 185. 35 Araújo JQ ; Carneiro J. W.de M. ; Araujo M. T. de ; Leite FHA ; Taranto AG Interaction between Artemisinin and Heme. A Density Functional Theory Study of Structures and Interaction Energies. Bioorg. Med. Chem 2008, 16 , 5021–5029.18375130 36 CP2K Open Source Molecular Dynamics. https://www.cp2k.org/quickstep (accessed 2022-06-27). 37 (a) Krack M ; Parrinello M In QUICKSTEP: Make the Atoms Dance; NIC Series; Forschungszentrum Juülich, 2004; p 29. (b) VandeVondele J ; Krack M ; Mohamed F ; Parrinello M ; Chassaing T ; Hutter J Quickstep: Fast and Accurate Density Functional Calculations Using a Mixed Gaussian and Plane Waves Approach. Comput. Phys. Commun 2005, 167 , 103–128. 38 (a) Laio A ; VandeVondele J ; Rothlisberger U A Hamiltonian Electrostatic Coupling Scheme for Hybrid Car–Parrinello Molecular Dynamics Simulations. J. Chem. Phys 2002, 116 , 6941–6947. (b) Laino T ; Mohamed F ; Laio A ; Parrinello M An Efficient Real Space Multigrid QM/MM Electrostatic Coupling. J. Chem. Theory Comput 2005, 1 , 1176–1184.26631661 39 (a) Becke AD Density-Functional Exchange-Energy Approximation with Correct Asymptotic Behavior. Phys. Rev. A 1988, 38 , 3098–3100. (b) Grimme S ; Antony J ; Ehrlich S ; Krieg H A Consistent and Accurate ab initio Parametrization of Density Functional Dispersion Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys 2010, 132 , 154104.20423165 40 VandeVondele J ; Hutter J Gaussian Basis Sets for Accurate Calculations on Molecular Systems in Gas and Condensed Phases. J. Chem. Phys 2007, 127 , 114105.17887826 41 Goedecker S ; Teter M ; Hutter J Separable Dual-Space Gaussian Pseudopotentials. Phys. Rev. B: Condens. Matter Mater. Phys 1996, 54 , 1703–1710. 42 Curran DP ; Tamine J Effects of Temperature on Atom Transfer Cyclization Reactions of Allylic α-Iodo Esters and Amides. J. Org. Chem 1991, 56 , 2746–2750. 43 (a) Miller BR ; McGee TD ; Swails JM ; Homeyer N ; Gohlke H ; Roitberg AE MMPBSA.Py: An Efficient Program for End-State Free Energy Calculations. J. Chem. Theory Comput 2012, 8 , 3314–3321.26605738 (b) For a previous study using MM-GBSA calculations to analyze active site residue–substrate interactions, see: Ref. 13d. 44 (a) Tang W ; Kwak Y ; Braunecker W ; Tsarevsky NV ; Coote ML ; Matyjaszewski K Understanding Atom Transfer Radical Polymerization: Effect of Ligand and Initiator Structures on the Equilibrium Constants. J. Am. Chem. Soc 2008, 130 , 10702–10713.18642811 (b) Lin CY ; Coote ML ; Gennaro A ; Matyjaszewski K Ab Initio Evaluation of the Thermodynamic and Electrochemical Properties of Alkyl Halides and Radicals and Their Mechanistic Implications for Atom Transfer Radical Polymerization. J. Am. Chem. Soc 2008, 130 , 12762–12774.18761460 45 The HOMO energies of a model Fe–porphine complex and Cu(TMPA)+ (TPMA: tris(2-pyridylmethyl)-amine) were computed at the B3LYP-D3/6–311+G(d,p)-LANL2TZ(f)/SMD(chlorobenzene)//B3LYP-D3/6–31G(d)-LANL2DZ level of theory. See Ref. 3. 46 For a computational study on the effects of HOMO energy of the transition metal complex on the rate of bromine atom abstraction, see: Fang C ; Fantin M ; Pan X ; de Fiebre K ; Coote ML ; Matyjaszewski K ; Liu P Mechanistically Guided Predictive Models for Ligand and Initiator Effects in Copper-Catalyzed Atom Transfer Radical Polymerization (Cu-ATRP). J. Am. Chem. Soc 2019, 141 , 7486–7497.30977644 47 QM/MM calculations of an alternative bromine atom abstraction pathway without the hydrogen bond with Q263 predict a 1.9 kcal/mol higher barrier than that involving Q263 via TS1. See Figure S16 in the SI for details. 48 The hydrogen bond distances in the metadynamics trajectories are longer than those in the QM/MM-optimized structures shown in Figure 4, because kinetic and potential energies were added to simulate the reaction at room temperature, whereas the QM/MM geometry optimizations obtain stationary points on the electronic energy surface.

PMC009xxxxxx/PMC9342916.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0413066 2830 Cell Cell Cell 0092-8674 1097-4172 35868277 9342916 10.1016/j.cell.2022.06.031 NIHMS1817594 Article What is a cell type and how to define it? Zeng Hongkui Allen Institute for Brain Science, Seattle, WA 98109, USA HongkuiZ@alleninstitute.org 23 6 2022 21 7 2022 21 7 2023 185 15 27392755 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Cell types are the basic functional units of an organism. They exhibit diverse phenotypic properties at multiple levels, making them challenging to define, categorize and understand. This Review provides an overview of the basic principles of cell types rooted in evolution and development, and discusses approaches to characterize and classify cell types and investigate how they contribute to the organism’s function, using the mammalian brain as a primary example. I propose a roadmap towards a conceptual framework and knowledge base of cell types that will enable a deeper understanding of the dynamic changes of cellular function under healthy and diseased conditions. In this Review, Zeng discusses how insights learned from the mammalian brain have begun to reveal generalizable organizing principles of cell types and proposes a roadmap based on these principles for taking a multilevel, iterative approach to define cell types and for generating a knowledge base of cell types across lifespan, species and the brain and body. pmcA cell is the basic unit of all living organisms (except for viruses) (Mazzarello, 1999). The evolution from unicellular to increasingly complex multicellular organisms involves multiplication of individual cells as well as groups of cells and diversification of the function of the cells. As such, billions of years of evolutionary process leads to the vast array of species whose diverse biological attributes are built upon their cellular compositions that exhibit similarities and differences both between species and among different organs within an individual organism (e.g., an animal or a plant). Thus, understanding the organization and function of cells within an organism lays the essential foundation for understanding how an organism works. Similarly, comparing the organization and function of cells between species allows understanding of functional diversity across species. Studies over the past century have revealed that cells within an organism can be grouped into types – cells within a type exhibit similar structure and function that are distinct from cells in other types (Arendt, 2008). Categorizing cells into types greatly reduces the complexity of investigating the organization and function of cells, especially for large organisms with billions to trillions of cells in the body, e.g., mammals. Researchers have measured a wide range of cellular properties and used these measurements to classify cell types (Petilla Interneuron Nomenclature Group et al., 2008; Regev et al., 2017; Zeng and Sanes, 2017). However, there has not been a consistent and standard definition of cell types even though it is critical for reproducible investigation. It is often unclear if cell types defined by different phenotypic features agree with each other, nor which feature is the “right” one to define cell types. Furthermore, lacking a systematic approach and effort, we do not know if all the cell types in an organism have been identified and where the gaps are. Recent advent in single cell transcriptomics is revolutionizing the way we understand cell types, with its unprecedented depth and scalability. It has been used to define cell types in a variety of species, tissue organs and brain regions (Armand et al., 2021; Svensson et al., 2020; Tanay and Sebe-Pedros, 2021). However, despite many illuminating studies it remains an open question to what extent transcriptomic clusters represent true cell types and what level of granularity is appropriate for defining cell types. Nonetheless, over the past few years, tremendous progress has been made and many new insights have been generated around these questions. In this review, I will mainly use the mammalian brain as an example (but also refer to other organs or species) to address key questions pertaining to the conceptual and operational definition of cell types. Approaches to characterize cell types Cell types in the brain and the body exhibit diverse properties in many modalities – molecular, morphological, physiological, and functional. Numerous studies in these different modalities in the brain over the past century, dating back to Ramón y Cajal and his contemporaries, have converged on a consistent high-level picture of cell type organization across brain regions (Fishell and Heintz, 2013; Markram et al., 2004; Masland, 2012; Mukamel and Ngai, 2019; Nelson et al., 2006; Petilla Interneuron Nomenclature Group et al., 2008; Sanes and Masland, 2015; Seung and Sumbul, 2014; Somogyi and Klausberger, 2005; Yuste et al., 2020; Zeng and Sanes, 2017). At the same time, cellular properties at individual cell level are highly heterogeneous, variations in different modalities do not necessarily exhibit high degrees of concordance, making it often impossible to define exactly what is a cell type and draw clear boundaries between “types”. In many cases, lacking a way to reproducibly label a cell type (typically using a molecular genetic approach) presents a major hurdle to relate different studies and findings to each other. To untangle this complexity, it is necessary to adopt approaches that provide comprehensive, unbiased, quantitative and standardizable measurements and are scalable to densely sample a sufficient number of cells within a brain region or tissue organ as well as across the entire brain and body to eventually reach completeness, and then perform data-driven computational clustering and analysis to obtain cell type classification. The Petilla convention to define criteria for defining cortical interneuron types represents a major community effort to specify such approaches (Petilla Interneuron Nomenclature Group et al., 2008). Given that physiological properties can take many different forms under different conditions, and functional properties are unknown or poorly defined for many types of cells, as well as the fact that these two modalities are better examined in vivo, it is challenging to scale up the physiological and functional approaches, such as in vivo electrode recording or functional imaging, in a comprehensive and unbiased manner as a primary way to define cell types. On the other hand, molecular and anatomical approaches are more suited for this purpose (Fig. 1A). Molecular approaches include the profiling of chromatin modifications (epigenomics), RNA transcripts (transcriptomics), and proteins (proteomics). Anatomical approaches include the characterization of the spatial distribution, morphology and connectivity of individual cells. Currently single-cell transcriptomics and connectomics (i.e., delineating the patterns of interconnections between individual neurons) are the two primary approaches that have the potential to meet the completeness requirement. Both approaches are now being realized in simpler model organisms including C. elegans (Taylor et al., 2021; White et al., 1986; Witvliet et al., 2021) and Drosophila (Hulse et al., 2021; Li et al., 2022; Scheffer et al., 2020), whereas in mammals transcriptomics is currently feasible and connectomics is still in development (Abbott et al., 2020). Transcriptomics by single-cell or single-nucleus RNA-sequencing (scRNA-seq or snRNA-seq) is now the most widely used approach to generate cell type taxonomies or atlases from many species, tissue organs and brain regions, due to its comprehensiveness and high dimensionality (i.e., profiling thousands of expressed genes per cell in a largely unbiased manner) as well as its high scalability (to hundreds of thousands or millions of cells). Transcriptomic cell atlases at the whole organism level have been generated for Drosophila (Li et al., 2022), Ciona (Cao et al., 2019a), and the nervous system of C. elegans (Taylor et al., 2021). The Human Cell Atlas community effort aims to create cell atlases for all organs in the human body (Lindeboom et al., 2021; Regev et al., 2017). The BRAIN Initiative cell census effort has the goal of creating high-resolution whole-brain cell type atlases for mouse, human and non-human primates (Brain Initiative Cell Census Network, 2021; Ecker et al., 2017; Ngai, 2022). A variety of transcriptomic cell atlases have been generated in mouse from many different regions of the nervous system, such as cortex, hippocampus, striatum, thalamus, hypothalamus, cerebellum, spinal cord, retina, etc. (Cembrowski et al., 2018; Hashikawa et al., 2020; Kozareva et al., 2021; Macosko et al., 2015; Marques et al., 2016; Phillips et al., 2019; Pool et al., 2020; Poulin et al., 2014; Ren et al., 2019; Romanov et al., 2017; Russ et al., 2021; Sathyamurthy et al., 2018; Saunders et al., 2018; Shekhar et al., 2016; Stanley et al., 2020; Tasic et al., 2018; Van Hove et al., 2019; Yao et al., 2021a; Yao et al., 2021b; Zeisel et al., 2018; Zeisel et al., 2015), and body organs (Han et al., 2018a; Jaitin et al., 2014; Tabula Muris et al., 2018), and increasingly more in human and non-human primates (Bakken et al., 2021; Darmanis et al., 2015; Drokhlyansky et al., 2020; Garcia et al., 2022; Han et al., 2022; Hodge et al., 2019; Kamath et al., 2022; Lake et al., 2016; Masuda et al., 2019; Tabula Sapiens et al., 2022; Winkler et al., 2022; Yang et al., 2022). Single-cell epigenomics, such as single-nucleus ATAC-seq (to characterize chromatin accessibility) or DNA methylation-sequencing, has also been used to generate cell type atlases for different brain regions that are consistent with transcriptomic cell atlases and further reveal cell type-specific gene and chromatin regulatory landscapes (Cusanovich et al., 2018; Lake et al., 2018; Li et al., 2021; Liu et al., 2021; Luo et al., 2017; Preissl et al., 2018; Yao et al., 2021a). Spatially resolved transcriptomics, including a variety of techniques based on in situ imaging, in situ capture or in situ sequencing (Close et al., 2021; Larsson et al., 2021; Lein et al., 2017; Moses and Pachter, 2022; Rao et al., 2021; Zhuang, 2021), is a powerful approach combining molecular and spatial characterization at single cell or near-single cell level, revealing spatial relationships between cell types in both local environment and global architecture (Chen et al., 2021; Moffitt et al., 2018; Ortiz et al., 2020; Rao et al., 2021; Wang et al., 2021; Zhang et al., 2021a). Other attributes in the transcriptomes, such as alternatively spliced variants, can provide further information and help to refine cell types (Booeshaghi et al., 2021). An area awaiting critical technology development is single-cell proteomics (Cho et al., 2022; Slavov, 2021), as the expression and subcellular distribution of proteins provides a crucial link between gene expression and cellular structure and function, and it may not have lock-step correlation with the transcriptome of the same cell type. A cell’s morphology (i.e., shape) and connectivity (especially for neurons) has been regarded as the most defining feature of brain cell types ever since Cajal, though its place may be overtaken by transcriptome. A cell’s morphology can be reconstructed from high-resolution light microscopy (LM) (coupled with colorimetric or fluorescent sparse labeling) (Gao et al., 2022; Jenett et al., 2012; Peng et al., 2021; Winnubst et al., 2019; Wolff and Rubin, 2018) or electron microscopy (EM) datasets (Hulse et al., 2021; Scheffer et al., 2020; Seung and Sumbul, 2014). Connections among individual neurons can be identified using approaches such as EM (Gour et al., 2021; Helmstaedter et al., 2013; Hildebrand et al., 2017; Hulse et al., 2021; Morgan et al., 2016; Scheffer et al., 2020; Schneider-Mizell et al., 2021; Turner et al., 2022; Witvliet et al., 2021), single-neuron trans-synaptic tracing (Schwarz and Remy, 2019) and barcoded connectomics (Chen et al., 2019; Clark et al., 2021; Gergues et al., 2020; Han et al., 2018b; Kebschull et al., 2016; Sun et al., 2021). Again, for definitive cell type classification, one needs to use a fully representative, rather than partial and biased, set of morphological and connectional features. In this regard, with the acquisition of whole brain EM connectomic and LM morphological datasets in Drosophila, refined cell type classification in the brain of this species has been primarily driven by morphology and connectivity (Hulse et al., 2021; Jenett et al., 2012; Scheffer et al., 2020; Wolff and Rubin, 2018). Most critically, these various approaches need to be integrated to achieve a coherent understanding of cell types and their function and to resolve issues such as which approach(es) (e.g., between transcriptomics and connectomics) can define cell types more clearly. The most common type of integration is to relate transcriptomic profiles with other modalities. Technically there are three ways to achieve such integration (Fig. 1B). First, conduct multimodal characterization from the same cell using approaches such as single-cell multi-omics (Zhu et al., 2020), Patch-seq which collects electrophysiological, morphological and transcriptomic data from a single patched cell (Berg et al., 2021; Cadwell et al., 2016; Fuzik et al., 2016; Gouwens et al., 2020; Lee et al., 2021; Munoz-Manchado et al., 2018; Scala et al., 2021), retrograde connectivity tracing coupled with single-cell molecular profiling (e.g., Retro-seq, Epi-retro-seq, or Retro-MERFISH) (Kim et al., 2020; Tasic et al., 2018; Zhang et al., 2021a; Zhang et al., 2021b), or in vivo calcium imaging followed by multiplexed FISH (Bugeon et al., 2021; Condylis et al., 2022; Lovett-Barron et al., 2020; von Buchholtz et al., 2021; Xu et al., 2020). Second, perform label transfer between independently collected datasets through “Rosetta stone” features, e.g., integration between single-cell transcriptomic and epigenomic datasets through marker genes and nearby chromatin modification sites (Armand et al., 2021; Yao et al., 2021a), or assigning molecular identities to neurons in EM and LM datasets using morphologies obtained from Patch-seq data. Integration between transcriptomics and epigenomics is now further empowered by various single-cell multi-omic techniques (Armand et al., 2021; Zhu et al., 2020). Third, create cell type-targeting genetic tools (e.g., transgenic lines or recombinant viral vectors) using marker genes, promoters and enhancer elements identified from transcriptomic and epigenomic cell atlases (Chan et al., 2017; Daigle et al., 2018; Dimidschstein et al., 2016; Graybuck et al., 2021; Hrvatin et al., 2019; Matho et al., 2021; Mich et al., 2021; Vormstein-Schneider et al., 2020), and use these tools for structural and functional studies. Currently available genetic tools are mostly targeting more coarse-level cell classes or subclasses, or a mixture of cell types. These tools have nonetheless proven to be tremendously powerful as the vast majority of our current knowledge of cell types in the brain and body and their functions has been derived from studies utilizing these tools. The emergence of comprehensive transcriptomic and epigenomic cell atlases now makes it possible to create highly specific tools targeting nearly all identified cell types, and even extended to non-genetic-model organisms and species (Ngai, 2022). This will have a paradigm-shifting effect to the study of function and dysfunction of broad biological systems. Overall, application of these approaches to characterize cell types in different brain regions and tissue organs as well as across species has begun to reveal generalizable organizing principles of cell types. Below I will discuss the large body of studies supporting these principles, and then conclude with a proposed roadmap based on these principles for taking a multilevel, iterative approach to define cell types and build an overarching knowledge base of cell types across the brain and body, across lifespan and across species. Cell types are the product of evolution The concept of cell types needs to be established based on where cell types originated and how they have diversified. Cell type classification has been compared to species classification (Stadler et al., 2021; Tanay and Sebe-Pedros, 2021; Zeng and Sanes, 2017). Indeed, species specialization is an overall culmination of the function of all the cell types within that species, thus they may follow similar evolutionary principles. There have been several ways proposed to classify species. One is based on the notion of reproductive isolation. However, this approach is not universally implementable, and many exceptions have also been found. A more fruitful approach is phylogenetic analysis, that is, comparing the relatedness between species using a wide range of structural and functional phenotypic features. Such analysis led to the foundational “tree of life” as we understand it today. Nonetheless, many issues remain unresolvable in the phylogenetic classification of species, often due to the highly specialized phenotypic features acquired by some species as they adapt to their ecological niches, as well as convergent phenotypic evolution in other cases, both of which could skew comparative analysis. Over the past decade, evolutionary approach based on comparative genomics (i.e., phylogenomics) has brought an entirely new paradigm to species classification, providing a systematic, rational, unbiased, universally applicable and extensible classification scheme (Murphy et al., 2021; Preuss and Wise, 2022; Stephan et al., 2022). Similarly, cell types are inherited through the genome. Relatedness between cell types reflects their evolutionary distance as they were created through cell type duplication and segregation events. It has been proposed that the formation of a new cell type identity requires the evolution of a unique cell type regulatory signature that includes a cell type-specific core regulatory complex (CoRC) of transcription factors, which defines the identity and coordinated gene expression pattern of the new cell type (Arendt et al., 2016). This set of master regulatory transcription factors, sometimes called terminal selectors, have been identified in a number of neuron types in worms, flies and mice (Hobert and Kratsios, 2019; Reilly et al., 2020). The master transcription factors should be identifiable when the transcriptomes of evolutionarily related cell types are compared. A large body of studies (see above) have now shown that clustering of single-cell transcriptomes can systematically categorize cells into putative types, many of which are consistent with existing knowledge and thus can be considered as bona fide cell types. Evolutionarily conserved cell types can be systematically identified by cross-species comparison of single-cell transcriptomic types in the brain (Bakken et al., 2021; Colquitt et al., 2021; Hodge et al., 2019; Kebschull et al., 2020; Krienen et al., 2020; Tosches et al., 2018; Yamagata et al., 2021). Thus, this approach appears to “make sense”; it is not coincidental, but strongly supports the notion that transcriptomes harbor the molecular genetic code for cell type identity. However, there are several challenges that must be surmounted to arrive at a complete and accurate evolutionary definition of cell types through cross-species comparisons. Accurate cross-species comparison of cell types at transcriptomic level requires well-annotated genomes, comparative gene ontologies and consistently high-quality transcriptomic data generation from many species (Tanay and Sebe-Pedros, 2021). Furthermore, species mostly diverged millions of years ago, as did cellular identities. Cell type homologies between related species are often discernible only at a relatively coarse level which do not fully capture the biological complexity. Many gaps also exist due to the extinction of intermediate species. These challenges could limit a deeper understanding of cell types (see below) and how they contribute to the body or brain function. On the other hand, one does not need to characterize cells from a large number of evolutionarily related species in order to define cell types. It is possible to gain a deep understanding of cell types from even a single species, since each species has evolved from its simpler ancestors through many rounds of cellular and regional duplications in which the newly created cell types and regions adopt new functions, and thus comparing between cell types and between regions within the species (in the same way as comparing between species) can reveal the evolutionary relationships between cell types as well. Then, we can expand the investigation into as many other species as possible, which will further clarify the description of cell types, their origins and how their functions manifest. Therefore, the first and foremost principle is that cell types are the product of evolution and cell type identities are encoded in the genome. Like phylogenomics for species classification, transcriptomic (and epigenomic) classification is a good proxy of cell type classification as the gene regulatory mechanisms that encode and maintain cell type identities are embedded in the transcriptomes and epigenomes. This core concept has led to a systematic delineation of the relationship between cell types both within a species and, increasingly, across species. At the same time, cell type conservation may be imposed more by function than natural selection directly as in organismal evolution. As such, evolution of individual cell types may be more complicated than organismal evolution as a whole, and it will be interesting to see if different cell types evolve in similar or different ways as the whole organism. Finally, a transcriptome also contains gene expression profiles that underlie arguably all phenotypic features of the cell at the time or state when the cell is characterized. What else are transcriptomes and transcriptomic clusters telling us? Hierarchical organization of transcriptomically-defined cell types Transcriptomically derived cell type taxonomies in the adult mammalian brain, with majority of the studies conducted in mouse, have consistently revealed a hierarchical organization of cell types (Fig. 2) (Brain Initiative Cell Census Network, 2021; Macosko et al., 2015; Romanov et al., 2017; Russ et al., 2021; Saunders et al., 2018; Shekhar et al., 2016; Tasic et al., 2016; Tasic et al., 2018; Yao et al., 2021b; Zeisel et al., 2018; Zeisel et al., 2015; Zeng and Sanes, 2017). The first (highest) level of branches is the separation of neuronal and various non-neuronal cell classes (Fig. 2A). For neurons, the second level of branches is driven by major brain structures/regions, and the third level comprises various cell subclasses and types within each major brain structure, although there may be cell types crossing or shared between brain structures due to cell migration during development. The basic architecture of the mammalian brain (Swanson, 2000, 2012) is composed of telencephalon, diencephalon, mesencephalon (midbrain) and rhombencephalon (hindbrain). Telencephalon (consisting of five major brain structures – isocortex, hippocampal formation, olfactory area, cortical subplate and cerebral nuclei) and diencephalon (including thalamus and hypothalamus) are collectively called forebrain. Midbrain is divided into tectum and tegmentum. And hindbrain is divided into pons, medulla and cerebellum. Within each of these major brain structures there are multiple regions and subregions, each with many cell types. A cell type can be specific to a subregion, a region or a major brain structure. Here I use isocortex (or simply called cortex) as an example to illustrate the organization of cell types within a major brain structure. Isocortex is composed of multiple cortical areas, each mediating sensory, motor or associational function. Transcriptomic cell type taxonomies from visual cortex and motor cortex display a similar organization (Fig. 2B) (Brain Initiative Cell Census Network, 2021; Tasic et al., 2018; Yao et al., 2021a). In each of these areas, there are two neuronal classes based on the dominant neurotransmitters they release, glutamatergic and GABAergic, as well as several non-neuronal classes. The glutamatergic excitatory neurons mostly have long-range axon projections to other cortical and/or subcortical regions. They are divided into nine subclasses based on their layer specificity and long-range projection patterns: L2/3 IT, L4/5 IT, L5 IT, L6 IT, Car3 IT, L5 ET, L5/6 NP, L6 CT, and L6b (IT: intratelencephalic projecting; ET: extratelencephalic projecting; NP: near-projecting; CT: corticothalamic projecting). The GABAergic inhibitory neurons mostly have their axon projections confined within the local area. They are divided into six subclasses named after canonical marker genes: Lamp5, Sncg, Vip, Sst, Sst-Chodl and Pvalb. Within each of the glutamatergic or GABAergic subclasses, as well as each non-neuronal class, there are several transcriptomic clusters or types, resulting in a total of ~110 transcriptomic cell types in each cortical area (Brain Initiative Cell Census Network, 2021; Tasic et al., 2018). This organization is highly consistent with the existing knowledge about cortical cell types which have been extensively studied in a variety of phenotypic modalities over the past 50 years (Harris and Shepherd, 2015; Tremblay et al., 2016; Yuste et al., 2020; Zeng and Sanes, 2017), suggesting that single-cell transcriptomics alone can faithfully capture the overall cell type organization at class and subclass levels, although the validity of transcriptomic clusters at the lowest branch level has yet to be fully tested. Comparison of transcriptomic cell types across different cortical areas reveals new insights. Glutamatergic neuron types are distinct between visual and motor cortical areas whereas GABAergic neuron types are shared between the two areas (Tasic et al., 2018). This dichotomy may be rooted in the developmental origins of these two cell classes. During development, glutamatergic neurons are generated within the cortex in which different areas are laid out by gradient expression of morphogens (Cadwell et al., 2019; O’Leary et al., 2007), whereas GABAergic neurons are generated in the subcortical ganglionic eminence and migrate into cortex (Hu et al., 2017; Lim et al., 2018). A larger transcriptomic study covering all areas from both isocortex and hippocampal formation (HPF) further identifies hundreds of transcriptomic types and a high degree of diversity in the glutamatergic neuron class across both brain structures (Fig. 2C) (Yao et al., 2021b). Within isocortex, cell types that are specific to a cortical area or shared among areas are both identified, and the shared cell types often exhibit gradient distribution or gradient gene expression across areas. This highly complex transcriptomic cell type landscape along the cortical sheet likely results from the series of cortical developmental events from “Protomap” to “Protocortex” (Cadwell et al., 2019; O’Leary et al., 2007). Compared between the two brain structures, the glutamatergic cell types in isocortex and HPF are highly distinct from each other, yet they also display one-to-one homology at subclass level suggesting a similar evolutionary origin (Yao et al., 2021b). This homologous relationship suggests that parallel neural networks can be formed by homologous sets of cell types. Overall, based on these findings, we may hypothesize that the adult-stage transcriptomic landscape can reveal the organization of cell types within and between brain regions that reflect their evolutionary and developmental histories. The hierarchical organization of transcriptomic cell types likely represents evolutionary origins of and distinctions between cell types; major branches represent earlier division of cell classes, and minor branches represent more recent segregation events. This hierarchical organization is laid out via an elaborate developmental program involving a series of highly coordinated processes and events. This hypothesis can be tested by studying the evolution and development of cell types (see below). Another prominent feature of the relationship between cell types revealed by transcriptomic studies is the coexistence of discrete and continuous variations between types. Continuous variations have been observed in a variety of forms in cortical excitatory and inhibitory neurons, medium spiny neurons in the striatum and excitatory projection neurons across different nuclei of the thalamus (Phillips et al., 2019; Stanley et al., 2020; Tasic et al., 2018; Yao et al., 2021b). Discrete variations exist among cell subclasses and major types that are usually at the higher branches of the hierarchy. Continuous variations are usually found among closely related transcriptomic clusters or subtypes at lower branches, such as the many IT neuron types across the cortical depth from L2/3 to L6 (Fig. 2C). Cells at opposite ends of the continuum have clearly distinct transcriptomic profiles, but the transition from one end to the other is gradual among the cells composing the continuum. This makes it difficult to subdivide the cells into types using statistical criteria and name an exact number of cell types. But the continuous variation itself is nonetheless biologically meaningful and needs to be properly represented in cell type descriptions. One way to better understand the significance of the continuous variation is to examine how it correlates with other modalities of cell type properties (see below). Regarding non-neuronal cells in the mammalian brain, there are multiple classes which can be divided into neural and non-neural groups (Fig. 2A–B) (Brain Initiative Cell Census Network, 2021; Zeisel et al., 2018). The non-neural group contains cell classes of the immune origin, i.e., microglia and border-associated macrophages (BAMs) (Butovsky and Weiner, 2018; Masuda et al., 2019; Munro et al., 2022; Prinz et al., 2019; Thion and Garel, 2020; Van Hove et al., 2019), and of the vascular origin, i.e., endothelial cells, smooth muscle cells (SMCs), pericytes, and vascular leptomeningeal cells (VLMCs) (Schaeffer and Iadecola, 2021; Sweeney et al., 2019; Vanlandewijck et al., 2018). The neural group contains cell classes of the neuroectoderm origin (same as the origin of neurons), including oligodendrocytes, oligodendrocyte progenitor cells (OPCs), astrocytes and ependymal cells (Ben Haim and Rowitch, 2017; Dimou and Simons, 2017; Escartin et al., 2021; Khakh and Deneen, 2019; Kuhn et al., 2019; Ortiz-Alvarez et al., 2019; Redmond et al., 2019). In brain transcriptomic cell type taxonomies, non-neuronal cells generally display less diversity than neurons, with little regional specificity except for astrocytes. There are two major subclasses of astrocytes, one specific to the telencephalon and the other to non-telencephalon regions, in addition to several other highly specialized astrocyte-like cell types such as Müller glia of the retina and Bergmann glia of the cerebellum (Zeisel et al., 2018). Immature and mature oligodendrocytes form a long continuous trajectory originating from OPCs, indicating coexistence of multiple states of gradually maturing oligodendrocytes (Marques et al., 2016; Zeisel et al., 2018). Like comparative genomics for species classification, single-cell transcriptomics is highly effective for cross-species comparison of cell types to reveal their evolutionary relationships. Comparative studies of cortical cell types among mouse, human and non-human primates (Bakken et al., 2021; Brain Initiative Cell Census Network, 2021; Hodge et al., 2019; Krienen et al., 2020) show that the hierarchical organization described above along with all the neuronal and non-neuronal cell classes and subclasses (major branches of the hierarchy) is well conserved across these mammalian species. Main differences across species lie in the heterogeneity of individual gene expression within each subclass and type, as well as the ambiguity of cross-species correspondence of the leaf-node transcriptomic types which likely will require multimodal characterization to clarify. Similarly, comparative transcriptomic studies reveal homologous cortical glutamatergic and GABAergic cell classes between mammal and reptile (Tosches et al., 2018), as well as homologies and variations of neuron subclasses and types in the cerebellar nuclei or the retina of mouse, chicken and primates (Kebschull et al., 2020; Yamagata et al., 2021). These studies further suggest that the hierarchical organization of brain cell types is a framework extensible to describing cell type evolution. Correspondence between transcriptomic cell types and other cellular properties Cell types are also considered to be the basic functional units of an organism. For the categorization of cell types based on their transcriptomes to be meaningful and to understand their relevance to the structure and function of the tissue organ where the cells reside, it is necessary to characterize other modalities of cellular properties. Multimodal correspondence of cell types in the mouse retina is a classic example where independent anatomical, functional and transcriptomic studies uncover similar numbers of neuron types (~130) and find that molecular profiles and anatomical distribution patterns (laminar specificity and mosaicism) are well correlated with visual response properties (Baden et al., 2020; Masland, 2012; Sanes and Masland, 2015; Seung and Sumbul, 2014; Shekhar and Sanes, 2021). Recent work from the BRAIN Initiative Cell Census Network (BICCN) in creating a multimodal cell census and atlas of the mammalian primary motor cortex represents the most comprehensive multimodal study to date, integrating transcriptomics with epigenomics, spatially resolved transcriptomics, morpho-electrical properties, and connectivity (Brain Initiative Cell Census Network, 2021). Integration of transcriptomic and epigenomic datasets using computational approaches allows consolidation of robust molecular cell types and identification of hundreds of thousands of cis-regulatory elements (CREs) associated with specific cell types (Yao et al., 2021a). Some of these CREs are associated with specific marker genes whereas others may represent past events. Integration of transcriptomics and MERFISH, a spatially resolved transcriptomic method, reveals the spatial organization of mouse motor cortex cell types (Zhang et al., 2021a). A major finding of the study is that, in addition to the laminar distribution of glutamatergic neuron subclasses as expected, even the GABAergic types within each subclass exhibit layer-selective localization, and the continuous variations among individual glutamatergic types or GABAergic types correlate well with their continuous distribution along cortical layers/depth (with a prominent example being that all the excitatory L2/3-L6 IT types line up along the cortical depth from L2/3 to L6) (Fig. 3A). Thus, a strong correspondence is demonstrated here between the continuous variations among cortical neuron types in the transcriptomic space and their continuous and directed spatial distribution patterns. Other MERFISH studies of neuron types in hypothalamus or nucleus accumbens also reveal strong correlation between transcriptomic specificity and anatomical/subregional specificity (Chen et al., 2021; Moffitt et al., 2018). Integration of transcriptomic, electrophysiological and morphological properties by Patch-seq reveals multimodal corresponding distinctions of mouse motor cortex cell types at subclass level (Scala et al., 2021). Within each subclass the morpho-electrical properties vary continuously along with the transcriptomic types. There is also additional heterogeneity of morpho-electrical properties within some transcriptomic types, indicating a more complex picture. Another Patch-seq study of mouse visual cortical GABAergic neurons also reveals a relatively high degree of corresponding continuous variations of transcriptomic types with their anatomical distribution along the cortical depth and the variations of their morpho-electrical properties (Fig. 3B) (Gouwens et al., 2020). To overcome heterogeneity at individual type level, a set of triple-modality MET types are defined at an intermediate level of granularity (between transcriptomic subclasses and types). These visual cortex GABAergic MET types show robust cross-modality concordance and mutual predictability. The vast majority of cortical and subcortical neuron types have long-range axon projections to form circuit networks throughout the brain. To examine the long-range axon projection specificity of transcriptomic cell types, Retro-seq and related methods (e.g., Epi-retro-seq, Retro-MERFISH) have been used (Kim et al., 2020; Tasic et al., 2018; Zhang et al., 2021a; Zhang et al., 2021b). Since a neuron type usually has multiple projection targets and a neuron within that type can choose a subset of those targets either specifically or randomly, a single-target-site Retro-seq assay is often insufficient to resolve the target specificity of a transcriptomic type except in special cases. Brain-wide complete reconstruction of single neuron morphology is currently the only approach to capture the full extent of a neuron’s axon projection pattern and define projection neuron types (Gao et al., 2022; Peng et al., 2021; Winnubst et al., 2019). A study using this approach in cortical excitatory neuron subclass-specific Cre driver lines (Peng et al., 2021) reveals distinct projection patterns between subclasses, e.g., not only between IT and ET but also between L2/3 IT and L5 IT neurons, confirming subclass level projection specificity (Fig. 3C). Within each subclass, the study also finds extensive heterogeneity among individual neurons; this heterogeneity reflects three axes of variations: regional specificity, topographic specificity and individual (potentially stochastic) variation, which do not readily correlate with transcriptomic types within the subclass. Thus, it remains an open question how axon projection patterns correlate with transcriptomic types, which needs to be addressed in future studies using approaches such as coupling complete morphology reconstruction with multiplexed FISH, or performing MAPseq/BARseq with sequencing of both starter cells and axon targets. It may also be necessary to extend such studies into developmental periods, to identify potentially clearer molecular correlates when the projection specificity is established (Klingler et al., 2021). Systematic investigation of connectivity among transcriptomic types at synaptic level and relating them to conventional studies where morphology and individual molecular markers were used to identify cell types is much needed to better understand the connectional specificity between transcriptomic types. It has been suggested that neuron types may be defined by their unique communication properties implemented as synaptic input-output patterns (Huang and Paul, 2019; Paul et al., 2017). The emerging large-scale EM datasets (e.g., from the MICrONS project, https://www.iarpa.gov/research-programs/microns) hold great promise to tackle synaptic-level connectivity between cell types and individual cells in the mammalian brain (Abbott et al., 2020). Perhaps a greater opportunity lies in the Drosophila field where comprehensive catalogs of both transcriptomic cell types and connectional cell types have been obtained independently (Hulse et al., 2021; Li et al., 2022; Scheffer et al., 2020) and a systematic comparison and cross-correlation between them may be realized soon. To compare functional properties among transcriptomic cell types, two general approaches have been taken – coupling in vivo calcium imaging with post hoc multiplexed FISH to decode the molecular identities of the imaged cells (Bugeon et al., 2021; Condylis et al., 2022; Lovett-Barron et al., 2020; von Buchholtz et al., 2021; Xu et al., 2020), or mapping immediate early gene (IEG) activation during sensory response or behavior using scRNA-seq or MERFISH (Hrvatin et al., 2018; Kim et al., 2019; Moffitt et al., 2018; Sathyamurthy et al., 2018; Wu et al., 2017). Using the former approach, it has been shown that GABAergic transcriptomic types in mouse visual cortex differ in their response to behavioral states (e.g., running versus resting), whereas visual response properties (e.g., orientation or direction selectivity) only differ at subclass level (Bugeon et al., 2021); in somatosensory cortex, higher sensory response is seen in a specific L2/3 IT excitatory transcriptomic type (Condylis et al., 2022). In hypothalamus, several studies using either of the two approaches in mice demonstrate that activated neurons during a specific behavioral state are often distributed across a range of transcriptomic cell types (Kim et al., 2019; Moffitt et al., 2018; Xu et al., 2020). Understanding the functional roles of different transcriptomic cell types is a huge undertaking. Obviously, studies mentioned here are just the beginning; many more will come in the future and will allow us to gain a much deeper understanding. In summary, for a definition of cell types to be meaningful, it must be associated with what cell types do. A transcriptomic cell type taxonomy must be linked to anatomical and functional information to evaluate the validity of the transcriptomic taxonomy and determine the appropriate level of granularity (since in theory transcriptomic clusters can be infinitely subdivided and the more cells profiled the more clusters can be obtained). So far it has been shown that transcriptomic types have excellent correspondence with their spatial distribution patterns. Since the spatial distribution pattern is defined during development, this suggests that transcriptomes may retain the developmental plan. At the same time, whether specific transcriptomic types (beyond the subclass level) have specific connectional or functional attributes or not is still unclear in many cases. Since transcriptomes are rich in containing the molecular correlates of all sorts of cellular properties, specific molecular signatures responsible for certain essential anatomical or functional features may be hidden below noise and will need to be brought out through supervised approaches and used to refine the classification of cell types towards more functional relevance. It is also necessary to trace back into development to identify potentially clearer molecular correlates as different connectional or functional properties may be established at different developmental time points. On the other hand, it is also reasonable to propose that some connectional or functional properties should not be used to define cell types, because they may be emerging properties arising from the interaction of a network of cell types, or from experience and/or activity dependent processes that represent a cell “state” rather than a defining feature for a cell “type”. Cell types versus cell states A key question arising when evaluating a transcriptomic taxonomy is whether some clusters actually represent a particular cell state – i.e., a transient or dynamically responsive property of a cell to a context – rather than a cell type, as a cell type can exist in different states. This is a difficult question since most of the phenotypic measurements including the single-cell transcriptome are only a one-time snapshot of the cell. However, one can compare transcriptomes collected from different time points or different behavioral, physiological or pathological states and see which clusters appear, disappear or shift under different conditions. Cell type-specific gene expression changes associated with different cell states may be seen during circadian cycles, variable metabolic states, development, aging, or under behavioral, pharmacological or diseased conditions (Fig. 4) (Mayr et al., 2019; Morris, 2019). Furthermore, individual variations within a species (e.g., within the human population) that are driven by genetic or environmental factors may be manifested as a variety of cell type or cell state variations. Studying the various states of cell types will enhance our ability to distinguish core gene sets (e.g., master transcription factors) maintaining cell type identities versus genes associated with specific functional states, and further our understanding of the diverse function of cell types as well as the biological basis of individual variability. The distinction between cell types and cell states is particularly challenging during development, as cells continually change their states and, at certain key time points, they may switch their cell type identities. Can single cell phenotypic properties such as transcriptomes distinguish types versus states? Although not absolute, it is reasonable to assume that transcriptomic changes tend to be more continuous during cell state transitions, and more abrupt or discrete when cells switch their types. More often, emergence of a new cell type during development is the consequence of cell division from which a daughter cell takes up a new cell type identity (Fig. 4). Trajectory analysis or lineage tracing coupled with single-cell transcriptomics across developmental time points has now often been used to identify the time course of emergence and maturation of each cell type, as well as the ancestor-descendant relationship across cell types that are present at different developmental stages (e.g., progenitors versus differentiated cells) (McKenna and Gagnon, 2019; Saelens et al., 2019; Tritschler et al., 2019; Wagner and Klein, 2020). Coordinated neuronal activities in brain circuits generate sensory perception and behavior. Specific neuronal populations activated during a particular perceptual or behavioral episode can be identified by screening for the activation of IEGs in them, via immunostaining, transgenic reporter lines, or single-cell or spatial transcriptomics in more recent years (DeNardo and Luo, 2017; Hrvatin et al., 2018; Moffitt et al., 2018; Wu et al., 2017). IEG activation leads to expression of downstream effectors, such as ion channels or synaptic proteins, that shape the cell states and remodel neuronal connections. Cell state changes in the brain are closely related to neural plasticity. Similarly, various diseased conditions can induce pathological changes in cell states in different brain regions or tissue organs. Numerous studies have revealed selective vulnerability of specific cell types for specific diseases. Pharmacological or genetic (e.g., CRISPR-based) perturbations in normal or diseased conditions, in combination with single-cell profiling (e.g., Perturb-seq), are powerful approaches to gain a mechanistic understanding of how disruptive or restorative cell state changes can affect cell type function or dysfunction (Adamson et al., 2016; Dixit et al., 2016; Jaitin et al., 2016; Replogle et al., 2022). Here I highlight a prominent feature of the non-neuronal cell types in the brain, which is that despite having lower diversity than neurons in baseline adult state, many non-neuronal cell types undergo pronounced changes, i.e., they exhibit many different cell states, under different physiological or diseased conditions. Astrocytes exhibit diverse morphological and physiological properties in different brain regions and contribute to essential functions in blood-brain barrier, synaptogenesis, neurotransmitter buffering, ion homeostasis, and secretion of neuroactive agents (Ben Haim and Rowitch, 2017; Khakh and Deneen, 2019). Astrocytes become reactive under pathological conditions. Reactive astrocytes undergo morphological, molecular, and functional changes in response to injury or CNS diseases; they may adopt multiple, heterogeneous states depending on context (Escartin et al., 2021). Oligodendrocytes are the myelinating cells of the central nervous system (CNS) that are generated from OPCs throughout life. Myelination process also exhibits activity-dependent plasticity (Monje, 2018). Oligodendrocyte pathology is evident in a range of disorders including multiple sclerosis, schizophrenia and Alzheimer’s disease (Kuhn et al., 2019). Regarding cerebrovascular cell types, recent single-cell transcriptomic studies in the human brain reveal gene expression changes in them that can impact blood-brain barrier integrity in Huntington’s and Alzheimer’s diseases (Garcia et al., 2022; Yang et al., 2022). Finally, microglia are the primary innate immune cells in the CNS and have a distinct developmental origin from peripheral immune cells. They are generated from mesodermal progenitors that arise from the yolk sac and are among the earliest residential cell types in the brain. Microglia display diverse and dynamic phenotypic states and play a plethora of roles in development, adulthood (homeostasis), aging and diseases (Butovsky and Weiner, 2018; Prinz et al., 2019; Thion and Garel, 2020). Single-cell transcriptomic studies reveal a relatively homogeneous adult microglia population, and greater heterogeneity of microglia states at different developmental stages, during aging and in pathological conditions (Hammond et al., 2019; Li et al., 2019; Masuda et al., 2019). In particular, microglia can be both responders to and inducers of various neurodegenerative, neuroinflammatory and neurodevelopmental diseases. Taken together, these studies paint a collective picture on how the variety of non-neuronal cell types actively respond to and contribute to different physiological and pathological changes in the brain. Cell type development A deep understanding of a subject often comes from understanding how it is built. The entire repertoire of cell types in the brain and the body is built through a sequential and parallel series of spatially and temporally coordinated developmental events starting from a single fertilized egg, the zygote. This developmental program carries out a remarkable implementation plan that unravels the identities of all cell types which are encoded in the genome through evolution. Transcriptional and epigenetic regulatory programs are unfolded from the genome sequences and drive a cascading series of cell proliferation and differentiation processes leading to the manifestation of diverse cellular phenotypes. In the developmental ontogeny of cell types, earlier-stage ancestral cell types are fewer and are more multipotent, they give rise to a larger number of descendant cell types with increasingly restricted fates. The developmental program rolls out not only a temporal but also an elaborate spatial plan, specifying the location of each tissue organ and the spatial organization of all the cell types within each. This is a highly dynamic spatiotemporal process involving specific cell-cell interactions, cell migration streams, and formation of niches and microenvironments that allow functional specialization. For the brain, the main series of events of brain development leading to the mature adult-stage cell types and circuits include: patterning and regionalization (laying out the master plan of brain architecture), neurogenesis and neuronal migration, gliogenesis and glia cell differentiation, neuronal differentiation and circuit formation (axonogenesis, dendritic arborization, synaptogenesis, myelination), and circuit refinement and plasticity (Fig. 4). Systematic single-cell transcriptomic, epigenomic and spatially resolved transcriptomic profiling with high temporal resolution, coupled with lineage tracing and other phenotypic characterization, holds tremendous potential to capture key sets of genes and genomic regulatory networks involved in these series of events and begin to resolve the extremely complex spatial and temporal transition of cell types and states leading to the adult-stage repertoire of cell types (Allaway et al., 2021; Bandler et al., 2022; Bhaduri et al., 2021; Cao et al., 2019b; Chen et al., 2022; Delgado et al., 2022; Di Bella et al., 2021; Klingler et al., 2021; La Manno et al., 2021; Romanov et al., 2020; Schmitz et al., 2022; Sharma et al., 2020; Shekhar et al., 2022; Tiklova et al., 2019; Zhu et al., 2018). Studying brain cell type development using these approaches will allow us to establish the developmental trajectory for each cell type from progenitors to transitional cell types and states to adult mature cells, discover the set of master transcription factors that define and maintain the identity of each cell type, and identify key events and molecular correlates that lead to the acquisition of a cell type’s specific connectional or functional properties. The generation and patterning of mouse cortex and spinal cord cell types are two example systems where extensive historical studies have uncovered several common principles (Cadwell et al., 2019; Catela et al., 2015; Jessell, 2000; O’Leary et al., 2007; Osseward and Pfaff, 2019; Sagner and Briscoe, 2019). First, opposing morphogen gradients establish the basic plan for cortex (anterior-posterior) or spinal cord (dorsal-ventral) patterning and provide instructive signals for the expression of complementary sets of transcription factor activators and repressors, which in turn define distinct neural progenitor domains within cortex or spinal cord. Second, driven by the transcription factor network, each type of neural progenitors generates a series of neuronal cell types. The set of cell type-defining transcription factors in a progenitor or a daughter cell can change with time, such that different neuronal types emanate from the same progenitor in a precisely timed “birth order”. Later in development, the same neural progenitors also generate non-neuronal cell types such as astrocytes and oligodendrocytes. Third, specific sets of cell adhesion molecules provide guidance cues for axon path finding and synapse formation, leading to the assembly of region- and cell type-specific local and global circuit networks. Fourth, patterned neural activities spontaneously emerged from the circuits and/or influenced by external inputs further sculpt synaptic connections and circuit organization to refine functional specificity of the circuit networks. In addition to these general principles, it is worth noting the many kinds of complexity already encountered. The development of a cell type may not follow a simple trajectory but involves multiple steps of divergent or convergent differentiation (Shekhar et al., 2022), the former due to the multipotency of progenitors or transitional cell types and the latter due to convergence of different transitional types. Developmental trajectory also is not the same as developmental lineage, as a lineage is defined as all the cells descended from a single precursor/progenitor and it has been shown in multiple systems that a progenitor can produce cells belonging to several neuronal and non-neuronal types, ordered by developmental timing (Agathocleous and Harris, 2009; Sagner and Briscoe, 2019; Sulston et al., 1983; Zeng and Sanes, 2017). Finally, there are transient cell types and circuits that mainly exist during development and have developmental stage-specific functions (Cossart and Garel, 2022; Molnar et al., 2020). All these observations, and more to be discovered, contribute to a nuanced understanding that cell type development is not a simple linear process, but a highly multifaceted process leading to the complex cell type landscape described in the above sections, which underlies the richness of cell type function. A comprehensive atlas of mammalian cell type development, likely first generated in mouse and then extended to other species including human (Haniffa et al., 2021), will provide the foundation for matching developmental events and their timelines across species, better understanding the evolutionary relationships between cell types, evaluating and guiding human iPSC and organoid in vitro development, and ultimately, transforming our investigation and treatment of developmental disorders. How to define cell types? In conclusion, cell types are the product of evolution, and they are the basic functional units of an organism. To unify these two concepts and define cell types properly, we need to take a multilevel approach, progressing from simple and singular to complex and multifaceted. In such a roadmap, with each iteration, the definition of cell types will become more mature and unified, and the repertoire of cell types defined will better align with the functional architecture of the organism. The logical first step is to use single-cell transcriptomics-based cell type taxonomy as the initial framework and anchor for defining cell types. The transcriptomic taxonomy contains evolutionarily rooted molecular signatures and allows effective label transfer and linking with all other modalities. Conversely, relating other cellular properties will help to refine transcriptomic types. The transcriptomic taxonomy organizes cell types in a hierarchical manner, laying out different levels of descriptions from major divisions at class and subclass levels to more granular and fuzzier divisions at type and subtype levels (due to the more prevailing continuous variations at the latter levels). To account for the biological reality, a hierarchical presentation of cell types is more meaningful than ascertaining an exact number of types. The transcriptomic taxonomy should be supported by comprehensive epigenomic and spatially resolved transcriptomic characterizations (Fig. 1A), to associate chromatin modification and gene regulatory elements to transcriptomic profiles and assign precise spatial distribution patterns to transcriptomic cell types. Second, we need to conduct comprehensive anatomical, physiological and functional studies of transcriptomic types using approaches that allow molecular identification of the cells under study (Fig. 1A–B). Such studies will help to resolve differing opinions in lumping or splitting cell types and provide rationales for determining the appropriate level of granularity in defining cell types. They will also provide a context for understanding cell type function and associated cell state changes. In particular, generating complete neuronal morphology reconstructions and comprehensive brain-wide connectomics datasets and relating them to transcriptomic types (Fig. 1A) will be extremely informative in understanding the ultimate synaptic-level brain architecture and its underlying organizing principles, which will lay the foundation for understanding circuit-based brain function. Third, we need to systematically study the entire developmental process of cell types, at least in mouse. Extending the above-mentioned approaches into development will reveal causal relationships and molecular mechanisms underlying the unique identities, connectivity or other forms of cell-cell interactions, and functions of the vast array of cell types. We should also extend such cell type studies into other species as much as possible, to further uncover evolutionary principles of cell type diversity and how it supports the common or species-specific biological functions including those of the human itself. The studies of cell types and evolution-development (Evo-Devo) are truly interdependent; to achieve meaningful progress one cannot just do one without the other. Finally, to put all these together, we need a conceptual framework and knowledge base to organize all the knowledge gained from these studies. A tantalizing idea of a “periodic table of cell types” has been proposed (Xia and Yanai, 2019). Considering the Evo-Devo root and the consequential hierarchical organization of cell types, here I suggest that a “tree of cell types” might be more appropriate for an overarching classification of cell types and delineation of their origins and relationships (Stadler et al., 2021; Tanay and Sebe-Pedros, 2021). One can define a tree of cell types for each species, covering its entire life span, and compare such trees across species. Obviously, the “tree” will be a very complex, multi-dimensional graph, and there will likely be multiple branches connecting each node to account for convergence, divergence and other multipronged interrelatedness. To make the tree of cell types widely applicable, it will be critical to adopt explicitly definable and standardized criteria, develop a common cell type ontology and nomenclature, and create computational tools to allow mapping and comparison across datasets as well as genetic tools to enable consistent access of cell types (Osumi-Sutherland, 2017; Yuste et al., 2020). To extract knowledge and insights from the vast amount of data, a list of associated rules, logics and principles will need to be established and articulated, and this will be greatly facilitated by computational modeling. Ultimately, this knowledge base of cell types, interweaving cell type function, development and evolution, will provide the blueprint of life to enable a deeper understanding of the dynamic changes of cellular function under a wide range of healthy and diseased conditions, and lead to innovations that improve human health in many ways. Acknowledgments I thank my colleagues in the NIH BRAIN Initiative Cell Census Network (BICCN) and at the Allen Institute for many insightful discussions that deepened and broadened my understanding of many topics covered in this review. This work is supported by Allen Institute for Brain Science and by NIH grant U19MH114830. Figure 1. Approaches to characterize cell types. (A) Molecular and anatomical approaches as primary ways of single-cell characterization include single-cell transcriptomics by sc/snRNA-seq, single-cell epigenomics exampled by snATAC-seq, spatially resolved transcriptomics exampled by MERFISH, full morphology reconstruction exampled by MouseLight (image adopted from Winnubst et al., 2019), EM connectomics (image adopted from Hulse et al., 2021), and barcoded connectomics exampled by BARseq (image adopted from Chen et al., 2019). (B) Cross-modality integrated approaches include Patch-seq (image adopted from Lee et al., 2021), retrograde tracing followed by molecular profiling, functional imaging followed by spatially resolved transcriptomics, using Patch-seq data as a Rosetta stone to assign molecular identities to neurons reconstructed from EM dataset (image adopted from Turner et al., 2022), and generation of enhancer based viral vectors (image adopted from Mich et al., 2021). Figure 2. Hierarchical organization of cell types. (A) A transcriptomic cell atlas for the mouse nervous system (image adopted from Zeisel et al., 2018). (B) A transcriptomic cell type taxonomy for the mouse primary motor cortex, with annotation (image adopted from Brain Initiative Cell Census Network, 2021). (C) UMAP representation of a transcriptomic cell type taxonomy for the glutamatergic neuron types in mouse isocortex and hippocampal formation, revealing discrete and continuous variations (image adopted from Yao et al., 2021b). Figure 3. Multimodal correspondence of cell type phenotypic properties. (A) MERFISH data from mouse motor cortex shows that continuous variation of glutamatergic IT transcriptomic types is correlated with their continuous spatial distribution along the cortical depth from L2/3 to L6 (image adopted from Zhang et al., 2021a). (B) Patch-seq data on GABAergic interneurons from mouse visual cortex shows correspondence between transcriptomic (T) types and morpho-electrical (ME) types (image adopted from Gouwens et al., 2020). (C) Brain-wide complete morphology reconstruction of cortical glutamatergic neurons shows distinct axon projection patterns between major transcriptomic types and further heterogeneity within each type (image adopted from Peng et al., 2021). Each color outlines the projection of one neuron within the type in each panel. Figure 4. Dynamic changes of cell types and states during development, aging and various physiological or pathological contexts. Major neuronal and non-neuronal classes are shown along the life stages of development, adulthood and aging. Neural progenitors generate different neuronal types, astrocytes and oligodendrocytes at different developmental timepoints, whereas microglia have a separate developmental origin. Major developmental events, various physiological states in adulthood, and different diseased states throughout the lifespan are shown below the timeline. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Declaration of Interests The author declares no competing interests. References Abbott LF , Bock DD , Callaway EM , Denk W , Dulac C , Fairhall AL , Fiete I , Harris KM , Helmstaedter M , Jain V , (2020). The Mind of a Mouse. Cell 182 , 1372–1376.32946777 Adamson B , Norman TM , Jost M , Cho MY , Nunez JK , Chen Y , Villalta JE , Gilbert LA , Horlbeck MA , Hein MY , (2016). A Multiplexed Single-Cell CRISPR Screening Platform Enables Systematic Dis of the Unfolded Protein Response. Cell 167 , 1867–1882 e1821.27984733 Agathocleous M , and Harris WA (2009). From progenitors to differentiated cells in the vertebrate retina. Annu Rev Cell Dev Biol 25 , 45–69.19575661 Allaway KC , Gabitto MI , Wapinski O , Saldi G , Wang CY , Bandler RC , Wu SJ , Bonneau R , and Fishell G (2021). Genetic and epigenetic coordination of cortical interneuron development. Nature 597 , 693–697.34552240 Arendt D (2008). The evolution of cell types in animals: emerging principles from molecular studies. Nature reviews Genetics 9 , 868–882. Arendt D , Musser JM , Baker CVH , Bergman A , Cepko C , Erwin DH , Pavlicev M , Schlosser G , Widder S , Laubichler MD , (2016). The origin and evolution of cell types. Nature reviews Genetics 17 , 744–757. Armand EJ , Li J , Xie F , Luo C , and Mukamel EA (2021). Single-Cell Sequencing of Brain Cell Transcriptomes and Epigenomes. Neuron 109 , 11–26.33412093 Baden T , Euler T , and Berens P (2020). Understanding the retinal basis of vision across species. Nature reviews Neuroscience 21 , 5–20.31780820 Bakken TE , Jorstad NL , Hu Q , Lake BB , Tian W , Kalmbach BE , Crow M , Hodge RD , Krienen FM , Sorensen SA , (2021). Comparative cellular analysis of motor cortex in human, marmoset and mouse. Nature 598 , 111–119.34616062 Bandler RC , Vitali I , Delgado RN , Ho MC , Dvoretskova E , Ibarra Molinas JS , Frazel PW , Mohammadkhani M , Machold R , Maedler S , (2022). Single-cell delineation of lineage and genetic identity in the mouse brain. Nature 601 , 404–409.34912118 Ben Haim L , and Rowitch DH (2017). Functional diversity of astrocytes in neural circuit regulation. Nature reviews Neuroscience 18 , 31–41.27904142 Berg J , Sorensen SA , Ting JT , Miller JA , Chartrand T , Buchin A , Bakken TE , Budzillo A , Dee N , Ding SL , (2021). Human neocortical expansion involves glutamatergic neuron diversification. Nature 598 , 151–158.34616067 Bhaduri A , Sandoval-Espinosa C , Otero-Garcia M , Oh I , Yin R , Eze UC , Nowakowski TJ , and Kriegstein AR (2021). An atlas of cortical arealization identifies dynamic molecular signatures. Nature 598 , 200–204.34616070 Booeshaghi AS , Yao Z , van Velthoven C , Smith K , Tasic B , Zeng H , and Pachter L (2021). Isoform cell-type specificity in the mouse primary motor cortex. Nature 598 , 195–199.34616073 Brain Initiative Cell Census Network (2021). A multimodal cell census and atlas of the mammalian primary motor cortex. Nature 598 , 86–102.34616075 Bugeon S , Duffield J , Dipoppa M , Ritoux A , Prankerd I , Nicolout-sopoulos D , Orme D , Shinn M , Peng H , Forrest H , (2021). A transcriptomic axis predicts state modulation of cortical interneurons. bioRxiv, 2021.2010.2024.465600. Butovsky O , and Weiner HL (2018). Microglial signatures and their role in health and disease. Nature reviews Neuroscience 19 , 622–635.30206328 Cadwell CR , Bhaduri A , Mostajo-Radji MA , Keefe MG , and Nowakowski TJ (2019). Development and Arealization of the Cerebral Cortex. Neuron 103 , 980–1004.31557462 Cadwell CR , Palasantza A , Jiang X , Berens P , Deng Q , Yilmaz M , Reimer J , Shen S , Bethge M , Tolias KF , (2016). Electrophysiological, transcriptomic and morphologic profiling of single neurons using Patch-seq. Nature biotechnology 34 , 199–203. Cao C , Lemaire LA , Wang W , Yoon PH , Choi YA , Parsons LR , Matese JC , Wang W , Levine M , and Chen K (2019a). Comprehensive single-cell transcriptome lineages of a proto-vertebrate. Nature 571 , 349–354.31292549 Cao J , Spielmann M , Qiu X , Huang X , Ibrahim DM , Hill AJ , Zhang F , Mundlos S , Christiansen L , Steemers FJ , (2019b). The single-cell transcriptional landscape of mammalian organogenesis. Nature 566 , 496–502.30787437 Catela C , Shin MM , and Dasen JS (2015). Assembly and function of spinal circuits for motor control. Annu Rev Cell Dev Biol 31 , 669–698.26393773 Cembrowski MS , Wang L , Lemire AL , Copeland M , DiLisio SF , Clements J , and Spruston N (2018). The subiculum is a patchwork of discrete subregions. eLife 7 . Chan KY , Jang MJ , Yoo BB , Greenbaum A , Ravi N , Wu WL , Sanchez-Guardado L , Lois C , Mazmanian SK , Deverman BE , (2017). Engineered AAVs for efficient noninvasive gene delivery to the central and peripheral nervous systems. Nature neuroscience 20 , 1172–1179.28671695 Chen A , Liao S , Cheng M , Ma K , Wu L , Lai Y , Qiu X , Yang J , Xu J , Hao S , (2022). Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA nanoball-patterned arrays. Cell 185 , 1777–1792 e21.35512705 Chen R , Blosser TR , Djekidel MN , Hao J , Bhattacherjee A , Chen W , Tuesta LM , Zhuang X , and Zhang Y (2021). Decoding molecular and cellular heterogeneity of mouse nucleus accumbens. Nature neuroscience 24 , 1757–1771.34663959 Chen X , Sun YC , Zhan H , Kebschull JM , Fischer S , Matho K , Huang ZJ , Gillis J , and Zador AM (2019). High-Throughput Mapping of Long-Range Neuronal Projection Using In Situ Sequencing. Cell 179 , 772–786 e719.31626774 Cho NH , Cheveralls KC , Brunner AD , Kim K , Michaelis AC , Raghavan P , Kobayashi H , Savy L , Li JY , Canaj H , (2022). OpenCell: Endogenous tagging for the cartography of human cellular organization. Science 375 , eabi6983.35271311 Clark IC , Gutierrez-Vazquez C , Wheeler MA , Li Z , Rothhammer V , Linnerbauer M , Sanmarco LM , Guo L , Blain M , Zandee SEJ , (2021). Barcoded viral tracing of single-cell interactions in central nervous system inflammation. Science 372 . Close JL , Long BR , and Zeng H (2021). Spatially resolved transcriptomics in neuroscience. Nature methods 18 , 23–25.33408398 Colquitt BM , Merullo DP , Konopka G , Roberts TF , and Brainard MS (2021). Cellular transcriptomics reveals evolutionary identities of songbird vocal circuits. Science 371 . Condylis C , Ghanbari A , Manjrekar N , Bistrong K , Yao S , Yao Z , Nguyen TN , Zeng H , Tasic B , and Chen JL (2022). Dense functional and molecular readout of a circuit hub in sensory cortex. Science 375 , eabl5981.34990233 Cossart R , and Garel S (2022). Step by step: cells with multiple functions in cortical circuit assembly. Nature reviews Neuroscience. Cusanovich DA , Hill AJ , Aghamirzaie D , Daza RM , Pliner HA , Berletch JB , Filippova GN , Huang X , Christiansen L , DeWitt WS , (2018). A Single-Cell Atlas of In Vivo Mammalian Chromatin Accessibility. Cell 174 , 1309–1324 e1318.30078704 Daigle TL , Madisen L , Hage TA , Valley MT , Knoblich U , Larsen RS , Takeno MM , Huang L , Gu H , Larsen R , (2018). A Suite of Transgenic Driver and Reporter Mouse Lines with Enhanced BrainCell-Type Targeting and Functionality. Cell 174 , 465–480 e422.30007418 Darmanis S , Sloan SA , Zhang Y , Enge M , Caneda C , Shuer LM , Hayden Gephart MG , Barres BA , and Quake SR (2015). A survey of human brain transcriptome diversity at the single cell level. Proc Natl Acad Sci U S A 112 , 7285–7290.26060301 Delgado RN , Allen DE , Keefe MG , Mancia Leon WR , Ziffra RS , Crouch EE , Alvarez-Buylla A , and Nowakowski TJ (2022). Individual human cortical progenitors can produce excitatory and inhibitory neurons. Nature 601 , 397–403.34912114 DeNardo L , and Luo L (2017). Genetic strategies to access activated neurons. Current opinion in neurobiology 45 , 121–129.28577429 Di Bella DJ , Habibi E , Stickels RR , Scalia G , Brown J , Yadollahpour P , Yang SM , Abbate C , Biancalani T , Macosko EZ , (2021). Molecular logic of cellular diversification in the mouse cerebral cortex. Nature 595 , 554–559.34163074 Dimidschstein J , Chen Q , Tremblay R , Rogers SL , Saldi GA , Guo L , Xu Q , Liu R , Lu C , Chu J , (2016). A viral strategy for targeting and manipulating interneurons across vertebrate species. Nature neuroscience 19 , 1743–1749.27798629 Dimou L , and Simons M (2017). Diversity of oligodendrocytes and their progenitors. Current opinion in neurobiology 47 , 73–79.29078110 Dixit A , Parnas O , Li B , Chen J , Fulco CP , Jerby-Arnon L , Marjanovic ND , Dionne D , Burks T , Raychowdhury R , (2016). Perturb-Seq: Dissecting Molecular Circuits with Scalable Single-Cell RNA Profiling of Pooled Genetic Screens. Cell 167 , 1853–1866 e1817.27984732 Drokhlyansky E , Smillie CS , Van Wittenberghe N , Ericsson M , Griffin GK , Eraslan G , Dionne D , Cuoco MS , Goder-Reiser MN , Sharova T , (2020). The Human and Mouse Enteric Nervous System at Single-Cell Resolution. Cell 182 , 1606–1622 e1623.32888429 Ecker JR , Geschwind DH , Kriegstein AR , Ngai J , Osten P , Polioudakis D , Regev A , Sestan N , Wickersham IR , and Zeng H (2017). The BRAIN Initiative Cell Census Consortium: Lessons Learned toward Generating a Comprehensive Brain Cell Atlas. Neuron 96 , 542–557.29096072 Escartin C , Galea E , Lakatos A , O’Callaghan JP , Petzold GC , Serrano-Pozo A , Steinhauser C , Volterra A , Carmignoto G , Agarwal A , (2021). Reactive astrocyte nomenclature, definitions, and future directions. Nature neuroscience 24 , 312–325.33589835 Fishell G , and Heintz N (2013). The neuron identity problem: form meets function. Neuron 80 , 602–612.24183013 Fuzik J , Zeisel A , Mate Z , Calvigioni D , Yanagawa Y , Szabo G , Linnarsson S , and Harkany T (2016). Integration of electrophysiological recordings with single-cell RNA-seq data identifies neuronal subtypes. Nature biotechnology 34 , 175–183. Gao L , Liu S , Gou L , Hu Y , Liu Y , Deng L , Ma D , Wang H , Yang Q , Chen Z , (2022). Single-neuron projectome of mouse prefrontal cortex. Nature neuroscience 25 , 515–529.35361973 Garcia FJ , Sun N , Lee H , Godlewski B , Mathys H , Galani K , Zhou B , Jiang X , Ng AP , Mantero J , (2022). Single-cell dissection of the human brain vasculature. Nature 603 , 893–899.35158371 Gergues MM , Han KJ , Choi HS , Brown B , Clausing KJ , Turner VS , Vainchtein ID , Molofsky AV , and Kheirbek MA (2020). Circuit and molecular architecture of a ventral hippocampal network. Nature neuroscience 23 , 1444–1452.32929245 Gour A , Boergens KM , Heike N , Hua Y , Laserstein P , Song K , and Helmstaedter M (2021). Postnatal connectomic development of inhibition in mouse barrel cortex. Science 371 . Gouwens NW , Sorensen SA , Baftizadeh F , Budzillo A , Lee BR , Jarsky T , Alfiler L , Baker K , Barkan E , Berry K , (2020). Integrated Morphoelectric and Transcriptomic Classification of Cortical GABAergic Cells. Cell 183 , 935–953 e919.33186530 Graybuck LT , Daigle TL , Sedeno-Cortes AE , Walker M , Kalmbach B , Lenz GH , Morin E , Nguyen TN , Garren E , Bendrick JL , (2021). Enhancer viruses for combinatorial cell-subclass-specific labeling. Neuron 109 , 1449–1464 e1413.33789083 Hammond TR , Dufort C , Dissing-Olesen L , Giera S , Young A , Wysoker A , Walker AJ , Gergits F , Segel M , Nemesh J , (2019). Single-Cell RNA Sequencing of Microglia throughout the Mouse Lifespan and in the Injured Brain Reveals Complex Cell-State Changes. Immunity 50 , 253–271 e256.30471926 Han L , Wei X , Liu C , Volpe G , Zhuang Z , Zou X , Wang Z , Pan T , Yuan Y , Zhang X , (2022). Cell transcriptomic atlas of the non-human primate Macaca fascicularis. Nature 604 , 723–731.35418686 Han X , Wang R , Zhou Y , Fei L , Sun H , Lai S , Saadatpour A , Zhou Z , Chen H , Ye F , (2018a). Mapping the Mouse Cell Atlas by Microwell-Seq. Cell 172 , 1091–1107 e1017.29474909 Han Y , Kebschull JM , Campbell RAA , Cowan D , Imhof F , Zador AM , and Mrsic-Flogel TD (2018b). The logic of single-cell projections from visual cortex. Nature 556 , 51–56.29590093 Haniffa M , Taylor D , Linnarsson S , Aronow BJ , Bader GD , Barker RA , Camara PG , Camp JG , Chedotal A , Copp A , (2021). A roadmap for the Human Developmental Cell Atlas. Nature 597 , 196–205.34497388 Harris KD , and Shepherd GM (2015). The neocortical circuit: themes and variations. Nature neuroscience 18 , 170–181.25622573 Hashikawa Y , Hashikawa K , Rossi MA , Basiri ML , Liu Y , Johnston NL , Ahmad OR , and Stuber GD (2020). Transcriptional and Spatial Resolution of Cell Types in the Mammalian Habenula. Neuron 106 , 743–758 e745.32272058 Helmstaedter M , Briggman KL , Turaga SC , Jain V , Seung HS , and Denk W (2013). Connectomic reconstruction of the inner plexiform layer in the mouse retina. Nature 500 , 168–174.23925239 Hildebrand DGC , Cicconet M , Torres RM , Choi W , Quan TM , Moon J , Wetzel AW , Scott Champion A , Graham BJ , Randlett O , (2017). Whole-brain serial-section electron microscopy in larval zebrafish. Nature 545 , 345–349.28489821 Hobert O , and Kratsios P (2019). Neuronal identity control by terminal selectors in worms, flies, and chordates. Current opinion in neurobiology 56 , 97–105.30665084 Hodge RD , Bakken TE , Miller JA , Smith KA , Barkan ER , Graybuck LT , Close JL , Long B , Johansen N , Penn O , (2019). Conserved cell types with divergent features in human versus mouse cortex. Nature 573 , 61–68.31435019 Hrvatin S , Hochbaum DR , Nagy MA , Cicconet M , Robertson K , Cheadle L , Zilionis R , Ratner A , Borges-Monroy R , Klein AM , (2018). Single-cell analysis of experience-dependent transcriptomic states in the mouse visual cortex. Nature neuroscience 21 , 120–129.29230054 Hrvatin S , Tzeng CP , Nagy MA , Stroud H , Koutsioumpa C , Wilcox OF , Assad EG , Green J , Harvey CD , Griffith EC , (2019). A scalable platform for the development of cell-type-specific viral drivers. eLife 8 . Hu JS , Vogt D , Sandberg M , and Rubenstein JL (2017). Cortical interneuron development: a tale of time and space. Development 144 , 3867–3878.29089360 Huang ZJ , and Paul A (2019). The diversity of GABAergic neurons and neural communication elements. Nature Reviews Neuroscience 20 , 563–572.31222186 Hulse BK , Haberkern H , Franconville R , Turner-Evans DB , Takemura SY , Wolff T , Noorman M , Dreher M , Dan C , Parekh R , (2021). A connectome of the Drosophila central complex reveals network motifs suitable for flexible navigation and context-dependent action selection. eLife 10 . Jaitin DA , Kenigsberg E , Keren-Shaul H , Elefant N , Paul F , Zaretsky I , Mildner A , Cohen N , Jung S , Tanay A , (2014). Massively parallel single-cell RNA-seq for marker-free decomposition of tissues into cell types. Science 343 , 776–779.24531970 Jaitin DA , Weiner A , Yofe I , Lara-Astiaso D , Keren-Shaul H , David E , Salame TM , Tanay A , van Oudenaarden A , and Amit I (2016). Dissecting Immune Circuits by Linking CRISPR-Pooled Screens with Single-Cell RNA-Seq. Cell 167 , 1883–1896 e1815.27984734 Jenett A , Rubin GM , Ngo TT , Shepherd D , Murphy C , Dionne H , Pfeiffer BD , Cavallaro A , Hall D , Jeter J , (2012). A GAL4-driver line resource for Drosophila neurobiology. Cell reports 2 , 991–1001.23063364 Jessell TM (2000). Neuronal specification in the spinal cord: inductive signals and transcriptional codes. Nature reviews Genetics 1 , 20–29. Kamath T , Abdulraouf A , Burris SJ , Langlieb J , Gazestani V , Nadaf NM , Balderrama K , Vanderburg C , and Macosko EZ (2022). Single-cell genomic profiling of human dopamine neurons identifies a population that selectively degenerates in Parkinson’s disease. Nature neuroscience 25 , 588–595.35513515 Kebschull JM , Garcia da Silva P , Reid AP , Peikon ID , Albeanu DF , and Zador AM (2016). High-Throughput Mapping of Single-Neuron Projections by Sequencing of Barcoded RNA. Neuron 91 , 975–987.27545715 Kebschull JM , Richman EB , Ringach N , Friedmann D , Albarran E , Kolluru SS , Jones RC , Allen WE , Wang Y , Cho SW , (2020). Cerebellar nuclei evolved by repeatedly duplicating a conserved cell-type set. Science 370 . Khakh BS , and Deneen B (2019). The Emerging Nature of Astrocyte Diversity. Annual review of neuroscience 42 , 187–207. Kim DW , Yao Z , Graybuck LT , Kim TK , Nguyen TN , Smith KA , Fong O , Yi L , Koulena N , Pierson N , (2019). Multimodal Analysis of Cell Types in a Hypothalamic Node Controlling Social Behavior. Cell 179 , 713–728 e717.31626771 Kim EJ , Zhang Z , Huang L , Ito-Cole T , Jacobs MW , Juavinett AL , Senturk G , Hu M , Ku M , Ecker JR , (2020). Extraction of Distinct Neuronal Cell Types from within a Genetically Continuous Population. Neuron 107 , 274–282 e276.32396852 Klingler E , Tomasello U , Prados J , Kebschull JM , Contestabile A , Galinanes GL , Fievre S , Santinha A , Platt R , Huber D , (2021). Temporal controls over inter-areal cortical projection neuron fate diversity. Nature 599 , 453–457.34754107 Kozareva V , Martin C , Osorno T , Rudolph S , Guo C , Vanderburg C , Nadaf N , Regev A , Regehr WG , and Macosko E (2021). A transcriptomic atlas of mouse cerebellar cortex comprehensively defines cell types. Nature 598 , 214–219.34616064 Krienen FM , Goldman M , Zhang Q , R CHDR , Florio M , Machold R , Saunders A , Levandowski K , Zaniewski H , Schuman B , (2020). Innovations present in the primate interneuron repertoire. Nature 586 , 262–269.32999462 Kuhn S , Gritti L , Crooks D , and Dombrowski Y (2019). Oligodendrocytes in Development, Myelin Generation and Beyond. Cells 8 . La Manno G , Siletti K , Furlan A , Gyllborg D , Vinsland E , Mossi Albiach A , Mattsson Langseth C , Khven I , Lederer AR , Dratva LM , (2021). Molecular architecture of the developing mouse brain. Nature 596 , 92–96.34321664 Lake BB , Ai R , Kaeser GE , Salathia NS , Yung YC , Liu R , Wildberg A , Gao D , Fung HL , Chen S , (2016). Neuronal subtypes and diversity revealed by single-nucleus RNA sequencing of the human brain. Science 352 , 1586–1590.27339989 Lake BB , Chen S , Sos BC , Fan J , Kaeser GE , Yung YC , Duong TE , Gao D , Chun J , Kharchenko PV , (2018). Integrative single-cell analysis of transcriptional and epigenetic states in the human adult brain. Nature biotechnology 36 , 70–80. Larsson L , Frisen J , and Lundeberg J (2021). Spatially resolved transcriptomics adds a new dimension to genomics. Nature methods 18 , 15–18.33408402 Lee BR , Budzillo A , Hadley K , Miller JA , Jarsky T , Baker K , Hill D , Kim L , Mann R , Ng L , (2021). Scaled, high fidelity electrophysiological, morphological, and transcriptomic cell characterization. eLife 10 . Lein E , Borm LE , and Linnarsson S (2017). The promise of spatial transcriptomics for neuroscience in the era of molecular cell typing. Science 358 , 64–69.28983044 Li H , Janssens J , De Waegeneer M , Kolluru SS , Davie K , Gardeux V , Saelens W , David FPA , Brbic M , Spanier K , (2022). Fly Cell Atlas: A single-nucleus transcriptomic atlas of the adult fruit fly. Science 375 , eabk2432.35239393 Li Q , Cheng Z , Zhou L , Darmanis S , Neff NF , Okamoto J , Gulati G , Bennett ML , Sun LO , Clarke LE , (2019). Developmental Heterogeneity of Microglia and Brain Myeloid Cells Revealed by Deep Single-Cell RNA Sequencing. Neuron 101 , 207–223 e210.30606613 Li YE , Preissl S , Hou X , Zhang Z , Zhang K , Qiu Y , Poirion OB , Li B , Chiou J , Liu H , (2021). An atlas of gene regulatory elements in adult mouse cerebrum. Nature 598 , 129–136.34616068 Lim L , Mi D , Llorca A , and Marin O (2018). Development and Functional Diversification of Cortical Interneurons. Neuron 100 , 294–313.30359598 Lindeboom RGH , Regev A , and Teichmann SA (2021). Towards a Human Cell Atlas: Taking Notes from the Past. Trends Genet 37 , 625–630.33879355 Liu H , Zhou J , Tian W , Luo C , Bartlett A , Aldridge A , Lucero J , Osteen JK , Nery JR , Chen H , (2021). DNA methylation atlas of the mouse brain at single-cell resolution. Nature 598 , 120–128.34616061 Lovett-Barron M , Chen R , Bradbury S , Andalman AS , Wagle M , Guo S , and Deisseroth K (2020). Multiple convergent hypothalamus-brainstem circuits drive defensive behavior. Nature neuroscience 23 , 959–967.32572237 Luo C , Keown CL , Kurihara L , Zhou J , He Y , Li J , Castanon R , Lucero J , Nery JR , Sandoval JP , (2017). Single-cell methylomes identify neuronal subtypes and regulatory elements in mammalian cortex. Science 357 , 600–604.28798132 Macosko EZ , Basu A , Satija R , Nemesh J , Shekhar K , Goldman M , Tirosh I , Bialas AR , Kamitaki N , Martersteck EM , (2015). Highly Parallel Genome-wide Expression Profiling of Individual Cells Using Nanoliter Droplets. Cell 161 , 1202–1214.26000488 Markram H , Toledo-Rodriguez M , Wang Y , Gupta A , Silberberg G , and Wu C (2004). Interneurons of the neocortical inhibitory system. Nature reviews Neuroscience 5 , 793–807.15378039 Marques S , Zeisel A , Codeluppi S , van Bruggen D , Mendanha Falcao A , Xiao L , Li H , Haring M , Hochgerner H , Romanov RA , (2016). Oligodendrocyte heterogeneity in the mouse juvenile and adult central nervous system. Science 352 , 1326–1329.27284195 Masland RH (2012). The neuronal organization of the retina. Neuron 76 , 266–280.23083731 Masuda T , Sankowski R , Staszewski O , Bottcher C , Amann L , Sagar , Scheiwe C , Nessler S , Kunz P , van Loo G , (2019). Spatial and temporal heterogeneity of mouse and human microglia at single-cell resolution. Nature 566 , 388–392.30760929 Matho KS , Huilgol D , Galbavy W , He M , Kim G , An X , Lu J , Wu P , Di Bella DJ , Shetty AS , (2021). Genetic dissection of the glutamatergic neuron system in cerebral cortex. Nature 598 , 182–187.34616069 Mayr U , Serra D , and Liberali P (2019). Exploring single cells in space and time during tissue development, homeostasis and regeneration. Development 146 . Mazzarello P (1999). A unifying concept: the history of cell theory. Nat Cell Biol 1 , E13–15.10559875 McKenna A , and Gagnon JA (2019). Recording development with single cell dynamic lineage tracing. Development 146 . Mich JK , Graybuck LT , Hess EE , Mahoney JT , Kojima Y , Ding Y , Somasundaram S , Miller JA , Kalmbach BE , Radaelli C , (2021). Functional enhancer elements drive subclass-selective expression from mouse to primate neocortex. Cell reports 34 , 108754.33789096 Moffitt JR , Bambah-Mukku D , Eichhorn SW , Vaughn E , Shekhar K , Perez JD , Rubinstein ND , Hao J , Regev A , Dulac C , (2018). Molecular, spatial, and functional single-cell profiling of the hypothalamic preoptic region. Science 362 . Molnar Z , Luhmann HJ , and Kanold PO (2020). Transient cortical circuits match spontaneous and sensory-driven activity during development. Science 370 . Monje M (2018). Myelin Plasticity and Nervous System Function. Annual review of neuroscience 41 , 61–76. Morgan JL , Berger DR , Wetzel AW , and Lichtman JW (2016). The Fuzzy Logic of Network Connectivity in Mouse Visual Thalamus. Cell 165 , 192–206.27015312 Morris SA (2019). The evolving concept of cell identity in the single cell era. Development 146 . Moses L , and Pachter L (2022). Museum of spatial transcriptomics. Nature methods 19 , 534–546.35273392 Mukamel EA , and Ngai J (2019). Perspectives on defining cell types in the brain. Current opinion in neurobiology 56 , 61–68.30530112 Munoz-Manchado AB , Bengtsson Gonzales C , Zeisel A , Munguba H , Bekkouche B , Skene NG , Lonnerberg P , Ryge J , Harris KD , Linnarsson S , (2018). Diversity of Interneurons in the Dorsal Striatum Revealed by Single-Cell RNA Sequencing and PatchSeq. Cell reports 24 , 2179–2190 e2177.30134177 Munro DAD , Movahedi K , and Priller J (2022). Macrophage compartmentalization in the brain and cerebrospinal fluid system. Sci Immunol 7 , eabk0391.35245085 Murphy WJ , Foley NM , Bredemeyer KR , Gatesy J , and Springer MS (2021). Phylogenomics and the Genetic Architecture of the Placental Mammal Radiation. Annu Rev Anim Biosci 9 , 29–53.33228377 Nelson SB , Sugino K , and Hempel CM (2006). The problem of neuronal cell types: a physiological genomics approach. Trends in neurosciences 29 , 339–345.16714064 Ngai J (2022). BRAIN 2.0: Transforming neuroscience. Cell 185 , 4–8.34995517 O’Leary DD , Chou SJ , and Sahara S (2007). Area patterning of the mammalian cortex. Neuron 56 , 252–269.17964244 Ortiz-Alvarez G , Daclin M , Shihavuddin A , Lansade P , Fortoul A , Faucourt M , Clavreul S , Lalioti ME , Taraviras S , Hippenmeyer S , (2019). Adult Neural Stem Cells and Multiciliated Ependymal Cells Share a Common Lineage Regulated by the Geminin Family Members. Neuron 102 , 159–172 e157.30824354 Ortiz C , Navarro JF , Jurek A , Martin A , Lundeberg J , and Meletis K (2020). Molecular atlas of the adult mouse brain. Sci Adv 6 , eabb3446.32637622 Osseward PJ 2nd , and Pfaff SL (2019). Cell type and circuit modules in the spinal cord. Current opinion in neurobiology 56 , 175–184.30954861 Osumi-Sutherland D (2017). Cell ontology in an age of data-driven cell classification. BMC bioinformatics 18 , 558.29322914 Paul A , Crow M , Raudales R , He M , Gillis J , and Huang ZJ (2017). Transcriptional Architecture of Synaptic Communication Delineates GABAergic Neuron Identity. Cell 171 , 522–539 e520.28942923 Peng H , Xie P , Liu L , Kuang X , Wang Y , Qu L , Gong H , Jiang S , Li A , Ruan Z , (2021). Morphological diversity of single neurons in molecularly defined cell types. Nature 598 , 174–181.34616072 Petilla Interneuron Nomenclature Group, Ascoli GA , Alonso-Nanclares L , Anderson SA , Barrionuevo G , Benavides-Piccione R , Burkhalter A , Buzsaki G , Cauli B , Defelipe J , (2008). Petilla terminology: nomenclature of features of GABAergic interneurons of the cerebral cortex. Nature reviews Neuroscience 9 , 557–568.18568015 Phillips JW , Schulmann A , Hara E , Winnubst J , Liu C , Valakh V , Wang L , Shields BC , Korff W , Chandrashekar J , (2019). A repeated molecular architecture across thalamic pathways. Nature neuroscience 22 , 1925–1935.31527803 Pool AH , Wang T , Stafford DA , Chance RK , Lee S , Ngai J , and Oka Y (2020). The cellular basis of distinct thirst modalities. Nature 588 , 112–117.33057193 Poulin JF , Zou J , Drouin-Ouellet J , Kim KY , Cicchetti F , and Awatramani RB (2014). Defining midbrain dopaminergic neuron diversity by single-cell gene expression profiling. Cell reports 9 , 930–943.25437550 Preissl S , Fang R , Huang H , Zhao Y , Raviram R , Gorkin DU , Zhang Y , Sos BC , Afzal V , Dickel DE , (2018). Single-nucleus analysis of accessible chromatin in developing mouse forebrain reveals cell-type-specific transcriptional regulation. Nature neuroscience 21 , 432–439.29434377 Preuss TM , and Wise SP (2022). Evolution of prefrontal cortex. Neuropsychopharmacology 47 , 3–19.34363014 Prinz M , Jung S , and Priller J (2019). Microglia Biology: One Century of Evolving Concepts. Cell 179 , 292–311.31585077 Rao A , Barkley D , Franca GS , and Yanai I (2021). Exploring tissue architecture using spatial transcriptomics. Nature 596 , 211–220.34381231 Redmond SA , Figueres-Onate M , Obernier K , Nascimento MA , Parraguez JI , Lopez-Mascaraque L , Fuentealba LC , and Alvarez-Buylla A (2019). Development of Ependymal and Postnatal Neural Stem Cells and Their Origin from a Common Embryonic Progenitor. Cell reports 27 , 429–441 e423.30970247 Regev A , Teichmann SA , Lander ES , Amit I , Benoist C , Birney E , Bodenmiller B , Campbell P , Carninci P , Clatworthy M , (2017). The Human Cell Atlas. eLife 6 . Reilly MB , Cros C , Varol E , Yemini E , and Hobert O (2020). Unique homeobox codes delineate all the neuron classes of C. elegans. Nature 584 , 595–601.32814896 Ren J , Isakova A , Friedmann D , Zeng J , Grutzner SM , Pun A , Zhao GQ , Kolluru SS , Wang R , Lin R , (2019). Single-cell transcriptomes and whole-brain projections of serotonin neurons in the mouse dorsal and median raphe nuclei. eLife 8 . Replogle JM , Saunders RA , Pogson AN , Hussmann JA , Lenail A , Guna A , Mascibroda L , Wagner EJ , Adelman K , Lithwick-Yanai G , (2022). Mapping information-rich genotype-phenotype landscapes with genome-scale Perturb-seq. Cell. Romanov RA , Tretiakov EO , Kastriti ME , Zupancic M , Haring M , Korchynska S , Popadin K , Benevento M , Rebernik P , Lallemend F , (2020). Molecular design of hypothalamus development. Nature 582 , 246–252.32499648 Romanov RA , Zeisel A , Bakker J , Girach F , Hellysaz A , Tomer R , Alpar A , Mulder J , Clotman F , Keimpema E , (2017). Molecular interrogation of hypothalamic organization reveals distinct dopamine neuronal subtypes. Nature neuroscience 20 , 176–188.27991900 Russ DE , Cross RBP , Li L , Koch SC , Matson KJE , Yadav A , Alkaslasi MR , Lee DI , Le Pichon CE , Menon V , (2021). A harmonized atlas of mouse spinal cord cell types and their spatial organization. Nature communications 12 , 5722. Saelens W , Cannoodt R , Todorov H , and Saeys Y (2019). A comparison of single-cell trajectory inference methods. Nature biotechnology 37 , 547–554. Sagner A , and Briscoe J (2019). Establishing neuronal diversity in the spinal cord: a time and a place. Development 146 . Sanes JR , and Masland RH (2015). The types of retinal ganglion cells: current status and implications for neuronal classification. Annual review of neuroscience 38 , 221–246. Sathyamurthy A , Johnson KR , Matson KJE , Dobrott CI , Li L , Ryba AR , Bergman TB , Kelly MC , Kelley MW , and Levine AJ (2018). Massively Parallel Single Nucleus Transcriptional Profiling Defines Spinal Cord Neurons and Their Activity during Behavior. Cell reports 22 , 2216–2225.29466745 Saunders A , Macosko EZ , Wysoker A , Goldman M , Krienen FM , de Rivera H , Bien E , Baum M , Bortolin L , Wang S , (2018). Molecular Diversity and Specializations among the Cells of the Adult Mouse Brain. Cell 174 , 1015–1030 e1016.30096299 Scala F , Kobak D , Bernabucci M , Bernaerts Y , Cadwell CR , Castro JR , Hartmanis L , Jiang X , Laturnus S , Miranda E , (2021). Phenotypic variation of transcriptomic cell types in mouse motor cortex. Nature 598 , 144–150.33184512 Schaeffer S , and Iadecola C (2021). Revisiting the neurovascular unit. Nature neuroscience 24 , 1198–1209.34354283 Scheffer LK , Xu CS , Januszewski M , Lu Z , Takemura SY , Hayworth KJ , Huang GB , Shinomiya K , Maitlin-Shepard J , Berg S , (2020). A connectome and analysis of the adult Drosophila central brain. eLife 9 . Schmitz MT , Sandoval K , Chen CP , Mostajo-Radji MA , Seeley WW , Nowakowski TJ , Ye CJ , Paredes MF , and Pollen AA (2022). The development and evolution of inhibitory neurons in primate cerebrum. Nature 603 , 871–877.35322231 Schneider-Mizell CM , Bodor AL , Collman F , Brittain D , Bleckert A , Dorkenwald S , Turner NL , Macrina T , Lee K , Lu R , (2021). Structure and function of axo-axonic inhibition. eLife 10 . Schwarz MK , and Remy S (2019). Rabies virus-mediated connectivity tracing from single neurons. Journal of neuroscience methods 325 , 108365.31330160 Seung HS , and Sumbul U (2014). Neuronal cell types and connectivity: lessons from the retina. Neuron 83 , 1262–1272.25233310 Sharma N , Flaherty K , Lezgiyeva K , Wagner DE , Klein AM , and Ginty DD (2020). The emergence of transcriptional identity in somatosensory neurons. Nature 577 , 392–398.31915380 Shekhar K , Lapan SW , Whitney IE , Tran NM , Macosko EZ , Kowalczyk M , Adiconis X , Levin JZ , Nemesh J , Goldman M , (2016). Comprehensive Classification of Retinal Bipolar Neurons by Single-Cell Transcriptomics. Cell 166 , 1308–1323 e1330.27565351 Shekhar K , and Sanes JR (2021). Generating and Using Transcriptomically Based Retinal Cell Atlases. Annu Rev Vis Sci 7 , 43–72.34228938 Shekhar K , Whitney IE , Butrus S , Peng YR , and Sanes JR (2022). Diversification of multipotential postmitotic mouse retinal ganglion cell precursors into discrete types. eLife 11 . Slavov N (2021). Single-cell protein analysis by mass spectrometry. Curr Opin Chem Biol 60 , 1–9.32599342 Somogyi P , and Klausberger T (2005). Defined types of cortical interneurone structure space and spike timing in the hippocampus. The Journal of physiology 562 , 9–26.15539390 Stadler T , Pybus OG , and Stumpf MPH (2021). Phylodynamics for cell biologists. Science 371 . Stanley G , Gokce O , Malenka RC , Sudhof TC , and Quake SR (2020). Continuous and Discrete Neuron Types of the Adult Murine Striatum. Neuron 105 , 688–699 e688.31813651 Stephan T , Burgess SM , Cheng H , Danko CG , Gill CA , Jarvis ED , Koepfli KP , Koltes JE , Lyons E , Ronald P , (2022). Darwinian genomics and diversity in the tree of life. Proc Natl Acad Sci U S A 119 . Sulston JE , Schierenberg E , White JG , and Thomson JN (1983). The embryonic cell lineage of the nematode Caenorhabditis elegans. Developmental biology 100 , 64–119.6684600 Sun YC , Chen X , Fischer S , Lu S , Zhan H , Gillis J , and Zador AM (2021). Integrating barcoded neuroanatomy with spatial transcriptional profiling enables identification of gene correlates of projections. Nature neuroscience 24 , 873–885.33972801 Svensson V , da Veiga Beltrame E , and Pachter L (2020). A curated database reveals trends in single-cell transcriptomics. Database (Oxford) 2020 . Swanson LW (2000). What is the brain? Trends in neurosciences 23 , 519–527.11074261 Swanson LW (2012). Brain architecture: understanding the basic plan, 2nd edn (New York, NY: Oxford University Press). Sweeney MD , Zhao Z , Montagne A , Nelson AR , and Zlokovic BV (2019). Blood-Brain Barrier: From Physiology to Disease and Back. Physiological reviews 99 , 21–78.30280653 Tabula Muris, C., Overall, c., Logistical, c., Organ, c., processing, Library, p., sequencing, Computational data, a., Cell type, a., Writing, g., (2018). Single-cell transcriptomics of 20 mouse organs creates a Tabula Muris. Nature 562 , 367–372.30283141 Tabula Sapiens C , Jones RC , Karkanias J , Krasnow MA , Pisco AO , Quake SR , Salzman J , Yosef N , Bulthaup B , Brown P , (2022). The Tabula Sapiens: A multiple-organ, single-cell transcriptomic atlas of humans. Science 376 , eabl4896.35549404 Tanay A , and Sebe-Pedros A (2021). Evolutionary cell type mapping with single-cell genomics. Trends Genet 37 , 919–932.34020820 Tasic B , Menon V , Nguyen TN , Kim TK , Jarsky T , Yao Z , Levi B , Gray LT , Sorensen SA , Dolbeare T , (2016). Adult mouse cortical cell taxonomy revealed by single cell transcriptomics. Nature neuroscience 19 , 335–346.26727548 Tasic B , Yao Z , Graybuck LT , Smith KA , Nguyen TN , Bertagnolli D , Goldy J , Garren E , Economo MN , Viswanathan S , (2018). Shared and distinct transcriptomic cell types across neocortical areas. Nature 563 , 72–78.30382198 Taylor SR , Santpere G , Weinreb A , Barrett A , Reilly MB , Xu C , Varol E , Oikonomou P , Glenwinkel L , McWhirter R , (2021). Molecular topography of an entire nervous system. Cell 184 , 4329–4347 e4323.34237253 Thion MS , and Garel S (2020). Microglial ontogeny, diversity and neurodevelopmental functions. Current opinion in genetics & development 65 , 186–194.32823206 Tiklova K , Bjorklund AK , Lahti L , Fiorenzano A , Nolbrant S , Gillberg L , Volakakis N , Yokota C , Hilscher MM , Hauling T , (2019). Single-cell RNA sequencing reveals midbrain dopamine neuron diversity emerging during mouse brain development. Nature communications 10 , 581. Tosches MA , Yamawaki TM , Naumann RK , Jacobi AA , Tushev G , and Laurent G (2018). Evolution of pallium, hippocampus, and cortical cell types revealed by single-cell transcriptomics in reptiles. Science 360 , 881–888.29724907 Tremblay R , Lee S , and Rudy B (2016). GABAergic Interneurons in the Neocortex: From Cellular Properties to Circuits. Neuron 91 , 260–292.27477017 Tritschler S , Buttner M , Fischer DS , Lange M , Bergen V , Lickert H , and Theis FJ (2019). Concepts and limitations for learning developmental trajectories from single cell genomics. Development 146 . Turner NL , Macrina T , Bae JA , Yang R , Wilson AM , Schneider-Mizell C , Lee K , Lu R , Wu J , Bodor AL , (2022). Reconstruction of neocortex: Organelles, compartments, cells, circuits, and activity. Cell 185 , 1082–1100 e1024.35216674 Van Hove H , Martens L , Scheyltjens I , De Vlaminck K , Pombo Antunes AR , De Prijck S , Vandamme N , De Schepper S , Van Isterdael G , Scott CL , (2019). A single-cell atlas of mouse brain macrophages reveals unique transcriptional identities shaped by ontogeny and tissue environment. Nature neuroscience 22 , 1021–1035.31061494 Vanlandewijck M , He L , Mae MA , Andrae J , Ando K , Del Gaudio F , Nahar K , Lebouvier T , Lavina B , Gouveia L , (2018). A molecular atlas of cell types and zonation in the brain vasculature. Nature 554 , 475–480.29443965 von Buchholtz LJ , Ghitani N , Lam RM , Licholai JA , Chesler AT , and Ryba NJP (2021). Decoding Cellular Mechanisms for Mechanosensory Discrimination. Neuron 109 , 285–298 e285.33186546 Vormstein-Schneider D , Lin JD , Pelkey KA , Chittajallu R , Guo B , Arias-Garcia MA , Allaway K , Sakopoulos S , Schneider G , Stevenson O , (2020). Viral manipulation of functionally distinct interneurons in mice, non-human primates and humans. Nature neuroscience 23 , 1629–1636.32807948 Wagner DE , and Klein AM (2020). Lineage tracing meets single-cell omics: opportunities and challenges. Nature reviews Genetics 21 , 410–427. Wang Y , Eddison M , Fleishman G , Weigert M , Xu S , Wang T , Rokicki K , Goina C , Henry FE , Lemire AL , (2021). EASI-FISH for thick tissue defines lateral hypothalamus spatio-molecular organization. Cell 184 , 6361–6377 e6324.34875226 White JG , Southgate E , Thomson JN , and Brenner S (1986). The structure of the nervous system of the nematode Caenorhabditis elegans. Philosophical transactions of the Royal Society of London Series B, Biological sciences 314 , 1–340.22462104 Winkler EA , Kim CN , Ross JM , Garcia JH , Gil E , Oh I , Chen LQ , Wu D , Catapano JS , Raygor K , (2022). A single-cell atlas of the normal and malformed human brain vasculature. Science 375 , eabi7377.35084939 Winnubst J , Bas E , Ferreira TA , Wu Z , Economo MN , Edson P , Arthur BJ , Bruns C , Rokicki K , Schauder D , (2019). Reconstruction of 1,000 Projection Neurons Reveals New Cell Types and Organization of Long-Range Connectivity in the Mouse Brain. Cell 179 , 268–281 e213.31495573 Witvliet D , Mulcahy B , Mitchell JK , Meirovitch Y , Berger DR , Wu Y , Liu Y , Koh WX , Parvathala R , Holmyard D , (2021). Connectomes across development reveal principles of brain maturation. Nature 596 , 257–261.34349261 Wolff T , and Rubin GM (2018). Neuroarchitecture of the Drosophila central complex: A catalog of nodulus and asymmetrical body neurons and a revision of the protocerebral bridge catalog. The Journal of comparative neurology 526 , 2585–2611.30084503 Wu YE , Pan L , Zuo Y , Li X , and Hong W (2017). Detecting Activated Cell Populations Using Single-Cell RNA-Seq. Neuron 96 , 313–329 e316.29024657 Xia B , and Yanai I (2019). A periodic table of cell types. Development 146 . Xu S , Yang H , Menon V , Lemire AL , Wang L , Henry FE , Turaga SC , and Sternson SM (2020). Behavioral state coding by molecularly defined paraventricular hypothalamic cell type ensembles. Science 370 . Yamagata M , Yan W , and Sanes JR (2021). A cell atlas of the chick retina based on single-cell transcriptomics. eLife 10 . Yang AC , Vest RT , Kern F , Lee DP , Agam M , Maat CA , Losada PM , Chen MB , Schaum N , Khoury N , (2022). A human brain vascular atlas reveals diverse mediators of Alzheimer’s risk. Nature 603 , 885–892.35165441 Yao Z , Liu H , Xie F , Fischer S , Adkins RS , Aldridge AI , Ament SA , Bartlett A , Behrens MM , Van den Berge K , (2021a). A transcriptomic and epigenomic cell atlas of the mouse primary motor cortex. Nature 598 , 103–110.34616066 Yao Z , van Velthoven CTJ , Nguyen TN , Goldy J , Sedeno-Cortes AE , Baftizadeh F , Bertagnolli D , Casper T , Chiang M , Crichton K , (2021b). A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell 184 , 3222–3241 e3226.34004146 Yuste R , Hawrylycz M , Aalling N , Aguilar-Valles A , Arendt D , Armananzas R , Ascoli GA , Bielza C , Bokharaie V , Bergmann TB , (2020). A community-based transcriptomics classification and nomenclature of neocortical cell types. Nature neuroscience 23 , 1456–1468.32839617 Zeisel A , Hochgerner H , Lonnerberg P , Johnsson A , Memic F , van der Zwan J , Haring M , Braun E , Borm LE , La Manno G , (2018). Molecular Architecture of the Mouse Nervous System. Cell 174 , 999–1014 e1022.30096314 Zeisel A , Munoz-Manchado AB , Codeluppi S , Lonnerberg P , La Manno G , Jureus A , Marques S , Munguba H , He L , Betsholtz C , (2015). Brain structure. Cell types in the mouse cortex and hippocampus revealed by single-cell RNA-seq. Science 347 , 1138–1142.25700174 Zeng H , and Sanes JR (2017). Neuronal cell-type classification: challenges, opportunities and the path forward. Nature reviews Neuroscience 18 , 530–546.28775344 Zhang M , Eichhorn SW , Zingg B , Yao Z , Cotter K , Zeng H , Dong H , and Zhuang X (2021a). Spatially resolved cell atlas of the mouse primary motor cortex by MERFISH. Nature 598 , 137–143.34616063 Zhang Z , Zhou J , Tan P , Pang Y , Rivkin AC , Kirchgessner MA , Williams E , Lee CT , Liu H , Franklin AD , (2021b). Epigenomic diversity of cortical projection neurons in the mouse brain. Nature 598 , 167–173.34616065 Zhu C , Preissl S , and Ren B (2020). Single-cell multimodal omics: the power of many. Nature Methods 17 , 11–14.31907462 Zhu Y , Sousa AMM , Gao T , Skarica M , Li M , Santpere G , Esteller-Cucala P , Juan D , Ferrandez-Peral L , Gulden FO , (2018). Spatiotemporal transcriptomic divergence across human and macaque brain development. Science 362 . Zhuang X (2021). Spatially resolved single-cell genomics and transcriptomics by imaging. Nature methods 18 , 18–22.33408406

PMC009xxxxxx/PMC9361154.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0401220 6459 Phys Med Biol Phys Med Biol Physics in medicine and biology 0031-9155 1361-6560 35594853 9361154 10.1088/1361-6560/ac71f1 NIHMS1826457 Article Fast and memory-efficient reconstruction of sparse Poisson data in listmode with non-smooth priors with application to time-of-flight PET Schramm Georg 1 Holler Martin 2 1 Department of Imaging and Pathology, Division of Nuclear Medicine, KU Leuven, Belgium 2 Institute of Mathematics and Scientific Computing, University of Graz, Austria. MH is a member of NAWI Graz (https://www.nawigraz.at) and BioTechMed Graz (https://biotechmedgraz.at). 3 8 2022 27 7 2022 27 7 2022 27 7 2023 67 15 10.1088/1361-6560/ac71f1This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Objective: Complete time of flight (TOF) sinograms of state-of-the-art TOF PET scanners have a large memory footprint. Currently, they contain ~4·109 data bins which amount to ~17 GB in 32 bit floating point precision. Moreover, their size will continue to increase with advances in the achievable detector TOF resolution and increases in the axial field of view. Using iterative algorithms to reconstruct such enormous TOF sinograms becomes increasingly challenging due to the memory requirements and the computation time needed to evaluate the forward model for every data bin. This is especially true for more advanced optimization algorithms such as the stochastic primal-dual hybrid gradient (SPDHG) algorithm which allows for the use of non-smooth priors for regularization using subsets with guaranteed convergence. SPDHG requires the storage of additional sinograms in memory, which severely limits its application to data sets from state-of-the-art TOF PET systems using conventional computing hardware. Approach: Motivated by the generally sparse nature of the TOF sinograms, we propose and analyze a new listmode (LM) extension of the SPDHG algorithm for image reconstruction of sparse data following a Poisson distribution. The new algorithm is evaluated based on realistic 2D and 3D simulationsn, and a real dataset acquired on a state-of-the-art TOF PET/CT system. The performance of the newly proposed LM SPDHG algorithm is compared against the conventional sinogram SPDHG and the listmode EM-TV algorithm. Main results: We show that the speed of convergence of the proposed LM-SPDHG is equivalent the original SPDHG operating on binned data (TOF sinograms). However, we find that for a TOF PET system with 400 ps TOF resolution and 25 cm axial FOV, the proposed LM-SPDHG reduces the required memory from approximately 56 GB to 0.7 GB for a short dynamic frame with 107 prompt coincidences and to 12.4 GB for a long static acquisition with 5 · 108 prompt coincidences. Significance: In contrast to SPDHG, the reduced memory requirements of LM-SPDHG enables a pure GPU implementation on state-of-the-art GPUs - avoiding memory transfers between host and GPU - which will substantially accelerate reconstruction times. This in turn will allow the application of LM-SPDHG in routine clinical practice where short reconstruction times are crucial. pmc1 Introduction A major challenge of image reconstruction in positron emission tomography (PET) is noise suppression since the acquired emission data suffer from high levels of Poisson noise due to limitations in acquisition time, injectable dose and scanner sensitivity. To limit the transfer of the data noise into the image during model-based iterative reconstruction (MBIR), different strategies exist. One possibility is to add a “smoothing” prior to the data fidelity term in the cost function that is being optimized. In general, we can formulate the resulting optimization problem for any imaging system where the acquired data follow a Poisson distribution as (1) argminx≥0∑i=1m(Px+s)i−dilog((Px+s)i)︸Di((Px+s)i)+βR(Kx), where x is the image to be reconstructed, P is the linear forward model, d are the acquired data and s are additive contaminations. ∑i=1mDi((Px+s)i) is the negative Poisson log-likelihood, i is the index of the data bin and m is the total number of data bins. In the specific case of time of flight (TOF) PET, P is the time of flight (TOF) PET forward model including the effects of attenuation, normalization and limited spatial resolution, d are the acquired prompt TOF PET coincidences (the emission sinogram), and s are the estimated random and scattered coincidences. R(K·) is a “smoothing prior” consisting of a generic linear operator K that calculates local differences and a proper, convex, lower-semicontinous function R. The level of regularization is controlled by the non-negative scalar factor β. A specific example for K would be the gradient operator ∇, e.g. approximated by finite forward differences in the discretized setting. Combining the gradient operator for K with the mixed L2-L1 norm for R leads to the well-known Total Variation (TV) prior [1]. The TV prior, as well as many other advanced smoothing priors aiming for edge-preservation such as e.g. Total Generalized Variation (TGV) [2], Joint T(G)V [3, 4] Parallel Level Sets [5, 6] or directional Total Variation (DTV) [7], require the use of non-smooth functions for R as instrumental building block. This prevents the use of simple and efficient purely gradient-based optimization algorithms to solve (1). PDHG and SPDHG for PET reconstruction with non-smooth priors Using the fact that z↦D(z)≔∑i=1mDi(zi) and z ↦ βR(z) are convex, lower-semicontinuous functions and thus are equal to their convex biconjugates D∗∗(z)=supy〈z,y〉−∑i=1mDi∗(yi) and (βR)**(z) = supw⟨z, w⟩ – (βR)*(w), respectively, where Di∗ and (βR)* are the convex conjugates, and that (βR)*(w) = βR*(w/β), we can rewrite (1) as the saddle point problem (2) argminx≥0supy,w〈Px+s,y〉+〈Kx,w〉−∑i=1mDi∗(yi)−βR∗(w∕β), introducing the dual variables y and w, and the convex dual of the Poisson log-likelihood given as (3) Di∗(yi)={−di+dilog(di1−yi)ifyi<1anddi>0,0ifyi≤1anddi=0,∞else.} Under mild assumptions on R, which hold for all of the above-mentioned smoothing priors, Problem (2) is equivalent to (1) and can be solved even for non-smooth priors using the generic primal-dual hybrid gradient (PDHG) algorithm by Chambolle and Pock [8]. PDHG is an iterative algorithm that requires the evaluation of the complete forward and adjoint operator in every update. The usage of the original PDHG algorithm to solve (2) for real-world state-of-the-art TOF PET systems, however, usually results in extremely long computation times, because the evaluation of P and PT for state-of-the-art TOF PET systems is computationally very demanding, and because several hundreds to thousands of updates are needed to obtain reasonable convergence. To overcome this limitation, Chambolle et al. published a stochastic extension of PDHG called SPDHG for saddle point problems that are separable in the dual variable in 2018 [9]. In contrast to PDHG, SPDHG has the advantage that the complete forward and adjoint operator are split into n subsets, and that, in every update, only a random subset of the forward and adjoint operator chosen according to a probability pk has to be evaluated. In [10], Ehrhardt et al. applied SPDHG to 3D non-TOF PET reconstruction with TV-like priors and showed that around 10 complete projections and back projections can be sufficient to obtain reasonable convergence using SPDHG with 252 subsets. Moreover, the authors also demonstrated that preconditioning further accelerates convergence. The resulting SPDHG algorithm to solve (2) is summarized in Algorithm 1, where the proximal operator for a convex function f using the weighted norm ‖⋅‖M2=〈M−1⋅,⋅〉 induced by a symmetric and positive-definite matrix M is defined as (4) proxjM(y)=argminv(12‖v−y‖M2+f(v)). For the convex conjugate of the negative Poisson log-likehood Di∗, this proximal operator can be calculated point-wise and is given by (5) (proxDi∗Si(y))i=proxDi∗Si(yi)=12(yi+1−(yi−1)2+4Sidi). Algorithm 1 SPDHG for PET reconstruction [10] 1:Initializex(=0),y(=0),(w=0),(Si)i,T,(pi)i,2:z¯=z=PTy+KTw3:repeat4:x←proj≥0(x−Tz¯)5:Selecti∈{1,…,n+1}randomly according to(pi)i6:ifi≤nthen7:yi+←proxDi∗Si(yi+Si(Pix+si))8:δz←PiT(yi+−yi)9:yi←yi+10:else11:w+←βproxR∗Si∕β((w+SiKx)∕β)12:δz←KT(w+−w)13:w←w+14:endif15:z←z+δz16:z¯←z+(δz∕pi)17:untilstopping criterion fulfilled18:returnx The proximal operator for R* obviously depends on the choice of R but can be also efficiently computed using point-wise operations for many popular choices of R. As mentioned in [10], Algorithm 1 converges if we use the preconditioned step sizes Sk=γdiag(ρPk1)Tk=γ−1diag(ρpkPkT1), and Sn+1=γρ‖K‖Tn+1=γ−1pn+1ρ‖K‖, setting T = mink=1…,n+1 Tk pointwise, and choosing 0 < ρ < 1 and γ > 0. Limitations of PDHG and SPDHG for PET reconstruction Whilst SPDHG is a big step forward for an efficient solution of (2) in terms of computational speed, SPDHG also comes with two main limitations. First, as discussed in Remark 2 of [10], a potential drawback of SPDHG is that it requires keeping at least one more complete (TOF) sinogram in memory (the dual variable y). Moreover, if the proposed preconditioning is used, a second complete (TOF) sinogram (the sequence of step sizes (Sk)k=1n) needs to be stored in memory. Storing one or two extra sinograms in memory of modern computers is not a major problem for static single-bed non-TOF PET data, where sinogram sizes are relatively small. However, for simultaneous multi-bed, dynamic or TOF PET data, the size of the complete data sinograms can be become problematic. This issue gets even more severe when aiming for a complete GPU implementation of Algorithm 1 since the available memory on state-of-the-art GPUs is nowadays much smaller compared to CPUs. As an example, for a TOF PET scanner with 25 cm axial FOV and a TOF resolution of ca. 400 ps, a complete unmashed static TOF sinogram for one bed position has approximately 4.4 · 109 data bins, requiring ca. 17 GB of memory in 32 bit floating point precision. Note that with improved TOF resolution and increasing axial field of view (e.g. total body PET scanners), the memory required to store a complete TOF sinogram will continue to substantially increase in the future. Second, PDHG and SPDHG only work with “binned” data. In the case of TOF PET reconstruction, that means that the acquired raw listmode data first need to be binned into TOF sinograms and that P and PT need to be evaluated using sinogram projectors. For most acquisitions with modern TOF PET scanners, this is inefficient both in terms of memory and computational time since the TOF emission data are extremely sparse. In contrast, storage and processing of the data in listmode format (event by event) is usually more efficient. Sparsity of TOF PET data Compared to non-TOF PET emission sinograms, TOF PET emission sinograms of most acquisitions with state-of-the-art TOF PET scanners are extremely sparse. This is because every geometrical line of response (LOR) has to be subdivided into several small TOF bins. To achieve sufficient sampling of the TOF information, the number of TOF bins has to be inversely proportional to the TOF resolution of the scanner. Consequently, for a fixed number of acquired prompt coincidences, the sparsity of the sinogram is proportional to the TOF resolution. As an example, for a typical 80 s acquisition of a liver bed position in an FDG scan with an injected dose of 323 MBq acquired 60 min p.i. on a state-of-the-art TOF PET/CT scanner with 20 cm axial FOV, more than 94% of the data (TOF sinogram) bins are empty. An example that demonstrates this extreme sparsity of TOF sinograms is shown in Fig. 1. For shorter frames, as present, e.g., in the early phase of dynamic scans or when respiratory or cardiac gating is used, the fraction of empty bins can even higher. Note that we expect that the sparsity of the TOF PET emission data of future PET systems will increase faster than linear compared to the improvement of the TOF resolution. This is because with better TOF resolution, every detected event carries more information such that fewer detected events are needed to reconstruct images with the similar variance [11]. Moreover, the method presented in this work is applicable not only to TOF PET image reconstruction, but to all imaging reconstrution problems with sparse data following a Poisson distribution such as (low count) SPECT or photon counting CT. Contributions and Aim To improve the efficiency of SPDHG in terms of required memory and computation time when reconstructing sparse TOF PET data, we propose and analyze a listmode extension of the SPDHG algorithm called LM-SPDHG that allows event-by-event processing using dedicated listmode forward and back projectors. We first derive LM-SPDHG from SPDHG and show that the convergence of LM-SPDHG is as fast as the convergence of SPDHG based on dedicated numeric examples in 2D. Moreover, we analyze the memory requirements for LM-SPDHG compared to SPDHG for typical scans acquired on state-of-the-art TOF PET scanners. We emphasize that the focus of this work is not on finding the optimal (non-smooth) prior or prior strength for a given clinical task in PET imaging. Instead, we aim to provide a framework enabling fast and memory efficient reconstruction of sparse TOF PET listmode data which will finally facilitate future research on the use of (non-smooth) priors in PET reconstruction. 2 Theory and Algorithms Before deriving LM-SPDHG, the next subsection first of all shows how to reduce the memory requirements of SPDHG when reconstructing sparse TOF sinograms. Memory-efficient SPDHG for sparse TOF sinograms As shown in [12], the memory requirements for SPDHG can be substantially reduced by choosing a better initialization of the dual variable y. From Eq. (5) we can observe that for data bins i where di = 0 (bins in the TOF emission sinogram with zero counts), proxDi∗(ai)=1 for ai ≥ 1 and proxDi∗(ai)=ai otherwise. Moreover, provided that yi ≥ 1, we see that ai = yi + Si(Pix + si) ≥ 1 since all other quantities are non-negative. Consequently, if we initialize all bins of y where the data sinogram d equals zero with one, these bins of y remain 1 during all iterations. This in turn means that these bins do not contribute to the update of δz, z, z¯, and x, since only the difference between y and y+ is backprojected in line 8 of Algorithm 1 meaning that all bins without data do not have to be kept in memory during the iteration loop (lines 3 until 17). The only place where the empty data bins contribute is the initialization of z and z¯ in line 2. However, this single back projection can be split into smaller chunks to also reduce the required memory of this step. While at a first glance, the initialization of y proposed above might seem artificial, in fact it directly corresponds to choosing the optimal value for those yi where di = 0. To see this, note that an optimal solution (x^, y^, w^) of (2) in particular needs to satisfy the optimality condition (6) Px^∈∂D∗(y^), where ∂D* is the subdifferential of D*. Since this is equivalent to y^∈∂D(Px^), and since D (the negative Poisson log likelihood) is differentiable this leads to the condition (7) y^=1−dPx^+s, where the division is to be understood point-wise. With this we see that the proposed initialization directly corresponds to choosing the optimal value for yi where di = 0, which is known explicitly in this case. As argued above, this improved initialization of y naturally reduces the memory requirements of SPDHG and also improves the speed of convergence when a “warm start” for x0 is chosen. The latter can, e.g., be achieved by applying one iteration and a reasonable amount of subsets of the EM-TV algorithm [13, 14]. Since in EM-TV every update is split into a classical EM step followed by a weighted denoising optimization problem in image space, EM-TV can be used in listmode as well by simply modifying the EM step. Listmode SPDHG As indicated by the name, emission data in listmode format are a chronological list N of detected events e ∈ N, where each event is characterized by a small set of (integer) numbers (e.g. the number of the two detectors, the discretized TOF difference, and a time stamp). To process the listmode data during reconstruction without binning it into a sinogram, we introduce the listmode forward operator PNLM mapping the image data x to a data-vector of dimension ∣N∣ via (8) (PNLMx)e=(Px)ie,for eache∈N, where ie is the sinogram bin in which event e was detected. Also, we denote by sLM with seLM=sie the listmode-based scatter and random estimate. Re-writing the gradient of the negative Poisson log likelihood using listmode data and the listmode operators is straightforward, such that any gradient-based PET reconstruction algorithm can be easily adapted to listmode data. For a Bayesian approach with prior R(K·) as in (1) this is less immediate, but we can show that indeed (1) can be equivalently re-written into a minimization problem involving only the listmode forward operator that is of the same form as (1) as shown in appendix A. This allows us to extend the SPDHG algorithm for listmode data and yields the algorithm as shown in Algorithm 2. Algorithm 2 LM-SPDHG for PET reconstruction 1:Inputevent listN,contamination listsN2:Calculateevent countsμefor each e inN(see text)3:Initializex,w,(Si)i,T,(pi)i4:InitializelistyN=1−(μN∕(PNLMx+sN))5:Preprocessingz¯=z=PT1−PLMT(yN−1)∕μN+KTw6:SplitlistsN,sNandyNintonsublistsNi,yNiandsNi7:repeat8:x←proj≥0(x−Tz¯)9:Selecti∈{1,…,n+1}randomly according to(pi)i10:ifi≤nthen11:yNi+←proxD∗Si(yNi+Si(PNiLMx+sNiLM))12:δz←PNiLMT(yNi+−yNiμNi)13:yNi←yNi+14:else15:w+←βproxR∗Si∕β((w+SiKx)∕β)16:δz←KT(w+−w)17:w←w+18:endif19:z←z+δz20:z¯←z+(δz∕pi)21:untilstopping criterion fulfilled22:returnx In contrast to the original SPDHG using binned data (sinograms), the forward and adjoint PET operators (lines 11 and 12) have been replaced by their listmode equivalents as in (8). Moreover, the dual variable for the data fidelity is replaced by the list yN, which has the same length as the measured event list N. If an event with a fixed sinogram bin ie occurs more than once in the event list N, it is also forward and back-projected multiple times in steps 11 and 12. To compensate for this fact, the algorithm divides by the event count1 μe before back projection in line 12. Note that (i) as shown in Fig. 1 in most standard acquisitions the event count of most events is 1 and that (ii) calculating the event count μe which is a prerequisit creates a small pre-processing overhead (step 2). However, when implemented on a modern GPU this overhead is small compared to the computation time needed to calculate all iterations. Moreover, it is in the similar order of the time needed to unlist the native listmode data into a sinogram which is a prerequisit and pre-processing overhead for SPDHG. Another difference of LM-SPDHG compared to SPDHG is the fact that we split the data into n subsets by assigning every n-th event of the complete event list N to the n-th sub list - as commonly done when defining subsets in listmode OSEM. In that way, we can think of the subset listmode forward operator PNiLM as the full forward operator P with a sensitivity reduced by a factor of n. Accordingly, we set the step sizes associated with the subset listmode PET operators to (9) Sk=γdiag(ρPNkLM1)Tk=γ−1diag(ρpkPT1∕n). At a first glance, the initialization of z and z¯ in step 5 of Algorithm 2 might look odd. However, the first part of the expression is equivalent to applying the adjoint sinogram operator PT to a sinogram initialized according to (7). Note that step 5 still requires to calculate a sinogram back projection of a unity sinogram (PT1) which we would like to avoid in listmode data processing. Unfortunately, avoiding this single sinogram backprojection is not possible. However, note that this “sensitivity image“ is also needed in the calculation of the step sizes Tk and also in the listmode EM-TV algorithm that we use to initialize x. Hence, we recommend pre-computing the sensitivity image (PT1) and storing it in memory. In summary, compared to SPDHG, LM-SPDHG as shown in Algorithm 2 has the advantage that (i) only lists (N, yN, sN, μN) instead of complete sinograms (y and d) have to be stored in memory and (ii) all projections and back projections can be performed using listmode projectors. This means that for all acquisitions where the number of detected events is substantially smaller than the number of data bins, the required memory and computation time is reduced. The latter depends on the actual implementation of the sinogram and listmode projectors and on the computational hardware. According to our experience using a state-of-the art GPU implementation of Joseph projectors [15] for a TOF PET scanner with 400 ps TOF resolution and 25 cm axial field of view, a forward and back projection is faster in listmode if approximately less than 3e8 events have to be processed. For 7e7 and 1e7 counts, the projections are approximately faster by a factor of 3 and 5, respectively2. Note that the difference in the computation time of the projection for sinogram and listmode depends on the specific implementation - especially on the optimization of the memory access - and the computational hardware. A detailed comparison of the required memory for SPDHG and LM-SPDHG for different typical PET acquisitions is listed in Table 1. 3 Numerical Experiments Recalling the fact that for most common TOF PET acquisitions, LM-SPDHG requires less memory and is also faster, one would probably prefer LM-SPDHG over SPDHG if the speed of convergence is the same. In the following, we describe a set of numerical experiments that we conducted to show that this is indeed the case. In the absence of an analytical solution to the optimization problems (1) and (2), we analyzed the convergence of LM-SPDHG and SPDHG with respect to a reference solution x* as done in [10]. Convergence was monitored by tracking the relative cost function (10) crel(x)=(c(x)−c(x∗))∕(c(x0)−c(x∗)), where c(x) is the cost function to be optimized in (1) and x0 is the initialization used for x. Moreover, convergence was also montiored in image space by tracking the peak signal-to-noise ratio with respect to x* (11) PSNR(x)=20log10(‖x∗‖∞∕MSE(x,x∗)), where MSE is the mean squared error. The reference solution was obtained by running the deterministic PDHG (SPDHG without subsets) for 20000 iterations. Since running PDHG with 20000 iterations using realistic 3D TOF PET data takes a very long time (approx. 250 h), all numerical convergence experiments were performed using simulated 2D TOF PET data. A 2D software brain phantom with a gray to white matter contrast of 4:1 was created based on the brainweb phantom [16] and used to generate simulated 2D TOF data including the effects of limited spatial resolution, attenuation and a flat contamination mimicking random and scattered coincidences with a contamination fraction of 42%. The geometry and TOF resolution of the simulated 2D PET system was chosen to mimic one direct plane of a state-of-the art TOF PET scanner with a ring diameter of 650 mm and a 400 ps TOF resolution (sinogram dimension: 357 radial bins, 224 projection angles, 27 TOF bins). Noisy simulated prompt emission TOF sinograms and corresponding listmode data were generated for two different count levels (5e5 and 5e6 prompt counts). Unless stated otherwise, we always applied the EM-TV algorithm, summarized in Algorithm 3, using 1 iteration and 28 subsets to initialize x0 and y0 according to (7). To solve the weighted denoising problem in step 8 of the EM-TV algorithm, we applied the accelerated PDHG algorithm [8] using 20 iterations. In all SPDHG and LM-SPDHG reconstructios, the step size ratio γ was set to 3/∥x0∥∞ and ρ was set to 0.999. When splitting the data into n subsets, the vector of probabilities determining whether an update with respect to a subset of the data or with respect to the prior is done was set to (12) pk={12nifk≤n12else,} such that on average an update with respect to the prior was done on every second update as suggested in [10]. In this work, we use the term iteration for 2n updates such that on average in every iteration the complete data are forward and backprojected once and n updates with respect to the prior are performed. As benchmark examples for non-smooth priors, we consider two “Total Variation like” priors in this work. These priors have the form (13) R(Kx)=‖Kx‖2,1=∑i∑j(Kx)ij2, where ∥Kx∥2,1 is the sum over all entries of the pointwise Euclidean norm of Kx (mixed L2-L1 norm). The proximal operator for the convex dual of this prior is given by (14) (proxR∗S(w))i=wimax(1,∣wi∣). Algorithm 3 listmode EM-TV for PET reconstruction [13, 14] 1:Inputevent listN,contamination listsN2:SplitlistsNandsNintonsublistsNi,andsNi3:Pre-computesensitivity imageg=PT14:repeat5:Select subseti6:z←xngPNiLMT1PNiLM+sNi(subset EM step)7:w←g∕(βx)8:x+←argminu≥0∑jwj2(uj−zj)2+R(Ku)9:untilstopping criterion fulfilled10:returnx For the linear operator K, we first used the finite forward difference operator resulting in the classical Total Variation (TV) prior [1]. Moreover, we also implemented the Directional Total Variation (DTV) prior incorporating structural information by only considering the component of the finite difference vector in every voxel that is perpendicular to the spatial gradient of a prior image in that voxel [7]. In our convergence experiments using DTV, we assumed perfect structural prior information meaning that the prior image used for DTV, was the image itself with a flipped contrast. Note that in real acquisitions the quality of the available structural prior information is of course inferior. However, we do not expect that this fact strongly affects the convergence properties of SPDHG and LM-SPDHG. To demonstrate that LM-SPDHG also works for realistic 3D TOF PET listmode data, we simulated and reconstructed 3D TOF PET data based on the 3D XCAT [17] phantom and a state-of-the-art TOF PET scanner with 20 cm axial field of view, a ring diameter of 650 mm and a 400 ps TOF resolution (sinogram dimension: 357 radial bins, 224 projection angles, 1296 direct and oblique planes, 27 TOF bins). As before, he data simulation included the effects of attenuation, finite resolution and a flat contamination with a contamination fraction of 42%. In total, 7e7 prompt counts were simulated corresponding approximately to a standard 80 s 1 h p.i. FDG liver bed position acquisition. Simulated listmode data were reconstructed using LM-SPDHG with 50 iterations and 224 subsets. Last but not least, we also reconstructed a real TOF PET data set of the NEMA image quality phantom acquired on a GE Discovery MI 4 ring TOF PET/CT [18] (20 cm axial FOV, sinogram dimension: 357 radial bins, 272 projection angles, 1261 direct and oblique planes, 29 TOF bins, 400 ps TOF resolution). The activity contrast between the sphere inserts and the background was 10:1 and the acquisition contained 4.7e7 prompt (2.2e7 true) coincidences. Reconstructions using a TV prior with β = 6 were performed with EM-TV using 100 iterations with 1 and 28 subsets, and with LM-SPDHG using 100 iterations and 224 subsets. We also performed a reconstruction of the same data set with LM-PDHG using 20000 iterations and 1 subset. This reconstruction was used as a reference to calculate the relative cost and PSNR. All reconstruction algorithms used in this work were implemented in python3 using the open-source “parallelproj” CUDA projector libraries available at https://github.com/gschramm/parallelproj. An Nvidia V100 GPU with 16 GB RAM was used to perform all reconstructions. Upon publication of this article, an open-source implementation of the LM-SPDHG algorithm will be made availalbe in the associated “pyparallelproj” python package. Preprocessing of the NEMA image quality phantom data set was performed using GE’s duetto PET reconstruction toolbox v2.17. SPDHG using a warm start vs a cold start Before investigating the convergence behavior of LM-SPDHG, we first tested whether a warm start could already help to lead to faster convergence of SPDHG using sinograms. Figure 2 shows the results of SPDHG reconstructions with a cold (x0 = 0 and y0 = 0) and warm start as described above for a data set with 3e5 true counts and a TV prior with β = 0.03. It can be seen that SPDHG with the warm start performs better in terms of relative cost and PSNR with respect to the reference reconstruction. A similar trend was observed for higher count levels, with different β values and using the DTV prior indicating that as expected the warm start helps to accelerate convergence in the early iterations. Convergence of LM-SPDHG compared to SPDHG Figure 3 summarizes the convergence comparison between sinogram SPDHG and LM-SPDHG for the same data set and prior as described in the previous subsection using the same warm start for both algorithms. As demonstrated by the convergence metrics and the reconstructed images, the convergence of sinogram SPDHG and LM-SPDHG is almost identical. This also holds for different count levels and for the structural DTV priors as shown in subfigures (b) and (c) and for different levels of regularization as shown in supplementary Figs. 1 and 2. For the example with the DTV prior shown in subfigure (c), the convergence in terms of PSNR seems to be even slightly faster with LM-SPDHG, which can be also seen in the difference images with respect to the reference reconstruction. Convergence of LM-SPDHG compared to listmode EM-TV A comparison between the convergence of LM-SPDHG using 224 subsets and listmode EM-TV using 1, 7 and 28 subsets is shown in Fig. 4. In all sub figures, LM-SPDHG converges much faster than EM-TV. Interestingly, when using listmode EM-TV with more than one subset, the convergence metrics saturate meaning that the algorithm remains on a limited cycle and the optimal solution is not reached. When using one subset, the convergence of EM-TV is much slower but does not seem to saturate after 100 iterations. Convergence vs number of subsets Figure 5 shows the convergence of LM-SPDHG as a function of the number of data subsets used for low (3e5) and high (3e6) counts and the TV and DTV prior. In general, using 224 subsets leads to faster convergence compared to 56 and 112 subsets. However, for the TV prior, especially at low counts, there is almost no difference between using 56, 112, and 224 subsets. In contrast, for the DTV prior, the difference between using 56, 112, and 224 subsets is more pronounced. For a fixed number of subsets, convergence is faster for high count compared to low count data sets. Reconstrution of 3D XCAT data Figure 6 shows the results of the reconstruction of the listmode data simulated from the 3D XCAT phantom using a TV prior with β = 0.03. From the plot of the cost function and the reconstructions shown in the middle, we can see that reasonable convergence is reached after approximately 25 iterations with 224 subsets. Note that in contrast to our 2D experiments, the cost function initially increases until the 3rd iteration before it decreases and stabilizes after around 25 iterations. As mentioned above, calculating the projections in listmode for 7e7 counts is roughly a factor of 3 faster compared to sinogram-based processing which substantially speeds up every iteration. Note that in our proof-of-concept implementation, the effective speed up is less than a factor of 3 since we did not implement the gradient operators and proximal mappings on a GPU yet which leads to overhead due to gradient-based updates. However, we expect that, once properly implemented on GPUs, this overhead will be small compared to the time needed to calculate the actual TOF projections. Reconstrution of NEMQ IQ phantom data Figure 7 shows the results of the reconstructions of the listmode data from the acquisition of the NEMA image quality phantom on the GE Discovey MI 4 ring TOF PET system. After 100 iterations, LM-SPDHG using 224 subsets has lower cost compared to both EM-TV reconstructions and is closest to the referene reconstruction in terms of PSNR. From the differene images it can be seen that the stochastic LM-SDPHG mostly differs from the reference in the background outside the phantom. In contrast, both EM-TV reconstructions show bigger residual differences within the phantom which are most pronounced in the “warm” background for EM-TV using 28 subsets and in the “cold” central insert for EM-TV using 1 subset. 4 Discussion The results of our numerical experiments presented in this work demonstrate that the speed of convergence of LM-SPDHG is essentially the same as the one of the original SPDHG using sinograms. For clinical acquisitions with state-of-the-art TOF PET scanners resulting in extremely sparse data, LM-SPDHG has two distinct advantages compared to SPDHG. First, during the iterations all forward and back projections can be performed in listmode which is faster compared to sinogram projectors for sparse TOF data. Second, as shown in Table 1, the memory requirements are substantially reduced such that even for an aquisition with 5e8 prompt coincidences only 12.5 GB of memory are needed. This actually enables a pure GPU-implementation of LM-SPDHG avoiding intermediate memory transfer between the host and a state-of-the-art GPU with a memory of approximately 16 GB. Note that in our proof of concept implementation of LM-SPDHG we used a hybrid computing approach where only the TOF PET forward and back projections where computed using a GPU. We expect that once properly implemented purely on a GPU, the computation time required for LM-SPDHG will further decrease since memory transfter between host and GPU is a bottleneck in our implementation. Note that a pure GPU implementation of the conventional SPDHG algorithm for modern TOF PET data is more complicated since more than 50 GB of GPU is required. In this proof of concept work, we only used two non-smooth TV-like priors to benchmark LM-SPDHG. Note that in general, SPDHG and LM-SPDHG can be also used for smooth priors as long as the proximal operator of the convex dual of the priors can be efficiently calculated which is e.g. the case for prior penalizing the squared L2 norm or Huber TV. However, for smooth priors the speed of convergence of (LM)-SPDHG should be benchmarked against other stochastic gradient-based optimization techniques such as the stochastic variance reduced gradient (SVRG) [19] or SAGA [20] which is beyond the scope of this work. We also note that in this work we used a scalar spatially-invariant prior strength β which leads to a spatially-variant local pertubation response (LPR). For pratical applications where a spatially-invariant reponse is desirable, a spatially-varient prior strength should be used as described in [21, 22]. We emphasize that the listmode EM-TV algorithm is still a very practical and useful algorithm to approximate the solution of the optimization problem (1) for listmode data and non-smooth priors. However, as shown in Fig. 4, it seems that when using ordered subsets for acceleration, EM-TV does not reach the optimal solution but rather remains on a limit cycle similar to the behaviour of OSEM. Whether the difference between the limit cycle and the true optimal solution is of importance for a given count level and clinical task, should be investigated in the future. As shown in Fig. 5, choosing the “optimal” number of subsets for LM-SPDHG seems to depend on the prior and the number of acquired counts. For a reconstruction of a high-count scan with DTV, it is possible that LM-SPDHG with more than 224 subsets converges even faster. A detailed investigation of the choice of the number of subsets for different acquistion and reconstruction scenarios is left for future research. Last but not least, we would like to note that according to our experience, SPDHG and LM-SPDHG for TOF PET reconstruction also converge for ρ slightly greater than one which might be of interest for practical applications where stopping as early as possible can be important. Suppl. Fig. 3 shows that in our 3D XCAT example, convergence can be further accelerated by using ρ = 8 and γ = 30∥x0∥,∞. A detailed investigation of the convergence speed as a function of ρ and γ potentially involving an iteration-dependent choice of ρ [23] is, however, beyond the scope of this manuscript and left for future research. 5 Conclusion For sparse TOF data, the proposed LM-SPDHG algorithm severely reduces the memory requirements and computation times compared to the original SPDHG enabling the application of LM-SPDHG in routine clinical practice where short reconstruction times are crucial. Supplementary Material pmbac71f1supp1.pdf Acknowledgements The authors wold like to thank Claire Delplancke for the insightful discussion on the initialization of (S)PDHG. Moreover, the authors would like to thank Prof. Kristof Baete for the acquisition of the NEMA data set and Dr. Ahmadreza Rezaei for preprocessing of the same data set. This work was supported in part by the NIH grant 1P41EB017183-01A1. Appendix A Listmode re-formulation of the sinogram-based minimization problem Here, as basis for the proposed LM-SPDHG algorithm, we provide details on the listmode re-formulation of the sinogram-based minimization problem (1). To this aim, we denote the dependence of Di() on the data di explicitly via D(·; di). Further, recall that for N the list of detected events, ie is the sinogram bin in which event e ∈ N was detected. With e0 ∉ N denoting an additional, artificial event summarizing zero measurements we then obtain ∑i=1mD((Px+s)i;di)=∑i=1m∑e∈N:ie=i(Px+s)iedie−log((Px+s)ie)+∑i:di=0(Px)i=∑e∈N(Px+s)iedie−log((Px+s)iedie)−∑e∈Nlog(die)+∑i:di=0(Px)i=∑e∈N∪{e0}D((PNLMx+sLM)edie;deLM)−∑e∈Nlog(die)=∑e∈N∪{e0}D(P^NLMx+s^LM)e;deLM)−∑e∈Nlog(die) where deLM=1 for e ∈ N, de0LM=0, sLM with seLM=sie is the list-mode random and scatter estimate and (PLMu)e={(Pu)ieife∈N∑i:di=0(Pu)iife=e0.} The last step uses the rescaled operator P^LM and the rescaled estimate s^LM defined as (P^LMu)e=(PLMu)e∕die and (s^LM)e=sLM∕die for e ∈ N, and (P^LMu)e0=(PLMu)e0 and (s^LM)e0=0, respectively. Now the last line of the above reformulation provides an equivalent data term which (after removing the constant ∑e∈Nlog(die)) can be used to obtain an equivalent list-mode-based reformulation of (1). Applying the SPDHG for this reformulation results in the proposed LM-SPDHG Algorithm 2. Note that there, the non-rescaled operator PLM and estimate sLM appear since i) on line 11 of the algorithm, the rescaling of PLM and sLM cancels out with the rescaling of PLM in the step size Sk of (9), and ii) the rescaling of PLM in line 12 is incorporated explicitly. Furthermore, note that we use the full forward operator P instead of PNkLM for the stepsize Tk, for reasons explained in the paragraph above the respective equation (9). Figure 1: Representative slices through a single view of an emission sinogram of a 80s [18F]FDG acquisition of a liver bed position. The scan was acquired 1 h post injection with an injected dose of 323 MBq on a GE DMI PET/CT with a TOF resolution of 400 ps (29 TOF bins). The horizontal and vertical axis represent the radial and axial direction (direct planes only), respectively. (top) sum over all TOF bins. (middle) central TOF bin 15/29 with 94% empty bins. (bottom) TOF bin 22/29 with 97% empty bins. Figure 2: Comparison of convergence of sinogram SPDHG using a cold and warm start for 3e5 counts and a TV prior with β = 0.03 using 224 subsets. For the warm start, x0 was taken from 1 EM-TV iteration with 28 subsets and y0 was calculated according to (7). Left column: relative cost and PSNR with respect to the reference solution as a function of the iterations. Second column: reference PDHG reconstruction using 20000 iterations (top) and initializer x0 used in the warm start. Third column: Reconstructions after 3 iterations with cold start (top) and warm start (bottom). Fourth column: Reconstructions after 20 iterations. Fifth and sixth column: Absolute differences of the SPDHG reconstructions with respect to the reference PDHG reconstruction. The range of the color map for the difference images is ±10% of the maximum of the reference reconstruction. Note that in the calculation of the relative cost the same x0 was used and that in both cases the same step size ratio γ was used. Moreover, in the very early iterations the relative cost of SPDHG is greater 1 meaning that the cost is worse compared the the initialization. This is due to the fact that (S)PDHG in general is a non-monotonic algorithm. Figure 3: Comparison of convergence of sinogram SPDHG and LM-SPDHG for using 244 subsets. All three subfigures (a) - (c) show the same quantities but for different count levels and priors. Left column: relative cost and PSNR to with respect to the reference solution as a function of the iterations for sinogram SPDHG (blue) and LM-SPDHG (orange). Second column: SPDHG (top) and LM-SPDHG (bottom) reconstruction after 100 iterations. Third column: absolute difference between SPDHG/LM-SPDHG and reference PDHG reconstruction. Forth column: reference PDHG reconstruction using 20000 iterations (top) and ground truth used to generate the data (bottom). Figure 4: Same as Fig. 3 but comparing the convergence of LM-SPDHG using 224 subsets and listmode EM-TV using 1, 7 and 28 subsets. The top row of images shows the reconstruction after 100 iterations and the bottom row shows the difference to the reference reconstruction which is shown in Fig. 3. Note that the width of the color window in the difference plots is five times wider compared to Fig. 3. Figure 5: Converge of LM-SPDHG for different number of subsets at two count levels for the TV and DTV prior. Note that when increasing the number of data subsets, the number of gradient updates per iteration increases as well. Figure 6: Convergence of LM-SPDHG for reconstruction of 3D TOF data generated from the XCAT phantom using 4e7 true (7e7) prompt counts and a TV prior with β = 0.03. Left column: evolution of cost function. Columns 2-3: transversal, sagittal and coronal slice of LM-SPDHG reconstruction after 20 iterations and 224 subsets. Columns 4-5: LM-SPDHG reconstruction after 100 iterations and 224 subsets. Columns 6-7: Ground truth image. Figure 7: Convergence of LM-SPDHG for reconstruction of real 3D TOF data from an acqusition of the NEMA image quality phantom on a GE Discovey MI 4 ring TOF PET system with a TV prior and β = 6 with 2.2e7 true (4.7e7 prompt) counts. First column, top two rows: transaxial and coronal slice of reference LM-PDHG reconstruction using 20000 iterations and 1 subset. Second column, top two rows: LM-SPDHG reconstruction using 100 iterations and 224 subsets. Third column, top two rows: EM-TV reconstruction using 100 iterations and 1 subset. Fourth column, top two rows: EM-TV reconstruction using 100 iterations and 28 subsets. Fist column, bottom two rows: relative cost and PSNR to reference reconstruction. Second, third, fourth column, bottom two rows: absolute difference compared to reference reconstruction. Table 1: Estimation of required memory for SPDHG and LM-SPDHG assuming an image size of (300,300,125) and a TOF sinogram with 357 radial elemnts, 224 views, 1981 direct and oblique planes, and 27 TOF bins corresponding to a TOF PET scanner with 400 ps TOF resolution and 25 cm axial FOV for three counts levels. 5e8 counts approximately correspond to a high count 20 min late static FDG brain scan, 7e7 counts to an 80 s static FDG body bed position 1 h p.i., and 1e7 counts to a short early frame in a dynamic acquisition. In this estimation, we assume that every TOF PET listmode even can be encoded using 10 bytes. counts 5e8 7e7 1e7 9 images 0.45 GB 0.45 GB 0.45 GB SPDHG 1 uint, 3 float sino. 55.6 GB 55.6 GB 55.6 GB LM-SPDHG 9 images 0.45 GB 0.45 GB 0.45 GB event list, 1 uint, 3 float lists 12.0 GB 1.7 GB 0.3 GB 9 images 0.45 GB 0.45 GB 0.45 GB EM-TV event list, 2 float list 9.0 GB 1.3 GB 0.2GB 1 the event count μe represents the number of times the event e occurs in the event list N. For TOF PET, a data bin is characterized by the combination of geometrical LOR and the TOF bin along the LOR (a bin in a TOF sinogram). 2 The reported computational times for the projections include the time needed to transfer the image and projection data to and from host to the GPU and thus correspond to a hybrid CPU/GPU computational model. References [1] Rudin LI , Osher S , and Fatemi E . “Nonlinear total variation based noise removal algorithms”. In: Physica D 60 .1-4 (1992), pp. 259–268. [2] Bredies K , Kunisch K , and Pock T . “Total generalized variation”. In: SIAM Journal on Imaging Sciences 3.3 (2010), pp. 492–526. doi: 10.1137/090769521. [3] Rigie DS and La Rivière PJ . “Joint reconstruction of multi-channel, spectral CT data via constrained total nuclear variation minimization.” In: Physics in medicine and biology 60.5 (2015), pp. 1741–62. doi: 10.1088/0031-9155/60/5/1741. [4] Knoll F , Holler M , Koesters T , “Joint MR-PET reconstruction using a multi-channel image regularizer”. In: IEEE Transactions on Medical Imaging 36.1 (2016), pp. 1–16. doi: 10.1109/TMI.2016.2564989. [5] Ehrhardt M , Markiewicz P , Liljeroth M , “PET Reconstruction with an Anatomical MRI Prior using Parallel Level Sets”. In: IEEE Transactions on Medical Imaging 35.9 (2016), pp. 2189–2199. doi: 10.1109/TMI.2016.2549601. [6] Schramm G , Holler M , Rezaei A , “Evaluation of Parallel Level Sets and Bowsher’s Method as Segmentation-Free Anatomical Priors for Time-of-Flight PET Reconstruction”. In: IEEE Transactions on Medical Imaging 37.2 (2017), pp. 590–603. doi: 10.1109/TMI.2017.2767940. [7] Ehrhardt MJ and Betcke MM . “Multicontrast MRI Reconstruction with Structure-Guided Total Variation”. In: SIAM Journal on Imaging Sciences 9.3 (2016), pp. 1084–1106. [8] Chambolle A and Pock T . “A first-order primal-dual algorithm for convex problems with applications to imaging”. In: Journal of Mathematical Imaging and Vision 40.1 (2011), pp. 120–145. doi: 10.1007/s10851-010-0251-1. [9] Chambolle A , Ehrhardt MJ , Richtarik P , “Stochastic primal-dual hybrid gradient algorithm with arbitrary sampling and imaging applications”. In: SIAM Journal on Optimization 28.4 (2018), pp. 2783–2808. doi: 10.1137/17M1134834. [10] Ehrhardt MJ , Markiewicz P , and Schönlieb CB . “Faster PET reconstruction with non-smooth priors by randomization and preconditioning”. In: Physics in Medicine and Biology 64.22 (2019). doi: 10.1088/1361-6560/ab3d07. [11] Tomitani T . “Image reconstruction and noise evaluation in photon time-of-flight assisted positron emission tomography”. In: IEEE Transactions on Nuclear Science 28 (1981), pp. 4581–4589. [12] Schramm G and Holler M . “Fast and memory-efficient reconstruction of sparse TOF PET data with non-smooth priors”. In: Proceedings of the 16th Virtual International Meeting on Fully 3D Image Reconstruction in Radiology and Nuclear Medicine. arXiv, 2021. [13] Sawatzky A , Brune C , Wübbeling F , “Accurate EM-TV Algorithm in PET with Low SNR”. In: IEEE Nuclear Science Symposium Conference Record. 2008, pp. 5133–5137. doi: 10.1109/NSSMIC.2008.4774392. [14] Sawatzky A , Brune C , Kösters T , “EM-TV Methods for Inverse Problems with Poisson Noise”. In: Level Set and PDE Based Reconstruction Methods in Imaging. Springer, 2013, p. 110. DOI: 10.1007/978-3-319-01712-9. [15] Joseph PM . “An Improved Algorithm for Reprojecting Rays Through Pixel Images”. In: IEEE Transactions on Medical Imaging 1.3 (1982), pp. 192–196. doi: 10.1109/TMI.1982.4307572. [16] Collins D , Zijdenbos A , Kollokian V , “Design and construction of a realistic digital brain phantom”. In: IEEE Transactions on Medical Imaging 17.3 (1998), pp. 463–468. doi: 10.1109/42.712135. [17] Segars WP , Sturgeon G , Mendonca S , “4D XCAT phantom for multimodality imaging research”. In: Medical Physics 37.9 (2010), pp. 4902–4915. doi: 10.1118/1.3480985. [18] Hsu DF , Ilan E , Peterson WT , “Studies of a next-generation silicon-photomultiplier-based time-of-flight PET/CT system”. In: Journal of Nuclear Medicine 58.9 (2017), pp. 1511–1518. doi: 10.2967/jnumed.117.189514. [19] Johnson R and Zhang T . “Accelerating Stochastic Gradient Descent using Predictive Variance Reduction”. In: Advances in Neural Information Processing Systems. Ed. by Burges CJC , Bottou L , Welling M , Vol. 26 . Curran Associates, Inc., 2013. [20] Defazio A , Bach F , and Lacoste-Julien S . “SAGA: A Fast Incremental Gradient Method With Support for Non-Strongly Convex Composite Objectives”. In: Advances in Neural Information Processing Systems. Ed. by Ghahramani Z , Welling M , Cortes C , Vol. 27 . Curran Associates, Inc., 2014. [21] Ahn S and Leahy RM . “Analysis of resolution and noise properties of nonquadratically regularized image reconstruction methods for PET”. In: IEEE Transactions on Medical Imaging 27.3 (2008), pp. 413–424. doi: 10.1109/TMI.2007.911549. [22] Tsai Y.-j. , Member S , Schramm G , “Benefits of Using a Spatially-Variant Penalty Strength With Anatomical Priors in PET Reconstruction”. In: IEEE Transactions on Medical Imaging 39.1 (2020), pp. 11–22. doi: 10.1109/TMI.2019.2913889. [23] Goldstein T , Li M , and Yuan X . “Adaptive primal-dual splitting methods for statistical learning and image processing”. In: Advances in Neural Information Processing Systems. 2015, pp. 2089–2097.

PMC009xxxxxx/PMC9405708.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7503056 4435 J Am Chem Soc J Am Chem Soc Journal of the American Chemical Society 0002-7863 1520-5126 35820104 9405708 10.1021/jacs.2c05266 NIHMS1830575 Article γ C-H Functionalization of Amines via Triple H-Atom Transfer of a Vinyl Sulfonyl Radical Chaperone Herbort James H. † Bednar Taylor N. † Chen Andrew D. RajanBabu T.V. http://orcid.org/0000-0001-8515-3740 Nagib David A. http://orcid.org/0000-0002-2275-6381 Department of Chemistry and Biochemistry, The Ohio State University, Columbus, Ohio 43210, United States † Author Contributions These authors contributed equally. Corresponding Author: nagib.1@osu.edu, rajanbabu.1@osu.edu 22 8 2022 27 7 2022 12 7 2022 27 7 2023 144 29 1336613373 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. A selective, remote desaturation has been developed to rapidly access homoallyl amines from their aliphatic precursors. The strategy employs a triple H-atom transfer (HAT) cascade, entailing (i) cobalt-catalyzed metal-HAT (MHAT), (ii) carbon-to-carbon 1,6-HAT, and (iii) Co-H regeneration via MHAT. A new class of sulfonyl radical chaperone – to rapidly access and direct remote, radical reactivity – enables remote desaturation of diverse amines, amino acids, and peptides with excellent site-, chemo-, and regio- selectivity. The key, enabling C-to-C HAT step in this cascade was computationally designed to satisfy thermodynamic (bond strength) and kinetic (polarity) requirements, and it has been probed via regioselectivity, isomerization, and competition experiments. We have also interrupted this radical transfer dehydrogenation to achieve a family of γ-selective C-Cl, C-CN, and C-N bond formations. Graphical Abstract pmcIntroduction Alkenes are highly versatile motifs in synthetic chemistry.1–6 Ideally, double C-H oxidation would install this valuable linchpin group with chemo- and regio- selectivity at positions distal to common functional groups.7 However, in addition to such selectivity targets, oxidative desaturation is difficult to achieve due to a tendency for over-oxidation of the alkene product. To overcome these challenges, we proposed a mild, redox-neutral desaturation of amines – enabled by a new radical chaperone that facilitates selective alkene transfer. We were inspired by metal H-atom transfer (MHAT) as a robust method to generate C-centered radicals via alkenes, silanes, and first-row metals (Figure 1a).8,9 Such MHAT-mediated strategies enable valuable chemo-, regio-, and stereo- selective alkene hydrofunctionalizations10–16 and isomerizations17 within complex molecules.18–21 Yet, this mechanism has not been coupled with intramolecular HAT to effect remote C-H functionalization. Toward this goal, we sought to develop a new class of radical chaperone to complement our β C-H aminations of alcohols via transient imidate radicals.22–24 In our new design, we envisioned temporary conversion of an amine to alkene A could enable remote, transfer dehydrogenation via a triple HAT cascade (Figure 1b). In this scheme, we sought to leverage our dual expertise in intramolecular HAT25 and Co catalysis26 to promote distal, site-selective desaturation. Our general mechanistic design entails sequential (1) MHAT of A to B, (2) C-to-C HAT of B to C, and (3) MHAT of C to D to generate, translocate, and terminate the respective radical intermediates. Moreover, we recognized an opportunity to interrupt the terminal MHAT and intercept the remote radical C to yield various γ C-H functionalizations E. Given the high value, yet ongoing limitations, of double C-H desaturation technologies,27–33 we set out to probe the utility of a new chaperone in addressing this synthetic challenge. Radical desaturations typically require adjacent functional groups (e.g. arene, heteroatom, carbonyl)34–38 as well as stoichiometric oxidants.39,40 Among the rare examples of catalytic or regioselective desaturation, it is instructive to note how oxidized precursors are used to access radicals centered on O (via O-O),41 N (via N-F, N-O),42,43 or C (via Ar-N3R, Ar-I, SiCH2I)44–46 (Figure 2a). Collectively, these pioneering strategies overcome the thermodynamic challenge of C-H abstraction by generating a strong O-H, N-H, primary or aryl C-H bond. However, weak benzylic or tertiary C-H bonds are still often required for efficient reactivity, likely due to kinetic effects arising from the requisite catalysts (e.g. Pd, Cu) or cationic intermediates. Design To address these ongoing challenges, we proposed three principal design elements: (i) an alkene-based radical chaperone must be easily installed and removed, (ii) redox-neutral, intramolecular transfer dehydrogenation can prevent over-oxidation, and (iii) a triple HAT cascade could permit robust desaturation of vicinal, secondary C-H bonds. We anticipated this triple HAT cascade to be critical, yet also most challenging. The requisite radical translocation by C-to-C HAT is difficult,47 since it lacks a thermodynamic driving force given the similarity in bond strength of the broken and formed C-H bonds (2° C-H: 98 kcal/mol) (Figure 2b). Moreover, a kinetic barrier exists due to the similar polarities of the two alkyl radicals involved in HAT (2° radical polarity: 0.7 eV).48,49 Electrophilic radicals are typically employed in HAT (c.f. O > 2.1 eV; N > 1.4 eV)25 to avoid competing side-reactions (e.g. reduction, dimerization, polymerization). Results and Discussion To facilitate a rapid evaluation of radical chaperone candidates, we computationally evaluated both of these kinetic (radical electrophilicity, ω)49 and thermodynamic (bond dissociation energy, BDE)50 parameters (Figure 2c; B3LYP-D3/6-311++G(d,p)). Among five representative examples (see more in SI), we observed the radical polarity (ω) is mismatched for effective secondary C-H HAT kinetics by several linkers, including methylene I (CH2), silyl II (SiMe2), and difluoro III (CF2). All three are significantly less electrophilic than iminyl radicals (ω < 1.4), which are among the least electrophilic radicals to efficiently promote HAT.51–53 Conversely, only α- carbonyl IV (CO) or sulfonyl V (SO2) radicals are computed to be electrophilic enough (ω ≥ 1.4) to effect rapid C-to-C HAT of 2°C-H bonds. To further enrich this kinetic analysis, thermodynamic favorability (ΔGHAT) was approximated by computing the difference in BDE of the abstracted C-H versus the C-H formed on the chaperone via HAT. Here, α- methylene I (CH2), silyl II (SiMe2), and carbonyl IV (CO) were found to be significantly endergonic – forming insufficiently strong C-H bonds (91-95 kcal/mol). Only difluoro III (CF2) and sulfonyl V (SO2) chaperones afford a strong enough C-H bond (97 kcal/mol) to render 2° C-H HAT nearly thermoneutral. Notably, physical experiments validated these computational predictions, wherein the only successful candidate to afford any remote desaturation was vinyl sulfonamide V, which is also the only linker calculated to have an HAT that is both polarity-matched and thermodynamically allowed at room temperature. In contrast, no desaturation was observed among the other candidates tested (I, II, IV; III is not easily synthesized). Notably, while an α-sulfone radical has been employed in 1,5-HAT,54 this is the first example of intramolecular HAT mediated by an α-sulfonamide radical. We suspect the divergent γ selectivity observed herein occurs via 1,6-HAT – due to elongated C-S-N bonds that strain a 1,5-HAT transition state, as Roizen observed for N-S-N bonds within N-centered sulfamate radicals.55–58 The development of this γ C-H desaturation is shown in Figure 3. Crucially, the radical initiator is easily installed onto primary or secondary amines in a single step by combination with commercially available 2-chloroethanesulfonyl chloride (1 equiv) and Et3N (3.5 equiv). Following transfer hydrogenation, the resulting EtSO2- (Es) is easily removed with Red-Al (NaAlH2•C6H14O4) or by the phosphonium-based method developed at Merck.59 These deprotection strategies are amenable to either secondary or tertiary sulfonamides. Thus, this facile addition and removal renders this vinyl sulfonamide an excellent radical chaperone. Our evaluation of this radical alkene transfer revealed the combination of Co(II) precatalyst Co1, Selectfluor oxidant to access Co(III), PhSiH3 to generate Co-H, and a DCE/tBuOH solvent mixture was optimal for enabling γ desaturation of vinyl sulfonamide 1. Variations in electronic and steric parameters of the salen ligand on Co show that subtle substituent effects may impact the rate of radical cage collapse or MHAT kinetics (entries 1-3).60 Although an oxidized version of precatalyst Co1 – as Co(III)Cl – affords comparable yields in the absence of Selectfluor, the Co(II) precatalyst/oxidant combination affords greater regioselectivity for the γ-δ desaturated amine (12:1 rr vs 4:1). The unique role of Selectfluor in providing high regioselectivity (for MHAT elimination away from the sulfonamide) remains unclear. Use of N-fluoro-collidine•BF4 as oxidant – to form Co(III)F alternatively – does not resurrect the rr, thus ruling out Cl vs F (or BF4) counterion effects (entries 4-5). Other hydrides and solvents, which have been employed in MHAT,8 were examined and found to be less effective (entries 6-7; see SI for full details). It is likely that the advantageous addition of a polar, protic solvent (tBuOH) either stabilizes the HAT transition state61 or generates an alkoxysilane in situ, as Shenvi has shown.62 Dilute conditions yield greater efficiency, likely by suppressing side-reactions such as H2 evolution and dimerization; O2 is also deleterious to this radical reaction (entries 8-9). Lastly, control experiments demonstrate the importance of each component, including the Co1 catalyst, silane, and oxidant (entries 10-12). Synthetic Scope With optimized conditions in hand, we next evaluated the generality of this alkene transfer (Figure 4). In addition to primary amines, we were pleased to see this strategy is also amenable to amides and carbamates (2-4). The higher regioselectivity observed in these cases (up to >20:1 rr (γ-δ vs β-γ) is likely due to increased steric repulsion in the transition state of the terminal MHAT step that disfavors abstraction of the β C-H (vs δ C-H). Other alicycles (cyclo-pentane, -heptane) and heterocycles (tetrahydropyran, piperidine) also cleanly afford γ desaturated analogs (5-8). For piperidine 8, steric control of the MHAT by the opposing NBoc (vs the Es) nearly reverses regioselectivity. The functional group tolerance of free alcohols, pyridines, aryl halides, and α or β substituents also showcase the chemoselectivity of this method (9-11) – including in forming contra-thermodynamic, terminal olefins.63,64 Notably, efficient desaturation of vicinal, secondary C-H bonds (12-13) is enabled by this C-to-C HAT – illustrating the broad synthetic utility of this triple HAT strategy. Even though there is less driving force for this abstraction of a 2° C-H (vs 3° C-H), subsequent MHAT formation of a disubstituted olefin likely affords the requisite thermodynamic driving force for this net alkene transfer from a vinyl sulfonamide. To further probe the generality and chemoselectivity of this radical chaperone strategy, a range of amines found within drug and natural products were desaturated (14-18). In these examples, multiple substitution patterns (α, β, γ, δ), rings (polycyclics, arenes), and divergent core functionality (anilines, endo-amide 18) are tolerated. Formation of non-conjugated systems (17 vs styrene; 1,4-diene 18 vs enone or 1,3-diene) are illustrative of the orthogonal selectivity of this desaturation method.43,45 As a solution to the central challenge of accessing medicinally relevant, unnatural amino acids,65 we were pleased to find several Cy-alanine (Cha) esters are selectively and efficiently dehydrogenated (19-21; >20:1 rr in all cases; see SI for x-ray structure of 19). Similarly, leucine (Leu) 22 and various di-peptides of Leu (23-25) are selectively desaturated at the γ C-H of the Leu residue, even in the presence of weak C-H bonds (Leu-Val, Leu-Phe) or unprotected indoles (Leu-Trp). Lastly, an additive robustness66 investigation was performed (on the reaction of 1 to 2) to explore tolerance of reactive functionalities and facilitate a comparison with Pd-catalyzed methods.45,46 Representative examples (see SI for full additive screen) show the desaturation uniquely permits the presence of aryl iodides, carboxylic acids, epoxides, and ketones – with minimal effect on yield or selectivity. The only major limitations observed are those groups that may interact with Co-H, such as aldehydes or alkynes. Although robust regioselectivity was observed in nearly all cases (accessing homoallyl amines via γ-δ desaturation), certain outlier substrates afford allyl amines via β-γ desaturation (Figure 5). These cases are either biased toward forming a more substituted β-3° olefin (26), or entail small, undifferentiated groups (e.g. isopentyl primary amine 27 or N-butyl secondary amine 28) that would afford terminal alkenes by MHAT of primary C-H bonds (vs longer groups that yield internal olefins 12). Smaller amino acids with β-3° centers similarly afford β-γ desaturated analogs, including of valine (Val) and Cy-glycine (Chg) (29-30). Finally, small amides afford α-β desaturation (31). Mechanistic Investigations To understand the origin of regioselectivity in this triple MHAT (and the switch in olefin position of the outliers), we designed several mechanistic experiments and HAT probes (Figure 6). First, isopentylamide 27, which only differs in its lack of α or β substitution compared with amines 9-11, 15-16 or Leu analogs 22-25, is the only of these 10 examples with β (vs γ) selectivity. We proposed this may result from Co-catalyzed isomerization of a transiently generated, typical γ alkene.17,67 To test this hypothesis, homoallyl amine 32 was prepared and subjected to reaction conditions (Fig 6a). In this case, 4:1 isomerization to trisubstituted olefin 27 recapitulates the 3:1 rr observed in the catalytic reaction (c.f. Fig 5). To explain why this post-reaction isomerization does not occur in most cases, we propose that α or β substituents on amines create steric bias for the MHAT elimination to afford terminal olefin 32 rather than allyl amines 27 (Fig 6b). Thus, while 1,6-HAT selectively affords a γ-radical in all cases, the less sterically hindered δ MHAT (vs β MHAT) affords kinetic selectivity for contra-thermodynamic desaturation. As support for this model (Fig 6c), increasingly larger α-substituents reverse β MHAT selectivity (4:1 β) to afford increasingly greater δ MHAT selectivity (6:1 δ) with α-Me and α-ester (>20:1 δ) amines. Next, we designed a series of intramolecular competition experiments to interrogate the regioselectivity of the key Csp3-Csp3 HAT (Fig 6d). To discern the effect of bond strength on HAT site-selectivity, we compared abstraction of 3° and 2° C-H bonds (33). As expected, tertiary C-H abstraction is significantly favored over secondary (13:1). With this baseline thermodynamic preference for 3° HAT in hand, we probed the kinetic preference of 1,5-, 1,6-, and 1,7-HAT by varying the 3° position. Consequently, we observed a relative reactivity trend of 1,6 > 1,5 > 1,7 HAT. The significant selectivity of 1,6- over 1,5-HAT (34, 7:1) is consistent with our hypothesis that elongated C-S-N bonds preclude 1,5-HAT by straining its transition state.55,56 The even greater preference for 1,6- over 1,7-HAT (35, >20:1) reflects an entropic barrier to the larger macrocyclic transition state.25 Taken together, these data suggest 1,6-HAT robustly abstracts the γ C-H in all cases and MHAT (dictated by sterics of α or β substituents) affords regioselectivity of the net desaturation – including by forming contra-thermodynamic terminal olefins in many cases. DFT calculations To gain further insight on the energetic preference for 1,6-HAT regioselectivity, we modeled the reaction coordinate for six competing HAT pathways within a single competition probe and computed the transition state (TS) energies of each feasible HAT event (Figure 7; ωB97X-D/6-311++G(d,p)/IEFPCM(DCE)). This permitted simultaneous examination of multiple variables at once – quantifying both site-selectivity of radical generation (3° > 2° > 1°) as well as HAT regioselectivity (1,6 > 1,5 > 1,7). As expected, we observed a kinetic preference for HAT of a 3° γ C-H bond (left box) over 2° γ C-H bonds (right box) by 2.3 kcal/mol. Additionally, 1,6-HAT is kinetically favored over both 1,5-HAT (by 4.0 kcal/mol) and 1,7-HAT (by 5.8 kcal/mol) for 3° C-H bonds (left). This preference can be explained by enthalpic (ring strain: 1,6 < 1,5) and entropic (reorganization energy: 1,6 < 1,7) forces, respectively.25 For the less differentiated, fully 2° C-H bond system (right), 1,6-HAT is similarly kinetically favored over both 1,5- and 1,7- HAT (by 1.3 kcal/mol). In all cases, the experimental reactivity trend of 1,6 > 1,5 > 1,7 HAT is also reflected in the relative stabilization of the respected distal radical products. This thermodynamic preference is pronounced for the γ 3° position (left) over δ 1° and β 2° (by 5.6 and 3.9 kcal/mol, respectively), and it is also significant (1.6-2.0 kcal/mol) for the fully 2° C-H bond case (right), likely due to radical destabilizing effects of the sulfonamide and terminal methyl groups. Further Classes of Reactivity Lastly, to showcase the broader utility of this new sulfonamide radical chaperone, we sought to complement this desaturation by also developing a series of γ C-H functionalizations (Figure 8). We recently interrupted the N-radical-mediated Hofmann-Löffler-Freytag reaction with Cu catalysts to realize δ C-H functionalization of amines.68,69 With the hopes of accessing complementary γ selectivity, we questioned if this Co-catalyzed C-to-C HAT strategy could also be interrupted. The central challenge to achieving remote functionalization in this manifold is differentiating between the two C-centered radicals. We reasoned that electron-deficient radical traps may preferentially react with the transposed alkyl radical C due to a polarity matched coupling of this nucleophilic intermediate (versus the electrophilic α- sulfonamide radical B). Carreira has shown Co-catalyzed MHAT may be coupled with electrophilic traps to afford alkene hydrofunctionalizations.11,70,71 To our delight, these new radical chaperones also enable interruption of this cascade by intramolecular HAT to realize γ C-Cl, C-CN, and C-N bond formation – by the addition of stoichiometric PhSiH3 and various radical traps (e.g. TsCl, TsCN, DBAD). To probe the generality of these new γ C-H functionalizations, we chose three representative substrates, including amines with either a 3° C-H (36-38) or 2° C-H (39-41) at the γ position, as well as the amino acid, Leu (42-44). Notably, C-H chlorination (with TsCl) and C-H cyanation (with TsCN) are as efficient as the desaturation reaction. And although hydrazination with the bulky reagent, DBAD, was less efficient, its viability nonetheless shows that other radical trap classes are also amenable to this γ C-H functionalization. Importantly, γ C-H selectivity was exclusively observed in all cases (>20:1 rr), further demonstrating the robust 1,6-HAT selectivity of the α-sulfonamide radical. Conclusions In summary, a new vinyl sulfonamide radical chaperone has been computationally designed and experimentally developed to realize chemoselective, remote desaturation of amines, amino acids, and peptides. This strategy utilizes a triple HAT cascade, consisting of 1,6-HAT sandwiched between two metal-mediated HAT events, to facilitate remote transfer dehydrogenation. The scope of this transformation displays broad functional group tolerance with a range of amines including natural products and drug-like motifs. The selectivity of intramolecular Csp3-Csp3 HAT from the α-sulfonamide radical was interrogated by competition experiments that reveal a strong preference for 1,6-HAT. Lastly, this triple HAT cascade was interrupted by addition of radical traps to realize remote C-Cl, C-CN, and C-N bond formation via polarity-matched couplings. We anticipate this new (i) radical chaperone and (ii) MHAT-HAT-MHAT cascade will each serve as platforms for inventing further, versatile tools for remote C-H functionalization. Supplementary Material SI ACKNOWLEDGMENT We thank the National Institutes of Health (R35 GM119812 to DAN, R35 GM139545 to TVR) for financial support. Calculations were performed using resources at the Ohio Supercomputer Center. Figure 1. Remote C-H functionalization via triple HAT. Figure 2. Design of a radical chaperone for γ C-H functionalization. (a) State-of-art, (b) Challenges, (c) Evaluation by computational predictions and experimental validation. Figure 3. Development of γ C-H desaturation of amines via vinyl sulfonamide. Figure 4. γ-Transfer dehydrogenation: scope and generality. rr denotes regioisomeric ratio (γ-δ:β-γ). See SI for full details. Figure 5. Allyl amines via β-γ desaturation of smaller amines. Figure 6. Mechanistic experiments probing isomerization, as well as MHAT and HAT regioselectivity. Figure 7. HAT regioselectivity: DFT calculations support experimental observations. Figure 8. Several classes of γ C-H functionalizations enabled by an interrupted radical cascade. The Supporting Information is available free of charge on the ACS Publications website at DOI: Experimental procedures and characterization for all new compounds (PDF) 1H and 13C NMR spectral data (PDF) Accession Codes CCDC 2180958 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_request@ccdc.cam.ac. uk, or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 223 336033. The authors declare no competing financial interest. REFERENCES (1) Qi X ; Diao T Nickel-Catalyzed Dicarbofunctionalization of Alkenes. ACS Catal. 2020, 10 , 8542–8556.33732540 (2) Badir SO ; Molander GA Developments in Photoredox/Nickel Dual-Catalyzed 1,2-Difunctionalizations. Chem 2020, 6 , 1327–1339.32542207 (3) Dhungana RK ; Sapkota RR ; Niroula D ; Giri R Walking Metals: Catalytic Difunctionalization of Alkenes at Nonclassical Sites. Chem. Sci 2020, 11 , 9757–9774.34094239 (4) Lan XW ; Wang NX ; Xing Y Recent Advances in Radical Difunctionalization of Simple Alkenes. European J. Org. Chem 2017, 2017 , 5821–5851. (5) McDonald RI ; Liu G ; Stahl SS Palladium(II)-Catalyzed Alkene Functionalization via Nucleopalladation: Stereochemical Pathways and Enantioselective Catalytic Applications. Chem. Rev 2011, 111 , 2981–3019.21428440 (6) Cheung KPS ; Sarkar S ; Gevorgyan V Visible Light-Induced Transition Metal Catalysis. Chem. Rev 2022, 122 , 1543–1625.34623151 (7) Peng B ; Maulide N The Redox-Neutral Approach to C-H Functionalization. Chem. Eur. J 2013, 19 , 13274–13287.24027042 (8) Crossley SWM ; Obradors C ; Martinez RM ; Shenvi RA Mn-, Fe-, and Co-Catalyzed Radical Hydrofunctionalizations of Olefins. Chem. Rev 2016, 116 , 8912–9000.27461578 (9) Green SA ; Crossley SWM ; Matos JLM ; Vásquez-Céspedes S ; Shevick SL ; Shenvi RA The High Chemofidelity of Metal-Catalyzed Hydrogen Atom Transfer. Acc. Chem. Res 2018, 51 , 2628–2640.30406655 (10) Mukaiyama T ; Isayama S ; Inoki S ; Kato K ; Yamada T ; Takai T Oxidation-Reduction Hydration of Olefins with Molecular Oxygen and 2-Propanol Catalyzed by Bis(Acetylacetonato)Cobalt(II). Chem. Lett 1989, 18 , 449–452. (11) Waser J ; Carreira EM Convenient Synthesis of Alkylhydrazides by the Cobalt-Catalyzed Hydrohydrazination Reaction of Olefins and Azodicarboxylates. J. Am. Chem. Soc 2004, 126 , 5676–5677.15125654 (12) Waser J ; Nambu H ; Carreira EM Cobalt-Catalyzed Hydroazidation of Olefins: Convenient Access to Alkyl Azides. J. Am. Chem. Soc 2005, 127 , 8294–8295.15941257 (13) Barker TJ ; Boger DL Fe(III)/NaBH4-Mediated Free Radical Hydrofluorination of Unactivated Alkenes. J. Am. Chem. Soc 2012, 134 , 13588–13591.22860624 (14) Shigehisa H ; Aoki T ; Yamaguchi S ; Shimizu N ; Hiroya K Hydroalkoxylation of Unactivated Olefins with Carbon Radicals and Carbocation Species as Key Intermediates. J. Am. Chem. Soc 2013, 135 , 10306–10309.23819774 (15) Lo JC ; Yabe Y ; Baran PS A Practical and Catalytic Reductive Olefin Coupling. J. Am. Chem. Soc 2014, 136 , 1304–1307.24428607 (16) Ma X ; Herzon SB Intermolecular Hydropyridylation of Unactivated Alkenes. J. Am. Chem. Soc 2016, 138 , 8718–8721.27384921 (17) Crossley SWM ; Barabé F ; Shenvi RA Simple, Chemoselective, Catalytic Olefin Isomerization. J. Am. Chem. Soc 2014, 136 , 16788–16791.25398144 (18) Ishikawa H ; Colby DA ; Seto S ; Va P ; Tam A ; Kakei H ; Rayl TJ ; Hwang I ; Boger DL Total Synthesis of Vinblastine, Vincristine, Related Natural Products, and Key Structural Analogues. J. Am. Chem. Soc 2009, 131 , 4904–4916.19292450 (19) Hu X ; Maimone TJ Four-Step Synthesis of the Antimalarial Cardamom Peroxide via an Oxygen Stitching Strategy. J. Am. Chem. Soc 2014, 136 , 5287–5290.24673099 (20) Zweig JE ; Kim DE ; Newhouse TR Methods Utilizing First-Row Transition Metals in Natural Product Total Synthesis. Chem. Rev 2017, 117 , 11680–11752.28525261 (21) Farney EP ; Feng SS ; Schäfers F ; Reisman SE Total Synthesis of (+)-Pleuromutilin. J. Am. Chem. Soc 2018, 140 , 1267–1270.29323492 (22) Wappes EA ; Nakafuku KM ; Nagib DA Directed β C–H Amination of Alcohols via Radical Relay Chaperones. J. Am. Chem. Soc 2017, 139 , 10204–10207.28741940 (23) Stateman LM ; Wappes EA ; Nakafuku KM ; Edwards KM ; Nagib DA Catalytic β C-H Amination: Via an Imidate Radical Relay. Chem. Sci 2019, 10 , 2693–2699.30996986 (24) Nakafuku KM ; Zhang Z ; Wappes EA ; Stateman LM ; Chen AD ; Nagib DA Enantioselective Radical C–H Amination for the Synthesis of β-Amino Alcohols. Nat. Chem 2020, 12 , 697–704.32572164 (25) Stateman LM ; Nakafuku KM ; Nagib DA Remote C-H Functionalization via Selective Hydrogen Atom Transfer. Synthesis 2018, 50 , 1569–1586.29755145 (26) Biswas S ; Parsutkar MM ; Jing SM ; Pagar VV ; Herbort JH ; Rajanbabu TV A New Paradigm in Enantioselective Cobalt Catalysis: Cationic Cobalt(I) Catalysts for Heterodimerization, Cycloaddition, and Hydrofunctionalization Reactions of Olefins. Acc. Chem. Res 2021, 54 , 4545–4564.34847327 (27) Goldman AS ; Roy AH ; Huang Z ; Ahuja R ; Schinski W ; Brookhart M Catalytic Alkane Metathesis by Tandem Alkane Dehydrogenation-Olefin Metathesis. Science 2006, 312 , 257–261.16614220 (28) Dobereiner GE ; Crabtree RH Dehydrogenation as a Substrate-Activating Strategy in Homogeneous Transition-Metal Catalysis. Chem. Rev 2010, 110 , 681–703.19938813 (29) Gunanathan C ; Milstein D Applications of Acceptorless Dehydrogenation and Related Transformations in Chemical Synthesis. Science 2013, 341 , 1229712.23869021 (30) West JG ; Huang D ; Sorensen EJ Acceptorless Dehydrogenation of Small Molecules through Cooperative Base Metal Catalysis. Nat. Commun 2015, 6 , 10093.26656087 (31) Dighe, S. U ; Juliá F ; Luridiana A ; Douglas JJ ; Leonori D . A Photochemical Dehydrogenative Strategy for Aniline Synthesis. Nature 2020, 584 , 75–81.32760044 (32) Wang Z ; Hu L ; Chekshin N ; Zhuang Z ; Qian S ; Qiao JX ; Yu J-Q Ligand-Controlled Divergent Dehydrogenative Reactions of Carboxylic Acids via C–H Activation. Science 2021, 374 , 1281–1285.34762490 (33) Spieß P ; Berger M ; Kaiser D ; Maulide N Direct Synthesis of Enamides via Electrophilic Activation of Amides. J. Am. Chem. Soc 2021, 143 , 10524–10529.34232035 (34) Parasram M ; Chuentragool P ; Sarkar D ; Gevorgyan V Photoinduced Formation of Hybrid Aryl Pd-Radical Species Capable of 1,5-HAT: Selective Catalytic Oxidation of Silyl Ethers into Silyl Enol Ethers. J. Am. Chem. Soc 2016, 138 , 6340–6343.27149524 (35) Zhou MJ ; Zhang L ; Liu G ; Xu C ; Huang Z Site-Selective Acceptorless Dehydrogenation of Aliphatics Enabled by Organophotoredox/Cobalt Dual Catalysis. J. Am. Chem. Soc 2021, 143 , 16470–16485.34592106 (36) Sun X ; Chen J ; Ritter T Catalytic Dehydrogenative Decarboxyolefination of Carboxylic Acids. Nat. Chem 2018, 10 , 1229–1233.30297751 (37) Cartwright KC ; Tunge JA Decarboxylative Elimination of N-Acyl Amino Acids via Photoredox/Cobalt Dual Catalysis. ACS Catal. 2018, 8 , 11801–11806. (38) Nguyen VT ; Nguyen VD ; Haug GC ; Dang HT ; Jin S ; Li Z ; Flores-Hansen C ; Benavides BS ; Arman HD ; Larionov OV Alkene Synthesis by Photocatalytic Chemoenzymatically Compatible Dehydrodecarboxylation of Carboxylic Acids and Biomass. ACS Catal. 2019, 9 , 9485–9498.35223139 (39) Breslow R ; Baldwin S ; Flechtner T ; Kalicky P ; Liu S ; Washburn W Remote Oxidation of Steroids by Photolysis of Attached Benzophenone Groups. J. Am. Chem. Soc 1973, 95 , 3251–3262.4708826 (40) Bigi MA ; Reed SA ; White MC Diverting Non-Haem Iron Catalysed Aliphatic C-H Hydroxylations towards Desaturations. Nat. Chem 2011, 3 , 216–222.21336327 (41) Cekovic Z ; Green MM Formation of Remote Double Bonds by Ferrous Sulfate-Cupric Acetate Promoted Decomposition of Alkyl Hydroperoxides. J. Am. Chem. Soc 1974, 96 , 3000–3002. (42) Stateman LM ; Dare RM ; Paneque AN ; Nagib DA Aza-Heterocycles via Copper-Catalyzed, Remote C–H Desaturation of Amines. Chem 2022, 8 , 210–224.35391915 (43) Jin W ; Yu S Photoinduced and Palladium-Catalyzed Remote Desaturation of Amide Derivatives. Org. Lett 2021, 23 , 6931–6935.34415772 (44) Voica AF ; Mendoza A ; Gutekunst WR ; Fraga JO ; Baran PS Guided Desaturation of Unactivated Aliphatics. Nat. Chem 2012, 4 , 629–635.22824894 (45) Parasram M ; Chuentragool P ; Wang Y ; Shi Y ; Gevorgyan V General, Auxiliary-Enabled Photoinduced Pd-Catalyzed Remote Desaturation of Aliphatic Alcohols. J. Am. Chem. Soc 2017, 139 , 14857–14860.28992686 (46) Chuentragool P ; Parasram M ; Shi Y ; Gevorgyan V General, Mild, and Selective Method for Desaturation of Aliphatic Amines. J. Am. Chem. Soc 2018, 140 , 2465–2468.29400959 (47) Sarkar S ; Cheung KPS ; Gevorgyan V C-H Functionalization Reactions Enabled by Hydrogen Atom Transfer to Carbon-Centered Radicals. Chem. Sci 2020, 11 , 12974–12993.34123240 (48) Parsaee F ; Senarathna MC ; Kannangara PB ; Alexander SN ; Arche PDE ; Welin ER Radical Philicity and Its Role in Selective Organic Transformations. Nat. Rev. Chem 2021, 5 , 486–499. (49) De Vleeschouwer F ; Van Speybroeck V ; Waroquier M ; Geerlings P ; De Proft F Electrophilicity and Nucleophilicity Index for Radicals. Org. Lett 2007, 9 , 2721–2724.17559221 (50) Mayer JM Understanding Hydrogen Atom Transfer: From Bond Strengths to Marcus Theory. Acc. Chem. Res 2011, 44 , 36–46.20977224 (51) Dauncey EM ; Morcillo SP ; Douglas JJ ; Sheikh NS ; Leonori D Photoinduced Remote Functionalisations by Iminyl Radical Promoted C–C and C–H Bond Cleavage Cascades. Angew. Chem. Int. Ed 2018, 57 , 744–748. (52) Morcillo SP ; Dauncey EM ; Kim JH ; Douglas JJ ; Sheikh NS ; Leonori D Photoinduced Remote Functionalization of Amides and Amines Using Electrophilic Nitrogen Radicals. Angew. Chem. Int. Ed 2018, 57 , 12945–12949. (53) Šakić D ; Zipse H Radical Stability as a Guideline in C-H Amination Reactions. Adv. Synth. Catal 2016, 358 , 3983–3991. (54) Masnyk M Radical Rearrangements of α-Iodoalkyl Phenyl Sulfones Involving 1,5 Hydrogen Atom Transfer Reactions. Tetrahedron Lett. 1997, 38 , 879–882. (55) Short MA ; Blackburn JM ; Roizen JL Modifying Positional Selectivity in C–H Functionalization Reactions with Nitrogen-Centered Radicals: Generalizable Approaches to 1,6-Hydrogen-Atom Transfer Processes. Synlett 2020, 31 , 102–116.33986583 (56) Nechab M ; Mondal S ; Bertrand MP 1,n-Hydrogen-Atom Transfer (HAT) Reactions in Which N≠5: An Updated Inventory. Chem. Eur. J 2014, 20 , 16034–16059.25345694 (57) Short MA ; Blackburn JM ; Roizen JL Sulfamate Esters Guide Selective Radical-Mediated Chlorination of Aliphatic C–H Bonds. Angew. Chem. Int. Ed 2018, 57 , 296–299. (58) Short MA ; Shehata MF ; Sanders MA ; Roizen JL Sulfamides Direct Radical-Mediated Chlorination of Aliphatic C-H Bonds. Chem. Sci 2020, 11 , 217–223. (59) Fier PS ; Kim S ; Maloney KM Reductive Cleavage of Secondary Sulfonamides: Converting Terminal Functional Groups into Versatile Synthetic Handles. J. Am. Chem. Soc 2019, 141 , 18416–18420.31693358 (60) Koenig TW ; Hay BP ; Finke RG Cage Effects in Organotransition Metal Chemistry: Their Importance in the Kinetic Estimation of Bond Dissociation Energies in Solution. Polyhedron 1988, 7 , 1499–1516. (61) Litwinienko G ; Beckwith ALJ ; Ingold KU The Frequently Overlooked Importance of Solvent in Free Radical Syntheses. Chem. Soc. Rev 2011, 40 , 2157–2163.21344074 (62) Obradors C ; Martinez RM ; Shenvi RA Ph(i-PrO)SiH2: An Exceptional Reductant for Metal-Catalyzed Hydrogen Atom Transfers. J. Am. Chem. Soc 2016, 138 , 4962–4971.26984323 (63) Zhao K ; Knowles RR Contra-Thermodynamic Positional Isomerization of Olefins. J. Am. Chem. Soc 2022, 144 , 137–144.34968043 (64) Occhialini G ; Palani V ; Wendlandt AE Catalytic, Contra-Thermodynamic Positional Alkene Isomerization. J. Am. Chem. Soc 2022, 144 , 145–152.34968044 (65) Blaskovich MAT Unusual Amino Acids in Medicinal Chemistry. J. Med. Chem 2016, 59 , 10807–10836.27589349 (66) Collins KD ; Glorius F A Robustness Screen for the Rapid Assessment of Chemical Reactions. Nat. Chem 2013, 5 , 597–601.23787750 (67) Matos JLM ; Green SA ; Chun Y ; Dang VQ ; Dushin RG ; Richardson P ; Chen JS ; Piotrowski DW ; Paegel BM ; Shenvi RA Cycloisomerization of Olefins in Water. Angew. Chem. Int. Ed 2020, 59 , 12998–13003. (68) Zhang Z ; Stateman LM ; Nagib DA δ C-H (Hetero)Arylation via Cu-Catalyzed Radical Relay. Chem. Sci 2019, 10 , 1207–1211.30774920 (69) Zhang Z ; Zhang X ; Nagib DA Chiral Piperidines from Acyclic Amines via Enantioselective, Radical-Mediated δ C–H Cyanation. Chem 2019, 5 , 3127–3134.32296736 (70) Gaspar B ; Carreira EM Mild Cobalt-Catalyzed Hydrocyanation of Olefins with Tosyl Cyanide. Angew. Chem. Int. Ed 2007, 46 , 4519–4522. (71) Gaspar B ; Carreira EM Catalytic Hydrochlorination of Unactivated Olefins with Para-Toluenesulfonyl Chloride. Angew. Chem. Int. Ed 2008, 47 , 5758–5760.

PMC009xxxxxx/PMC9414303.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101614637 41946 Cancer Immunol Res Cancer Immunol Res Cancer immunology research 2326-6066 2326-6074 34785506 9414303 10.1158/2326-6066.CIR-21-0454 NIHMS1854643 Article Identification of Immunogenic MHC Class II Human HER3 Peptides that Mediate Anti-HER3 CD4+ Th1 Responses and Potential Use as a Cancer Vaccine Basu Amrita 1 Albert Gabriella K. 1 Awshah Sabrina 1 Datta Jashodeep 2 Kodumudi Krithika N. 13 Gallen Corey 1 Beyer Amber 1 Smalley Keiran S.M. 45 Rodriguez Paulo C. 6 Duckett Derek R. 7 Forsyth Peter A. 8 Soyano Aixa 9 Koski Gary K. 10 Costa Ricardo Lima Barros 9 Han Heather 9 Soliman Hatem 9 Lee Marie Catherine 9 Kalinski Pawel 11 Czerniecki Brian J. 169 1 Clinical Science Division, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida. 2 Department of Surgery, University of Miami Miller School of Medicine, Sylvester Comprehensive Cancer Center, Miami, Florida. 3 Department of Oncological Sciences, University of South Florida, Tampa, Florida. 4 Department of Cutaneous Oncology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida. 5 Department of Tumor Biology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida. 6 Department of Immunology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida. 7 Department of Drug Discovery, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida. 8 Department of NeuroOncology and the NeuroOncology Program, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida. 9 Department of Breast Oncology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida. 10 Department of Biological Sciences, Kent State University, Kent, Ohio. 11 Department of Immunology, Roswell Park Comprehensive Cancer Center, New York, New York. A. Basu and G.K. Albert contributed equally to this article. Authors’ Contributions A. Basu: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. G.K. Albert: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. S. Awshah: Validation, investigation, methodology, writing–review and editing. J. Datta: Conceptualization, formal analysis, validation, investigation, methodology, writing–review and editing. K.N. Kodumudi: Conceptualization, formal analysis, supervision, validation, investigation, visualization, methodology, project administration, writing–review and editing. C. Gallen: Investigation, methodology, writing–review and editing. A. Beyer: Investigation. K.S.M. Smalley: Writing–review and editing. P.C. Rodriguez: Writing–review and editing. D.R. Duckett: Writing–review and editing. P.A. Forsyth: Writing–review and editing. A. Soyano: Writing–review and editing. G.K. Koski: Writing–review and editing. R.L.B. Costa: Writing–review and editing. H. Han: Writing–review and editing. H. Soliman: Writing–review and editing. M.C. Lee: Writing–review and editing. P. Kalinski: Writing–review and editing. B.J. Czerniecki: Conceptualization, resources, supervision, funding acquisition, validation, writing–original draft, project administration, writing–review and editing. Corresponding Author: Brian J. Czerniecki, Department of Breast Oncology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL 33612. Brian.Czerniecki@moffitt.org 4 12 2022 1 2022 16 11 2021 09 1 2023 10 1 108125 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. The HER3/ERBB3 receptor is an oncogenic receptor tyrosine kinase that forms heterodimers with EGFR family members and is overexpressed in numerous cancers. HER3 overexpression associates with reduced survival and acquired resistance to targeted therapies, making it a potential therapeutic target in multiple cancer types. Here, we report on immunogenic, promiscuous MHC class II–binding HER3 peptides, which can generate HER3-specific CD4+ Th1 antitumor immune responses. Using an overlapping peptide screening methodology, we identified nine MHC class II–binding HER3 epitopes that elicited specific Th1 immune response in both healthy donors and breast cancer patients. Most of these peptides were not identified by current binding algorithms. Homology assessment of amino acid sequence BLAST showed >90% sequence similarity between human and murine HER3/ERBB3 peptide sequences. HER3 peptide–pulsed dendritic cell vaccination resulted in anti-HER3 CD4+ Th1 responses that prevented tumor development, significantly delayed tumor growth in prevention models, and caused regression in multiple therapeutic models of HER3-expressing murine tumors, including mammary carcinoma and melanoma. Tumors were robustly infiltrated with CD4+ T cells, suggesting their key role in tumor rejection. Our data demonstrate that class II HER3 promiscuous peptides are effective at inducing HER3-specific CD4+ Th1 responses and suggest their applicability in immunotherapies for human HER3-overexpressing tumors. pmcIntroduction Oncodrivers are proteins overexpressed in tumor cells that promote proliferation and growth, but counteract cellular senescence, contributing to tumor malignancy. Such oncodriver addiction in tumor cells makes these proteins a promising target for developing new immunotherapies relevant to a large number of patients. Human ERBB3 receptor tyrosine kinase 3 (HER3/ERBB3) is a member of the ERBB family of growth receptors that has several ligands, including heregulin and neuregulin. Although lacking intrinsic kinase activity, HER3 is a critical heterodimerization partner for other members of the family (EGFR and HER2) contributing to the growth, proliferation, and survival of tumor cells (1). HER3 is an established oncodriver in multiple cancers because HER3 overexpression has been detected in breast, melanoma, colorectal, prostate, lung, and ovarian cancers (2). The HER2–HER3 heterodimer is responsible for the most potent ligand-induced tyrosine phosphorylation and downstream signaling via the PI3K–AKT pathway, utilizing the six p85-binding motifs present in HER3 intracellular domain (3, 4). In HER2-overexpressing (HER2pos) breast cancer, HER3 is identified as one of the most prominent inducers of therapy escape, leading to trastuzumab resistance and hyperactivation of PI3K–AKT-mediated signaling (3). Transcriptional/translational upregulation of HER3 is linked to resistance to MEK/RAF inhibitors in melanoma (5), castration-resistant prostate cancer (6), platinum-resistant ovarian cancer (7), and EGFR TKI-resistant non–small cell lung cancer (NSCLC; refs. 8, 9). In triple-negative breast cancer (TNBC), HER3 overexpression has been independently identified as a prognostic marker of poor survival (10, 11), and combined antagonism of EGFR and HER3 enhances responses to PI3K–AKT inhibitors in TNBC preclinical and clinical samples (12). Multiple monoclonal antibodies targeting HER3 are currently being tested at various stages of preclinical and clinical studies; however, therapeutic potential of a HER3-specific immunotherapy has not been comprehensively tested. Developing HER3-specific cellular immunotherapy can be a novel and efficient treatment strategy for multiple cancer types overexpressing HER3 in order to improve patient prognosis and survival. Although current immunotherapies focus on cytotoxic CD8+ T-cell activity, “helper signals” from CD4+ T cells have been deemed essential for the proliferation, recruitment, and effector function of CD8+ T cells in tumors (13, 14). Help from CD4+ T cells is necessary for memory T-cell survival during recall expansion (15), and IL-2 secreted from CD4+ T cells in the tumor microenvironment has been shown to increase CD8+ T-cell proliferation and granzyme B production (16). Absence of CD4+ T-cell help affects survival and clonal expansion of CD8+ T cells, and defective recall responses by CD8+ T cells have been reported in CD4−/− mice (17). In a head and neck cancer model, HER3-specific helper T-cell responses induce cytolytic CD4+ T-cell activity against tumor cells in an HLA-DR–restricted manner (18). These studies highlight the potential for developing CD4+ T cell–based immunotherapy for cancer, and as previously observed by our lab and others, oncodrivers represent an excellent choice for developing such therapeutic strategies (19). A previous study from our lab demonstrates gradual loss of HER3-specific CD4+ Th1 immune responses in the peripheral circulation of TNBC patients compared with healthy donors and other breast cancer subtypes, and this loss of immune response negatively associates with patient outcome (20). In the current study, we investigated whether promiscuous MHC class II HER3 peptides could be identified to develop an immunotherapy strategy for HER3-overexpressing cancers. Herein, we present a peptide screening methodology for identifying promiscuous class II epitopes that are capable of inducing tumor-specific CD4+ T-cell responses. We identified nine immunogenic class II epitopes and investigated the therapeutic efficacy of HER3 peptide–pulsed type I DC (HER3-DC1) treatment in multiple preclinical models of breast cancer and melanoma and in both preventive and therapeutic settings. Materials and Methods HER3 expression in cancer and correlation with survival The Cancer Genome Atlas (TCGA) RNA sequencing (RNA-seq) expression data (FPKM) for HER3 (ERBB3) in 12 types of cancers were downloaded from the Genomic Data Commons Data Portal (GDC; https://portal.gdc.cancer.gov, RRID:SCR_014514). Project IDs included in analysis were as follows: TCGA-BLCA, bladder urothelial carcinoma (n = 408); TCGA-BRCA, breast invasive carcinoma (n = 1092); TCGA-COAD, colon adenocarcinoma (n = 456); TCGA-HNSC, head and neck squamous cell carcinoma (n = 501); TCGA-LUAD, lung adenocarcinoma (n = 515); TCGA-LUSC, lung squamous cell carcinoma (n = 501); TCGA-OV, ovarian serous cystadenocarcinoma (n = 376); TCGA-PAAD, pancreatic adenocarcinoma (n = 177); TCGA-PRAD, prostate adenocarcinoma (n = 496); TCGA-SKCM, skin cutaneous melanoma (n = 468); and TCGA-STAD, stomach adenocarcinoma (n = 380). Only samples with available sequence reads were chosen for further analysis. HER3 mRNA expression in primary and normal tissues was assessed and graphed as a boxplot using a customized R script (Supplementary File). For tumor samples, only data from primary tissues were used. An unpaired t test was used for differential analysis of the data set. Survival plots were generated by the cBioPortal (https://www.cbioportal.org, RRID:SCR_014555) survival tool. Samples were sorted in the descending order, according to the median value of the HER3 gene expression, and the list was divided into two groups in the middle of the list: High expression was defined as the top 50% of the list and low expression as the bottom 50% of the list. Peptide library generation The amino acid sequence for native human HER3 protein (1,342 amino acids, Uniprot Accession number P21860) was divided into two peptide libraries: the extracellular domain (ECD, aa 20–643) and intracellular domain (ICD, aa 665–1,342). Libraries consisted of peptide fragments 15 amino acids in length, with a 10-amino acid overlap between adjacent sequences, generating 123 ECD and 134 ICD peptides (GenScript). BRAF-negative control peptide was purchased from GenScript. Peptides were reconstituted according to solubility instructions (GenScript) to 1 mg/mL concentrations. ICD peptides p37, p54, and p55 were unable to be synthesized. Identified immunogenic human HER3 peptides were used for testing in preclinical mouse models as well, due to significant sequence homology. Human samples Autologous monocyte fractions were obtained from leukapheresis products from the peripheral blood of normal donors and breast cancer patients. Normal donor samples (n = 11) were purchased through the Cell Therapies Core at H. Lee Moffitt Cancer Center and Research Institute, which performed the elutriation to generate monocyte- and lymphocyte-rich fractions. Breast cancer patient elutriation fractions were obtained under the Moffitt Cancer Center (MCC) 19000 protocol and collected during 2017–2021. Inclusion criteria included samples from newly diagnosed, operable breast cancer patients with: (i) incident invasive primary ductal breast cancer prior to surgery with Her2/neu overexpression via IHC; (ii) age over 18; and (iii) consent to participate. Patients with bilateral breast cancer were not excluded. Exclusion criteria included: (i) patients with no plan for definitive surgery; (ii) no detectable residual breast disease after diagnostic biopsy; and (iii) patients unable to receive neoadjuvant herceptin-based chemotherapy. The control group was composed of unaffected, seemingly healthy individuals with inclusion criteria: (i) no prior history of systemic cancer treatment (i.e., chemotherapy or hormone therapy); (ii) age over 18; (iii) consent to participate; and (iv) BIRADS 1–3 breast imaging within the past 12 months. Exclusion criteria included: (i) active diagnosis of autoimmune disease (i.e., rheumatoid arthritis, lupus, Sjogren syndrome) and (ii) systemic steroid use within the past 12 months. Elutriation fractions were washed in PBS (cat. #SH30028LS, Fisher Scientific) and lysed with ACK RBC lysis buffer containing NH4Cl (0.15M, cat. #A649–500; Fisher Scientific), KHCO3 (10 mmol/L, cat. #P184–500, Fisher Scientific) and Na2EDTA (0.5 mmol/L; cat. #E7889–100 mL, Sigma-Aldrich), with pH 7.2, when necessary to remove remaining RBCs. Processed fractions were stored in liquid nitrogen until further use. Breast cancer patient (n = 10) and healthy control (n = 6) samples were used for in vitro validation of identified HER3 class II peptides. All clinical samples were obtained by informed written consent from the subjects, following an Institutional Review Board–approved protocol. All research involving human subjects was performed in accordance with the Declaration of Helsinki and was approved by the IRB at Advarra. Human DC generation Monocytes were differentiated into dendritic cells (DC) through the addition of recombinant human (rh)GM-CSF (50 ng/mL; cat. #215-GM-050, R&D Systems) and rhIL4 (10 ng/mL; cat. #204-IL-050, R&D Systems) and cultured in macrophage serum-free media (cat. #12065074; Thermo Fisher Scientific) for 24 hours at 37°C and 5% CO2. DCs were then pulsed with HER3 ECD or ICD peptides (10 μg/mL), a negative control (BRAF class II p8), or left unpulsed. After 24 hours, DCs were matured and polarized to a type I phenotype (DC1) with the addition of rhIFN-γ (1,000 IU/mL; cat. #285IF-100, R&D Systems), followed by LPS (0.2 μg/mL; cat. #L4391–1MG, Sigma-Aldrich) 6 hours prior to harvest. Mature DC1 cells were harvested after 72 hours of culture. Immature DCs (iDC) remained in rhGM-CSF and rhIL4 for 48 hours prior to harvesting. DCs were preserved in 10% DMSO (cat. #D5879–500 mL, VWR) and 90% human AB serum (cat. #S11550, Atlanta Biologicals) and stored in liquid nitrogen until use. Culture medium Complete media consisted of RPMI-1640 growth media (cat. #MT-10–040-CM, Corning) supplemented with 10% heat-inactivated fetal bovine serum (FBS; cat. #S11550 Atlanta Biologicals), 0.1 mmol/L nonessential amino acids (cat. #25–025-CI, Corning), 1 mmol/L sodium pyruvate (cat. #MT-25–000-CI, Corning), 100 mg/mL streptomycin, 100 U/mL penicillin (cat. #MT-30–002-CI, Fisher Scientific), 50 mg/mL gentamycin (cat. #15750–060, Gibco), 0.5 mg/mL amphotericin B (cat. #400–104, GeminiBio), and 0.05 mmol/L 2-mercaptoethanol (cat. #21985–023, Invitrogen). Peptide screening Mature HER3-DC1 were cocultured in complete media with autologous lymphocytes at a 1:10 ratio in a 12-well tissue culture plate and incubated at 37°C and 5% CO2. The lymphocyte-rich apheresis fractions were processed as described above. After 24 hours, rhIL2 (5 IU/mL) was added to induce the proliferation of CD4+ T cells. Following 8 to 10 days of coculture, T cells were harvested and restimulated in a 96-well plate with iDCs (105 T cells with 104 iDCs for 10:1 ratio) pulsed with the corresponding HER3 class II peptide, a negative class II control (BRAF class II p8, 10 μg/mL), or unpulsed iDCs. Culture supernatants were collected after 24 hours, and IFN-γ production was measured using a Human Quantikine IFN-γ ELISA kit (cat. #SIF50, R&D Systems) after 1:20 or 1:30 sample dilution. The screening process proceeded using a 10-peptide pool, 5-peptide pool, and individual peptide (peptide concentration 10 μg/mL throughout the screening sequence) sequential scheme (Supplementary Fig. S1A) for both ECD and ICD libraries, separately. Libraries were first screened in pools of 10 peptides and run on three independent donor samples. The 10-peptide pools that showed ≥1.5-fold increase in IFN-γ production in two of the three donors were further pursued in 5-peptide pools. Common positive 5-peptide pools were then tested as individual peptides and screened in additional donor samples to identify peptides that could reproducibly induce an increased CD4+ Th1 immune response. Due to the variability in donor sample size and the progressive nature of the peptide screening methodology, the sample number tested for each identified peptide was inconsistent; however, individual peptide candidates were screened across a minimum of five donor samples, with a threshold response rate of ≥1.5-fold to be considered immunogenic. Whole HER3 ECD/ICD protein restimulation Immunogenicity of the identified HER3 class II epitopes was confirmed through sensitization of CD4+ T cells with HER3-DC1 as described above, followed by restimulation with iDCs pulsed with the corresponding HER3 class II peptide, a negative peptide control (BRAF class II p8), native HER3 ECD (cat. #NBP2-52128-0.05 mg, Novus Biologicals) or ICD (cat. #10201-H20B1, SinoBiological) whole protein sequence, or a negative whole protein control (Hemocyanin-Keyhole Limpet Native protein; cat. #SRP6195, Sigma). The same number of T cells, HER3-DC1, and iDCs were used for this experiment and 10 μg/mL of whole protein and/or peptide was used throughout this experiment. Culture supernatants were collected after 24 hours, and IFN-γ production was measured using the Human Quantikine IFN-γ ELISA kit after 1:20 or 1:30 sample dilution. ELISPOT assays To evaluate anti-HER3 CD4+ Th1 immune responses in breast cancer patients, IFN-γ production was compared between healthy controls and patient samples using the human IFN-γ ELISPOT kit (cat. #HIFNgp-1M/10, Cellular Technologies Limited). ELISPOT plates precoated with human IFN-γ capture antibody were incubated with nine HER3 class II peptides (4 μg/well), media only (untreated/negative control), or anti-human CD3 (Orthoclone OKT3, cat. #73337989, Johnson and Johnson, treated/positive control, 15 ng/mL). Cryopreserved PBMCs were plated (2 × 105 cells/well) in CTL-Test medium supplemented with 1% L-glutamine (provided with the kit) and incubated at 37°C, 5% CO2 for 48 hours. Per manufacturer’s protocol, plates were washed, and detection antibody (anti-human IFN-γ Biotin; 100 mg/mL) was added to each well. After incubation at room temperature for 2 hours, 1:1,000-diluted streptavidin-AP was added and incubated for 30 minutes. Plates were washed again before substrate solution was added, and plates were incubated for 15 minutes to allow for color development. Plates were washed with tap water, dried overnight at room temperature, and spot-forming cells (SFC) were counted using an automated reader (Immunospot Cellular Technology Limited). Because inter-replicate variability in ELISPOT was low, an empiric method of determining antigen-specific responses was used as previously shown (21). In brief, positive response to an individual class II HER3 peptide was defined as (i) threshold minimum of 20 SFC/2 × 105 cells in experimental wells after subtracting unstimulated background; and (ii) approximately 2-fold increase of antigen-specific SFCs over background. Three metrics of CD4+ Th1 response were defined for each group: (i) overall anti-HER3 responsivity (proportion of patients responding to ≥1 peptide), (ii) response repertoire (mean number of reactive peptides), and (iii) cumulative response across nine identified class II HER3 peptides (total SFC/1 × 106 cells). HLA typing of donor samples Lymphocyte fractions of healthy donors (n = 11) used in peptide screening HER3 ECD and ICD libraries were sent for HLA-DRB1, HLA-DQB1, and HLA-DPB1 allele typing (American Red Cross). Alleles expressed by each donor were cross-referenced with the CD4+ Th1 response induced by each identified HER3 class II peptide. Each HER3 peptide tested against a donor expressing a particular HLA allele was indicated with a (+) if ≥1 instance of this allele demonstrated a peptide-specific Th1 immune response (IFN-γ production increased ≥1.5-fold compared to negative control) to the corresponding peptide. Because HER3 ECD and ICD libraries were screened using different donor samples, this resulted in two groupings. NetMHCIIpan 4.0 prediction algorithm Epitope prediction was done using NetMHCIIpan 4.0 (22) as implemented by www.iedb.org (23). A series of 15-mer peptides were serially generated (one amino acid difference) from the HER3 protein. Each peptide was then predicted for its binding affinity (nmol/L) with each HLA allele type by identifying a 9-mer binding core. Peptide candidates with affinity <500 nmol/L were labeled WB (weak binding) and <50 nmol/L as SB (strong binding). MixMHCIIpred prediction algorithm Epitope prediction was done using the MixMHCIIpred algorithm (24, 25). A series of random 12–25 amino acid (aa) peptides were generated from the HER3 protein sequence by the algorithm. A percentile rank was provided for each of the nine HER3 peptides out of all 12–25aa random peptides, for each of the 38 HLA allele types. Best score for percentile rank was 0 and the worst score was 100. The best alleles with the highest binding affinity for each peptide were also identified. Cell lines and reagents The human TNBC cell lines MDA-MB-231 (ATCC HTB-26, RRID: CVCL_0062), MDA-MB-468 (ATCC HTB-132, RRID:CVCL_0419), HCC1143 (ATCC CRL-2321, RRID:CVCL_1245), Hs578T (ATCC HTB-126, RRID:CVCL_0332), BT-549 (ATCC HTB-122, RRID: CVCL_1092), SK-BR-3 (ATCC HTB-30, RRID: CVCL_0033) and mouse breast cancer cell lines 4T1 (ATCC CRL-2539, RRID: CVCL_0125) and EO771 (ATCC CRL-3461, RRID:CVCL_GR23) were obtained from the American Type Culture Collection. The TUBO murine breast cancer cell line (a kind gift from Dr. Wei Zen Wei, Wayne State University) was cloned from a spontaneous mammary tumor in BALB/c mice transgenic for the rat HER2/neu gene (BALB-HER2/neuT; ref. 26). The M05 cell line was a kind gift from Dr. Shari Pilon-Thomas (Moffitt Cancer Center), and cells were grown in media containing 0.8 mg/mL G418 (neomycin; cat. #30234CR, Corning; refs. 27, 28). M05 is derived from the B16 murine melanoma cell line, transfected to express ovalbumin, and transgene expression was maintained using G418 as the selection antibiotic in the culture media. Cells were grown at 37°C in a humidified 5% CO2 incubator in complete media. All cell lines used in the study tested negative for mycoplasma using a mycoplasma kit (PlasmoTest, cat. #rep-pt1, InvivoGen). Cell lines at passages 2–5 were used for all experiments. For all in vitro experiments, recombinant human and mouse IFN-γ were purchased from R&D Biosystems (human: cat. #285IF-100; mouse: cat. #485-MI-100). BLASTP analysis Amino acid sequence homology between identified human HER3 peptides and murine Erbb3 sequence (UniProtKB; Q61526) was evaluated by using the BLASTP program (http://blast.ncbi.nlm.nih.gov/Blast.cgi?PROGRAM=blastp&PAGE_TYPE=BlastSearch&LINK_LOC=blasthome RRID:SCR_001010). The percentage of amino acid sequence homology has been reported. Western blots For protein expression, cells were seeded (4 × 105 cells/well for human TNBC cells, 2 × 105 cells/well for mouse TNBC cells) in six-well plates and cell lysates were prepared in RIPA buffer (cat. #20–188, EMD Millipore) containing protease inhibitor cocktail (10 μL/mL; cat. #P8340–1ML, Sigma-Aldrich) and phosphatase inhibitor (100 μL/mL; cat. #A32957, Pierce) by incubating for 20 minutes at 4°C. Cell lysates were centrifuged at 14,000 × g for 20 minutes at 4°C, and total protein in the supernatant was collected and stored at −86°C until further use. Protein concentration was measured using Bradford protein assay (cat. #5000006, Bio-Rad). For Western blotting, 20 μg protein was resolved on 4% to 12% SDS-PAGE (GenScript) and transferred onto PVDF membranes (cat. #IPVH00010, Millipore) using eBlot L1 wet transfer system (GenScript). Membranes were blocked with 5% BSA/TBS-T for 1 hour at room temperature, and then incubated with following primary antibodies overnight at 4°C: HER3 (cat. #12708S), phospho-Stat1 (Tyr701; 58D6; cat. #9167S, Cell Signaling Technology), cleaved caspase-3 (Asp175; cat. #9661S, Cell Signaling Technology), phospho-p44/42 MAPK (Erk1/2; Thr202/Tyr204; cat. N#9101S, Cell Signaling Technology; RRID:AB_2315112), phospho-Akt (Ser473; cat. #9271S, Cell Signaling Technology; RRID:AB_329825; all at 1:1,000 dilution), and β-actin (C4; cat. #sc-47778, Santa Cruz Biotechnology; RRID:AB_2714189; 1:3,000 dilution). Membranes were probed with anti-rabbit IgG HRP-linked secondary antibody (cat. #7074S, Cell Signaling Technology, RRID:AB_2099233) or goat anti-mouse IgG (H + L)–HRP conjugated secondary antibody (cat. #172–1011, Bio-Rad, RRID:AB_11125936; 1:3,000 dilution) for 1 hour at room temperature. Protein expression was detected with ECL Western blotting substrate (cat. #32106, Pierce) using an AmershamTM Imager 600 image acquisition system. Analysis and quantification of Western blot images were performed using ImageJ software (http://rsb.info.nih.gov/ij/, RRID:SCR_003070). Immunofluorescence Cells were seeded (4 × 105 cells/well for human cells (MDA-MB-468, SK-BR-3), 2 × 105 cells/well for mouse cells (4T1, TUBO, and M05) in six-well plates, each well containing three 12-mm round glass coverslips (cat. #12-545-80, Fisher Scientific). After cells reached 70% to 80% confluence, cells were washed twice in PBS, fixed with 4% paraformaldehyde for 15 minutes at room temperature, and washed three times with PBS. Cells were permeabilized with 0.02% Triton X-100 (cat. #T8787—50 mL, Sigma; in PBS) for 10 minutes at room temperature, washed three times with PBS, and blocked with 5% BSA/PBS blocking buffer for 1 hour at room temperature. Cells were incubated with primary anti-HER3 (cat. #12708S, Cell Signaling Technology) in 3% BSA/PBS (1:500 dilution) overnight at 4°C. The next day, cells were washed in PBS three times and incubated in Alexa Fluor 594-conjugated goat-anti-rabbit secondary antibody (1:5,000 dilution; cat. #8889S, Cell Signaling Technology) and FITC-conjugated goat anti-mouse secondary antibody (1:5,000 dilution; cat. #115-095-003 Jackson ImmunoResearch Inc, RRID: AB_2338589) for 1 hour at room temperature. Cells were washed three times in PBS, and the coverslips were mounted onto sterile glass slides using VECTA-SHIELD Antifade Mounting Medium with DAPI (cat. #H-1200, Vector Laboratories). Slides were sealed with clear nail varnish and allowed to cure overnight at 4°C in the dark. Immunofluorescence images were obtained using Zeiss Apotome.2 fluorescence microscope (Carl Zeiss Inc.). Mouse DC generation Bone marrow was harvested from 6- to 8-week-old BALB/c (RRID: IMSR_ORNL:BALB/cRl) and C57BL/B6 (RRID: IMSR_ JAX:000664) mice (purchased from Charles River Laboratories), as described previously (29). Briefly, femurs and tibias were harvested from mice, bone marrow cells were flushed to obtain a single-cell suspension in PBS, and red blood cells were lysed using ACK lysis buffer. Cells (2 × 106 cells/mL) were then cultured in complete media containing recombinant human Flt3L (25 ng/mL; cat. #10778–670, VWR/Pepro-Tech) and recombinant mouse IL6 (30 ng/mL; cat. #406ml025, R&D Systems), and incubated for 6 days at 37°C and 5% CO2. On day 6, cells were harvested, washed with RPMI-1640, and cultured with recombinant mouse GM-CSF (50 ng/mL; cat. #415-ML-050, R&D Systems) and recombinant mouse IL4 (10 ng/mL; cat. #404ml050, R&D Systems) overnight for DC differentiation. The cells were matured with DC1-polarizing cytokines: CPG/ODN1826, a TLR9 agonist (10 ng/mL; cat. #NC9685794, InVivoGen), and LPS, a TLR4 agonist (20 ng/mL). DCs were pulsed with the nine immunogenic class II human HER3 peptides (10 μg/mL) 6 to 8 hours later and harvested the following morning prior to injection into mice. Mouse models Female BALB/c and C57BL/6 mice (6–8 weeks old) were housed at the Animal Research Facility of the H. Lee Moffitt Cancer Center and Research Institute. The study protocol was designed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH. The protocol was reviewed and approved by the Institutional Animal Care and Use Committee at the University of South Florida. Mice were observed daily and were euthanized by CO2 inhalation at the end of the study or if a solitary subcutaneous tumor exceeded endpoint (250 mm2 for 4T1 and 400 mm2 for TUBO and M05), following the American Veterinary Medical Association Guidelines. All efforts were made to minimize suffering. Preventive models BALB/c and C57BL/6 mice (n = 8–10/group) were vaccinated with either unpulsed mature DC1 cells or HER3-DC1 cells subcutaneously (1 × 106 cells/mouse) twice a week for a total of six doses. Two weeks after the last DC injection, naïve and vaccinated BALB/c mice received either 4T1 (50,000 cells/mouse) or TUBO (250,000 cells/mouse) tumor cells, and C57BL/6 mice received M05 (300,000 cells/mouse) tumor cells, administered subcutaneously on the opposite flank to the vaccination site. Tumor size was measured using caliper and recorded every 2 to 3 days. Mouse tumor area was determined by the formula length × width (mm2). Each experiment was performed three times. Therapeutic models For therapeutic models, either 4T1 (50,000 cells/mouse) or TUBO (250,000 cells/mouse) were injected subcutaneously in BALB/c mice, and M05 cells (300,000 cells/mouse) were injected in C57BL/6 mice (n = 8–10 mice/group). After 7 to 10 days (7 days for 4T1 and TUBO, 10 days for M05) when tumors were palpable, mice received intratumoral injection of either unpulsed mature DC1 cells or HER3-DC1 (2 × 106 cells/mouse for 4T1 and 1 × 106 cells/mouse for TUBO and M05, in 50 μL PBS) once (TUBO and M05) or twice (4T1) a week for a total of six doses. Control mice received PBS intratumoral injection. Tumor growth was measured with caliper and recorded every 2 to 3 days, and tumor area was determined by the formula length × width (mm2). Each experiment was performed three times. CD4+ T-cell depletion in therapeutic models BALB/c mice were injected intraperitoneally with monoclonal CD4 depletion antibody [InVivoMAb, anti-mouse CD4 (GK1.5); cat. #BE0003–1, Bio X Cell; RRID:AB_1107636; 300 μg/mouse] starting 3 days before subcutaneous TUBO tumor cell injection (250,000 cells/mouse) and was continued twice a week until endpoint. When tumors were palpable, mice were randomized into two groups. One group of mice continued receiving CD4 depletion antibody only, and the other group received once-weekly intratumoral HER3-DC1 (1 × 106 cells/mouse, in 50 μL PBS) along with CD4 depletion antibody. Another group of TUBO tumor–bearing mice received intratumoral HER3-DC1 only, and the control group of mice received PBS intratumorally. Tumor growth was measured with caliper and recorded every 2 to 3 days, and tumor area was determined by the formula length × width (mm2). Flow cytometry For functional analyses, mice (n = 3/group) were sacrificed 2 weeks after the last HER3-DC1 injection, and tumors, spleen, and lymph nodes (tumor-draining and nondraining) were collected under sterile conditions for in vitro assays. Single-cell suspensions were prepared from the tumor samples by enzymatic digestion with HBSS (Fisher Scientific; cat. #MT-21–022-CM), containing 1 mg/mL collagenase (cat. #C9891 and C-5138), 0.1 mg/mL DNase I (cat. #DN25), and 2.5 U/mL hyaluronidase (cat. #H-6254–1G; all purchased from Millipore Sigma), by constant stirring for 2 hours at room temperature. Tumor digests were strained through a 100-μm cell strainer, and ACK lysis buffer was used to remove red blood cells from tumor digests. Resulting cell suspensions were further strained through a 70-μm and a 30-μm cell strainer to generate single-cell suspension. Splenocytes and lymph nodes were processed in PBS. Samples were macerated using the back of a syringe plunger in a 10-cm culture dish, followed by red blood cell lysis using ACK lysis buffer. After lysis, cell suspensions were strained through a 70-μm and a 30-μm cell strainer sequentially. Cell suspensions were centrifuged at 1,500 rpm for 5 minutes, and cells were resuspended in complete media and counted before plating. To analyze distribution and phenotype of immune cell populations, 1 × 106 cells were incubated with Live/Dead Zombie near IR (cat. #423106, BioLegend) for 30 minutes at room temperature. After washing once with PBS, cells were stained with the following mouse antibodies for surface expression analysis: CD3-APC (cat. #553066, Clone 145–2C11, BD Biosciences), CD45-PE/Cy7 (cat. #103114, Clone 30-F11, BioLegend), or CD45-BUV395 (cat. #564279, Clone 30-F11, BD Biosciences), CD4-BUV805 (cat. #564922, Clone GK1.5, BioLegend), CD8-Pacific Blue (cat. #558106, Clone 53–6.7, BD Biosciences), CD44-FITC (cat. #553133, Clone IM7, BD Biosciences), CD62L-BUV395 (cat. #740218, Clone MEL-14, BD Biosciences). Cells were incubated in the antibody solution made in staining buffer, which was made with PBS (without calcium or magnesium, 100 mL/L; cat. #14200166, GIBCO), sodium azide (1 g/L; cat. #S2002, Sigma) and BSA (1% final W/V; cat. #BP1605–100, Fisher Scientific) in water, for 20 minutes on ice and protected from light, following the manufacturer’s instructions. Flow cytometry acquisition was performed using an LSRII (BD Biosciences) cytometer, and FACS data analysis was performed with FlowJo software (FlowJo, RRID:SCR_008520). Functional restimulation/coculture assays To investigate antigen specificity following HER3-DC1 administration in mice, spleens and lymph nodes were harvested 2 weeks after the last vaccination, processed as stated above, and splenocytes from control and vaccinated/treated mice were plated in 48-well tissue culture plates at 2 × 106 cells/mL in 1% FBS-supplemented RPMI and rested for 24 hours at 37°C and 5% CO2. After 24 hours, cells were either pulsed (2 μg/mL) with individual HER3 class II immunogenic peptides, a negative class II control peptide, or left unpulsed. For lymph nodes, lymphocytes were plated in a 96-well plate with DCs, previously matured as described above, and pulsed with HER3 class II peptides, a negative class II control (OT-II), or left unpulsed (lymphocyte:DC ratio 10:1). Following 72 hours of incubation, culture supernatant was collected, and IFN-γ secretion was measured after 1:20 dilution of the supernatant, using mouse IFN-γ Quantikine ELISA Kit (cat. #PMIF00, R&D Systems). The same protocol was followed for all mouse models included in the study. Intracellular staining for IFN-γ secretion Tumors were collected (n = 3–5/group) from control and HER3-DC1–treated TUBO tumor–bearing mice. Tumor-infiltrating lymphocytes (TIL) were isolated following the protocol described above and cocultured with mature DC1 cells pulsed with individual HER3 peptides (10:1 TIL:DC; i.e., 106 TIL:105 DC in 1 mL total volume). Intracellular staining was performed using the BD Cytofix/Cytoperm Plus Fixation/Permeabilization Kit with BD GolgiPlug Protein transport inhibitor containing Brefeldin A (cat. #555028, BD Biosciences). Briefly, 6 hours after TIL:DC coculture, GolgiPlug was added to inhibit intracellular protein transport (1 μL/106 cells) for 12 hours. Cells were harvested the next day, and surface staining with CD45-BUV395 (cat. #564279, Clone 30-F11, BD Biosciences), CD4-BUV805 (cat. #612900, Clone GK1.5, BD Biosciences), and CD8-Pacific Blue (cat. #558106, Clone 53–6.7, BD Biosciences) was performed as described above. Cells were fixed and permeabilized following the manufacturer’s protocol and were stained for intracellular IFN-γ-PE (cat. #554412, Clone XMG1.2, BD Biosciences). Acquisition was performed using an LSRII (BD Biosciences) cytometer, and FACS data analysis was performed with FlowJo software (FlowJo). Statistical analyses To compare immune response generated by the peptides, ELISA data were analyzed by multiple t test, without correction for multiple comparisons, using GraphPad Prism software (RRID:SCR_002798). Each row was analyzed individually, without assuming consistent standard deviation. Data are represented as mean SEM. For analyzing immune response across HLA alleles, statistical significance was determined using Mann–Whitney test (ns, P > 0.05), and data are represented as mean ± SEM. To compare tumor growth between groups, data were analyzed using multiple t test without correction for multiple comparisons. Each row was analyzed individually, without assuming a consistent standard deviation, and data are represented as mean ± SEM. A log-rank (Mantel–Cox) test was used to determine differences between the survival curves. Unpaired two-tailed t test was performed to analyze Western blot data. For all analyses, significance threshold was considered as *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001. Results HER3 is overexpressed in multiple cancers HER3 is an established oncodriver that contributes to the growth, proliferation, and survival of cancer cells (30, 31). Analysis of HER3 RNA-seq expression data in 12 cancers from the GDC demonstrated significantly increased HER3 expression in bladder (P = 0.0045), breast (P = 1.15E−12), lung adenocarcinoma (P = 1.77E−19), prostate (P = 6.63E−11), and stomach cancers (P = 1.31E–5; Fig. 1A). Although not statistically significant, a similar trend of high HER3 expression was also noted in pancreatic cancer (P = 0.1843). Conversely, HER3 mRNA expression in tumor tissues was significantly lower than healthy tissues in HNC (P = 7.27E−7) and lung squamous cell carcinoma (P = 1.82E−7), indicating cancer type specificity. Due to lack of data on normal tissue HER3 mRNA expression in TCGA database, comparison of ovarian serous cystadenocarcinoma and melanoma could not be performed; and no change in HER3 expression was noted in colon adenocarcinoma (P = 0.1259) or esophageal cancer (P = 0.758). We used the cBioPortal survival tool to further investigate if elevated/suppressed HER3 expression correlated with overall patient survival. Whereas in breast cancer high versus low HER3 expression had no statistically significant correlation with overall survival (P = 0.182), high HER3 expression showed a significantly negative correlation with overall survival in melanoma (P = 0.0246; Fig. 1B–C). Although HER3 mRNA expression was not significantly different between normal pancreas and pancreatic adenocarcinoma, a significant negative correlation was observed between high HER3 expression and overall survival (P = 6.248E−3) in pancreatic cancer (Supplementary Fig. S1B). A positive correlation between high HER3 expression and overall survival was indicated in head and neck squamous cell carcinoma (P = 3.913E−3; Supplementary Fig. S1C). We did not observe a statistically significant correlation between HER3 expression and overall survival in any of the other cancers investigated (Supplementary Fig. S1D–S1K). These results suggested an impact of increased expression of HER3 in specific malignancies. Identification of HER3 class II peptide epitopes To identify immunogenic class II epitopes from the HER3 protein, human DCs were pulsed with overlapping 15-mer peptides from a HER3 library, rapidly matured into a type I phenotype (DC1), and cocultured with autologous naïve CD4+ T cells (Supplementary Fig. S2). Following restimulation of peptide-primed CD4+ T cells with the matching class II peptide or negative control, supernatants were collected to measure peptide-specific Th1 responses by IFN-γ secretion. Peptide libraries were sequentially screened in pools of 10-peptides, 5-peptides, and individual peptides based on significant fold increases in IFN-γ compared with control (see Materials and Methods and Supplementary Fig. S1A). Both ECD and ICD HER3 peptide libraries were screened separately on three healthy donor samples with two representative screenings reported for ECD (samples 1–2; Fig. 1D and E) and ICD libraries (sample 3–4; Fig. 1F and G) to identify immunogenic HER3 peptides and confirm reproducibility of peptide-specific Th1 immune responses. Screening of HER3 ECD 10-peptide pools revealed p11–20, p81–90, and p91–100 as inducing a significant increase in Th1 immune responses compared with the negative control (Fig. 1D and E). Further breakdown into 5-peptide pools showed p11–15, p81–85, and p91–95 had a comparable significant increase in IFN-γ. Lastly, the corresponding 15 individual peptides were screened, identifying peptides p12 (P = 0.0281), p81 (P = 0.0041), p84 (P = 0.0023), and p91 (P = 0.0161) in sample 1 and likewise, p12 (P = 0.0052), p81 P = 0.0021), p84 (P = 0.0009), and p91 (P = 0.0012) in p31–40, p51–60, and p81–90 demonstrated sample 2. Similarly, significant increases in Th1 immune responses when screening HER3 ICD 10-peptide pools (Fig. 1F and G). Of the three donor samples used in HER3 ICD 10-peptide pool screening, the first donor sample demonstrated an overall lack in response to pool p41–50 for both the control- and peptide-stimulated T cells (Supplementary Fig. S3A). Prior to continued screening of HER3 ICD 10-peptide pools, all peptides within the pool p41–50 were screened individually, with p41 inducing a significant increase (P = 0.0158) in IFN-γ (Supplementary Fig. S3B) and was included as an epitope candidate. To validate this discovery, individual peptides from two additional unresponsive pools were screened in the ECD (Supplementary Fig. S3C–S3D) and ICD (Supplementary Fig. S3E and S3F) libraries; no other positive peptides were identified. 5-Peptide pools were then screened, which revealed p36–40, p51–55, and p86–90 as common positive pools. Thirteen peptides were then screened individually, in addition to the five individual peptides within pool p56–60, which showed a significant increase in IFN-γ in sample 3. Overall, five individual HER3 ICD peptides demonstrated a common significant increase in IFN-γ compared with the negative control: p38 (P = 0.0055), p52 (P = 0.0119), p86 (P = 0.0020), and p89 (P = 0.0028) in sample 3 and, similarly, p38 (P = 0.0043), p41 (P = 0.0124), p52 (P = 0.0018), p86 (P = 0.0016), and p89 (P = 0.0012) in sample 4. Taken together, four HER3 ECD and five HER3 ICD peptides were identified as potential class II epitopes through the sequential peptide screening of HER3 ECD and ICD peptide libraries: HER3aa56–70, HER3aa401–415, HER3aa416–430, and HER3aa451–465, HER3aa850–864, HER3aa865–879, HER3aa920–934, HER3aa1090–1104, and HER3aa1105–1119 (Supplementary Table S1). Screening schema for the ECD and ICD libraries at each step are shown in Fig. 2A–B. The nine HER3 class II peptides were recognized as candidate HER3 epitopes and screened across 6–13 additional samples (Supplementary Table S1) to confirm the reproducibility of the peptide-specific CD4+ Th1 immune responses (Supplementary Fig. S4A–S4F). HER3 epitopes induce HER3-specific Th1 immune responses in vitro To confirm the identified HER3 peptides were competent epitopes in vitro, HER3-peptide-primed CD4+ T cells were restimulated with the corresponding whole HER3 ECD or ICD protein. HER3 ECD peptides demonstrated a significant increase in IFN-γ production when peptide-primed CD4+ T cells were restimulated with the matching class II peptide [(p12, P = 0.0181), (p81 = 0.0075), (p84, P = 0.0025), (p91, P = 0.00004)], and responses were comparable to that of restimulation with the full HER3 ECD protein domain [(p12, P = 0.0417), (p81,(P = 0.0121), (p84, P = 0.0036), p91 (P = 0.0155); Fig. 2C]. HER3 ICD peptide–primed CD4+ Th1 cells restimulated with the matching HER3 class II peptides similarly showed a significant response compared with the peptide negative control [(p38 (P = 0.0012), (p41, P = 0.0021), (p52, P = 0.0035), (p86, P = 0.0008), (p89, P = 0.0021)], which was analogous to that of peptide-primed CD4+ Th1 cells restimulated with the full HER3 ICD protein domain [(p38, P = 0.0131), (p41, P = 0.0191), (p52, P = 0.0430), (p86, P = 0.0072), (p89, P = 0.0159); Fig. 2D]. To determine if the identified HER3 peptides induced comparable anti-HER3 Th1 immune responses in healthy donor and breast cancer patient samples in vitro, PBMCs from healthy donor controls (n = 6) and breast cancer patients (n = 10) were sensitized to the nine HER3 peptides, and IFN-γ responses were evaluated by ELISPOT (Fig. 2E). There was no significant difference between healthy donor controls and breast cancer patients in anti-HER3 responsiveness (P = 0.0820), response repertoire (P = 0.4717), or cumulative responses (P = 0.3132). Together, these results indicate CD4+ T cells specific to the nine HER3 class II peptide candidates could reproducibly elicit the desired HER3-specific Th1 immune response in different donor and breast cancer patient samples and were competent epitopes in vitro. HER3 class II peptides represent promiscuous epitopes To evaluate if the identified HER3 class II peptides were restricted to specific HLA alleles, donor samples used in both ECD and ICD peptide library screenings were analyzed for MHC class II HLA allele expression (Supplementary Table S2). A total of 35 different alleles were expressed across HLA-DR (13 alleles), HLA-DP (11 alleles), and HLA-DQ (11 alleles), with minimal overlap in expression across donor samples (≤3 samples expressed the same HLA allele). Allele expression was cross-referenced with the corresponding HER3 ECD and HER3 ICD (Supplementary Table S3) peptide–specific Th1 immune responses. Together, these data confirmed high variability in HLA expression across samples and demonstrates the promiscuous binding of identified HER3 peptides across multiple MHC II HLA alleles, indicating the application of identified HER3 epitopes for a HER3-DC1 vaccine with capabilities for widespread application. Class II predictive algorithm did not identify HER3 peptides as candidate epitopes MHC II prediction algorithm, NetMHCIIpan 4.0, was used to predict class II peptides from the full HER3 protein sequence, and predicted results were compared with those obtained from the experimental peptide screening. NetMHCIIpan predicted approximately 458 class II peptide sequences (15-mer) with >700 different 9-mer binding cores exhibiting strong binding affinity (<50 nmol/L; Table 1). HER3 ECD p91 was the only peptide with predicted strong binding affinity with multiple HLA alleles. The remaining HER3 peptides were predicted to establish only weak binding interactions (50–500 nmol/L) with approximately 1–19 different HLA alleles. However, the MixMHCIIpred algorithm predicted a strong binding affinity for all nine HER3 peptides, with p91 showing the strongest binding affinity to most of the 38 tested HLA alleles (Table 1). Most of the peptides showed strong binding affinity toward a handful of HLA alleles, although each showed significant immunologic responses when tested experimentally. These results indicate the inability of the NetMHCIIpan prediction algorithm, as well as inconsistencies between algorithms, to identify the majority of HER3 class II peptides as candidate epitopes from the full HER3 protein and supports the indispensability of an empirical approach to epitope identification for tumor antigens. Antigen-specific responses and delayed tumor growth by preventive vaccination To investigate the antitumor immune response generated by the nine identified human HER3 peptides, we investigated both preventive and therapeutic efficacy of HER3 peptide–pulsed DC1 (HER3-DC1) in preclinical murine models. Because we observed high HER3 expression in breast cancer and melanoma compared with normal tissues in our TCGA analysis, we chose murine mammary carcinoma models 4T1 and TUBO, which represent human TNBC and HER2pos breast cancer, respectively, and the M05 murine melanoma model. Western blots showed HER3 protein expression in all three cell lines, with TUBO cells showing the highest and 4T1 showing the lowest HER3 expression (Fig. 3A). Immunofluorescence staining showed HER3 expression in 4T1 and TUBO cells, primarily confined to the cellular surface of both cell lines (Fig. 3B). In human TNBC and HER2pos breast cancer cells, we observed HER3 expression, although the expression level varied considerably across cell lines (Supplementary Fig. S5A and S5B). To test the human peptides in murine models, we tested the amino acid sequence similarity between the identified human HER3 peptides and full-length murine Erbb3 amino acid sequence using the BLASTP program. We observed 100% sequence homology between murine Erbb3 and human HER3 peptides ECD p12, ECD p81, ECD p84, ICD p38, and ICD p86; 93% sequence homology with ICD p41, ICD p52, and ICD p89; and 87% sequence homology with ECD p91 (Supplementary Table S4). Next, we evaluated the preventive efficacy of HER3-DC1 in BALB/c mice by vaccination with either unpulsed DC1 or HER3-DC1, followed by 4T1 or TUBO tumor challenge. To investigate immune response and antigen specificity of the response to HER3 peptides in murine models, splenocytes from vaccinated and naïve control mice were restimulated with the HER3 peptides for 72 hours ex vivo, followed by an IFN-γ ELISA using culture supernatants. We observed significantly higher IFN-γ secretion from HER3-DC1–vaccinated mice compared with control mice (Fig. 3C; Supplementary Fig. S5C). No significant difference in splenocyte IFN-γ secretion from unpulsed DC1 vaccinated mice compared with controls was observed, suggesting immune responses were specifically stimulated by HER3 peptides and was not a result of nonspecific immune stimulation by mature, nonprimed DC1 vaccination. Similarly, IFN-γ in supernatants from cocultures of the lymph node–derived lymphocytes with DC1 pulsed with cognate HER3 peptides showed significantly increased IFN-γ secretion from HER3-DC1–vaccinated BALB/c mice compared with naïve controls (Fig. 3D; Supplementary Fig. S5D). These data suggest that the development of systemic immune responses in HER3-DC1–vaccinated mice is HER3 peptide–specific and not induced by unpulsed mature DC1. Two weeks after the last vaccination, HER3-DC1–vaccinated or unvaccinated control mice were challenged with either 4T1 or TUBO tumor to compare preventive efficacy of HER3-DC1 vaccine. HER3-DC1–vaccinated mice challenged with 4T1 showed significantly delayed tumor growth (P = 0.007) and extended survival, whereas there was no significant difference in tumor growth and survival between control and unpulsed DC1-vaccinated mice (P = 0.9782; Fig. 3E; Supplementary Fig. S5E). Tumor growth in HER3-DC1–vaccinated mice was also significantly delayed compared with the unpulsed DC1 mice (P = 0.0086), suggesting the role of anti-HER3 Th1 response in preventing tumor growth over unpulsed DC1 vaccine. In the TUBO model, we observed significantly delayed tumor growth in HER3-DC1–vaccinated mice compared with the unvaccinated controls (P = 0.0191; Fig. 3F). Despite clear preventive benefit and systemic immune response after HER3-DC1 vaccination, the NetMHCIIPan4.0 algorithm did not predict binding affinity of any of the nine HER3 peptides (Supplementary Table S5), further highlighting the limitation of false-negative predictions by algorithms. These data suggest HER3-DC1 vaccination stimulates HER3 antigen–specific immune responses and offers preventive benefit to delay tumor growth in both TNBC and HER2pos breast cancer models. Intratumoral HER3 DCs delay tumor growth and enhance immune infiltration Next, we evaluated the therapeutic efficacy of HER3-DC1 in the aggressive murine mammary carcinoma model 4T1 (32), which mimics human TNBC. Intratumoral HER3-DC1 administration significantly delayed tumor growth compared with controls (P = 0.0019) and unpulsed DC1-treated mice (P = 0.0183), indicating therapeutic benefit of HER3-DC1 (Fig. 4A and B), and no significant difference was noted in tumor growth between control and unpulsed DC1–treated groups. We also observed a significant increase in survival rate in the HER3-DC1–treated group compared with control mice (P = 0.0327; Fig. 4C). Compared with the intratumoral route of delivery, subcutaneous HER3-DC administration did not show therapeutic benefit in the 4T1 model (Supplementary Fig. S5F and S5G). We then investigated changes in the intratumoral immune landscape after intratumoral delivery of HER3-DC1, as well as in the tumor-draining lymph nodes, by flow cytometry. The flow gating strategy to identify CD4+ and CD8+ TILs and their subtypes is shown in Supplementary Fig. S6. Flow analysis revealed significantly higher CD4+ (P = 0.03) and CD8+ (P = 0.0042) T-cell in filtration per milligram of tumor in HER3-DC1–treated mice compared with control tumors (Fig. 4D; Supplementary Fig. S7A). We observed a significant increase in the CD62L−CD44+ effector memory (EM, P = 0.0304), CD62L+CD44+ central memory (CM, P = 0.0057), and CD62L−CD44+ effector (P = 0.0412) T-cell populations in HER3-DC1–treated tumors compared with controls (Fig. 4E; Supplementary Fig. S7B). We then compared immune phenotypes of cells isolated from the draining lymph nodes of the HER3-DC1 versus control mice and observed no statistically significant increases in total CD4+ (P = 0.6492) and CD8+ (P = 0.7699) populations in the HER3-DC1 group compared with controls (Fig. 4F), and no statistically significant increases in CD4+ CM, EM, and effector T-cell abundance in the HER3-DC1 group compared with the controls was seen (Fig. 4G). To evaluate systemic immune responses after intratumoral HER3-DC1 administration, lymph nodes harvested from control and treated mice were cocultured with DC1 cells pulsed with individual HER3 peptides or OT-II (as negative control) for 72 hours. ELISA with culture supernatants revealed significantly higher IFN-γ secretion in the HER3-DC1 group, compared with the control, for all nine peptides (Fig. 4H). Lastly, we investigated HER3 protein expression from in vivo tumor samples by immunoblotting to determine if intratumoral DC administration altered receptor protein expression in tumor cells. Total HER3 protein expression was significantly reduced in HER3-DC1–treated tumors compared with control tumors (P = 0.0369; Fig. 4I), suggesting molecular cross-talk between HER3-DC1–induced immune responses and oncodriver signaling can affect receptor protein expression in 4T1 tumor cells. HER3-DC1 delays tumor growth and enhances immune infiltration in the TUBO model HER3 is the most potent dimerization partner of HER2 that facilitates downstream signaling, contributing to the growth and proliferation of tumor cells. Hyperactivation of HER3 has also been identified as one of the primary mechanisms of therapeutic resistance to HER2-targted therapies in HER2pos breast cancer (33, 34). Therefore, we investigated the therapeutic efficacy of HER3-DC1 in a HER2-resistant TUBO murine mammary tumor model (35). We observed significantly delayed tumor growth in mice that received intratumoral HER3-DC1 compared to control- (P = 0.0028) and unpulsed DC1-treated mice (P = 0.0185). Tumor regression occurred in 30% of HER3-DC1–treated mice, and growth was significantly delayed in the remaining mice, compared with the controls (Fig. 5A; Supplementary Fig. S8A). This resulted in a significant improvement in survival of HER3-DC1–treated mice compared with the control (median survival 64 days vs. 46 days, P = 0.0002) and unpulsed DC1-treated groups (median survival 64 days vs. 49 days, P = 0.0015; Fig. 5B). To evaluate immune infiltration into the tumor microenvironment after HER3-DC1 intratumoral delivery, we collected tumors from control- and HER3-DC1–treated mice and analyzed immune cell abundance by flow cytometry. Compared with control tumors, we observed significantly higher infiltration of CD4+ (P = 0.0449) and CD8+ (P = 0.0201) T in cells tumors from HER3-DC1–treated mice (Fig. 5C; Supplementary Fig. S8B). Within the CD4+ population, we noted significantly amplified abundance of CD62L+CD44+ CM (P = 0.0446) and CD62L−CD44+ EM (P = 0.0458), but not CD62L−CD44− effector T cells (Fig. 5D). We also observed significantly increased CD8+ T cells, specifically CD62L+CD44+ CM (P = 0.0131), and CD62L−CD44+ EM T cells (P = 0.033) in tumors from HER3-DC1–treated mice compared with the controls (Supplementary Fig. S8C). When CD4+ T cells were selectively depleted, tumor growth was comparably faster than the control group, and the antitumor effect of HER3-DC1 was completely abolished (Fig. 5E), highlighting the necessity of CD4+ T cells for functional activity of the class II peptide–pulsed HER3-DC1. When TILs from these mice were cocultured with DC1 pulsed with individual HER3 peptides, a significantly higher percentage of CD4+IFN-γ+ cells in HER3-DC1–treated TILs were noted in response to each of the HER3 peptides compared with the untreated controls. These data indicate HER3-specific IFN-γ secretion by CD4+ T cells following HER3-DC1 administration (Fig. 5F; Supplementary Fig. S8D). On the contrary, no statistically significant difference was noted in the percentage of CD8+IFN-γ+ cells between groups (Fig. 5G). In HER3-DC1–treated lymph nodes, we observed minimal, but not statistically significant, increase in the number of CD4+ and CD8+ T cells (Supplementary Fig. S8E). Together, these data suggest intratumoral HER3-DC1 can stimulate CD4+ immune responses, which, in turn, can activate CD8+ T cells. Coculture of the lymph node immune cells with HER3 peptide–pulsed DC1 for 72 hours confirmed antigen specificity, with significantly augmented IFN-γ secretion in the HER3-DC1–treated group compared with controls (Fig. 5H). Similarly, elevated IFN-γ secretion was noted when HER3-DC1 splenocytes were restimulated with HER3 peptides, compared with the controls (Supplementary Fig. S8F). Next, we examined whether HER3-DC1 intratumoral administration alters the molecular identity of TUBO tumor cells. Western blots on tumor cell lysates revealed downregulation of HER3 protein in HER3-DC1–treated tumors compared with controls. We also observed diminished phosphorylated AKT in HER3-DC1 tumors, along with an upregulation of apoptosis marker, cleaved caspase-3, compared to the control tumors (Fig. 5I). Downregulation of phosphorylated p44/42 MAPK in HER3-DC1 tumors was also noted, whereas no difference in HER3 and phospho-p44/42MAPK expression was observed in tumors from unpulsed DC1-treated mice compared to the controls (Fig. 5J). These data suggest HER3-DC1–induced Th1 immune responses can modulate molecular signaling to interfere with oncodriver expression and activation and induce tumor cell apoptosis. HER3-DC1 vaccination prevents tumor development in the M05 model Because TCGA and cBioPortal analyses indicated high HER3 expression in skin cutaneous melanoma compared with normal tissue and a negative correlation between high HER3 expression and overall patient survival, we tested the preventive and therapeutic efficacy of HER3-DC1 in an M05 murine melanoma model, derived from the original B16 melanoma cell line transfected with the ovalbumin gene (36). Immunoblotting of M05 cell lysates showed moderate HER3 protein expression (Fig. 3B), whereas substantial surface expression of HER3 was detected by immunofluorescence (Fig. 6A). Therefore, we tested the preventive efficacy of HER3-DC1 in C57BL/6 mice. Compared with controls, splenocytes from HER3-DC1–vaccinated mice showed significantly enhanced peptide-specific immune responses when restimulated with HER3 peptides (Fig. 6B). Similarly, when immune cells isolated from the lymph nodes of vaccinated mice were cocultured with HER3 peptide–pulsed DC1 ex vivo, we observed significantly higher IFN-γ secretion compared with naïve controls (Fig. 6C). When naïve and vaccinated mice were challenged with M05 tumor 2 weeks after the last vaccination, tumor growth was significantly delayed, and survival was improved in vaccinated mice compared with naïve controls (P = 0.0004; Fig. 6D; Supplementary Fig. S8G). These data suggest HER3-DC1 vaccination induces peptide-specific, systemic immune responses in C57BL/6 mice. Intratumoral HER3-DC1 diminishes tumor growth in the M05 melanoma model Next, we investigated the therapeutic efficacy of intratumoral HER3-DC1 treatment in a subcutaneous M05 murine melanoma model. An OT-II–pulsed DC1 group was included as a positive control, since the M05 cell line expresses ovalbumin. Compared with controls and unpulsed DC1-treated mice, HER3-DC1 treatment significantly delayed tumor growth (Fig. 6E; control vs. HER3-DC1, P ≤ 0.000001; unpulsed DC1 vs. HER3-DC1, P < 0.0001). HER3-DC1 treatment was comparable with OT-II-DC1 treatment in terms of tumor growth and tumor regression. We observed tumor regression in 20% of HER3-DC1–treated mice, whereas tumor growth was significantly slower in 40% of the remaining mice, with a steady growth in the rest. This resulted in significantly improved survival in HER3-DC1–treated mice compared with controls (P ≤ 0.0001, median survival control = 25 days, HER3-DC1 undefined; Fig. 6F). We next assessed how intratumoral HER3-DC1 treatment modulated the tumor immune landscape. Like the murine mammary carcinoma models, we observed significantly increased CD4+ T-cell infiltration in HER3-DC1 mice compared with the controls (P = 0.0339). A significantly enhanced CD8+ T-cell infiltration was also noted in the HER3-DC1–treated group, which could contribute to the antitumor efficacy of the intratumoral HER3-DC1 observed (P = 0.0079; Fig. 6G). For CD4+T-cell subsets, EM (P = 0.0208) and CM (P = 0.0199) populations were significantly enhanced in the HER3-DC1 group, with no differences in effector T-cell populations (Fig. 6H). We investigated the systemic immune responses mediated by intratumoral HER3-DC1 treatment. Lymph node immune cells from control and treated mice were assessed for lymphoid markers. As previously seen in the breast cancer models, we observed significantly enhanced CD4+ T cells in HER3-DC1–treated mice (P = 0.0085), which was comparable to the positive control OT-II-DC1 group; however, no significant difference was noted in the frequency of CD8+ T cells (Fig. 6I). Within the CD4+ population, we noted significantly increased EM (P = 0.0058), CM (P = 0.0052), and effector (P = 0.0016) T cells (Fig. 6J). We evaluated IFN-γ secretion by immune cells residing in the lymph node of treated and control mice. As previously seen in the breast cancer models, we noted significantly enhanced IFN-γ secretion in the HER3-DC1 group compared with the control, when lymph node immune cells were cocultured with HER3 peptide–pulsed DCs (Fig. 6K). In the OT-II-DC1–treated group, a significant positive response was noted only when cocultured with the OT-II peptide–pulsed DC1 (P = 0.00567), but not the HER3 peptides, indicating the antigen specificity of the response. We also detected significantly higher IFN-γ secretion from splenocytes isolated from the HER3-DC1–treated M05 mice compared with controls (Supplementary Fig. S8H). Together, these data suggest HER3-DC1 induces potent antitumor immune response in a HER3+ murine melanoma model, resulting in tumor regression/delayed growth and improved survival by enhancing intratumoral and systemic CD4+ T cells and secretion of IFN-γ. Discussion HER3/ERBB3, a member of the ERBB family of growth receptors, has gained momentum as a therapeutic target, owing to its multifaceted role in tumor development, growth, and therapy resistance (1). In TCGA, HER3 expression was found to predict worse survival in melanoma patients. Interestingly, in response to BRAF/MEK inhibitor treatment, melanoma cells are shown to adapt and escape therapy by upregulating HER3 expression via the FOXP3 transcription factor (5). DC1 pulsed with promiscuous MHC class II HER3 peptides demonstrated a dramatic therapeutic effect in a murine melanoma model that expressed increased HER3, suggesting the possibility of clinically developing a HER3 immunotherapy in melanoma patients expressing HER3, especially those treated with BRAF/MEK inhibitors to possibly prevent escape through HER3. We did not find a significant impact of high HER3 expression on breast cancer survival, which could be attributed to the small sample size. However, high HER3 expression has been previously identified as a prognostic marker of poor overall and disease-free survival in TNBC (10) and has been shown to contribute to therapy resistance (37). Upregulated HER3 expression and activation have also been documented as one of the primary compensatory mechanisms allowing therapy escape in HER2pos breast cancer via restimulation of the PI3K/AKT pathway (38). We utilized a HER2pos mammary tumor model resistant to HER2-targeted therapies, which was sensitive to HER3-mediated CD4+ Th1 immune response. This raises the possibility that HER3 peptides can be used to overcome resistant HER2 populations. CD4+ T cells are critical players in regulating antitumor immune responses, and Th1 cells secreting IFN-γ contribute to elimination of tumors through multiple mechanisms, including antibody class switching, CD8+ T-cell help, modulation of innate effectors, such as macrophages and NK cells, and maturation of DCs (39). This study showed increased production of IFN-γ by CD4+ Th1 cells in both human CD4+ T cells ex vivo and in murine preclinical models of mammary carcinoma and melanoma. CD4+ T cells accumulated in the tumors of mice treated with HER3 peptide–pulsed DC1, suggesting a contribution of the CD4+ response in tumor regression. We validated the significance of CD4+ T cells in stimulating antitumor immune responses by class II peptide–pulsed DC1 treatment. HER3-DC1 intratumoral administration significantly enhanced CD4+ T-cell infiltration, and restimulation with HER3 peptide–pulsed DC resulted in increased intracellular IFN-γ secretion by those CD4+ TILs. This observation highlights the CD4-dependent mechanism of action of the HER3-DC1. In the 4T1 and TUBO models, HER3-DC1 intratumoral injection and subsequent Th1 immune responses result in delayed tumor growth and downregulation of HER3 expression in tumor cells, which indicates possible correlation at the mechanistic level between HER3 expression and the antitumor effect. We observed significantly decreased total HER3 protein in the HER3-DC1 tumors obtained from both 4T1 and TUBO models, suggesting that intratumoral delivery of HER3-DC1 may induce cellular cross-talk downstream and inhibit HER3 protein expression, which contributes to tumor growth and proliferation. For TUBO tumors, we observed HER3-DC1 tumors had downregulation of phospho-AKT and phospho-p44/42MAPK, the primary signaling effectors downstream of HER3, suggesting HER3-DC1 may interfere with activation and function of HER3 in tumors. Also, prominent upregulation of the apoptosis marker cleaved caspase-3 indicated intratumoral cell death due to HER3-DC1. The Th1 cytokine IFN-γ can downregulate expression of HER family of receptor proteins in vitro (35) and, hence, HER3-DC1 treatment may reduce HER3 expression by CD4+ Th1 stimulation. The peptide library screening method described here identified promiscuous class II epitopes that were not clearly predicted by current binding algorithms. Only ECD p91 was predicted to have high binding affinity, whereas none of the remaining eight peptides were indicated to be high-affinity binders. Previous reports have shown high rates of false negatives for immunogenic peptide epitopes that are intermediate- or low-affinity MHC binders (40–43). The MixMHCIIpred algorithm predicted strong binding affinity of all nine peptides identified empirically. However, it should be considered that binding affinity alone is not the only predictor of immunologic response generated by immunogenic peptides, and hence, the large number of random peptides generated by the algorithm that showed a strong binding affinity will still need to be tested individually and experimentally for immune responses in vitro, to avoid false-positive epitopes. Absence of defined cutoff percentile rank also leads to subjective interpretation of strong versus weak binding. Our identified peptides showed promiscuous binding across multiple MHC II HLA alleles, suggesting broad application across patients, whereas MixMHCIIpred algorithm showed variation of binding affinity of the peptides across HLA alleles. Lastly, the stark difference of predictive accuracy between the two algorithms used in this study further highlights the efficiency of our empirical approach to accurately identify immunologically responsive peptides. MHC I prediction algorithms have shown success in identifying immunogenic class I epitopes that can elicit potent CD8+ T-cell responses in vivo; however, the identification of MHC class II epitopes that activate CD4+ T cells has been less successful (24, 41–46). This discrepancy can be attributed to the highly polymorphic nature of MHC II molecules, which results in different binding patterns across the diverse HLA alleles (43–45, 47–49). Unlike MHC I, MHC II molecules consist of an open binding groove, allowing for longer peptides of 9–25 amino acid length to bind at different locations across the groove’s open surface (43, 45, 46, 48). Our peptide screening attempted to circumvent several limitations in MHC II prediction algorithms to identify immunogenic class II epitopes for use in therapeutic development. First, a library of 15-mer peptides was generated for the protein of interest, with a 10-amino acid overlap between adjacent sequences. This design aimed to account for differences in the binding pattern across the MHC II open binding groove. Second, screening the full peptide library sequentially in pools of 10 peptides, 5 peptides, and individual peptides rapidly highlighted regions within the protein that may harbor immunogenic peptide sequences. Screening the full library essentially surveys all regions of the protein for epitope candidates, minimizing the probability of false negatives. Third, peptide candidates were screened across numerous donor samples to ensure the peptide-specific immune responses were reproducible and demonstrated promiscuous binding for widespread therapeutic application. Lastly, a key benefit to this methodology is that it takes an entirely experimental approach to epitope identification by recapitulating the in vivo processes of antigen processing and presentation, T-cell priming, and immune activation to isolate peptides capable of acting as competent epitopes in vitro and in vivo. Because this overlapping peptide library can be created and rapidly tested, it offers a novel avenue to MHC II epitope discovery against other known oncodrivers. We have, thus, begun to interrogate such oncodrivers for CD4+ Th1 peptide epitopes. Our study may also open the avenue for potential use of the DC vaccine platform to generate oncodriver-specific CD4+ T cells for novel adoptive T-cell therapies. In summary, using an overlapping 15-mer peptide library derived from the HER3 oncodriver, we identified promiscuous MHC II epitopes capable of driving an anti-HER3 CD4+ Th1 response that can have therapeutic impact in breast cancer and melanoma, and can be developed for cancer therapy in these diseases. Supplementary Material Supplementary Figures Supplementary File: R Module Supplementary Tables Supplementary Ledgends Acknowledgments This work was supported by Department of Defense Award W81XWH-19-1-0675, awarded to B.J. Czerniecki, P.A. Forsyth, P.C. Rodriguez, and P. Kalinski. This work was also supported by Department of Defense Award W81XWH-16-1-0385, awarded to B.J. Czerniecki and G.K. Koski, and Pennies in Action, awarded to B.J. Czerniecki and G.K. Koski. This work was supported in part by the Flow Cytometry Core, Analytic Microscopy Core Facility, Cell Therapies Core, and Vivarium Services Core at the H. Lee Moffitt Cancer Center and Research Institute, an NCI-designated Comprehensive Cancer Center (P30-CA076292). Authors’ Disclosures K.N. Kodumudi reports a patent for PCT/US2020/050689 pending to Moffitt Cancer Center. K.S.M. Smalley reports personal fees from Elsevier outside the submitted work. P.A. Forsyth reports personal fees from AbbVie Inc., Bayer, Bristol Myers Squibb, Boehringer Ingelheim, Inovio, NCI-NIH-NCRI, Novocure, Novellus, Physical Sciences Oncology Network, Tocagen, Ziopharm, and grants from CDMRP, Department of Defense, Moffitt Cancer Center of Excellence, NIH/NCI, Pfizer, and State of FL Bankhead Coley outside the submitted work. H. Han reports personal fees from Lilly and grants from Department of Defense outside the submitted work. H. Soliman reports personal fees from PUMA, Novartis, Eisai, Seattle Genetics, and AstraZeneca outside the submitted work. B.J. Czerniecki reports grants from CDMRP DOD during the conduct of the study; other support from ImmunoRestoration outside the submitted work; and a patent for identification of immunogenic MHC class II peptides for immune-based therapy (patent# 10,829,538) issued to the University of Pennsylvania. No disclosures were reported by the other authors. Figure 1. HER3 expression in cancer, and HER3 peptide screening of ECD and ICD class II peptide libraries. A, Expression of HER3 mRNA in normal (N) and tumor (T) tissues obtained from RNA-seq data from the GDC across cancer types (see Materials and Methods). B, Correlation between percentage of HER3 expression and overall patient survival (in months) in breast cancer. Samples were sorted in the descending order of HER3 expression and put into two groups: high HER3 (red) and low HER3 (blue). C, Correlation of overall patient survival with high HER3 (red) versus low HER3 (blue) expression in melanoma. P value indicated in individual graphs. D and E, IFN-γ production at each screening step for sample 1 [normal donor (ND) 8; D] and sample 2 (ND 9; E) when stimulated with ECD peptides. F and G, IFN-γ production at each screening step for sample 3 (ND 3; F) and sample 4 (ND 5; G) upon stimulation with ICD peptides. D–G, IFN-γ response to negative peptide control (black) compared with HER3 peptides (red) with an immunogenic response threshold of ≥1.5-fold increase. Data represented as mean ± SEM with statistical significance determined using a multiple t test without correction for multiple comparisons. Each row was analyzed individually, without assuming consistent SD.*, P < 0.05; **, P < 0.01; ***, P < 0.001. Figure 2. HER3 class II peptides demonstrate anti-HER3 immune responses in vitro. A and B, ECD and ICD peptide libraries were screened sequentially in 10-peptide pools, 5-peptide pools, and individual peptides with an immunogenic response threshold of ≥1.5-fold increase in IFN-γ production between peptide and control restimulated CD4+ T cells (see Fig. 1), ultimately identifying four ECD and five ICD HER3 class II peptides. Each schematic is representative of the combined responses across samples used in the peptide screening (n = 3), indicating reproducible significant immunogenic response compared with the class II control in ≥2 samples and number of donor responses in parentheses (red). C and D, Peptide-primed CD4+ T cells were restimulated with matching class II peptide (HER3 ECD or HER3 ICD), class II–negative control (peptide control), whole HER3 domain protein (WP HER3 ECD or WP HER3 ICD), or whole protein control (WP control). Data represented as mean SEM with statistical significance determined using multiple t test without correction for multiple comparisons. Each row was analyzed individually, without assuming a consistent SD.*, P < 0.05; **, P < 0.01; ***, P < 0.001. E, PBMCs from healthy donors (HD, black bar, n = 6) or breast cancer patients (Patient, red bar, n = 10) were individually stimulated with the nine HER3 class II peptides and analyzed via IFN-y ELISPOT. Left, percentage of subjects responding to ≥1 HER3 peptide (anti-HER3 responsivity). Middle, mean number of peptides inducing anti-HER3–specific immunity (response repertoire). Right, total IFN-γ spots (mean total SFC/1e6 cells) from stimulation with HER3 peptides (cumulative response). Data represented as mean ± SEM with statistical significance determined using Mann–Whitney test. ns, not significant. Figure 3. HER3-DC1 vaccination elicits peptide-specific immune response and delays tumor growth. A, Immunoblotting of murine tumor cell lines 4T1, TUBO, and M05 to detect HER3. β-Actin: loading control. B, Immunofluorescence for HER3 (red) and nucleus (DAPI, blue) in 4T1 and TUBO murine mammary tumor cells (image magnification: 1,200). C and D, Individual HER3 peptide–specific immune responses in spleens (C) and lymph node–derived immune cells (D) from control (black), unpulsed mature DC1 (blue), and HER3-DC1 (red) vaccinated BALB/c mice (n = 3). Spleens were processed, and splenocytes were restimulated with the HER3 peptides for 72 hours to detect IFN-γ by ELISA. Lymph node–derived lymphocytes were cocultured with DC1 pulsed with individual HER3 peptides for 72 hours to detect IFN-γ by ELISA. E, Tumor growth after 4T1 tumor challenge in control (black), unpulsed mature DC1 (blue), and HER3-DC1 (red) vaccinated mice (n = 7–10 mice/group). Mice were challenged 2 weeks after the last vaccination and were monitored until endpoint., control versus HER3-DC1; #, unpulsed DC1 versus HER3-DC1. F, TUBO tumor growth in control (black) and HER3-DC1 (red) vaccinated mice (n = 7–10 mice/group). Mice were challenged 2 weeks after the last vaccination. Data represented as mean ± SEM with statistical significance determined using multiple t test without correction for multiple comparisons. Each row analyzed individually, without assuming a consistent SD.*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ##, P ≤ 0.01. Figure 4. Intratumoral HER3-DC1 administration elicits peptide-specific immune responses and delays tumor growth. A, Tumor growth in the 4T1 murine mammary carcinoma model. BALB/c mice bearing subcutaneous 4T1 tumors received either intratumoral PBS (black), unpulsed mature DC1 (blue), or HER3 peptide–pulsed DC1 (red; n = 10 mice/group), starting on day 7 when tumors were palpable. Tumor growth was monitored until endpoint and was compared between control and HER3-DC1, as well as between unpulsed DC1 and HER3-DC1., control versus HER3-DC1; #, unpulsed DC1 versus HER3-DC1. B, Individual tumor growth for each mouse from control (black)-, unpulsed DC1 (blue)–, and HER3-DC1 (red)–treated groups. C, Percent survival in the 4T1 mouse model. Control, black; unpulsed DC1, blue; HER3-DC1, red. D, Intratumoral CD3+CD4+ and CD3+CD8+ T-cell infiltration per milligram of tumor in control (black)-, unpulsed DC1 (blue)–, and HER3-DC1 (red)–treated mice. Absolute number of immune cells was compared between control and HER3-DC1 groups. E, Frequency of CD62L+CD44+ central memory (CM), CD62L−CD44+ effector memory (EM), and CD62L−CD44− effector (Eff) T-cell populations within intratumoral CD4+ cells between control- (black) and HER3-DC1–treated (red) tumors. The unpulsed DC1 (blue) group was not included in any statistical analyses. F, Absolute number of CD3+CD4+ and CD3+CD8+ T cells in lymph nodes of control (black)-, intratumoral unpulsed DC1 (blue)–, and HER3-DC1 (red)–treated mice. Cell numbers were compared in control versus HER3-DC1 groups. Data shown are the representative from three independent experiments. G, Absolute numbers of CD4+ CM, EM, and Eff T-cell populations in lymph nodes of control (black), unpulsed DC1 (blue), and HER3-DC1 (red) mice. Data shown are the representative from three independent experiments. H, Lymphocytes from the lymph nodes of control and treated mice were cocultured with DC1 pulsed with individual HER3 or OT-II (negative control) peptides. Culture supernatants were collected after 72 hours, and IFN-γ was measured by ELISA (control: black bar; HER3-DC1: red bar). I, Total protein isolated from in vivo tumor samples was analyzed by Western blotting to compare HER3 protein expression after intratumoral HER3-DC1 (green) administration with respect to the control (black). β-Actin: loading control. Data represented as mean ± SEM with statistical significance determined using multiple t test without correction for multiple comparisons. Each row was analyzed individually, without assuming a consistent SD. A log-rank (Mantel–Cox) test was used to determine differences between the survival curves. Unpaired two-tailed t test was performed to analyze Western blot data. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; #, P ≤ 0.01. Figure 5. Intratumoral HER3-DC1 delays tumor growth and enhances immune infiltration in a HER2pos TUBO therapeutic model in a CD4-dependent manner. A, Tumor growth in the TUBO murine mammary carcinoma model. BALB/c mice were injected with TUBO tumor cells, and on day 7, mice received either PBS control (black), unpulsed mature DC1 (blue), or HER3-DC1 (red) intratumorally once weekly for six doses (n = 10 mice/group). Tumor growth was monitored until endpoint and compared in control versus HER3-DC1 (*) and unpulsed DC1 versus HER3-DC1 (#) mice. B, Percent survival in TUBO mouse model. Control: black; unpulsed DC1: blue; HER3-DC1: red. C, CD3+CD4+ and CD3+CD8+ T cells per milligram of tumors from mice (A) after intratumoral DC injection was compared between control (black) and HER3-DC1 (red) groups. No statistical analyses were performed for the unpulsed DC1 (blue) mice (n = 3/group). D, Abundance of CD4+ central memory (CD62L+CD44+ CM), effector memory (CD62L−CD44+ EM), and effector (CD62L−CD44− Eff) T cells in control (black) versus HER3-DC1 mice (red) per milligram of tumor tissue. Data shown are the representative from three independent experiments. E, Tumor growth of TUBO tumors after CD4 depletion. BALB/c mice were injected with anti-CD4 antibodies 3 days before subcutaneous TUBO tumor injection. When tumors were palpable, mice received either PBS control (black), intratumoral HER3-DC1 once weekly (red) for six doses, CD4 depletion antibody alone (blue; continued twice weekly until endpoint), or HER3-DC1 (green) with CD4 depletion. Tumor growth was monitored until endpoint. F and G, Percentage of CD4+IFN-γ+ (F) and CD8+IFN-γ+ (G) TILs in the tumors from control (black) versus HER3-DC1 (red) mice from E. H, Coculture of the lymph node immune cells with HER3 peptide–pulsed DC1 for 72 hours to detect IFN-γ via ELISA. Control: black bar; unpulsed DC1: blue bar; HER3-DC1: red bar. I, Western blot for HER3, phosphorylated AKT (phAKT), and cleaved caspase-3 (clCasp-3) with total protein isolated from control- and HER3-DC1–treated TUBO tumors. β-Actin: loading control. J, Western blot for HER3 and phosphorylated p44/42 MAPK (ph-p44/42 MAPK) from control-, unpulsed DC1–, and HER3-DC1–treated TUBO tumors. β-Actin: loading control. Data represented as mean ± SEM with statistical significance determined using multiple t test without correction for multiple comparisons. Each row was analyzed individually, without assuming a consistent SD. A log-rank (Mantel–Cox) test was used to determine differences between the survival curves.*, P ≤ 0.05; **, P ≤ 0.01; **, P ≤ 0.001; #, P ≤ 0.01. Figure 6. Vaccination prevents tumor development in a preventive HER3+ melanoma model and diminishes tumor growth and improves survival in a therapeutic HER3+ melanoma model. A, Immunofluorescence for HER3 (red) surface expression in M05 cells. DAPI (blue): nucleus (image magnification: 1,200×). B, HER3 peptide–specific immune responses (n = 3–5/group). Splenocytes from control (black) and vaccinated (red) mice were restimulated with individual HER3 peptides for 72 hours to detect IFN-γ by ELISA. C, Lymph node lymphocytes from control (black) and HER3-DC1 (red) vaccinated mice were cocultured with individual HER3 peptide–pulsed DCs to detect antigen-specific immune response in HER3-DC1 vaccinated mice by ELISA. D, Preventive model: Two weeks after the last vaccine, C57BL/6 mice (n = 10 mice/group) were challenged with M05 tumor cells. Tumor growth was monitored until endpoint in control (black) versus HER3-DC1 vaccinated (red) mice. E, Therapeutic setting: C57BL/6 mice were injected subcutaneously with the M05 murine melanoma cells in the left flank, and upon palpable tumor formation on day 10, mice were randomized into four groups (n = 10 mice/group). Tumor growth was monitored in mice receiving PBS (black), unpulsed DC1 (blue), HER3-DC1 (red), and OT-II peptide–pulsed DC1 (green) intratumorally once weekly for 6 weeks. Tumor growth was compared between control and HER3-DC1 (*) and unpulsed DC1 vs. HER3-DC1 (#) groups. Individual tumor growth curve for each mouse shown on the right. F, Percent survival in M05 mouse model for control (black)-, unpulsed DC1 (blue)–, HER3-DC1 (red)–, and OT-II-DC1 (green)–treated mice. G, Intratumoral infiltration of CD3+CD4+ and CD3+CD8+ T cells was compared between control (black) and HER3-DC1 (red) mice from E. H, CD4+ central memory (CD62L+CD44+ CM) and effector memory (CD62L−CD44+ EM) T-cell infiltration per milligram of tumors in control (black) versus HER3-DC1 (red) mice from E. I, Absolute number of CD3+CD4+ and CD3+CD8+ T cells in the lymph nodes of control-versus HER3-DC1–treated mice from E. J, Abundance of CD4+ CM, EM, and effector (Eff) T cells in the lymph nodes of control versus treated mice. For G–J, Unpulsed DC1 (blue) and OT-II-DC1 (green) groups were not included in the statistical analyses. K, Lymph node lymphocytes from control and treated mice (E) were cocultured with individual HER3 peptide–pulsed DC1 for 72 hours to detect IFN-γ by ELISA. Responses were compared between control and HER3-DC1 groups for HER3 peptides, and control versus OT-II-DC1 groups to OT-II peptide–pulsed DCs. Data shown are the representative from three independent experiments and are represented as mean SEM with statistical significance determined using multiple t test without correction for multiple comparisons. Each row was analyzed individually, without assuming a consistent SD. A log-rank (Mantel–Cox) test was used to determine differences between the survival curves.*, P ≤ 0.05; **, P ≤ 0.01; **, P ≤ 0.001; ***, P ≤ 0.0001; ####, P ≤ 0.0001. Table 1. Prediction of binding affinity from algorithms for identified HER3 peptides. Top, Class II–binding affinity predictions (nmol/L) from NetMHCIIpan 4.0 algorithm for nine identified HER3 peptides. Bottom, Class II–binding affinity predictions from MixMHCIIpred algorithm for nine identified HER3 peptides. The “BestAllele” column shows specific HLA allele(s) that had the best binding prediction. The “%Rank_best (12–25aa)” column shows the percentile rank among all randomly generated peptide sizes 12–25 amino acid; best score = 0, worst score = 100. The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734. Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/). References 1. Haikala HM , Janne PA . Thirty years of HER3: from basic biology to therapeutic interventions. Clin Cancer Res 2021;27 :3528–39.33608318 2. Ocana A , Vera-Badillo F , Seruga B , Templeton A , Pandiella A , Amir E . HER3 overexpression and survival in solid tumors: a meta-analysis. J Natl Cancer Inst 2013;105 :266–73.23221996 3. Gala K , Chandarlapaty S . Molecular pathways: HER3 targeted therapy. Clin Cancer Res 2014;20 :1410–6.24520092 4. Pinkas-Kramarski R , Soussan L , Waterman H , Levkowitz G , Alroy I , Klapper L , Diversification of Neu differentiation factor and epidermal growth factor signaling by combinatorial receptor interactions. EMBO J 1996;15 : 2452–67.8665853 5. Abel EV , Basile KJ , Kugel CH 3rd , Witkiewicz AK , Le K , Amaravadi RK , Melanoma adapts to RAF/MEK inhibitors through FOXD3-mediated upregulation of ERBB3. J Clin Invest 2013;123 :2155–68.23543055 6. Zhang Z , Karthaus WR , Lee YS , Gao VR , Wu C , Russo JW , Tumor microenvironment-derived NRG1 promotes antiandrogen resistance in prostate cancer. Cancer Cell 2020;38 :279–96.32679108 7. Bezler M , Hengstler JG , Ullrich A . Inhibition of doxorubicin-induced HER3–PI3K–AKT signalling enhances apoptosis of ovarian cancer cells. Mol Oncol 2012;6 :516–29.22841590 8. Recondo G , Bahcall M , Spurr LF , Che J , Ricciuti B , Leonardi GC , Molecular mechanisms of acquired resistance to MET tyrosine kinase inhibitors in patients with MET exon 14-mutant NSCLC. Clin Cancer Res 2020;26 :2615–25.32034073 9. Lee Y , Ma J , Lyu H , Huang J , Kim A , Liu B . Role of erbB3 receptors in cancer therapeutic resistance. Acta Biochim Biophys Sin 2014;46 :190–8.24449784 10. Bae SY , Choi YL , Kim S , Kim M , Kim J , Jung SP , HER3 status by immunohistochemistry is correlated with poor prognosis in hormone receptor-negative breast cancer patients. Breast Cancer Res Treat 2013;139 :741–50.23722313 11. Ogden A , Bhattarai S , Sahoo B , Mongan NP , Alsaleem M , Green AR , Combined HER3-EGFR score in triple-negative breast cancer provides prognostic and predictive significance superior to individual biomarkers. Sci Rep 2020;10 :3009.32080212 12. Tao JJ , Castel P , Radosevic-Robin N , Elkabets M , Auricchio N , Aceto N , Antagonism of EGFR and HER3 enhances the response to inhibitors of the PI3K–Akt pathway in triple-negative breast cancer. Sci Signal 2014;7 :ra29. 13. Laidlaw BJ , Craft JE , Kaech SM . The multifaceted role of CD4(+) T cells in CD8 (+) T cell memory. Nat Rev Immunol 2016;16 :102–11.26781939 14. Basu A , Ramamoorthi G , Albert G , Gallen C , Beyer A , Snyder C , Differentiation and regulation of TH cells: a balancing act for cancer immunotherapy. Front Immunol 2021;12 :669474.34012451 15. Novy P , Quigley M , Huang X , Yang Y . CD4 T cells are required for CD8 T cell survival during both primary and memory recall responses. J Immunol 2007;179 : 8243–51.18056368 16. Bos R , Sherman LA . CD4+ T-cell help in the tumor milieu is required for recruitment and cytolytic function of CD8+ T lymphocytes. Cancer Res 2010;70 : 8368–77.20940398 17. Shedlock DJ , Shen H . Requirement for CD4 T cell help in generating functional CD8 T cell memory. Science 2003;300 :337–9.12690201 18. Kumai T , Ohkuri T , Nagato T , Matsuda Y , Oikawa K , Aoki N , Targeting HER-3 to elicit antitumor helper T cells against head and neck squamous cell carcinoma. Sci Rep 2015;5 :16280.26538233 19. Lowenfeld L , Mick R , Datta J , Xu S , Fitzpatrick E , Fisher CS , Dendritic cell vaccination enhances immune responses and induces regression of HER2(pos) DCIS independent of route: results of randomized selection design trial. Clin Cancer Res 2017;23 :2961–71.27965306 20. Fracol M , Datta J , Lowenfeld L , Xu S , Zhang PJ , Fisher CS , Loss of anti-HER-3 CD4+ T-helper type 1 immunity occurs in breast tumorigenesis and is negatively associated with outcomes. Ann Surg Oncol 2017;24 :407–17.27663569 21. Datta J , Rosemblit C , Berk E , Showalter L , Namjoshi P , Mick R , Progressive loss of anti-HER2 CD4(+) T-helper type 1 response in breast tumorigenesis and the potential for immune restoration. Oncoimmunology 2015;4 :e1022301.26451293 22. Reynisson B , Barra C , Kaabinejadian S , Hildebrand WH , Peters B , Nielsen M . Improved prediction of MHC II antigen presentation through integration and motif deconvolution of mass spectrometry MHC eluted ligand data. J Proteome Res 2020;19 :2304–15.32308001 23. Vita R , Overton JA , Greenbaum JA , Ponomarenko J , Clark JD , Cantrell JR , The immune epitope database (IEDB) 3.0. Nucleic Acids Res 2015;43 : D405–12.25300482 24. Racle J , Michaux J , Rockinger GA , Arnaud M , Bobisse S , Chong C , Robust prediction of HLA class II epitopes by deep motif deconvolution of immunopeptidomes. Nat Biotechnol 2019;37 :1283–6.31611696 25. Moore TV , Nishimura MI . Improved MHC II epitope prediction—a step towards personalized medicine. Nat Rev Clin Oncol 2020;17 :71–2.31836878 26. Rovero S , Amici A , Di Carlo E , Bei R , Nanni P , Quaglino E , DNA vaccination against rat her-2/Neu p185 more effectively inhibits carcinogenesis than transplantable carcinomas in transgenic BALB/c mice. J Immunol 2000; 165 :5133–42.11046045 27. Falo LD , Kovacsovicsbankowski M , Thompson K , Rock KL . Targeting antigen into the phagocytic pathway in-vivo induces protective tumor-immunity. Nat Med 1995;1 :649–53.7585145 28. Kodumudi KN , Weber A , Sarnaik AA , Pilon-Thomas S . Blockade of myeloid-derived suppressor cells after induction of lymphopenia improves adoptive T cell therapy in a murine model of melanoma. J Immunol 2012;189 :5147.23100512 29. Cintolo JA , Datta J , Xu S , Gupta M , Somasundaram R , Czerniecki BJ . Type I-polarized BRAF-pulsed dendritic cells induce antigen-specific CD8+ T cells that impact BRAF-mutant murine melanoma. Melanoma Res 2016;26 :1–11.26451873 30. Mishra R , Patel H , Alanazi S , Yuan L , Garrett JT . HER3 signaling and targeted therapy in cancer. Oncol Rev 2018;12 :355.30057690 31. Majumder A , Sandhu M , Banerji D , Steri V , Olshen A , Moasser MM . The role of HER2 and HER3 in HER2-amplified cancers beyond breast cancers. Sci Rep 2021;11 :9091.33907275 32. Pulaski BA , Ostrand-Rosenberg S . Mouse 4T1 breast tumor model. Curr Protoc Immunol 2001;Chapter 20 :Unit 20.2 . 33. Yang L , Li Y , Shen E , Cao F , Li L , Li X , NRG1-dependent activation of HER3 induces primary resistance to trastuzumab in HER2-overexpressing breast cancer cells. Int J Oncol 2017;51 :1553–62.29048656 34. Sergina NV , Rausch M , Wang D , Blair J , Hann B , Shokat KM , Escape from HER-family tyrosine kinase inhibitor therapy by the kinase-inactive HER3. Nature 2007;445 :437–41.17206155 35. Jia Y , Kodumudi KN , Ramamoorthi G , Basu A , Snyder C , Wiener D , Th1 cytokine interferon gamma improves response in HER2 breast cancer by modulating the ubiquitin proteasomal pathway. Mol Ther 2021;29 :1541–56.33412308 36. Liu H , Weber A , Morse J , Kodumudi K , Scott E , Mullinax J , T cell mediated immunity after combination therapy with intralesional PV-10 and blockade of the PD-1/PD-L1 pathway in a murine melanoma model. PLoS One 2018;13 : e0196033.29694419 37. Sinevici N , Ataeinia B , Zehnder V , Lin K , Grove L , Heidari P , HER3 differentiates basal from claudin type triple negative breast cancer and contributes to drug and microenvironmental induced resistance. Front Oncol 2020;10 :554704.33330026 38. Chakrabarty A , Sanchez V , Kuba MG , Rinehart C , Arteaga CL . Feedback upregulation of HER3 (ErbB3) expression and activity attenuates antitumor effect of PI3K inhibitors. Proc Natl Acad Sci U S A 2012;109 :2718–23.21368164 39. Castro F , Cardoso AP , Goncalves RM , Serre K , Oliveira MJ . Interferon-gamma at the crossroads of tumor immune surveillance or evasion. Front Immunol 2018;9 :847.29780381 40. Mazor R , Tai CH , Lee B , Pastan I . Poor correlation between T-cell activation assays and HLA-DR binding prediction algorithms in an immunogenic fragment of Pseudomonas exotoxin A. J Immunol Methods 2015;425 :10–20.26056938 41. Dhanda SK , Karosiene E , Edwards L , Grifoni A , Paul S , Andreatta M , Predicting HLA CD4 immunogenicity in human populations. Front Immunol 2018;9 :1369.29963059 42. Soria-Guerra RE , Nieto-Gomez R , Govea-Alonso DO , Rosales-Mendoza S . An overview of bioinformatics tools for epitope prediction: implications on vaccine development. J Biomed Inform 2015;53 :405–14.25464113 43. Wang P , Sidney J , Dow C , Mothe B , Sette A , Peters B . A systematic assessment of MHC class II peptide binding predictions and evaluation of a consensus approach. PLoS Comput Biol 2008;4 :e1000048.18389056 44. Jurtz V , Paul S , Andreatta M , Marcatili P , Peters B , Nielsen M . NetMHCpan-4.0: improved peptide-MHC class I interaction predictions integrating eluted ligand and peptide binding affinity data. J Immunol 2017;199 :3360–8.28978689 45. Chaves FA , Lee AH , Nayak JL , Richards KA , Sant AJ . The utility and limitations of current Web-available algorithms to predict peptides recognized by CD4 T cells in response to pathogen infection. J Immunol 2012;188 :4235–48.22467652 46. Andreatta M , Trolle T , Yan Z , Greenbaum JA , Peters B , Nielsen M . An automated benchmarking platform for MHC class II binding prediction methods. Bioinformatics 2018;34 :1522–8.29281002 47. Kropshofer H , Singer T . Overview of cell-based tools for preclinical assessment of immunogenicity of biotherapeutics. J Immunotoxicol 2006;3 :131–6.18958693 48. Nielsen M , Lund O , Buus S , Lundegaard C . MHC class II epitope predictive algorithms. Immunology 2010;130 :319–28.20408898 49. Zhang L , Udaka K , Mamitsuka H , Zhu S . Toward more accurate pan-specific MHC-peptide binding prediction: a review of current methods and tools. Brief Bioinform 2012;13 :350–64.21949215

PMC009xxxxxx/PMC9444141.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9802571 20730 Mol Cell Mol Cell Molecular cell 1097-2765 1097-4164 35654044 9444141 10.1016/j.molcel.2022.05.008 NIHMS1827300 Article A pro-metastatic tRNA fragment drives Nucleolin oligomerization and stabilization of its bound metabolic mRNAs Liu Xuhang 1 Mei Wenbin 1 Padmanaban Veena 1 Alwaseem Hanan 2 Molina Henrik 2 Passarelli Maria C. 1 Tavora Bernardo 1 Tavazoie Sohail F. 145 1 Laboratory of Systems Cancer Biology, The Rockefeller University, New York, NY 10065, USA. 2 Proteomics Resource Center, The Rockefeller University, New York, NY 10065, USA. 4 Lead contact AUTHOR CONTRIBUTIONS X.L. and S.F.T conceived and designed the project and supervised all experiments. X.L. designed and performed experiments. X.L. and W.M. conducted bioinformatic and statistical analyses. X.L. and H.A. performed metabolite profiling. X.L. and H.M. performed proteomic analysis. X.L., V.P., and B.T. performed mouse experiments. X.L. and S.F.T. wrote the manuscript with input from all authors. 5 Correspondence: sohail.tavazoie@rockefeller.edu 19 8 2022 21 7 2022 01 6 2022 21 7 2023 82 14 26042617.e8 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY Stress-induced cleavage of transfer RNAs (tRNAs) into tRNA-derived fragments (tRFs) occurs across organisms from yeast to humans, yet its mechanistic underpinnings and pathological consequences remain poorly defined. Small RNA profiling revealed increased abundance of a Cysteine tRNA fragment (5’-tRFCys) during breast cancer metastatic progression. 5’-tRFCys was required for efficient breast cancer metastatic lung colonization and cancer cell survival. We identified Nucleolin as the direct binding partner of 5’-tRFCys. 5’-tRFCys promoted the oligomerization of Nucleolin and its bound metabolic transcripts Mthfd1l and Pafah1b1 into a higher-order transcript stabilizing ribonucleoprotein complex, which protected these transcripts from exonucleolytic degradation. Consistent with this, Mthfd1l and Pafah1b1 mediated pro-metastatic and metabolic effects downstream of 5’-tRFCys—impacting folate, one-carbon and phosphatidylcholine metabolism. Our findings reveal that a tRF can promote oligomerization of an RNA binding protein into a transcript stabilizing ribonucleoprotein complex, thereby driving specific metabolic pathways underlying cancer progression. tRNA fragment tRF Nucleolin breast cancer metastasis post-transcriptional transcript stability oligomerization Pafah1b1 Mthfd1l pmcINTRODUCTION Transfer RNA-derived RNA fragments (tRFs) are a novel class of small non-coding RNAs cleaved from mature or precursor transfer RNAs (tRNAs). TRFs are classified into distinct groups based on their sites of origin within tRNAs. 5’- and 3’-tRNA halves arise from cleavage within the anticodon loop by Angiogenin, RNase L, or RNase 1 (Donovan et al., 2017; Fu et al., 2009; Nechooshtan et al., 2020; Yamasaki et al., 2009). Additionally, 5’- and 3’-tRF-18 are 18-nt fragments generated by Dicer (Babiarz et al., 2008; Cole et al., 2009), while 3’U tRFs are cleaved by RNase Z from the 3’ end of precursor tRNAs (Lee et al., 2009). TRFs are important regulators of a number of molecular and cellular processes (Oberbauer and Schaefer, 2018; Schimmel, 2018). For example, 5’-tRFs are transmitted via sperm to mouse zygotes and alter embryonic gene expression (Chen et al., 2016; Sharma et al., 2016). 3’-tRFLys inhibits transposition of retroviruses by serving as a decoy for tRNALys (Schorn et al., 2017). 3’-tRFAla enhances translation by upregulating the expression of a key ribosomal protein gene (Kim et al., 2017). TRFs were also implicated in cancer progression and cell proliferation (Goodarzi et al., 2015; Honda et al., 2015; Shao et al., 2017). For example, in response to hypoxia, a subset of tRFs containing a common motif were found to become upregulated in cancer cells and bound the RNA-binding protein Y-box Binding Protein 1 (YBX1)—competitively inhibiting YBX1 binding to its growth-promoting target transcripts and consequently suppressing cancer progression (Goodarzi et al., 2015). Despite their demonstrated roles in various cellular processes, the molecular mechanisms underlying tRF function remain poorly defined. Nucleolin is an evolutionarily conserved RNA binding protein that plays crucial roles in multiple molecular processes (Mongelard and Bouvet, 2007). Nucleolin is essential for ribosomal biogenesis, ribosomal RNA (rRNA) transcription and processing, and assembly of ribosomes (Ginisty et al., 1998; Ginisty et al., 1999; Serin et al., 1996). Nucleolin also regulates multiple steps in messenger RNA (mRNA) processing, including splicing, stabilization, nucleus-cytoplasmic transport, and translation (Abdelmohsen and Gorospe, 2012). Recent studies have further implicated Nucleolin in the synthesis of microRNAs (Pichiorri et al., 2013; Pickering et al., 2011). In keeping with its pleiotropic roles in post-transcriptional control, Nucleolin is essential for the proliferation and survival of mammalian cells (Ugrinova et al., 2007) and self-renewal of embryonic stem cells (Percharde et al., 2018). Despite decades of research, the full spectrum of RNAs directly bound to Nucleolin has remained unclear. During cancer progression, there is positive selection for cancer cells that overcome multiple restrictive physiological barriers within the primary and metastatic microenvironments. These barriers are bypassed when rare cells modulate gene expression programs to enable multiple adaptive phenotypes (Hanahan and Weinberg, 2011). A critical mechanism by which such gene expression states are achieved is through alterations in the expression levels of specific small RNAs, which enables pro-metastatic post-transcriptional gene expression responses (Loo et al., 2015; Mendell and Olson, 2012; Nabet et al., 2017; Pencheva et al., 2012; Truitt and Ruggero, 2017). We hypothesized that certain tRFs might become induced during metastatic progression. We reasoned that by studying how such tRFs shape gene expression programs, we might uncover insights into their mechanisms of action. We herein report that a tRNA fragment, 5’-tRFCys, derived from the 5’ half of Cysteine tRNA is upregulated in multiple highly metastatic breast cancer cell lines. We find that increased expression of 5’-tRFCys associates with worse clinical outcomes of breast cancer patients and is required for efficient breast cancer metastatic colonization and survival. We find that this fragment directly binds to Nucleolin, enhances the interaction of Nucleolin with specific target transcripts, and promotes the formation of an oligomeric Nucleolin/transcript complex—collectively stabilizing the Nucleolin-bound transcripts. 5’-tRFCys and Nucleolin promote breast cancer metastasis by stabilizing transcripts encoding Pafah1b1 and Mthfd1l, key metabolic enzymes. Moreover, administration of an antisense oligonucleotide targeting 5’-tRFCys suppressed metastatic colonization. Our results uncover a 5’-tRFCys/Nucleolin axis as a driver of breast cancer metastatic progression and demonstrate that a tRF can enhance oligomerization of an RNA binding protein into a stabilizing higher-order ribonucleoprotein complex that regulates the metabolic state of the cell. RESULTS 5’-tRFCys is upregulated during breast cancer progression and metastasis To search for tRFs that may regulate breast cancer metastasis, we performed next-generation sequencing of small RNAs isolated from three isogenic mouse breast cancer cell lines with distinct metastatic capacities— highly metastatic 4T1, poorly metastatic 4TO7, and non-metastatic 67NR cells (Aslakson and Miller, 1992; Dexter et al., 1978). Hierarchical clustering based on sample-to-sample distance revealed that the expression levels of small RNAs including tRFs and microRNAs were sufficiently informative to classify samples into their respective highly, poorly, and non-metastatic groups (Figure S1A). Among different classes of tRFs, the most abundant species were 5’-tRNA halves (Figure S1B). We found 5’-tRFCys, derived from the 5’ half of multiple tRNACysGCA isodecoders, to be one of the most upregulated tRNA halves in highly metastatic 4T1 cells relative to isogenic poorly metastatic 4TO7 cells (Figure 1A). Quantitative reverse transcription PCR (RT-qPCR) using a 5’-tRFCys specific stem-loop primer independently confirmed that expression of 5’-tRFCys was significantly elevated in 4T1 relative to 4TO7 or 67NR cells (Figure 1B). A similar increase in 5’-tRFCys expression was observed in an independent in vivo selected EO771-LM3 highly metastatic cell line relative to its parental EO771-Par poorly metastatic isogenic line (Figure S1C). These results reveal that 5’-tRFCys is upregulated in breast cancer cells of high metastatic potential. Analysis of small RNA sequencing data from the Cancer Genome Atlas (TCGA) (Cancer Genome Atlas, 2012; Pliatsika et al., 2018) revealed that high expression of 5’-tRFCys correlated with poor survival of breast cancer patients (Figure 1C; P=0.011; n=1066). Importantly, 5’-tRFCys expression remained significantly correlated with poor human breast cancer survival when common co-variates such as age at diagnosis, clinical stage, subtype, the status of hormone receptors, and HER2 status were taken into account in multivariable Cox analyses (Figures S1D and S1E; p=0.041 and p=0.04; n=978 and n=671 respectively). Additionally, 5’-tRFCys expression was found to be significantly elevated in breast cancer samples relative to matched normal breast tissues in the TCGA cohort (Figure 1D; p=1.82e-11; n=101). These results reveal that 5’-tRFCys expression is increased during breast cancer progression. 5’-tRFCys is required for efficient metastatic lung colonization of breast cancer cells To test whether 5’-tRFCys plays a functional role in breast cancer metastasis, we inhibited the activity of 5’-tRFCys using two distinct antisense locked nucleic acid (LNA) oligonucleotides (Figure S2A). Transfection of either antisense LNA oligonucleotide into either of two highly metastatic mouse breast cancer cell lines (4T1 and EO771-LM3) significantly reduced metastasis in tail-vein colonization assays (Figures 2A, 2B and S2B). Additionally, inhibiting 5’-tRFCys by transducing an antisense tough-decoy molecule (Haraguchi et al., 2009; Xie et al., 2012) into the highly metastatic MDA-MB-231-LM2 human breast cancer cell line or a human breast cancer patient-derived xenograft organoid (PDXO) line significantly suppressed metastatic colonization in immunocompromised NOD-scid gamma (NSG) mice (Figures S2C and S2D). Furthermore, inhibiting 5’-tRFCys significantly reduced primary tumor growth when breast cancer cells were implanted in the mammary fat pads of mice (Figure 2C). Lastly, bi-weekly injections of a 5’-tRFCys antisense LNA oligo into mice after the establishment of breast cancer micro-metastases in the lung significantly suppressed metastatic colonization (Figure 2D). Collectively, these results reveal that 5’-tRFCys plays a critical role in breast tumor growth and metastatic colonization. 5’-tRFCys enhances breast cancer cell survival We next studied the cellular mechanism by which 5’-tRFCys promotes metastasis. Immunostaining of metastatic lung sections revealed no significant difference in the abundance of Ki-67 or Endomucin positive cells (Figures S2E and S2F) upon 5’-tRFCys inhibition, suggesting that neither proliferation nor angiogenesis was impacted. In contrast, we observed a significant increase in the number of cleaved Caspase-3 positive foci in metastatic lung nodules (Figure 2E) and in the fraction of apoptotic cells in primary mammary tumors upon 5’-tRFCys inhibition (Figure 2F). Consistent with this finding, 5’-tRFCys inhibition significantly increased Caspase3/7 activity in 4T1 and EO771-LM3 metastatic cells and in two human breast cancer PDXO lines in vitro (Figures 2G, 2H, S2G, and S2H). These findings reveal that 5’-tRFCys promotes primary tumor growth and metastatic colonization by enhancing the survival of breast cancer cells. Nucleolin directly binds 5’-tRFCys Small non-coding RNAs generally mediate their effects through interactions with RNA-binding proteins (RBPs). We searched for protein(s) that directly interact with 5’-tRFCys using a biochemical approach. We UV irradiated cells in order to crosslink proteins to RNA substrates and used a biotinylated 5’-tRFCys antisense oligonucleotide to pull down 5’-tRFCys-interacting partners from 4T1 cell lysates. Mass spectrometry (MS) identified Nucleolin as the most enriched protein in the 5’-tRFCys antisense pulldown relative to the antisense control (Figure 3A). Notably, Nucleolin was the only protein identified by MS to be reproducibly enriched in the 5’-tRFCys antisense pulldowns from three different mouse and human breast cancer cell lines (Figure 3B). In reciprocal pulldown experiments using an anti-Nucleolin antibody, we observed a 250–500-fold enrichment of 5’-tRFCys from Nucleolin immunoprecipitates (IP) relative to the control IP in both mouse and human breast cancer cells (Figure 3C). These experiments demonstrate that Nucleolin is an evolutionarily conserved binding partner of 5’-tRFCys. To test whether Nucleolin directly binds 5’-tRFCys, we performed high-throughput sequencing of RNA isolated by crosslinking immunoprecipitation (HITS-CLIP) (Moore et al., 2014) in 4T1 cells. Notably, unlike canonical HITS-CLIP, exogenous ribonuclease (RNase) was not used in this experiment in order to preserve the natural ends of small RNAs. Prior studies have shown that the fraction of reads with crosslinking induced modification sites (CIMS) correlates with the binding affinity of an RBP for its substrate RNA . Bioinformatic analyses revealed that 5’-tRNA halves were the most enriched type of tRFs bound to Nucleolin (Figure S3A). Notably, several of the most enriched 5’-tRNA halves were derived from tRNACys loci (Figure 3D). These results reveal a high binding affinity for Nucleolin towards 5’-tRFCys, implicating Nucleolin as a direct binding partner of 5’-tRFCys. Motif analysis of sequences surrounding CIMS sites revealed the enrichment of two motifs (Figure 3E). The first C-rich motif is highly similar to the Nucleolin recognition element, which was previously identified by in vitro biochemical assays and plays a critical role in mediating Nucleolin’s binding to the 5’ external transcribed spacers (ETS) of pre-45S rRNA transcripts (Ghisolfi-Nieto et al., 1996). The second G-rich motif has not been previously reported. 5’-tRFCys contains a CIMS site right next to this motif (Figure 3F), suggesting that this motif may mediate Nucleolin binding to 5’-tRFCys. Interestingly, we noted another highly similar G-rich motif at the 5’ end of 5’-tRFCys. To test whether these motifs are required for Nucleolin binding to 5’-tRFCys, we mutated each individually or in combination. Electrophoretic mobility shift assays (EMSA) with purified Nucleolin protein (Figure S3B) revealed that mutating either G-rich motif of 5’-tRFCys dramatically reduced Nucleolin’s binding (Figure 3G). Notably, both motifs are highly conserved among almost all human and mouse 5’-tRFCys isomers (Figure S3C). Collectively, these results reveal a critical role for these conserved G-rich motifs in mediating Nucleolin’s binding to 5’-tRFCys. 5’-tRFCys facilitates recruitment of Nucleolin to target transcripts Given the well-established role of Nucleolin as an essential protein in ribosomal RNA (rRNA) biogenesis (Srivastava and Pollard, 1999), we next examined whether the binding of 5’-tRFCys to Nucleolin impacts translation. We observed that neither rRNA synthesis nor global translation was impaired upon 5’-tRFCys inhibition (Figures S4A and S4B). Considering that Nucleolin has also been previously implicated in regulating mRNA stability and translation (Abdelmohsen et al., 2011; Sengupta et al., 2004; Takagi et al., 2005; Zaidi and Malter, 1995), we postulated that 5’-tRFCys might regulate Nucleolin binding to specific mRNAs. To test this hypothesis, we performed HITS-CLIP of Nucleolin in cells transfected with either scrambled control or 5’-tRFCys antisense LNA oligonucleotide. Surprisingly, in addition to the well-established targets such as rRNAs and snoRNAs, we found that Nucleolin bound a broad spectrum of mRNA transcripts (Figure S4C), with most binding peaks residing within 5’ untranslated regions (UTRs) (Figure S4D). Interestingly, Nucleolin binding to a subset of peaks was significantly reduced upon 5’-tRFCys inhibition (Figure 4A; p<2.2e-16), suggesting that 5’-tRFCys promoted Nucleolin’s binding to a subset of transcripts. To examine the impact of depleting 5’-tRFCys on its target transcripts, we quantified proteome changes by mass spectrometry, focusing on genes whose transcripts exhibited enhanced 5’-tRFCys-mediated Nucleolin binding. The protein levels of a subset of these genes were significantly reduced upon 5’-tRFCys inhibition (Figure 4B; p=7.4e-3). These results suggest that 5’-tRFCys promotes Nucleolin binding to 5’-UTRs of target transcripts, thereby enhancing gene expression. In line with this hypothesis, we also found that Nucleolin-bound transcripts were significantly downregulated upon Nucleolin depletion (Figure S4E; p=4.8e-13), suggesting that Nucleolin binding stabilizes its bound target transcripts. To identify targets regulated by 5’-tRFCys/ Nucleolin binding, we examined transcriptome and translatome changes upon 5’-tRFCys inhibition with two distinct 5’-tRFCys antisense LNAs by RNA Sequencing (RNA-Seq) and Ribosome Profiling (Ribo-Seq), respectively. Inhibition of 5’-tRFCys by the two distinct 5’-tRFCys antisense LNAs led to highly similar gene expression changes (ρ = 0.717; p< 2.2e-16) (Figure S4F). We found that both protein abundance and the abundance of ribosome-protected fragments correlated significantly with their transcript levels (Figures 4C and S4G), suggesting that most Nucleolin-bound transcripts were regulated at the post-transcriptional rather than translational level. By searching for genes whose Nucleolin binding was enhanced by 5’-tRFCys and whose expression levels were downregulated in 5’-tRFCys inhibited cells, we identified two of the most downregulated Nucleolin-bound transcripts—Pafah1b1 and Mthfd1l— as candidate downstream effectors (Figure 4C). Platelet-activating factor acetylhydrolase 1 beta 1 (Pafah1b1) is a regulatory subunit of platelet-activating factor acetylhydrolase 1 beta (Karasawa and Inoue, 2015), whereas Methylenetetrahydrofolate Dehydrogenase 1 Like (Mthfd1l) is a metabolic enzyme that plays a critical role in one-carbon metabolism (Ducker and Rabinowitz, 2017; Yang and Vousden, 2016). We found that Pafah1b1 and Mthfd1l became downregulated upon inhibition of 5’-tRFCys (Figures 4D, S4H, and S4K) or depletion of Nucleolin (Figures 4E and S4I) in both mouse and human breast cancer cells. Consistent with results from the aforementioned genome-wide analyses, Nucleolin predominantly bound Pafah1b1 and Mthfd1l in the 5’ UTRs of these transcripts in a 5’-tRFCys-dependent manner (Figure 4F and 4G). RT-qPCR assays revealed that 5’-tRFCys also enhanced Nucleolin binding to both transcripts in human breast cancer cells (Figure 4H). RNA Sequencing and RT-qPCR revealed that both transcripts became reduced upon 5’-tRFCys inhibition (Figures 4F, 4G and 4I). Ribo-Seq analyses revealed a similar reduction in the number of ribosome-protected fragments (Figures 4F and 4G), suggesting that 5’-tRFCys-mediated regulation of these two target candidates occurs at the post-transcriptional rather than the translational level. Nucleolin binding stabilizes its target transcripts To test whether Nucleolin directly regulates target transcript stability, we performed tethering experiments. We placed increasing copies of boxB hairpins in the 5’ UTR of a luciferase reporter and fused Nucleolin to the λN peptide (Figure S4J), which specifically binds the boxB hairpin (De Gregorio et al., 1999; Lazinski et al., 1989). Consistent with our hypothesis, tethering Nucleolin to the 5’UTR of a luciferase reporter transcript significantly increased the intracellular abundance of that transcript in a boxB and λN peptide-dependent manner (Figure 4K), revealing that Nucleolin binding to a transcript can increase its stability. Moreover, Nucleolin tethering also significantly increased reporter gene expression even in the presence of multiple copies of boxB hairpins, whose strong secondary structures are known to impede translational initiation in the 5’ UTR and reduce reporter expression in a copy-number dependent manner (Figure 4L) (Lytle et al., 2007). Together, these results reveal that Nucleolin binding to the 5’ UTR of target transcripts enhances transcript stability. We hypothesized that 5’-tRFCys might facilitate the recruitment of Nucleolin to its target transcripts, thereby enhancing transcript stability. To test this hypothesis, we performed RNA stability assays by quantifying transcript abundance upon 5’-tRFCys inhibition in 4T1 and MDA-MB-231-LM2 cells treated with the RNA Polymerase II inhibitor—5,6-dichloro-1-beta-D-ribofuranosylbenzimidazole (DRB). Inhibition of 5’-tRFCys significantly reduced the stability of both Pafah1b1 and Mthfd1l in both mouse and human breast cancer cells (Figures 4J and S4M). To provide further evidence for the direct regulation of transcript stability by 5’-tRFCys, we transfected in vitro transcribed luciferase reporter transcripts that contained either Pafah1b1 or Mthfd1l 5’ UTRs. Dual-luciferase assays revealed a significant reduction in the luminescence signal upon 5’-tRFCys inhibition (Figure S4N). Moreover, analyses of the TCGA dataset revealed that Nucleolin and 5’-tRFCys expression significantly correlated with the aggregate expression of Mthfd1l and Pafah1b1 in breast cancer patients, respectively (Figures S4O and S4P). These data support a model in which 5’-tRFCys enhances the stability of its target transcripts by promoting Nucleolin binding to their 5’ UTRs. 5’-tRFCys promotes oligomerization of Nucleolin and its pro-metastatic transcripts Given that Nucleolin directly binds both 5’-tRFCys and its target transcripts in vivo, we next asked how 5’-tRFCys regulates the interaction of Nucleolin with its target transcripts. A prior study reported that Nucleolin self-interacts (Chen et al., 2012). Indeed, anti-FLAG IP from cells transfected with HA-tagged Nucleolin with or without FLAG-tagged Nucleolin confirmed that Nucleolin self-interacts in vivo (Figure 5A). Interestingly, this self-interaction is RNA-dependent since the interaction was abolished by pre-treating lysates with RNase A before immunoprecipitation (Figure 5A). Given these findings, we hypothesized that 5’-tRFCys mediates the assembly of Nucleolin into a higher-order complex with its target transcripts. To test this hypothesis, we established an in-vitro Nucleolin complex assembly assay using 5’-radiolabeled 5’-tRFCys and/or Nucleolin’s target transcripts, with purified Nucleolin protein. We then resolved the reaction products by performing native polyacrylamide gel electrophoresis (PAGE) and detected the assembled complexes by autoradiography. In the presence of Mg2+, Nucleolin first formed two low-molecular-weight complexes with Pafab1b1 mRNA and 5’-tRFCys, which we referred to as complex Am and At, respectively (Figures S5A and S5B), as both complexes migrated to similar positions and their assembly was independent of either Mg2+ or incubation at an elevated temperature. Surprisingly, as the amount of Nucleolin protein was increased, complexes Am and At gradually disappeared while a higher molecular weight complex D emerged. This suggested that complexes Am and At represent precursors of complex D (Figures S5A and S5B). Quantitative analysis of complex D assembly revealed the Hill coefficient, a measure of cooperative binding, to be 1.4 and 3.5 for 5’-tRFCys and Pafab1b1, respectively (Figure S5C). This result suggests cooperativity in complex D assembly from complex Am for Pafah1b1. We calculated maximum specific binding (Bmax) of complex D to be only 0.1 (Figure S5C), suggesting that the formation of complex D is inefficient under the experimental condition tested. After extensive optimization, we found that complex D assembly could be substantially enhanced by using Nucleolin immunoprecipitates (IP) instead of purified Nucleolin protein, suggesting the presence of one or more unidentified factors that facilitate complex D formation. We also found that complex D assembly required Mg2+ and incubation at an elevated temperature because its assembly was suppressed by EDTA or incubation at 4 °C (Figures 5B-5D). We hypothesized that complex Am/At and D represent RNA-Nucleolin monomers and oligomers, respectively. To test this hypothesis, we performed chemical crosslinking and western blotting with an anti-Nucleolin antibody after the assembly assay. Consistent with this hypothesis, we detected only monomers under experimental conditions permissive for complex Am/At assembly (i.e., in the presence of EDTA) (Figure 5E). On the contrary, under experimental conditions favorable for complex D assembly (i.e., in the presence of Mg2+ at 30 °C), we detected not only Nucleolin monomers, but also dimers, tetramers, and hexamers (Figure 5E). The addition of increasing amounts of micrococcal nuclease before the assembly assay gradually turned Nucleolin oligomers back into monomers (Figure S5D), supporting an essential role for RNA in promoting Nucleolin oligomerization. Kinetic analyses revealed that 5’-tRFCys assembled complex D much faster than Pafah1b1 (Figures 5F, S5E, and S5F). Furthermore, quantitative analyses showed that 5’-tRFCys exhibits a lower Kd than Pafah1b1 in assembling complex D (Figures 5G, S5G, and S5H). Importantly, the Hill coefficient for 5’-tRFCys and Pafab1b1 is 1.2 and 2.1, respectively (Figure 5G), suggesting Pafab1b1 exhibits much higher cooperativity than 5’-tRFCys in assembling complex D. Taken together, these results suggest that 5’-tRFCys may drive Nucleolin oligomerization with its target transcripts. To test this hypothesis, we added wild-type or Nucleolin-binding deficient mutant 5’-tRFCys, along with Pafah1b1 or Mthfd1l transcript to the assembly assay. Indeed, we found that the addition of only wild-type but not mutant 5’-tRFCys promoted complex D assembly (Figures 5H and S5I). Additionally, more Nucleolin tetramers and hexamers were detected when we supplemented 5’-tRFCys along with Pafah1b1 to the assembly assay than Pafah1b1 alone (Figures 5I and S5J). Overall, these results reveal that 5’-tRFCys facilitates the oligomerization of RNA-Nucleolin monomers. Nucleolin oligomerization protects pro-metastatic transcripts from exonucleolytic degradation Given that Nucleolin binds its target transcripts in either monomeric or oligomeric states, we next assessed the functional difference of these states. Considering that Nucleolin preferentially binds the 5’ UTR of its target transcripts, we hypothesized that oligomerization might protect its target transcripts from exonucleolytic degradation. As such, we tested the sensitivity of Nucleolin’s target transcripts to a 5’->3’ exonuclease under experimental conditions permissive for Nucleolin oligomerization. Remarkably, we found that Pafah1b1 and Mthfd1l target transcripts were protected by Nucleolin from degradation by the 5’->3’ exonuclease when the complex assembly assay was performed under experimental conditions favoring its oligomerization in comparison to conditions favoring only monomeric Nucleolin (Figure S5K). Such protection was also observed for two additional Nucleolin direct targets—Pkp4 and Scaf8 (Figure S5K). Thus, in comparison to monomeric Nucleolin, oligomeric Nucleolin better protects its target transcripts from exonucleolytic degradation. Pafah1b1 and Mthfd1l function downstream of 5’-tRFCys to promote breast cancer metastasis Having shown that the expression levels of both Pafah1b1 and Mthfd1l are enhanced by 5’-tRFCys, we next asked whether Pafah1b1 and Mthfd1l are downstream effectors of the pro-metastatic phenotype of 5’-tRFCys, given that both genes were found to promote the progression of multiple cancer types in prior studies (Agarwal et al., 2019; Lee et al., 2017; Lo et al., 2012; Locasale, 2013; Selcuklu et al., 2012; Zimdahl et al., 2014). Epistasis experiments revealed that overexpression of Pafah1b1 or Mthfd1l rescued survival defects caused by 5’-tRFCys inhibition in vivo, as quantified by cleaved Caspase 3/7 activity (Figures 6A and 6B). Additionally, overexpression of either Pafah1b1 or Mthfd1l rescued metastatic lung colonization defects upon 5’-tRFCys inhibition in both human and mouse breast cancer cells (Figures 6C, 6D, S6A, and S6B), consistent with these genes functioning downstream of 5’-tRFCys in promoting breast cancer metastasis. Together, these results reveal that Pafah1b1 and Mthfd1l enhance breast cancer cell survival and are downstream effectors of 5’-tRFCys-mediated metastatic colonization. Platelet-activating factor acetylhydrolase 1 beta 1 (Pafah1b1) is a regulatory subunit of the type I Platelet-activating factor acetylhydrolase complex (Arai et al., 2002). Given that platelet-activating factor is a lipid, we performed untargeted metabolomic lipid profiling by mass spectrometry to search for lipids that changed in abundance in a 5’-tRFCys- and Pafah1b1-dependent manner. Interestingly, we found that inhibition of 5’-tRFCys significantly reduced the abundance of multiple phosphatidylcholine (PC) species (Figure S6C), which are upstream of PAFAH1B1 in the PC synthesis pathway (Figure S6D). Importantly, this reduction in PC species was rescued by Pafah1b1 overexpression (Figure S6C). Conversely, CRISPR deletion of Pafah1b1 also significantly reduced intracellular PC levels (Figure S6E). PC was reported to become elevated in breast tumors relative to normal breast tissues and that PC levels positively correlate with poor breast cancer patient outcomes (Hilvo et al., 2011). Additionally, perturbing PC synthesis was observed to induce apoptosis in cancer cells (Ridgway, 2013). Collectively, these results suggest that Pafah1b1 may promote metastasis through the regulation of phosphatidylcholine synthesis. On the other hand, Mthfd1l is a key metabolic enzyme in the folate cycle (Figure S6F). We thus performed untargeted metabolite profiling to search for Mthfd1l downstream metabolite effectors that may promote metastasis. Consistent with our finding that Mthfd1l is the downstream effector of 5’-tRFCys and the critical role of Mthfd1l in one-carbon metabolism (Ducker et al., 2016; Ducker and Rabinowitz, 2017; Jain et al., 2012; Minton et al., 2018; Zheng et al., 2018), we identified several metabolites in the folate cycle (glycine) and interconnected metabolomic pathways (e.g., inosine monophosphate (IMP), carbamoyl-aspartate (CA), S-adenosylhomocysteine (SAH) and S-adenosylmethionine (SAM)), that became significantly increased in both control and Mthfd1l-rescued cells relative to the 5’-tRFCys inhibited cells (Figures 6E and S6F). We further found that supplementing both formate and glycine, two key metabolites of the folate cycle, partially rescued the cell survival defect in 5’-tRFCys-suppressed cells (Figure S6G). These results collectively reveal that 5’-tRFCys regulates one-carbon metabolism through Mthfd1l, thereby contributing to metastasis. Elevated expression of 5’-tRFCys targets correlates with poor breast cancer clinical outcomes We next sought to determine if the expression levels of 5’-tRFCys target transcripts correlated with clinical outcomes of human breast cancer patients. Breast cancer patients whose tumors expressed higher levels of Pafah1b1 and Mthfd1l exhibited significantly reduced survival relative to those whose tumors expressed lower levels of these genes (Figures 6F and 6G; P<1e-4 and P=2.2e-3, respectively; n=1097). Additionally, increased expression of a 5’-tRFCys target-gene signature significantly correlated with poor survival of breast cancer patients in both TCGA (Figure S6H; p=0.024; n=1007) and the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) datasets (Figure S6I; P=0.02; n=1671). Together, these findings provide human pathological association support for a role of this tRF/Pafah1b1/Mthfd11 axis in post-transcriptional control of metabolic response in driving breast cancer progression. DISCUSSION In vivo and in vitro experiments revealed that 5’-tRFCys promoted metastatic lung colonization of breast cancer cells by enhancing cancer cell survival. We found that two metabolic enzymes—Pafah1b1 and Mthfd1l—act downstream of 5’-tRFCys to enhance cancer cell survival and metastasis. Pafah1b1 and two catalytic enzymes, Pafah1b2 and Pafah1b3, form the type I Platelet-activating factor acetylhydrolase complex (Arai et al., 2002). Pafah1b1 is proposed to control the platelet-activating factor (PAF) levels by regulating the enzymatic activity of Pafah1b2 and Pafah1b3 (Arai et al., 2002). Consistent with our findings, prior studies reported that Pafah1b2 and Pafah1b3 promote cancer progression and metastasis (Ma et al., 2018; Mulvihill et al., 2014). Recent studies have revealed a critical role for Mthfd1l in promoting the progression of several cancer types (Agarwal et al., 2019; Eich et al., 2019; Lee et al., 2017; Selcuklu et al., 2012). Our findings are consistent with the emerging recognition of the impact of post-transcriptional regulation on metabolic control under pathological conditions (Loo et al., 2015; Sullivan et al., 2018; Zhu et al., 2011) and underscore the critical role that dysregulated metabolism plays in tumor biology (Cantor and Sabatini, 2012; DeBerardinis and Chandel, 2016). While 5’-tRFCys was reported to bind YBX-1 in a study using rabbit reticulocyte lysates (Ivanov et al., 2011), we found that Nucleolin is the primary in-vitro and in-vivo binding partner of 5’-tRFCys in both human and mouse breast cancer cells. HITS-CLIP experiments and EMSA assays revealed that Nucleolin directly binds 5’-tRFCys by engaging its two G-rich motifs. A prior study using Selection of Ligands by Exponential Enrichment (SELEX) proposed that Nucleolin binds RNAs through an evolutionarily conserved motif (Ginisty et al., 2000). Interestingly, upon close inspection of this previously reported group of selected sequences, we identified the enrichment of a highly similar G-rich motif to the one identified by our study. We speculate that the same G-rich motif identified herein to bind Nucleolin may contribute to Nucleolin binding to the SELEX-enriched sequences in that prior study. Although Nucleolin is best known for its role in rRNA biogenesis and ribosome assembly, we found that Nucleolin binds a previously unappreciated range of RNAs. Interestingly, most binding sites in mRNAs reside within 5’ UTRs, suggesting a direct role for Nucleolin in post-transcriptional regulation. In addition to the well-established role of RNA binding proteins in regulating the stability of transcripts via 3’ UTR-binding, recent studies have uncovered a similar role for RNA binding proteins at 5’ UTR sites (Arribere and Gilbert, 2013). Indeed, tethering of Nucleolin to the 5’ UTR increased the abundance of mRNA transcripts in vivo. Additionally, complex D, which represents an RNA-Nucleolin oligomer, better protected RNAs from exonucleolytic degradation than an RNA-Nucleolin monomer in vitro. Together, these results are consistent with a model whereby Nucleolin oligomerization plays a critical role in stabilizing its bound target transcripts. Our findings reveal that a specific tRF can function in mammalian post-transcriptional gene regulation of metabolic genes by driving oligomerization of an RNA binding protein into a stabilizing ribonucleoprotein complex. 5’-tRFCys exhibits a higher affinity for Nucleolin and assembles complex D faster than its target transcripts, while target transcripts display higher binding cooperativity in Nucleolin oligomerization than 5’-tRFCys, suggesting 5’-tRFCys may act as a nucleator for oligomerization from monomeric mRNA-Nucleolin complexes. Our study thus provides the first example wherein a small RNA promotes the oligomerization of an RNA binding protein, thereby stabilizing its target transcripts (Figure 6H). Our finding contrasts with the classical small RNA-mediated gene regulation model wherein a small RNA such as a siRNA or miRNA controls gene expression post-transcriptionally through base pairing interactions with target transcripts. Interestingly, a recent study in bacteria reported that tRFs could activate transcription of the RNA repair operon by promoting oligomerization of the RtcR transcriptional activator, a critical step in operon activation (Hughes et al., 2020). Collectively, our study motivates the search for additional tRFs that could regulate gene expression by promoting the oligomerization of other RNA binding proteins. Given that both complex D assembly and Nucleolin oligomerization require Mg2+ and incubation at an elevated temperature and that Nucleolin IP is more efficient at oligomerization than purified Nucleolin protein, it is tempting to speculate that one or more enzymes that interact with Nucleolin may catalyze complex D assembly and Nucleolin oligomerization. Future studies are required to identify the enzyme(s) responsible for this transformation. Limitations of the Study While we found that Nucleolin binding at the 5’ UTR primarily enhances transcript stability under our experimental conditions, it is possible that Nucleolin binding may play additional roles in post-transcriptional regulation of mRNAs under different experimental conditions, as previously reported (Abdelmohsen et al., 2011; Izumi et al., 2001; Takagi et al., 2005). Additionally, while we could detect up to hexameric forms of Nucleolin, we cannot exclude the possibility that oligomeric complexes of higher orders may exist. Defining and characterizing the molecular components and functional interactions of this oligomeric Nucleolin complex will require future biochemical and structural studies. STAR*METHODS RESOURCE AVAILABILITY Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Sohail Tavazoie (sohail.tavazoie@rockefeller.edu). Materials availability Plasmids generated in this study will be deposited to Addgene. This study did not generate new unique reagents. Data and Code Availability HITS-CLIP, small RNA-Seq, RNA-Seq, and Ribo-Seq data in this study have been deposited at NCBI GEO and are publicly available as of the date of publication. Accession numbers are listed in the key resources table. This paper does not report original code. All software is commercially available or cited in previous publications. Previously published expression datasets used in this study are available through The Cancer Genome Atlas (TCGA). Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. EXPERIMENTAL MODEL AND SUBJECT DETAILS Cell lines 67NR and 4TO7 cells (breast mammary carcinoma) were generously provided by W. P. Schiemann (Case Comprehensive Cancer Center). MDA-MB-231-LM2 (Minn et al., 2005), 67NR, 4TO7, and 4T1 (breast mammary carcinoma, ATCC, CRL-2539) cells were maintained in DMEM supplemented with 10% FBS, sodium pyruvate, L-glutamine, Pen/Strep, and Amphotericin. EO771 (breast mammary carcinoma, ATCC, CRL-3461) and its metastatic lung derivative EO771-LM3 were cultured in RPMI supplemented with 10% FBS, sodium pyruvate, L-glutamine, Pen/Strep, Amphotericin, and 10 mM HEPES. Experiments were performed 24–48 hours post-transfection. All cell lines were authenticated by short-term repeat (STR) analysis and routinely tested for the presence of mycoplasma and were confirmed to be free of mycoplasma contamination. Patient-derived xenograft organoids Breast cancer patient-derived xenograft organoids (PDXO) were obtained from A. L. Welm (DeRose et al., 2011; Sikora et al., 2014). Organoids were embedded in reduced growth factor matrigel (BD Biosciences). The growth media was made from advanced DMEM/F12 (Thermo Fisher Scientific) supplemented with 5% FBS (Thermo Fisher Scientific), 10 mM HEPES, GlutaMAX (Thermo Fisher Scientific), 1 ug/ml Hydrocortisone, 0.05 mg/ml Gentamicin, and 10 ug/ml hEGF. This medium was kept at 4°C for up to 2 weeks and before addition to PDXO cultures supplemented with 10µM Rock Inhibitor Y-27632 (Selleck Chemicals), 100 ng/ml FGF2 (R&D Systems), 1 mM N-Acetyl-L-cysteine (Sigma Aldrich), and 10 mM Heregulin Beta-1 (PeproTech). PDXO cells were transduced with pHIV-Luc-ZsGreen (a gift from B. Welm; Addgene# 39196), pLKO.1-Scr-TD, or pLKO.1-Cys-TD as previously described (Tavora et al., 2020). Lentivirus was concentrated using Lenti-X Concentrator (Takara) and re-suspended in PDXO culture media supplemented with 10 μg/ml polybrene (Sigma Aldrich). Organoids were dissociated and resuspended in the PDXO culture media previously used to resuspend the lentivirus and transferred to a 24 well ultra-low attachment plate (Corning) and centrifuged at 600 g for 1 hour at room temperature and then incubated at 370C for 6 hours. Organoids were expanded in matrigel for another two weeks, and then zsGreen positive organoids were sorted using a BD FACSAria II cell sorter and expanded again in matrigel. Alternatively, organoids were selected with Puromycin until all non-transduced organoids were non-viable. Animal studies All animal work was conducted in accordance with a protocol approved by the Institutional Animal Care and Use Committee (IACUC) at The Rockefeller University. All animals were housed and fed using the university’s standard husbandry protocols. For experimental metastasis assay, 4T1, EO771-LM3, MDA-MB-231-LM2, and PDXO cells labeled with a luciferase and GFP reporter or a luciferase and ZsGreen reporter were injected into 6–10 week-old female Balb/c (The Jackson Laboratory, 000651), C57Bl/6 (The Jackson Laboratory, 000664), and immunocompromised NOD-scid-gamma (NSG) (The Jackson Laboratory, 005557) mice respectively via tail vein. Tumor growth was monitored weekly by luminescence imaging using an IVIS Lumina II (Caliper Life Science). Lung nodules were detected by H&E staining. To monitor Caspase 3/7 activity in vivo, the lung bioluminescence signal was measured using VivoGlo Caspase 3/7 Substrate (Promega) and normalized to the bioluminescence signal from D-luciferin (GoldBio). For mammary fatpad growth assays, 150,000 4T1 cells were resuspended in 1:1 mixture of PBS with Matrigel (Corning) and injected into one of the fourth mammary fatpads of mice in a total volume of 100 μL. Tumor size was measure with digital calipers and tumor volume was calculated using the formula widtĥ2 x length x π/6. METHOD DETAILS Molecular cloning FLAG-tagged mouse Nucleolin cDNA was cloned into pcDNA3. Mouse Mthfd1l or Pafah1b1 cDNAs were cloned into the pLJC6-EF1core-Blast vector. The 5’ UTR of mouse 5’-tRFCys targets Pafah1b1 and Mthfd1l or non-target GAPDH were cloned into the psiCHECK2 vector (Promega). Synthetic 5’-tRF-Cys antisense sequence or a scrambled control sequence embedded in the tough decoy backbone (Haraguchi et al., 2009; Xie et al., 2012) were cloned into the pLKO.1-Puro vector (a gift from R. Weinberg; Addgene #8453) to generate pLKO.1-Cys-TD and pLKO.1-Scr-TD, respectively. RNA isolation Unless otherwise stated, RNAs were isolated using TRIzol and RNeasy MINI kit (Qiagen) with the following modifications to isolate both large (>200 nt) and small (<200 nt) RNAs. In brief, chloroform-extracted lysates were supplemented with the addition of 1.5 volumes of ethanol before lysates were passed through the column, and the column was washed twice with only buffer RPE. Transfection of antisense LNA oligos To inhibit 5’-tRFCys, 20–50 nM scrambled control or 5’-tRFCys antisense locked nucleic acid (LNA) oligos were transfected with Lipofectamine 2000 (Thermo Fisher Scientific) into cells cultured in Opti-MEM (Thermo Fisher Scientific) for 5–6 hours before transfection media was replaced with regular culture media. Lentiviral transduction Lentiviral viruses were produced as described previously (Liu et al., 2014). In brief, 293LTV cells were co-transfected with pLKO.1-Puro-Scr-TD, pLKO.1-Puro-Cys-TD, pLJC6-EF1core-Blast-mPafah1b1, or pLJC6-EF1core-Blast-mMthfd1l, plus pSPAX2 and pMD2.G (gifts from D. Trono; Addgene #12260 and #12259) using Lipofectamine 2000. After 48 hours, lentiviruses were concentrated using Lenti-X concentrators (Takara). 4T1 or MDA-MB-231 cells were transduced with concentrated viruses in the presence of 8 ug/ml of polybrene (Sigma) overnight or spin transduced at 500g at 32 °C for 90 min followed by incubation at 37 °C for 6 hours. Transduced cells were selected with 2–3 ug/ml of Puromycin or 3 ug/ml of Blasticidin 2 days post-transduction until all non-transduced cells were dead. Cell viability and apoptosis assays Cells were seeded in a 24-well plate and transfected with scrambled control or 5’-tRFCys antisense LNA as described above. Cell viability and Caspase activities were detected with CellTiter-Glo 2.0 and CaspaseGlo-3/7 assay (Promega) two days post-transfection respectively. Caspase 3/7 activity was determined by dividing the CaspaseGlo-3/7 signal by the CellTiter-Glo 2.0 signal. Alternatively, cells were seeded in a 6-well plate and transfected with scrambled control or 5’-tRFCys antisense LNA as described above; subsequently, the fraction of apoptotic cells was determined using PE Annexin V Apoptosis Detection Kit I (BD Bioscience) with an Attune Nxt Flow Cytometer (Thermo Fisher Scientific). Immunostaining Immunostaining of mouse lung sections was performed as previously described (Tavora et al., 2020) with the following antibodies: anti-Ki-67 (Abcam), anti-Endomucin (Santa Cruz Biotechnology), anti-cleaved Caspase3 (Asp175) (Cell Signaling Technology), anti-rabbit Alexa Fluor 488, and anti-rat Alexa Fluor 555 (Thermo Fisher Scientific). Images were acquired using a Zeiss LSM 880 confocal microscope. H&E staining of mouse lung sections was performed by Histoserv Inc. Identification of 5’-tRFCys bound proteins Antisense RNA purification was adapted from (McHugh et al., 2015). In brief, 4T1, MDA-MB-231-LM2 or MCF7 cells were exposed to 254 nm of UV radiation at 0.4 mJ/cm2 and lysed in 500 ul of lysis buffer (10 mM Tris-HCl pH 7.5, 500 mM LiCl, 0.5% dodecyl maltoside, 0.2% SDS, 0.1% sodium deoxycholate) plus cOmplete Protease Inhibitor, EDTA-free (Roche) and 0.5 U/ul of RNasin Plus (Promega). Cells were lysed by passage through a G-26 needle five times before the addition of 2.4 ul of DNase salt solution (0.5 M MgCl2 and 0.1 M CaCl2) and 20 U of TURBO DNase (Thermo Fisher Scientific). Lysates were incubated at 37 °C for 10 min before 10 mM EDTA, 5 mM EGTA, and 2.5 mM Tris(2-carboxyethyl)phosphine hydrochloride (TCEP) were added to inactivate the reaction. Lysates were then mixed with 1.5X hybridization buffer (6 M urea, 15 mM Tris-HCl, pH 7.5, 7.5 mM EDTA, 750 mM LiCl, 0.75% dodecyl maltoside, 0.3% SDS and 0.15% sodium deoxycholate) supplemented with 3.75 mM TCEP and centrifuged at 15,000 rpm at 4 °C for 10 min. Supernatants were mixed with biotinylated (scramble control or 5’-tRFCys) antisense oligos and Dynabeads MyOne Streptavidin C1 (Thermo Fisher Scientific) and incubated for 2 hours. Beads were washed 5 times with 1X hybridization buffer (4 M urea, 10 mM Tris-HCl, pH 7.5, 5 mM EDTA, 500 mM LiCl, 0.5% dodecyl maltoside, 0.2% SDS, and 0.1% sodium deoxycholate), twice with 2M urea and pulled down proteins were identified by reversed phase nano-LC-MS/MS (EasyLC 1200 coupled to a Fusion Lumos, Thermo Scientific) at the Rockefeller University Proteomics Resource Center. CLIP RT-qPCR 4T1 cells were lysed with lysis buffer (1×PBS (no Mg2+ or Ca2+), 0.1% SDS, 0.5% deoxycholate and 0.5% IGEPAL CA-630) plus cOmplete, EDTA-free protease inhibitorand 0.5 U/ul RNasin Plus RNase inhibitor. 40 ul of RQ1 DNase I (Promega) was added and the lysate was shaken at 37 °C at 1,000 rpm for 10 min. The lysate was spun at 16,000 g at 4 °C for 10 min. The supernatant was transferred to a fresh tube. Dynabeads M-280 Sheep Anti-Rabbit IgG (Thermo Fisher Scientific) coupled to either a rabbit IgG control or an anti-Nucleolin antibody (Cell Signaling Technology) was added and rotated in the cold room for 1 hour. The supernatant was discarded, and beads were washed twice with lysis buffer, twice with high salt wash buffer (5×PBS (no Mg2+ or Ca2+), 0.1% SDS, 0.5% deoxycholate, 0.5% IGEPAL CA-630) and twice with PNK buffer (50 mM Tris-HCl, pH 7.5, 10 mM MgCl2, 0.5% IGEPAL CA-630). The beads were drained of all liquids before 2 mg/ml of Proteinase K (New England Biolabs) diluted in PK buffer (50 mM Tris-HCl pH 7.5, 10 mM MgCl2, 0.5% IGEPAL CA-630) plus 0.2% SDS was added and shaken at 1,000 rpm at 50 °C for 30 min. RNAs were purified sequentially with acid phenol-chloroform (Thermo Fisher Scientific) and chloroform before overnight precipitation with ethanol in the presence of 20 ug glycogen. Precipitated RNAs were quantified using Taqman miRNA assays with custom primers to detect 5’-tRFCys (Thermo Fisher Scientific). Purification of Nucleolin and Nucleolin-containing complexes To purify Nucleolin protein, 293LTV cells transfected with FLAG-tagged Nucleolin were lysed with lysis buffer (20 mM Tris-HCl, pH 7.5, 200 mM NaCl, 1% Triton-X-100, 10% Glycerol) plus cOmplete EDTA-free protease inhibitor. Cells were sonicated 10 seconds three times and centrifuged at 16,000 g at 4 °C for 10 minutes. Next, the supernatant was mixed with FLAG M2 agarose beads (Sigma) for 1–2 hours in the cold room. The beads were washed with lysis buffer three times and then with high-salt wash buffer (20 mM Tris-HCl, pH 7.5, 1 M NaCl, 1% Triton-X-100, 5% Glycerol, cOmplete EDTA-free protease inhibitor) three times. Nucleolin protein was eluted with FLAG peptide and dialyzed in dialysis buffer (20 mM HEPES, 100 mM KCl and 10% Glycerol). A similar protocol was used to purify Nucleolin-containing complexes, except that lysis and wash steps were both performed with lysis buffer containing 0.5% IGEPAL CA-630 instead of Triton-X-100. Small RNA sequencing and analysis Libraries were constructed from total RNAs isolated from 67NR, 4TO7, or 4T1 cells using RealSeq-AC (SomaGenics) according to the manufacturer’s instruction. All high-throughput sequencing libraries were sequenced for 75 cycles using HiSeq 2500 or NextSeq 500 at the Rockefeller University Genomics Resource Center. Adapters were removed from the 3’ end of reads using cutadapt (Martin, 2011). To identify differentially expressed tRNA fragments, trimmed reads were aligned using bwa (Li and Durbin, 2009) to the mouse mature tRNA space, which consists of all mouse tRNA sequences downloaded from GtRNAdB (Chan and Lowe, 2009; 2016) with introns and CCA-tail removed and identical sequences collapsed. Mapped reads were counted using featureCounts (Liao et al., 2014). Differentially expressed tRNA fragments were determined using DESeq2 (Love et al., 2014). RNA sequencing and analysis Total RNA was depleted of ribosomal RNAs using NEBNext rRNA Depletion Kit (Human/Mouse/Rat) (New England Biolabs) before being used for library construction using NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs). Adaptors were removed from the 3’ end of reads using cutadapt (Martin, 2011). Trimmed reads were aligned to the mouse genome (mm10) using STAR (Dobin et al., 2013). Uniquely mapped reads were counted using featureCounts (Liao et al., 2014), and differentially expressed genes were determined using DESeq2 (Love et al., 2014). Clinical association analyses For survival analysis, statistical significance was determined by the Mantel-Cox log-rank test using the ‘survival’ and ‘survminer’ R packages. Multivariable analyses of the impact of clinical and molecular features on the survival of breast cancer patients were performed according to the Cox proportional hazards model using the ‘survival’ R package. A 5’-tRFCys target signature composed of 23 genes was defined by merging the list of Nucleolin bound genes from HITS-CLIP with the list of significantly downregulated genes upon inhibition with two distinct 5’-tRFCys antisense LNA oligos in RNA-Seq. This aggregate expression signature was then used to make Kaplan–Meier curves. For expression correlation analysis, breast tumors in the TCGA dataset were filtered by their ESTIMATE (Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data) scores (Yoshihara et al., 2013) to retain samples with high tumor purity. Correlation of 5’-tRFCys and protein-coding gene expression was then determined by Pearson’s correlation. HITS-CLIP and analysis HITS-CLIP libraries were constructed as previously described (Moore et al., 2014). To identify small RNAs directly bound by Nucleolin, HITS-CLIP was performed without any RNase. Analysis was done using CLIP Tool Kit (Shah et al., 2017). Piranha was used to identify Nucleolin-bound peaks with a background threshold of 0.95 (Uren et al., 2012). The identified peaks from all samples were merged before reads aligned to the peaks were counted using featureCounts (Liao et al., 2014). Proteomics analysis 4T1 cells transfected with scrambled control or 5’-tRFCys antisense LNA were lysed in lysis buffer (20 mM Tris-HCl, pH 7.5, 100 mM KCl, 1 mM EDTA, 0.5% NP-40) supplemented with cOmplete EDTA-free protease inhibitor and 1 mM DTT. Lysates were sonicated 10 seconds three times and centrifuged at 16,000 g at 4 °C for 10 minutes. The supernatant was sent to identify differentially expressed proteins by reversed phase nano-LC-MS/MS (Ultimate 3000 coupled to a Q-Exactive HF, Thermo Scientific) at the Rockefeller University Proteomics Resource Center. Ribosome profiling and analysis Ribosome profiling (Ribo-Seq) libraries were constructed as described previously (McGlincy and Ingolia, 2017). Adaptors were removed from the 3’ end of reads using cutadapt (Martin, 2011). Trimmed reads were aligned to the mouse genome (mm10) using STAR (Dobin et al., 2013). Both uniquely- and primary multi-mapped reads aligned to the annotated coding sequence of protein-coding genes were counted using featureCounts (Liao et al., 2014). Differentially translated genes were determined using DESeq2 (Love et al., 2014). Preparation of Nucleolin target transcripts and 5’-tRFCys DNA templates for Nucleolin-bound regions were PCR amplified or chemically synthesized by Integrated DNA Technologies and in-vitro transcribed with MEGAshortscript T7 transcription kit (Thermo Fisher Scientific). In-vitro transcribed products and chemically synthesized 5’-tRFCys (Integrated DNA Technologies) were purified on a 10% TBE-Urea PAGE gel (Thermo Fisher Scientific) before dephosphorylated with Quick CIP (New England Biolabs) and 5’ radiolabeled using T4 PNK (New England Biolabs). Labeling reactions were purified with acid-phenol chloroform and chloroform before overnight precipitation with ethanol supplemented with 200 mM NaCl and glycogen. Electrophoretic mobility shift assay Purified Nucleolin protein was incubated with 5’ radiolabeled 5’-tRFCys and/or in-vitro transcribed 5’-tRFCys targets in binding buffer (20 mM HEPES, pH 7, 50 mM KCl, 1 mM DTT, 1 mM EDTA) at 4 °C for 10 min before 3% of Ficoll-400 (Sigma) was added and assembled Nucleolin complexes were resolved on a 10% TBE PAGE gel (Thermo Fisher Scientific) at 4 °C. Nucleolin complex assembly assay Nucleolin protein or IP was incubated with 5’ radiolabeled 5’-tRFCys and/or 5’-tRFCys target transcripts in binding buffer (20 mM HEPES, pH 7, 50 mM KCl, 1 mM DTT, 1 mM EDTA) or reaction buffer (20 mM HEPES, pH 7, 50 mM KCl, 1 mM DTT, 10 mM MgCl2) at 30 °C for 10–20 min unless otherwise stated before 3% of Ficoll-400 was added. Assembled Nucleolin complexes were resolved on a 7% native polyacrylamide gel at 4 °C. Exonucleolytic degradation assay After assembling Nucleolin complexes in reaction buffer (20 mM HEPES, pH 7, 50 mM KCl, 1 mM DTT, 10 mM MgCl2) at 4 or 30 °C to preferentially assemble complex A and D respectively, 0.5 ul of Terminator (Lucigen) was added and incubated at 30 °C for 10 min. RNAs were purified using TRIzol (Thermo Fisher Scientific) and resolved on a 10% TBE-Urea PAGE gel (Thermo Fisher Scientific). Western blotting and chemical crosslinking to detect Nucleolin oligomers Western blotting was performed essentially as previously described (Liu et al., 2012), except that transfers were completed using an XCell II Blot Module (Thermo Fisher Scientific). To detect Nucleolin oligomers, Nucleolin IP was incubated in reaction buffer (20 mM HEPES, pH 7, 50 mM KCl, 1 mM DTT, 10 mM MgCl2) at 30 °C for 10–20 min before crosslinking with ethylene glycol bis(succinimidyl succinate) (EGS) at room temperature for 30 min. Crosslinking was stopped with the addition of 50 mM Tris-HCl, pH 7.5, and incubation at room temperature for 15 min. Samples were separated on a 3–8% Tris-Acetate NuPAGE gel (Thermo Fisher Scientific), transferred to a 0.45 um Immobilon-P PVDF membrane (Millipore) overnight in the cold room, and immunoblotted with an anti-Nucleolin antibody (Cell Signaling). Dual-luciferase assay pcDNA3-FLUC containing zero, two, or five boxB sites were constructed from phRL-TK-5boxB-sp36 (a gift from W. Filipowicz; Addgene #115365) (Pillai et al., 2004) and pcDNA3-RLUC-POLIRES-FLUC (a gift from N. Sonenberg; Addgene #45642) (Poulin et al., 1998) using Gibson assembly. HA-mNcl and λNHA-mNcl were constructed by subcloning mNcl from pcDNA3-mNcl-FLAG into pCIneo-deltaHA and pCIneo-deltaNHA (gifts from W. Filipowicz; Addgene #115360, #115359) (Pillai et al., 2004). Firefly luciferase reporters and λNHA-LacZ (a gift from W. Filipowicz; Addgene #115363) (Pillai et al., 2004), HA-mNcl or λNHA-mNcl were co-transfected with pNL1.1.CMV (Promega) that expressed NanoLuc luciferase into unlabeled 4T1 cells using Lipofectamine 2000 (Life Technologies). Dual-luciferase assay was performed two days post-transfection using the Nano-Glo Dual-Luciferase Reporter Assay (Promega) in a Synergy Neo2 microplate reader (Biotek). Alternatively, the 5’ UTR of the renilla luciferase gene was replaced with Nucleolin-bound regions from the target transcript or GAPDH. Then, the renilla luciferase and firefly luciferase genes were in-vitro transcribed using HiScribe T7 RNA Synthesis Kit (NEB Biolabs) supplemented with N1-Methylpseudo-UTP (TriLink Biotechnologies), polyadenylated with E. coli poly(A) polymerase (NEB Biolabs), and capped and 2’-O-methylated with Vaccinia Capping System (NEB Biolabs) and mRNA Cap 2'-O-Methyltransferase (NEB Biolabs) respectively. Then, LNA oligos were transfected into unlabeled 4T1 cells using Lipofectamine 2000 (Thermo Fisher Scientific) as described above. The next day, firefly and renilla luciferase reporter transcripts were co-transfected into unlabeled 4T1 cells using Lipofectamine RNAiMax (Thermo Fisher Scientific) for 6–8 hours before dual luciferase assay was performed using the Dual-Glo Luciferase Assay (Promega) in a Synergy Neo2 microplate reader (Biotek). Metabolite profiling Cells cultured in 6-well plates were transfected with 50 nM of scrambled control or 5’-tRF-Cys antisense LNA oligos using Lipofectamine 2000 as described above. After 24 hours, cells were washed with 0.9% NaCl twice and lysed with pre-chilled extraction buffer composed of 480 ul of methanol mixed with 120 ul of H2O and 1 uM of pre-mixed Heavy Amino Acid mix (Cambridge Isotope Laboratories). Lysates were transferred to an Eppendorf tube and vortexed at 4 °C for 10 min before being centrifuged at 16,000 g at 4 °C for 10 min. The supernatant was transferred to a fresh Eppendorf tube and dried using a nitrogen air evaporator. The dried pellet was analyzed by LC-MS/MS at the Rockefeller University Proteomics Resource Center. The pellets were resuspended in 60 μl of 50% acetonitrile, vortexed for 10 seconds, centrifuged for 15 minutes at 20,000 g at 4°C and 5 μl of the supernatant was injected onto the LC-MS in a randomized sequence. Chromatographic separation and mass spectrometry analysis was acquired as previously described (Soula et al., 2020). Relative quantification of polar metabolites was performed using Skyline (Pino et al., 2020) with the maximum mass and retention time tolerance set to 2 ppm and 12 s, respectively, referencing an in-house library of polar metabolite standards. QUANTIFICATION AND STATISTICAL ANALYSIS Details about statistical analysis and samples sizes are described in the figure legends or in the corresponding method details section. Statistical analysis was done with Graphpad Prism (GraphPad Software) or R. Data in bar charts or line graphs are represented as mean ± SEM. ns: not significant, *p < 0.05, **p < 0.01, ***p < 0.001. Supplementary Material Suppl_Table1 Suppl_figures_and_legends Suppl_Table2 ACKNOWLEDGEMENTS We thank Siavash Kurdistani and Kivanc Birsoy for their critical comments on our manuscript. We are also grateful to members of the Tavazoie lab for helpful discussions. We thank K. Sawicka and R. Darnell for their assistance with polysome profiling. We thank C. Zhao, C. Lai, and N. Nnatubeugo at the Genomics Resource Center, S. Mazel and the Flow Cytometry Center, the Bio-Imaging Center, the High-Throughput Screening Resource Center and the Comparative Biology Center for their technical support. X.L. was supported by a Bristol Meyers Squibb postdoctoral fellowship. M.C.P was supported by a Medical Scientist Training Program grant from NIH under award number T32GM007739 to the Weill Cornell/Rockefeller/Sloan Kettering Tri-Institutional MD-PhD program, and by an F30 Predoctoral Fellowship from the NIH under award number 1F30CA247026. This study was supported by NIH grant 5R01CA215491. The research of S.F.T. was supported in part by a Faculty Scholar grant from the HHMI, by the DOD Collaborative Scholars and Innovators Award (W81XWH-12–1-0301), Pershing Square Sohn Cancer Research Alliance award, Breast Cancer Research Foundation award, Emerald Foundation, NIH grant U54CA261701, and the Black Family Metastasis Center. Figure 1. 5’-tRFCys is upregulated during breast cancer progression and metastasis. A. A scatter plot depicting log2FoldChange (log2FC) for 5’-tRNA fragments (5’-tRFs) isolated from highly metastatic 4T1, poorly metastatic 4TO7, and non-metastatic 67NR cells. 5’-tRNA halves that were significantly upregulated only in 4T1 but not 4TO7 or 67NR cells are marked in red. The blue dashed lines denote -0.5 and 0.5, respectively. B. Quantification of 5’-tRFCys levels by RT-qPCR from mouse breast cancer cells with differing metastatic capacities (N=3). All data hereafter are represented as mean ± s.e.m. Data points represent biological replicates. Representative results from at least two independent experiments are shown, except mouse experiments, most of which were performed once with cohorts of mice. Statistical significance was determined by a one-tailed t-test with Welch’s correction. *, p<0.05. C. Kaplan-Meier curves depicting survival probability of breast cancer patients (n=1066) in the TCGA cohort stratified by 5’-tRFCys expression levels. Statistical significance was determined by the Mantel-Cox log-rank test. D. 5’-tRFCys expression in breast tumors and matched normal breast tissues in the TCGA cohort (n=101). Statistical significance was determined by paired t-test. Figure 2. 5’-tRFCys promotes breast cancer metastasis and enhances breast cancer cell survival. A, B. Bioluminescence imaging plots of metastatic lung colonization by 4T1 (A) and in-vivo selected lung metastatic EO771-LM3 (B) cells transfected with scrambled control (Scr-LNA) or a 5’-tRFCys antisense LNA oligonucleotide (5’-Cys-LNA1). Representative bioluminescence images and H&E stained lung sections for each cohort are shown (N=5–7). C. Tumor growth rates for 4T1 cells transfected with scrambled control (Scr-LNA) or a 5’-tRFCys antisense LNA oligo (5’-Cys-LNA1) and implanted into mammary fat pads of Balb/c mice (N=9–10). D. Bioluminescence imaging plot of metastatic lung colonization in Nod scid gamma (NSG) mice by MDA-MB-231-LM2 cells. One day following tail-vein injection of MDA-MB-231-LM2 cells, mice were intravenously treated with an LNA targeting 5'-tRFCys (12.5 mg/kg dose) or a mock PBS control and twice weekly thereafter. Arrows denote days that mice received the intravenous treatments. Representative bioluminescence images and H&E-stained lung sections for each cohort are shown (N=7–8). E. Left, quantification of cleaved Caspase 3 positive cells in metastatic lung nodules normalized by areas of lung nodules. Mice were injected with 4T1 cells transduced with a 5’-tRFCys antisense (5’-Cys-TD) or a scrambled control tough decoy (Scr-TD) (N=3). Each data point represents the average of at least 10 different image measurements from one mouse lung section. Right, representative confocal images of anti-cleaved Caspase 3 staining from mouse lung sections. The dashed line delineates metastatic nodules. Scale bar: 10 um. F. Quantification of the fraction of apoptotic cells in 4T1 tumors transfected with scrambled control (Scr-LNA) or a 5’-tRFCys antisense LNA oligo (5’-Cys-LNA1) one day after implantation into mammary fat pads of Balb/c mice (N=9–10). G, H. Quantification of Caspase3/7 activities in 4T1 (G) cells or a human breast cancer patient-derived xenograft organoid line PDXO-1 (H) upon 5’-tRFCys inhibition (N=3–8). Statistical significance in mouse (A-D, F) and cell biology (E, G, H) experiments was determined by a one-tailed Mann-Whitney test and t-test with Welch’s correction, respectively. *, p<0.05; **, p<0.01; ***, p<0.001. Figure 3. Nucleolin is a direct binding partner of 5’-tRFCys. A. Volcano plot depicting log2Fold Change (log2FC) values versus –log10Pvalue in protein abundance quantified from pulldowns conducted with a biotinylated 5’-tRFCys antisense oligo (5’-Cys-AS) in comparison to a biotinylated scrambled control oligo (Scr-AS) in 4T1 cells. B. Venn diagram showing the number of proteins that were enriched by more than twofold in the 5’-tRFCys antisense pulldown compared to the control in distinct human (MCF7, MDA-MB-231-LM2) and mouse (4T1) breast cancer cell lines. C. Quantification by Taqman RT-qPCR assays of the amount of 5’-tRFCys pulled down by an anti-Nucleolin antibody or normal rabbit serum (NRS) from human (MDA-MB-231-LM2) or mouse (4T1) breast cancer cells. Statistical significance was determined by one-tail t-tests with Welch’s correction. ***, p<0.001. D. Dot plot depicting 5’-tRNA halves with the highest fraction of crosslinking induced modification sites (CIMS) in the non-RNase-treated Nucleolin CLIP libraries. Reads derived from the tRNACys loci are marked in red. E.The two most enriched motifs identified using Nucleolin-bound CLIP tags with the highest fractions of CIMS sites in non-RNase-treated Nucleolin CLIP libraries. F.Genome browser view of Nucleolin-CLIP tags and CIMS sites in a representative tRNACys locus from two non-RNase treated libraries. The blue and green boxes denote the two G-rich motifs and the anticodon (AC), respectively. The red rectangle marks the CIMS site. The Y-axis in the CLIP tag track represents reads per million (RPM), while that in the CIMS track represents the number of CIMS sites in CLIP tags. G.Electrophoresis mobility shift assay (EMSA) in the presence of an EDTA-containing buffer using purified Nucleolin protein and 5’ radiolabeled wild-type (WT) or mutant (MUT) 5’-tRFCys that contain mutations in the 5’ (5’MUT), middle (mid-MUT) or both (double-MUT) G-rich motifs. Figure 4. 5’-tRFCys promotes Nucleolin binding to its target transcripts to enhance their stability. A. Density plot of the log10 CLIP tag counts in Nucleolin-bound peaks identified from cells transfected with the control LNA (Scr-LNA) or 5’-tRFCys antisense LNA (5’-Cys-LNA1) oligos. Statistical significance was determined by Kolmogorov–Smirnov test. B. Cumulative distribution function (CDF) of the log2FC in protein abundance between 5’-tRFCys suppressed and control cells for all transcripts stratified by whether their Nucleolin binding was enhanced by 5’-tRFCys (red) or not (grey). Statistical significance was determined by the Kolmogorov–Smirnov test. C.Scatter plot comparing log2FC in the number of ribosome-protected fragments (RPFs) and log2FC in the number of RNA-Seq reads between 5’-tRFCys suppressed (5’-Cys-LNA1) and control cells (Scr-LNA) for all transcripts stratified by whether Nucleolin binding was enhanced by 5’-tRFCys (red) or not (grey). ρ, Spearman’s correlation coefficient. D, E. Quantification of 5’-tRFCys targets’ protein abundances upon inhibition of 5’-tRFCys (D) or depletion of Nucleolin (E). See also Figures S4H and S4I. F, G. Genome browser view of aligned Nucleolin (Ncl) CLIP tags (orange), RNA-Seq reads (red), and Ribo-Seq reads (green) within the 5’ UTR of Mthfd1l (F) and Pafah1b1 (G). The Y-axis represents RPM. TSS, transcription start site. H. Quantification by RT-qPCR of 5’-tRFCys target transcripts bound by Nucleolin immunoprecipitated from MDA-MB-231-LM2 cells transfected with either scrambled control (Scr-LNA) or a 5’-tRFCys antisense LNA oligonucleotide (5’-Cys-LNA1). I. Quantification by RT-qPCR of 5’-tRFCys target transcript abundance in MDA-MB-231-LM2 cells transfected with either scrambled control (Scr-LNA) or a 5’-tRFCys antisense LNA oligonucleotide (5’-Cys-LNA1). J. Quantification by RT-qPCR of the log2FC in Pafah1b1 and Mthfd1l transcript abundance in control (Scr-LNA) or 5’-tRFCys suppressed cells (5’-Cys-LNA1) treated with the RNA Polymerase II inhibitor, 5,6-dichloro-1-beta-D-ribofuranosylbenzimidazole (DRB). Statistical significance was determined by one-tail t-tests with Welch’s correction. *, p<0.05; **, p<0.01; ***, p<0.001. Figure 5. 5’-tRFCys promotes complex D assembly and Nucleolin oligomerization. A. Representative images of western blots of anti-FLAG immunoprecipitates (IP) from cells transfected with either HA-tagged Nucleolin alone or together with FLAG-tagged Nucleolin in the presence or absence of RNase A. B, C. Native gel analysis of complexes assembled from Pafah1b1 (B) or 5’-tRFCys (C) using Nucleolin IP in the presence or absence of Mg2+. Asterisk denotes an RNA-protein complex that was detected only with Nucleolin IP but not Nucleolin protein. D. Native gel analysis of complexes assembled from Pafah1b1 using Nucleolin IP at different temperatures. E. Western blot of Nucleolin using Nucleolin IP that was incubated at 30 °C with or without Mg2+ before crosslinking with ethylene glycol bis (succinimidyl succinate). F. Quantification of the kinetics of complex D assembly using Nucleolin IP with either Pafah1b1 or 5’-tRFCys. See also Figures S5E and S5F. G. Quantification of complex D assembly using increasing amounts of Nucleolin IP with either Pafah1b1 or 5’-tRFCys. Bmax, specific maximum binding. h, Hill coefficient. Kd, equilibrium dissociation constant. See also Figures S5G and S5H. H. Native gel analysis of Nucleolin complexes assembled using Nucleolin IP with either Pafah1b1 alone, or together with wild-type (WT) or a Nucleolin binding deficient 5’-tRFCys (MUT). I. Quantification of different forms of Nucleolin assembled using Nucleolin IP with Pafah1b1 alone or together with 5’-tRFCys. Figure 6. Pafah1b1 and Mthfd1l function downstream of 5’-tRFCys to promote breast cancer metastasis. A, B. Quantification of Caspase3/7 activity in 4T1 cells transfected with a control LNA (Scr-LNA), a 5’-tRFCys antisense LNA (5’-Cys-LNA1) either alone or together with overexpression of Pafah1b1 (Pafah1b1-rescued) (A) or Mthfdl1 (Mthfd1l-rescued) (B). C, D. Bioluminescence quantification of metastatic lung colonization in mice injected with 4T1 cells transfected with a control LNA (Scr-LNA), a 5’-tRFCys antisense LNA (5’-Cys-LNA1) either alone or together with overexpression of Pafah1b1 (Pafah1b1-rescued) (C) or Mthfdl1 (Mthfd1l-rescued) (D). Representative H&E stained lung sections for each cohort are shown (N=6–13). E. Heatmap showing z-scores for the abundance of metabolites that were significantly changed upon 5’-tRFCys inhibition in 4T1 cells. F, G. Kaplan-Meier curves depicting survival probability of breast cancer patients in the TCGA cohort (n=1097) stratified by their expression of Pafah1b1 (F) or Mthfd1l (G). Statistical significance was determined by Mantel-Cox log-rank test. H. A schematic model depicting the mechanism underlying 5’-tRFCys promotion of breast cancer metastasis. KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Pafah1b1 Abcam Cat# ab2607, RRID:AB_2299251 Mthfd1l Cell Signaling Technology Cat# 14999, RRID:AB_2798681 Nucleolin (D4C7O) Cell Signaling Technology Cat# 14574, RRID:AB_2798519 β-Tubulin (9F3) Cell Signaling Technology Cat# 2128, RRID:AB_823664 Hsc70 Proteintech Cat# 10654–1-AP, RRID:AB_2120153 Ki-67 Abcam Cat# ab15580, RRID:AB_443209 Cleaved Caspase-3 (Asp175) Cell Signaling Technology Cat# 9661, RRID:AB_2341188 Endomucin (V.7C7) Santa Cruz Biotechnology Cat# sc-65495, RRID:AB_2100037 DYKDDDDK Tag (D6W5B) Cell Signaling Technology Cat# 14793, RRID:AB_2572291 Goat anti-Rabbit IgG (H+L), Alexa Fluor 488 Thermo Fisher Scientific Cat# A-11008, RRID:AB_143165 Goat anti-Rat IgG (H+L) Cross, Alexa Fluor 555 Thermo Fisher Scientific Cat# A-21434, RRID:AB_141733 Bacterial and Virus Strains TOP10 Thermo Fisher Scientific C404010 5-alpha New England Biolabs C2987 XL10-Gold Stratagene 200315 Biological Samples mouse lungs this study NA Chemicals, Peptides, and Recombinant Proteins ethylene glycol bis(succinimidyl succinate) (EGS) Thermo Fisher Scientific 21565 D-Tube Dialyzer Mini,  MWCO=6–8 kDa Sigma Aldrich 71504–3 Lenti-X Concentrator Takara 631231 TaqMan MicroRNA Reverse Transcription Kit Thermo Fisher Scientific 4366597 Chloroform Sigma Aldrich C2432 Acrylamide: Bis-Acrylamide 19:1 (40%) Fisher Scientific BP1406–1 Acid Phenol:Chloroform, pH 4.5 Thermo Fisher Scientific AM9720 Water, Optima LC/MS Grade Fisher Scientific W6–4 Metabolomics Amino Acid Mix Standard Cambridge Isotope Laboratories MSK-A2–1.2 Methanol, LC-MS grade Fisher Scientific A456–4 E.coli Poly (A) Polymerase New England Biolabs M0276 mRNA Cap2'-O-Methyltransferase New England Biolabs M0366 Vaccinia Capping System New England Biolabs M2080 Advanced DMEM/F-12 Thermo Fisher Scientific 12634010 TaqMan Fast Universal PCR Master Mix (2X), no AmpErase™ UNG Thermo Fisher Scientific 4366072 Fetal Bovine Serum Thermo Fisher Scientific 10437010 GlutaMAX Thermo Fisher Scientific 35050061 Hydrocortisone Sigma Aldrich H0888 Gentamicin Genesee Scientific 25–533 hEGF Sigma Aldrich E9644 Y-27632 Selleck Chemicals S1049 FGF2 R&D Systems 4114-TC-01M N-Acetyl-L-cysteine Sigma Aldrich A7250 Heregulin Beta-1 PeproTech 100–03 Polybrene Sigma Aldrich TR-1003-G 24 well ultra-low attachment plate Corning 3473 Tryple Express Enzyme (1X), Phenol Red Thermo Fisher Scientific 12605010 Matrigel BD Biosciences 356231 10% TBE-Urea Gels, 10 well Thermo Fisher Scientific EC6875BOX cOmplete, EDTA-free Protease Inhibitor Cocktail Roche 11873580001 RNasin Plus Ribonuclease Inhibitor Promega N2611 TURBO DNase Thermo Fisher Scientific AM2239 Tris(2-carboxyethyl)phosphine hydrochloride Sigma Aldrich 75259 Dynabeads MyOne Streptavidin C1 Thermo Fisher Scientific 65601 PhosSTOP Roche 4906845001 Lipofectamine 2000 Thermo Fisher Scientific 11668019 Ficoll 400 Sigma Aldrich F2637 Lipofectamine RNAiMax Thermo Fisher Scientific 13778100 ANTI-FLAG M2 Affinity Gel Sigma Aldrich A2220 NuPAGE 3–8% Tris-Acetate Gel Thermo Fisher Scientific EA0378BOX NuPAGE 4 to 12%, Bis-Tris Gel Thermo Fisher Scientific NP0321BOX T4 Polynucleotide Kinase (PNK) New England Biolabs M0201 Dynabeads M-280 Sheep Anti-Rabbit IgG Thermo Fisher Scientific 11203D Invitrogen Novex HiMark Pre-stained Protein Standard Thermo Fisher Scientific LC5699 Terminator 5′-Phosphate Dependent Exonuclease Lucigen TER51020 Poly(A) Polymerase New England Biolabs M0276S N1-Methylpseudouridine-5'-Triphosphate TriLink N-1081 Vaccinia Capping System New England Biolabs M2080S mRNA Cap 2´-O-Methyltransferase New England Biolabs M0366S Critical Commercial Assays CellTiter-Glo 2.0 Assay, Promega, CellTiter-Glo 2.0 Assay Promega G9241 HiScribe T7 High Yield RNA Synthesis Kit New England Biolabs E2040 MEGAshortscript T7 transcription kit Thermo Fisher Scientific AM1354 Direct-zol RNA MiniPrep Zymo Research R2050 Monarch RNA Cleanup Kit New England Biolabs T2040 Caspase-Glo 3/7 Assay Promega G8091 NEBNext rRNA Depletion Kit (Human/Mouse/Rat) New England Biolabs E6310 NEBNext Ultra II Directional RNA Library Prep Kit New England Biolabs E7760 RNeasy MINI kit Qiagen 74104 QIAGEN Plasmid Plus Midi Kit Qiagen 12943 PE Annexin V Apoptosis Detection Kit I BD Biosciences 559763 Nano-Glo Dual-Luciferase Reporter Assay Promega N1610 Dual-Glo Luciferase Assay Promega E2920 RealSeq-AC SomaGenics NA D-Luciferin GoldBio LUCK-100 VivoGlo Caspase3/7 Substrate Promega P1782 Deposited Data RNA-Seq and Ribo-Seq with mouse 4T1 cells this study GSE173373 Experimental Models: Cell Lines 67NR W. P. Schiemann NA 4TO7 W. P. Schiemann NA 4T1 ATCC Cat# CRL-2539, RRID:CVCL_0125 MDA-MB-231-LM2 Minn et al., 2005 NA EO771 ATCC Cat# CRL-3461, RRID:CVCL_GR23 Experimental Models: Organisms/Strains C57BL/6J The Jackson Laboratory RRID: IMSR_JAX:000664 BALB/c The Jackson Laboratory RRID: IMSR_JAX:000651 NOD scid gamma The Jackson Laboratory RRID: IMSR_JAX:005557 Oligonucleotides Primers and RNA sequences See Suppl Table S1 NA Recombinant DNA pcDNA3-FLAG-mNcl this study NA psiCHECK2-mPafah1b1 this study NA psiCHECK2-mMthfd1l this study NA psiCHECK2-mGapdh this study NA pLKO.1-Puro-Scr-TD this study NA pLKO.1-Puro-Cys-TD this study NA pSPAX2 D. Trono RRID:Addgene_12260 pMD2.G D. Trono RRID:Addgene_12259 pcDNA3-mNcl-FLAG this study NA pHIV-Luc-ZsGreen Bryan Welm RRID:Addgene_39196 pCIneo-deltaNHA-LacZ Pillai et al., 2004 RRID:Addgene_115363 pLJC6-EF1core-Blast-mPafah1b1 this study NA pLJC6-EF1core-Blast-mMthfd1l this study NA pCIneo-deltaNHA-mNcl this study NA pCIneo-HA-mNcl this study NA pcDNA3-FLUC this study NA pcDNA3–2BoxB-FLUC this study NA pcDNA3–5BoxB-FLUC this study NA Software and Algorithms STAR Dobin et al., 2013 https://github.com/alexdobin/STAR bwa Li and Durbin, 2009 http://bio-bwa.sourceforge.net/ DESeq2 Love et al., 2014 https://bioconductor.org/packages/release/bioc/html/DESeq2.html CLIP Tool Kit Shah et al., 2017 https://zhanglab.c2b2.columbia.edu/index.php/CTK_Documentation Piranha Uren et al., 2012 http://smithlabresearch.org/software/piranha/ R R https://www.r-project.org/ Prism GraphPad https://www.graphpad.com/ cutadapt Martin, 2011a https://cutadapt.readthedocs.io/en/stable/ Fiji ImageJ https://imagej.net/Fiji Integrative Genome Viewer Broad Institute https://software.broadinstitute.org/software/igv/ featureCounts Liao et al., 2014 http://subread.sourceforge.net/ Skyline Pino et al., 2020 https://skyline.ms/project/home/begin.view? DECLARATION OF INTERESTS S.F.T. is a cofounder, shareholder and member of the scientific advisory board of Inspirna. The remaining authors declare no competing interests. REFERENCES Abdelmohsen K , and Gorospe M (2012). RNA-binding protein nucleolin in disease. RNA Biol 9 , 799–808. 10.4161/rna.19718.22617883 Abdelmohsen K , Tominaga K , Lee EK , Srikantan S , Kang MJ , Kim MM , Selimyan R , Martindale JL , Yang X , Carrier F , (2011). Enhanced translation by Nucleolin via G-rich elements in coding and non-coding regions of target mRNAs. Nucleic Acids Res 39 , 8513–8530. 10.1093/nar/gkr488.21737422 Agarwal S , Behring M , Hale K , Al Diffalha S , Wang K , Manne U , and Varambally S (2019). MTHFD1L, A Folate Cycle Enzyme, Is Involved in Progression of Colorectal Cancer. Transl Oncol 12 , 1461–1467. 10.1016/j.tranon.2019.07.011.31421459 Arai H , Koizumi H , Aoki J , and Inoue K (2002). Platelet-activating factor acetylhydrolase (PAF-AH). J Biochem 131 , 635–640. 10.1093/oxfordjournals.jbchem.a003145.11983068 Arribere JA , and Gilbert WV (2013). Roles for transcript leaders in translation and mRNA decay revealed by transcript leader sequencing. Genome Res 23 , 977–987. 10.1101/gr.150342.112.23580730 Aslakson CJ , and Miller FR (1992). Selective events in the metastatic process defined by analysis of the sequential dissemination of subpopulations of a mouse mammary tumor. Cancer Res 52 , 1399–1405.1540948 Babiarz JE , Ruby JG , Wang Y , Bartel DP , and Blelloch R (2008). Mouse ES cells express endogenous shRNAs, siRNAs, and other Microprocessor-independent, Dicer-dependent small RNAs. Genes Dev 22 , 2773–2785. 10.1101/gad.1705308.18923076 Cancer Genome Atlas N (2012). Comprehensive molecular portraits of human breast tumours. Nature 490 , 61–70. 10.1038/nature11412.23000897 Cantor JR , and Sabatini DM (2012). Cancer cell metabolism: one hallmark, many faces. Cancer Discov 2 , 881–898. 10.1158/2159-8290.CD-12-0345.23009760 Chan PP , and Lowe TM (2009). GtRNAdb: a database of transfer RNA genes detected in genomic sequence. Nucleic Acids Res 37 , D93–97. 10.1093/nar/gkn787.18984615 Chan PP , and Lowe TM (2016). GtRNAdb 2.0: an expanded database of transfer RNA genes identified in complete and draft genomes. Nucleic Acids Res 44 , D184–189. 10.1093/nar/gkv1309.26673694 Chen J , Guo K , and Kastan MB (2012). Interactions of nucleolin and ribosomal protein L26 (RPL26) in translational control of human p53 mRNA. J Biol Chem 287 , 16467–16476. 10.1074/jbc.M112.349274.22433872 Chen Q , Yan M , Cao Z , Li X , Zhang Y , Shi J , Feng GH , Peng H , Zhang X , Zhang Y , (2016). Sperm tsRNAs contribute to intergenerational inheritance of an acquired metabolic disorder. Science 351 , 397–400. 10.1126/science.aad7977.26721680 Cole C , Sobala A , Lu C , Thatcher SR , Bowman A , Brown JW , Green PJ , Barton GJ , and Hutvagner G (2009). Filtering of deep sequencing data reveals the existence of abundant Dicer-dependent small RNAs derived from tRNAs. RNA 15 , 2147–2160. 10.1261/rna.1738409.19850906 De Gregorio E , Preiss T , and Hentze MW (1999). Translation driven by an eIF4G core domain in vivo. EMBO J 18 , 4865–4874. 10.1093/emboj/18.17.4865.10469664 DeBerardinis RJ , and Chandel NS (2016). Fundamentals of cancer metabolism. Sci Adv 2 , e1600200. 10.1126/sciadv.1600200.27386546 DeRose YS , Wang G , Lin YC , Bernard PS , Buys SS , Ebbert MT , Factor R , Matsen C , Milash BA , Nelson E , (2011). Tumor grafts derived from women with breast cancer authentically reflect tumor pathology, growth, metastasis and disease outcomes. Nat Med 17 , 1514–1520. 10.1038/nm.2454.22019887 Dexter DL , Kowalski HM , Blazar BA , Fligiel Z , Vogel R , and Heppner GH (1978). Heterogeneity of tumor cells from a single mouse mammary tumor. Cancer Res 38 , 3174–3181.210930 Dobin A , Davis CA , Schlesinger F , Drenkow J , Zaleski C , Jha S , Batut P , Chaisson M , and Gingeras TR (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29 , 15–21. 10.1093/bioinformatics/bts635.23104886 Donovan J , Rath S , Kolet-Mandrikov D , and Korennykh A (2017). Rapid RNase L-driven arrest of protein synthesis in the dsRNA response without degradation of translation machinery. RNA 23 , 1660–1671. 10.1261/rna.062000.117.28808124 Ducker GS , Chen L , Morscher RJ , Ghergurovich JM , Esposito M , Teng X , Kang Y , and Rabinowitz JD (2016). Reversal of Cytosolic One-Carbon Flux Compensates for Loss of the Mitochondrial Folate Pathway. Cell Metab 24 , 640–641. 10.1016/j.cmet.2016.09.011.27732838 Ducker GS , and Rabinowitz JD (2017). One-Carbon Metabolism in Health and Disease. Cell Metab 25 , 27–42. 10.1016/j.cmet.2016.08.009.27641100 Eich ML , Rodriguez Pena MDC , Chandrashekar DS , Chaux A , Agarwal S , Gordetsky JB , Ferguson JE , Sonpavde GP , Netto GJ , and Varambally S (2019). Expression and Role of Methylenetetrahydrofolate Dehydrogenase 1 Like (MTHFD1L) in Bladder Cancer. Transl Oncol 12 , 1416–1424. 10.1016/j.tranon.2019.07.012.31401334 Fu H , Feng J , Liu Q , Sun F , Tie Y , Zhu J , Xing R , Sun Z , and Zheng X (2009). Stress induces tRNA cleavage by angiogenin in mammalian cells. FEBS Lett 583 , 437–442. 10.1016/j.febslet.2008.12.043.19114040 Ghisolfi-Nieto L , Joseph G , Puvion-Dutilleul F , Amalric F , and Bouvet P (1996). Nucleolin is a sequence-specific RNA-binding protein: characterization of targets on pre-ribosomal RNA. J Mol Biol 260 , 34–53. 10.1006/jmbi.1996.0380.8676391 Ginisty H , Amalric F , and Bouvet P (1998). Nucleolin functions in the first step of ribosomal RNA processing. EMBO J 17 , 1476–1486. 10.1093/emboj/17.5.1476.9482744 Ginisty H , Serin G , Ghisolfi-Nieto L , Roger B , Libante V , Amalric F , and Bouvet P (2000). Interaction of nucleolin with an evolutionarily conserved pre-ribosomal RNA sequence is required for the assembly of the primary processing complex. J Biol Chem 275 , 18845–18850. 10.1074/jbc.M002350200.10858445 Ginisty H , Sicard H , Roger B , and Bouvet P (1999). Structure and functions of nucleolin. J Cell Sci 112 (Pt 6 ), 761–772.10036227 Goodarzi H , Liu X , Nguyen HC , Zhang S , Fish L , and Tavazoie SF (2015). Endogenous tRNA-Derived Fragments Suppress Breast Cancer Progression via YBX1 Displacement. Cell 161 , 790–802. 10.1016/j.cell.2015.02.053.25957686 Hanahan D , and Weinberg RA (2011). Hallmarks of cancer: the next generation. Cell 144 , 646–674. 10.1016/j.cell.2011.02.013.21376230 Haraguchi T , Ozaki Y , and Iba H (2009). Vectors expressing efficient RNA decoys achieve the long-term suppression of specific microRNA activity in mammalian cells. Nucleic Acids Res 37 , e43. 10.1093/nar/gkp040.19223327 Hilvo M , Denkert C , Lehtinen L , Muller B , Brockmoller S , Seppanen-Laakso T , Budczies J , Bucher E , Yetukuri L , Castillo S , (2011). Novel theranostic opportunities offered by characterization of altered membrane lipid metabolism in breast cancer progression. Cancer Res 71 , 3236–3245. 10.1158/0008-5472.CAN-10-3894.21415164 Honda S , Loher P , Shigematsu M , Palazzo JP , Suzuki R , Imoto I , Rigoutsos I , and Kirino Y (2015). Sex hormone-dependent tRNA halves enhance cell proliferation in breast and prostate cancers. Proc Natl Acad Sci U S A 112 , E3816–3825. 10.1073/pnas.1510077112.26124144 Hughes KJ , Chen X , Burroughs AM , Aravind L , and Wolin SL (2020). An RNA Repair Operon Regulated by Damaged tRNAs. Cell Rep 33 , 108527. 10.1016/j.celrep.2020.108527.33357439 Ivanov P , Emara MM , Villen J , Gygi SP , and Anderson P (2011). Angiogenin-induced tRNA fragments inhibit translation initiation. Mol Cell 43 , 613–623. 10.1016/j.molcel.2011.06.022.21855800 Izumi RE , Valdez B , Banerjee R , Srivastava M , and Dasgupta A (2001). Nucleolin stimulates viral internal ribosome entry site-mediated translation. Virus Res 76 , 17–29. 10.1016/s0168-1702(01)00240–4.11376843 Jain M , Nilsson R , Sharma S , Madhusudhan N , Kitami T , Souza AL , Kafri R , Kirschner MW , Clish CB , and Mootha VK (2012). Metabolite profiling identifies a key role for glycine in rapid cancer cell proliferation. Science 336 , 1040–1044. 10.1126/science.1218595.22628656 Karasawa K , and Inoue K (2015). Overview of PAF-Degrading Enzymes. Enzymes 38 , 1–22. 10.1016/bs.enz.2015.09.006.26612643 Kim HK , Fuchs G , Wang S , Wei W , Zhang Y , Park H , Roy-Chaudhuri B , Li P , Xu J , Chu K , (2017). A transfer-RNA-derived small RNA regulates ribosome biogenesis. Nature 552 , 57–62. 10.1038/nature25005.29186115 Lazinski D , Grzadzielska E , and Das A (1989). Sequence-specific recognition of RNA hairpins by bacteriophage antiterminators requires a conserved arginine-rich motif. Cell 59 , 207–218. 10.1016/0092-8674(89)90882-9.2477156 Lee D , Xu IM , Chiu DK , Lai RK , Tse AP , Lan Li L , Law CT , Tsang FH , Wei LL , Chan CY , (2017). Folate cycle enzyme MTHFD1L confers metabolic advantages in hepatocellular carcinoma. J Clin Invest 127 , 1856–1872. 10.1172/JCI90253.28394261 Lee YS , Shibata Y , Malhotra A , and Dutta A (2009). A novel class of small RNAs: tRNA-derived RNA fragments (tRFs). Genes Dev 23 , 2639–2649. 10.1101/gad.1837609.19933153 Li H , and Durbin R (2009). Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25 , 1754–1760. 10.1093/bioinformatics/btp324.19451168 Liao Y , Smyth GK , and Shi W (2014). featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30 , 923–930. 10.1093/bioinformatics/btt656.24227677 Liu X , Jin DY , McManus MT , and Mourelatos Z (2012). Precursor microRNA-programmed silencing complex assembly pathways in mammals. Mol Cell 46 , 507–517. 10.1016/j.molcel.2012.03.010.22503104 Liu X , Zheng Q , Vrettos N , Maragkakis M , Alexiou P , Gregory BD , and Mourelatos Z (2014). A MicroRNA precursor surveillance system in quality control of MicroRNA synthesis. Mol Cell 55 , 868–879. 10.1016/j.molcel.2014.07.017.25175028 Lo FY , Chen HT , Cheng HC , Hsu HS , and Wang YC (2012). Overexpression of PAFAH1B1 is associated with tumor metastasis and poor survival in non-small cell lung cancer. Lung Cancer 77 , 585–592. 10.1016/j.lungcan.2012.05.105.22749159 Locasale JW (2013). Serine, glycine and one-carbon units: cancer metabolism in full circle. Nat Rev Cancer 13 , 572–583. 10.1038/nrc3557.23822983 Loo JM , Scherl A , Nguyen A , Man FY , Weinberg E , Zeng Z , Saltz L , Paty PB , and Tavazoie SF (2015). Extracellular metabolic energetics can promote cancer progression. Cell 160 , 393–406. 10.1016/j.cell.2014.12.018.25601461 Love MI , Huber W , and Anders S (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15 , 550. 10.1186/s13059-014-0550-8.25516281 Lytle JR , Yario TA , and Steitz JA (2007). Target mRNAs are repressed as efficiently by microRNA-binding sites in the 5' UTR as in the 3' UTR. Proc Natl Acad Sci U S A 104 , 9667–9672. 10.1073/pnas.0703820104.17535905 Ma C , Guo Y , Zhang Y , Duo A , Jia Y , Liu C , and Li B (2018). PAFAH1B2 is a HIF1a target gene and promotes metastasis in pancreatic cancer. Biochem Biophys Res Commun 501 , 654–660. 10.1016/j.bbrc.2018.05.039.29758199 Martin M (2011). Cutadapt Removes Adapter Sequences From High-Throughput Sequencing Reads. EMBnet.journal 17 , 10–12. 10.14806/ej.17.1.200. McGlincy NJ , and Ingolia NT (2017). Transcriptome-wide measurement of translation by ribosome profiling. Methods 126 , 112–129. 10.1016/j.ymeth.2017.05.028.28579404 McHugh CA , Chen CK , Chow A , Surka CF , Tran C , McDonel P , Pandya-Jones A , Blanco M , Burghard C , Moradian A , (2015). The Xist lncRNA interacts directly with SHARP to silence transcription through HDAC3. Nature 521 , 232–236. 10.1038/nature14443.25915022 Mendell JT , and Olson EN (2012). MicroRNAs in stress signaling and human disease. Cell 148 , 1172–1187. 10.1016/j.cell.2012.02.005.22424228 Minn AJ , Gupta GP , Siegel PM , Bos PD , Shu W , Giri DD , Viale A , Olshen AB , Gerald WL , and Massague J (2005). Genes that mediate breast cancer metastasis to lung. Nature 436 , 518–524. 10.1038/nature03799.16049480 Minton DR , Nam M , McLaughlin DJ , Shin J , Bayraktar EC , Alvarez SW , Sviderskiy VO , Papagiannakopoulos T , Sabatini DM , Birsoy K , and Possemato R (2018). Serine Catabolism by SHMT2 Is Required for Proper Mitochondrial Translation Initiation and Maintenance of Formylmethionyl-tRNAs. Mol Cell 69 , 610–621 e615. 10.1016/j.molcel.2018.01.024.29452640 Mongelard F , and Bouvet P (2007). Nucleolin: a multiFACeTed protein. Trends Cell Biol 17 , 80–86. 10.1016/j.tcb.2006.11.010.17157503 Moore MJ , Zhang C , Gantman EC , Mele A , Darnell JC , and Darnell RB (2014). Mapping Argonaute and conventional RNA-binding protein interactions with RNA at single-nucleotide resolution using HITS-CLIP and CIMS analysis. Nat Protoc 9 , 263–293. 10.1038/nprot.2014.012.24407355 Mulvihill MM , Benjamin DI , Ji X , Le Scolan E , Louie SM , Shieh A , Green M , Narasimhalu T , Morris PJ , Luo K , and Nomura DK (2014). Metabolic profiling reveals PAFAH1B3 as a critical driver of breast cancer pathogenicity. Chem Biol 21 , 831–840. 10.1016/j.chembiol.2014.05.008.24954006 Nabet BY , Qiu Y , Shabason JE , Wu TJ , Yoon T , Kim BC , Benci JL , DeMichele AM , Tchou J , Marcotrigiano J , and Minn AJ (2017). Exosome RNA Unshielding Couples Stromal Activation to Pattern Recognition Receptor Signaling in Cancer. Cell 170 , 352–366 e313. 10.1016/j.cell.2017.06.031.28709002 Nechooshtan G , Yunusov D , Chang K , and Gingeras TR (2020). Processing by RNase 1 forms tRNA halves and distinct Y RNA fragments in the extracellular environment. Nucleic Acids Res 48 , 8035–8049. 10.1093/nar/gkaa526.32609822 Oberbauer V , and Schaefer MR (2018). tRNA-Derived Small RNAs: Biogenesis, Modification, Function and Potential Impact on Human Disease Development. Genes (Basel) 9 . 10.3390/genes9120607. Pencheva N , Tran H , Buss C , Huh D , Drobnjak M , Busam K , and Tavazoie SF (2012). Convergent multi-miRNA targeting of ApoE drives LRP1/LRP8-dependent melanoma metastasis and angiogenesis. Cell 151 , 1068–1082. 10.1016/j.cell.2012.10.028.23142051 Percharde M , Lin CJ , Yin Y , Guan J , Peixoto GA , Bulut-Karslioglu A , Biechele S , Huang B , Shen X , and Ramalho-Santos M (2018). A LINE1-Nucleolin Partnership Regulates Early Development and ESC Identity. Cell 174 , 391–405 e319. 10.1016/j.cell.2018.05.043.29937225 Pichiorri F , Palmieri D , De Luca L , Consiglio J , You J , Rocci A , Talabere T , Piovan C , Lagana A , Cascione L , (2013). In vivo NCL targeting affects breast cancer aggressiveness through miRNA regulation. J Exp Med 210 , 951–968. 10.1084/jem.20120950.23610125 Pickering BF , Yu D , and Van Dyke MW (2011). Nucleolin protein interacts with microprocessor complex to affect biogenesis of microRNAs 15a and 16. J Biol Chem 286 , 44095–44103. 10.1074/jbc.M111.265439.22049078 Pillai RS , Artus CG , and Filipowicz W (2004). Tethering of human Ago proteins to mRNA mimics the miRNA-mediated repression of protein synthesis. RNA 10 , 1518–1525. 10.1261/rna.7131604.15337849 Pino LK , Searle BC , Bollinger JG , Nunn B , MacLean B , and MacCoss MJ (2020). The Skyline ecosystem: Informatics for quantitative mass spectrometry proteomics. Mass Spectrom Rev 39 , 229–244. 10.1002/mas.21540.28691345 Pliatsika V , Loher P , Magee R , Telonis AG , Londin E , Shigematsu M , Kirino Y , and Rigoutsos I (2018). MINTbase v2.0: a comprehensive database for tRNA-derived fragments that includes nuclear and mitochondrial fragments from all The Cancer Genome Atlas projects. Nucleic Acids Res 46 , D152–D159. 10.1093/nar/gkx1075.29186503 Poulin F , Gingras AC , Olsen H , Chevalier S , and Sonenberg N (1998). 4E-BP3, a new member of the eukaryotic initiation factor 4E-binding protein family. J Biol Chem 273 , 14002–14007. 10.1074/jbc.273.22.14002.9593750 Ridgway ND (2013). The role of phosphatidylcholine and choline metabolites to cell proliferation and survival. Crit Rev Biochem Mol Biol 48 , 20–38. 10.3109/10409238.2012.735643.23350810 Schimmel P (2018). The emerging complexity of the tRNA world: mammalian tRNAs beyond protein synthesis. Nat Rev Mol Cell Biol 19 , 45–58. 10.1038/nrm.2017.77.28875994 Schorn AJ , Gutbrod MJ , LeBlanc C , and Martienssen R (2017). LTR-Retrotransposon Control by tRNA-Derived Small RNAs. Cell 170 , 61–71 e11. 10.1016/j.cell.2017.06.013.28666125 Selcuklu SD , Donoghue MT , Rehmet K , de Souza Gomes M , Fort A , Kovvuru P , Muniyappa MK , Kerin MJ , Enright AJ , and Spillane C (2012). MicroRNA-9 inhibition of cell proliferation and identification of novel miR-9 targets by transcriptome profiling in breast cancer cells. J Biol Chem 287 , 29516–29528. 10.1074/jbc.M111.335943.22761433 Sengupta TK , Bandyopadhyay S , Fernandes DJ , and Spicer EK (2004). Identification of nucleolin as an AU-rich element binding protein involved in bcl-2 mRNA stabilization. J Biol Chem 279 , 10855–10863. 10.1074/jbc.M309111200.14679209 Serin G , Joseph G , Faucher C , Ghisolfi L , Bouche G , Amalric F , and Bouvet P (1996). Localization of nucleolin binding sites on human and mouse pre-ribosomal RNA. Biochimie 78 , 530–538. 10.1016/0300-9084(96)84759-6.8915542 Shah A , Qian Y , Weyn-Vanhentenryck SM , and Zhang C (2017). CLIP Tool Kit (CTK): a flexible and robust pipeline to analyze CLIP sequencing data. Bioinformatics 33 , 566–567. 10.1093/bioinformatics/btw653.27797762 Shao Y , Sun Q , Liu X , Wang P , Wu R , and Ma Z (2017). tRF-Leu-CAG promotes cell proliferation and cell cycle in non-small cell lung cancer. Chem Biol Drug Des 90 , 730–738. 10.1111/cbdd.12994.28378898 Sharma U , Conine CC , Shea JM , Boskovic A , Derr AG , Bing XY , Belleannee C , Kucukural A , Serra RW , Sun F , (2016). Biogenesis and function of tRNA fragments during sperm maturation and fertilization in mammals. Science 351 , 391–396. 10.1126/science.aad6780.26721685 Sikora MJ , Cooper KL , Bahreini A , Luthra S , Wang G , Chandran UR , Davidson NE , Dabbs DJ , Welm AL , and Oesterreich S (2014). Invasive lobular carcinoma cell lines are characterized by unique estrogen-mediated gene expression patterns and altered tamoxifen response. Cancer Res 74 , 1463–1474. 10.1158/0008-5472.CAN-13-2779.24425047 Soula M , Weber RA , Zilka O , Alwaseem H , La K , Yen F , Molina H , Garcia-Bermudez J , Pratt DA , and Birsoy K (2020). Metabolic determinants of cancer cell sensitivity to canonical ferroptosis inducers. Nat Chem Biol 16 , 1351–1360. 10.1038/s41589-020-0613-y.32778843 Srivastava M , and Pollard HB (1999). Molecular dissection of nucleolin's role in growth and cell proliferation: new insights. FASEB J 13 , 1911–1922.10544174 Sullivan WJ , Mullen PJ , Schmid EW , Flores A , Momcilovic M , Sharpley MS , Jelinek D , Whiteley AE , Maxwell MB , Wilde BR , (2018). Extracellular Matrix Remodeling Regulates Glucose Metabolism through TXNIP Destabilization. Cell 175 , 117–132 e121. 10.1016/j.cell.2018.08.017.30197082 Takagi M , Absalon MJ , McLure KG , and Kastan MB (2005). Regulation of p53 translation and induction after DNA damage by ribosomal protein L26 and nucleolin. Cell 123 , 49–63. 10.1016/j.cell.2005.07.034.16213212 Tavora B , Mederer T , Wessel KJ , Ruffing S , Sadjadi M , Missmahl M , Ostendorf BN , Liu X , Kim JY , Olsen O , (2020). Tumoural activation of TLR3-SLIT2 axis in endothelium drives metastasis. Nature 586 , 299–304. 10.1038/s41586-020-2774-y.32999457 Truitt ML , and Ruggero D (2017). New frontiers in translational control of the cancer genome. Nat Rev Cancer 17 , 332. 10.1038/nrc.2017.30. Ugrinova I , Monier K , Ivaldi C , Thiry M , Storck S , Mongelard F , and Bouvet P (2007). Inactivation of nucleolin leads to nucleolar disruption, cell cycle arrest and defects in centrosome duplication. BMC Mol Biol 8 , 66. 10.1186/1471-2199-8-66.17692122 Uren PJ , Bahrami-Samani E , Burns SC , Qiao M , Karginov FV , Hodges E , Hannon GJ , Sanford JR , Penalva LO , and Smith AD (2012). Site identification in high-throughput RNA-protein interaction data. Bioinformatics 28 , 3013–3020. 10.1093/bioinformatics/bts569.23024010 Xie J , Ameres SL , Friedline R , Hung JH , Zhang Y , Xie Q , Zhong L , Su Q , He R , Li M , (2012). Long-term, efficient inhibition of microRNA function in mice using rAAV vectors. Nat Methods 9 , 403–409. 10.1038/nmeth.1903.22388288 Yamasaki S , Ivanov P , Hu GF , and Anderson P (2009). Angiogenin cleaves tRNA and promotes stress-induced translational repression. J Cell Biol 185 , 35–42. 10.1083/jcb.200811106.19332886 Yang M , and Vousden KH (2016). Serine and one-carbon metabolism in cancer. Nat Rev Cancer 16 , 650–662. 10.1038/nrc.2016.81.27634448 Yoshihara K , Shahmoradgoli M , Martinez E , Vegesna R , Kim H , Torres-Garcia W , Trevino V , Shen H , Laird PW , Levine DA , (2013). Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun 4 , 2612. 10.1038/ncomms3612.24113773 Zaidi SH , and Malter JS (1995). Nucleolin and heterogeneous nuclear ribonucleoprotein C proteins specifically interact with the 3'-untranslated region of amyloid protein precursor mRNA. J Biol Chem 270 , 17292–17298. 10.1074/jbc.270.29.17292.7615529 Zheng Y , Lin TY , Lee G , Paddock MN , Momb J , Cheng Z , Li Q , Fei DL , Stein BD , Ramsamooj S , (2018). Mitochondrial One-Carbon Pathway Supports Cytosolic Folate Integrity in Cancer Cells. Cell 175 , 1546–1560 e1517. 10.1016/j.cell.2018.09.041.30500537 Zhu H , Shyh-Chang N , Segre AV , Shinoda G , Shah SP , Einhorn WS , Takeuchi A , Engreitz JM , Hagan JP , Kharas MG , (2011). The Lin28/let-7 axis regulates glucose metabolism. Cell 147 , 81–94. 10.1016/j.cell.2011.08.033.21962509 Zimdahl B , Ito T , Blevins A , Bajaj J , Konuma T , Weeks J , Koechlein CS , Kwon HY , Arami O , Rizzieri D , (2014). Lis1 regulates asymmetric division in hematopoietic stem cells and in leukemia. Nat Genet 46 , 245–252. 10.1038/ng.2889.24487275

PMC009xxxxxx/PMC9450512.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101676030 44856 Cell Chem Biol Cell Chem Biol Cell chemical biology 2451-9456 2451-9448 35709754 9450512 10.1016/j.chembiol.2022.05.008 NIHMS1817660 Article Evidence that HDAC7 Acts as an Epigenetic “Reader” of AR Acetylation through NCoR-HDAC3 Dissociation Zhang Yuchen Andrade Rafael Hanna Anthony A. Pflum Mary Kay H. * Department of Chemistry, Wayne State University, 5101 Cass Ave. Detroit, MI 48202 Author Contribution: M.K.P conceived the project. Y.Z. performed all experiments, except the NCoR binding studies in the absence of cellular proteins (Figure 3B–C), immunohistochemistry (Figure S6 and S9B), and AR acetylation studies (Figure S11) performed by R.A. and AR peptide synthesis and repetitive AR peptide-dependent NCoR binding assays (Figure S5) performed by A.H.. M.K.P., Y.Z., and R.A. wrote the manuscript. * to whom correspondence should be sent and Lead Contact: pflum@wayne.edu 5 7 2022 21 7 2022 15 6 2022 21 7 2023 29 7 11621173.e5 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Summary Histone deacetylase (HDAC) proteins are epigenetic regulators that govern a wide variety of cellular events. With a role in cancer formation, HDAC inhibitors have emerged as anti-cancer therapeutics. Among the eleven metal-dependent class I, II and IV HDAC proteins targeted by inhibitor drugs, class IIa HDAC4, 5, 7, and 9 harbor low deacetylase activity and are hypothesized to be “reader” proteins, which bind to post-translationally acetylated lysine. However, evidence linking acetyllysine binding to a downstream functional event is lacking. Here, we report for the first time that HDAC4, 5 and 7 dissociated from corepressor NCoR in the presence of an acetyllysine-containing peptide, consistent with reader function. Documenting the biological consequences of this possible reader function, mutation of a critical acetylation site regulated Androgen Receptor (AR) transcriptional activation function through HDAC7-NCoR-HDAC3 dissociation. The data document the first evidence consistent with epigenetic reader functions of class IIa HDAC proteins. eTOC blurb Zhang et al. report evidence for the epigenetic reader function of Histone Deacetylase 7 (HDAC7). A model is proposed where transcription by Androgen Receptor (AR) is activated by binding of the AR K630 acetylation site to HDAC7 and dissociation of the repressive NCoR-HDAC3 complex Graphical Abstract histone deacetylase HDAC7 NCoR androgen receptor epigenetic reader pmcIntroduction Lysine acetylation has emerged as a prominent posttranslational modification with impact on numerous cellular functions (Choudhary et al., 2014). As a well-studied example, acetylation of nucleosomal histone proteins influences gene expression through chromatin remodeling (Hebbes et al., 1988), which is a key mechanism in epigenetic regulation. The removal of the acetyl group from acetyllysine is catalyzed by histone deacetylase (HDAC) proteins. Numerous studies showed a correlation between overexpression of HDAC proteins and human tumor formation (West and Johnstone, 2014). Elevated HDAC activities resulted in reduced global acetylation of histone H4 and altered gene expression, which is a hallmark of cancer onset (Fraga et al., 2005). Due to the role in cancer progression, four HDAC inhibitors are approved as cancer therapeutics, including SAHA (Vorinostat or suberoyl anilide hydroxamic acid) (Hesham et al., 2018). The 18 human HDAC proteins comprise four classes (de Ruijter et al., 2003). Class I (HDAC1, 2, 3, and 8), II (HDAC4, 5, 6, 7, 9, and 10) and IV (HDAC11) bear a catalytic metal ion, whereas class III (Sirtuins 1–7) utilize NAD+ as the cosubstrate (Houtkooper et al., 2012). Class II is further subdivided into class IIa (HDAC4, 5, 7, and 9) and IIb (HDAC6 and 10) based on size and catalytic activity. Class IIa HDAC proteins bear a unique structural zinc binding site that is located adjacent to the active site (Bottomley et al., 2008b; Schuetz et al., 2008). Importantly, class IIa proteins harbor almost no deacetylase activity due to mutation of a catalytic Tyr to His (Clocchiatti et al., 2011; Jones et al., 2008; Lahm et al., 2007). We note that the mutation does not rule out the possibility that class IIa HDAC proteins can serve as a deacylase towards substrates yet to be identified. However, the current evidence suggests that class IIa proteins regulate histone deacetylation and gene expression by interacting with other proteins, including the multi-protein complex containing the nuclear receptor corepressor NCoR/SMRT and HDAC3 (Fischle et al., 2002b; Guenther et al., 2001; Li et al., 2000). Through NCoR/SMRT binding, class IIa HDAC proteins act as scaffolds to recruit active HDAC3 and regulate histone-mediated gene expression. Whereas the scaffolding function of Class IIa HDAC proteins is well established, the purpose of the inactive catalytic domain in that scaffolding activity remains unclear. One hypothesis suggests that HDAC4, 5, 7, and 9 act as “reader” proteins, similar to bromodomain-containing proteins (Arrowsmith et al., 2012; Bradner et al., 2010; Lobera et al., 2013). With bromodomains, acetyllysine binding results in a downstream effect. For example, the bromodomain of acetyltransferase p300/CBP aids in acetyllysine-dependent histone substrate recruitment (Manning et al., 2001), which leads to elevate histone acetylation and transcriptional activation (Ebrahimi et al., 2019). As this example illustrates, the characterization of bromodomains as “readers” of acetylation has provided critical mechanistic insights into epigenetic regulation of gene expression. Importantly, bromodomains have emerged as targets for drug development (Smith and Zhou, 2016). Despite speculation, little evidence supports the hypothesis that class IIa HDAC proteins maintain reader function. An acetyllysine-containing peptide inhibited class IIa deacylation of non-natural trifluoroacetyllysine substrates (Bradner et al., 2010), which suggests that acetyllysine binds to the active site of class IIa HDAC proteins. In addition, the presence of HDAC inhibitors or mutation of amino acids in the inactive catalytic domain disrupted the HDAC4-NCoR/SMRT complex (Bottomley et al., 2008b; Gaur et al., 2016; Kim et al., 2015), which suggests that active site binding influences the scaffolding activity of class IIa HDAC proteins. These combined results implicate a model where class IIa HDAC proteins act as readers through active site binding of acetyllysine-containing proteins to promote NCoR/SMRT dissociation. However, no experimental evidence yet documents acetyl-lysine dependent disruption of HDAC-NCoR/SMRT binding. More importantly, no acetyllysine-containing protein is known to bind to the active site of a class IIa HDAC in a “reader” manner. Therefore, no cellular mechanism links acetyllysine-containing protein binding to HDAC7-NCoR association and a downstream biological consequence. To date, any possible reader function of class IIa HDAC proteins remains unsubstantiated. Androgen receptor (AR) is a transcriptional activator regulated by binding to a hormone ligand, such as dihydrotestosterone (DHT) (Gao et al., 2005). The transcriptional activity of AR is influenced by HDAC4 and HDAC7 (Karvonen et al., 2006; Yang et al., 2011). With HDAC4, the E3 ubiquitin ligase activity of HDAC4 facilitates sumoylation to inhibit AR activities (Yang et al., 2011). In contrast, HDAC7 regulates AR activity through co-localization with the NCoR-HDAC3 complex (Karvonen et al., 2006). Related, AR-mediated transcription hinges on the acetylation of K630 or K632/633, with loss of transactivation upon mutation of K630 or K632/633 (Fu et al., 2000; Fu et al., 2002). Moreover, wild type AR showed 10-fold reduced binding to NCoR than the K630A mutant (Fu et al., 2002). Taken together, prior data suggests that acetylation-dependent AR transcriptional activity involves the HDAC7-NCoR-HDAC3 complex, although a molecular mechanism linking AR function to HDAC7 and NCoR is unknown. Based on the prior literature, we hypothesized that class IIa HDAC proteins act as reader proteins to influence the transcriptional activity of AR. Consistent with the reader hypothesis, HDAC4, 5, and 7, but not HDAC9, bound NCoR in an acetyl-lysine-dependent manner. An assessment of binding affinity confirmed the stable binding of an acetyllysine-containing peptide with HDAC7. To provide a biological context for the class IIa HDAC-NCoR reader function, dissociation of the HDAC7-NCoR complex was dependent on the K630 acetylation site of AR, and the transcriptional activity of AR required both an intact HDAC7 active site and K630 acetylation site. In total, the work here reports the first evidence for the epigenetic reader function of HDAC7, where AR transcriptional activity is regulated through acetyllysine-dependent HDAC7-NCoR complex dissociation. Results Inhibitor-dependent class IIa HDAC-NCoR binding A key feature of epigenetic reader proteins, such as bromodomains, is that a binding event leads to a downstream function. Related to class IIa HDAC proteins, prior work documented that HDAC inhibitors disrupted the HDAC4-NCoR/SMRT-HDAC3 complex (Bottomley et al., 2008b; Gaur et al., 2016). In addition, gain-of-function (GOF) mutant HDAC4 bound more tightly to HDAC inhibitors and demonstrated more dramatic loss of NCoR binding than wild type (Gaur et al., 2016; Hudson et al., 2015). This prior data suggests a reader function for HDAC4, where active site binding results in dissociation of NCoR/SMRT. However, the other three class IIa HDAC isoforms, HDAC5, HDAC7, and HDAC9, had not yet been tested. To initially confirm and extend the prior data, SAHA inhibitor-dependent binding of both wild type and GOF mutants of all four class IIa HDAC isoforms to NCoR was assessed using a gel analysis. First, GOF mutants of HDAC4, 5, 7, and 9 were generated using Quickchange mutagenesis (Figure S1A–B)(Lahm et al., 2007; Ling et al., 2012; Schuetz et al., 2008), and the deacetylase activity of GOF mutants of HDAC4 and HDAC7 was confirmed (Figure S1C, lanes 4 and 6). The GOF mutant is an ideal negative control for reader function because the active mutant is still capable of binding to a substrate, but only transiently, assuring that mutation maintains protein structure and function. Next, Flag-tagged WT or GOF mutant HDAC4, 5, 7 and 9 were overexpressed in HEK293 cells and then immunoprecipitated to assess NCoR binding. Consistent with prior data (Bottomley et al., 2008b; Gaur et al., 2016; Ling et al., 2012), both wild type and GOF mutant HDAC4 bound NCoR (Figure 1A, lanes 2 and 4), which was disrupted by SAHA (Figure 1A, lane 3 and 5). Similar to HDAC4, HDAC5 and 7 showed similar strong NCoR binding (Figure 1B and 1C, lanes 2 and 4), with reduction in the presence of SAHA (Figure 1B and 1C, lanes 3 and 5). GOF mutants of HDAC4, 5 and 7 showed more dramatic SAHA-dependent loss of HDAC-NCoR interaction (Figure 1, lanes 2 versus 3) compared with wild type (Figure 1, lanes 4 versus 5), consistent with prior work showing that SAHA is 300-fold more potent towards the HDAC7 GOF mutant than the wild type (Schuetz et al., 2008). In contrast to the other class IIa HDAC isoforms, neither the wild type nor GOF mutant of HDAC9 bound NCoR (Figure 1D, lanes 2–5). The inability for HDAC9 to bind NCoR is in agreement with a previous study (Joshi et al., 2013). To thoroughly characterize the inhibitory effect of SAHA on NCoR binding, dose-dependent SAHA binding inhibition was assessed with HDAC4 and 7. HDAC4 and 7 were selected as representative class IIa HDAC proteins here, given their role in AR activity, as discussed later. Dose dependent reduction in NCoR binding was observed with both WT and GOF mutants of HDAC4 and 7 (Figure S2E and S2F), with 100 μM necessary to disrupt binding. These results suggest that the binding modes of NCoR to HDAC4 and 7 are similar, with dose dependent HDAC inhibitor-mediated disruption of NCoR binding. Acetyllysine-dependent class IIa HDAC-NCoR binding SAHA and acetyllysine bind to the active site of HDAC proteins in a similar manner, with coordination to the catalytic metal (Darkin-Rattray et al., 1996; Finnin et al., 1999; Vannini et al., 2007). Given their similar active site binding, acetyllysine-containing peptides might disrupt NCoR binding to the class IIa HDAC proteins similar to SAHA, which would be consistent with possible reader function. To explore the possibility that acetyllysine containing peptides cause HDAC-NCoR dissociation, similar to the SAHA inhibitor, peptides derived from histone 4 lysine 12 (H4K12) were synthesized, including acetylated Ac-Leu-Gly-Lys(Ac)-NH2 (LGKAc) and nonacetylated Ac-Leu-Gly-Lys-NH2 (LGK). These short histone-derived peptides are commonly used in commercial HDAC activity assays, making them an ideal general acetyllysine-containing substrate. NCoR binding experiments were then performed with HDAC4, 5 and 7 in the presences of the peptides. The LGKAc peptide caused a dramatic disruption of HDAC-NCoR complex (Figure 2A–C, lane 4) compared to untreated control samples (Figure 2A–C, lane 2), similar to SAHA (Figure 2A–C, lane 5). Importantly, unacetylated LGK failed to disrupt the HDAC-NCoR complex (Figure 2A–C, lane 3). Quantification of three independent trials showed significant reduction in NCoR binding to all three HDAC isoforms in the presence of the LGKAc peptide, but not the unacetylated LGK peptide (Figure 2E), showing the reproducibility of the experiment. Interestingly, the first version of the peptide that was tested, Ac-Ala-Lys(Ac)-Leu-OH (AKAcL-OH), which contains a free carboxylate terminus, did not disrupt the HDAC7-NCoR complex (Figure S3E). We speculate that electrostatic repulsion between Asp residues at the HDAC7 active site entrance (D759 and D626) (Fischle et al., 2002b; Guenther et al., 2001) and the C-terminal carboxylic acid prevented peptide binding, suggesting that class IIa HDAC proteins prefer binding to internal acetyllysine residues. These results demonstrate that HDAC4, 5 and 7 bind to acetyllysine to dissociate NCoR. To further characterize acetyllysine-mediated NCoR-HDAC disruption, dose-dependent binding experiments were performed, similar to SAHA. NCoR-HDAC dissociation was observed with 1 mM LGKAc peptide (Figure S3F–G). The reduced dissociation potency of LGKAc compared to SAHA (1 mM with LGKAc versus 0.1 mM with SAHA) is consistent with the low micromolar IC50 values reported for acetyllysine peptides as inhibitors (Bradner et al., 2010), but the mid-nanomolar KD and IC50 values reported for SAHA inhibition (Lauffer et al., 2013; Negmeldin et al., 2018; Negmeldin and Pflum, 2017). Binding assessment of the HDAC7-acetyllysine peptide complex A critical prerequisite of reader function is that acetyllysine-containing peptides or proteins bind stably to class IIa HDAC proteins, as was previously observed with bromodomain proteins (Jacobson et al., 2000). Acetyllysine-dependent disruption of the class IIa HDAC-NCoR interaction (Figure 2) suggests stable binding between class IIa HDAC proteins and acetyllysine-containing peptides. Likewise, prior work documenting that acetyllysine-containing peptides acted as inhibitors of class IIa deacetylation using an unnatural trifluoroacetyllysine substrate (Bradner et al., 2010) is also consistent with class IIa HDAC-acetyllysine interactions. However, no report to date characterizes the binding affinity of an acetyl-lysine peptide to a class IIa catalytic domain. To provide evidence for stable interaction between class IIa HDAC proteins and acetyllysine-containing peptides, the binding affinity of the LGKAc peptide to HDAC7 was assessed. We selected HDAC7 for these biophysical studies given the focus of this project on the unique relationship between HDAC7 and AR, as discussed below. The catalytic domain of HDAC7 (cdHDAC7) was expressed in bacteria and purified (Figure S4A–B). Purified cdHDAC7, along with acetylated LGKAc or unacetylated LGK peptides, were subjected to analysis by Bio-Layer Interferometry (BLI). While the LGK peptide did not incur observable binding, even at 5 μM concentration (Figure S4C–E), the LGKAc peptide showed dose dependent association and dissociation with cdHDAC7. The dissociation constant (KD) from three independent trials was in the range from 0.72 ± 0.1 μM to 1.0 ± 0.1 μM (Figure S4C–E). The observed sub-micromolar binding affinity is considerably lower than the 1 mM concentration of LGKAc peptide necessary to dissociate NCoR from class IIa HDAC proteins in lysates (Figure S3F–G). We speculate that the LGKAc peptide might be susceptible to both degradation by proteases and deacetylation by active HDAC proteins in the cell lysate, resulting in the requirement for elevated concentrations. AR K630 is Critical for HDAC7-NCoR Dissociation In addition to stable acetyllysine binding, a critical aspect of reader proteins is that binding by an acetyllysine-containing protein leads to a downstream biological event. HDAC7 regulates AR activity through NCoR-HDAC3 binding (Karvonen et al., 2006), and the transcriptional activity of AR is dependent on acetylation at a critical K630 residue (Fu et al., 2000; Fu et al., 2002). Despite the connections between AR activity, NCoR binding, and K630 acetylation, a molecular mechanism accounting for these activities has not yet been tested. We hypothesize that acetyl-K630 of AR binds directly to the HDAC7 active site to dissociate the NCoR-HDAC3 complex and activate transcription, consistent with a reader function for HDAC7. As a first test of this hypothesis, we assessed whether an acetyllysine-containing peptide derived from AR would disrupt the NCoR-HDAC7 complex, similar to the LGKAc peptide. Two peptides including residues 625 to 637 of the AR sequence were synthesized: acetylated TLGARKAcLKKLGNL (AR K630Ac) and unacetylated TLGARKLKKLGN (AR K630, where K630 is underlined in both sequences). AR K630Ac caused dramatic NCoR dissociation from HDAC7 (Figure 3A, lane 6), whereas AR K630 did not (Figure 3A, lane 2), consistent with a role for AR acetylation in HDAC7-NCoR complex disruption. To further test the hypothesis that AR acetylation affects the HDAC7-NCoR complex, NCoR binding assays were performed in the presence of overexpressed full-length AR wild type or unacetylated K630R mutant. The presence of co-expressed wild type AR resulted in the loss of NCoR in the HDAC7-Flag immunoprecipitation (Figure 3B, lane 7). In contrast, co-expression of unacetylated AR K630R mutant did not disrupt the HDAC7-NCoR complex (Figure 3B, lane 6), consistent with the model where K630Ac binds the HDAC7 active site to dissociate NCoR. Likewise, NCoR remained bound to the GOF HDAC7 mutant in the presence of either wild type or K630R AR (Figure 3B, lanes 2 and 3) due to active deacetylation and loss of “reader” function. Interestingly, AR co-immunoprecipitated with HDAC7 in all samples (Figure 3B, lanes 2–9), which suggests that AR interacts with HDAC7 independently of acetylation-induced active site binding. These finding provide the first evidence that HDAC7 acts as a reader of AR acetylation to affect NCoR-HDAC3 association. An alternative mechanistic hypothesis explaining the AR K630-dependent NCoR-HDAC7 dissociation is that cellular acetyllysine-binding factors, such as transcriptional coactivators, interact with acetylated-K630 AR to physically disrupt NCoR binding. In this case, SAHA or acetyllysine-containing peptides could alter AR acetylation through inhibition of cellular HDAC proteins to influence transcriptional coactivator interaction, instead of direct HDAC7 active site binding, to dissociate NCoR. To experimentally test this alternative hypothesis, overexpressed wild type or GOF mutant HDAC7 were immunoprecipitated in the presence of overexpressed AR and then washed with a high salt (500 mM) buffer to remove unbound cellular proteins, including coactivators and deacetylases. As expected, NCoR remained bound to the samples in this initial immunoprecipitation (Figure S5C), documenting the absence of NCoR-disrupting factors. Next, SAHA was added to the purified AR-HDAC7-NCoR complex to assess if NCoR dissociates through direct HDAC7 binding in the absence of cellular proteins. SAHA incubation resulted in the NCoR dissociation with both wild type (Figure 3C, compare lanes 2 and 3) and GOF mutant HDAC7 (Figure 3C, compare lanes 4 and 5), similar to earlier NCoR binding studies (Figure 1C and 3B). Quantification of NCoR protein levels from three independent trials documented reproducible SAHA-dependent NCoR loss (Figure 3D). Without the influence of cellular coactivators or deacetylases, the data are consistent with direct binding of SAHA to the HDAC7 active site, which further confirms the reader function of HDAC7. Finally, to confirm the interaction between HDAC7 and AR in cells, their cellular localization was studied using immunohistochemistry and confocal microscopy. As expected, HDAC7 showed strong colocalization with AR in the absence and presence of dihydrotestosterone (DHT) ligand (Figure S6), which is consistent with earlier coimmunoprecipitation data (Figure 3B). Interestingly, AR and HDAC7 resided in both the nucleus and cytoplasm without DHT treatment, whereas they are found primarily in the nucleus with DHT treatment (Figure S6), similar to a previous report (Karvonen et al., 2006). The colocalization data is consistent with acetylation-independent interaction of HDAC7 and AR. AR transcriptional activity is dependent on HDAC7 With AR acetylation-dependent disruption of the HDAC7-NCoR complex established, the next step to test the reader function of HDAC7 was to assess HDAC7-dependent AR transcriptional activation function. AR-mediated transcription was monitored in the presence of HDAC7 using a gene reporter system. HEK293 cells were selected due to the absence of endogenous AR expression (Alimirah et al., 2006), which allows expression of wild type and mutant AR. As expected, AR-dependent transcription increased roughly 10-fold in the presence of DHT ligand (100 ± 10 %, Figure 4A, lane 2), compared to the absence (10 ± 1%, Figure 4A, lane 1). Whereas co-expression of wild type HDAC7 did not influence DHT-induced transcription (98 ± 1%, Figure 4A, lanes 2 and 3), co-expression of GOF mutant HDAC7 significantly decreased transcription (56 ± 1%, Figure 4A, lane 4). These observations are consistent with earlier co-immunoprecipitation data where GOF mutant HDAC7 was unable to disrupt NCoR binding (Figure 3B), resulting in continued repression of transcription through HDAC3 recruitment. To confirm the importance of AR K630 acetylation in transcriptional activation, as observed previously (Fu et al., 2002), AR-mediated transactivation activity was reduced to basal levels with co-expression of AR K630R (Figure 4A, lanes 7–12), showing acetylation-dependent AR transcription. As a positive control, co-expression of HDAC4 inhibited AR activity (43 ± 1 %, Figure 4A, lane 6), likely due to AR sumoylation (Yang et al., 2011). As a negative control, the presence of HDAC9 GOF mutant did not affect AR transcriptional activity (93 ± 3 %, Figure 4A, lane 5). This transcriptional data document the dependence of AR gene expression activity on both HDAC7 and K630, consistent with a reader function for HDAC7. To confirm the unique role of HDAC7 in AR-mediated transcription, all class IIa HDAC proteins were tested in the reporter assay. Whereas HDAC7 showed transcriptional signal dependent on deacetylase activity (Figure 4B, lanes 7 and 8), the other class IIa HDAC proteins did not (Figure 4B, lanes 3–6, 9–10). The reporter assay data provide evidence that HDAC7 uniquely acts as a reader of AR acetylation. NCoR represses transcription primarily by recruiting HDAC3 to genomic DNA and deacetylating nucleosomal histones (Fischle et al., 2002b). To test the HDAC3 dependence of AR transcriptional activity, the same AR-dependent gene reporter assay was performed in the presence of HDAC3-selective inhibitor, RGFP966. Whereas GOF mutant HDAC7 reduced DHT-induced AR transcription (Figure 4C, lane 4) compared to wild type (Figure 4C, lane 3) or untreated (Figure 4C, lane 2) samples, RGFP966 restored AR-mediated transcription in the presence of GOF mutant HDAC7 to levels similar as untreated samples (Figure 4C, lane 6). As a control, the HDAC1/2 dual selective inhibitor SHI:12 did not influence GOF mutant repressed AR-mediated transcription (Figure 4C, lane 5), despite prior work showing that AR associates with HDAC1 or HDAC2 to repress transcription (Chng et al., 2012). The HDAC inhibitor study confirmed that AR-mediated gene repression by HDAC7 is dependent on HDAC3, consistent with recruitment of the NCoR-HDAC3 complex. To extend the study of HDAC7-dependent AR transcriptional activity to a biologically-relevant context, mRNA levels of two AR-regulated genes, SPRF5 and Wnt16 (Tanner et al., 2011), were assessed in the presence of expressed AR and HDAC7 using reverse transcription/polymerase chain reaction (RT-PCR) analysis. Upon DHT treatment, both genes showed robust upregulation (Figure 5A, lanes 4–6) compared to the untreated control (Figure 5A, lanes 1–3). While the presence of wild type HDAC7 only modestly reduced mRNA levels of both genes (Figure 5A, lanes 7–9), the presence of HDAC7 GOF mutant significant reduced mRNA levels of both genes (Figure 5A, lanes 10–12). Quantification of triplicates from three independent trials showed that GOF mutant HDAC7 reduced SPRF5 expression to 44 ± 1 % (Figure 5B, column 4) and Wnt16 expression to 40 ± 1 % (Figure 5C, column 4) compared to untransfected cells (Figures 5B and 5C, column 2). The RT-PCR results confirmed that HDAC7 reader function is required for transcription of AR-dependent cellular genes. ER and AR Similarly Interact with HDAC7 and NCoR Finally, we extended these studies to a second nuclear receptors to test if HDAC7 reader function affects other nuclear receptor family members. Estrogen receptor α (ER) is a transcriptional activator regulated by binding to ligands, such as estradiol (E2) (Heldring et al., 2007). Similar to AR, ER transcriptional activity is influenced by HDAC4 and HDAC7 through direct interaction (Leong et al., 2005; Malik et al., 2010). Additionally, ER recruits NCoR/SMRT corepressors to regulate transcriptional function (Fu et al., 2003; Varlakhanova et al., 2010). Despite the connections between ER, class IIa HDAC proteins, and NCoR/SMRT corepressors, similar to AR, no study has yet directly probed their possible acetylation-dependent interactions. Based on the AR data, we hypothesized that ER acetylation affects the HDAC7-NCoR complex. To probe this hypothesis, NCoR binding assays were performed in the presence of overexpressed full-length ER wild type and either HDAC7 WT or gain-of-function (GOF) mutant. NCoR bound HDAC7 GOF mutant to a greater extent than HDAC7 WT (Figure 6, lanes 2 vs 4), which was disrupted in the presence of SAHA (Figure 6, lanes 3 and 5). Similar to AR, ER robustly bound to HDAC7 WT or GOF in the absence or presence of SAHA treatment (Figure 6, lanes 2–5), suggesting that ER also interacts with HDAC7 independently of acetylation-induced active site binding. In fact, immunohistochemistry documented that HDAC7 colocalized with ER in the absence and presence of ligand (Figure S9B). In this case, ER predominantly resided in the nucleus with HDAC7 both in the presence and absence of E2 ligand, which is consistent with previous work (Monje et al., 2001). Based on the similar data, we speculated that both AR and ER interact with HDAC7 to regulate acetylation-dependent NCoR binding. Discussion HDAC proteins are epigenetic repressors due to deacetylation of nucleosomal histone substrates (Krämer et al., 2001). However, class IIa HDAC proteins cannot function directly as acetylation-dependent epigenetic enzymes since they are catalytically inactive (Bradner et al., 2010; Lahm et al., 2007; Schuetz et al., 2008). Prior work documented that class IIa HDAC4, 5 and 7 interact with the repressive NCoR/SMRT-HDAC3 complex to regulate transcription (Fischle et al., 2001; Mottis et al., 2013; Wong et al., 2014). Here we demonstrate for the first time that binding of HDAC4, 5 and 7 to NCoR is acetyllysine-dependent, which is consistent with possible “reader” function. The “reader” model proposes that binding of acetyllysine into the active sites of HDAC4, 5 and 7 releases the NCoR/SMRT-HDAC3 complex (Figure 7A). By controlling HDAC3-mediated deacetylation, HDAC4, 5 and 7 indirectly control gene expression. Several prior studies are consistent with the proposed model of HDAC7 reader function. An HDAC inhibitor disrupted class IIa HDAC binding to an NCoR/SMRT-derived peptide (Hudson et al., 2015), which suggests that active site binding molecules (either an inhibitor or acetyllysine) affect HDAC-NCoR association. Similarly, HDAC3 dissociated from HDAC4 after incubation with an HDAC inhibitor, which could be due to the loss of NCoR association after active site binding by the inhibitor (Bottomley et al., 2008b). Deacylation by class IIa HDAC proteins was inhibited by an acetyllysine-containing substrate (Bradner et al., 2010), which is consistent with HDAC active site binding by acetyllysine. Finally, chemoproteomics studies reported that class IIa HDAC proteins (HDAC4, 5, and 7), but not NCoR or HDAC3, were enriched by inhibitor-bound beads (Bantscheff et al., 2011; Lobera et al., 2013). In total, prior reports are consistent with the proposed model where HDAC7 acts as a reader of AR acetylation to modulate gene expression. To thoroughly characterize the reader function of HDAC7, the dissociation constant of an acetyllysine peptide bound to HDAC7 was determined to be 0.7–1.0 μM using BLI analysis. Although no prior literature reports the binding affinity of a class IIa HDAC to an acetyllysine-containing peptide, the inhibitory potency (IC50) of acetyllysine-containing peptides in deacetylase assays with class IIa HDAC proteins is 0.64 and 2.3 μM (Bradner et al., 2010), similar to the binding data. To compare the affinities observed here with HDAC7 to bromodomain-containing reader proteins, dissociation constants of 1.4 – 39 μM were determined using isothermal calorimetry for acetyllysine-containing peptides bound to bromodomain-containing TAFII250, which showed a 1:1 stoichiometry (Jacobson et al., 2000). NMR titration was also used to assess the binding affinity of acetyllysine-containing peptides to bromodomain-containing P/CAF (350 μM), GCN5 (900 μM), and CBP (50 μM) proteins (Dhalluin et al., 1999; Hudson et al., 2000; Mujtaba et al., 2004). In total, the affinity of HDAC7 for an acetyllysine-containing peptide observed here is similar or better than those reported for bromodomain-containing reader proteins, further corroborating possible reader function. Prior HDAC7 structural studies proposed that acetyllysine-containing substrates interact with both the active site and the nearby structural zinc metal region that binds SMRT/NCoR (Desravines et al., 2017; Schuetz et al., 2008). Similarly with HDAC4, a SMRT-derived peptide bound a cleft between the structural zinc domain and the active site (Kim et al., 2015; Park et al., 2018). The proximity of the structural zinc region to the active site suggests that SMRT/NCoR-class Ila HDAC protein complexes are sensitive to acetyllysine binding. In addition, inhibitor binding to the HDAC4 active site resulted in conformational changes in the structural zinc region, which could explain the inhibitor-dependent HDAC4-NCoR/SMRT disruption (Bottomley et al., 2008b). Based on these prior structural studies, a conformational change in the NCoR/SMRT interaction domain after active site binding, or a blocking of the active site cleft, or both, might be responsible for release of NCoR/SMRT. Further structural studies are needed to test this hypothesis. Whereas HDAC4, 5, and 7 showed inhibitor and acetyllysine sensitive NCoR association (Mottis et al., 2013; Wong et al., 2014), HDAC9 did not. In fact, there is debate in the literature regarding the HDAC9-NCoR/SMRT interaction. Ectopically expressed HDAC9 bound NCoR by co-immunoprecipitation (Petrie et al., 2003), and HDAC9 bound an NCoR-derived peptide, similar to HDAC4, 5, and 7 (Hudson et al., 2015). However, proteomics analysis after immunoprecipitation documented that HDAC4, 5 and 7 exist in a cluster of proteins containing NCoR, whereas HDAC9 was in a separate cluster absent of NCoR (Joshi et al., 2013). Additional studies with HDAC9 are needed to identify reader activity, which might be distinct from HDAC4, 5, and 7 To demonstrate the biological relevance of HDAC7 reader function, we provide evidence that HDAC7 acts as a reader of AR acetylation to activate AR-mediated transcription. According to the proposed model, unacetylated AR binds to the HDAC7-NCoR-HDAC3 complex (Figure 7B). In this transcriptionally inhibited state, HDAC7 acts as a regulatory scaffold to recruit the repressive NCoR-HDAC3 complex to AR-bound genomic DNA. Upon acetylation of AR at K630, the newly formed acetyllysine residue binds into the active site of HDAC7 to dissociate the NCoR-HDAC3 complex from AR (Figure 7C). Without histone deacetylation by HDAC3, the presence of AR results in activation of transcription (Figure 7C), likely through binding to coactivator proteins, such as acetyltransferases. With this model, AR acts both as a DNA-binding transcriptional activator and as a recruiter of repressive HDAC7-NCoR-HDAC3 to influence transcription via two complementary mechanisms. In support of this model is the fact that activation of AR by DHT treatment induced nuclear localization of both AR and HDAC7 (Figure S6). An interesting observation in these studies is that AR binds to HDAC7 independently of acetylation (Figure 3B). Similarly, a prior co-localization study demonstrating that AR and HDAC7 interact independently of NCoR (Karvonen et al., 2006). Taken together, HDAC7 reader function might be facilitated by “pre-binding” of substrates, which ensures that acetyllysine is in proximity to the active site for efficient binding. In fact, previous mass spectrometry data indicated that most acetylation occurs with a low median stoichiometric ratio of 0.02% (Hansen et al., 2019). Similarly, the acetyllysine levels of AR after immunoprecipitation were low or unobservable (Figure S11), consistent with a low stoichiometry. Given the low AR acetylation levels, yet prominent effects by K630 on transcription, we hypothesize that pre-binding of AR to HDAC7 promotes a high local concentration of acetylated K630 on AR near the HDAC7 active site to facilitate binding and NCoR release. In support of this hypothesis, prior work with bromodomains and HDAC8 documented the role of interactions outside of the acetyllysine binding domains in reader function (Castaneda et al., 2017; Mujtaba et al., 2002; Owen et al., 2000). Additional reader substrates of class IIa HDAC proteins will need to be identified to explore whether acetylation-independent pre-binding is a common feature of their acetyllysine binding function. Acetylation is critical for AR-mediated transcription since K630R or K632/633R mutants of AR were weak transcriptional activators (Fu et al., 2003; Fu et al., 2000; Fu et al., 2002). In addition, wild type AR displayed 10 times worse NCoR binding than the K630A AR mutant (Fu et al., 2002), which is consistent with acetylation-dependent NCoR-AR interaction and the proposed model (Figure 7B). Related to the HDAC7-AR interaction, fluorescence staining showed that HDAC7 and AR colocalized upon treatment of testosterone, although without NCoR (Karvonen et al., 2006). Hollenberg et al. suggested a direct interaction between AR and NCoR (Cheng et al., 2002; Hodgson et al., 2005), although their study did not preclude the possibility that AR exerts its interaction with NCoR through scaffolding HDAC7. In total, prior data on AR-mediated transcriptional regulation is consistent with the proposed model involving HDAC7 reader activity. Many nuclear receptors, including estrogen receptor (ER) and thyroid receptor (TR), share a common acetylation motif with AR (Wang et al., 2011). Acetylation is also critical to regulate transcription by ER and TR (Sanchez-Pacheco et al., 2009; Wang et al., 2001). We speculate that additional receptors interact with class IIa HDAC proteins to mediate acetylation-dependent transcriptional activation. In fact, here we provide evidence that HDAC7-NCoR disruption by SAHA occurs in the presence of ER (Figure 6), similar to AR. Like HDAC7, HDAC5 and HDAC9 directly interacted with ER to inhibit its transcriptional activity by recruiting NCoR/SMRT (van Rooij et al., 2010). Also similar to AR, ER requires K266 and K268 for full transcriptional activity (Wang et al., 2001), suggesting that one or both of these sites are involved in class IIa reader binding. TR also interacts with SMRT dependent on the presence of three lysine residues (Sanchez-Pacheco et al., 2009). Future studies are needed to fully understand the relationship between class IIa HDAC proteins and the transcriptional activity of the hormone receptor family. HDAC4, 5, and 7 dissociated from NCoR dependent on both inhibitor and acetyllysine binding (Figures 1 and 2), consistent with prior work (Mottis et al., 2013; Wong et al., 2014), suggesting that all three isoforms have reader functions. However, only HDAC7 affected the transcriptional activity of AR as a function of reader activity (Figure 4B). The combined data suggest that HDAC7 uniquely collaborates with AR to regulate transcriptional activity. We hypothesize that HDAC4, 5, and 7 cooperate with different transcription factors to modulate epigenetic regulatory mechanisms. Additional studies are needed to identify the unique transcription factors and/or enhancer complexes coordinating with HDAC4, 5, and 7 for reader-mediated epigenetic regulation. Significance The first evidence of the epigenetic “reader” function of HDAC7 is reported. While prior work proposed an acetyllysine-dependent reader function for class IIa HDAC proteins (Arrowsmith et al., 2012; Bradner et al., 2010; Lobera et al., 2013), here experimental data documents that HDAC7 influences NCoR-HDAC3- and K630-dependent AR transcriptional activity, consistent with the proposed model (Figure 7). To further test this model of reader function, future work will focus on identifying additional acetyllysine-binding substrates of class IIa HDAC proteins. Given the biological importance of reader proteins, such as bromodomains (Smith and Zhou, 2016), and the use of HDAC inhibitors in the clinic (Eckschlager et al., 2017), a full understanding of the epigenetic reader functions of class IIa HDAC proteins will be critical to fully characterize the roles of HDAC proteins in cell biology, in addition to assisting HDAC-targeted drug development efforts. Limitations of the study As the initial study reporting a reader function of a Class IIa HDAC protein, additional studies are needed to further test the proposed reader mechanism. In fact, a limitation of this study was the need to use overexpressed wild type and mutant HDAC and AR proteins. Overexpressed proteins are commonly used to overcome low abundance or allow use of mutants (Bottomley et al., 2008a; Fischle et al., 1999; Guenther et al., 2001; Karvonen et al., 2006; Kim et al., 2015; Yang et al., 2011), and results with endogenous or stably expressed proteins have reproduced those with overexpressed proteins (Fischle et al., 2002a; Karvonen et al., 2006; Yang et al., 2011). Future work exploring the reader functions of Class IIa HDAC proteins using endogenous proteins in additional biological contexts will provide necessary confirmation. STAR Methods RESOURCE AVAILABILITY Lead Contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Mary Kay H. Pflum (Pflum@wayne.edu) Materials availability Plasmids generated in this study are available upon request. Data and Code availability All repetitive trials and raw data are reported in the supplementary information file. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. EXPERIMENTAL MODEL AND SUBJECT DETAILS Cell line HEK293 cells (human embryonic kidney cells, ATCC; authenticated annually using short tandem repeat profiling) were grown in Dulbecco’s Modified Eagle’s Medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (Life technologies) and 1% antibiotic/antimycotic (Hyclone) at 37°C in a 5% CO2 incubator. Jetprime reagent (VWR) was used for transfection of plasmid DNA (2 μg) into HEK293 cells (2 × 106 cells) at 70% confluency. When HDAC7 and AR expression constructs were co-transfected, both plasmids (2 μg of each) were used with the same conditions. After a 24 h growth period, cells were treated with SAHA (10 μM) for 24 h to induce robust acetylation, and then harvested and washed once with DPBS (Dulbecco’s Phosphate Buffered Saline, Hyclone, 10 mM Na2HPO4, 1.8 mM KH2PO4 pH 7.4, 137 mM NaCl, 2.7 mM KCl). Cells were either used immediately or stored at −80 °C as a cell pellet. METHOD DETAILS Plasmids pcDNA3.1-Flag HDAC4, 5, 7 and 9 were described earlier (Fischle et al., 1999; Yang et al., 2011; Yuan et al., 2010; Zhang et al., 2004). GOF mutants of HDAC4, 5, 7, and 9 were created using QuickChange site-directed mutagenesis (Agilent, 200521) with primers in Figure S1. Sequences were confirmed by DNA sequencing. pcDNA3-AR and pcDNA3-AR K630R were described earlier (Fu et al., 2002). pLR-ARE-LUC and pLR-Renilla-LUC plasmids were described earlier (Zhang et al., 2016). pcDNA-HA-ER WT was a gift from Sarat Chandarlapaty (Addgene plasmid number 49498). HDAC Deacetylase Activity Assay Transfection of plasmid DNA (pBJ5-Flag wild type HDAC1 or HDAC1 C151A (2 μg))(Wambua et al., 2014), pcDNA3.1-Flag wild type HDAC4 or HDAC4 H976Y (2 μg)), pcDNA3.1-Flag wild type HDAC7 or HDAC7 H842Y (2 μg)) into HEK293 cells (2×106) was carried out as described. Following transfection, cells were lysed using lysis buffer (500 μL; 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 10% glycerol, and 0.5% Triton X-100) containing 1× protease inhibitor (GenDepot) for 30 min at 4 °C with rocking. Cell debris was removed by centrifugation (13.2 × 103 rpm) for 10 min at 4 °C and total protein concentrations from the soluble fraction was determine using Bradford reagent (Bio-Rad, cat# 5000205). Lysates (1 mg total protein) containing Flag-tagged HDAC proteins were immunoprecipitated using prewashed Anti-Flag M2 beads (25 μL) as described. Proteins after immunoprecipitation (20% of total) were assayed using HDAC-Glo™ kit (G648B-C, Promega) following the manufacturer’s instruction. Briefly, bound beads were washed with lysis buffer twice. The HDAC-Glo™ substrate (1 mL) and developer (1 μL) were first premixed to form the HDAC-Glo™ reagents. Then, to monitor deacetylase activity, the HDAC-Glo™ reagent (2 μL for HDAC1 and 1μL for HDAC4 and HDAC7) and HDAC-Glo™ buffer (48 μL) were added to the prewashed bound beads and incubated for 25 min at room temperature with rocking. Deacetylase activity was measured as luminescent signal using a GeniosPlus Fluorimeter (Tecan) at optimal gain. The remaining of the precipitated HDAC proteins (80% of total) was washed three times with lysis buffer, followed by 10% SDS-PAGE separation of proteins, transfer to a PVDF membrane, and visualization with Flag (F3165, Sigma) or an NCoR (5948S, Cell Signaling) antibodies using FluorChemQ gel imager (ProteinSimple). Immunoprecipitation Immunoprecipitation was performed as previously described with minor changes (36, 37). Following transfection as described above, cells were lysed using lysis buffer (500 μL; 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 10% glycerol, and 0.5% Triton X-100) containing 1× protease inhibitor (GenDepot) for 30 min at 4 °C with rocking. Cell debris was removed by centrifugation (13.2 × 103 rpm) for 10 min at 4 °C and total protein concentrations from the soluble fraction was determine using Bradford reagent (Bio-Rad, cat#5000205). For the immunoprecipitation, lysates containing wild type or mutant Flag-tagged HDAC4, 5, 7 or 9 (2 mg of total protein, 1 mL of total volume) were mixed with pre-washed Flag M2 beads (25 μL bead slurry, Sigma) and rocked for 3 h at 4°C). For SAHA or acetyllysine competition during immunoprecipitation (Figures 1–3), an equal amount of the same lysate sample was immunoprecipitated as described in the presence of active site inhibitor SAHA (100 μM or concentration indicated in Figure S2 in <2% DMSO), Ac-LGKAc-NH2 peptide (1 mM or concentration indicated in Figure S3 in ultra-pure water; 95% purity; MALDI-TOF mass spectrometry - expected [M+H]+ = 400.45; observed [M+H]+ = 400.30), or Ac-LGK-NH2 peptide (1 mM or concentration indicated in Figure S3 in ultra-pure water; 95% purity; MALDI-TOF mass spectrometry - expected [M+H]+ = 358.45; observed [M+H]+ = 358.25), AR K630Ac peptide (1 mM or concentration indicated in Figure S5 in ultra-pure water; 95% purity; MALDI-TOF mass spectrometry - expected [M+H]+ = 1495.92; observed [M+H]+ = 1496.07), AR K630 peptide (1 mM or concentration indicated in Figure S5 in ultra-pure water; 96% purity; MALDI-TOF mass spectrometry - expected [M+H]+ = 1453.91; observed [M+H]+ = 1454.02). To assess AR acetylation levels (Figure S11), lysates (1 mg of total protein, 500 μL of total volume) were mixed with pre-washed Protein A/G beads (50 μL bead slurry, Santa Cruz Biotechnology) and primary AR antibody (sc-7305 Santa Cruz Biotechnology) diluted according to the manufacturer protocol. After immunoprecipitation, beads were washed three times with wash buffer (1 mL; 50 mM Tris-Cl pH 8.0, 500 mM NaCl, 10% glycerol, 0.5% triton-X-100) with centrifugation (5,000 rcf for 1 min) to collect the beads after each wash. For SAHA competition after immunoprecipitation (Figure 3C–D), beads after the initial washes were further washed with wash buffer (1 mL) supplemented with SAHA (100 μM in >0.002% DMSO) or DMSO carrier (>0.002%), with centrifugation (5,000 rcf for 1 min) to collect the beads after each of three washes. Following washing, bound proteins were eluted and denatured by boiling for 8 min at 95°C in SDS loading dye (Bio-Rad, Laemmli Sample Buffer, 65.8 mM Tris-HCl, pH 6.8, 26.3% (w/v) glycerol, 2.1% SDS and 0.01% bromophenol blue) supplemented with β-mercaptoethanol (10% v/v). Proteins were separated by 10% SDS-PAGE, followed by transfer to a PVDF membrane (Immobilon P, Millipore) and visualization with Flag (F3165, Sigma), NCoR (5948S, Cell Signaling), AR (5153S, Cell Signaling), ER (8644S, Cell Signaling), or acetyllysine (9441S, Cell Signaling) antibodies using the FluorChemQ gel imager (ProteinSimple). Expression and Purification of the Catalytic Domain of Human Histone Deacetylase 7 The expression and purification of the Catalytic Domain of Human HDAC7 (cdHDAC7) procedure was adapted from a previously published protocol (Schuetz et al., 2008). The bacterial expression construct encoding His-tagged cdHDAC7 DNA in the pET28a-LIC backbone was obtained from Addgene (plasmid #51340) and was transformed into Rosetta2(DE3) competent cells using the heat shock method. Terrific broth (1 L, 1.2% Tryptone, 2.4% Yeast Extract and 0.5% Glycerol) supplemented with HEPES (100 mM, pH 7.5), kanamycin antibiotic (30 g/mL), and Zn(OAc)2 (100 μM) was inoculated with the transformed cells and grown at 37°C with shaking (250 rpm). When the optical density (OD) of the cell culture reached roughly 1.0, the temperature was lowered to 20°C and protein expression was induced by addition of isopropyl β-d-1-thiogalactopyranoside (IPTG, 0.4 mM). After 16 hr of protein expression at 37°C with shaking (250 rpm), the cell pellet was collected by centrifugation (5,000 rpm) at 4°C for 20 min and then suspended in lysis buffer (50 mL; 25 mM Tris pH 8, 500 mM NaCl, 5% Glycerol, 0.1% CHAPS, 1 mM DTT and 0.1 mM PMSF). Cells were lysed by sonication (6 × 30 s pulses with 30 s pauses in between), and then the cell debris after lysis was removed by centrifugation (40,000 rpm) for 30 min. For protein purification, Ni-NTA beads (4 mL, Qiagen, cat #30450) were first washed with washing buffer (40 mL; 25 mM Tris pH 8, 150 mM NaCl, 5 % Glycerol and 20 mM Imidazole). The soluble fraction of the lysate was added to the washed Ni-NTA beads and rocked for one hour at 4°C to allow protein binding. After binding, the unbound proteins from the lysate were allowed to flow through the column, followed by washing of the beads five times with washing buffer (10 mL) at room temperature. Bound proteins were eluted from the beads with elution buffer (4 mL; 25 mM Tris pH 8, 150 mM NaCl, 5% Glycerol and 300 mM Imidazole) five times at room temperature. The combined elution solution was loaded onto a HiTrap Q HP anion exchange column (GE Healthcare, cat #17-1154-01) using ÄKTA P-920 purification system. The flow rate was set to 1 mL/min and a gradient from 50 mM to 400 mM NaCl in HEPES was used to elute bound proteins. Protein purification was assessed using 10 % SDS-PAGE separation and visualization of proteins by Coomassie blue staining (Bio-Rad, cat #1610436). Fractions containing pure cdHDAC7 were combined and dialyzed (14 kDa cut off, Sigma Aldrich, cat #D9777) against HEPES buffer (25 mM HEPES pH 8, 200 mM NaCl) containing 10 % glycerol in preparation for biotinylation and bio-layer interferometry instrument (BLI) analysis. Biotinylation of purified cdHDAC7 Purified cdHDAC7 was biotinylated to facilitate binding to the super streptavidin coated sensor (SSC sensor) for BLI analysis. The purified cdHDAC7 protein (0.5 mL of a 60 μM solution) was incubated with biotin-NHS-PEG4 (1.5 μL of a 20 mM solution, Thermo Fisher, cat #A39259) with a 1:1 protein-to-biotin ratio (0.3 μM of each) in a total of 0.5 mL to prevent over-biotinylation. The reaction was incubated on ice for 2 h before dialysis against HEPES buffer containing 10% glycerol using a slid-A-lyzer (10 kDa cut off, Thermo Fisher, cat #66380) to remove the biotin-NHS-PEG4 reagent. Dialyzed samples were fast frozen with liquid nitrogen in aliquots and stored at −80°C before use. Binding Assessment of cdHDAC7-LGK(Ac) using bio-layer interferometry (BLI) In preparation for BLI analysis, the biotinylated cdHDAC7 protein was thawed on ice and subjected to dialysis against HEPES buffer using a slid-A-lyzer (10 kDa cut off, Thermo Fisher, cat #66380) to remove glycerol. BLI analysis was carried out using an Octet Red 96 instrument (FortéBio). All kinetics experiments were performed at 25°C while stirred at 1000 rpm to create homogeneous mixtures. Biotinylated cdHDAC7 (0.1 mg/mL) was loaded onto the SSC sensor (FortéBio, cat #18–5057) until the binding signal reached 7 nm. Acetylated LGK(Ac) peptide was used at 20 nM, 1 μM, and 5 μM concentrations, whereas the unacetylated control peptide LGK was used only at a 5 μM concentration. The binding experiments were performed as follows: initial equilibration after loading was set to 1200 s and the baseline was then stabilized for 60 s. Peptide was then allowed to associate for 1200 s, followed by dissociation for 1200 s. FortéBio Octet Data Acquisition and Analysis Software version 7.0 was utilized to perform data acquisition and processing. Global fitting applied to acetylated LGK(Ac) peptide binding curves. Immunohistochemistry Staining HEK293 cells (3 × 106) were seeded onto coverslips in a 6-well plate and grown in DMEM (2 mL) supplemented with 10% fetal bovine serum (Life technologies) and 1% antibiotic/antimycotic (Hyclone) at 37°C in a 5% CO2 incubator. After 24 h, cells were co-transfected with expression plasmids encoding WT HDAC7-Flag (0.64 μg) and either pcDNA3-AR or pcDNA-HA-ER plasmids (0.64 μg) using Jetprime reagent (VWR) according to the manufacturer protocol. After another 24 h, cells transfected with AR were incubated with DHT ligand (1 nM in 0.1% DMSO) or with the DMSO vehicle (0.1% v/v). Similarly, cells transfected with ER were incubated with E2 ligand (10 pM in 0.1% ethanol) or with the ethanol vehicle (0.1% v/v). Following 24 h incubation with ligands or vehicle, cells were washed with DPBS (2 mL; Thermo Fisher, SH30028FS) once, fixed with 4% paraformaldehyde (2 mL; Santa Cruz, sc-281692) for 10 minutes, washed three times with DPBS (2 mL), and fixed with ice cold methanol (2 mL) for 3 minutes. Fixed cells were washed three times with DPBS (2 mL) and permeabilized with permeabilization buffer (2 mL; 0.3% Triton X-100 in PBS, DPBS) for 10 minutes at room temperature. Following permeabilization, cells were washed three times with DPBS (2 mL) and blocked with SuperBlock™ Buffer (2 mL; Thermo Fisher, cat # 37515) for one hour at room temperature. After the blocking buffer was removed, cells were incubated with mouse anti-Flag (Sigma Aldrich, cat #F3156) and either rabbit anti-AR (Cell Signaling, cat #5153) or rabbit anti-ER antibody (Cell Signaling, cat #8644) overnight in DPBST (2 mL; 0.1% Triton X-100 in DPBS). After incubation with the primary antibodies, cells were washed three times with DPBS and incubated with anti-mouse Alexa Fluor 488 (Thermo Fisher, cat #A-11029) and anti-rabbit Alexa Fluor 594 secondary antibody (Thermo Fisher, cat #A-11037) mix in PBST (2 mL) for one hour at room temperature. Antibody-stained cells were washed with DPBS (2 mL) three times and counterstained with DAPI (2 mL; 300 nM in PBS; Thermo Fisher, cat #D21490) for five minutes. Cells were then washed three times with DPBS (2 mL), mounted on a glass slide with a drop of ProLong® Gold Antifade Reagent (Thermo Fisher, cat #P10144) and visualized using a Leica TCS SP8 inverted confocal microscope at a resolution of 2048 × 2048 and a line averaging of 32. Images were obtained using an Acousto Optical Beam Splitter (AOBS) and an Acousto Optical Tunable Filter (AOTF) with excitation lasers of 405, 488, 552 and 632 nm. Pictures were acquired using the LAS_X acquisition software with detection ranges of 415 to 485 nm for DAPI, 555 to 625 nm for Alexa Fluor 594, and 695–765 nm for Alexa Fluor 488. Pictures in the manuscript and supplemental information are reported exactly as they were obtained. Reporter Gene Assay HEK293 cells (200,000) were seeded in 6-well cell culture plates and grown in phenol red-free DMEM supplemented with 10% charcoal-stripped fetal bovine serum (A3382101, Fisher) at 37 °C in a 5% CO2 incubator. Transfections were performed using JetPrime reagent according to manufacturer instructions using pcDNA3-AR (0.32 μg), pLR-ARE-LUC (0.64 μg), pLR-Ren-LUC (0.32 μg) and HDAC-Flag expression construct (0.64 μg). Five hours after transfection, cells were washed once with DPBS (1 mL). Then, media (3 mL) in the absence (0.1% DMSO) or presence of DHT (1 nM on 0.1% DMSO) was added. Some reactions also included small molecule inhibitor (10 μM of SAHA, SHI:1–2, or RGFP966). Cells were incubated for 36 h before harvesting using passive lysis buffer (500 μL, E1960, Promega) by rocking at room temperature for 15 min. Lysates were generated based on manufacture instruction and immediately assayed using the Dual Luciferase Assay (E1960, Promega) using a Geniosplus Fluorimeter (Tecan). Activity is reported in relative light unit (RLU), which were determined as the ratio of inducible firefly luciferase luminescence from pLR-ARE-LUS divided by the luminescence of the renilla luciferase control from pLR-Ren-LUC. Mean and standard error were calculated from three independent trials. RT-PCR of AR-Regulated Gene Expression HEK293 cells were seeded at a density of 200,000 cells per well in 6-well cell culture plates and grown in phenol red-free DMEM supplemented with 10% charcoal-stripped fetal bovine serum (Fisher, cat #A3382101) at 37 °C in a 5% CO2 incubator. After growth of 24 h, cells were transfected with pcDNA3-AR (0.32 μg) and either pcDNA3.1-Flag HDAC7 or HDAC7 H842Y (0.64 μg), as described. After 5 h of growth, the media was removed and replaced with media in the absence (0.1% DMSO) or presence of DHT (1 nM on 0.1% DMSO). Cells were grown for another 36 h before media was removed and TRIzol™ reagent (800 μL, Thermo Fisher, cat #15596026) was added to the cell directly to denature the protein. The resulting mixture was diluted with chloroform (200 μL), followed by gentle shaking and incubation for 5 min. Cell debris was precipitated by centrifuging at 12,000 rpm for 10 min at 4°C. The supernatant after centrifugation (300 μL) was diluted with isopropanol (300 μL) and cooled at −20°C for 1 hr, followed by centrifugation at 4°C at 12,000 rpm for 10 min. The supernatant was removed carefully followed by washing of the RNA pellet with 70% ethanol (700 μL). The precipitated RNA was dried at RT for 5 min, followed by reconstitution with dd water (150 μL). RNA quantity was determined at 260 nm using a NanoDrop™ Spectrophotometer. RNA (700 ng) was mixed with random primer (1 μL, New England Biolabs, cat #S1330S), dNTP (1 μL, 10 mM), and ddH2O in a total volume of 10 μL. The RNA/Primer mixture was denatured for 5 min at 65°C, followed by addition of 10X amplification buffer (2 μL, New England Biolabs, cat #B0537S), RNase inhibitor (0.5 μL, New England Biolabs, cat #M0314S), and reverse transcriptase (0.25 μL, New England Biolabs, cat #M0380S) in a total volume of 20μL. Reverse transcription occurred by incubating the mixture for 5 min at 25°C for annealing and 30 min at 55°C for synthesis. The resulting cDNA (1 μL) was mixed with forward and reverse primers (1.5 μL of each, 10 μM for each primer) for the indicated genes (Table S1) and Taq 2X Master Mix (7.5 μL, New England Biolabs, cat #M0270L) in a total volume of 15 μL. The PCR protocol entailed initial denaturation for 5 min at 95°C, followed by cycles of 1 min at 55°C for annealing, 1 min at 68°C for extension, and 1 min at 95°C for denaturation. After 35 cycles for SFRP5 and Wnt16, or 25 cycles for GAPDH, a 7 min final extension at 68°C was performed. The PCR product was resolved on a 2% agarose gel supplemented with ethidium bromide (Sigma Aldrich, E1510) and visualized with a ChemiDoc gel imager (Bio-Rad). Band intensities were analyzed using ImageJ (Schneider et al., 2012). Intensities of each band was normalized against the GAPDH band by dividing the target gene RT-PCR band intensity by the corresponding GAPDH band intensity. DHT induction sample bands were averaged and set to 100% for comparison. All the other samples were normalized based on the DHT induction samples to give the relative intensity. QUANTIFICATION AND STATISTICAL ANALYSIS To quantify protein levels, bands in the gel images were quantified using ImageJ (Schneider et al., 2012); the whole lane was selected and plotted to show the intensity of the protein bands in that lane, which was then quantified by measuring the area under the curve above the background. The band intensities were normalized to one of the samples to generate normalized percentage values, with the raw data shown in the supplementary information. For the gene expression assays, mean light intensity data were normalized as a percentage of the DHT sample, with the raw data shown in the supplementary information. As indicated in the figure legends, the mean and standard error of at least 3 independent trials (all trials are provided as supplementary information) are shown in all figures. Student t-test analysis was applied to the normalized percentage values using Prism software (GraphPad Software, Inc.). One-tailed (Figure 3D) or two-tailed (all other figures) p values were determined from at least 3 three independent experiments using a 95% confidence interval with N.S. = not significant (p > 0.05), * = p < 0.05, ** = p < 0.01, *** = p < 0.001, and *** = p < 0.0001. Supplementary Material 2 Acknowledgement: We thank the National Institutes of Health (GM121061 and GM131821) and Wayne State University for funding, X. Zhang (Kamanos Cancer Institute, Wayne State University) for HDAC4, 5, 7, and 9 expression plasmids and technical support, R. G. Pestell (Pennsylvania Cancer and Regenerative Medicine Center) for the AR and AR K630R expression plasmids, J. T. Koh (University of Delaware) for the ARE-LUC and Ren-LUC gene reporter plasmids, and I. Gomes, K. Herath, and J. Knoff for comments on the manuscript. Figure 1. HDAC inhibitor-dependent HDAC-NCoR binding. (A) Flag-tagged wild type (WT) or GOF HDAC4 (A), HDAC5 (B), HDAC7 (C), or HDAC9 (D) were overexpressed in HEK293 cells, which were then treated with SAHA (10 μM) to induce acetylation. After lysis, Flag-tagged HDAC proteins were immunoprecipitated (IP) in the presence or absence of SAHA. Bound proteins were resolved by SDS-PAGE, followed by western blot analysis with NCoR and Flag antibodies. As a gel migration control, lysate (Lys) from transfected cells were included. All trials include a bead binding control using lysates without expression of HDAC-Flag (IgG). Activity assays associated with the wild type and mutant HDAC proteins are shown in Figure S1. Additional independent trials and SAHA dose dependence are shown in Figure S2. Figure 2. NCoR binds to HDAC4, 5, and 7 in an acetyllysine-dependent manner. Flag-tagged HDAC4 (A), HDAC5 (B), and HDAC7 (C) were overexpressed in HEK293 cells, which were then treated with SAHA (10 μM) to induce acetylation. After lysis, anti-Flag beads were used to immunoprecipitate (IP) the HDAC proteins either in the absence or presence of 1 mM LGK (K), LGKAc (KAc), or SAHA (SA). Bound proteins from the IP were resolved by SDS-PAGE, followed by Western blot analysis with NCoR and Flag antibodies. Lysate (Lys) from transfected cells were included as a gel migration control. All trials include a bead binding control using lysates without expression of HDAC-Flag (IgG). Additional independent trials and dose dependence are shown in Figure S3A–C. Binding affinity studies are shown in Figure S4. (D) NCoR band intensities from three independent trials with HDAC4, 5 and 7 (parts A-C and S73A-C) were quantified and normalized to untreated (set to 100%), with the student t test applied to show significance. N.S. = not significant (p > 0.05), and ** = p < 0.01. Raw data are shown in Figure S3D. Figure 3. AR influenced HDAC7-NCoR association. (A) Flag-tagged HDAC7 was expressed in HEK293 cells, which were then treated with SAHA (10 μM) to induce acetylation, followed by lysis, immunoprecipitation (IP) in the presence of different mM concentrations of AR K630Ac or AR WT peptide, SDS-PAGE separation, and western blot analysis with NCoR and Flag (HD7-F) antibodies. Repetitive independent trials are shown in Figure S5A. (B) HDAC7 wild type (WT) or GOF mutant were co-expressed with AR WT or K630R mutant in HEK293 cells, followed by lysis, immunoprecipitation (IP) of HDAC-Flag from the lysates, SDS-PAGE separation, and western blot analysis with NCoR, AR, or Flag (HD7-F) antibodies. As a gel migration control, lysate (Lys) from transfected cells were included. Repetitive independent trials are shown in Figure S5B. (C) HDAC7 wild type (WT) or GOF mutant (Y) were co-expressed with AR WT in HEK293 cells, which were then treated with SAHA (10 μM) to induce acetylation. After lysis and immunoprecipitation (IP), bound proteins were washed with a high salt (500 mM) buffer, before SAHA (100 μM) or DMSO vehicle was added and further washed. Bound proteins were separated by SDS-PAGE and visualized with NCoR, AR, or Flag (HD7-F) antibodies. Repetitive independent trials are shown in Figures S5C. (D) NCoR proteins levels from three independent trials from part C and Figures S5C were quantified, normalized to samples without SAHA (set to 100%), and plotted with mean and standard error shown (Figure S5D). * = p<0.05, ** = p < 0.01. All trials include a bead binding control using lysates without expression of HDAC7-Flag (IgG). Figure 4. GOF mutant HDAC7 decreased AR-mediated transcription. Wild type and GOF mutant HDAC4, HDAC5, HDAC7, or HDAC9 were co-expressed with wild type AR (A, B, and C) or K630R mutant AR (A only), as indicated, in the presence of the AR-dependent reporter gene constructs ARE-LUC and global expression control construct Ren-LUC (all samples). Cells were then treated without or with DHT (1 nM). HDAC1/2-selective inhibitor SHI-1:2 (10 μM) or HDAC3-selective inhibitor (RGFP996, 10 μM) were also included in part C. Light signal due to luciferase expression was measure 36 h after DHT treatment and normalized to Ren luciferase levels. Mean light intensity data and standard error normalized as a percentage of the DHT sample (lane 2, set to 100%) from at least three independent trials are shown, with the data provided in Figure S7. Student t-test analysis was applied, where NS = not significant (p > 0.05), * = p < 0.05, ** = p < 0.01, ***= p < 0.001, and ***= p < 0.0001. Representative gel images show protein expression levels in the samples, where GAPDH (GDH) was included as a load control. Figure 5. HDAC7-dependent gene expression by AR using RT-PCR. (A) Wild type AR was co-expressed with wild type (WT) or GOF mutant (GF) HDAC7. Cells were then untreated (0.01% DMSO only) or treated with DHT (1 nM in 0.01% DMSO) for 36 hours. After lysis, total cDNA was synthesized from mRNA with reverse transcriptase and mRNA levels of SPRF5 and Wnt16 were measure by RT-PCR reaction from the total cDNA. Representative gel images show triplicate reactions from three independent trials. (B and C) Signals were quantified from part A using ImageJ 1.47v and normalized to the DHT alone sample (lane 2, set to 100%) to calculate mean percentage and standard error. Mean light intensity data and standard error from at least three independent trials are shown, with the data provided in Figure S8. Student t-test analysis was applied, where NS = not significant (p > 0.05), * = p < 0.05, ** = p < 0.01. Figure 6. HDAC7-NCoR dissociation in the presence of ER. HDAC7 wild type (WT) or GOF mutant were co-expressed with wild type ER in HEK293 cells, followed by lysis, immunoprecipitation (IP) of HDAC-Flag from the lysates, and western blot analysis with NCoR, Flag (HD7-F), ER antibodies. Repetitive independent trials are shown in Figure S9. All trials include a bead binding control using lysates without expression of HDAC-Flag (IgG). Figure 7. Proposed model of class IIa HDAC “reader” function. (A) HDAC4, 5, and 7 can act as epigenetic ‘readers’ that recruit the NCoR-HDAC3 complex, which is disrupted when acetyllysine binds to the inactive active site. (B) HDAC7 bridges unacetylated AR and NCoR to repress the activity of AR through HDAC3 recruitment. This acetylation-independent binding of AR to the HDAC7-NCoR-HDAC3 complex might be direct or indirect through associated proteins (AP), and is likely mediated outside of the HDAC7 active site. (C) Once AR is acetylated on its critical lysine K630, acetyl-K630 binds the HDAC7 active site to dissociate the NCoR-HDAC3 complex. Without epigenetic repression by HDAC3, AR-mediated transcription is active. Key resources table REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies anti-Flag Sigma F3165 anti-NCoR Cell Signaling 5948 anti-AR Cell Signaling 5153 anti-ER Cell Signaling 8644 anti-acetyllysine Cell Signaling 9441S anti-mouse Alexa Fluor 488 Thermo Fisher A-11029 anti-rabbit Alexa Fluor 594 Thermo Fisher A-11037 Anti-Flag M2 beads Sigma A2220 Bacterial and virus strains Rosetta2(DE3) Sigma 70954 Chemicals, peptides, and recombinant proteins LGKAc peptide GenScript USA, Inc. Custom synthesis LGK peptide GenScript USA, Inc. Custom synthesis DAPI Thermo Fisher D21490 ProLong® Gold Antifade Reagent Thermo Fisher P10144 Jetprime transfection reagent VWR 89129-924 Critical commercial assays QuickChange site-directed mutagenesis Agilent 200521 HDAC-Glo™ assay Promega G648B-C Bradford reagent Bio-Rad 5000205 Dual Luciferase Assay Promega E1960 Experimental models: Cell lines HEK293 ATCC CRL-1573 Oligonucleotides See Figure S1A and S8A for oligonucleotide sequences This paper N/A Recombinant DNA pcDNA3.1-HDAC4-Flag Addgene Addgene #13821 pcDNA3.1-Flag HDAC5 Addgene Addgene #13822 pcDNA3.1-Flag HDAC7 Addgene Addgene #13824 2NVR (pET28a-LIC-HDAC7 catalytic domain) Addgene Addgene #51340 pcDNA3.1-Flag HDAC9 Zhang et al., 2004 N/A pcDNA3.1-Flag HDAC1 C151A Wambua et al. 2014 N/A pcDNA3.1-Flag HDAC4 H976Y This paper N/A pcDNA3.1-Flag HDAC5 H1006Y This paper N/A pcDNA3.1-Flag HDAC7 H842Y This paper N/A pcDNA3.1-Flag HDAC9 H956Y This paper N/A pcDNA3-AR Fu et al., 2022 N/A pcDNA3-AR(K630A) Fu et al., 2022 N/A pLR-ARE-LUC Zhang et al., 2016 N/A pLR-Renilla-LUC Zhang et al., 2016 N/A pcDNA-HA-ER WT Addgene Addgene # 49498 Other FluorChemQ gel imager ProteinSimple Octet Red 96 BLI instrument FortéBio SSC sensor for BLI instrument FortéBio 18-5057 TCS SP8 inverted confocal microscope Leica Geniosplus Fluorimeter Tecan ChemiDoc gel imager Bio-Rad Highlights HDAC4, 5 and 7 dissociated from corepressor NCoR dependent on acetyllysine peptides Dissociation of the HDAC7-NCoR complex depended on the K630 acetylation site of AR AR transcriptional activity required an intact HDAC7 active site and AR K630 The evidence is consistent with Class IIa HDAC7 epigenetic reader function Declaration of interest: Authors declare no competing interests. Inclusion and Diversity: One or more of the authors of this paper self-identifies as an underrepresented ethnic minority in science. One or more of the authors of this paper self-identifies as a member of the LGBTQ+ community. One or more of the authors of this paper self-identifies as living with a disability. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. References Alimirah F , Chen J , Basrawala Z , Xin H , and Choubey D (2006). DU-145 and PC-3 human prostate cancer cell lines express androgen receptor: implications for the androgen receptor functions and regulation. FEBS Lett 580 , 2294–2300.16580667 Arrowsmith CH , Bountra C , Fish PV , Lee K , and Schapira M (2012). Epigenetic protein families: a new frontier for drug discovery. Nat Rev Drug Discov 11 , 384–400.22498752 Bantscheff M , Hopf C , Savitski MM , Dittmann A , Grandi P , Michon AM , Schlegl J , Abraham Y , Becher I , Bergamini G , (2011). Chemoproteomics profiling of HDAC inhibitors reveals selective targeting of HDAC complexes. Nature biotechnology 29 , 255–265. Bottomley MJ , Lo Surdo P , Di Giovine P , Cirillo A , Scarpelli R , Ferrigno F , Jones P , Neddermann P , De Francesco R , Steinkuhler C , (2008a). Structural and Functional Analysis of the Human HDAC4 Catalytic Domain Reveals a Regulatory Structural Zinc-binding Domain. The Journal of biological chemistry 283 , 26694–26704.18614528 Bottomley MJ , Lo Surdo P , Di Giovine P , Cirillo A , Scarpelli R , Ferrigno F , Jones P , Neddermann P , De Francesco R , Steinkuhler C , (2008b). Structural and functional analysis of the human HDAC4 catalytic domain reveals a regulatory structural zinc-binding domain. J Biol Chem 283 , 26694–26704.18614528 Bradner JE , West N , Grachan ML , Greenberg EF , Haggarty SJ , Warnow T , and Mazitschek R (2010). Chemical phylogenetics of histone deacetylases. Nature chemical biology 6 , 238–243.20139990 Castaneda CA , Wolfson NA , Leng KR , Kuo YM , Andrews AJ , and Fierke CA (2017). HDAC8 substrate selectivity is determined by long- and short-range interactions leading to enhanced reactivity for full-length histone substrates compared with peptides. The Journal of biological chemistry 292 , 21568–21577.29109148 Cheng S , Brzostek S , Lee SR , Hollenberg AN , and Balk SP (2002). Inhibition of the dihydrotestosterone-activated androgen receptor by nuclear receptor corepressor. Mol Endocrinol 16 , 1492–1501.12089345 Chng KR , Chang CW , Tan SK , Yang C , Hong SZ , Sng NY , and Cheung E (2012). A transcriptional repressor co-regulatory network governing androgen response in prostate cancers. The EMBO journal 31 , 2810–2823.22531786 Choudhary C , Weinert BT , Nishida Y , Verdin E , and Mann M (2014). The growing landscape of lysine acetylation links metabolism and cell signalling. Nat Rev Mol Cell Biol 15 , 536–550.25053359 Clocchiatti A , Florean C , and Brancolini C (2011). Class IIa HDACs: from important roles in differentiation to possible implications in tumourigenesis. J Cell Mol Med 15 , 1833–1846.21435179 Darkin-Rattray SJ , Gurnett AM , Myers RW , Dulski PM , Crumley TM , Allocco JJ , Cannova C , Meinke PT , Colletti SL , Bednarek MA , (1996). Apicidin: a novel antiprotozoal agent that inhibits parasite histone deacetylase. Proc Natl Acad Sci U S A 93 , 13143–13147.8917558 de Ruijter AJ , van Gennip AH , Caron HN , Kemp S , and van Kuilenburg AB (2003). Histone deacetylases (HDACs): characterization of the classical HDAC family. Biochem J 370 , 737–749.12429021 Desravines DC , Serna Martin I , Schneider R , Mas PJ , Aleksandrova N , Jensen MR , Blackledge M , and Hart DJ (2017). Structural Characterization of the SMRT Corepressor Interacting with Histone Deacetylase 7. Sci Rep 7 , 3678.28623264 Dhalluin C , Carlson JE , Zeng L , He C , Aggarwal AK , and Zhou MM (1999). Structure and ligand of a histone acetyltransferase bromodomain. Nature 399 , 491–496.10365964 Ebrahimi A , Sevinc K , Gurhan Sevinc G , Cribbs AP , Philpott M , Uyulur F , Morova T , Dunford JE , Goklemez S , Ari S , (2019). Bromodomain inhibition of the coactivators CBP/EP300 facilitate cellular reprogramming. Nature chemical biology 15 , 519–528.30962627 Eckschlager T , Plch J , Stiborova M , and Hrabeta J (2017). Histone Deacetylase Inhibitors as Anticancer Drugs. Int J Mol Sci 18 . Finnin MS , Donigian JR , Cohen A , Richon VM , Rifkind RA , Marks PA , Breslow R , and Pavletich NP (1999). Structures of a histone deacetylase homologue bound to the TSA and SAHA inhibitors. Nature 401 , 188–193.10490031 Fischle W , Dequiedt F , Fillion M , Hendzel MJ , Voelter W , and Verdin E (2001). Human HDAC7 histone deacetylase activity is associated with HDAC3 in vivo. J Biol Chem 276 , 35826–35835.11466315 Fischle W , Dequiedt F , Hendzel MJ , Guenther MG , Lazar MA , Voelter W , and Verdin E (2002a). Enzymatic Activity Associated with Class II HDACs Is Dependent on a Multiprotein Complex Containing HDAC3 and SMRT/N-CoR. Molecular cell 9 , 45–57.11804585 Fischle W , Dequiedt F , Hendzel MJ , Guenther MG , Lazar MA , Voelter W , and Verdin E (2002b). Enzymatic activity associated with class IIHDACs is dependent on a multiprotein complex containing HDAC3 and SMRT/N-CoR. Molecular Cell 9 , 45–57.11804585 Fischle W , Emiliani S , Hendzel MJ , Nagase T , Nomura N , Voelter W , and Verdin E (1999). A new family of human histone deacetylases related to Saccharomyces cerevisiae HDA1p. The Journal of biological chemistry 274 , 11713–11720.10206986 Fraga MF , Ballestar E , Villar-Garea A , Boix-Chornet M , Espada J , Schotta G , Bonaldi T , Haydon C , Ropero S , Petrie K , (2005). Loss of acetylation at Lys16 and trimethylation at Lys20 of histone H4 is a common hallmark of human cancer. Nat Genet 37 , 391–400.15765097 Fu M , Rao M , Wang C , Sakamaki T , Wang J , Di Vizio D , Zhang X , Albanese C , Balk S , Chang C , (2003). Acetylation of androgen receptor enhances coactivator binding and promotes prostate cancer cell growth. Mol Cell Biol 23 , 8563–8575.14612401 Fu M , Wang C , Reutens AT , Wang J , Angeletti RH , Siconolfi-Baez L , Ogryzko V , Avantaggiati ML , and Pestell RG (2000). p300 and p300/cAMP-response element-binding protein-associated factor acetylate the androgen receptor at sites governing hormone-dependent transactivation. J Biol Chem 275 , 20853–20860.10779504 Fu M , Wang C , Wang J , Zhang X , Sakamaki T , Yeung YG , Chang C , Hopp T , Fuqua SA , Jaffray E , (2002). Androgen receptor acetylation governs trans activation and MEKK1-induced apoptosis without affecting in vitro sumoylation and trans-repression function. Mol Cell Biol 22 , 3373–3388.11971970 Gao WQ , Bohl CE , and Dalton JT (2005). Chemistry and structural biology of androgen receptor. Chemical Reviews 105 , 3352–3370.16159155 Gaur V , Connor T , Sanigorski A , Martin SD , Bruce CR , Henstridge DC , Bond ST , McEwen KA , Kerr-Bayles L , Ashton TD , (2016). Disruption of the Class IIa HDAC Corepressor Complex Increases Energy Expenditure and Lipid Oxidation. Cell reports 16 , 2802–2810.27626651 Guenther MG , Barak O , and Lazar MA (2001). The SMRT and N-CoR corepressors are activating cofactors for histone deacetylase 3. Mol Cell Biol 21 , 6091–6101.11509652 Hansen BK , Gupta R , Baldus L , Lyon D , Narita T , Lammers M , Choudhary C , and Weinert BT (2019). Analysis of human acetylation stoichiometry defines mechanistic constraints on protein regulation. Nature Communications 10 , 1055. Hebbes TR , Thorne AW , and Crane-Robinson C (1988). A direct link between core histone acetylation and transcriptionally active chromatin. The EMBO journal 7 , 1395–1402.3409869 Heldring N , Pike A , Andersson S , Matthews J , Cheng G , Hartman J , Tujague M , Ström A , Treuter E , Warner M , (2007). Estrogen Receptors: How Do They Signal and What Are Their Targets. Physiological Reviews 87 , 905–931.17615392 Hesham HM , Lasheen DS , and Abouzid KAM (2018). Chimeric HDAC inhibitors: Comprehensive review on the HDAC-based strategies developed to combat cancer. Med Res Rev 38 , 2058–2109.29733427 Hodgson MC , Astapova I , Cheng S , Lee LJ , Verhoeven MC , Choi E , Balk SP , and Hollenberg AN (2005). The androgen receptor recruits nuclear receptor CoRepressor (N-CoR) in the presence of mifepristone via its N and C termini revealing a novel molecular mechanism for androgen receptor antagonists. J Biol Chem 280 , 6511–6519.15598662 Houtkooper RH , Pirinen E , and Auwerx J (2012). Sirtuins as regulators of metabolism and healthspan. Nat Rev Mol Cell Biol 13 , 225–238.22395773 Hudson BP , Martinez-Yamout MA , Dyson HJ , and Wright PE (2000). Solution structure and acetyl-lysine binding activity of the GCN5 bromodomain. Journal of molecular biology 304 , 355–370.11090279 Hudson GM , Watson PJ , Fairall L , Jamieson AG , and Schwabe JW (2015). Insights into the Recruitment of Class IIa Histone Deacetylases (HDACs) to the SMRT/NCoR Transcriptional Repression Complex. J Biol Chem 290 , 18237–18244.26055705 Jacobson RH , Ladurner AG , King DS , and Tjian R (2000). Structure and function of a human TAF(II)250 double bromodomain module. Science 288 , 1422–1425.10827952 Jones P , Altamura S , De Francesco R , Gallinari P , Lahm A , Neddermann P , Rowley M , Serafini S , and Steinkuhler C (2008). Probing the elusive catalytic activity of vertebrate class IIa histone deacetylases. Bioorg Med Chem Lett 18 , 1814–1819.18308563 Joshi P , Greco TM , Guise AJ , Luo Y , Yu F , Nesvizhskii AI , and Cristea IM (2013). The functional interactome landscape of the human histone deacetylase family. Mol Syst Biol 9 , 672.23752268 Karvonen U , Janne OA , and Palvimo JJ (2006). Androgen receptor regulates nuclear trafficking and nuclear domain residency of corepressor HDAC7 in a ligand-dependent fashion. Exp Cell Res 312 , 3165–3183.16860317 Kim GS , Jung HE , Kim JS , and Lee YC (2015). Mutagenesis Study Reveals the Rim of Catalytic Entry Site of HDAC4 and −5 as the Major Binding Surface of SMRT Corepressor. PLoS One 10 , e0132680.26161557 Krämer OH , Göttlicher M , and Heinzel T (2001). Histone deacetylase as a therapeutic target. Trends in Endocrinology & Metabolism 12 , 294–300.11504668 Lahm A , Paolini C , Pallaoro M , Nardi MC , Jones P , Neddermann P , Sambucini S , Bottomley MJ , Lo Surdo P , Carfi A , (2007). Unraveling the hidden catalytic activity of vertebrate class IIa histone deacetylases. Proc Natl Acad Sci U S A 104 , 17335–17340.17956988 Lauffer BE , Mintzer R , Fong R , Mukund S , Tam C , Zilberleyb I , Flicke B , Ritscher A , Fedorowicz G , Vallero R , (2013). Histone deacetylase (HDAC) inhibitor kinetic rate constants correlate with cellular histone acetylation but not transcription and cell viability. J Biol Chem 288 , 26926–26943.23897821 Leong H , Sloan JR , Nash PD , and Greene GL (2005). Recruitment of histone deacetylase 4 to the N-terminal region of estrogen receptor alpha. Mol Endocrinol 19 , 2930–2942.16051668 Li JW , Wang J , Wang JX , Nawaz Z , Liu JM , Qin J , and Wong JM (2000). Both corepressor proteins SMRT and N-CoR exist in large protein complexes containing HDAC3. Embo Journal 19 , 4342–4350.10944117 Ling H , Peng L , Seto E , and Fukasawa K (2012). Suppression of centrosome duplication and amplification by deacetylases. Cell Cycle 11 , 3779–3791.23022877 Lobera M , Madauss KP , Pohlhaus DT , Wright QG , Trocha M , Schmidt DR , Baloglu E , Trump RP , Head MS , Hofmann GA , (2013). Selective class IIa histone deacetylase inhibition via a nonchelating zinc-binding group. Nature chemical biology 9 , 319–325.23524983 Malik S , Jiang S , Garee JP , Verdin E , Lee AV , O’Malley BW , Zhang M , Belaguli NS , and Oesterreich S (2010). Histone deacetylase 7 and FoxA1 in estrogen-mediated repression of RPRM. Mol Cell Biol 30 , 399–412.19917725 Manning ET , Ikehara T , Ito T , Kadonaga JT , and Kraus WL (2001). p300 forms a stable, template-committed complex with chromatin: role for the bromodomain. Mol Cell Biol 21 , 3876–3887.11359896 Monje P , Zanello S , Holick M , and Boland R (2001). Differential cellular localization of estrogen receptor alpha in uterine and mammary cells. Mol Cell Endocrinol 181 , 117–129.11476946 Mottis A , Mouchiroud L , and Auwerx J (2013). Emerging roles of the corepressors NCoR1 and SMRT in homeostasis. Genes Dev 27 , 819–835.23630073 Mujtaba S , He Y , Zeng L , Farooq A , Carlson JE , Ott M , Verdin E , and Zhou M-M (2002). Structural Basis of Lysine-Acetylated HIV-1 Tat Recognition by PCAF Bromodomain. Molecular Cell 9 , 575–586.11931765 Mujtaba S , He Y , Zeng L , Yan S , Plotnikova O , Sachchidanand , Sanchez R , Zeleznik-Le NJ , Ronai Z , and Zhou MM (2004). Structural mechanism of the bromodomain of the coactivator CBP in p53 transcriptional activation. Molecular cell 13 , 251–263.14759370 Negmeldin AT , Knoff JR , and Pflum MKH (2018). The structural requirements of histone deacetylase inhibitors: C4-modified SAHA analogs display dual HDAC6/HDAC8 selectivity. Eur J Med Chem 143 , 1790–1806.29150330 Negmeldin AT , and Pflum MKH (2017). The structural requirements of histone deacetylase inhibitors: SAHA analogs modified at the C5 position display dual HDAC6/8 selectivity. Bioorg Med Chem Lett 27 , 3254–3258.28648461 Owen DJ , Ornaghi P , Yang JC , Lowe N , Evans PR , Ballario P , Neuhaus D , Filetici P , and Travers AA (2000). The structural basis for the recognition of acetylated histone H4 by the bromodomain of histone acetyltransferase gcn5p. The EMBO journal 19 , 6141–6149.11080160 Park SY , Kim GS , Hwang HJ , Nam TH , Park HS , Song J , Jang TH , Lee YC , and Kim JS (2018). Structural basis of the specific interaction of SMRT corepressor with histone deacetylase 4. Nucleic Acids Res 46 , 11776–11788.30321390 Petrie K , Guidez F , Howell L , Healy L , Waxman S , Greaves M , and Zelent A (2003). The histone deacetylase 9 gene encodes multiple protein isoforms. J Biol Chem 278 , 16059–16072.12590135 Sanchez-Pacheco A , Martinez-Iglesias O , Mendez-Pertuz M , and Aranda A (2009). Residues K128, 132, and 134 in the thyroid hormone receptor-alpha are essential for receptor acetylation and activity. Endocrinology 150 , 5143–5152.19819978 Schneider CA , Rasband WS , and Eliceiri KW (2012). NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9 , 671–675.22930834 Schuetz A , Min J , Allali-Hassani A , Schapira M , Shuen M , Loppnau P , Mazitschek R , Kwiatkowski NP , Lewis TA , Maglathin RL , (2008). Human HDAC7 harbors a class IIa histone deacetylase-specific zinc binding motif and cryptic deacetylase activity. J Biol Chem 283 , 11355–11363.18285338 Smith SG , and Zhou MM (2016). The Bromodomain: A New Target in Emerging Epigenetic Medicine. ACS Chem Biol 11 , 598–608.26596782 Tanner MJ , Welliver RC Jr. , Chen M , Shtutman M , Godoy A , Smith G , Mian BM , and Buttyan R (2011). Effects of androgen receptor and androgen on gene expression in prostate stromal fibroblasts and paracrine signaling to prostate cancer cells. PLoS One 6 , e16027.21267466 van Rooij E , Fielitz J , Sutherland LB , Thijssen VL , Crijns HJ , Dimaio MJ , Shelton J , De Windt LJ , Hill JA , and Olson EN (2010). Myocyte enhancer factor 2 and class II histone deacetylases control a gender-specific pathway of cardioprotection mediated by the estrogen receptor. Circ Res 106 , 155–165.19893013 Vannini A , Volpari C , Gallinari P , Jones P , Mattu M , Carfi A , De Francesco R , Steinkuhler C , and Di Marco S (2007). Substrate binding to histone deacetylases as shown by the crystal structure of the HDAC8-substrate complex. EMBO Rep 8 , 879–884.17721440 Varlakhanova N , Snyder C , Jose S , Hahm JB , and Privalsky ML (2010). Estrogen receptors recruit SMRT and N-CoR corepressors through newly recognized contacts between the corepressor N terminus and the receptor DNA binding domain. Mol Cell Biol 30 , 1434–1445.20065040 Wambua MK , Nalawansha DA , Negmeldin AT , and Pflum MK (2014). Mutagenesis Studies of the 14 Å Internal Cavity of Histone Deacetylase 1: Insights towards the Acetate Escape Hypothesis and Selective Inhibitor Design. Journal of medicinal chemistry 57 , 642–650.24405391 Wang C , Fu M , Angeletti RH , Siconolfi-Baez L , Reutens AT , Albanese C , Lisanti MP , Katzenellenbogen BS , Kato S , Hopp T , (2001). Direct acetylation of the estrogen receptor alpha hinge region by p300 regulates transactivation and hormone sensitivity. J Biol Chem 276 , 18375–18383.11279135 Wang C , Tian L , Popov VM , and Pestell RG (2011). Acetylation and nuclear receptor action. J Steroid Biochem Mol Biol 123 , 91–100.21167281 West AC , and Johnstone RW (2014). New and emerging HDAC inhibitors for cancer treatment. J Clin Invest 124 , 30–39.24382387 Wong MM , Guo C , and Zhang J (2014). Nuclear receptor corepressor complexes in cancer: mechanism, function and regulation. Am J Clin Exp Urol 2 , 169–187.25374920 Yang Y , Tse AK , Li P , Ma Q , Xiang S , Nicosia SV , Seto E , Zhang X , and Bai W (2011). Inhibition of androgen receptor activity by histone deacetylase 4 through receptor SUMOylation. Oncogene 30 , 2207–2218.21242980 Yuan Z , Peng L , Radhakrishnan R , and Seto E (2010). Histone deacetylase 9 (HDAC9) regulates the functions of the ATDC (TRIM29) protein. J Biol Chem 285 , 39329–39338.20947501 Zhang X , Wharton W , Yuan Z , Tsai SC , Olashaw N , and Seto E (2004). Activation of the growth-differentiation factor 11 gene by the histone deacetylase (HDAC) inhibitor trichostatin A and repression by HDAC3. Mol Cell Biol 24 , 5106–5118.15169878 Zhang Y , Mantravadi PK , Jobbagy S , Bao W , and Koh JT (2016). Antagonizing the Androgen Receptor with a Biomimetic Acyltransferase. ACS Chem Biol 11 , 2797–2802.27548116

PMC009xxxxxx/PMC9454337.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0413066 2830 Cell Cell Cell 0092-8674 1097-4172 35798006 9454337 10.1016/j.cell.2022.03.045 NIHMS1833281 Article Therapeutic in vivo delivery of gene editing agents Raguram Aditya 123† Banskota Samagya 123† Liu David R. 123* 1 Merkin Institute of Transformative Technologies in Healthcare, Broad Institute of MIT and Harvard, Cambridge, Massachusetts, USA 2 Department of Chemistry and Chemical Biology, Harvard University, Cambridge, Massachusetts, USA 3 Howard Hughes Medical Institute, Harvard University, Cambridge, Massachusetts, USA * Correspondence to: drliu@fas.harvard.edu † These authors contributed equally. Author order was decided by a coin flip. 3 9 2022 21 7 2022 06 7 2022 21 7 2023 185 15 28062827 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Summary In vivo gene editing therapies offer the potential to treat the root causes of many genetic diseases. Realizing the promise of therapeutic in vivo gene editing requires the ability to safely and efficiently deliver gene editing agents to relevant organs and tissues in vivo. Here, we review current delivery technologies that have been used to enable therapeutic in vivo gene editing, including viral vectors, lipid nanoparticles, and virus-like particles. Since no single delivery modality is likely to be appropriate for every possible application, we compare the benefits and drawbacks of each method and highlight opportunities for future improvements. eToC blurb – Liu Therapeutic gene editing has made exciting progress in the past few years, in part through key advances in in vivo delivery technologies. David Liu and colleagues review the essential characteristics required for efficient in vivo gene editing delivery vehicles, discussing the pros and cons of viral vectors, lipid nanoparticles, and virus-like particles. pmcIntroduction The ability to precisely manipulate and edit the sequence of DNA in human cells could enable powerful new classes of genomic medicines. Millions of people worldwide suffer from heritable genetic disorders (Korf et al., 2019), the root causes of which could in principle be corrected by therapeutic DNA editing agents. While traditional gene augmentation therapies can treat some autosomal recessive or haploinsufficiency disorders by providing a functional copy of a gene, gene editing therapies can directly correct pathogenic mutations in genomic DNA,. As such, gene editing in principle could treat a much wider range of genetic diseases, including autosomal dominant disorders, conditions that arise from too little or too much of a gene product, or other conditions for which simple overexpression of a gene cannot optimally rescue the disease. Even for conditions that could be addressed with existing gene augmentation or gene silencing strategies, gene editing therapies that install mutations to increase or decrease the expression of a target gene could achieve the same effect with a one-time treatment, offering the possibility of a permanent cure. More broadly, the risk of suffering from certain major diseases such as coronary heart disease, even in individuals without pathogenic mutations, can be modulated by precise modification of target genes, raising the possibility that gene editing (if shown to be sufficiently safe and efficacious) may one day be used to decrease disease risk in the general population. The promise of therapeutic gene editing has motivated intense efforts to bring gene editing therapies to the clinic. Recent advances include the development of robust tools for gene editing in mammalian cells, including programmable nucleases, base editors, and prime editors (Anzalone et al., 2020; Doudna, 2020; Newby and Liu, 2021). These gene editing agents have been widely applied to treat numerous disorders with a genetic component across a variety of animal models (Newby and Liu, 2021; Rees et al., 2021). Some of these therapeutic gene editing strategies have already entered clinical trials, with promising early results (Gillmore et al., 2021), and many additional clinical and pre-clinical gene editing programs are underway. Most current gene editing clinical trials involve ex vivo editing (Wang et al., 2020) in which cells are removed from a patient’s body, edited while outside the body, and reintroduced into the patient. This approach is feasible for some important cell types, including hematopoietic stem cells (HSCs) (Ferrari et al., 2021), but most cell types are not amenable to ex vivo manipulation and transplantation into patients. In vivo gene editing, where cells are edited directly within the body, offers the greatest promise for treating genetic disorders. However, in vivo gene editing requires the ability to efficiently and safely deliver editing agents to a large enough fraction of relevant cells directly within the body, which can present a major challenge. Therapeutic methods for delivering gene editing agents in vivo must efficiently target desired cells and deliver sufficient quantities of editing agents into those cells. To achieve this goal, numerous delivery technologies have been developed and tested in mice, non-human primates (NHPs), and other animals (Newby and Liu, 2021; Taha et al., 2022; van Haasteren et al., 2020; Wang et al., 2020; Wei et al., 2020a; Yin et al., 2017a). Two promising delivery strategies—adeno-associated virus (AAV) delivery and lipid nanoparticle (LNP) delivery—have shown initial successes in recent in vivo gene editing clinical trials (Gillmore et al., 2021). These developments suggest that current state-of-the-art delivery methods have the potential to enable powerful new in vivo gene editing therapies in the coming years. In this review, we summarize three types of gene editing agents that have been used for therapeutic in vivo gene editing and overview essential characteristics of efficient in vivo delivery vehicles. We then describe methods that are commonly used to deliver therapeutic gene editing agents in vivo, with a focus on viral and non-viral delivery methods currently used in ongoing clinical trials, including AAV and LNP delivery. Finally, we discuss virus-like particle (VLP) delivery, a promising emerging method that combines key benefits of both viral and non-viral delivery. As no single delivery approach is likely to be ideal for all possible applications, we note advantages and disadvantages of each strategy and highlight opportunities for further development. Therapeutic gene editing strategies Several types of gene editing agents have been used for in vivo gene editing. Within space limitations, we provide below a brief overview of these gene editing agents (Figure 1) and then discuss approaches for delivering such agents in vivo. Modern methods for therapeutic gene editing have been reviewed extensively elsewhere (Anzalone et al., 2020; Rees and Liu, 2018), and we direct the reader to these reviews for additional details. Nucleases Until recently, all robust strategies for performing gene editing in mammalian cells involved using a nuclease to generate a double-strand break (DSB) at a specific location in genomic DNA. Meganucleases, zinc finger nucleases (ZFNs), and transcription activator-like effector nucleases (TALENs) were the first enzymes used for gene editing in mammalian cells (Urnov, 2018). However, because these editing agents rely on complex protein:DNA interactions to bind specific DNA sequences prior to cutting, a new editing protein must be designed and constructed for each new genomic target site of interest. In practice, this design and/or construction process can be time- and resource-intensive. CRISPR-Cas nucleases revolutionized gene editing because these enzymes can recognize their targets by simple RNA:DNA base-pairing interactions between the target DNA and a single-guide RNA (sgRNA) molecule loaded inside the Cas protein (Doudna, 2020; Jinek et al., 2012). This remarkable feature allows researchers to program CRISPR-Cas nucleases to target and cut different genomic loci simply by changing the sequence of a ~20 base-pair portion of the sgRNA, without needing to design a new Cas protein. Currently, CRISPR-Cas nucleases are by far the most widely used enzymes for generating targeted DSBs in mammalian cells. In mammalian cells, DSBs are most frequently repaired by non-homologous end joining (NHEJ) (Lieber, 2010) or microhomology-mediated end joining (MMEJ) (Seol et al., 2018), both of which result in an uncontrollable distribution of small insertions and deletions (indels) at the target site. In some situations, DSBs can be repaired via homology-directed repair (HDR) (San Filippo et al., 2008). This process can be templated by an exogenous DNA donor that can contain any arbitrary sequence flanked by regions of homology to the target site and can, in principle, be used to install any desired sequence in genomic DNA. However, HDR is inefficient in most cell types, including non-dividing cells, and successful HDR editing outcomes are generally accompanied by a substantial amount of indels (Chapman et al., 2012; Cox et al., 2015). For these reasons, nuclease-mediated gene editing is a poor general strategy for precisely correcting a mutated gene back to the wild-type sequence with minimal undesired editing byproducts. Instead, nuclease-generated DSBs are most useful for disrupting coding sequences of genes by introducing frameshift mutations, or for disrupting regulatory motifs in non-coding sequences to modulate gene expression. Indeed, the vast majority of in vivo gene editing with nucleases involves the disruption of targeted genomic loci to induce therapeutic effects (Cox et al., 2015; Li et al., 2020a). Importantly, DSB-mediated indel outcomes are not controllable and typically consist of mixtures of numerous different detectable sequence outcomes (Nambiar et al., 2022), each of which can have a different impact on disease biology. While programmable nucleases have been applied therapeutically, several undesired consequences of nuclease-mediated gene editing have been reported. In addition to small indels at the target site, generating DSBs in genomic DNA can lead to large deletions (Kosicki et al., 2018; Song et al., 2020), chromosomal translocations (Giannoukos et al., 2018; Stadtmauer et al., 2020; Turchiano et al., 2021; Webber et al., 2019), chromothripsis (Leibowitz et al., 2021), or other chromosomal abnormalities (Alanis-Lobato et al., 2021). Indeed, nuclease-edited chimeric antigen receptor (CAR)-T cells contained chromosomal translocations that persisted in vivo (Stadtmauer et al., 2020). These editing outcomes, while rare, pose safety risks that could negatively impact certain clinical uses of nucleases. These drawbacks of nuclease editing, combined with the fact that nuclease-initiated HDR is inefficient in most therapeutically relevant cell types, have motivated the development of alternative strategies for more precise gene editing. Base editors Base editors (BEs) overcome many of the limitations of nucleases by enabling precise gene correction through single-nucleotide conversions in genomic DNA without requiring DSBs (Rees and Liu, 2018). BEs are composed of DNA-modifying enzymes fused to programmable DNA-binding domains, and many BEs of different types have been reported to date. The first reported BEs were cytosine base editors (CBEs), which perform targeted C•G-to-T•A conversions and consist of cytidine deaminases fused to catalytically impaired Cas enzymes and uracil glycosylase inhibitors (UGIs) (Komor et al., 2016; Nishida et al., 2016). In canonical CBEs, the catalytically impaired Cas enzyme first binds to a specific genomic locus without generating a DSB. Base pairing between the guide RNA and the target DNA strand exposes a single-stranded DNA bubble that is accessible to deamination by the fused cytidine deaminase domain. Because the fused cytidine deaminase is specific for single-stranded DNA substrates, deamination is restricted to a small window within the exposed DNA strand. Deamination of cytosine generates uracil, which is partially protected from base excision by the fused UGIs, resulting in a U•G mismatch at the target DNA locus. The catalytically impaired Cas enzyme selectively nicks the unedited G-containing strand only—without creating a DSB—which biases cellular mismatch repair to replace the unedited strand by using the edited strand as a template. The resulting U•A base pair is eventually converted into a T•A base pair by cellular repair machinery. If the uracil-containing intermediate is not protected and instead the excision of uracil is promoted, the result is often a C•G-to-G•C conversion rather than a C•G-to-T•A conversion (Komor et al., 2017). This phenomenon was exploited to engineer C•G-to-G•C base editors (CGBEs) (Chen et al., 2021a; Koblan et al., 2021a; Kurt et al., 2021). Adenine base editors (ABEs) perform A•T-to-G•C conversions via an analogous mechanism (Gaudelli et al., 2017). Since no known natural enzyme catalyzes the deamination of deoxyadenosine that is needed to base edit adenine in DNA, all ABEs described to date are derived from laboratory-evolved deoxyadenosine deaminases fused to catalytically impaired Cas enzymes (Gaudelli et al., 2017; Gaudelli et al., 2020; Richter et al., 2020). ABEs are an especially useful class of base editor because they reverse the most common type of pathogenic point mutation (C•G to T•A), which accounts for approximately half of known pathogenic single-nucleotide polymorphisms (SNPs) (Rees and Liu, 2018). Mitochondria have their own genomes, and mutations in mitochondrial DNA cause many genetic diseases (Gorman et al., 2016). Editing of mitochondrial DNA is challenged by the lack of efficient methods to deliver guide RNAs into mitochondria, precluding the efficient use of CRISPR-Cas-based gene editing systems. As a result, until recently, only targeted gene destruction, rather than precise gene editing, was possible in mitochondria. The development of DdCBEs, a special class of CBEs that use a unique cytidine deaminase domain that operates on double-stranded DNA substrates, enabled the first precise editing of mitochondrial DNA in living systems (Mok et al., 2020). DdCBEs use TALE proteins instead of Cas domains to direct deamination to a particular DNA locus, which enables CRISPR-free base editing of mitochondrial DNA, in addition to nuclear DNA (Guo et al., 2022; Lee et al., 2021; Lim et al., 2022; Mok et al., 2020; Silva-Pinheiro et al., 2022). Since the development of the original CBE and ABE, hundreds of base editors with different properties have been reported by many laboratories (Anzalone et al., 2020; Rees and Liu, 2018). BEs have been applied therapeutically for various ex vivo and in vivo gene editing applications to correct disease-causing point mutations or to install single-nucleotide variants that prevent or rescue disease phenotypes (Newby and Liu, 2021). Prime editors While BEs can, in principle, correct the majority of pathogenic SNPs (Rees and Liu, 2018), they cannot perform all possible single-nucleotide conversions and also cannot mediate targeted insertions or deletions. To address these limitations, we developed prime editors (PEs), which enable the programmable installation of any single-nucleotide conversion, small insertion, small deletion, or a combination thereof, without generating DSBs (Anzalone et al., 2019). PEs consist of a reverse transcriptase fused to a Cas9 nickase domain and use an engineered prime editing guide RNA (pegRNA) to both direct the Cas9 nickase to a specific target locus and encode the particular edit of interest. PEs first nick the non-target DNA strand and use the resulting free 3’ end to prime reverse transcription using the pegRNA extension as a template. After the desired edit is incorporated into the newly synthesized strand, an additional nick can be used to bias cellular DNA repair to replace the unedited strand by using the edited strand as a template. Several examples of in vivo gene editing using PEs have been reported (Newby and Liu, 2021). Recent improvements to both the PE protein and pegRNA have enabled highly efficient prime editing in various cell types and should be useful for future in vivo prime editing efforts (Chen et al., 2021b; Nelson et al., 2021). Essential characteristics of efficient in vivo delivery vehicles Gene editing agents can be delivered into cells either as DNA or mRNA encoding their expression or directly as proteins or (if a guide RNA is required) ribonucleoproteins (RNPs). In all cases, successful in vivo delivery of gene editing agents requires overcoming several biological and molecular barriers to the intracellular delivery of macromolecules. Specifically, an efficient in vivo delivery vehicle must (1) package and protect its cargo from sequestration or destruction before it enters cells, (2) bind desired cells, (3) traverse the target cell membrane to access the cellular interior, and (4) release its cargo into the appropriate intracellular compartment (Figure 2). Most robust in vivo delivery vehicles encapsulate their cargos in protein or lipid shells to protect them from sequestration or degradation prior to cell entry (Mitchell et al., 2021). This protection enables the cargo to survive in circulation or at the site of administration until the vehicle encounters the target cell types. Additionally, delivery vehicles must avoid recognition by the immune system, as immune activation can cause them to be targeted for degradation. Some vehicle compositions are prone to activating the complement system, which can lead to vehicle clearance by phagocytic immune cells, and antibody-mediated recognition of delivery vehicles can also lead to undesirable phagocytic clearance (Hoshyar et al., 2016; Mitchell et al., 2021). Before delivering their cargos into cells, in vivo delivery vehicles must first be able to bind those cells within the body. This ability to access target cells is highly dependent on the route of administration of the delivery vehicle. Many intravenously injected vehicles can efficiently access some tissues, such as the liver, but cannot efficiently access others, such as the central nervous system (CNS), due to intrinsic biological barriers (e.g., the blood-brain barrier) (Daneman and Prat, 2015). Locally injecting vehicles into the CNS (e.g., via intrathecal injection) or eye (e.g., via subretinal injection) can circumvent biological barriers and enable access to certain important cell populations (Bottros and Christo, 2014; Peng et al., 2017). However, the ability to physically access a particular cell type does not guarantee efficacious delivery to that cell type. Delivery vehicles must be able to target desired cells, and typically use parts of their surfaces or specific targeting moieties to engage receptors on the surfaces of target cells and promote subsequent cell entry (Mitchell et al., 2021; Paunovska et al., 2022). Following successful engagement of target cells, delivery vehicles must enter those cells by crossing the cell membrane. In many cases, binding of delivery vehicles to cell surface receptors promotes endocytosis of those delivery vehicles into endosomes (Bareford and Swaan, 2007; Kazmierczak et al., 2020). Vehicles and cargos that remain sequestered in endosomes will eventually be degraded (Smith et al., 2019; Varkouhi et al., 2011). Therefore, successful vehicles must escape endosomes to release their cargos outside of endosomes and into the cell cytosol. Many successful delivery vehicles exploit the acidic environment of endosomes to trigger changes in the vehicle’s structure that promote endosomal escape and cargo release (Mitchell et al., 2021; Staring et al., 2018). Importantly, efficient delivery vehicles must be stable enough to protect their cargo while outside of cells, but must be able to disassemble and release their cargo after entering cells and escaping endosomes. Over multiple decades, researchers have identified and engineered several classes of delivery vehicles that can overcome these complex molecular obstacles to intracellular delivery. Current state-of-the-art delivery systems, including viral vectors, LNPs, and VLPs, can satisfy these key criteria for efficient in vivo delivery vehicles and are therefore well suited for the in vivo delivery of gene editing agents. Viral delivery Viruses naturally evolved to overcome barriers to in vivo delivery and can natively deliver nucleic acid cargos to many cell types. Because of these favorable characteristics, viruses are promising vehicles for delivering gene editing agents. Many viral vectors have been developed for in vivo gene therapy applications and used to deliver therapeutic genes in over 1,000 clinical trials (Ginn et al., 2018). Most in vivo gene editing applications have utilized adeno-associated viruses (AAVs), and a few pre-clinical studies have used lentiviruses or adenoviruses. Notably, an ongoing clinical trial uses AAVs to deliver gene editing agents into the eye to treat a form of genetic blindness (Sheridan, 2018). Below we provide insights into recent examples of in vivo gene editing using viral vectors and highlight opportunities for future advances. Adeno-associated virus (AAV) delivery The adeno-associated virus (AAV) is a ~25-nm non-enveloped virus composed of 60 copies of viral proteins VP1, VP2, and VP3 assembled into an icosahedral capsid (Drouin and Agbandje-McKenna, 2013). AAVs package a single-stranded DNA genome of ~5 kb (Naso et al., 2017; Wu et al., 2010). Because AAVs have been used to deliver in vivo gene therapies in animal models of human disease (Deverman et al., 2018; Wang et al., 2019), in clinical trials (Mendell et al., 2021), and in FDA-approved therapies (Mendell et al., 2017; Russell et al., 2017), they are currently the most popular viral vectors for delivering macromolecular therapeutics encoded as DNA. AAV delivery offers many advantages. AAVs have a well-understood and favorable safety profile, are highly biocompatible, and can carry payloads efficiently to a variety of clinically relevant tissues, including the eye (Maguire et al., 2008), liver, brain (Wang et al., 2019), cardiac muscle, and skeletal muscle (Wang et al., 2005). Furthermore, different naturally occurring AAV capsid serotypes can be used to direct AAVs to transduce various tissues in vivo (Wu et al., 2006). Laboratory evolution and rational engineering of AAV capsids have further expanded the available tissue-targeting specificities of AAVs (Asokan et al., 2010; Byrne et al., 2020; Dalkara et al., 2013; Deverman et al., 2016; Goertsen et al., 2022; Li et al., 2008b; Maheshri et al., 2006; Shen et al., 2013; Tabebordbar et al., 2021; Zinn et al., 2015), although few engineered or laboratory-evolved AAVs have entered the clinic as of this writing. The availability of numerous AAV serotypes allows researchers to choose an appropriate serotype for different applications that require targeting distinct cell populations in vivo. The size of the nucleic acid cargo is an important consideration when using AAV, as it has a packaging capacity of only ~5 kb of DNA (Dong et al., 1996; Wu et al., 2010). The AAV vector genome must be flanked by two inverted terminal repeats (ITRs) that are required for packaging the vector genome during AAV production, which leaves ~4.7 kb for a transgene cassette. This packaging capacity limits the potential scope of AAVs as delivery vehicles for gene editing agents, as most BEs and PEs that use a canonical S. pyogenes Cas9 (SpCas9) DNA-targeting domain are too large to fit into a single AAV. In addition to packaging DNA encoding the editing agent and, if needed, guide RNA(s), AAVs must also encode promoters driving editor and guide RNA expression, and cis-regulatory elements for efficient activity in vivo. These additional components further increase the required transgene size and limit the effective packaging capacity of a single AAV. To overcome these size limitations, researchers have developed several approaches that enable gene editing agents to be packaged into AAV vectors. Development of dual-AAV strategies that effectively reconstitute full-length proteins To address the packaging limitations of AAV, multiple groups (Chemello et al., 2021; Chen et al., 2020; Chew et al., 2016; Fine et al., 2015; Levy et al., 2020; Li et al., 2008a; Lim et al., 2020; Ryu et al., 2018; Truong et al., 2015; Villiger et al., 2018; Xu et al., 2021) developed strategies for splitting gene editing agents into two halves, such that each half can be packaged separately into individual AAV vectors. These two AAVs are then administered simultaneously, and in cells that are co-transduced by both AAVs, reconstitution of the full-length gene editing agent is achieved via molecular mechanisms acting at either the DNA, pre-mRNA, or protein levels (Tornabene and Trapani, 2020). Both mRNA and protein trans-splicing strategies have been used to reconstitute full-length gene editing agents that are split into two AAVs. Kim and colleagues used an mRNA trans-splicing strategy to deliver ABEs into mice (Ryu et al., 2018); intramuscular injection of these AAVs into a mouse model of Duchenne muscular dystrophy (DMD) yielded 3.3% base editing. Multiple laboratories (Chemello et al., 2021; Chen et al., 2020; Chew et al., 2016; Fine et al., 2015; Levy et al., 2020; Li et al., 2008a; Lim et al., 2020; Truong et al., 2015; Villiger et al., 2018; Xu et al., 2021) developed split-intein systems in which gene editing agents are reconstituted via protein trans-splicing. In these systems, gene editing agents are split into two halves, each fused to a split intein, and then packaged into two separate AAV capsids. In co-transduced cells, both halves of the editor protein are expressed, and dimerization of the split inteins promotes a partial or complete trans-protein splicing reaction that reconstitutes the full-length editor protein (Aranko et al., 2014). When we compared mRNA trans-splicing and protein trans-splicing methods, the split intein-mediated protein reconstitution strategy provided, on average, 4.5-fold higher base editing efficiency across multiple tissues in mice (Levy et al., 2020). This efficiency difference likely arises because of the two-step process required for successful trans-mRNA splicing that involves AAV genome concatemerization (Duan et al., 2001) followed by transcription and splicing of the ITR sequences, which have been reported to destabilize pre-mRNA (Xu et al., 2004). Therefore, the split intein-mediated protein reconstitution strategy is potentially a simpler and more robust strategy for splitting gene editing agents for dual-AAV delivery. Several studies successfully used the split-intein dual-AAV strategy and achieved editing efficiencies ranging from 9–60% across various therapeutic organs, including the liver, eye, CNS, cardiac muscle, and skeletal muscle (Koblan et al., 2021b; Lau and Suh, 2017; Levy et al., 2020; Rothgangl et al., 2021; Villiger et al., 2018; Yeh et al., 2020). We applied the dual-AAV base editing strategy in a mouse model of Hutchinson-Gilford progeria syndrome (HGPS) and corrected the C•G-to-T•A mutation in the LMNA gene responsible for HGPS (Koblan et al., 2021b). We achieved up to 30% correction of the gene in heart tissue and observed a large reduction in the amount of progerin protein in most tissues examined. Recently, Schwank and coworkers used a dual-AAV9 strategy to deliver an ABE targeting Pcsk9 into mice and achieved 60% base editing in the bulk liver, with a 6.8-fold reduction of serum Pcsk9 protein and a 3.3-fold reduction in serum cholesterol (Rothgangl et al., 2021). Split-intein dual-AAVs have also been used in the CNS to knock out mutant Huntington (HTT) gene (Yang et al., 2017), to correct the disease-causing mutation in a mouse model of Niemann-Pick disease (Levy et al., 2020), and to introduce strop codons in SOD1 to slow disease progression in a mouse model of amyotrophic lateral sclerosis (ALS) (Gaj et al., 2017). Dual-AAVs have also been used to achieve therapeutic levels of gene editing in the skeletal muscle, eye, and ear (Chemello et al., 2021; Jo et al., 2021; Ryu et al., 2018; Yeh et al., 2020). PEs, which are ~1 kb larger than corresponding BEs, also need to be split into multiple AAVs for successful delivery, and some early reports of in vivo prime editing have used split-intein AAV vectors (Böck et al., 2022; Liu et al., 2021a; Zheng et al., 2022). Schwank and coworkers reported 14% prime editing at the Dnmt1 test site in the mouse liver with dual-AAV8 vectors (Böck et al., 2022), and Xue and coworkers reported 6% prime editing in the mouse liver (Zheng et al., 2022). Dual-AAV delivery of PEs currently yields lower editing efficiency compared to dual-AAV delivery of Cas9 nuclease or BEs, but recent improvements to both the PE protein and pegRNA will likely be useful for improving in vivo prime editing efficiencies (Chen et al., 2021b; Nelson et al., 2021). Development of single-AAV vectors enabled by smaller Cas orthologs While dual-AAV approaches described above have mediated therapeutic editing in mouse models of human disease, single-AAV delivery would offer critical advantages for research and clinical use by simplifying manufacturing and characterization. A single-AAV delivery strategy can also reduce the total dose of AAV required to achieve a desired level of gene editing. Moreover, single-AAV approaches might enable increased editing efficiencies in tissues that are currently difficult to transduce by obviating the need for simultaneous transduction of multiple AAVs. Efforts to identify smaller orthologs of Cas9 or to generate small engineered Cas9 variants have enabled single-AAV delivery of CRISPR gene editing agents (Kim et al., 2017; Shams et al., 2021; Wang et al., 2020). The Cas9 nuclease from Staphylococcus aureus (SaCas9) is commonly used in single-AAV approaches as it has a gene size of 3.2 kb, which can be packaged in a single AAV along with one or two sgRNA expression cassettes. Zhang and coworkers harnessed an SaCas9 nuclease-encoding single-AAV vector to knock out Pcsk9 and reduce serum cholesterol in mice (Ran et al., 2015). In addition, an ongoing clinical trial uses a subretinally administered single AAV to deliver SaCas9 nuclease and two sgRNAs to delete a disease-causing mutation in the CEP290 gene in patients suffering from Leber’s congenital amaurosis 10 (LCA10) (Maeder et al., 2019; Sheridan, 2018). More recently, the discovery of other compact Cas9 variants such as Nme2Cas9 (3.24kb, PAM=N4CC) (Edraki et al., 2019; Liu et al., 2021b), CjCas9 (2.95kb, PAM=N4RYAC) (Kim et al., 2017; Li et al., 2020b), and SauriCas9 (3.18kb, PAM=N2GG) (Hu et al., 2020) has increased the number of Cas9 enzymes that can in principle be packaged into single-AAV vectors. These compact Cas9 variants have also broadened the targeting scope of single-AAV gene editing agents beyond that of SaCas9 (3.16kb, PAM=NNGRRT) or engineered variants such as SaKKH (3.16kb, PAM=NNNRRT) (Kleinstiver et al., 2015). In one example, Kim and coworkers developed a single-AAV system using CjCas9 nuclease and administered it subretinally into a mouse model of age-related macular degeneration to knock out VEGF-A (Kim et al., 2017). They observed 20% indels in the retina and retinal pigment epithelium (RPE) cells, which enabled therapeutic rescue with reduced neovascularization. Recently, multiple groups (Davis et al., 2022; Zhang et al., 2022) have used these smaller Cas9 variants to develop single-AAV approaches for packaging BEs that enable higher base editing efficiencies with lower total AAV doses compared to dual-AAV systems. As AAVs can cause dose-limiting toxicity in patients (Kuzmin et al., 2021), reducing total AAV dosing can increase therapeutic potential. Single-AAV approaches thus may offer favorable safety profiles compared to dual-AAV systems. Minimizing long-term expression of gene editing agents following AAV delivery One of the outstanding limitations of AAV delivery is that it results in persistent cargo expression in transduced cells. Because AAV genomes are maintained episomally in the nucleus, expression can persist for years (Chu et al., 2003; Vassalli et al., 2003). While this prolonged cargo expression is desirable for gene augmentation therapy applications, it is undesirable for gene editing applications as persistent expression of gene editing agents increases the risks of various types of off-target editing (Anzalone et al., 2020; Doman et al., 2018). Moreover, prolonged expression of Cas9, a non-human protein, can trigger an immune response in which edited cells that express Cas9 are targeted for destruction by the immune system (Charlesworth et al., 2019; Chew et al., 2016; Crudele and Chamberlain, 2018; Wagner et al., 2019; Wagner et al., 2021). To address this issue, various strategies for transiently expressing AAV-delivered gene editing agents have been developed. Multiple research groups have developed self-inactivating CRISPR/Cas9 AAV systems that use a guide RNA targeting the Cas9 enzyme. This strategy introduces indels into the AAV genome as long as the editor is expressed, which inactivates editor expression over time (Ibraheim et al., 2021; Li et al., 2019a; Li et al., 2019b). Although this approach functioned in mice without significantly reducing on-target editing, editor expression was not completely diminished. Additionally, cleaved AAV products were found to integrate at the on-target genomic locus, raising safety concerns. In another approach, Chen and coworkers developed a unique CBE variant that only becomes active at the on-target site by proteolysis of a fused deaminase inhibitory domain, which limits the presence of active CBE in cells (Wang et al., 2021). This strategy enabled efficient on-target base editing in the mouse liver with no observable off-target DNA or RNA base editing above background levels. A particularly noteworthy strategy for temporally regulating the expression of AAV-delivered gene editing agents was reported recently by Davidson and coworkers (Monteys et al., 2021). They developed a universal switch called Xon that exploits small molecule-controlled alternative RNA splicing to precisely control AAV transgene expression. In the presence of the small molecule inducer, drug-modulated splicing results in mRNA that includes the exon containing the start codon and leads to full-length protein expression from the AAV genome. However, in the absence of the small molecule, the exon containing the start codon is excluded from the mRNA, which prevents successful protein expression. In this study, the authors also developed a smaller version of Xon that can fit along with SaCas9 into a single AAV and showed successful temporal control of gene editing in vivo (Monteys et al., 2021). Given this promising result and the use of a clinical small molecule to trigger the Xon system, this approach, as well as others that temporally restrict AAV expression, could be useful for in vivo therapeutic gene editing. In addition to strategies for temporally regulating the expression of AAV-delivered gene editing agents, researchers have also developed methods to spatially control AAV cargo expression. By using tissue-specific AAV capsids, promoters, or miRNAs, expression of the gene editor cargo can be limited to a particular tissue, which will minimize the potential for off-target editing in non-target tissues. Although naturally occurring or laboratory-evolved AAV capsids have expanded the tissue-targeting scope of AAVs to many tissues, the ability to target a particular tissue does not necessarily entail specificity for that particular tissue over others (Deverman et al., 2016; Tabebordbar et al., 2021). Using tissue-specific promoters to drive editor expression is an attractive strategy, but the size limitation imposed by AAV packaging with Cas9 limits promoter choices. Another strategy to modulate cargo expression in different tissues is to incorporate binding sites for an endogenous miRNA in the 3’UTR of the Cas9 expression cassette (Xiao et al., 2019). In this approach, Cas9 protein expression can be silenced in tissues that highly express the miRNA, and expression will be limited to tissues that lack miRNA expression. Sontheimer, Niopek, and their respective coworkers combined the miRNA approach with natural inhibitors of Cas proteins known as anti-CRISPRs (Acrs) to limit Cas9 expression to tissues that express the miRNA (Hoffmann et al., 2019; Lee et al., 2019). In this approach, Acr protein expression is silenced in the presence of a miRNA, which in turn allows expression of Cas9 only in tissues that highly express that miRNA. Overall, spatially and temporally controlling the expression of AAV-delivered gene editing agents offers useful strategies to maximize gene editing specificity and thus may improve the safety of future therapeutic applications. Lentiviral delivery Lentiviruses (LVs) are enveloped viruses derived from HIV-1 that are made replication-incompetent by deletions in the 3’ LTR and by splitting the necessary components for virus production into multiple constructs (Dull et al., 1998; Naldini et al., 1996). LVs deliver RNA cargo that is reverse-transcribed and stably integrated semi-randomly into the genome of transduced cells. Integrase-deficient lentiviral vectors (IDLVs) have also been engineered, in which the integrase domain has been inactivated so that the viral cDNA persists episomally following reverse transcription (Wanisch and Yanez-Munoz, 2009). LVs have been used primarily for ex vivo gene delivery, mostly in HSCs and T cells, and are currently used in two FDA-approved chimeric antigen receptor (CAR)-T therapies (Mullard, 2017). LVs possess several advantages that make them attractive for genome editing. First, LVs can accommodate up to 10 kb of cargo DNA (Sweeney and Vink, 2021), which is sufficient to package virtually all known gene editing agents into a single vector. The large cargo packaging capacity of LVs also makes them well suited for multiplex genome editing using CRISPR-based agents, which requires the packaging of multiple sgRNA expression cassettes (Kabadi et al., 2014). Second, LVs can efficiently transduce both dividing and non-dividing cells (Kumar et al., 2001). Third, IDLV genomes can also be used as HDR templates (Lombardo et al., 2007). Finally, the tropism of lentiviruses can be readily modulated by changing the envelope glycoprotein used to pseudotype the virions (Cronin et al., 2005; Joglekar and Sandoval, 2017). There are few examples of using LVs for in vivo gene editing. Palczewski and coworkers administered an ABE- and sgRNA-encoding LV subretinally to correct a premature stop codon in the Rpe65 gene in a mouse model of Leber congenital amaurosis (Suh et al., 2021). A single dose of lentivirus injected into 4-week-old mice resulted in 15% base editing at the target site and restored near-normal levels of visual function. In vivo delivery using LVs to other organs, including the bone marrow, brain, and liver, has also been demonstrated, although these applications are limited to gene augmentation therapy and not gene editing (Dalsgaard et al., 2018; Milone and O'Doherty, 2018; Richter et al., 2017). A significant disadvantage of using LVs for in vivo delivery applications is the potential for genomic integration, which can lead to detrimental outcomes. Although episomal transgenes from IDLV vectors are designed to be non-integrating, they retain residual genome integration frequencies (Kymalainen et al., 2014; Wang et al., 2010) and also still lead to prolonged expression of the editing agent, which increases risks associated with off-target editing. Notably, the use of LVs in in vivo gene augmentation therapy clinical studies has raised concerns about genotoxicity, immunogenicity, and the high cost of manufacturing (Milone and O'Doherty, 2018), all of which may limit the use of LVs for in vivo gene editing applications. Adenoviral delivery Adenovirus (Ad) is an icosahedral non-enveloped virus, 90–100nm in size, with a large (36 kb) genome (Lee et al., 2017). Adenoviruses deliver DNA cargo that is then episomally maintained in the nucleus of transduced cells (Lee et al., 2017). Ad is the most commonly used viral vector in gene therapy clinical trials worldwide (accounting for >20%), primarily because of its large cargo packaging capacity, well-defined biology, genetic stability, high transduction efficiency, and ability to be produced at high titers on a large scale (Lee et al., 2017). Moreover, there are 57 known serotypes of Ad that infect humans (Lee et al., 2017) and ~100 that infect primates (Nelson and Gersbach, 2016), allowing researchers to modulate Ad tropism by using different capsids. In 2017, Musunuru and coworkers used an Ad to systemically deliver an early-generation CBE into mice, resulting in 28% base editing of Pcsk9 in the liver and a 28% reduction in cholesterol levels four weeks after injection (Chadwick et al., 2017). In another study, Lieber and coworkers used Ads to deliver ABEs to HSCs in vivo (Li et al., 2021). The ABEs were designed to disrupt repressor binding sites in the fetal hemoglobin promoter, which can upregulate fetal hemoglobin expression as a potential therapeutic strategy to treat sickle cell disease and β-thalassemia (Li et al., 2021). This study was the first to report therapeutic in vivo base editing in hematopoietic stem and progenitor cells (HSPCs). Their approach used two Ads: one Ad delivered the base editor cargo and MGMJP140K, a selectable marker, flanked by inverted repeats for genomic integration, and the second Ad contained the transposase and recombinase machinery required to integrate the selection marker into the genome of transduced cells. After 16 weeks post-Ad treatment and four selection rounds, the researchers observed 20% editing of the target site in HSPCs, which led to therapeutic levels of fetal hemoglobin expression. Ads have also been used recently for in vivo prime editing applications. Schwank and coworkers used an Ad to deliver PE2 without the RNaseH domain to neonatal or adult mice (Böck et al., 2022). They observed 58% and 36% prime editing in the hepatocytes of neonates and adult mice, respectively. While using Ads to deliver gene editing agents has yielded efficient in vivo editing, it has also led to the generation of neutralizing antibodies against Cas9 (Wang et al., 2015), potentially due to the immunogenic nature of the vector. Hence, drawbacks of using Ads for in vivo gene editing applications include immunogenicity and its inherently high adjuvant nature that can lead to T cell-mediated cytotoxicity (Geutskens et al., 2000; Raper et al., 2003). Efforts to make the virus “stealth-like” by minimizing the expression of viral antigens can significantly reduce its immunogenicity (Lee et al., 2017). The use of adenoviruses as COVID-19 vaccines has generated new excitement around the technology. However, broader applications of Ad for in vivo gene editing will require further engineering efforts. The future of in vivo gene editing using viral vectors Overall, viral vectors have shown great promise for delivering gene editing agents in vivo across many pre-clinical studies and one ongoing clinical trial. To date, viral vectors offer some of the highest gene editing efficiencies observed across many organs due to their inherent abilities to potently transduce diverse cell types in vivo and deliver their nucleic acids cargos. Future improvements to viral vectors will require careful efforts to overcome the challenges outlined above, including the immunogenicity of the vector, prolonged expression of the gene editing agent, off-target gene editing, potential for genomic integration, manufacturing cost, and dose-limiting toxicity (Figure 3). Vector engineering approaches to improve the potency and tissue specificity could reduce the required dose and reduce the cost of manufacturing of viral delivery platforms. Methods to durably silence cargo expression after on-target editing will also substantially improve the safety profile of viral delivery. As discussed below, hybrid viral and non-viral strategies could offer the best of both worlds by combining the robust efficiency of viral delivery with the transient nature of non-viral delivery approaches. Lipid nanoparticle (LNP) delivery Lipid nanoparticles (LNPs) have grown increasingly popular as non-viral vehicles for delivering gene editing agents in vivo. For decades, LNPs have been used to deliver nucleic acid cargos, including siRNAs and therapeutic mRNAs (Cullis and Hope, 2017; Paunovska et al., 2022). To deliver their encapsulated payloads into target cells, they first enter cells through endocytosis, escape endosomes by disrupting endosomal membranes after endosome acidification, and subsequently gain access to the target cell cytosol (Gilleron et al., 2013; Wittrup et al., 2015). LNPs are completely synthetic and are typically composed of four components: a cationic or ionizable lipid, a helper lipid, a polyethylene glycol (PEG)-lipid, and cholesterol (Paunovska et al., 2022) (Figure 4). Varying the identities of these components can yield LNPs with different properties, including distinct pharmacokinetic profiles and abilities to target different cell types (Paunovska et al., 2022). Following extensive development and optimization, LNPs have been approved for use in humans by the US FDA, including via intravenous administration to deliver therapeutic siRNAs to hepatocytes (Adams et al., 2018) and via intramuscular administration to deliver mRNA vaccines (Baden et al., 2021; Polack et al., 2020). As discussed below, LNPs are already being used in a clinical trial to deliver Cas9 nuclease mRNA to the liver and are poised to become a delivery vehicle of choice for many clinical in vivo gene editing applications. LNPs for liver delivery Most intravenously administered nanoparticles accumulate in the liver (Paunovska et al., 2022). Specifically, many LNPs become coated with ApoE lipoproteins in the bloodstream, which leads to LNP uptake by hepatocytes mediated by ApoE:LDL receptor interactions (Akinc et al., 2019; Paunovska et al., 2022). For these reasons, LNPs have thus far been most commonly used to deliver therapeutic cargos to the liver. While LNPs were originally optimized for delivering siRNAs, key advances in LNP formulations enabled efficient encapsulation and delivery of mRNAs instead of siRNAs. Anderson and coworkers optimized a lipid formulation for mRNA delivery (Kauffman et al., 2015) and used this formulation to deliver SpCas9 nuclease mRNA to the mouse liver (Yin et al., 2016). The SpCas9-encoding mRNA was chemically modified to include pseudouridine and 5-methylcytidine, which was important for increasing mRNA stability and reducing cellular innate immune responses to foreign RNAs (Kariko et al., 2008). Initially, only SpCas9 mRNA was delivered within LNPs; an sgRNA expression cassette and HDR donor DNA template were provided on an AAV8 vector that was co-administered along with the LNPs (Yin et al., 2016). This approach led to 24% indels and 0.8% correction of a tyrosinemia-causing mutation in the bulk mouse liver, which was sufficient to cure the disease through the increased fitness of edited hepatocytes and the eventual replacement of non-edited liver cells with edited ones in the treated animals. Subsequently, Anderson and coworkers demonstrated that chemically modifying the sgRNA to include a specific combination of 2’OMe, 2’F, and phosphorothioate linkages enabled more efficient editing when these sgRNAs were encapsulated in LNPs (Yin et al., 2017b). In mice, an intravenous injection of LNPs that co-encapsulated SpCas9 mRNA and chemically modified sgRNAs targeting Pcsk9 led to 80% editing and a reduction of serum Pcsk9 to undetectable levels (Yin et al., 2017b). These results were some of the first to demonstrate the promise of using LNPs encapsulating Cas9 nuclease mRNA and chemically modified sgRNAs to achieve therapeutic levels of gene editing in mice. Many other groups have also developed LNP formulations that enable efficient delivery of Cas9 nuclease mRNA and sgRNAs to the mouse liver. Siegwart and coworkers developed zwitterionic amino lipid (ZAL) nanoparticles, which successfully co-delivered Cas9 nuclease mRNA and sgRNA to the liver of nuclease reporter mice (Miller et al., 2017). Dong, Tan, and coworkers developed TT3-based lipid-like nanoparticles (LLNs) and demonstrated that they could achieve 30% indels at Pcsk9 in the liver after intravenously injecting LLN-encapsulated Cas9 nuclease mRNA and sgRNA into mice (Jiang et al., 2017). Xu and coworkers used bioreducible LNPs containing integrated disulfide bonds to encapsulate Cas9 nuclease mRNA and sgRNA, achieving 20% editing of Pcsk9 and 39% editing of Angptl3 in the mouse liver (Liu et al., 2019; Qiu et al., 2021). Collectively, these studies demonstrate that various LNP compositions can support efficient delivery of Cas9 nuclease mRNA and sgRNAs to the liver. In an ongoing phase 1 clinical trial, Gillmore and coworkers recently demonstrated the efficacy of an mRNA LNP approach for in vivo liver gene editing in six patients with hereditary transthyretin amyloidosis (Gillmore et al., 2021). Knockdown of transthyretin (TTR) protein levels reduces ongoing TTR amyloid formation, which can improve disease outcomes (Adams et al., 2018). Previously, the researchers had reported that LNP-delivered Cas9 nuclease mRNA and a mouse Ttr-targeting sgRNA successfully disrupted Ttr in the liver and led to a substantial and durable reduction in Ttr protein levels (Finn et al., 2018). Additional preclinical studies in cynomolgus monkeys demonstrated 73% TTR disruption in the liver and a corresponding >94% reduction in serum TTR protein that was sustained over a period of 12 months (Gillmore et al., 2021). The clinical data revealed that patients who had received the intravenously administered LNP-based drug at a dose of 0.1 mg/kg or 0.3 mg/kg exhibited reductions in serum TTR levels of 53% or 87%, respectively, with minimal adverse effects reported. These results were the first to establish in vivo Cas9 nuclease gene editing in the liver mediated by RNA-encapsulating LNPs as a therapeutic strategy in humans. In addition to delivering Cas9 nuclease mRNA, LNPs have also been used to deliver base editor mRNA to the livers of mice and non-human primates. Xue and coworkers observed 12.5% base editing of a tyrosinemia mutation in the mouse liver mediated by LNP delivery of ABE mRNA (Jiang et al., 2020), and Schwank and coworkers observed 10% base editing of a phenylketonuria mutation in the mouse liver mediated by LNP delivery of SaCas9-BE3 mRNA (Villiger et al., 2021). Recently, separate studies led by Kathiresan, Schwank, and their respective coworkers reported highly efficient (>60%) base editing to disrupt a splice site in PCSK9 in the livers of mice and cynomolgus monkeys, which was mediated by LNP delivery of ABE mRNA (Musunuru et al., 2021; Rothgangl et al., 2021). A single LNP administration in cynomolgus monkeys led to a substantial (90%) and sustained (>8 months) knockdown of serum PCSK9 protein and a 60% reduction in blood cholesterol (Musunuru et al., 2021). These promising preclinical results using ABE mRNA LNPs in NHPs, combined with the promising clinical results using Cas9 nuclease mRNA LNPs in human patients, suggest that LNPs could be used in the future to mediate in vivo liver base editing treatments for indications such as hypercholesterolemia and other genetic liver diseases. LNPs for non-liver delivery Because most intravenously administered LNPs naturally accumulate in the liver, achieving non-liver gene editing mediated by LNPs is challenging (Wei et al., 2020a). One approach for subverting the natural liver-targeting nature of LNPs is to administer them by local injection rather than intravenous injection. Multiple laboratories have previously reported successful nuclease editing and base editing in the mouse inner ear and retina following local administration of lipid-encapsulated RNPs (Gao et al., 2018; Jang et al., 2021; Yeh et al., 2018; Zuris et al., 2015). However, the ability to use systemically administered LNPs to deliver gene editing agents to non-liver tissues would greatly expand the therapeutic applicability of LNP delivery. Many groups have pursued the development of LNPs that target non-liver tissues. Dahlman and coworkers developed strategies for simultaneously screening hundreds of different LNPs in vivo to identify LNP compositions that enable non-liver delivery (Dahlman et al., 2017; Sago et al., 2018). These strategies mark distinct LNP formulations with unique DNA barcodes, inject pooled barcoded LNP libraries into mice, and sequence the barcodes extracted from a tissue of interest to reveal the identity of the LNP(s) that enabled delivery to that tissue. Using these strategies, Dahlman and coworkers identified LNPs that delivered Cas9 nuclease mRNA and sgRNA in mice to splenic endothelial cells as efficiently as to hepatocytes (Sago et al., 2018). Siegwart and coworkers developed selective organ targeting (SORT) LNPs by adding an additional charged lipid component to modulate the internal charge of the particles without substantially disrupting the standard four-component nature of efficient LNPs (Cheng et al., 2020; Wei et al., 2020a). They found that changing the charge and concentration of this additional component was sufficient to direct LNPs to either the lung or spleen without targeting the liver in mice (Cheng et al., 2020). These SORT LNPs were used to deliver Cas9 mRNA and sgRNA specifically to the lung, achieving 15% editing of bulk lung tissue. Some SORT LNPs can be formulated with permanently cationic lipids and therefore can be assembled in neutral instead of acidic buffers, which enabled packaging of Cas9 RNPs into LNPs for the first time (Wei et al., 2020b). Together, these studies and others have demonstrated that LNP compositions can be altered to modulate tissue-targeting capabilities, although specific rules for retargeting LNPs in this way remain unknown. Another strategy for directing LNPs to non-liver tissues involves conjugating targeting groups such as antibody fragments to the surface of LNPs (Kedmi et al., 2018; Paunovska et al., 2022; Veiga et al., 2018). A particularly noteworthy example of this strategy was recently reported by Epstein and coworkers, who used intravenously administered anti-CD5 antibody-conjugated LNPs to target T cells and transiently generate chimeric antigen receptor T cells that could treat cardiac injury in mice (Rurik et al., 2022). While these active targeting approaches have not yet been applied to deliver gene editing agents to non-liver tissues, they offer the potential to enable non-liver in vivo gene editing using systemically administered LNPs in the near future. Advantages of LNP delivery and future prospects LNP delivery offers several advantages over viral delivery, especially when delivering gene editing agents. LNP delivery results in transient expression of gene editing agents, which is known to minimize the potential for off-target editing relative to prolonged expression from episomal or integrated viral genomes (Newby and Liu, 2021). Prolonged expression of gene editing agents could also result in immune recognition of edited cells, which might impact the long-term persistence of edited cells (Wagner et al., 2021). Additionally, since LNPs are synthetic, the immunogenicity of LNPs is much lower than that of viruses and can support repeat dosing in some cases (Kenjo et al., 2021). Currently used LNP components are typically biodegradable and non-toxic in vivo (Maier et al., 2013; Witzigmann et al., 2020). Doses of LNPs that are sufficient to support robust gene editing have not shown significant adverse effects in mice or NHPs and have thus far shown good safety profiles in humans (Gillmore et al., 2021). Importantly, LNP manufacturing for large-scale production has been demonstrated to be feasible (Schoenmaker et al., 2021), opening up avenues for additional clinical programs that use LNP delivery for in vivo gene editing. The development of LNPs that enable efficient non-liver delivery remains a critical goal for the therapeutic gene editing field. Understanding the mechanisms by which different LNP formulations enable different tissue-targeting properties might enable better methodologies for engineering new LNPs with desired targeting capabilities (Dilliard et al., 2021). Cell types of high interest include hematopoietic stem cells (HSCs), as LNPs capable of delivering gene editing agents to bone marrow HSCs following an intravenous or intraosseous injection could revolutionize the treatment of genetic blood disorders by obviating the need to harvest, edit ex vivo, and transplant patient HSCs. Overall, given their recent successes as delivery vehicles for multiple types of therapeutic RNAs in humans (Adams et al., 2018; Baden et al., 2021; Gillmore et al., 2021; Polack et al., 2020), LNPs are likely to be used extensively for in vivo gene editing in the liver and potentially in other organs. Virus-like particle (VLP) delivery Virus-like particles (VLPs) have emerged as potentially promising vehicles for delivering gene editing agents. VLPs are non-infectious assemblies of viral proteins that package desired cargo mRNAs, proteins, or RNPs in addition to or instead of viral genetic material (Lyu et al., 2020). Because VLPs are derived from existing viral scaffolds, they exploit natural properties of viruses that enable efficient intracellular delivery, including their ability to encapsulate cargos, escape endosomes, and be reprogrammed to target different cell types. However, unlike viruses, VLPs transiently deliver gene editing agents as mRNA or protein instead of as DNA, which substantially reduces the risks of off-target gene editing and viral genome integration (Chandler et al., 2017). For these reasons, VLPs are attractive vehicles for delivering gene editing agents as they can offer key benefits of both viral and non-viral delivery. Nearly all reported VLP architectures for delivering mRNA or protein cargos are based on retroviruses, as retroviruses possess several characteristics that are ideal for VLPs. Immature retroviral particles are spherical and typically lack rigid structural symmetry (Zhang et al., 2015), which allows increased flexibility to encapsulate desired cargos compared to most non-enveloped icosahedral viruses. Furthermore, the large particle diameter (100–200 nm) of retroviruses (Zhang et al., 2015) provides more physical space for packaging large cargos such as Cas9. Finally, retroviruses are inherently modular with respect to cell targeting and cargo packaging; cell-type specificity is dictated by the envelope glycoproteins, and cargo packaging is controlled by the capsid proteins (Cronin et al., 2005). This modularity suggests that a VLP capsid architecture that efficiently packages desired cargo could be readily combined with various existing envelope glycoproteins that are currently used to modulate the tropism of retroviruses. While retroviral VLPs have been explored for decades as delivery vehicles for mRNAs and proteins, recent efforts were the first to realize the potential of VLPs to mediate efficient in vivo delivery of gene editing agents. mRNA-packaging VLPs Packaging a desired mRNA cargo within VLPs requires a molecular mechanism by which a specific mRNA can be recognized by viral capsid proteins and subsequently incorporated into virions. Retroviral RNA genomes contain a packaging signal (Ψ) that directs the encapsulation of viral RNA into virions, and thus an mRNA cargo engineered to contain Ψ should similarly be incorporated into virions. Some of the first mRNA-packaging VLPs designed by Baum and coworkers used Ψ to encapsulate Cre-recombinase mRNA into murine leukemia virus (MLV) particles (Galla et al., 2004). Importantly, this Ψ-containing mRNA was additionally modified so that it would not be reverse transcribed by the MLV reverse transcriptase, resulting in transient delivery of mRNA rather than stable integration of viral cDNA into the genomes of transduced cells. However, only two copies of Ψ-containing RNA could be packaged per viral particle, which motivated the development of alternative strategies to package greater amounts of mRNA cargo into VLPs. To improve the mRNA-packaging potential of VLPs, Pagès and coworkers used the interaction between the MS2 coat protein (MS2cp) and MS2 aptamer (MS2apt) to direct packaging of mRNA cargo into modified HIV-1 particles (Prel et al., 2015). In their designs, they replaced the ZF2 domain of the HIV-1 nucleocapsid with MS2cp and included twelve copies of MS2apt at the 3’ end of a luciferase mRNA cargo. This approach enabled the packaging of 5-6 copies of luciferase mRNA per VLP, an improvement over Ψ-mediated RNA packaging. Therefore, the strategy of modifying retroviral capsid proteins to include MS2cp and cargo mRNAs to include MS2apt was adopted as a promising way to generate mRNA-packaging VLPs (Figure 5). Several groups have since demonstrated the use of MS2cp/MS2apt to package Cas9 nuclease mRNA into VLPs. Galla and coworkers fused two copies of MS2cp to the C-terminus of MLV gag along with two copies of MS2apt within the 3’ UTR of SpCas9 mRNA and at the 3’ end of the sgRNA (Knopp et al., 2018). This strategy enabled successful delivery to HEK293T cells, Jurkat cells, and primary human fibroblasts, but insufficient delivery of the sgRNA limited gene editing efficiencies. Lu and coworkers fused two copies of MS2cp directly downstream of the HIV-1 nucleocapsid ZF2 domain along with one copy of MS2apt within the 3’ UTR of SaCas9 mRNA (Lu et al., 2019). In this system, the SaCas9 sgRNA was encoded in a separate IDLV that was used to co-transduce target cells along with SaCas9 mRNA-containing VLPs. This SaCas9 VLP plus IDLV system exhibited efficient editing in HEK293T cells, but its efficiency was not evaluated in other cell types or in vivo. Cai and coworkers developed a similar system (mLPs) for packaging SpCas9 mRNA into HIV-1 VLPs (Ling et al., 2021). In mLPs, one copy of MS2cp is fused to the N-terminus of HIV-1 gag-pol and six copies of MS2apt are added within the 3’ UTR of SpCas9 mRNA. Additionally, Cai and coworkers produced all-in-one mLPs that packaged both SpCas9 mRNA as well as an IDLV genome expressing an SpCas9 sgRNA. mLPs displayed efficient editing in HEK293T, NIH3T3, K562, and Jurkat cells. Notably, a single subretinal injection of mLPs into mice mediated 44% knockout of Vegfa in retinal pigment epithelial (RPE) cells, which was sufficient to prevent wet age-related macular degeneration. In a separate study, Cai and coworkers also demonstrated that an intracorneal injection of mLPs loaded with SpCas9 mRNA and two sgRNA expression cassettes cured herpetic stromal keratitis in mice by simultaneously targeting two essential herpesvirus genes (Yin et al., 2021). These results highlight the in vivo therapeutic utility of Cas9 nuclease mRNA-packaging VLPs for the treatment of ocular diseases. One drawback of using mRNA-packaging VLPs for the delivery of Cas9-based gene editing agents is that there are various challenges associated with sgRNA delivery. Guide RNAs that are not chemically modified are rapidly degraded unless they are protected by complexing with Cas9 protein (Allen et al., 2020). Guide RNAs packaged alongside Cas9 mRNAs in VLPs may therefore be substantially degraded before Cas9 protein is synthesized in the transduced cells. Although sgRNA expression cassettes encoded on IDLVs enable efficient editing by Cas9 mRNA VLPs, such cassettes persist as episomal DNA in transduced cells. As previously noted, while IDLVs exhibit substantially minimized rates of genomic integration compared to integration-competent lentiviral vectors, they still support a detectable frequency of genomic integration (Kymalainen et al., 2014; Wang et al., 2010), which increases the risks of this approach. Protein- or RNP-packaging VLPs Packaging desired protein or RNP cargos within VLPs requires a strategy for localizing target proteins into VLPs as they form. To accomplish this, researchers have fused desired cargo proteins to viral structural proteins, including at various locations within retroviral gag polyproteins; this strategy directs the cargo into virions during the capsid self-assembly process (Kaczmarczyk et al., 2011; Voelkel et al., 2010) (Figure 5). In most cases, the gag and cargo are linked by a short peptide sequence that is cleaved by the co-encapsulated viral protease following virion maturation (after the cargo is successfully packaged), enabling cargo release into the transduced cells (Kaczmarczyk et al., 2011; Voelkel et al., 2010). This approach has been used to package and deliver various protein cargos within VLPs (Cai et al., 2014). Similar strategies were used to package Cas9 nuclease protein into VLPs. Manjunath and coworkers reported VLPs that contained SpCas9 fused to the N-terminus of HIV-1 gag-pol via a HIV-1 protease-cleavable linker and expressed sgRNAs from a cassette encoded on a co-packaged lentiviral genome (Choi et al., 2016). This VLP construct achieved 14–28% editing in Jurkat cells. Doudna and coworkers reported related VLPs that contained SpCas9 fused to the C-terminus of HIV-1 gag via a HIV-1 protease-cleavable linker (Hamilton et al., 2021). In this Cas9-VLP construct, the sgRNA was either encoded on a co-packaged lentiviral genome or expressed from a non-lentiviral plasmid during VLP production. The latter approach for sgRNA packaging relies on the high affinity between Cas9 and its sgRNA, which enables gag–Cas9 fusions to be loaded with sgRNAs prior to packaging within VLPs. These Cas9-VLPs enabled up to 90% editing in Jurkat cells and up to 60% editing in primary human T cells. These efficiencies represented substantial improvements over previous results, likely because of the improved Cas9 fusion orientation at the C-terminus rather than the N-terminus of HIV-1 gag, as well as the ability to package Cas9/sgRNA RNPs rather than Cas9 protein alone. Additionally, Doudna and coworkers demonstrated that Cas9-VLPs could be targeted to specific T cell subpopulations by pseudotyping the particles with different envelope glycoproteins (Hamilton et al., 2021). Ricci and coworkers also leveraged gag–Cas9 fusions to generate RNP-packaging VLPs (Mangeot et al., 2019). These VLPs, also termed “nanoblades”, contained SpCas9 fused to the C-terminus of MLV gag via a MLV protease-cleavable linker and expressed sgRNAs from a non-viral plasmid during VLP production to enable direct packaging of Cas9 RNPs. Nanoblades displayed efficient editing in vitro in HEK293T cells (80-90%), primary human T cells (30%), primary human HSPCs (40%), and other cell types. Notably, a single intravenous injection of nanoblades into mice achieved up to 10% editing in the liver, representing the first demonstration of the in vivo efficacy of Cas9 nuclease RNP-packaging VLPs. Other groups have developed strategies for packaging Cas9 RNPs into VLPs that do not involve gag–Cas9 fusions. Indikova and Indik fused Cas9 to the C-terminus of HIV-1 VPR, an accessory protein that is packaged into HIV-1 particles via interactions with the p6 domain of HIV-1 gag (Indikova and Indik, 2020). This VLP construct achieved >90% editing in HEK293T cells but lower efficiency in primary human T cells (15%) compared to the Doudna group’s HIV-1 gag-fused Cas9-VLPs. Lu and coworkers utilized aptamer and aptamer binding protein interactions, which were previously used to package Cas9 mRNA within VLPs, to instead package Cas9 RNPs (Lu et al., 2021; Lyu et al., 2019). They replaced the tetraloop of the sgRNA with a com aptamer, fused a Com-binding protein directly downstream of the HIV-1 nucleocapsid ZF2 domain, and expressed these constructs along with free Cas9 protein in VLP producer cells. In this approach, RNP packaging is driven by the sgRNA:VLP capsid interaction and requires Cas9 protein to complex with aptamer-containing sgRNA prior to RNP loading into particles. Lu and coworkers also showed that this strategy could be used to package adenine base editor RNPs in addition to Cas9 nuclease RNPs (Lyu et al., 2021). These VLP constructs achieved 70-80% editing in HEK293T cells. Hotta and coworkers employed a distinct strategy for RNP packaging that used the small molecule AP21967 (a rapamycin analog) to dimerize FRB–Cas9 fusions with FKBP12–HIV-1gag fusions during particle formation in producer cells (Gee et al., 2020). This strategy, also termed “NanoMEDIC”, mediated 40% deletion of dystrophin exon 45 in DMD patient-derived iPSCs and 6% deletion of exon 45 in gastrocnemius muscle tissue following intramuscular injection into mice. While various Cas9 RNP-packaging VLPs exhibited promising efficiencies in vitro, all of the systems described above were either not tested in vivo or exhibited low in vivo efficacy (<10% editing). We recently developed engineered VLPs (eVLPs) based on Moloney MLV (MMLV) that package Cas9 nuclease or base editor RNPs and mediate potent, therapeutic levels of gene editing across multiple organs in mice (Banskota et al., 2022). We identified key bottlenecks that limit VLP potency in vivo and engineered solutions in eVLPs to overcome these bottlenecks. First, we engineered the protease-cleavable linker sequence between the MMLV gag and protein cargo to improve cargo release after eVLP maturation while minimizing premature cleavage prior to particle formation. Next, we added nuclear export sequences to modulate the localization of the MMLV gag–cargo fusion selectively in producer cells and substantially improve cargo loading into eVLPs. Finally, we engineered an optimal stoichiometry of viral structural components (MMLV gag-pro-pol) and cargo to maximize eVLP efficiencies. eVLPs mediated efficient gene editing in vitro in HEK293T cells (>95% editing), 3T3 fibroblasts (>80% editing), Neuro-2a cells (>90% editing), primary human and mouse fibroblasts (>90% editing), and primary human T cells (50–60% editing). eVLPs successfully delivered base editor RNPs in vivo following a local injection into the mouse brain, resulting in 5% bulk editing and 60% editing in cells enriched for VLP transduction. Additionally, a single subretinal injection of eVLPs into a mouse model of genetic blindness efficiently corrected the disease-causing point mutation (20–30% editing in mouse RPE cells) and improved visual function. Finally, a single intravenous injection of eVLPs into mice achieved 63% editing of Pcsk9 in the liver and 78% knockdown of serum Pcsk9 levels, which are comparable to Pcsk9 editing and knockdown efficiencies observed with AAV and LNP mRNA delivery. These results demonstrate that rational engineering of VLP architectures to improve potency was required to enable efficient in vivo editing by RNP-packaging VLPs and establish eVLPs as the most potent RNP-packaging VLPs reported to date. Advantages of VLP delivery and future prospects A major advantage of using VLPs to deliver gene editing agents is that VLP delivery results in minimal off-target editing. We and others have demonstrated that VLPs offer substantially minimized off-target editing relative to plasmid and viral delivery in vitro (Banskota et al., 2022; Mangeot et al., 2019). Additionally, we recently demonstrated that RNP-packaging eVLPs offer minimal off-target editing in vivo, including minimized off-target DNA base editing in the mouse liver relative to AAV delivery and minimized off-target RNA base editing in the mouse retina relative to lentiviral delivery (Banskota et al., 2022). Given that the latest generation of RNP-packaging VLPs exhibit on-target editing efficiencies comparable to those achieved by mRNA-packaging VLPs or LNPs, we anticipate that RNP-packaging VLPs will be the preferred delivery vehicles for many applications due to the fact that they offer the shortest exposure to gene editing agents and therefore the lowest potential for off-target editing. We and others have demonstrated that the cell-type specificity of VLPs in vitro can be altered by using different envelope glycoproteins (Banskota et al., 2022; Hamilton et al., 2021; Mangeot et al., 2019). To further improve the broad therapeutic applicability of VLP delivery, it will be important to demonstrate delivery to additional organs, which could be achieved in part by using different envelope glycoproteins or other targeting moieties. The ability to do so would realize the full potential of VLPs as a delivery modality that combines the programmable tropism of viruses with the transient delivery of mRNAs and RNPs. While we showed that systemic administration of eVLPs was non-toxic in mice, future studies should further characterize the safety profile of VLPs in vivo. Because all of the VLPs reviewed above are derived from viral scaffolds, the immunogenicity of VLPs should also be evaluated. Recently, Zhang and coworkers reported that the mammalian retrovirus-like protein PEG10 can be programmed to package desired mRNA cargos, including Cas9 nuclease (Segel et al., 2021). With in vivo validation and further development to improve delivery efficiency, the PEG10-based "SEND” platform could potentially offer minimized immunogenicity relative to retroviral VLPs, as it uses an endogenous mammalian protein scaffold. Finally, it will be critical to establish the feasibility of scaling up VLP production to quantities required for pre-clinical studies in large animal models and beyond. If successful, such studies could pave the way for the use of VLPs in the clinic as delivery vehicles for gene editing agents that offer several of the most important features of both viral and non-viral delivery technologies. Future prospects and conclusions As shown by the examples summarized above, the era of therapeutic in vivo gene editing in humans has already arrived. Extensive development and optimization of CRISPR-Cas technologies have yielded robust tools for gene editing, including programmable nucleases, base editors, and prime editors. Pairing these gene editing tools with efficient in vivo delivery methods, including viral vectors, LNPs, and VLPs, has led to numerous demonstrations of in vivo gene editing, from proof-of-concept applications in animal models to therapeutic outcomes in humans. With current in vivo delivery modalities, gene editing agents can be readily delivered to cells in the liver via intravenous injection and to cells in the eye via intraocular injection. For this reason, in vivo gene editing therapies in the near future will likely treat diseases that can be addressed through editing the liver or eye. Efficient delivery to non-liver tissues following intravenous administration remains a major challenge for most delivery vehicles. The use of naturally occurring and newly engineered AAV capsids is a promising strategy for targeting non-liver tissues, including the CNS (Goertsen et al., 2022), skeletal muscle (Tabebordbar et al., 2021), and heart (Koblan et al., 2021b). Analogous strategies could prove useful for retargeting VLPs to target new cell types by using different envelope glycoproteins. While systematic rules for reformulating LNPs to target different cell populations are not well understood, emerging methods for conjugating targeting moieties to the surface of LNPs could prove especially useful (Paunovska et al., 2022). For every distinct therapeutic application of in vivo gene editing, it will also be important to understand whether tissue-specific targeting and editing is required, or if targeting and editing a desired tissue in addition to the liver is acceptable. Cell type-specific delivery within a particular tissue could offer advantages for certain therapeutic applications (Kwon et al., 2020; Nance et al., 2019; Tabebordbar et al., 2016). Importantly, in vivo gene editing strategies should only target somatic cell populations and must always avoid collateral germline editing, as clinical human germline editing raises serious ethical concerns (Lander et al., 2019; Saha et al., 2021). Immunogenicity concerns associated with the in vivo delivery of gene editing agents are complex and remain to be characterized comprehensively. Preexisting immunity to delivery vehicles could interfere with in vivo gene editing therapies, as preexisting antibodies could directly neutralize viral vectors (Verdera et al., 2020; Weber, 2021). Preexisting cellular immunity to Cas9 or other components of gene editing agents could lead to immune-mediated clearance of transduced and edited cells (Crudele and Chamberlain, 2018; Wagner et al., 2021). Over time, prolonged expression of gene editing agents in edited cells could provoke adaptive immune responses (Wagner et al., 2021), which could also lead to clearance of transduced and edited cells. While a single transient administration of a gene editing agent in vivo in the absence of any preexisting immunity has been shown to be effective (Finn et al., 2018; Gillmore et al., 2021; Musunuru et al., 2021), such an administration could in some cases trigger an adaptive immune response that would limit the efficacy of repeat dosing or the use of related editing agents in the future (Rothgangl et al., 2021). These concerns highlight the advantages of delivery methods that can support a one-time, transient, and potent delivery of a therapeutic gene editing agent in vivo, such as LNPs and VLPs. An additional advantage of transiently delivering gene editing agents is that transient delivery leads to reduced off-target editing compared to prolonged delivery (Anzalone et al., 2020; Banskota et al., 2022; Doman et al., 2020; Newby and Liu, 2021). Minimizing off-target gene editing in vivo is especially important, as even highly rare off-target editing events could install cancer-causing mutations. While viral delivery generally leads to prolonged expression, methods to turn off the expression of the gene editing agent after on-target editing is complete could be useful for reducing off-target editing. mRNA delivery (e.g., by LNPs) is transient and offers favorable on-target vs. off-target editing profiles, but RNP delivery minimizes the potential for off-target editing as it offers the shortest exposure to gene editing agents (Banskota et al., 2022; Newby et al., 2021). However, current methods for efficiently delivering RNPs to multiple organs in vivo are limited to eVLPs. As RNP delivery vehicles are the most attractive from a safety perspective, the development of improved RNP delivery vehicles for therapeutic in vivo gene editing will likely be highly impactful. As additional in vivo gene editing therapies move rapidly toward the clinic, the availability of robust in vivo delivery methods will be critical. Future advances in delivery technologies will help to enable a wide range of in vivo gene editing therapies and potentially other macromolecular therapeutic approaches. Acknowledgements This work was supported by NIH UG3AI150551, U01AI142756, R35GM118062, RM1HG009490, HHMI, and the Bill and Melinda Gates Foundation. A.R. gratefully acknowledges support from an NSF graduate research fellowship. We thank Dr. Anahita Vieira for assistance editing the manuscript. Some images in Figures 2, 3, and 4 were created with BioRender.com. Figure 1. Overview of therapeutic gene editing technologies Nucleases create targeted double-strand DNA breaks (DSBs), which generally lead to uncontrolled mixtures of insertions and deletions (indels) that are useful for gene disruption. In certain types of dividing cells, DSBs in the presence of a DNA donor template can also lead to homology-directed repair (HDR) outcomes that can support gene correction, though indel byproducts typically accompany HDR outcomes. Base editors mediate targeted C•G-to-T•A, A•T-to-G•C, or C•G-to-G•C conversions with minimal indel byproducts. Prime editors enable targeted single-nucleotide conversions, insertions, deletions, and combinations thereof with minimal indel byproducts. See also Anzalone et al., 2020 for a more detailed description of gene editing mechanisms. Figure 2. Requirements for efficient in vivo delivery of gene editing agents (A) An appropriate delivery vehicle (gray circles) for gene editing agents must efficiently encapsulate DNA or mRNA encoding gene editing agents, or gene editing proteins or ribonucleoproteins (RNPs). Delivery vehicles must protect their cargos from sequestration or degradation in vivo prior to encountering target cells. (B) Delivery vehicles must bind target cells, typically by engaging cell surface receptors with complementary molecules on the surface of the delivery vehicle. (C) Delivery vehicles must traverse the target cell membrane, typically through receptor-mediated endocytosis. (D) Following endocytosis, delivery vehicles must either escape endosomes and release their cargo, or fuse with endosomes to release their cargo into the target cell cytosol. The cargo must then be trafficked to the appropriate cellular compartment (typically the nucleus) for successful gene editing to occur. Figure 3. Overview and comparison of viral delivery methods (A) Adeno-associated viruses are single-stranded DNA viruses with cargo capacity of 5 kb. (B) Lentiviral vectors are enveloped viruses with that package a single-stranded RNA genome of up to 10 kb. (C) Adenoviral vectors are double-stranded DNA viruses with a packaging capacity of 8 kb that can be expanded to 36 kb in “gutless” vectors devoid of all the viral protein-coding genes. Figure 4. Lipid nanoparticle (LNP) delivery LNPs consist of four key components and can efficiently encapsulate various RNAs. Encapsulated mRNAs are typically modified by including alternative nucleotides during in vitro transcription, such as N1-methylpseudouridine, to increase cellular stability after delivery. Encapsulated guide RNAs are chemically modified at various positions, including with 2’-O-methylation and phosphorothioate linkages, which enhance the stability of the guide RNA. Figure 5. Virus-like particle (VLP) delivery General schematic of the most important components of mRNA-packaging VLPs (left) and protein- or RNP-packaging VLPs (right). In both types of VLPs, the retroviral gag and gag-pro-pol polyproteins provide structural stability and the viral protease required for cleaving the polyproteins into distinct subunits during particle maturation. In mRNA-packaging VLPs, fusion of gag with an RNA-binding protein (RBP) enables encapsulation of mRNA cargo containing the RNA aptamer recognized by the RBP. If necessary, a guide RNA is typically encoded on an integration-deficient lentivirus (IDLV) genome. In RNP-packaging VLPs, fusion of gag with protein cargo via a viral protease-cleavable linker directs encapsulation of protein into particles as they form. Cleavage of the linker after particle maturation enables the release of free protein cargo into transduced cells. When packaging RNPs, guide RNAs can be co-packaged into particles due to the intrinsic affinity between the Cas9 protein and its guide RNA. In engineered VLPs (eVLPs), cargo packaging, release, and localization have been optimized through protein engineering (Banskota et al., 2022). This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Declaration of Interests The authors have filed patent applications on gene editing technologies and delivery technologies through the Broad Institute of MIT and Harvard. D.R.L. is a consultant and cofounder of Beam Therapeutics, Prime Medicine, Pairwise Plants, Editas Medicine, Chroma Medicine, and Envelop Therapeutics, companies that use and/or deliver genome editing or genome engineering agents. References Adams D , Gonzalez-Duarte A , O'Riordan WD , Yang CC , Ueda M , Kristen AV , Tournev I , Schmidt HH , Coelho T , Berk JL , (2018). Patisiran, an RNAi Therapeutic, for Hereditary Transthyretin Amyloidosis. N Engl J Med 379 , 11–21.29972753 Akinc A , Maier MA , Manoharan M , Fitzgerald K , Jayaraman M , Barros S , Ansell S , Du X , Hope MJ , Madden TD , (2019). The Onpattro story and the clinical translation of nanomedicines containing nucleic acid-based drugs. Nat Nanotechnol 14 , 1084–1087.31802031 Alanis-Lobato G , Zohren J , McCarthy A , Fogarty NME , Kubikova N , Hardman E , Greco M , Wells D , Turner JMA , and Niakan KK (2021). Frequent loss of heterozygosity in CRISPR-Cas9-edited early human embryos. Proc Natl Acad Sci U S A 118 . Allen D , Rosenberg M , and Hendel A (2020). Using Synthetically Engineered Guide RNAs to Enhance CRISPR Genome Editing Systems in Mammalian Cells. Front Genome Ed 2 , 617910.34713240 Anzalone AV , Koblan LW , and Liu DR (2020). Genome editing with CRISPR-Cas nucleases, base editors, transposases and prime editors. Nat Biotechnol 38 , 824–844.32572269 Anzalone AV , Randolph PB , Davis JR , Sousa AA , Koblan LW , Levy JM , Chen PJ , Wilson C , Newby GA , Raguram A , (2019). Search-and-replace genome editing without double-strand breaks or donor DNA. Nature 576 , 149–157.31634902 Aranko AS , Wlodawer A , and Iwai H (2014). Nature's recipe for splitting inteins. Protein Eng Des Sel 27 , 263–271.25096198 Asokan A , Conway JC , Phillips JL , Li C , Hegge J , Sinnott R , Yadav S , DiPrimio N , Nam HJ , Agbandje-McKenna M , (2010). Reengineering a receptor footprint of adeno-associated virus enables selective and systemic gene transfer to muscle. Nat Biotechnol 28 , 79–82.20037580 Baden LR , El Sahly HM , Essink B , Kotloff K , Frey S , Novak R , Diemert D , Spector SA , Rouphael N , Creech CB , (2021). Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine. N Engl J Med 384 , 403–416.33378609 Banskota S , Raguram A , Suh S , Du SW , Davis JR , Choi EH , Wang X , Nielsen SC , Newby GA , Randolph PB , (2022). Engineered virus-like particles for efficient in vivo delivery of therapeutic proteins. Cell 185 , 250–265 e216.35021064 Bareford LM , and Swaan PW (2007). Endocytic mechanisms for targeted drug delivery. Adv Drug Deliv Rev 59 , 748–758.17659804 Böck D , Rothgangl T , Villiger L , Schmidheini L , Matsushita M , Mathis N , Ioannidi E , Rimann N , Grisch-Chan HM , Kreutzer S , (2022). In vivo prime editing of a metabolic liver disease in mice. Sci Transl Med 14 , eabl9238.35294257 Bottros MM , and Christo PJ (2014). Current perspectives on intrathecal drug delivery. J Pain Res 7 , 615–626.25395870 Byrne LC , Day TP , Visel M , Strazzeri JA , Fortuny C , Dalkara D , Merigan WH , Schaffer DV , and Flannery JG (2020). In vivo-directed evolution of adeno-associated virus in the primate retina. JCI Insight 5 . Cai Y , Bak RO , and Mikkelsen JG (2014). Targeted genome editing by lentiviral protein transduction of zinc-finger and TAL-effector nucleases. Elife 3 , e01911.24843011 Chadwick AC , Wang X , and Musunuru K (2017). In Vivo Base Editing of PCSK9 (Proprotein Convertase Subtilisin/Kexin Type 9) as a Therapeutic Alternative to Genome Editing. Arterioscler Thromb Vasc Biol 37 , 1741–1747.28751571 Chandler RJ , Sands MS , and Venditti CP (2017). Recombinant Adeno-Associated Viral Integration and Genotoxicity: Insights from Animal Models. Hum Gene Ther 28 , 314–322.28293963 Chapman JR , Taylor MR , and Boulton SJ (2012). Playing the end game: DNA double-strand break repair pathway choice. Mol Cell 47 , 497–510.22920291 Charlesworth CT , Deshpande PS , Dever DP , Camarena J , Lemgart VT , Cromer MK , Vakulskas CA , Collingwood MA , Zhang L , Bode NM , (2019). Identification of preexisting adaptive immunity to Cas9 proteins in humans. Nat Med 25 , 249–254.30692695 Chemello F , Chai AC , Li H , Rodriguez-Caycedo C , Sanchez-Ortiz E , Atmanli A , Mireault AA , Liu N , Bassel-Duby R , and Olson EN (2021). Precise correction of Duchenne muscular dystrophy exon deletion mutations by base and prime editing. Sci Adv 7 . Chen L , Park JE , Paa P , Rajakumar PD , Prekop HT , Chew YT , Manivannan SN , and Chew WL (2021a). Programmable C:G to G:C genome editing with CRISPR-Cas9-directed base excision repair proteins. Nat Commun 12 , 1384.33654077 Chen PJ , Hussmann JA , Yan J , Knipping F , Ravisankar P , Chen PF , Chen C , Nelson JW , Newby GA , Sahin M , (2021b). Enhanced prime editing systems by manipulating cellular determinants of editing outcomes. Cell 184 , 5635–5652 e5629.34653350 Chen Y , Zhi S , Liu W , Wen J , Hu S , Cao T , Sun H , Li Y , Huang L , Liu Y , (2020). Development of Highly Efficient Dual-AAV Split Adenosine Base Editor for In Vivo Gene Therapy. Small Methods 4 . Cheng Q , Wei T , Farbiak L , Johnson LT , Dilliard SA , and Siegwart DJ (2020). Selective organ targeting (SORT) nanoparticles for tissue-specific mRNA delivery and CRISPR-Cas gene editing. Nat Nanotechnol 15 , 313–320.32251383 Chew WL , Tabebordbar M , Cheng JK , Mali P , Wu EY , Ng AH , Zhu K , Wagers AJ , and Church GM (2016). A multifunctional AAV-CRISPR-Cas9 and its host response. Nat Methods 13 , 868–874.27595405 Choi JG , Dang Y , Abraham S , Ma H , Zhang J , Guo H , Cai Y , Mikkelsen JG , Wu H , Shankar P , (2016). Lentivirus pre-packed with Cas9 protein for safer gene editing. Gene Ther 23 , 627–633.27052803 Chu D , Sullivan CC , Weitzman MD , Du L , Wolf PL , Jamieson SW , and Thistlethwaite PA (2003). Direct comparison of efficiency and stability of gene transfer into the mammalian heart using adeno-associated virus versus adenovirus vectors. J Thorac Cardiovasc Surg 126 , 671–679.14502138 Cox DB , Platt RJ , and Zhang F (2015). Therapeutic genome editing: prospects and challenges. Nat Med 21 , 121–131.25654603 Cronin J , Zhang XY , and Reiser J (2005). Altering the tropism of lentiviral vectors through pseudotyping. Curr Gene Ther 5 , 387–398.16101513 Crudele JM , and Chamberlain JS (2018). Cas9 immunity creates challenges for CRISPR gene editing therapies. Nat Commun 9 , 3497.30158648 Cullis PR , and Hope MJ (2017). Lipid Nanoparticle Systems for Enabling Gene Therapies. Mol Ther 25 , 1467–1475.28412170 Dahlman JE , Kauffman KJ , Xing Y , Shaw TE , Mir FF , Dlott CC , Langer R , Anderson DG , and Wang ET (2017). Barcoded nanoparticles for high throughput in vivo discovery of targeted therapeutics. Proc Natl Acad Sci U S A 114 , 2060–2065.28167778 Dalkara D , Byrne LC , Klimczak RR , Visel M , Yin L , Merigan WH , Flannery JG , and Schaffer DV (2013). In vivo-directed evolution of a new adeno-associated virus for therapeutic outer retinal gene delivery from the vitreous. Sci Transl Med 5 , 189ra176. Dalsgaard T , Cecchi CR , Askou AL , Bak RO , Andersen PO , Hougaard D , Jensen TG , Dagnaes-Hansen F , Mikkelsen JG , Corydon TJ , (2018). Improved Lentiviral Gene Delivery to Mouse Liver by Hydrodynamic Vector Injection through Tail Vein. Mol Ther Nucleic Acids 12 , 672–683.30092403 Daneman R , and Prat A (2015). The blood-brain barrier. Cold Spring Harb Perspect Biol 7 , a020412.25561720 Davis JR , Wang X , Witte I , Huang TP , Levy JM , Raguram A , Banskota S , Seidah N , Musunuru K , and Liu DR (2022). Efficient in vivo delivery of adenine base editors in a single adeno-associated virus. in revision. Deverman BE , Pravdo PL , Simpson BP , Kumar SR , Chan KY , Banerjee A , Wu WL , Yang B , Huber N , Pasca SP , (2016). Cre-dependent selection yields AAV variants for widespread gene transfer to the adult brain. Nat Biotechnol 34 , 204–209.26829320 Deverman BE , Ravina BM , Bankiewicz KS , Paul SM , and Sah DWY (2018). Gene therapy for neurological disorders: progress and prospects. Nat Rev Drug Discov 17 , 641–659.30093643 Dilliard SA , Cheng Q , and Siegwart DJ (2021). On the mechanism of tissue-specific mRNA delivery by selective organ targeting nanoparticles. Proc Natl Acad Sci U S A 118 . Doman JL , Raguram A , Newby GA , and Liu DR (2020). Evaluation and minimization of Cas9-independent off-target DNA editing by cytosine base editors. Nat Biotechnol 38 , 620–628.32042165 Dong JY , Fan PD , and Frizzell RA (1996). Quantitative analysis of the packaging capacity of recombinant adeno-associated virus. Hum Gene Ther 7 , 2101–2112.8934224 Doudna JA (2020). The promise and challenge of therapeutic genome editing. Nature 578 , 229–236.32051598 Drouin LM , and Agbandje-McKenna M (2013). Adeno-associated virus structural biology as a tool in vector development. Future Virol 8 , 1183–1199.24533032 Duan D , Yue Y , and Engelhardt JF (2001). Expanding AAV packaging capacity with trans-splicing or overlapping vectors: a quantitative comparison. Mol Ther 4 , 383–391.11592843 Dull T , Zufferey R , Kelly M , Mandel RJ , Nguyen M , Trono D , and Naldini L (1998). A third-generation lentivirus vector with a conditional packaging system. J Virol 72 , 8463–8471.9765382 Edraki A , Mir A , Ibraheim R , Gainetdinov I , Yoon Y , Song CQ , Cao Y , Gallant J , Xue W , Rivera-Perez JA , (2019). A Compact, High-Accuracy Cas9 with a Dinucleotide PAM for In Vivo Genome Editing. Mol Cell 73 , 714–726 e714.30581144 Ferrari S , Vavassori V , Canarutto D , Jacob A , Castiello MC , Javed AO , and Genovese P (2021). Gene Editing of Hematopoietic Stem Cells: Hopes and Hurdles Toward Clinical Translation. Front Genome Ed 3 , 618378.34713250 Fine EJ , Appleton CM , White DE , Brown MT , Deshmukh H , Kemp ML , and Bao G (2015). Trans-spliced Cas9 allows cleavage of HBB and CCR5 genes in human cells using compact expression cassettes. Sci Rep 5 , 10777.26126518 Finn JD , Smith AR , Patel MC , Shaw L , Youniss MR , van Heteren J , Dirstine T , Ciullo C , Lescarbeau R , Seitzer J , (2018). A Single Administration of CRISPR/Cas9 Lipid Nanoparticles Achieves Robust and Persistent In Vivo Genome Editing. Cell Rep 22 , 2227–2235.29490262 Gaj T , Ojala DS , Ekman FK , Byrne LC , Limsirichai P , and Schaffer DV (2017). In vivo genome editing improves motor function and extends survival in a mouse model of ALS. Sci Adv 3 , eaar3952.29279867 Galla M , Will E , Kraunus J , Chen L , and Baum C (2004). Retroviral pseudotransduction for targeted cell manipulation. Mol Cell 16 , 309–315.15494317 Gao X , Tao Y , Lamas V , Huang M , Yeh WH , Pan B , Hu YJ , Hu JH , Thompson DB , Shu Y , (2018). Treatment of autosomal dominant hearing loss by in vivo delivery of genome editing agents. Nature 553 , 217–221.29258297 Gaudelli NM , Komor AC , Rees HA , Packer MS , Badran AH , Bryson DI , and Liu DR (2017). Programmable base editing of A*T to G*C in genomic DNA without DNA cleavage. Nature 551 , 464–471.29160308 Gaudelli NM , Lam DK , Rees HA , Sola-Esteves NM , Barrera LA , Born DA , Edwards A , Gehrke JM , Lee SJ , Liquori AJ , (2020). Directed evolution of adenine base editors with increased activity and therapeutic application. Nat Biotechnol 38 , 892–900.32284586 Gee P , Lung MSY , Okuzaki Y , Sasakawa N , Iguchi T , Makita Y , Hozumi H , Miura Y , Yang LF , Iwasaki M , (2020). Extracellular nanovesicles for packaging of CRISPR-Cas9 protein and sgRNA to induce therapeutic exon skipping. Nat Commun 11 , 1334.32170079 Geutskens SB , van der Eb MM , Plomp AC , Jonges LE , Cramer SJ , Ensink NG , Kuppen PJ , and Hoeben RC (2000). Recombinant adenoviral vectors have adjuvant activity and stimulate T cell responses against tumor cells. Gene Ther 7 , 1410–1416.10981668 Giannoukos G , Ciulla DM , Marco E , Abdulkerim HS , Barrera LA , Bothmer A , Dhanapal V , Gloskowski SW , Jayaram H , Maeder ML , (2018). UDiTaS, a genome editing detection method for indels and genome rearrangements. BMC Genomics 19 , 212.29562890 Gilleron J , Querbes W , Zeigerer A , Borodovsky A , Marsico G , Schubert U , Manygoats K , Seifert S , Andree C , Stoter M , (2013). Image-based analysis of lipid nanoparticle-mediated siRNA delivery, intracellular trafficking and endosomal escape. Nat Biotechnol 31 , 638–646.23792630 Gillmore JD , Gane E , Taubel J , Kao J , Fontana M , Maitland ML , Seitzer J , O'Connell D , Walsh KR , Wood K , (2021). CRISPR-Cas9 In Vivo Gene Editing for Transthyretin Amyloidosis. N Engl J Med 385 , 493–502.34215024 Ginn SL , Amaya AK , Alexander IE , Edelstein M , and Abedi MR (2018). Gene therapy clinical trials worldwide to 2017: An update. J Gene Med 20 , e3015.29575374 Goertsen D , Flytzanis NC , Goeden N , Chuapoco MR , Cummins A , Chen Y , Fan Y , Zhang Q , Sharma J , Duan Y , (2022). AAV capsid variants with brain-wide transgene expression and decreased liver targeting after intravenous delivery in mouse and marmoset. Nat Neurosci 25 , 106–115.34887588 Gorman GS , Chinnery PF , DiMauro S , Hirano M , Koga Y , McFarland R , Suomalainen A , Thorburn DR , Zeviani M , and Turnbull DM (2016). Mitochondrial diseases. Nat Rev Dis Primers 2 , 16080.27775730 Guo J , Chen X , Liu Z , Sun H , Zhou Y , Dai Y , Ma Y , He L , Qian X , Wang J , (2022). DdCBE mediates efficient and inheritable modifications in mouse mitochondrial genome. Mol Ther Nucleic Acids 27 , 73–80.34938607 Hamilton JR , Tsuchida CA , Nguyen DN , Shy BR , McGarrigle ER , Sandoval Espinoza CR , Carr D , Blaeschke F , Marson A , and Doudna JA (2021). Targeted delivery of CRISPR-Cas9 and transgenes enables complex immune cell engineering. Cell Rep 35 , 109207.34077734 Hoffmann MD , Aschenbrenner S , Grosse S , Rapti K , Domenger C , Fakhiri J , Mastel M , Borner K , Eils R , Grimm D , (2019). Cell-specific CRISPR-Cas9 activation by microRNA-dependent expression of anti-CRISPR proteins. Nucleic Acids Res 47 , e75.30982889 Hoshyar N , Gray S , Han H , and Bao G (2016). The effect of nanoparticle size on in vivo pharmacokinetics and cellular interaction. Nanomedicine (Lond) 11 , 673–692.27003448 Hu Z , Wang S , Zhang C , Gao N , Li M , Wang D , Wang D , Liu D , Liu H , Ong SG , (2020). A compact Cas9 ortholog from Staphylococcus Auricularis (SauriCas9) expands the dNa targeting scope. PLoS Biol 18 , e3000686.32226015 Ibraheim R , Tai PWL , Mir A , Javeed N , Wang J , Rodriguez TC , Namkung S , Nelson S , Khokhar ES , Mintzer E , (2021). Self-inactivating, all-in-one AAV vectors for precision Cas9 genome editing via homology-directed repair in vivo. Nat Commun 12 , 6267.34725353 Indikova I , and Indik S (2020). Highly efficient 'hit-and-run' genome editing with unconcentrated lentivectors carrying Vpr.Prot.Cas9 protein produced from RRE-containing transcripts. Nucleic Acids Res 48 , 8178–8187.32619241 Jang HK , Jo DH , Lee SN , Cho CS , Jeong YK , Jung Y , Yu J , Kim JH , Woo JS , and Bae S (2021). High-purity production and precise editing of DNA base editing ribonucleoproteins. Sci Adv 7 . Jiang C , Mei M , Li B , Zhu X , Zu W , Tian Y , Wang Q , Guo Y , Dong Y , and Tan X (2017). A non-viral CRISPR/Cas9 delivery system for therapeutically targeting HBV DNA and pcsk9 in vivo. Cell Res 27 , 440–443.28117345 Jiang T , Henderson JM , Coote K , Cheng Y , Valley HC , Zhang XO , Wang Q , Rhym LH , Cao Y , Newby GA , (2020). Chemical modifications of adenine base editor mRNA and guide RNA expand its application scope. Nat Commun 11 , 1979.32332735 Jinek M , Chylinski K , Fonfara I , Hauer M , Doudna JA , and Charpentier E (2012). A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science 337 , 816–821.22745249 Jo DH , Jang H-K , Cho CS , Han JH , Ryu G , Jung Y , Bae S , and Kim JH (2021). Therapeutic adenine base editing corrects nonsense mutation and improves visual function in a mouse model of Leber congenital amaurosis. bioRxiv. Joglekar AV , and Sandoval S (2017). Pseudotyped Lentiviral Vectors: One Vector, Many Guises. Hum Gene Ther Methods 28 , 291–301.28870117 Kabadi AM , Ousterout DG , Hilton IB , and Gersbach CA (2014). Multiplex CRISPR/Cas9-based genome engineering from a single lentiviral vector. Nucleic Acids Res 42 , e147.25122746 Kaczmarczyk SJ , Sitaraman K , Young HA , Hughes SH , and Chatterjee DK (2011). Protein delivery using engineered virus-like particles. Proc Natl Acad Sci U S A 108 , 16998–17003.21949376 Kariko K , Muramatsu H , Welsh FA , Ludwig J , Kato H , Akira S , and Weissman D (2008). Incorporation of pseudouridine into mRNA yields superior nonimmunogenic vector with increased translational capacity and biological stability. Mol Ther 16 , 1833–1840.18797453 Kauffman KJ , Dorkin JR , Yang JH , Heartlein MW , DeRosa F , Mir FF , Fenton OS , and Anderson DG (2015). Optimization of Lipid Nanoparticle Formulations for mRNA Delivery in Vivo with Fractional Factorial and Definitive Screening Designs. Nano Lett 15 , 7300–7306.26469188 Kazmierczak Z , Szostak-Paluch K , Przybylo M , Langner M , Witkiewicz W , Jedruchniewicz N , and Dabrowska K (2020). Endocytosis in cellular uptake of drug delivery vectors: Molecular aspects in drug development. Bioorg Med Chem 28 , 115556.32828419 Kedmi R , Veiga N , Ramishetti S , Goldsmith M , Rosenblum D , Dammes N , Hazan-Halevy I , Nahary L , Leviatan-Ben-Arye S , Harlev M , (2018). A modular platform for targeted RNAi therapeutics. Nat Nanotechnol 13 , 214–219.29379205 Kenjo E , Hozumi H , Makita Y , Iwabuchi KA , Fujimoto N , Matsumoto S , Kimura M , Amano Y , Ifuku M , Naoe Y , (2021). Low immunogenicity of LNP allows repeated administrations of CRISPR-Cas9 mRNA into skeletal muscle in mice. Nat Commun 12 , 7101.34880218 Kim E , Koo T , Park SW , Kim D , Kim K , Cho HY , Song DW , Lee KJ , Jung MH , Kim S , (2017). In vivo genome editing with a small Cas9 orthologue derived from Campylobacter jejuni. Nat Commun 8 , 14500.28220790 Kleinstiver BP , Prew MS , Tsai SQ , Nguyen NT , Topkar VV , Zheng Z , and Joung JK (2015). Broadening the targeting range of Staphylococcus aureus CRISPR-Cas9 by modifying Pam recognition. Nat Biotechnol 33 , 1293–1298.26524662 Knopp Y , Geis FK , Heckl D , Horn S , Neumann T , Kuehle J , Meyer J , Fehse B , Baum C , Morgan M , (2018). Transient Retrovirus-Based CRISPR/Cas9 All-in-One Particles for Efficient, Targeted Gene Knockout. Mol Ther Nucleic Acids 13 , 256–274.30317165 Koblan LW , Arbab M , Shen MW , Hussmann JA , Anzalone AV , Doman JL , Newby GA , Yang D , Mok B , Replogle JM , (2021a). Efficient C*G-to-G*C base editors developed using CRISPRi screens, target-library analysis, and machine learning. Nat Biotechnol 39 , 1414–1425.34183861 Koblan LW , Erdos MR , Wilson C , Cabral WA , Levy JM , Xiong ZM , Tavarez UL , Davison LM , Gete YG , Mao X , (2021b). In vivo base editing rescues Hutchinson-Gilford progeria syndrome in mice. Nature 589 , 608–614.33408413 Komor AC , Kim YB , Packer MS , Zuris JA , and Liu DR (2016). Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature 533 , 420–424.27096365 Komor AC , Zhao KT , Packer MS , Gaudelli NM , Waterbury AL , Koblan LW , Kim YB , Badran AH , and Liu DR (2017). Improved base excision repair inhibition and bacteriophage Mu Gam protein yields C:G-to-T:A base editors with higher efficiency and product purity. Sci Adv 3 , eaao4774.28875174 Korf BR , Pyeritz RE , and Grody WW (2019). Nature and Frequency of Genetic Disease. 47–51. Kosicki M , Tomberg K , and Bradley A (2018). Repair of double-strand breaks induced by CRISPR-Cas9 leads to large deletions and complex rearrangements. Nat Biotechnol 36 , 765–771.30010673 Kumar M , Keller B , Makalou N , and Sutton RE (2001). Systematic determination of the packaging limit of lentiviral vectors. Hum Gene Ther 12 , 1893–1905.11589831 Kurt IC , Zhou R , Iyer S , Garcia SP , Miller BR , Langner LM , Grunewald J , and Joung JK (2021). CRISPR C-to-G base editors for inducing targeted DNA transversions in human cells. Nat Biotechnol 39 , 41–46.32690971 Kuzmin DA , Shutova MV , Johnston NR , Smith OP , Fedorin VV , Kukushkin YS , van der Loo JCM , and Johnstone EC (2021). The clinical landscape for AAV gene therapies. Nat Rev Drug Discov 20 , 173–174.33495615 Kwon JB , Ettyreddy AR , Vankara A , Bohning JD , Devlin G , Hauschka SD , Asokan A , and Gersbach CA (2020). In Vivo Gene Editing of Muscle Stem Cells with Adeno-Associated Viral Vectors in a Mouse Model of Duchenne Muscular Dystrophy. Mol Ther Methods Clin Dev 19 , 320–329.33145368 Kymalainen H , Appelt JU , Giordano FA , Davies AF , Ogilvie CM , Ahmed SG , Laufs S , Schmidt M , Bode J , Yanez-Munoz RJ , (2014). Long-term episomal transgene expression from mitotically stable integration-deficient lentiviral vectors. Hum Gene Ther 25 , 428–442.24483952 Lander ES , Baylis F , Zhang F , Charpentier E , Berg P , Bourgain C , Friedrich B , Joung JK , Li J , Liu D , (2019). Adopt a moratorium on heritable genome editing. Nature 567 , 165–168.30867611 Lau CH , and Suh Y (2017). In vivo genome editing in animals using AAV-CRISPR system: applications to translational research of human disease. F1000Res 6 , 2153.29333255 Lee CS , Bishop ES , Zhang R , Yu X , Farina EM , Yan S , Zhao C , Zheng Z , Shu Y , Wu X , (2017). Adenovirus-Mediated Gene Delivery: Potential Applications for Gene and Cell-Based Therapies in the New Era of Personalized Medicine. Genes Dis 4 , 43–63.28944281 Lee H , Lee S , Baek G , Kim A , Kang BC , Seo H , and Kim JS (2021). Mitochondrial DNA editing in mice with DddA-TALE fusion deaminases. Nat Commun 12 , 1190.33608520 Lee J , Mou H , Ibraheim R , Liang SQ , Liu P , Xue W , and Sontheimer EJ (2019). Tissue-restricted genome editing in vivo specified by microRNA-repressible anti-CRISPR proteins. RNA 25 , 1421–1431.31439808 Leibowitz ML , Papathanasiou S , Doerfler PA , Blaine LJ , Sun L , Yao Y , Zhang CZ , Weiss MJ , and Pellman D (2021). Chromothripsis as an on-target consequence of CRISPR-Cas9 genome editing. Nat Genet 53 , 895–905.33846636 Levy JM , Yeh WH , Pendse N , Davis JR , Hennessey E , Butcher R , Koblan LW , Comander J , Liu Q , and Liu DR (2020). Cytosine and adenine base editing of the brain, liver, retina, heart and skeletal muscle of mice via adeno-associated viruses. Nat Biomed Eng 4 , 97–110.31937940 Li A , Lee CM , Hurley AE , Jarrett KE , De Giorgi M , Lu W , Balderrama KS , Doerfler AM , Deshmukh H , Ray A , (2019a). A Self-Deleting AAV-CRISPR System for In Vivo Genome Editing. Mol Ther Methods Clin Dev 12 , 111–122.30619914 Li C , Georgakopoulou A , Mishra A , Gil S , Hawkins RD , Yannaki E , and Lieber A (2021). In vivo HSPC gene therapy with base editors allows for efficient reactivation of fetal gamma-globin in beta-YAC mice. Blood Adv 5 , 1122–1135.33620431 Li F , Hung SSC , Mohd Khalid MKN , Wang JH , Chrysostomou V , Wong VHY , Singh V , Wing K , Tu L , Bender JA , (2019b). Utility of Self-Destructing CRISPR/Cas Constructs for Targeted Gene Editing in the Retina. Hum Gene Ther 30 , 1349–1360.31373227 Li H , Yang Y , Hong W , Huang M , Wu M , and Zhao X (2020a). Applications of genome editing technology in the targeted therapy of human diseases: mechanisms, advances and prospects. Signal Transduct Target Ther 5 , 1.32296011 Li J , Sun W , Wang B , Xiao X , and Liu XQ (2008a). Protein trans-splicing as a means for viral vector-mediated in vivo gene therapy. Hum Gene Ther 19 , 958–964.18788906 Li W , Asokan A , Wu Z , Van Dyke T , DiPrimio N , Johnson JS , Govindaswamy L , Agbandje-McKenna M , Leichtle S , Redmond DE Jr. , (2008b). Engineering and selection of shuffled AAV genomes: a new strategy for producing targeted biological nanoparticles. Mol Ther 16 , 1252–1260. Li X , Qian X , Wang B , Xia Y , Zheng Y , Du L , Xu D , Xing D , DePinho RA , and Lu Z (2020b). Programmable base editing of mutated TERT promoter inhibits brain tumour growth. Nat Cell Biol 22 , 282–288.32066906 Lieber MR (2010). The Mechanism of Double-Strand DNA Break Repair by the Nonhomologous DNA End-Joining Pathway. Annual Review of Biochemistry 79 , 181–211. Lim CKW , Gapinske M , Brooks AK , Woods WS , Powell JE , Zeballos CM , Winter J , Perez-Pinera P , and Gaj T (2020). Treatment of a Mouse Model of ALS by In Vivo Base Editing. Mol Ther 28 , 1177–1189.31991108 Lim K , Cho SI , and Kim JS (2022). Nuclear and mitochondrial DNA editing in human cells with zinc finger deaminases. Nat Commun 13 , 366.35042880 Ling S , Yang S , Hu X , Yin D , Dai Y , Qian X , Wang D , Pan X , Hong J , Sun X , (2021). Lentiviral delivery of co-packaged Cas9 mRNA and a Vegfa-targeting guide RNA prevents wet age-related macular degeneration in mice. Nat Biomed Eng 5 , 144–156.33398131 Liu J , Chang J , Jiang Y , Meng X , Sun T , Mao L , Xu Q , and Wang M (2019). Fast and Efficient CRISPR/Cas9 Genome Editing In Vivo Enabled by Bioreducible Lipid and Messenger RNA Nanoparticles. Adv Mater 31 , e1902575.31215123 Liu P , Liang SQ , Zheng C , Mintzer E , Zhao YG , Ponnienselvan K , Mir A , Sontheimer EJ , Gao G , Flotte TR , (2021a). Improved prime editors enable pathogenic allele correction and cancer modelling in adult mice. Nat Commun 12 , 2121.33837189 Liu Z , Chen S , Jia Y , Shan H , Chen M , Song Y , Lai L , and Li Z (2021b). Efficient and high-fidelity base editor with expanded PAM compatibility for cytidine dinucleotide. Sci China Life Sci 64 , 1355–1367.33420918 Lombardo A , Genovese P , Beausejour CM , Colleoni S , Lee YL , Kim KA , Ando D , Urnov FD , Galli C , Gregory PD , (2007). Gene editing in human stem cells using zinc finger nucleases and integrase-defective lentiviral vector delivery. Nat Biotechnol 25 , 1298–1306.17965707 Lu B , Javidi-Parsijani P , Makani V , Mehraein-Ghomi F , Sarhan WM , Sun D , Yoo KW , Atala ZP , Lyu P , and Atala A (2019). Delivering SaCas9 mRNA by lentivirus-like bionanoparticles for transient expression and efficient genome editing. Nucleic Acids Res 47 , e44.30759231 Lu Z , Yao X , Lyu P , Yadav M , Yoo K , Atala A , and Lu B (2021). Lentiviral Capsid-Mediated Streptococcus pyogenes Cas9 Ribonucleoprotein Delivery for Efficient and Safe Multiplex Genome Editing. CRISPR J 4 , 914–928.33733873 Lyu P , Javidi-Parsijani P , Atala A , and Lu B (2019). Delivering Cas9/sgRNA ribonucleoprotein (RNP) by lentiviral capsid-based bionanoparticles for efficient 'hit-and-run' genome editing. Nucleic Acids Res 47 , e99.31299082 Lyu P , Lu Z , Cho SI , Yadav M , Yoo KW , Atala A , Kim JS , and Lu B (2021). Adenine Base Editor Ribonucleoproteins Delivered by Lentivirus-Like Particles Show High On-Target Base Editing and Undetectable RNA Off-Target Activities. CRISPR J 4 , 69–81.33616436 Lyu P , Wang L , and Lu B (2020). Virus-Like Particle Mediated CRISPR/Cas9 Delivery for Efficient and Safe Genome Editing. Life (Basel) 10 . Maeder ML , Stefanidakis M , Wilson CJ , Baral R , Barrera LA , Bounoutas GS , Bumcrot D , Chao H , Ciulla DM , DaSilva JA , (2019). Development of a gene-editing approach to restore vision loss in Leber congenital amaurosis type 10. Nat Med 25 , 229–233.30664785 Maguire AM , Simonelli F , Pierce EA , Pugh EN Jr. , Mingozzi F , Bennicelli J , Banfi S , Marshall KA , Testa F , Surace EM , (2008). Safety and efficacy of gene transfer for Leber's congenital amaurosis. N Engl J Med 358 , 2240–2248.18441370 Maheshri N , Koerber JT , Kaspar BK , and Schaffer DV (2006). Directed evolution of adeno-associated virus yields enhanced gene delivery vectors. Nat Biotechnol 24 , 198–204.16429148 Maier MA , Jayaraman M , Matsuda S , Liu J , Barros S , Querbes W , Tam YK , Ansell SM , Kumar V , Qin J , (2013). Biodegradable lipids enabling rapidly eliminated lipid nanoparticles for systemic delivery of RNAi therapeutics. Mol Ther 21 , 1570–1578.23799535 Mangeot PE , Risson V , Fusil F , Marnef A , Laurent E , Blin J , Mournetas V , Massourides E , Sohier TJM , Corbin A , (2019). Genome editing in primary cells and in vivo using viral-derived Nanoblades loaded with Cas9-sgRNA ribonucleoproteins. Nat Commun 10 , 45.30604748 Mendell JR , Al-Zaidy S , Shell R , Arnold WD , Rodino-Klapac LR , Prior TW , Lowes L , Alfano L , Berry K , Church K , (2017). Single-Dose Gene-Replacement Therapy for Spinal Muscular Atrophy. N Engl J Med 377 , 1713–1722.29091557 Mendell JR , Al-Zaidy SA , Rodino-Klapac LR , Goodspeed K , Gray SJ , Kay CN , Boye SL , Boye SE , George LA , Salabarria S , (2021). Current Clinical Applications of In Vivo Gene Therapy with AAVs. Mol Ther 29 , 464–488.33309881 Miller JB , Zhang S , Kos P , Xiong H , Zhou K , Perelman SS , Zhu H , and Siegwart DJ (2017). Non-Viral CRISPR/Cas Gene Editing In Vitro and In Vivo Enabled by Synthetic Nanoparticle Co-Delivery of Cas9 mRNA and sgRNA. Angew Chem Int Ed Engl 56 , 1059–1063.27981708 Milone MC , and O'Doherty U (2018). Clinical use of lentiviral vectors. Leukemia 32 , 1529–1541.29654266 Mitchell MJ , Billingsley MM , Haley RM , Wechsler ME , Peppas NA , and Langer R (2021). Engineering precision nanoparticles for drug delivery. Nat Rev Drug Discov 20 , 101–124.33277608 Mok BY , de Moraes MH , Zeng J , Bosch DE , Kotrys AV , Raguram A , Hsu F , Radey MC , Peterson SB , Mootha VK , (2020). A bacterial cytidine deaminase toxin enables CRISPR-free mitochondrial base editing. Nature 583 , 631–637.32641830 Monteys AM , Hundley AA , Ranum PT , Tecedor L , Muehlmatt A , Lim E , Lukashev D , Sivasankaran R , and Davidson BL (2021). Regulated control of gene therapies by drug-induced splicing. Nature 596 , 291–295.34321659 Mullard A (2017). FDA approves first CAR T therapy. Nat Rev Drug Discov 16 , 669. Musunuru K , Chadwick AC , Mizoguchi T , Garcia SP , DeNizio JE , Reiss CW , Wang K , Iyer S , Dutta C , Clendaniel V , (2021). In vivo CRISPR base editing of PCSK9 durably lowers cholesterol in primates. Nature 593 , 429–434.34012082 Naldini L , Blomer U , Gallay P , Ory D , Mulligan R , Gage FH , Verma IM , and Trono D (1996). In vivo gene delivery and stable transduction of nondividing cells by a lentiviral vector. Science 272 , 263–267.8602510 Nambiar TS , Baudrier L , Billon P , and Ciccia A (2022). CRISPR-based genome editing through the lens of DNA repair. Mol Cell 82 , 348–388.35063100 Nance ME , Shi R , Hakim CH , Wasala NB , Yue Y , Pan X , Zhang T , Robinson CA , Duan SX , Yao G , (2019). AAV9 Edits Muscle Stem Cells in Normal and Dystrophic Adult Mice. Mol Ther 27 , 1568–1585.31327755 Naso MF , Tomkowicz B , Perry WL 3rd , and Strohl WR (2017). Adeno-Associated Virus (AAV) as a Vector for Gene Therapy. BioDrugs 31 , 317–334.28669112 Nelson CE , and Gersbach CA (2016). Engineering Delivery Vehicles for Genome Editing. Annu Rev Chem Biomol Eng 7 , 637–662.27146557 Nelson JW , Randolph PB , Shen SP , Everette KA , Chen PJ , Anzalone AV , An M , Newby GA , Chen JC , Hsu A , (2021). Engineered pegRNAs improve prime editing efficiency. Nat Biotechnol. Newby GA , and Liu DR (2021). In vivo somatic cell base editing and prime editing. Mol Ther 29 , 3107–3124.34509669 Newby GA , Yen JS , Woodard KJ , Mayuranathan T , Lazzarotto CR , Li Y , Sheppard-Tillman H , Porter SN , Yao Y , Mayberry K , (2021). Base editing of haematopoietic stem cells rescues sickle cell disease in mice. Nature 595 , 295–302.34079130 Nishida K , Arazoe T , Yachie N , Banno S , Kakimoto M , Tabata M , Mochizuki M , Miyabe A , Araki M , Hara KY , (2016). Targeted nucleotide editing using hybrid prokaryotic and vertebrate adaptive immune systems. Science 353 . Paunovska K , Loughrey D , and Dahlman JE (2022). Drug delivery systems for RNA therapeutics. Nat Rev Genet. Peng Y , Tang L , and Zhou Y (2017). Subretinal Injection: A Review on the Novel Route of Therapeutic Delivery for Vitreoretinal Diseases. Ophthalmic Res 58 , 217–226.28858866 Polack FP , Thomas SJ , Kitchin N , Absalon J , Gurtman A , Lockhart S , Perez JL , Perez Marc G , Moreira ED , Zerbini C , (2020). Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. N Engl J Med 383 , 2603–2615.33301246 Prel A , Caval V , Gayon R , Ravassard P , Duthoit C , Payen E , Maouche-Chretien L , Creneguy A , Nguyen TH , Martin N , (2015). Highly efficient in vitro and in vivo delivery of functional RNAs using new versatile MS2-chimeric retrovirus-like particles. Mol Ther Methods Clin Dev 2 , 15039.26528487 Qiu M , Glass Z , Chen J , Haas M , Jin X , Zhao X , Rui X , Ye Z , Li Y , Zhang F , (2021). Lipid nanoparticle-mediated codelivery of Cas9 mRNA and single-guide RNA achieves liver-specific in vivo genome editing of Angptl3. Proc Natl Acad Sci U S A 118 . Ran FA , Cong L , Yan WX , Scott DA , Gootenberg JS , Kriz AJ , Zetsche B , Shalem O , Wu X , Makarova KS , (2015). In vivo genome editing using Staphylococcus aureus Cas9. Nature 520 , 186–191.25830891 Raper SE , Chirmule N , Lee FS , Wivel NA , Bagg A , Gao GP , Wilson JM , and Batshaw ML (2003). Fatal systemic inflammatory response syndrome in a ornithine transcarbamylase deficient patient following adenoviral gene transfer. Mol Genet Metab 80 , 148–158.14567964 Rees HA , and Liu DR (2018). Base editing: precision chemistry on the genome and transcriptome of living cells. Nat Rev Genet 19 , 770–788.30323312 Rees HA , Minella AC , Burnett CA , Komor AC , and Gaudelli NM (2021). CRISPR-derived genome editing therapies: Progress from bench to bedside. Mol Ther 29 , 3125–3139.34619370 Richter M , Stone D , Miao C , Humbert O , Kiem HP , Papayannopoulou T , and Lieber A (2017). In Vivo Hematopoietic Stem Cell Transduction. Hematol Oncol Clin North Am 31 , 771–785.28895846 Richter MF , Zhao KT , Eton E , Lapinaite A , Newby GA , Thuronyi BW , Wilson C , Koblan LW , Zeng J , Bauer DE , (2020). Phage-assisted evolution of an adenine base editor with improved Cas domain compatibility and activity. Nat Biotechnol 38 , 883–891.32433547 Rothgangl T , Dennis MK , Lin PJC , Oka R , Witzigmann D , Villiger L , Qi W , Hruzova M , Kissling L , Lenggenhager D , (2021). In vivo adenine base editing of PCSK9 in macaques reduces LDL cholesterol levels. Nat Biotechnol 39 , 949–957.34012094 Rurik JG , Tombacz I , Yadegari A , Mendez Fernandez PO , Shewale SV , Li L , Kimura T , Soliman OY , Papp TE , Tam YK , (2022). CAR T cells produced in vivo to treat cardiac injury. Science 375 , 91–96.34990237 Russell S , Bennett J , Wellman JA , Chung DC , Yu ZF , Tillman A , Wittes J , Pappas J , Elci O , McCague S , (2017). Efficacy and safety of voretigene neparvovec (AAV2-hRPE65v2) in patients with RPE65-mediated inherited retinal dystrophy: a randomised, controlled, open-label, phase 3 trial. Lancet 390 , 849–860.28712537 Ryu SM , Koo T , Kim K , Lim K , Baek G , Kim ST , Kim HS , Kim DE , Lee H , Chung E , (2018). Adenine base editing in mouse embryos and an adult mouse model of Duchenne muscular dystrophy. Nat Biotechnol 36 , 536–539.29702637 Sago CD , Lokugamage MP , Paunovska K , Vanover DA , Monaco CM , Shah NN , Gamboa Castro M , Anderson SE , Rudoltz TG , Lando GN , (2018). High-throughput in vivo screen of functional mRNA delivery identifies nanoparticles for endothelial cell gene editing. Proc Natl Acad Sci U S A 115 , E9944–E9952.30275336 Saha K , Sontheimer EJ , Brooks PJ , Dwinell MR , Gersbach CA , Liu DR , Murray SA , Tsai SQ , Wilson RC , Anderson DG , (2021). The NIH Somatic Cell Genome Editing program. Nature 592 , 195–204.33828315 San Filippo J , Sung P , and Klein H (2008). Mechanism of eukaryotic homologous recombination. Annu Rev Biochem 77 , 229–257.18275380 Schoenmaker L , Witzigmann D , Kulkarni JA , Verbeke R , Kersten G , Jiskoot W , and Crommelin DJA (2021). mRNA-lipid nanoparticle COVID-19 vaccines: Structure and stability. Int J Pharm 601 , 120586.33839230 Segel M , Lash B , Song J , Ladha A , Liu CC , Jin X , Mekhedov SL , Macrae RK , Koonin EV , and Zhang F (2021). Mammalian retrovirus-like protein PEG10 packages its own mRNA and can be pseudotyped for mRNA delivery. Science 373 , 882–889.34413232 Seol JH , Shim EY , and Lee SE (2018). Microhomology-mediated end joining: Good, bad and ugly. Mutat Res 809 , 81–87.28754468 Shams A , Higgins SA , Fellmann C , Laughlin TG , Oakes BL , Lew R , Kim S , Lukarska M , Arnold M , Staahl BT , (2021). Comprehensive deletion landscape of CRISPR-Cas9 identifies minimal RNA-guided DNA-binding modules. Nat Commun 12 , 5664.34580310 Shen S , Horowitz ED , Troupes AN , Brown SM , Pulicherla N , Samulski RJ , Agbandje-McKenna M , and Asokan A (2013). Engraftment of a galactose receptor footprint onto adeno-associated viral capsids improves transduction efficiency. J Biol Chem 288 , 28814–28823.23940044 Sheridan C (2018). Go-ahead for first in-body CRISPR medicine testing. Nature Biotechnology. Silva-Pinheiro P , Nash PA , Van Haute L , Mutti CD , Turner K , and Minczuk M (2022). In vivo mitochondrial base editing via adeno-associated viral delivery to mouse post-mitotic tissue. Nat Commun 13 , 750.35136065 Smith SA , Selby LI , Johnston APR , and Such GK (2019). The Endosomal Escape of Nanoparticles: Toward More Efficient Cellular Delivery. Bioconjug Chem 30 , 263–272.30452233 Song Y , Liu Z , Zhang Y , Chen M , Sui T , Lai L , and Li Z (2020). Large-Fragment Deletions Induced by Cas9 Cleavage while Not in the BEs System. Mol Ther Nucleic Acids 21 , 523–526.32711379 Stadtmauer EA , Fraietta JA , Davis MM , Cohen AD , Weber KL , Lancaster E , Mangan PA , Kulikovskaya I , Gupta M , Chen F , (2020). CRISPR-engineered T cells in patients with refractory cancer. Science 367 . Staring J , Raaben M , and Brummelkamp TR (2018). Viral escape from endosomes and host detection at a glance. J Cell Sci 131 . Suh S , Choi EH , Leinonen H , Foik AT , Newby GA , Yeh WH , Dong Z , Kiser PD , Lyon DC , Liu DR , (2021). Restoration of visual function in adult mice with an inherited retinal disease via adenine base editing. Nat Biomed Eng 5 , 169–178.33077938 Sweeney NP , and Vink CA (2021). The impact of lentiviral vector genome size and producer cell genomic to gag-pol mRNA ratios on packaging efficiency and titre. Mol Ther Methods Clin Dev 21 , 574–584.34095341 Tabebordbar M , Lagerborg KA , Stanton A , King EM , Ye S , Tellez L , Krunnfusz A , Tavakoli S , Widrick JJ , Messemer KA , (2021). Directed evolution of a family of AAV capsid variants enabling potent muscle-directed gene delivery across species. Cell 184 , 4919–4938 e4922.34506722 Tabebordbar M , Zhu K , Cheng JKW , Chew WL , Widrick JJ , Yan WX , Maesner C , Wu EY , Xiao R , Ran FA , (2016). In vivo gene editing in dystrophic mouse muscle and muscle stem cells. Science 351 , 407–411.26721686 Taha EA , Lee J , and Hotta A (2022). Delivery of CRISPR-Cas tools for in vivo genome editing therapy: Trends and challenges. J Control Release 342 , 345–361.35026352 Tornabene P , and Trapani I (2020). Can Adeno-Associated Viral Vectors Deliver Effectively Large Genes? Hum Gene Ther 31 , 47–56.31916856 Truong DJ , Kuhner K , Kuhn R , Werfel S , Engelhardt S , Wurst W , and Ortiz O (2015). Development of an intein-mediated split-Cas9 system for gene therapy. Nucleic Acids Res 43 , 6450–6458.26082496 Turchiano G , Andrieux G , Klermund J , Blattner G , Pennucci V , El Gaz M , Monaco G , Poddar S , Mussolino C , Cornu TI , (2021). Quantitative evaluation of chromosomal rearrangements in gene-edited human stem cells by CAST-Seq. Cell Stem Cell 28 , 1136–1147 e1135.33626327 Urnov FD (2018). Genome Editing B.C. (Before CRISPR): Lasting Lessons from the "Old Testament". CRISPR J 1 , 34–46.31021186 van Haasteren J , Li J , Scheideler OJ , Murthy N , and Schaffer DV (2020). The delivery challenge: fulfilling the promise of therapeutic genome editing. Nat Biotechnol 38 , 845–855.32601435 Varkouhi AK , Scholte M , Storm G , and Haisma HJ (2011). Endosomal escape pathways for delivery of biologicals. J Control Release 151 , 220–228.21078351 Vassalli G , Bueler H , Dudler J , von Segesser LK , and Kappenberger L (2003). Adeno-associated virus (AAV) vectors achieve prolonged transgene expression in mouse myocardium and arteries in vivo: a comparative study with adenovirus vectors. Int J Cardiol 90 , 229–238.12957756 Veiga N , Goldsmith M , Granot Y , Rosenblum D , Dammes N , Kedmi R , Ramishetti S , and Peer D (2018). Cell specific delivery of modified mRNA expressing therapeutic proteins to leukocytes. Nat Commun 9 , 4493.30374059 Verdera HC , Kuranda K , and Mingozzi F (2020). AAV Vector Immunogenicity in Humans: A Long Journey to Successful Gene Transfer. Mol Ther 28 , 723–746.31972133 Villiger L , Grisch-Chan HM , Lindsay H , Ringnalda F , Pogliano CB , Allegri G , Fingerhut R , Haberle J , Matos J , Robinson MD , (2018). Treatment of a metabolic liver disease by in vivo genome base editing in adult mice. Nat Med 24 , 1519–1525.30297904 Villiger L , Rothgangl T , Witzigmann D , Oka R , Lin PJC , Qi W , Janjuha S , Berk C , Ringnalda F , Beattie MB , (2021). In vivo cytidine base editing of hepatocytes without detectable off-target mutations in RNA and DNA. Nat Biomed Eng 5 , 179–189.33495639 Voelkel C , Galla M , Maetzig T , Warlich E , Kuehle J , Zychlinski D , Bode J , Cantz T , Schambach A , and Baum C (2010). Protein transduction from retroviral Gag precursors. Proc Natl Acad Sci U S A 107 , 7805–7810.20385817 Wagner DL , Amini L , Wendering DJ , Burkhardt LM , Akyuz L , Reinke P , Volk HD , and Schmueck-Henneresse M (2019). High prevalence of Streptococcus pyogenes Cas9-reactive T cells within the adult human population. Nat Med 25 , 242–248.30374197 Wagner DL , Peter L , and Schmueck-Henneresse M (2021). Cas9-directed immune tolerance in humans-a model to evaluate regulatory T cells in gene therapy? Gene Ther 28 , 549–559.33574580 Wang D , Mou H , Li S , Li Y , Hough S , Tran K , Li J , Yin H , Anderson DG , Sontheimer EJ , (2015). Adenovirus-Mediated Somatic Genome Editing of Pten by CRISPR/Cas9 in Mouse Liver in Spite of Cas9-Specific Immune Responses. Hum Gene Ther 26 , 432–442.26086867 Wang D , Tai PWL , and Gao G (2019). Adeno-associated virus vector as a platform for gene therapy delivery. Nat Rev Drug Discov 18 , 358–378.30710128 Wang D , Zhang F , and Gao G (2020). CRISPR-Based Therapeutic Genome Editing: Strategies and In Vivo Delivery by AaV Vectors. Cell 181 , 136–150.32243786 Wang L , Xue W , Zhang H , Gao R , Qiu H , Wei J , Zhou L , Lei YN , Wu X , Li X , (2021). Eliminating base-editor-induced genome-wide and transcriptome-wide off-target mutations. Nat Cell Biol 23 , 552–563.33972728 Wang W , Bartholomae CC , Arens A , Gabriel R , Glimm H , von Kalle C , and Schmidt M (2010). Genome Wide Characterization of Insertion Profiles of Integration Deficient Lentiviral Vectors Using Two-Sided Non-Restrictive and Standard LAM-PCR. Molecular Therapy 18 . Wang Z , Zhu T , Qiao C , Zhou L , Wang B , Zhang J , Chen C , Li J , and Xiao X (2005). Adeno-associated virus serotype 8 efficiently delivers genes to muscle and heart. Nat Biotechnol 23 , 321–328.15735640 Wanisch K , and Yanez-Munoz RJ (2009). Integration-deficient lentiviral vectors: a slow coming of age. Mol Ther 17 , 1316–1332.19491821 Webber BR , Lonetree CL , Kluesner MG , Johnson MJ , Pomeroy EJ , Diers MD , Lahr WS , Draper GM , Slipek NJ , Smeester BA , (2019). Highly efficient multiplex human T cell engineering without double-strand breaks using Cas9 base editors. Nat Commun 10 , 5222.31745080 Weber T (2021). Anti-AAV Antibodies in AAV Gene Therapy: Current Challenges and Possible Solutions. Front Immunol 12 , 658399.33815421 Wei T , Cheng Q , Farbiak L , Anderson DG , Langer R , and Siegwart DJ (2020a). Delivery of Tissue-Targeted Scalpels: Opportunities and Challenges for In Vivo CRISPR/Cas-Based Genome Editing. ACS Nano 14 , 9243–9262.32697075 Wei T , Cheng Q , Min YL , Olson EN , and Siegwart DJ (2020b). Systemic nanoparticle delivery of CRISPR-Cas9 ribonucleoproteins for effective tissue specific genome editing. Nat Commun 11 , 3232.32591530 Wittrup A , Ai A , Liu X , Hamar P , Trifonova R , Charisse K , Manoharan M , Kirchhausen T , and Lieberman J (2015). Visualizing lipid-formulated siRNA release from endosomes and target gene knockdown. Nat Biotechnol 33 , 870–876.26192320 Witzigmann D , Kulkarni JA , Leung J , Chen S , Cullis PR , and van der Meel R (2020). Lipid nanoparticle technology for therapeutic gene regulation in the liver. Adv Drug Deliv Rev 159 , 344–363.32622021 Wu Z , Asokan A , and Samulski RJ (2006). Adeno-associated virus serotypes: vector toolkit for human gene therapy. Mol Ther 14 , 316–327.16824801 Wu Z , Yang H , and Colosi P (2010). Effect of genome size on AAV vector packaging. Mol Ther 18 , 80–86.19904234 Xiao Y , Muhuri M , Li S , Qin W , Xu G , Luo L , Li J , Letizia AJ , Wang SK , Chan YK , (2019). Circumventing cellular immunity by miR142-mediated regulation sufficiently supports rAAV-delivered OVA expression without activating humoral immunity. JCI Insight 5 . Xu L , Zhang C , Li H , Wang P , Gao Y , Mokadam NA , Ma J , Arnold WD , and Han R (2021). Efficient precise in vivo base editing in adult dystrophic mice. Nat Commun 12 , 3719.34140489 Xu Z , Yue Y , Lai Y , Ye C , Qiu J , Pintel DJ , and Duan D (2004). Trans-splicing adeno-associated viral vector-mediated gene therapy is limited by the accumulation of spliced mRNA but not by dual vector coinfection efficiency. Hum Gene Ther 15 , 896–905.15353044 Yang S , Chang R , Yang H , Zhao T , Hong Y , Kong HE , Sun X , Qin Z , Jin P , Li S , (2017). CRISPR/Cas9-mediated gene editing ameliorates neurotoxicity in mouse model of Huntington's disease. J Clin Invest 127 , 2719–2724.28628038 Yeh WH , Chiang H , Rees HA , Edge ASB , and Liu DR (2018). In vivo base editing of post-mitotic sensory cells. Nat Commun 9 , 2184.29872041 Yeh WH , Shubina-Oleinik O , Levy JM , Pan B , Newby GA , Wornow M , Burt R , Chen JC , Holt JR , and Liu DR (2020). In vivo base editing restores sensory transduction and transiently improves auditory function in a mouse model of recessive deafness. Sci Transl Med 12 . Yin D , Ling S , Wang D , Dai Y , Jiang H , Zhou X , Paludan SR , Hong J , and Cai Y (2021). Targeting herpes simplex virus with CRISPR-Cas9 cures herpetic stromal keratitis in mice. Nat Biotechnol 39 , 567–577.33432198 Yin H , Kauffman KJ , and Anderson DG (2017a). Delivery technologies for genome editing. Nat Rev Drug Discov 16 , 387–399.28337020 Yin H , Song CQ , Dorkin JR , Zhu LJ , Li Y , Wu Q , Park A , Yang J , Suresh S , Bizhanova A , (2016). Therapeutic genome editing by combined viral and non-viral delivery of CRISPR system components in vivo. Nat Biotechnol 34 , 328–333.26829318 Yin H , Song CQ , Suresh S , Wu Q , Walsh S , Rhym LH , Mintzer E , Bolukbasi MF , Zhu LJ , Kauffman K , (2017b). Structure-guided chemical modification of guide RNA enables potent non-viral in vivo genome editing. Nat Biotechnol 35 , 1179–1187.29131148 Zhang H , Bamidele N , Liu P , Ojelabi O , Gao XD , Rodriguez T , Cheng H , Xie J , Gao G , Wolfe SA , (2022). Adenine Base Editing in vivo with a Single Adeno-Associated Virus Vector. bioRxiv. Zhang W , Cao S , Martin JL , Mueller JD , and Mansky LM (2015). Morphology and ultrastructure of retrovirus particles. AIMS Biophys 2 , 343–369.26448965 Zheng C , Liang SQ , Liu B , Liu P , Kwan SY , Wolfe SA , and Xue W (2022). A flexible split prime editor using truncated reverse transcriptase improves dual-AAV delivery in mouse liver. Mol Ther. Zinn E , Pacouret S , Khaychuk V , Turunen HT , Carvalho LS , Andres-Mateos E , Shah S , Shelke R , Maurer AC , Plovie E , (2015). In Silico Reconstruction of the Viral Evolutionary Lineage Yields a Potent Gene Therapy Vector. Cell Rep 12 , 1056–1068.26235624 Zuris JA , Thompson DB , Shu Y , Guilinger JP , Bessen JL , Hu JH , Maeder ML , Joung JK , Chen ZY , and Liu DR (2015). Cationic lipid-mediated delivery of proteins enables efficient protein-based genome editing in vitro and in vivo. Nat Biotechnol 33 , 73–80.25357182

PMC009xxxxxx/PMC9462387.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7503056 4435 J Am Chem Soc J Am Chem Soc Journal of the American Chemical Society 0002-7863 1520-5126 35834763 9462387 10.1021/jacs.2c05385 NIHMS1832668 Article Flipping the switch: reverse-demand voltage-sensitive fluorophores McCann Jack T. ‡ Benlian Brittany R. § Yaeger-Weiss Susanna K. ‡ Knudson Isaac J. ‡ He Minyi ‡ Miller Evan W. ‡§†* ‡ Department of Chemistry, University of California, Berkeley, California 94720, United States. § Department of Molecular & Cell Biology, University of California, Berkeley, California 94720, United States. † Department of Helen Wills Neuroscience Institute, University of California, Berkeley, California 94720, United States. * Corresponding Author evanwmiller@berkeley.edu 2 9 2022 27 7 2022 14 7 2022 27 7 2023 144 29 1305013054 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Fluorescence microscopy with fluorescent reporters that respond to environmental cues are a powerful method for interrogating biochemistry and biophysics in living systems. Photoinduced electron transfer (PeT) is commonly used as a trigger to modulate fluorescence in response to changes in the biological environment. PeT based indicators rely either on PeT into the excited state (acceptor PeT) or out of the excited state (donor PeT). Our group has been developing voltage-sensitive fluorophores (VF dyes) that respond to changes in biological membrane potential (Vm). We hypothesize that the mechanism of voltage sensitivity arises from acceptor PeT (a-PeT) from an electron-rich aniline-containing molecular wire into the excited state fluorophore, resulting in decreased fluorescence at negative Vm. Here we reverse the direction of electron flow to access donor-excited PeT (d-PeT) VF dyes by introducing electron-withdrawing (EWG), rather than electron-rich molecular wires. EWG-containing VF dyes show voltage-sensitive fluorescence, but with the opposite polarity: hyperpolarizing Vm now give fluorescence increases. We use a combination of computation and experiment to design and synthesize five d-PeT VF targets, two of which are voltage sensitive. Graphical Abstract pmcBiological membrane potentials (Vm) arise from differences in ion concentrations across a selectively-permeable lipid bilayer and are a defining feature of life.1 Visualizing cellular Vm with fluorescent indicators offers a powerful complement to traditional electrode methods and seeks to address problems of low-throughput, poor spatial resolution, and invasiveness associated with electrophysiology.2 Fluorescent dyes have long been used to monitor biologically relevant analytes, reactions, and properties. Modulating photoinduced electron transfer (PeT) is a powerful method for designing fluorescent reporters.3-6 By controlling PeT into or out of the excited state of a fluorophore based on the binding or reaction withanalytes like ions7-8 and reactive metabolites,9-11 PeT provides a generalizable trigger for designing fluorescent reporters. Our group has been exploring the application of PeT-based triggers for monitoring Vm.12 We postulate that voltage sensitivity within Voltage-sensitive Fluorophores (or VF dyes) arises from a Vm-sensitive electron transfer (Scheme 1);13-14 therefore, the direction of the electron transfer matters. At hyperpolarized Vm, the electron moves from a molecular wire buried in the plasma membrane into a fluorophore on the extracellular face PeT is occurring and the dye is dim. At depolarized Vm, the voltage decreases the rate of PeT, allowing fluorescence to occur, and the dye brightens. Consistent with this hypothesis, VF dyes possess fluorescence turn-on responses upon membrane depolarization,12 nanosecond response kinetics,15 and voltage-dependent fluorescence lifetimes.16 To date, all VF dyes make use of an aniline-containing molecular wire to achieve voltage sensitivity in an acceptor-PeT (a-PeT) configuration in which the fluorophore acts as the electron acceptor. However, if the hypothesis about the mechanism of voltage sensing is correct, replacing the electron-rich aniline with an electron-withdrawing group (EWG) should decrease the frontier molecular orbital energies of the wire and enable donor-excited PeT (d-PeT) (Scheme 1, Figure 1a-b). In this configuration, hyperpolarized Vm decreases the rate of PeT. This results in fluorescence brightening at hyperpolarized potentials and would provide the first example of a molecular sensor architecture with bi-directional electron flow for sensing in a-PeT or d-PeT configurations. Here we show that electron-poor molecular wires with EWG substituents can be incorporated into a VF dye scaffold, reversing the direction of electron flow, and inverting the sign of the fluorescence response to Vm changes. We calculate the HOMO/LUMO energies of a series of EWG-containing molecular wires, synthesize 5 new EWG-VF dyes, characterize their spectroscopic properties, and evaluate their voltage sensitivity in mammalian cells. Two of the new dyes show voltage sensitivity, but with an inverted polarity relative to previously reported aniline-containing VF dyes.17-19 To investigate the possibility of reversing the polarity of VF dyes through d-PeT, we performed DFT calculations to estimate the relative HOMO/LUMO energies of the orthogonal fluorophore and molecular wire systems. Complete EWG-VFs were modeled in two components, the fluorophore and the molecular wire (Figure 1c, d, S1). Geometries were optimized using def2-TZVP/ωB97XD, and calculated HOMO/LUMO values (Figure 1e) were normalized to a shared sulfonate orbital (Figure S2) to allow direct orbital energy comparison between molecules.20-21 The HOMO/LUMO values calculated using the individual components gave values that matched well with those calculated using the entire VF dye (Figure S1). Component-based computationscut CPU time by nearly two-thirds and enable a mix-and-match comparison of fluorophores and molecular wires. For aniline-substituted VF2.1.Cl,17-19 the molecular wire and fluorophore HOMO is higher than the fluorophore HOMO, with a HOMO-HOMO (H-H) gap of approximately −0.05 eV, indicating the possibility of a-PeT (wire-to-fluorophore). Conversely, the molecular wire LUMO of VF2.1.Cl is higher than the LUMO of 2',7'-dichloro-3-sulfonofluorescein, with a LUMO-LUMO (L-L) gap of approximately +0.27 eV, indicating that d-PeT (fluorophore-to-wire) is unlikely for this molecule (Figure 1). For EWG-VF dyes like 4-NO2-VF, we find the complementary configuration: the wire LUMO of 4-NO2-VF is lower than the fluorescein LUMO by −0.58 eV, indicating the possibility of d-PeT.20 We calculated the orbital energies for other EWG-containing molecular wires, 2,4-diNO2, 3-NO2, 4-CN, and 4-SO2Me (Scheme 2, Figure 1). Both 4-NO2-VF and 2,4-diNO2-VF possess L-L gaps of around −0.6 eV or larger. 3-NO2-VF has an intermediate L-L gap value of −0.33 eV, while the L-L gaps for 4-CN-VF and 4-SO2Me-VF decrease substantially to −0.18 and −0.07 eV, respectively (Figure 1, Table 1). For comparison, the L-L and H-H gaps for non-voltage sensitive VF2.0.Cl (4-H), are +0.12 eV and +0.61 eV, respectively. Based on these data, we hypothesize that strongly EWG substitutions, like 4-NO2 and 2,4-diNO2, might enable d-PeT VF dyes because of their large, negative L-L gaps. To test the hypothesis that VF dyes could be “run in reverse”, we synthesized 5 different EWG-substituted stilbene derivatives. 4-NO2, 3-NO2, and 4-CN stilbenes could all be accessed via sequential Wittig/Heck/Wittig reactions, with the Wittig reactions carried out with K2CO3 as the base (Scheme S1). We synthesized 2,4-diNO2 and 4-SO2Me stilbene (2b, 2e) via a Horner-Wadsworth-Emmons reaction (Scheme S2), since we observed decomposition under Wittig conditions. Pd-catalyzed cross coupling of the EWG-substituted stilbenes (2a-e) with 5-bromo-2',7'-dichloro-3-sulfono-fluorescein (1) afforded EWG-VF dyes (3a-e). All of the EWG-VF dyes display an invariant absorption maximum at 511 nm and emission maximum at 527 nm, owing to the common fluorophore (Figure 2). The molecular wire absorbance varies for EWG-VF dyes (Figure 2, Table 1). The quantum yield of fluorescence (Φfl) for EWG-VF dyes ranges from 0.09 to 0.71 (Table 1). Both 4-NO2-VF and 2,4-diNO2-VF have low Φfl values, indicating a high degree of PeT quenching. The pKa values of the phenolic oxygen of fluorescein of the EWG-VF dyes range from 4.8 to 5.7 (Figure S3, Table S2). In the presence of high concentrations of thiols (2 mM glutathione), EWG-VFs display stability comparable to VF2.1.Cl (p = 0.4, Figure S4). All of the EWG-VF dyes stain the plasma membranes of HEK293T cells (Figure 3a-f). Widefield epifluorescence microscopy reveals a “chicken-wire” pattern of cellular staining, indicating localization to the plasma membrane. 4-NO2-VF has the brightest membrane staining, 2-fold brighter than VF2.1.Cl (Figure 3a, f, g, Table 1). We assessed the voltage sensitivity of the new, EWG-VF dyes in HEK293T cells using whole-cell patch clamp electrophysiology (Figure 4a-e). EWG-VF dyes with 4-nitro substituents show voltage sensitivity, but with a reverse polarity. Unlike VF2.1.Cl, which displays a fluorescence increase upon membrane depolarization, the fluorescence of 4-NO2-VF and 2,4-diNO2-VF decreases upon depolarization and becomes brighter upon hyperpolarization (Figure 4f). Despite a higher nominal voltage sensitivity, 2,4-diNO2-VF shows very low cellular fluorescence (Figure 3a, b, S5), ~40-folder lower than 4-NO2-VF, making 4-NO2-VF the most useful EWG-VF for cellular voltage imaging. 4-NO2-VF is capable of monitoring evoked action potentials (APs) in cultured rat hippocampal neurons (Figure S6). In a single trial, 4-NO2-VF reports on APs with an average ΔF/F of −1.3% (±0.14, n = 17, standard deviation) and a signal-to-noise ratio (SNR) of 9.1 (±1.0, n = 17). The response to neuronal AP depolarization is a fluorescence decrease, showing that d-PeT indicators function in complex cellular contexts. At hyperpolarizing potentials more negative than −100 mV, 4-NO2-VF becomes even brighter, achieving a turn on response for hyperpolarization of +2.4% (Figure 5a-b). At extreme hyperpolarized potentials, the optical response deviates from the linearity observed between −100 mV and +100 mV (Figure 5b). Hyperpolarized potentials play important roles in inhibitory neurotransmission and more broadly, in the physiology of mitochondria, where resting mitochondrial potentials are in the range of −100 to −200 mV.22-23 Finally, we show that 4-NO2-VF can be used in simultaneous, two-color mapping of Vm dynamics along with a far-red voltage indicator, BeRST-1 (Figure 5c-e, S7).24 4-NO2-VF exactly follows the time course of BeRST-1, indicating that the d-PeT method provides fast response kinetics and is compatible with two color imaging (Figure 5f). In summary, we report the design, synthesis, and validation of d-PeT based VFs for voltage imaging. DFT was used to determine the orbital energies of molecular wires with EWGs and 2',7'-dichloro-3-sulfonofluorescein. Five d-PeT VFs were synthesized to test the computational method. Both 4-NO2-VF and 2,4-diNO2-VF are voltage sensitive, acting as turn-on indicators for hyperpolarization. While a-PeT and d-PeT have been used to design fluorescent reporters,5 this is the first demonstration that HOMO/LUMO levels can be tuned to access both a-PeT and d-PeT for detection of a biologically-relevant analyte with the same fluorophore scaffold. Future directions include improving voltage sensitivity by pairing electron-deficient molecular wires with electron-rich fluorophores and applying reverse VFs to biological contexts where the Vm polarity is switched compared to plasma membranes, for example in mitochondria25 and organelles.26 Supplementary Material SI ACKNOWLEDGMENT E.W.M. acknowledges support from the Camille Dreyfus Foundation and NIH (R35GM119855). J.T.M., B.R.B., and I.J.K. were supported, in part, by a training grant from NIH (T32GM066698). S.K.Y.-W. was supported, in part, by a training grant from NIH (T32GM008295). We thank Dr. Hasan Celik and the staff of the College of Chemistry NMR facility for their assistance. Instruments in CoC-NMR are supported in part by NIH S10OD024998. 900MHz NMR supported by NIH GM68933. We thank Drs. Kathleen Durkin and Dave Small for their assistance with computations. The CoC-MGCF is supported by NIH S10OD023532. Figure 1. Controlling photoinduced electron transfer processes in voltage-sensitive fluorophores. Frontier molecular orbital diagrams for fluorescein and either a) electron-rich molecular wires that exhibit acceptor-excited PeT (a-PeT) or b) electron-poor molecular wires that exhibit donor-excited PeT (d-PeT). c) Structure and LUMO of 2',7'-dichloro-3-sulfonofluorescein. d) Structure and LUMO of variousmolecular wires. This specific example is 4-NO2-VF, but R can equal any of the substituents indicated in panel (e). e) Plot of calculated energy levels (eV) of 2',7'-dichloro-3-sulfonofluorescein and various molecular wires with the indicated R group. Figure 2. UV-vis and emission spectra of reverse VF dyes. Plot of relative intensity vs wavelength for reverse VF dyes. Absorbance is shown in solid lines; emission is shown in dashed lines. Spectra are normalized to the λmax and acquired with 1.25 μM dye in 0.1M KOH in EtOH. Excitation provided at 480 nm. Figure 3. Live cell imaging with reverse VF dyes. Widefield epifluorescence images of HEK293T cells treated with either a) 4-NO2-VF (6), b) 2,4-diNO2-VF (13), c) 3-NO2-VF (17), d) 4-SO2Me-VF (23), e) 4-CN-VF (27), or f) VF2.1.Cl (4-NMe2). All dyes were loaded at 250 nM. Scale bar is 20 μm. g) Plot of cellular fluorescence intensity of HEK293T cells loaded with the indicated dye. Data are mean ± S.E.M. for n = 6 independent experiments. For each experiment, or coverslip, we analyzed between 80 – 100 cells and took the mean fluorescence intensity. Figure 4. Voltage sensitivity of EWG-containing VF dyes in living cells. Plots of ΔF/F vs time for HEK293T cells loaded with either a) 4-NO2-VF, b) 2,4-diNO2-VF, c) 3-NO2-VF, d) 4-CN-VF, or e) 4-SO2Me-VF. Cells were held at −60 mV under whole-cell voltage-clamp conditions and then stepped to potentials ranging from −100 mV to −100 mV in 20 mV increments. f) Plot of ΔF/F per 100 mV vs Vm (in mV) for 4-NO2-VF (green, n= 8), 2,4-diNO2-VF (magenta, n = 5), 3-NO2-VF (red, n = 5), 4-CN-VF (blue, n = 3), 4-SO2Me-VF (gray, n = 4), or VF2.1.Cl (black). Data are mean ± S.E.M. VF2.1.Cl data is from Turnbull, et al.21 Figure 5. Voltage sensitivity of 4-NO2-VF at hyperpolarized potentials. a) Plot of ΔF/F vs time for HEK293T cells loaded with 4-NO2-VF. Cells were held at −60 mV under whole-cell voltage-clamp conditions and stepped to potentials ranging from +100 mV to −200 mV in 20 mV increments. b) Plot of ΔF/F vs Vm (in mV) for 4-NO2-VF. Data are mean ± SEM for n = 5 cells. Fluorescence images of a HEK cell stained with c) BeRST-1 and d) 4-NO2-VF. e) Plot of ΔF/F vs time for the same cell imaged with both BeRST and 4-NO2-VF. f) Zoomed-in region of e. Scheme 1. PeT in Voltage-sensitive Fluorophores (VF) dyes Scheme 2. Synthesis of EWG-containing VF dyes Table 1. Properties of EWG-VFs Compound R ΔF/Fa LUMO (eV)b L-L Gap (eV)c λabsd (nm) Rel. Cell Brightnesse Φflf 3a 4-NO2 −3.8 ± 0.1 7.69 −0.58 387 2.0 ± 0.1 0.23 3b 2,4-diNO2 −8.2 ± 0.7 7.45 −0.79 397 0.13 ± 0.02 0.10 3c 3-NO2 −0.4 ± 0.01 7.94 −0.33 342 0.76 ± 0.2 0.71 3d 4-CN −0.4 ± 0.1 8.09 −0.18 366 0.87 ± 0.1 0.71 3e 4-SO2Me 0 ± 0.7 8.21 −0.07 360 0.095 ± 0.01 0.51 VF2.1.Cl 4-NMe2 25 ± 1 8.54 +0.27 389g 1.0 ± 0.04 0.12g VF2.0.Cl H 0 8.40 +0.12 360g --- 0.83g a per 100 mV in HEK293T cells. b Values normalized to respective sulfonate orbital. c Difference between LUMO of 2',7'-dichloro-3-sulfonofluorescein and corresponding molecular wire. d For molecular wire; acquired in 0.1 M EtOH-KOH. e Determined in HEK293T cells; all values relative to brightness of VF2.1.Cl. f Determined in EtOH-KOH. g Values from Boggess, et al.17 Supporting Information. Experimental procedures, NMR spectra, and supporting figures. This material is available free of charge via the Internet at http://pubs.acs.org. REFERENCES 1. Kandel ER ; Schwartz JH ; Jessell TM ; Siegelbaum SA ; Hudspeth AJ ; Mack S , Principles of neural science. 5th ed. ed.; McGraw-Hill Education LLC: New York, N.Y, 2013. 2. Peterka DS ; Takahashi H ; Yuste R , Imaging voltage in neurons. Neuron 2011, 69 (1 ), 9–21.21220095 3. de Silva PA ; Gunaratne NHQ ; McCoy CP , A molecular photoionic AND gate based on fluorescent signalling. Nature 1993, 364 (6432 ), 42–44. 4. Daly B ; Ling J ; de Silva AP , Current developments in fluorescent PET (photoinduced electron transfer) sensors and switches. Chemical Society Reviews 2015, 44 (13 ), 4203–4211.25695939 5. Nagano T , Development of fluorescent probes for bioimaging applications. Proc Jpn Acad Ser B Phys Biol Sci 2010, 86 (8 ), 837–847. 6. Lavis LD , Teaching Old Dyes New Tricks: Biological Probes Built from Fluoresceins and Rhodamines. Annual Review of Biochemistry 2017, 86 (1 ), 825–843. 7. Minta A ; Kao JPY ; Tsien RY , Fluorescent indicators for cytosolic calcium based on rhodamine and fluorescein chromophores. Journal of Biological Chemistry 1989, 264 (14 ), 8171–8178.2498308 8. Burdette SC ; Walkup GK ; Spingler B ; Tsien RY ; Lippard SJ , Fluorescent Sensors for Zn2+ Based on a Fluorescein Platform: Synthesis, Properties and Intracellular Distribution. Journal of the American Chemical Society 2001, 123 (32 ), 7831–7841.11493056 9. Kojima H ; Nakatsubo N ; Kikuchi K ; Kawahara S ; Kirino Y ; Nagoshi H ; Hirata Y ; Nagano T , Detection and Imaging of Nitric Oxide with Novel Fluorescent Indicators: Diaminofluoresceins. Analytical Chemistry 1998, 70 (13 ), 2446–2453.9666719 10. Tanaka K ; Miura T ; Umezawa N ; Urano Y ; Kikuchi K ; Higuchi T ; Nagano T , Rational Design of Fluorescein-Based Fluorescence Probes. Mechanism-Based Design of a Maximum Fluorescence Probe for Singlet Oxygen. Journal of the American Chemical Society 2001, 123 (11 ), 2530–2536.11456921 11. Ueno T ; Urano Y ; Setsukinai K.-i. ; Takakusa H ; Kojima H ; Kikuchi K ; Ohkubo K ; Fukuzumi S ; Nagano T , Rational Principles for Modulating Fluorescence Properties of Fluorescein. Journal of the American Chemical Society 2004, 126 (43 ), 14079–14085.15506772 12. Liu P ; Miller EW , Electrophysiology, Unplugged: Imaging Membrane Potential with Fluorescent Indicators. Accounts of Chemical Research 2020, 53 (1 ), 11–19.31834772 13. de Silva AP ; Gunaratne HQN ; Habib-Jiwan J-L ; McCoy CP ; Rice TE ; Soumillion J-P , New Fluorescent Model Compounds for the Study of Photoinduced Electron Transfer: The Influence of a Molecular Electric Field in the Excited State. Angewandte Chemie International Edition in English 1995, 34 (16 ), 1728–1731. 14. Li L.-s. , Fluorescence Probes for Membrane Potentials Based on Mesoscopic Electron Transfer. Nano Letters 2007, 7 (10 ), 2981–2986.17880257 15. Beier HT ; Roth CC ; Bixler JN ; Sedelnikova AV ; Ibey BL , Visualization of Dynamic Sub-microsecond Changes in Membrane Potential. Biophysical Journal 2019, 116 (1 ), 120–126.30579565 16. Lazzari-Dean JR ; Gest AMM ; Miller EW , Optical estimation of absolute membrane potential using fluorescence lifetime imaging. eLife 2019, 8 , e44522.31545164 17. Boggess SC ; Lazzari-Dean JR ; Raliski BK ; Mun DM ; Li AY ; Turnbull JL ; Miller EW , Fluorescence lifetime predicts performance of voltage sensitive fluorophores in cardiomyocytes and neurons. RSC Chemical Biology 2021, 2 (1 ), 248–258.34212146 18. Woodford CR ; Frady EP ; Smith RS ; Morey B ; Canzi G ; Palida SF ; Araneda RC ; Kristan WB ; Kubiak CP ; Miller EW ; Tsien RY , Improved PeT Molecules for Optically Sensing Voltage in Neurons. Journal of the American Chemical Society 2015, 137 (5 ), 1817–1824.25584688 19. Miller EW ; Lin JY ; Frady EP ; Steinbach PA ; Kristan WB ; Tsien RY , Optically monitoring voltage in neurons by photoinduced electron transfer through molecular wires. Proceedings of the National Academy of Sciences 2012, 109 (6 ), 2114–2119. 20. Chi W ; Chen J ; Liu W ; Wang C ; Qi Q ; Qiao Q ; Tan TM ; Xiong K ; Liu X ; Kang K ; Chang Y-T ; Xu Z ; Liu X , A General Descriptor ΔE Enables the Quantitative Development of Luminescent Materials Based on Photoinduced Electron Transfer. Journal of the American Chemical Society 2020, 142 (14 ), 6777–6785.32182060 21. Turnbull JL ; Benlian BR ; Golden RP ; Miller EW , Phosphonofluoresceins: Synthesis, Spectroscopy, and Applications. Journal of the American Chemical Society 2021, 143 (16 ), 6194–6201.33797899 22. Alpert NM ; Guehl N ; Ptaszek L ; Pelletier-Galarneau M ; Ruskin J ; Mansour MC ; Wooten D ; Ma C ;Takahashi K ; Zhou Y ; Shoup TM ; Normandin MD ; El Fakhri G , Quantitative in vivo mapping of myocardial mitochondrial membrane potential. PLOS ONE 2018, 13 (1 ), e0190968.29338024 23. Gerencser AA ; Chinopoulos C ; Birket MJ ; Jastroch M ; Vitelli C ; Nicholls DG ; Brand MD , Quantitative measurement of mitochondrial membrane potential in cultured cells: calcium-induced de- and hyperpolarization of neuronal mitochondria. The Journal of physiology 2012, 590 (12 ), 2845–71.22495585 24. Huang Y-L ; Walker AS ; Miller EW , A Photostable Silicon Rhodamine Platform for Optical Voltage Sensing. Journal of the American Chemical Society 2015, 137 (33 ), 10767–10776.26237573 25. Klier PEZ ; Martin JG ; Miller EW , Imaging Reversible Mitochondrial Membrane Potential Dynamics with a Masked Rhodamine Voltage Reporter. J Am Chem Soc 2021, 143 (11 ), 4095–4099.33710896 26. Saminathan A ; Devany J ; Veetil AT ; Suresh B ; Pillai KS ; Schwake M ; Krishnan Y , A DNA-based voltmeter for organelles. Nature Nanotechnology 2021, 16 (1 ), 96–103.

PMC009xxxxxx/PMC9469798.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0413066 2830 Cell Cell Cell 0092-8674 1097-4172 35839759 9469798 10.1016/j.cell.2022.06.029 NIHMS1833280 Article Metabolic analysis as a driver for discovery, diagnosis and therapy DeBerardinis Ralph J. 1 Keshari Kayvan R. 2 1 Howard Hughes Medical Institute and Children’s Medical Center Research Institute, University of Texas Southwestern Medical Center 2 Memorial Sloan Kettering Cancer Center Correspondence: Ralph J. DeBerardinis: ralph.deberardinis@utsouthwestern.edu; Kayvan R. Keshari: rahimikk@mskcc.org Lead contact: Ralph J. DeBerardinis: ralph.deberardinis@utsouthwestern.edu AUTHOR CONTRIBUTIONS R.J.D. and K.R.K. decided together on the topics most important to cover. They also wrote the paper and designed the figures. 2 9 2022 21 7 2022 14 7 2022 21 7 2023 185 15 26782689 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY: Metabolic anomalies contribute to tissue dysfunction. Current metabolism research spans from organelles to populations, and new technologies can accommodate investigation across these scales. Here we review recent advancements in metabolic analysis, including small-scale metabolomics techniques amenable to organelles and rare cell types, functional screening to explore how cells to respond to metabolic stress, and imaging approaches to assess metabolic perturbations non-invasively in disease. We discuss how metabolomics provides an informative phenotypic dimension that complements genomic analysis in Mendelian and non-Mendelian disorders. We also outline pressing challenges and how addressing them may further clarify the biochemical basis of human disease. Metabolism metabolomics stable isotopes genomics molecular imaging magnetic resonance positron emission tomography pmcINTRODUCTION Many phenotypes in humans and other organisms involve reprogramming of metabolism. Disease-associated metabolic perturbations can be fixed (e.g. defined by germline mutations) or reversible (e.g. nutritional deficiencies, transient tissue hypoxia), and may involve simple defects confined to particular cell types or complex alterations of systemic homeostasis. Most of the common causes of death in the developed world – heart disease, stroke, diabetes, cancer, and others – are characterized by metabolic changes that contribute to tissue dysfunction. The importance of metabolism in basic cellular processes and the large fraction of the genome devoted to the metabolic network explain why altered metabolism is so prominent in disease. These connections make it appealing to identify metabolic features related to phenotypic variation. Recent advances have made it possible to characterize disease-associated metabolic alterations in detail. This has involved applying emerging technologies to metabolism research and repurposing established techniques such as stable isotope tracing to probe disease-associated metabolic perturbations in vivo. These efforts have produced new therapeutic targets and insights about the mechanistic basis of metabolic diseases. A case in point involves human cancers with mutations in isocitrate dehydrogenase-1 or -2 (IDH1, IDH2). Cancer-associated mutations in IDH1 and IDH2 occur in gliomas and other malignancies (Mardis et al., 2009; Parsons et al., 2008). Metabolomic profiling revealed that mutant IDH1 and IDH2 convert α-ketoglutarate to D-2-hydroxyglutartate (D-2HG), resulting in massive accumulation of this usually scarce metabolite (Dang et al., 2009). D-2HG interferes with histone and DNA demethylases that use α-ketoglutarate as a cofactor (Figueroa et al., 2010), and failure to activate gene expression programs required for differentiation is thought to promote malignancy in IDH-mutant cells (Rohle et al., 2013; Wang et al., 2013). It is now possible to follow D-2HG levels clinically to track disease progression (Choi et al., 2016; Intlekofer et al., 2018), and drugs that inhibit the mutant enzymes are used to treat leukemia (DiNardo et al., 2018). Therefore, discovery of a disease-associated metabolic alteration produced insights into pathophysiology and changed clinical care. This review aims to provide biologists with a survey of recent methodological advances in metabolism, emphasizing four with particular relevance to human disease and potential to enable many new discoveries in the coming decade. We also discuss challenges that need to addressed to fuel the next wave of breakthroughs. ADVANCE 1 - SMALL-SCALE METABOLOMICS TO ASSESS METABOLITE LEVELS IN DISCRETE, BIOLOGICALLY IMPORTANT CELLULAR SUBSETS Metabolomics on bulk tissues has provided a wealth of knowledge about metabolic perturbation in disease. But it is clear from single cell RNA studies that an appreciation for the metabolism of individual cells within a complex microenvironment is critical to understanding pathogenesis (Xiao et al., 2019). This is important for four reasons. First, different cell types within the same tissue have specialized metabolic properties. Metabolic compartmentation between neurons and astrocytes in the brain is an example of this concept. Second, the location of cells within a microenvironment influences their metabolism. Examples include compartmentalized metabolic fluxes within the hepatic lobule, and the impact of proximity to the blood supply on cancer cell metabolism within tumors. Third, some scarce cells, e.g. stem and progenitor cells, are thought to be metabolically distinct from the rest of the tissue, but it is nearly impossible to infer their metabolic activities from bulk analysis. Fourth, even within a single cell, metabolism is compartmentalized within organelles and other structures to achieve precise regulation. Recent advances have addressed the challenge of assessing metabolism at small scale. Single Cells and Rare Cell Populations Given that traditional metabolomics approaches require 105-106 cells, small-scale metabolomics has been driven by the need to characterize rare cell populations. Rare cells can be analyzed through rapid enrichment by flow cytometry at cold temperatures, sorting directly into extraction medium followed by mass spectrometry, yielding up to 160 metabolites in hematopoietic stem cells and circulating tumor cells (DeVilbiss et al., 2021). This approach revealed the surprising stability of many metabolites during the 12-minute preparation, and demonstrated that purine monophosphates become depleted when melanoma cells enter the circulation, possibly reflecting adaptation to stresses encountered after escape from the tumor. Metabolomics can be coupled to sensitive metabolic flux measurements, including isotope tracing with hyperpolarized nuclear magnetic resonance (NMR). This provides insight into quantitative rates of metabolite turnover. Recent work has shown that this can provide flux measurements in as few as 9,000 cells and be applied to the study of flow-sorted cancer stem cells when coupled to a microcoil system (Jeong et al., 2017). Beyond direct metabolite measurements, other profiling approaches can provide information about the metabolic network, sometimes at single-cell resolution. Single cell RNA sequencing can determine the levels of metabolically relevant genes within complex tissues (Arguello et al., 2020; Artyomov and Van den Bossche, 2020). Proteomic methods such as Cytometry/time of flight mass spectrometry (CyTOF) provide a richer appreciation for metabolic heterogeneity within immune cell populations. Other work has exploited optical approaches using FRET sensors to quantify metabolites like amino acids and redox cofactors (Cameron et al., 2016; Tao et al., 2017; Wu et al., 2021). Together with live-cell imaging, this provides real-time information about how metabolic features relate to cell biology. These approaches require cellular engineering and are generally incompatible with high-throughput screening. Traditional optical imaging (e.g. two-photon microscopy) for endogenous metabolites provides a “probeless” strategy and can be used to interrogate redox cofactors such as NAD(H) and FADH(H) (Blacker et al., 2014; Walsh et al., 2013). While these approaches require further development, coupling them with rapidly developing spatial mass spectrometry imaging (MSI) methods could aid in the assessment of real time flux in small-scale applications. Spatially Resolved Metabolism Isolation provides a means of enriching cells of interest, but eliminates the ability to spatially map metabolites within a tissue. Multiple approaches have been pioneered in the last 10 years to address this need. Leveraging the increased sensitivity of modern mass spectrometry systems, many imaging approaches have been placed in tandem to raster scan and generate spatial metabolomic data. Matrix assisted laser-desorption ionization (MALDI) has become the most widely utilized spatial mass spectrometry approach and is typically conducted by sectioning a sample to 5-10μm thickness and placing the section on a moving stage where a mass spectrum is recorded in a rasterized manner to reconstruct a 3D dataset (2D spatial and 1D spectral)(Chaurand et al., 2006). The spatial resolution (50-100μm) continues to be limiting for true single cell analysis. Interesting methods have been developed for cells in culture with application to tissue sections that attempt to infer single cell measurement (e.g. SpaceM (Rappez et al., 2021)), and this will be a key area for further development. Spatially resolved MSI using MALDI was instrumental in assessing novel metabolism in both the setting of normal brain function and disease. After the surprising finding that glucosamine constitutes a significant fraction of the monosaccharides composing glycogen in the mouse brain, MALDI MSI revealed that defects in glycogen breakdown resulted in co-localized glycogen accumulation and depletion of N-linked glycans (Sun et al., 2021). In genetic mouse models of squamous cell carcinoma, MALDI MSI provided evidence for a subset of CD34+ cells containing increased glutathione abundance, conferring oxidative stress resistance and tumorigenic potential in vivo (Choi et al., 2021). MALDI MSI has pushed into isotope tracing with recent studies annotating metabolite labeling in regions of the mouse brain (Wang et al., 2022). It is still challenging to infer true metabolic flux given the static nature of these experiments, but a wealth of comparisons can now be made in the same region. Metabolic flux is driven by substrate and cofactor concentrations which can vary dramatically across compartments of the cell. While the holy grail of spatial MSI approaches would be submicron resolution to assess distinct compartments within the cell, this is not currently possible. But intermediate approaches have been developed and proven to be informative. Building on approaches to “pull down” cells of interest, multiple groups have taken advantage of immunoprecipitation to enrich organelles for metabolic analysis. MitoIP quantifies the abundance of metabolites within the mitochondria (Chen et al., 2016). Artifacts associated with conventional, time-consuming mitochondrial enrichments using centrifugation and other techniques are minimized by the rapid (under 15 minutes) immunopurification approach used by Mito-IP (Chen et al., 2017). Mitochondrial metabolomics has aided in the discovery of mitochondrial transporters for glutathione (Wang et al., 2021) and NAD (Kory et al., 2020). The technique has been extended to lysosomes (Abu-Remaileh et al., 2017) and peroxisomes (Ray et al., 2020). These approaches require large numbers of cells (e.g. 107 cells), though it is likely that this can be improved with some of the high-sensitivity mass spectrometry approaches used in rare cell applications. Altogether, the methods discussed in this section have improved our understanding of metabolism by making it possible to explore metabolic features of discrete cellular populations and subcellular compartments. The conceptual advance emerging from these studies is that despite the undeniable utility of performing metabolic analyses on bulk tissues, some crucial aspects of metabolism are only observable through advanced techniques that allow the investigator to focus on metabolically-distinct regions within complex environments. ADVANCE 2: OPEN-ENDED TECHNIQUES TO IDENTIFY NEW MECHANISMS OF METABOLIC REGULATION. The reactions that comprise the metabolic network have been known for decades. But questions persist about how context-dependent metabolic requirements are established. Screening methods have become increasingly common in metabolism research, and the open-ended nature of these approaches has produced many non-intuitive discoveries. This section discusses two screening techniques: functional genomics using libraries optimized to probe metabolism, and chemical library screens tailored to identify metabolic vulnerabilities. Functional Genomics The large number of “metabolic” genes – over 2,000 in humans (Gillespie et al., 2022) – means that a thorough analysis demands high-throughput methods. Techniques in mammalian cell functional genomics allow investigators to identify metabolic vulnerabilities in an unbiased fashion without pre-existing hypotheses about which pathways are most important. An early example used a lentiviral short-hairpin RNA library tailored to target metabolic enzymes and transporters (Possemato et al., 2011). The library was expressed in breast cancer cells, then the cells were implanted orthotopically in mice. After tumor formation, massively parallel sequencing revealed that the gene encoding phosphoglycerate dehydrogenase (PHGDH) is required for tumor growth in this model. The genomic region containing PHGDH is amplified in over half of estrogen-receptor negative breast cancers, and some other cancers as well (Locasale et al., 2011; Possemato et al., 2011). PHGDH catalyzes a reaction in de novo synthesis of serine and glycine, intermediates that contribute to cancer cell growth. Many subsequent studies have explored the importance of PHGDH in cancer, including the recent finding that a paucity of serine in the brain microenvironment increases dependence on PHGDH and renders brain metastases sensitive to PHGDH inhibition (Ngo et al., 2020). CRISPR-based screens (Figure 1) have been used to identify many metabolic liabilities, sometimes solving long-standing conundrums in cell biology. Although oxidative phosphorylation (OxPhos) by the electron transport chain (ETC) supports many metabolic functions, cells with defective OxPhos can proliferate if provided with exogenous pyruvate (King and Attardi, 1989). Why pyruvate is so important under these circumstances had been puzzling. A CRISPR screen in Jurkat cells containing guide RNAs against ~3,000 metabolic genes revealed that GOT1, the cytosolic glutamate-oxaloacetate transaminase that produces aspartate from oxaloacetate, becomes essential during OxPhos inhibition (Birsoy et al., 2015). Aspartate levels decline in the absence of pyruvate, but exogenous aspartate compensates either for pyruvate depletion or GOT1 deficiency. This led to the surprising conclusion that a major function of the ETC in proliferating cells is to produce aspartate rather than energy. Pyruvate serves as an electron acceptor that ultimately enables production of oxaloacetate for transamination (Birsoy et al., 2015; Sullivan et al., 2015). Subsequent studies revealed that hypoxic regions of tumors contain insufficient levels of aspartate to support cell growth (Garcia-Bermudez et al., 2018). Similar screens have identified enzymes that produce lethal toxin-induced oxidative stress and new mechanisms of redox homeostasis (Garcia-Bermudez et al., 2019; Reczek et al., 2017). All these discoveries relied on the unbiased, global assessment of metabolic dependencies provided by functional genomics. The tumor microenvironment (TME) influences metabolism, so many enzymes required for cell growth in vivo are dispensable in culture, and vice versa (Davidson et al., 2016). Two recent studies compared liabilities from parallel screens in pancreatic ductal adenocarcinoma (PDAC) cells conducted in culture and mice (Biancur et al., 2021; Zhu et al., 2021). After introducing the library, the cells were either allowed to proliferate in culture or implanted subcutaneously to form tumors in immunocompetent mice. Surprisingly, many liabilities during in vivo growth were also observed in culture. But other pathways were selectively required for growth in vivo. Among these, heme biosynthesis was detected by both teams, indicating that heme availability limits PDAC cell growth in vivo but not in culture. Modulating Metabolism with Small Molecules A few hundred inhibitors against metabolic enzymes and nutrient transporters exist as tool compounds or, in some cases, drugs in the clinical arena. Chemical libraries enriched for these inhibitors have been used to identify synthetically lethal interactions, uncovering unexpected roles for metabolism in the cellular responses to blockade of other processes. For example, this chemical screening approach revealed that glutathione depletion synergizes with deubiquitinase inhibition in breast cancer, perhaps reflecting the importance of deubiquitination in mitigating the accumulation of misfolded proteins during oxidative stress (Harris et al., 2019). Inhibition of NAMPT, an enzyme in NAD+ salvage, sensitized triple-negative breast cancer cells to apoptosis by BH3 mimetics (Daniels et al., 2021). The NAD+ depletion caused by NAMPT inhibition reduces adenine levels and primes cells for apoptosis. Metabolites bind allosteric sites on enzymes, and this is a common mechanism to regulate pathway activity. Although many such interactions are already known, an unbiased approach to search for metabolite-protein interactions could uncover novel modes of regulation. One recent approach used equilibrium dialysis and mass spectrometry to detect metabolites from a library that can bind to purified proteins (Hicks K.G., 2021). Screening 33 proteins uncovered hundreds of interactions and several new regulatory sites resolved through atomic-resolution co-crystal structures containing the metabolite and protein. In one such interaction, long-chain acyl-CoAs were found to bind lactate dehydrogenase A (LDHA) and inhibit its activity at low-micromolar concentrations. The findings imply a new mode of regulation between lactate and fatty acid metabolism, exerted at the level of LDHA. Many other protein-metabolite interactions uncovered by this screening assay remain to be explored. The primary conceptual advance of these genetic and chemical screening methods over the last decade is that they enable detailed, efficient queries of the metabolic network without a priori hypotheses about interactions and liabilities. In unbiased screens like these, the benchmark of success is that they produce discoveries that would not have arisen purely from existing knowledge. The discoveries cited above demonstrate ways that unbiased screening has enriched our understanding of metabolism. ADVANCE 3: COMBINING GENOMICS AND METABOLOMICS TO UNDERSTAND THE GENETIC BASIS OF METABOLIC VARIABILITY AND DISEASE Outbred populations contain a remarkable amount of metabolic heterogeneity, much of it genetically defined and some associated with disease. It has been a long-standing challenge to identify genomic variants that contribute to metabolic heterogeneity and relate them to disease. Measuring a few key metabolites is a long-established component of the metabolic disease workup. But more recently, analytical advances in genomics and metabolomics have made it possible to perform a holistic and integrated assessment of the genetic basis of metabolic variation. This section describes conceptual and practical advances in this area. Genetic Modifiers of Metabolism in Mice and Humans Inbred mice are instrumental in defining the basis of metabolic phenotypes. However these strains are useful specifically because of the lack of genomic and metabolic heterogeneity among individuals. To capitalize on the mouse as a model system to understand genetic determinants of metabolism, eight inbred strains representing most of the genetic diversity in commonly-used lab mice were intercrossed to create the Diversity Outbred (DO) stocks (Threadgill et al., 2011). Individual DO mice differ from each other by millions of single nucleotide polymorphisms, making it possible to use genetics to identify variants associated with metabolic features. More than 300 DO mice were phenotyped for over 3,000 plasma lipidomic features. Lipid abundance was then mapped to quantitative trait loci in the genome, de-orphaning many lipid species and identifying new aspects of regulation (Linke et al., 2020). In another study, pancreatic islets from DO mice were analyzed ex vivo for their ability to release insulin in response to secretagogues, revealing several loci that contribute to variability in insulin secretion (Keller et al., 2019). Similar approaches have been used in humans, capitalizing on natural variation across populations and combining genome-wide association studies with metabolomics. These efforts demonstrated that polymorphisms in genes encoding metabolic enzymes account for a substantial fraction of natural variation in metabolite abundance in plasma from healthy people (Gieger et al., 2008; Illig et al., 2010). Many effects are phenotypically silent beyond metabolite abundance, but some genotype-metabolite associations involve genes previously connected to common, multifactorial disorders like coronary artery disease, hypertension and diabetes (Suhre et al., 2011). The Dallas Heart Study (DHS), a large multiethnic population-based probability sample, has for 20 years used deep metabolic phenotyping and genome-wide association studies to detect variants accounting for disease-associated metabolic perturbations (Victor et al., 2004). This approach led to the discovery that loss-of-function variants in PCSK9 are associated with reduced plasma low-density lipoprotein levels and protection against coronary heart disease (Cohen et al., 2006). This insight stimulated development of PCSK9 inhibitors, which are now used to treat refractory hypercholesterolemia (Sabatine et al., 2017). Another DHS advance was the association of variants in PNPLA3, which encodes a triacylglycerol lipase, with hepatic fat content and liver inflammation, hallmarks of nonalcoholic fatty liver disease (NAFLD) (Romeo et al., 2008). This discovery launched hundreds of studies investigating the mechanisms linking PNPLA3 to hepatic fat, and how to develop NAFLD therapies by targeting PNPLA3. Metabolomics in Rare Human Diseases Metabolomics is increasingly used to characterize rare human phenotypes, particularly as whole-exome and whole-genome sequencing (WES, WGS) have become clinically available to diagnose Mendelian diseases. WES and WGS are unmatched in their ability to detect potentially pathogenic genomic variants. But a challenge with sequence-first approaches is to interpret variants of uncertain significance; that is, sequence variants that differ from the consensus but with unknown effects on gene function. In this scenario, metabolomics provides a useful dimension of deep phenotyping. Conventional workup for inborn errors by clinical biochemical laboratories focuses on a few dozen metabolites, but modern metabolomics can report hundreds of metabolites at once. This makes it possible to observe rare patterns of metabolic anomalies in patients with nonspecific phenotypes, and to relate these to rare genomic variants uncovered by WES/WGS (Figure 2). In a recent example, we observed a pattern of plasma metabolic abnormalities that were difficult to ascribe to a defect in any single enzyme. WES revealed variants in LIPT1, which encodes the lipoyltransferase that adds a covalently-bound cofactor to several 2-ketoacid dehydrogenases. The pattern of metabolomic abnormalities was then easily recognizable as a combined defect in lipoylation-dependent enzymes, and this was confirmed in functional assays in patient-derived cells (Ni et al., 2019). Metabolomic-Assisted Pathway Screening (MAPS) is a clinical diagnostic test based on metabolomic profiling in plasma and envisioned as a companion to WES/WGS. In a series of 170 patients assessed by both WES and MAPS, the metabolomics data contributed to interpretation of genomic variants in over 40% of cases (Alaimo et al., 2020). Confirming the metabolic effects of sequence variants – or lack thereof – is of diagnostic value because it facilitates interpreting these variants in future patients with unexplained phenotypes. Newborn screening identifies pre-symptomatic patients with treatable inborn errors and other diseases by testing all babies for diagnostic biomarkers (often metabolites). Because most disorders covered by these programs have a genetic basis, it is reasonable to compare DNA analysis to conventional metabolite detection in terms of diagnostic yield. A re-analysis of samples from over 1,000 babies with positive screens for inborn errors of metabolism found that biochemical analysis outperformed sequencing as a stand-alone test, both in sensitivity and specificity (Adhikari et al., 2020). The DNA analysis’ lack of sensitivity, largely caused by patients lacking variants that could be detected by WES, is problematic because the goal of newborn screening is to identify every baby eligible for treatment. Metabolomics in patients with known inborn errors could help uncover new biomarkers amenable to newborn screening. This would be useful in treatable diseases not included on newborn screens because they lack diagnostic biomarkers, and in diseases that may become treatable in the future. Overall, these examples of combining genomics with metabolic profiling have provided insights into the basis of heritable metabolic heterogeneity, and in some cases led to new therapies for common metabolic diseases such as dyslipidemia. In the case of rare monogenic disorders, broad metabolomic profiling has improved the utility of clinical genomics, providing a deep phenotyping dimension that clarifies the significance of genomic variants. ADVANCE 4: TRANSLATIONAL METABOLIC IMAGING While metabolic mechanisms can be readily teased out in experimental models, most analytical methods are destructive or require long periods of data acquisition, limiting feasibility in patients. New approaches to in vivo metabolic imaging have provided a means to leverage such metabolic changes to better understand human biology. These encompass approaches in nuclear medicine (herein predominantly positron emission tomography, PET) and magnetic resonance that have the ability to resolve metabolic fluxes in humans. Radioactive methods for metabolic imaging Historically, PET has been the standard for probing human metabolism. From the onset of the most widely used tracer, 18F-flurodeoxyglucose (FDG), PET scans have been leveraged to infer changes in glucose metabolism (Ido et al., 1978), particularly in oncology and cardiology. While the radioactive dose administered for a PET scan is quite low, it limits the number of PET scans that can be used in a given patient, particularly in children. Recent work has led to the development of whole body PET scanners (EXPLORER) (Badawi et al., 2019) that provide enhanced sensitivity and a 10-fold reduction in radioactive dose. The gain in signal strength can be utilized to acquire time-dependent data as opposed to the summation of accumulated ion counts in traditional PET. Moreover, with time varying signals throughout the body and improved sensitivity, one can better localize the signal using time of flight (TOF) PET imaging (Surti, 2015). Since localization of the radionuclide is a function of the accuracy of measuring coincident photons hitting the PET detector 180 degrees apart, TOF PET imaging leverages a highly accurate measurement of the timing of photon detection (on the order of picoseconds) to improve localization. Bringing these advances together and complementing them with new reconstruction approaches, TOF PET in the EXPLORER has the potential to fit kinetic rate models at the systems level across a wide range of tissues, promising total body measurements. FDG PET detects radionuclide uptake and retention but is limited in its ability to resolve specific metabolic reactions, leading to ambiguity in its interpretation. FDG is “metabolically trapped” after phosphorylation by hexokinase (Gallagher et al., 1978) and does not proceed through glycolysis due to the lack of the C2 hydroxyl necessary for isomerization to fructose-6-phosphate (F6P). This results in accumulation as G6P, further metabolism into the PPP or dephosphorylation and export. Recent work has focused on the development of new tracers able to probe other, potentially more specific metabolic reactions (e.g. 18F-glutamine (Lieberman et al., 2011), 18F-FACBC (Shoup et al., 1999) and 18F-acetate (Ponde et al., 2007)). Understanding how these tracers accumulate is key for the interpretation of these images. For example, 18F-glutamine is not metabolically converted to other products and informs on glutamine abundance rather than glutamine metabolism (Venneti et al., 2015). In the setting of glutaminase inhibition, 18F-glutamine accumulates in response to on-target inhibition, suggesting that the pool size increases (Abu Aboud et al., 2017; Viswanath et al., 2021; Zhou et al., 2017). With the appropriate context, such readouts can be used to inform on the metabolic consequences of perturbation in vivo, but critically require a full biochemical understanding of the probe’s mechanism of action. 18F analogues, while harboring fairly long lifetimes, are limited in that they do not exhibit the same metabolism as the natural substrates of metabolic reactions. For this reason, 13N and 11C analogues have been re-explored for use in metabolic imaging studies. For example, 11C acetate has the potential not only to inform on fatty acid metabolism but also varying degrees of acetylation, now of increasing interest considering the tethering of epigenetics to metabolism (Comerford et al., 2014). Moreover, although this approach is under-utilized, radioisotope labeling at specific positions can further inform on flux selection to better interpret metabolic images. For example, incorporation of 11C at the C1 position of lactate is retained in tissues with a large lactate pool, but upon generation of pyruvate and subsequent metabolism through PDH, it is lost as CO2, heavily weighting the signal to lactate pool size. In contrast, incorporation of 11C at the C3 position of lactate provides a means of incorporating lactate metabolism in the TCA cycle as the label is retained beyond PDH (Figure 3). These approaches have the potential to provide high sensitivity for metabolic trapping in vivo, though their nuclear half-lives can be limiting (13N – t1/2~10 min and 11C t1/2~20.3 min). This makes it essential that these tracers be radio-synthesized on site (Parent and McConathy, 2018). Ultimately, position matters and sophisticated chemical methods are needed to rapidly synthesize and purify such precursors. Stable isotope methods for metabolic imaging PET methods provide high sensitivity but relatively low specificity and spatial localization. MRI in contrast provides exquisite spatial resolution and specificity but lacks the sensitivity typically needed to quantify metabolic processes. Several MR methods pioneered in recent years have sought to address the growing need to study metabolism in humans. Stable isotope tracing is a versatile approach that provides information about metabolic activity in vivo. Recent efforts using stable isotope-labeled nutrients have revealed new aspects of disease-related metabolic activity, including the unexpected role of serine in hepatic lipogenesis in mice and the use of lactate as a fuel in aggressive human lung tumors (Faubert et al., 2017; Zhang et al., 2021). Conveniently, the same isotopes (2H, 13C and 15N) used for tracing are detectable using MRI and magnetic resonance spectroscopy (MRS). While sensitivity is still limited for such tracing approaches, work in the setting of cancer as well as other diseases has demonstrated the ability to infuse isotopically labeled nutrients and follow their metabolic conversion non-invasively. For example, in the brain, fluxes from substrates such as 13C-acetate and 13C-glucose can be used to measure recycling rates of glutamine and glutamate (Boumezbeur et al., 2010; Lebon et al., 2002). This can be extended further to measuring TCA cycle flux and anapleorsis in human liver (Befroy et al., 2014). 2H is transferred from 2H-glucose to 2H-lactate in human brain tumors, while 2H-glx (a combination of glutamine, glutamate and GSH) is a more prominent product in normal brain (De Feyter et al., 2018). While these approaches provide a means of measuring a rate, like conventional MRS they continue to suffer from both spatial resolution and time required to acquire such data due to low signal to noise ratio (SNR). Newer approaches combining stable isotope tracers with indirect 1H MRS detection may be able to overcome this in small fields of view in the brain, though extending outside of the brain is extremely challenging (Rich et al., 2020). Long scan times, motion artifacts and field inhomogeneities can severely limit SNR and make it challenging to extend these approaches to large fields of view. More work is needed to develop the MRI detection coils and pulse sequences required to overcome these limitations, though this is an emerging field of metabolic research that has promise. By far the most exciting advance in the area of imaging metabolic activity in situ with isotope tracers has been the development of hyperpolarized magnetic resonance imaging (Keshari and Wilson, 2014; Kurhanewicz et al., 2019). Hyperpolarization refers to a massive increase in the spin polarization of a target nucleus beyond its Boltzmann state for a given magnetic field. These hyperpolarized spins are detected using a conventional MRI system, enhancing the signal by thousands of fold. While there are now a multitude of ways to generate hyperpolarized nuclei, the end goal of all such approaches has been to generate a hyperpolarized nutrient which can be infused into a living system to measure enzymatic conversion. Recent work has shown that such approaches can be leveraged to follow the metabolic conversion of many substrates including glucose (Rodrigues et al., 2014), fructose (Keshari et al., 2009), glutamine (Salamanca-Cardona et al., 2017) and most significantly pyruvate (Golman and Petersson, 2006; Nelson et al., 2013) in tumors. 13C pyruvate labeled in the C1 position has proven capable of revealing fluxes both through reduction to lactate and oxidation to CO2 and further equilibration to bicarbonate in vivo, providing the potential to measure rate constants for these reactions at spatial resolutions that rival radionuclide approaches (Cunningham et al., 2016). Most notably, 13C pyruvate reduction to lactate has been shown to correlate with cancer grade (Granlund et al., 2020) and oxidation to bicarbonate has provided exquisite maps of cardiac function with potential applications in heart disease. Other approaches are in the process of translation, including 13C dehydroascorbate to image oxidative stress (Keshari et al., 2011) and 13C fumarate to assess necrosis (Gallagher et al., 2009). The major limitation of HP MRI is the lifetime of the substrate and its products in vivo, currently on the order of a few minutes, thus restricting both the range of molecules that can be probed and the reaction kinetic lifetime that can be observed. Taken together, these rapidly evolving nuclear and magnetic resonance approaches are approaching prime time. In the last 10 years, true breakthroughs in these techniques have provided new opportunities to study metabolic dimensions of disease. Stable isotope tracing using deuterium MRI in the brain has provided a means to annotate glycolytic utilization without the need for radioactivity (De Feyter et al., 2018). HP MRI has created a platform which can measure flux to a specific product in vivo, defining the source of the oncometabolite 2-HG (Salamanca-Cardona et al., 2017); connecting altered metabolic flux to loss of PTEN in prostate cancer patients (Granlund et al., 2020); and uncovering control of LDH expression by FOXM1 in breast cancer (Ros et al., 2020). OPPORTUNITIES FOR FUTURE DISCOVERY Building on a legacy of discovery that defined the human metabolic network, the technological and conceptual advances described above have given rise to new insights into the interface between metabolism and disease. These advances have occurred across scales, leveraging tools to study metabolism from the subcellular level all the way to human populations. There are many exciting opportunities for future discoveries across these scales as well, and a few are discussed below. Localization of pathways to membrane-enclosed organelles like the mitochondria is a well-known component of metabolic regulation. But there is increasing appreciation that some pathways are also associated into “metabolons,” macromolecular complexes that are not otherwise sequestered from the rest of the cell. Organizing pathways in this manner is appealing because it implies several ways to regulate flux. The intimate co-localization of enzymes catalyzing sequential reactions could facilitate substrate channeling or increase pathway efficiency by eliminating the need for intermediates to diffuse from one enzyme to the next. Post-translational modifications could regulate protein inclusion or exclusion into these metabolons, thereby altering pathway activity. As an example, the purinosome is a biomolecular condensate of enzymes from de novo purine biosynthesis. It is formed through liquid-liquid phase separation in cultured cells and proposed to facilitate metabolite channeling (Pedley et al., 2022). We do not know the full set of functional mammalian metabolons, how they are regulated to match metabolic supply and demand, and which if any are relevant to disease. Research into biomolecular condensates and other mechanisms of subcellular localization contributing to metabolism may be aided by advanced cellular imaging methods to assess protein localization and macromolecular structures in live tissues, and new intracellular sensors that can visualize metabolite distribution. Understanding how metabolism is regulated within tissues, particularly how metabolic interactions among functionally distinct cell types contribute to overall performance of the tissue, will require better resolution of metabolic and orthogonal properties in situ. MALDI and related MSI techniques already allow some metabolic features to be localized within tissue sections, but it is still difficult to assign features to specific cell types within complex tissues. Metabolic MSI would benefit from further method development to superimpose several spatially-resolved features in the same sample, essentially multiplexing transcriptomic, proteomic, metabolomic and possibly even isotope enrichment features with high-resolution histological data. In principle, this could provide information about metabolism, potentially at the single cell level and assigned to individual cell types. Finally, because identifying and quantifying metabolic mechanisms drives our understanding of disease, localizing metabolic perturbations in patients is key to our ability to leverage these findings to address disease. Rapid advances in clinical metabolic imaging have already begun, and we anticipate that new developments will allow us to illuminate pathways in an unprecedented way. Given the complementary nature of clinical imaging tools, e.g. metabolic MRI and PET, the fusion of these modalities with each other, and the use of a broad range of metabolic probes, could provide a holistic view of systems metabolism. Facilitating the interrogation of metabolism across scales, from organelles to patients, has the potential to transform how we approach research, diagnostics and therapy in metabolic disease. ACKNOWLEDGMENTS We thank Javier Garcia-Bermudez and members of the DeBerardinis and Keshari labs for inspiration and critique. R.J.D. is supported by the Howard Hughes Medical Institute and grants from the N.C.I. (R35CA22044901, P50CA196516-02 and 2P50CA070907-21A1) and The Cancer Prevention and Research Institute of Texas (RP180778). K.R.K. is supported by grants from the N.C.I. (R01CA237466, R01CA252037, R01CA248364, R01CA249294, and Cancer Center Support Grant P30CA008748). The drafts of Figures 1-3 were generated using Biorender. Figure 1: High-throughput screening to identify mechanisms by which cells respond to metabolic stress. (A) Here a guide RNA (gRNA) library induces loss-of-function mutations against metabolic genes. The library is introduced to cultured cells, producing a mixed population. The cells are grown in the presence or absence of a metabolic stress, and sequencing identifies gRNAs whose abundance changes selectively under stress. In this example, several gRNAs are depleted (lost) during standard culture. These gRNAs target genes required for growth under standard conditions. Blue cells increase in number (gain) because the gRNA in these cells targets a growth-suppressive gene. A different distribution occurs in cells exposed to metabolic stress, highlighted by the dashed box, with some gRNAs lost (purple) or gained (green) only under stress. (B) Cells use the red pathway to produce a metabolite, M5, required for growth. The pathway also produces the mildly toxic M6. The purple pathway is dispensable in conventional culture but can produce M5 when levels of the nutrient increase, for example if E1 is mutated or blocked with an inhibitor. A CRISPR screen like the one in (A) would generate the results shown on the right. Figure 2: Integrating metabolomics and genomics to identify the molecular basis of metabolic anomalies. Broad metabolomic profiling provides an orthogonal approach to whole exome and whole genome sequencing in patients whose phenotypes make it difficult to pinpoint the responsible gene or pathway. This is valuable when sequence variants of uncertain significance coincide with metabolomic anomalies in a related pathway, increasing confidence that the sequence variants alter gene function. In the illustrated scenario, exome/genome sequencing reveals that the patient has a variant of uncertain significance in the gene encoding enzyme E2. In the patient’s blood, accumulation of metabolites M1 and M2 upstream of E2, and depletion of metabolite M3 downstream of E2, imply that the variant impairs E2’s function. This hypothesis could be validated by performing experiments to directly test the impact of the E2 variant in cells and mice, and by identifying other patients with the same metabolomic-genomic findings. Figure 3: Isotopomer-selective in vivo PET imaging as a means to reveal differential metabolism. Many molecular imaging approaches are geared toward in vivo annotation of metabolism. Radioactive approaches, while extremely sensitive, suffer from the inability to discern specific metabolites. This can be circumvented by using isotopmers where the radionuclide is positioned at different locations within the molecule, designed to report different aspects of metabolic activity. In the example, lactate is labeled with 11C in either C3 (Tracer 1, blue) or C1 (Tracer 2, orange) to illustrate the loss of 11C1 signal when PDH-mediated decarboxylation takes place, releasing 11CO2. This would produce 11C PET images of differential intensity at different locations in the same patient, e.g. in individual lesions suspected to be malignant. Blue lesions would be the highest intensity, and the relative intensity of the orange lesions would reflect the percent of the flux incorporated into the TCA cycle through PDH as opposed to total lactate oxidation. DECLARATION OF INTERESTS R.J.D. is a founder and advisor at Atavistik Bio, and serves on the Scientific Advisory Boards of Agios Pharmaceuticlas, Vida Ventures, Droia Ventures and Nirogy Therapeutics. K.R.K. is co-founder of Atish Technologies, and serves on the Scientific Advisory Boards of NVision Imaging Technologies and Imaginostics. He holds patents related to imaging and leveraging cellular metabolism. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. REFERENCES Abu Aboud O , Habib SL , Trott J , Stewart B , Liang S , Chaudhari AJ , Sutcliffe J , and Weiss RH (2017). Glutamine Addiction in Kidney Cancer Suppresses Oxidative Stress and Can Be Exploited for Real-Time Imaging. Cancer research 77 , 6746–6758. 10.1158/0008-5472.CAN-17-0930.29021138 Abu-Remaileh M , Wyant GA , Kim C , Laqtom NN , Abbasi M , Chan SH , Freinkman E , and Sabatini DM (2017). Lysosomal metabolomics reveals V-ATPase- and mTOR-dependent regulation of amino acid efflux from lysosomes. Science 358 , 807–813. 10.1126/science.aan6298.29074583 Adhikari AN , Gallagher RC , Wang Y , Currier RJ , Amatuni G , Bassaganyas L , Chen F , Kundu K , Kvale M , Mooney SD , (2020). The role of exome sequencing in newborn screening for inborn errors of metabolism. Nat Med 26 , 1392–1397. 10.1038/s41591-020-0966-5.32778825 Alaimo JT , Glinton KE , Liu N , Xiao J , Yang Y , Reid Sutton V , and Elsea SH (2020). Integrated analysis of metabolomic profiling and exome data supplements sequence variant interpretation, classification, and diagnosis. Genet Med 22 , 1560–1566. 10.1038/s41436-020-0827-0.32439973 Arguello RJ , Combes AJ , Char R , Gigan JP , Baaziz AI , Bousiquot E , Camosseto V , Samad B , Tsui J , Yan P , (2020). SCENITH: A Flow Cytometry-Based Method to Functionally Profile Energy Metabolism with Single-Cell Resolution. Cell metabolism 32 , 1063–1075 e1067. 10.1016/j.cmet.2020.11.007.33264598 Artyomov MN , and Van den Bossche J (2020). Immunometabolism in the Single-Cell Era. Cell metabolism 32 , 710–725. 10.1016/j.cmet.2020.09.013.33027638 Badawi RD , Shi H , Hu P , Chen S , Xu T , Price PM , Ding Y , Spencer BA , Nardo L , Liu W , (2019). First Human Imaging Studies with the EXPLORER Total-Body PET Scanner. Journal of nuclear medicine : official publication, Society of Nuclear Medicine 60 , 299–303. 10.2967/jnumed.119.226498.30733314 Befroy DE , Perry RJ , Jain N , Dufour S , Cline GW , Trimmer JK , Brosnan J , Rothman DL , Petersen KF , and Shulman GI (2014). Direct assessment of hepatic mitochondrial oxidative and anaplerotic fluxes in humans using dynamic 13C magnetic resonance spectroscopy. Nature medicine 20 , 98–102. 10.1038/nm.3415. Biancur DE , Kapner KS , Yamamoto K , Banh RS , Neggers JE , Sohn ASW , Wu W , Manguso RT , Brown A , Root DE , (2021). Functional Genomics Identifies Metabolic Vulnerabilities in Pancreatic Cancer. Cell Metab 33 , 199–210 e198. 10.1016/j.cmet.2020.10.018.33152323 Birsoy K , Wang T , Chen WW , Freinkman E , Abu-Remaileh M , and Sabatini DM (2015). An Essential Role of the Mitochondrial Electron Transport Chain in Cell Proliferation Is to Enable Aspartate Synthesis. Cell 162 , 540–551. 10.1016/j.cell.2015.07.016.26232224 Blacker TS , Mann ZF , Gale JE , Ziegler M , Bain AJ , Szabadkai G , and Duchen MR (2014). Separating NADH and NADPH fluorescence in live cells and tissues using FLIM. Nat Commun 5 , 3936. 10.1038/ncomms4936.24874098 Boumezbeur F , Mason GF , de Graaf RA , Behar KL , Cline GW , Shulman GI , Rothman DL , and Petersen KF (2010). Altered brain mitochondrial metabolism in healthy aging as assessed by in vivo magnetic resonance spectroscopy. J Cereb Blood Flow Metab 30 , 211–221. 10.1038/jcbfm.2009.197.19794401 Cameron WD , Bui CV , Hutchinson A , Loppnau P , Graslund S , and Rocheleau JV (2016). Apollo-NADP(+): a spectrally tunable family of genetically encoded sensors for NADP(+). Nature methods 13 , 352–358. 10.1038/nmeth.3764.26878383 Chaurand P , Norris JL , Cornett DS , Mobley JA , and Caprioli RM (2006). New developments in profiling and imaging of proteins from tissue sections by MALDI mass spectrometry. J Proteome Res 5 , 2889–2900. 10.1021/pr060346u.17081040 Chen WW , Freinkman E , and Sabatini DM (2017). Rapid immunopurification of mitochondria for metabolite profiling and absolute quantification of matrix metabolites. Nat Protoc 12 , 2215–2231. 10.1038/nprot.2017.104.29532801 Chen WW , Freinkman E , Wang T , Birsoy K , and Sabatini DM (2016). Absolute Quantification of Matrix Metabolites Reveals the Dynamics of Mitochondrial Metabolism. Cell 166 , 1324–1337 e1311. 10.1016/j.cell.2016.07.040.27565352 Choi C , Raisanen JM , Ganji SK , Zhang S , McNeil SS , An Z , Madan A , Hatanpaa KJ , Vemireddy V , Sheppard CA , (2016). Prospective Longitudinal Analysis of 2-Hydroxyglutarate Magnetic Resonance Spectroscopy Identifies Broad Clinical Utility for the Management of Patients With IDH-Mutant Glioma. J Clin Oncol 34 , 4030–4039. 10.1200/JCO.2016.67.1222.28248126 Choi JE , Sebastian C , Ferrer CM , Lewis CA , Sade-Feldman M , LaSalle T , Gonye A , Lopez BGC , Abdelmoula WM , Regan MS , (2021). A unique subset of glycolytic tumour-propagating cells drives squamous cell carcinoma. Nat Metab 3 , 182–195. 10.1038/s42255-021-00350-6.33619381 Cohen JC , Boerwinkle E , Mosley TH Jr. , and Hobbs HH (2006). Sequence variations in PCSK9, low LDL, and protection against coronary heart disease. N Engl J Med 354 , 1264–1272. 10.1056/NEJMoa054013.16554528 Comerford SA , Huang Z , Du X , Wang Y , Cai L , Witkiewicz AK , Walters H , Tantawy MN , Fu A , Manning HC , (2014). Acetate dependence of tumors. Cell 159 , 1591–1602. 10.1016/j.cell.2014.11.020.25525877 Cunningham CH , Lau JY , Chen AP , Geraghty BJ , Perks WJ , Roifman I , Wright GA , and Connelly KA (2016). Hyperpolarized 13C Metabolic MRI of the Human Heart: Initial Experience. Circ Res 119 , 1177–1182. 10.1161/CIRCRESAHA.116.309769.27635086 Dang L , White DW , Gross S , Bennett BD , Bittinger MA , Driggers EM , Fantin VR , Jang HG , Jin S , Keenan MC , (2009). Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature 462 , 739–744. 10.1038/nature08617.19935646 Daniels VW , Zoeller JJ , van Gastel N , McQueeney KE , Parvin S , Potter DS , Fell GG , Ferreira VG , Yilma B , Gupta R , (2021). Metabolic perturbations sensitize triple-negative breast cancers to apoptosis induced by BH3 mimetics. Sci Signal 14 . 10.1126/scisignal.abc7405. Davidson SM , Papagiannakopoulos T , Olenchock BA , Heyman JE , Keibler MA , Luengo A , Bauer MR , Jha AK , O'Brien JP , Pierce KA , (2016). Environment Impacts the Metabolic Dependencies of Ras-Driven Non-Small Cell Lung Cancer. Cell Metab 23 , 517–528. 10.1016/j.cmet.2016.01.007.26853747 De Feyter HM , Behar KL , Corbin ZA , Fulbright RK , Brown PB , McIntyre S , Nixon TW , Rothman DL , and de Graaf RA (2018). Deuterium metabolic imaging (DMI) for MRI-based 3D mapping of metabolism in vivo. Sci Adv 4 , eaat7314. 10.1126/sciadv.aat7314.30140744 DeVilbiss AW , Zhao Z , Martin-Sandoval MS , Ubellacker JM , Tasdogan A , Agathocleous M , Mathews TP , and Morrison SJ (2021). Metabolomic profiling of rare cell populations isolated by flow cytometry from tissues. Elife 10 . 10.7554/eLife.61980. DiNardo CD , Stein EM , de Botton S , Roboz GJ , Altman JK , Mims AS , Swords R , Collins RH , Mannis GN , Pollyea DA , (2018). Durable Remissions with Ivosidenib in IDH1-Mutated Relapsed or Refractory AML. N Engl J Med 378 , 2386–2398. 10.1056/NEJMoa1716984.29860938 Faubert B , Li KY , Cai L , Hensley CT , Kim J , Zacharias LG , Yang C , Do QN , Doucette S , Burguete D , (2017). Lactate Metabolism in Human Lung Tumors. Cell 171 , 358–371 e359. 10.1016/j.cell.2017.09.019.28985563 Figueroa ME , Abdel-Wahab O , Lu C , Ward PS , Patel J , Shih A , Li Y , Bhagwat N , Vasanthakumar A , Fernandez HF , (2010). Leukemic IDH1 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair hematopoietic differentiation. Cancer Cell 18 , 553–567. 10.1016/j.ccr.2010.11.015.21130701 Gallagher BM , Fowler JS , Gutterson NI , MacGregor RR , Wan CN , and Wolf AP (1978). Metabolic trapping as a principle of oradiopharmaceutical design: some factors resposible for the biodistribution of [18F] 2-deoxy-2-fluoro-D-glucose. Journal of nuclear medicine : official publication, Society of Nuclear Medicine 19 , 1154–1161.214528 Gallagher FA , Kettunen MI , Hu DE , Jensen PR , Zandt RI , Karlsson M , Gisselsson A , Nelson SK , Witney TH , Bohndiek SE , (2009). Production of hyperpolarized [1,4-13C2]malate from [1,4-13C2]fumarate is a marker of cell necrosis and treatment response in tumors. Proceedings of the National Academy of Sciences of the United States of America 106 , 19801–19806. 10.1073/pnas.0911447106.19903889 Garcia-Bermudez J , Baudrier L , Bayraktar EC , Shen Y , La K , Guarecuco R , Yucel B , Fiore D , Tavora B , Freinkman E , (2019). Squalene accumulation in cholesterol auxotrophic lymphomas prevents oxidative cell death. Nature 567 , 118–122. 10.1038/s41586-019-0945-5.30760928 Garcia-Bermudez J , Baudrier L , La K , Zhu XG , Fidelin J , Sviderskiy VO , Papagiannakopoulos T , Molina H , Snuderl M , Lewis CA , (2018). Aspartate is a limiting metabolite for cancer cell proliferation under hypoxia and in tumours. Nat Cell Biol 20 , 775–781. 10.1038/s41556-018-0118-z.29941933 Gieger C , Geistlinger L , Altmaier E , Hrabe de Angelis M , Kronenberg F , Meitinger T , Mewes HW , Wichmann HE , Weinberger KM , Adamski J , (2008). Genetics meets metabolomics: a genome-wide association study of metabolite profiles in human serum. PLoS Genet 4 , e1000282. 10.1371/journal.pgen.1000282.19043545 Gillespie M , Jassal B , Stephan R , Milacic M , Rothfels K , Senff-Ribeiro A , Griss J , Sevilla C , Matthews L , Gong C , (2022). The reactome pathway knowledgebase 2022. Nucleic Acids Res 50 , D687–D692. 10.1093/nar/gkab1028.34788843 Golman K , and Petersson JS (2006). Metabolic imaging and other applications of hyperpolarized 13C1. Acad Radiol 13 , 932–942. 10.1016/j.acra.2006.06.001.16843845 Granlund KL , Tee SS , Vargas HA , Lyashchenko SK , Reznik E , Fine S , Laudone V , Eastham JA , Touijer KA , Reuter VE , (2020). Hyperpolarized MRI of Human Prostate Cancer Reveals Increased Lactate with Tumor Grade Driven by Monocarboxylate Transporter 1. Cell metabolism 31 , 105–114 e103. 10.1016/j.cmet.2019.08.024.31564440 Harris IS , Endress JE , Coloff JL , Selfors LM , McBrayer SK , Rosenbluth JM , Takahashi N , Dhakal S , Koduri V , Oser MG , (2019). Deubiquitinases Maintain Protein Homeostasis and Survival of Cancer Cells upon Glutathione Depletion. Cell Metab 29 , 1166–1181 e1166. 10.1016/j.cmet.2019.01.020.30799286 Hicks KG , C. AA , Schubert HL , Hackett SR , Berg JA , Leonard PG , Ajalla Aleixo MA , Blevins A , Barta P , Tilley S , Fogarty S , Winter JM , Ahn H-C , Allen KN , Block S , Cardoso IA , Ding J , Dreveny I , Gasper C , Ho Q , Matsuura A , Palladino MJ , Prajapati S , Sun P , Tittmann K , Tolan DR , Unterlass J , VanDemark AP , Vander Heiden MG , Webb BA , Yun C-H , Zhap P , Hill CP , Nonato MC , Muller FL , Gottschling DE , Cox JE , Rutter J (2021). Protein-Metabolite Interactomics Reveals Novel Regulation of Carbohydrate Metabolism. BioRxiv. 10.1101/2021.08.28.458030. Ido T , Wan C-N , Casella V , Fowler J , Wolf A , Reivich M , and Kuhl D (1978). Labeled 2- deoxy-D-glucose analogs, 18F-labeled 2-deoxy-2-fluoro-D-glucose, 2-deoxy-2-fluoro-D-mannose and 14C-2-deoxy-2-fluoro-S-glucose. J Labelled Compd and Radiopharm 14 , 171–183. Illig T , Gieger C , Zhai G , Romisch-Margl W , Wang-Sattler R , Prehn C , Altmaier E , Kastenmuller G , Kato BS , Mewes HW , (2010). A genome-wide perspective of genetic variation in human metabolism. Nat Genet 42 , 137–141. 10.1038/ng.507.20037589 Intlekofer AM , Shih AH , Wang B , Nazir A , Rustenburg AS , Albanese SK , Patel M , Famulare C , Correa FM , Takemoto N , (2018). Acquired resistance to IDH inhibition through trans or cis dimer-interface mutations. Nature 559 , 125–129. 10.1038/s41586-018-0251-7.29950729 Jeong S , Eskandari R , Park SM , Alvarez J , Tee SS , Weissleder R , Kharas MG , Lee H , and Keshari KR (2017). Real-time quantitative analysis of metabolic flux in live cells using a hyperpolarized micromagnetic resonance spectrometer. Sci Adv 3 , e1700341. 10.1126/sciadv.1700341.28630930 Keller MP , Rabaglia ME , Schueler KL , Stapleton DS , Gatti DM , Vincent M , Mitok KA , Wang Z , Ishimura T , Simonett SP , (2019). Gene loci associated with insulin secretion in islets from non-diabetic mice. J Clin Invest 129 , 4419–4432. 10.1172/JCI129143.31343992 Keshari KR , Kurhanewicz J , Bok R , Larson PE , Vigneron DB , and Wilson DM (2011). Hyperpolarized 13C dehydroascorbate as an endogenous redox sensor for in vivo metabolic imaging. Proceedings of the National Academy of Sciences of the United States of America 108 , 18606–18611. 10.1073/pnas.1106920108.22042839 Keshari KR , and Wilson DM (2014). Chemistry and biochemistry of 13C hyperpolarized magnetic resonance using dynamic nuclear polarization. Chemical Society reviews 43 , 1627–1659. 10.1039/c3cs60124b.24363044 Keshari KR , Wilson DM , Chen AP , Bok R , Larson PE , Hu S , Van Criekinge M , Macdonald JM , Vigneron DB , and Kurhanewicz J (2009). Hyperpolarized [2-13C]-fructose: a hemiketal DNP substrate for in vivo metabolic imaging. J Am Chem Soc 131 , 17591–17596. 10.1021/ja9049355.19860409 King MP , and Attardi G (1989). Human cells lacking mtDNA: repopulation with exogenous mitochondria by complementation. Science 246 , 500–503. 10.1126/science.2814477.2814477 Kory N , Uit de Bos J , van der Rijt S , Jankovic N , Gura M , Arp N , Pena IA , Prakash G , Chan SH , Kunchok T , (2020). MCART1/SLC25A51 is required for mitochondrial NAD transport. Sci Adv 6 . 10.1126/sciadv.abe5310. Kurhanewicz J , Vigneron DB , Ardenkjaer-Larsen JH , Bankson JA , Brindle K , Cunningham CH , Gallagher FA , Keshari KR , Kjaer A , Laustsen C , (2019). Hyperpolarized (13)C MRI: Path to Clinical Translation in Oncology. Neoplasia 21 , 1–16. 10.1016/j.neo.2018.09.006.30472500 Lebon V , Petersen KF , Cline GW , Shen J , Mason GF , Dufour S , Behar KL , Shulman GI , and Rothman DL (2002). Astroglial contribution to brain energy metabolism in humans revealed by 13C nuclear magnetic resonance spectroscopy: elucidation of the dominant pathway for neurotransmitter glutamate repletion and measurement of astrocytic oxidative metabolism. J Neurosci 22 , 1523–1531.11880482 Lieberman BP , Ploessl K , Wang L , Qu W , Zha Z , Wise DR , Chodosh LA , Belka G , Thompson CB , and Kung HF (2011). PET imaging of glutaminolysis in tumors by 18F-(2S,4R)4-fluoroglutamine. Journal of nuclear medicine : official publication, Society of Nuclear Medicine 52 , 1947–1955. 10.2967/jnumed.111.093815.22095958 Linke V , Overmyer KA , Miller IJ , Brademan DR , Hutchins PD , Trujillo EA , Reddy TR , Russell JD , Cushing EM , Schueler KL , (2020). A large-scale genome-lipid association map guides lipid identification. Nat Metab 2 , 1149–1162. 10.1038/s42255-020-00278-3.32958938 Locasale JW , Grassian AR , Melman T , Lyssiotis CA , Mattaini KR , Bass AJ , Heffron G , Metallo CM , Muranen T , Sharfi H , (2011). Phosphoglycerate dehydrogenase diverts glycolytic flux and contributes to oncogenesis. Nat Genet 43 , 869–874. 10.1038/ng.890.21804546 Mardis ER , Ding L , Dooling DJ , Larson DE , McLellan MD , Chen K , Koboldt DC , Fulton RS , Delehaunty KD , McGrath SD , (2009). Recurring mutations found by sequencing an acute myeloid leukemia genome. N Engl J Med 361 , 1058–1066. 10.1056/NEJMoa0903840.19657110 Nelson SJ , Kurhanewicz J , Vigneron DB , Larson PE , Harzstark AL , Ferrone M , van Criekinge M , Chang JW , Bok R , Park I , (2013). Metabolic imaging of patients with prostate cancer using hyperpolarized [1-(1)(3)C]pyruvate. Science translational medicine 5 , 198ra108. 10.1126/scitranslmed.3006070. Ngo B , Kim E , Osorio-Vasquez V , Doll S , Bustraan S , Liang RJ , Luengo A , Davidson SM , Ali A , Ferraro GB , (2020). Limited Environmental Serine and Glycine Confer Brain Metastasis Sensitivity to PHGDH Inhibition. Cancer discovery 10 , 1352–1373. 10.1158/2159-8290.CD-19-1228.32571778 Ni M , Solmonson A , Pan C , Yang C , Li D , Notzon A , Cai L , Guevara G , Zacharias LG , Faubert B , (2019). Functional Assessment of Lipoyltransferase-1 Deficiency in Cells, Mice, and Humans. Cell reports 27 , 1376–1386 e1376. 10.1016/j.celrep.2019.04.005.31042466 Parent E , and McConathy J (2018). Overview of Positron-Emission Tomography Tracers for Metabolic Imaging. In Imaging and Metabolism, Lewis JS , and Keshari KR , eds. (Springer International Publishing), pp. 47–80. 10.1007/978-3-319-61401-4_3. Parsons DW , Jones S , Zhang X , Lin JC , Leary RJ , Angenendt P , Mankoo P , Carter H , Siu IM , Gallia GL , (2008). An integrated genomic analysis of human glioblastoma multiforme. Science 321 , 1807–1812. 10.1126/science.1164382.18772396 Pedley AM , Pareek V , and Benkovic SJ (2022). The Purinosome: A Case Study for a Mammalian Metabolon. Annu Rev Biochem. 10.1146/annurev-biochem-032620-105728. Ponde DE , Dence CS , Oyama N , Kim J , Tai YC , Laforest R , Siegel BA , and Welch MJ (2007). 18F-fluoroacetate: a potential acetate analog for prostate tumor imaging--in vivo evaluation of 18F-fluoroacetate versus 11C-acetate. Journal of nuclear medicine : official publication, Society of Nuclear Medicine 48 , 420–428.17332620 Possemato R , Marks KM , Shaul YD , Pacold ME , Kim D , Birsoy K , Sethumadhavan S , Woo HK , Jang HG , Jha AK , (2011). Functional genomics reveal that the serine synthesis pathway is essential in breast cancer. Nature 476 , 346–350. 10.1038/nature10350.21760589 Rappez L , Stadler M , Triana S , Gathungu RM , Ovchinnikova K , Phapale P , Heikenwalder M , and Alexandrov T (2021). SpaceM reveals metabolic states of single cells. Nature methods 18 , 799–805. 10.1038/s41592-021-01198-0.34226721 Ray GJ , Boydston EA , Shortt E , Wyant GA , Lourido S , Chen WW , and Sabatini DM (2020). A PEROXO-Tag Enables Rapid Isolation of Peroxisomes from Human Cells. iScience 23 , 101109. 10.1016/j.isci.2020.101109.32417403 Reczek CR , Birsoy K , Kong H , Martinez-Reyes I , Wang T , Gao P , Sabatini DM , and Chandel NS (2017). A CRISPR screen identifies a pathway required for paraquat-induced cell death. Nature chemical biology 13 , 1274–1279. 10.1038/nchembio.2499.29058724 Rich LJ , Bagga P , Wilson NE , Schnall MD , Detre JA , Haris M , and Reddy R (2020). (1)H magnetic resonance spectroscopy of (2)H-to-(1)H exchange quantifies the dynamics of cellular metabolism in vivo. Nat Biomed Eng 4 , 335–342. 10.1038/s41551-019-0499-8.31988460 Rodrigues TB , Serrao EM , Kennedy BW , Hu DE , Kettunen MI , and Brindle KM (2014). Magnetic resonance imaging of tumor glycolysis using hyperpolarized 13C-labeled glucose. Nature medicine 20 , 93–97. 10.1038/nm.3416. Rohle D , Popovici-Muller J , Palaskas N , Turcan S , Grommes C , Campos C , Tsoi J , Clark O , Oldrini B , Komisopoulou E , (2013). An inhibitor of mutant IDH1 delays growth and promotes differentiation of glioma cells. Science 340 , 626–630. 10.1126/science.1236062.23558169 Romeo S , Kozlitina J , Xing C , Pertsemlidis A , Cox D , Pennacchio LA , Boerwinkle E , Cohen JC , and Hobbs HH (2008). Genetic variation in PNPLA3 confers susceptibility to nonalcoholic fatty liver disease. Nat Genet 40 , 1461–1465. 10.1038/ng.257.18820647 Ros S , Wright AJ , D'Santos P , Hu DE , Hesketh RL , Lubling Y , Georgopoulou D , Lerda G , Couturier DL , Razavi P , (2020). Metabolic Imaging Detects Resistance to PI3Kalpha Inhibition Mediated by Persistent FOXM1 Expression in ER(+) Breast Cancer. Cancer cell 38 , 516–533 e519. 10.1016/j.ccell.2020.08.016.32976773 Sabatine MS , Giugliano RP , Keech AC , Honarpour N , Wiviott SD , Murphy SA , Kuder JF , Wang H , Liu T , Wasserman SM , (2017). Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease. N Engl J Med 376 , 1713–1722. 10.1056/NEJMoa1615664.28304224 Salamanca-Cardona L , Shah H , Poot AJ , Correa FM , Di Gialleonardo V , Lui H , Miloushev VZ , Granlund KL , Tee SS , Cross JR , (2017). In Vivo Imaging of Glutamine Metabolism to the Oncometabolite 2-Hydroxyglutarate in IDH1/2 Mutant Tumors. Cell metabolism 26 , 830–841 e833. 10.1016/j.cmet.2017.10.001.29056515 Shoup TM , Olson J , Hoffman JM , Votaw J , Eshima D , Eshima L , Camp VM , Stabin M , Votaw D , and Goodman MM (1999). Synthesis and evaluation of [18F]1-amino-3- fluorocyclobutane-1-carboxylic acid to image brain tumors. Journal of nuclear medicine : official publication, Society of Nuclear Medicine 40 , 331–338.10025843 Suhre K , Shin SY , Petersen AK , Mohney RP , Meredith D , Wagele B , Altmaier E , CardioGram, Deloukas P , Erdmann J , (2011). Human metabolic individuality in biomedical and pharmaceutical research. Nature 477 , 54–60. 10.1038/nature10354.21886157 Sullivan LB , Gui DY , Hosios AM , Bush LN , Freinkman E , and Vander Heiden MG (2015). Supporting Aspartate Biosynthesis Is an Essential Function of Respiration in Proliferating Cells. Cell 162 , 552–563. 10.1016/j.cell.2015.07.017.26232225 Sun RC , Young LEA , Bruntz RC , Markussen KH , Zhou Z , Conroy LR , Hawkinson TR , Clarke HA , Stanback AE , Macedo JKA , (2021). Brain glycogen serves as a critical glucosamine cache required for protein glycosylation. Cell metabolism 33 , 1404–1417 e1409. 10.1016/j.cmet.2021.05.003.34043942 Surti S (2015). Update on time-of-flight PET imaging. Journal of nuclear medicine : official publication, Society of Nuclear Medicine 56 , 98–105. 10.2967/jnumed.114.145029.25525181 Tao R , Zhao Y , Chu H , Wang A , Zhu J , Chen X , Zou Y , Shi M , Liu R , Su N , (2017). Genetically encoded fluorescent sensors reveal dynamic regulation of NADPH metabolism. Nature methods 14 , 720–728. 10.1038/nmeth.4306.28581494 Threadgill DW , Miller DR , Churchill GA , and de Villena FP (2011). The collaborative cross: a recombinant inbred mouse population for the systems genetic era. ILAR J 52 , 24–31. 10.1093/ilar.52.1.24.21411855 Venneti S , Dunphy MP , Zhang H , Pitter KL , Zanzonico P , Campos C , Carlin SD , La Rocca G , Lyashchenko S , Ploessl K , (2015). Glutamine-based PET imaging facilitates enhanced metabolic evaluation of gliomas in vivo. Science translational medicine 7 , 274ra217. 10.1126/scitranslmed.aaa1009. Victor RG , Haley RW , Willett DL , Peshock RM , Vaeth PC , Leonard D , Basit M , Cooper RS , Iannacchione VG , Visscher WA , (2004). The Dallas Heart Study: a population-based probability sample for the multidisciplinary study of ethnic differences in cardiovascular health. Am J Cardiol 93 , 1473–1480. 10.1016/j.amjcard.2004.02.058.15194016 Viswanath V , Zhou R , Lee H , Li S , Cragin A , Doot RK , Mankoff DA , and Pantel AR (2021). Kinetic Modeling of (18)F-(2S,4R)4-Fluoroglutamine in Mouse Models of Breast Cancer to Estimate Glutamine Pool Size as an Indicator of Tumor Glutamine Metabolism. Journal of nuclear medicine : official publication, Society of Nuclear Medicine 62 , 1154–1162. 10.2967/jnumed.120.250977.33277391 Walsh AJ , Cook RS , Manning HC , Hicks DJ , Lafontant A , Arteaga CL , and Skala MC (2013). Optical metabolic imaging identifies glycolytic levels, subtypes, and early-treatment response in breast cancer. Cancer research 73 , 6164–6174. 10.1158/0008-5472.CAN-13-0527.24130112 Wang F , Travins J , DeLaBarre B , Penard-Lacronique V , Schalm S , Hansen E , Straley K , Kernytsky A , Liu W , Gliser C , (2013). Targeted inhibition of mutant IDH2 in leukemia cells induces cellular differentiation. Science 340 , 622–626. 10.1126/science.1234769.23558173 Wang L , Xing X , Zeng X , Jackson SR , TeSlaa T , Al-Dalahmah O , Samarah LZ , Goodwin K , Yang L , McReynolds MR , (2022). Spatially resolved isotope tracing reveals tissue metabolic activity. Nature methods 19 , 223–230. 10.1038/s41592-021-01378-y.35132243 Wang Y , Yen FS , Zhu XG , Timson RC , Weber R , Xing C , Liu Y , Allwein B , Luo H , Yeh HW , (2021). SLC25A39 is necessary for mitochondrial glutathione import in mammalian cells. Nature 599 , 136–140. 10.1038/s41586-021-04025-w.34707288 Wu D , Harrison DL , Szasz T , Yeh CF , Shentu TP , Meliton A , Huang RT , Zhou Z , Mutlu GM , Huang J , and Fang Y (2021). Single-cell metabolic imaging reveals a SLC2A3-dependent glycolytic burst in motile endothelial cells. Nat Metab 3 , 714–727. 10.1038/s42255-021-00390-y.34031595 Xiao Z , Dai Z , and Locasale JW (2019). Metabolic landscape of the tumor microenvironment at single cell resolution. Nat Commun 10 , 3763. 10.1038/s41467-019-11738-0.31434891 Zhang Z , TeSlaa T , Xu X , Zeng X , Yang L , Xing G , Tesz GJ , Clasquin MF , and Rabinowitz JD (2021). Serine catabolism generates liver NADPH and supports hepatic lipogenesis. Nat Metab 3 , 1608–1620. 10.1038/s42255-021-00487-4.34845393 Zhou R , Pantel AR , Li S , Lieberman BP , Ploessl K , Choi H , Blankemeyer E , Lee H , Kung HF , Mach RH , and Mankoff DA (2017). [(18)F](2S,4R)4-Fluoroglutamine PET Detects Glutamine Pool Size Changes in Triple-Negative Breast Cancer in Response to Glutaminase Inhibition. Cancer research 77 , 1476–1484. 10.1158/0008-5472.CAN-16-1945.28202527 Zhu XG , Chudnovskiy A , Baudrier L , Prizer B , Liu Y , Ostendorf BN , Yamaguchi N , Arab A , Tavora B , Timson R , (2021). Functional Genomics In Vivo Reveal Metabolic Dependencies of Pancreatic Cancer Cells. Cell Metab 33 , 211–221 e216. 10.1016/j.cmet.2020.10.017 33152324

PMC009xxxxxx/PMC9479679.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9805750 33377 European J Org Chem European J Org Chem European journal of organic chemistry 1434-193X 36120398 9479679 10.1002/ejoc.202200302 NIHMS1801578 Article Exploring Glycan Binding Specificity of Odorranalectin by Alanine Scanning Library Singh YashoNandini a Cudic Predrag a Cudic Maré a a Department of Chemistry and Biochemistry, Florida Atlantic University, 777 Glades Road, Boca Raton, Florida 33431, United States mcudic@fau.edu; pcudic@fau.edu 26 4 2022 27 7 2022 22 4 2022 27 7 2023 2022 28 e202200302This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Fluorescently labelled alanine scan analogues of odorranalectin (OL), a cyclic peptide that exhibits lectin like properties, were screened for binding BSA-conjugated monosaccharides using an enzyme-linked lectin assay (ELLA). Results revealed that Lys5, Phe7, Tyr9, Gly12, Leu14, and Thr17 were crucial for binding BSA-L-fucose, BSA-D-galactose and BSA-N-acetyl-D-galactosamine. Notably, Ala substitution of Ser3, Pro4, and Val13 resulted in higher binding affinities compared to the native OL. The obtained data also indicated that Arg8 plays an important role in differentiation of binding for BSA-L-fucose/D-galactose from BSA-N-acetyl-D-galactosamine. The thermodynamics of binding of the selected alanine analogues was evaluated by isothermal titration calorimetry. Low to moderate binding affinities were determined for the tetravalent MUC1 glycopeptide and asialofetuin, respectively, and high for the fucose rich polysaccharide, fucoidan. The thermodynamic profile of interactions with asialofetuin exhibits shift to an entropy-driven mechanism compared to the fucoidan, which displayed an enthalpyentropy compensation, typically associated with the carbohydratelectin recognition process. Roles of amino acids in odorranalectin sequence, a cyclic peptide with lectin-like properties, for the glycan binding was evaluated by alanine scanning approach. Results revealed residues crucial for binding L-fucose, D-galactose and N-acetyl-D-galactosamine (Lys5, Phe7, Tyr9, Gly12, Leu14, and Thr17), carbohydrate binding specificity (Phe7 and Arg8), and improved affinity compared to the native odorranalectin (Ser3, Pro4, and Val13). Alanine scanning Cyclic peptides Fluorescent probes Glycan binding Thermodynamics pmcIntroduction Aberrant glycosylation in cancer, involves changes in glycosylation patterns of cell surface and secreted glycoproteins.[1–3] Some of the changes include loss of or overexpression of certain glycans, increased expression of incomplete or truncated glycans, de novo expression of terminal sialylated glycans, and altered fucosylation.[4,5] For example, in epithelial carcinoma, a hallmark feature of mucins is their aberrant glycosylation, leading to the expression of truncated mucin-type O-glycans, with the simplest one being the Tn antigen (αGalNAc-Thr/Ser).[6,7] In hepatocellular carcinoma (HCC), fucosylated alpha-fetoprotein (AFP) is one of the most representative types of glycan-related cancer biomarkers, marked by increased levels of core-fucosylation (α1–6) and terminal fucosylation (α1–2), such as Lewis blood-group antigens (Lex/y and Lea/b).[8] These glycans are essential components of malignant and metastatic phenotype of many human cancers[2,3,5] that apart from being indicators of cancer progression, also mediate several processes involved in tumor cell proliferation and metastasis.[9,10] Therefore, targeting these glycans can prove to be a novel strategy for cancer prevention and therapeutics. Lectins are extensively being used for targeting, separating, and reliably identifying glycans that have been utilized as biomarkers for breast, prostate, and liver cancers.[11–13] Previous studies have shown their ability to selectively bind specific sugar moieties of targeted cells and drive drug delivery into or across cell barriers through receptor-mediated endocytosis.[14,15] However, most lectins have large molecular weights that hamper their targeting efficacy and often lead to high in vivo immunogenicity and potent red cell agglutination activity.[16] Odorranalectin (OL) is a small lectin-mimicking peptide (17 amino acid residues), isolated from the skin secretions of Odorrana grahami, that exhibits increased affinity towards L-fucose on the cell membrane and to a lesser degree towards D-galactose and N-acetyl-D-galactosamine (GalNAc), constituents of mucin-type tumor-associated glycans.[17,18] Studies have shown that OL has the potential for intranasal drug delivery to the brain, due to its recognition of fucosylated epitopes located on the olfactory epithelium of nasal mucosa.[19–21] In addition, OL displays low immunogenicity compared to other lectins, such as wheat germ agglutinin (WGA), often used to modify the surface of drug carriers for improving absorption on nasal mucosa.[22] Although OL is a promising candidate for targeted drug delivery, very little is known about the role of individual amino acid residues involved in the carbohydrate binding and recognition.[17,18] Among the various techniques, alanine (Ala) scanning has been frequently used to determine the structural and functional roles of amino acid residues in a protein or peptide of interest.[23–26] Herein, for the first time, we describe the synthesis of an Ala scan OL library, including the flexible tail. Ala scan analogues were analyzed for their secondary structure content using circular dichroism (CD) spectroscopy. Their binding affinities were assessed qualitatively against BSA-conjugated monosaccharides (L-fucose, D-galactose, and D-GalNAc) and mucin (MUC1) glycopeptides carrying the Tn antigen using a fluorescence-based binding assay. The Ala scan analogues that exhibited high binding affinity in comparison to OL were further quantitatively evaluated by screening against asialofetuin, MUC1-Tn glycopeptides, and fucoidan, using isothermal titration calorimetry (ITC), to obtain thermodynamic binding parameters and deconvolute their binding mechanisms. Results and Discussion Peptide Synthesis and Circular Dichroism (CD) Analysis of Odorranalectin (1) and its Ala Scan Analogues (2–17). The solid phase peptide synthesis (SPPS) of OL and fluorescently labelled OL peptide 1 containing a disulfide bridge has been previously described by our group.[17] The same strategy was applied in the current study to the synthesis of the Ala scan analogues of OL (Scheme 1, Table 1). Each amino acid position was substituted with Ala, starting from the C-terminus (Thr17Ala, 2) and ending at the N-terminus (Tyr1Ala, 14), except for Cys6 and Cys16, required for disulfide bridge. In addition to monosubstituted analogues 2-14, double (Ser3Ala/Val13Ala, 16 and Ser3Ala/Pro4Ala, 17) and triple (Ser3Ala/Pro4Ala/Val13Ala, 15) substituted Ala analogues were synthesized to determine if the binding effects of each individual positions could be amplified. A fluorescence tag, 5(6)-carboxyfluorescein (FAM), was incorporated at the N-terminus of all peptides. Prior to the addition of the FAM tag, a PEG linker (20 atoms) was coupled to avoid interference of the tag within the ligand’s binding interface and to improve solubility in aqueous buffers. Fluorescently labelled OL and its Ala scan analogues, (1-17)-FAM, were cleaved from the resin under standard acidic conditions and their purity was confirmed by their RP-HPLC elution profiles and MALDI-TOF MS analyses (Table 1 and see the Supporting Information, Pages S3–19). Depending on the amino acid substituted with Ala, there were slight differences in the RP-HPLC retention times (tR) of the Ala analogues compared to native OL 1 (Table 1). For example, upon substituting non-polar amino acids with Ala, such as Leu14 (3), Val13 (4), Pro10 (7), and Phe7 (10), the tR decreased (ΔtR = 0.8–0.5 min) compared to 1, with the largest difference observed for substituted Val13 (4) residue (ΔtR = 0.8 min). On contrary, substituting charged and polar amino acids with Ala, such as Asn11 (6), Arg8 (9), and Lys5 (11), the tR increased (ΔtR = 0.6–1.3 min) compared to 1, with the largest difference observed for substituting Arg8 (9) (ΔtR = 1.1 min) and Lys5 (11) (ΔtR = 1.3 min) residues, respectively. Hence, as expected, substituting non-polar residues with Ala, increased the overall hydrophilicity whereas substituting charged and polar amino acids with Ala, increased the overall hydrophobicity of the analogues compared to native OL 1. Circular dichroism (CD) spectroscopy was used to investigate the conformational characteristics of 1 and its Ala scan analogues, with and without the PEG linker and FAM tag, in 10 mM sodium phosphate buffer, pH 7.4 (see the Supporting Information, Pages S29–34). The CD spectrum of 1 is characterized by a minimum at 190 nm and a weak maximum at 230 nm, characteristic of an overall β-turn conformation, in agreement with our earlier reported conformational analysis of OL.[17] The Ala scan analogues of OL (2–17) exhibited very similar CD spectra to 1. In order to estimate the contribution of specific secondary conformational components, the measured spectra were analyzed using the Beta Structure Selection (BeStSel) web server.[27,28] The percentages of the secondary structural elements were 13–15% β-turn, 20–38% anti-parallel β-sheet, and 49–55% random coil, in agreement with data previously reported.[18] The amount of α-helix and parallel β-sheet was <10% (see the Supporting Information, Tables S1 and S2, Pages S30 and S33). The presence of the FAM tag and Ala substitution did not significantly alter the conformational ratio between β-turn, anti-parallel β-sheet, and random coil. Screening with BSA-Conjugated Monosaccharides and MUC1-Tn Glycopeptides Using a Fluorescence-Based Binding Assay. The fluorescein-labelled Ala scan analogues of OL, (2-17)-FAM, were screened against BSA-conjugated L-fucose, D-galactose and N-acetyl-D-galactosamine to identify specific side chain contributions to binding using enzyme-linked lectin assay (ELLA).[17,29,30] Briefly, microtiter plates were immobilized with BSA-conjugated monosaccharides, blocked with 3% BSA to minimize non-specific binding, and incubated with either fluorescein-labelled odorranalectin (1-FAM) or its Ala scan analogues (2-17)-FAM. A fucose-specific lectin, fluorescein-labelled Aleuria aurantia (AAL) and a N-acetyl-D-galactosamine-specific lectin, fluorescein-labelled Soybean agglutinin (SBA) or SBA-Alexa Fluor, were used as controls. After removal of excess lectin, wells were washed with PBS and the fluorescence intensities were measured at 517 nm. The assay was performed in three replicates for each concentration point in 96-well plates and the average intensities after background subtraction were plotted against BSA-conjugated monosaccharide concentration. The binding of fluorescein-labelled OL (1-FAM) showed a strong preference for L-fucose and D-galactose, and to a lesser degree, for the N-acetyl-D-galactosamine (Figure 1 and see the Supporting Information, Figures S1–S3, Page S35) as previously reported by us.[17] In comparison to 1-FAM, AAL and SBA lectins showed specificity and 3–4-fold higher affinity for L-fucose and N-acetyl-D-galactosamine, respectively (Figure 1 and see the Supporting Information, Figures S1 and S3, Page 35). Six Ala scan analogues, 14-FAM (Tyr1Ala), 11-FAM (Lys5Ala), 8-FAM (Tyr9Ala), 5-FAM (Gly12Ala), 3-FAM (Leu14Ala), and 2-FAM (Thr17Ala) showed a marked decrease in binding to all three monosaccharides tested. The most pronounced decrease in affinity (10-fold) was observed for Tyr1Ala (14-FAM) and Tyr9Ala (8-FAM) in comparison to 1-FAM. In the case of 8-FAM, the effect could be rationalized by loss of two hydrogen bonds formed by Tyr9, that appear to stabilize the structure of OL.[18] The observed large reduction in binding for 14-FAM analogue is less predictable. Our previously published docking studies highlighted the importance of Tyr9 and Arg8 residues in binding L-fucose, D-galactose, and D-galactosamine.[17] According to the NMR titration experiments, the Tyr1 residue is not a key residue for monosaccharide binding.[18] Thus, we hypothesize that the observed effect is non-specific and possibly due to the Tyr residue being directly linked to the pegylated fluorescent tag. The remaining four analogues, 11-, 5-, 3- and 2-FAM have either a charged (Lys5) and polar (Thr17) or hydrophobic (Phe7 and Leu14) side chains replaced by Ala. Lys5 and Thr17 residues are most likely to be directly involved in binding because of their ability for hydrogen bonding, whereas Phe7 and Leu14, have hydrophobic side chains that can also contribute to the stability of the binding complex through interactions with the hydrophobic aliphatic C-H groups of monosaccharides.[31] Side chains of Pro10 and Asn11 are less likely to be part of the functional binding epitope of OL as their Ala substitutions (7-FAM and 6-FAM) did not significantly impede binding to monosaccharides (less than 3-fold). In agreement with previous studies, Ala substitution at Phe7 (10-FAM) drastically lowered binding towards L-fucose,[17,18] however, binding towards D-galactose and N-acetyl-D-galactosamine was largely unaffected (Figure 1 and see the Supporting Information, Figures S1–3, Page S35). Ala substitution at position Arg8 (9-FAM) resulted in 5-fold decrease in binding affinity for L-fucose and D-galactose; however, there was no significant change in N-acetyl-D-galactosamine binding compared to the 1-FAM (Figure 1 and see the Supporting Information, Figures S1–3, Page S35). We hypothesize that decreased electron density on the nitrogen atom due to the presence of the acetamido group in N-acetyl-D-galactosamine, leads to a very weak or no hydrogen bonding with Arg8. Conversely, the presence of hydroxyl groups (-OH) in L-fucose and D-galactose facilitate hydrogen bonding with Arg8. Unexpectedly, three alanine analogues 13-FAM (Ser3Ala), 12-FAM (Pro4Ala), and 4-FAM (Val13Ala) exhibited higher binding (2.5-fold) toward all three monosaccharides (Figure 1 and see the Supporting Information, Page S35). 13-FAM and 12-FAM have the Ala substitution located in the flexible N-terminal part of OL, and 4-FAM has the Val13 substituted with Ala. Although Val13 and Gly12 participate in hydrogen bonding with Tyr9, it is apparent that only Gly12Ala substitution (5-FAM) negatively affects binding. Thus, glycine flexible nature allows conformational changes to occur in OL for effective binding.[32] Our results show that Ser3, Pro4, and Val13 are not essential for binding, and we hypothesize that the structural rearrangements brought by Ala substitution either increase the binding by stabilization (rigidification) of the native conformation of OL (folded state) and/or by destabilization of the unfolded state by decreasing its entropy.[33–35] Overall, binding assay with BSA-conjugated monosaccharides provided better insights into key amino acids positions involved in binding of OL to monosaccharides. Ala substitutions that disrupt binding indicate that residues such as Lys5, Phe7, Arg8, Leu14, and Thr17 directly interact with monosaccharides. Residues such as Tyr9 and Gly12 suggest to play a role in supporting OL conformation that is preorganized for more efficient carbohydrate binding. Ala scanning also revealed two amino acid side chains that were able to affect the carbohydrate binding specificity. For example, our results show that Phe7 is essential for binding L-fucose and to a much lesser degree D-galactose and N-acetyl-D-galactosamine, whereas Arg8 is essential for binding L-fucose and D-galactose and to a much lesser degree, to N-acetyl-D-galactosamine. Ala substitution in certain region of OL does not significantly alter monosaccharide binding. Analogs 7-FAM and 6-FAM that have Pro10 and Asn11 residues replaced with Ala shoved comparable affinities toward tested monosaccharides as 1-FAM. Lastly, substitution of Ser3, Pro4, and Val13, with Ala leads to increased binding. In contrast to their single Ala substitution, the triple substituted analogue at positions Ser3, Pro4, and Val13 (15-FAM), showed a significant decrease in binding (approximately 10-fold) compared to 1-FAM (Table 1, Figure 2, and see the Supporting Information, Page S36). Two combinations of double Ala substitution, Ser3Ala/Val13Ala (16-FAM) and Ser3Ala/Pro4Ala (17-FAM), were also tested (Table 1, Figure 2, and see the Supporting Information, Pages S37–38). These analogues showed slightly better binding than the triple substituted analogue 15-FAM, but it was still significantly lower in comparison to 1-FAM and analogues with a single Ala substitution (4-FAM, 12-FAM, and 13-FAM) (Figure 2). In general, changes in binding affinity observed for analogues with the single Ala substitution (4-FAM, 12-FAM, and 13-FAM) do not mirror the changes observed for analogues with the double or triple Ala substitution (15-FAM-17-FAM), suggesting that these effects are not additive. Apart from BSA-conjugated monosaccharides, two synthetic MUC1 model glycopeptides, 18 (HGVT*SAPDTRPAPGSTAPPA) and 19 (HGVS*APDT*RPAPGS*T*APPA), bearing the Tn antigen (T*/S* = Thr-/Ser-O-α-N-acetyl-D-galactosamine) were tested for binding with Ala scan analogues, 4-FAM, 10-FAM, 12-FAM, and 13-FAM, that showed comparable or higher binding than the native 1-FAM for BSA-N-acetyl-D-galactosamine (Figure 3 and see the Supporting Information, Pages S39). The fluorescence-based binding assay protocol was used as described earlier. SBA lectin was used as a control. The average fluorescence intensities were obtained after background subtraction of wells containing lectin/lectinomimics and non-glycosylated MUC1 peptide 20 (HGVTSAPDTRPAPGSTAPPA). The glycopeptides were plated at two concentrations (100 and 250 μg/mL) (see the Supporting Information, Pages S39). In comparison to BSA-fucose and BSA-galactose, native 1-FAM displayed very weak binding to 18, similar to BSA-N-acetyl-D-galactosamine. The other Ala scan analogues showed comparable (4-FAM) or higher binding (10-FAM, 12-FAM, and 13-FAM) affinities than 1-FAM, further confirming that residues Val13, Phe7, Pro4, and Ser3 are not essential for binding of N-acetyl-D-galactosamine. As the number of glycosylation sites increased, the binding of the selected analogues to tetrasubstituted glycopeptide 19 increased compared to 18 (Figure 3). This effect was more pronounced for 13-FAM, where binding increased almost 9-fold compared to 1-FAM. Our results clearly show that Ala substitution of Ser3 leads to a significant increase in binding to BSA-conjugated monosaccharides and MUC1 glycopeptides. In contrast, Ala substitutions of Pro4 (12-FAM), Phe7 (10-FAM), and Val13 (4-FAM) are less crucial for carbohydrate binding. Binding Analysis of Select Alanine Scan Analogues with ASF, MUC1-Tn (Glyco)peptides, and Fucoidan by Isothermal Titration Calorimetry (ITC). Thermodynamics of binding interactions of OL 1 and OL-derived lectinomimics (4, 9, 10, 12, and 13) was assessed using three model ligands by ITC. Asialofetuin (ASF) was used as a model glycoprotein because of its well-characterized N- and O-linked oligosaccharide structures with the terminal D-galactose.[36] Tetravalent MUC1 glycopeptide bearing N-acetyl-D-galactosamine displays truncated, cancer specific Tn antigen, and fucoidan was chosen to further examine the fucose-selectivity. OL and Ala scan analogues were used without the PEG linker and fluorescent label, respectively (see the Supporting Information, Pages S17–21). The thermodynamic binding parameters for ASF are summarized in Table 2. OL 1 bound to ASF in μM range [Kd = 148 μM, Table 2 and Figure 4(a)], in agreement with our previously reported data.[17] Compared to OL 1, the dissociation constant (Kd) for ASF binding to Pro4Ala (12) and Ser13Ala (13) decreased by ~2-fold, and to Val13Ala (4) by ~1.5 fold, respectively. The Phe7Ala (10) analogue showed slightly weaker binding than OL 1, while the complete loss of affinity was obtained for Arg8Ala (9) analogue [Table 2, Figure 4, see the Supporting Information, Figure S4(f) for Arg8Ala (9), Page S41]. These results are in full agreement with the screening data with BSA-conjugated monosaccharides, that identified Arg8 as an important residue for binding L-fucose and D-galactose, Phe7 was unaffected, and the three residues that increased affinity for D-galactose were Pro4, Ser3, and to a lesser extent, Val13. Interestingly, the low enthalpic contributions to binding, ranging from -2.38 kcal/mol for 12 (Pro4Ala) to −1.78 kcal/mol for 13 (Ser3Ala), seems to be compensated by a gain in the entropic contribution [Table 2 and Figure 4(d) and 4(e)]. This effect can be due to the substitution of rigid Pro4 with more flexible Ala residue in case of 12, and the loss of hydrogen bonding with the solvent by replacement of Ser3 with Ala in 13.[37] Moreover, solvation effects, such as solvent re-organization, or the release of tightly bound water upon ligand binding can contribute significantly to the entropic term.[38,39] The obtained binding stoichiometries (n-value) were reflective of titrations reaching full saturation, revealing the functional valency of ASF defined as the inverse of the obtained n value (1/n).[40] A weak binding interaction was observed for OL 1 and analogue 13 (Ser3Ala) with tetravalent MUC1 glycopeptide 19 (see the Supporting Information Figure S5(a) and (c) and Table S3, Page S42]. The binding affinities could not be accurately determined because saturation points were not reached regardless of the OL 1 present in the cell or the syringe during the titration. There was no binding interaction detected by ITC for either OL 1 or the Ala scan analogs (4, 9, 10, 12 and 13) with mono-glycosylated 18 or non-glycosylated 20 MUC1 peptides, respectively (data not shown). We can speculate that the low valency of MUC1-Tn glycopeptides, mono- (18) or even tetra-valent presentation of Tn glycan (19) may have played the role. Interestingly, even though 9 did not show any significant binding to ASF, it displayed weak binding to MUC1 glycopeptide 19 [see the Supporting Information, Figure S5(b) and Table S3, Page S42], suggesting Arg8 being slightly more selective for binding D-galactose than N-acetyl-D-galactosamine. To further examine fucose-selectivity, the Ala-scanning analogs of OL were also screened against fucoidan, a sulfated polysaccharide isolated from sea alga, as a model system containing terminal fucose.[41,42] Fucoidan contains 34–44% of sulfated fucose and the rest is a mixture of small proportions of galactose, mannose, xylose and uronic acids, along with acetyl groups and proteins.[43,44] We have compared binding of AAL lectin to BSA-conjugated fucose and fucoidan using a fluorescence-based binding assay. A 5-fold increase in binding towards fucoidan compared to BSA-conjugated fucose was observed (see the Supporting Information, Page S40). Furthermore, fucoidan showed similar binding trends to BSA-fucose in the fluorescence-based binding assay with the Ala scan analogues (see the Supporting Information, Pages S35 and S40). Select Ala scan analogues (4, 12, and 13) that showed higher binding than OL (1) with fucoidan in the fluorescence-binding assay were further analyzed for their thermodynamic binding parameters using ITC (Table 3). Titration experiments of 1, 4, 12, and 13 revealed binding by fucoidan in the low μM range. Compared to 1, the Kd for fucoidan to Ser3Ala (13) decreased by ~2-fold (Kd = 0.31 μM) while no significant difference was observed for Val13Ala (4) and Pro4Ala (12) (Figure 5). The enthalpy (DH) of the binding interaction between 1, 4, and 12 and fucoidan ranged from -21.7 to -20.8 kcal/mol whereas the entropy (-TΔS) ranged from 3.9 to 12.9 kcal/mol. Compared to the thermodynamic parameters obtained with ASF, interactions with fucoidan displayed a shift towards an enthalpy-driven process, which is typical for interactions between lectins and carbohydrate ligands.[45,46] Interestingly, 13 displayed a similar trend in DH and -TΔS values as observed with ASF where a stronger binding interaction was accompanied by a decrease in the favorable ΔH value compensated by a gain in the entropic contribution. The overall free energy (ΔG) for all analogues remained in the range from -8.34 to -8.88 kcal/mol. Conclusion In summary, Ala scan approach was applied to determine the importance of each amino acid residue in the 17-mer sequence of OL for binding carbohydrates of biological relevance. The ability of Ala-scan analogues to bind carbohydrates was qualitatively analyzed using a fluorescence-based binding assay that utilized BSA-conjugated monosaccharides (L-fucose, D-galactose, and N-acetyl-D-galactosamine), known to bind native OL.[17,18] As expected, the most disruptive Ala substitution that depleted binding were Lys5Ala (11), Tyr9Ala (8), Gly12Ala (5), and Thr17Ala (1), indicating the ligand-binding interface of OL. Arg8 (9) was critical for binding L-fucose and D-galactose but showed weaker specificity for N-acetyl-D-galactosamine, and Phe7 was critical for binding L-fucose, while its Ala substitution analogue (10) had no effect on binding D-galactose and N-acetyl-D-galactosamine compared to 1. Substitution of Leu14 by Ala (3) clearly shows that Leu hydrophobic side-chain is essential for binding all monosaccharides. Hence, Ala substitutions either disrupted the direct interaction with monosaccharides (Lys5Ala 11, Phe7Ala 10, Arg8Ala 9, and Leu14Ala 3) or disrupted a conformation that was necessary for glycan-binding (Tyr9Ala 8, Gly12Ala 5, and Thr17Ala 1). Ala substitutions at Pro10 (7) and Asn11 (6) did not significantly alter binding compared to the native OL. On contrary, Ala analogues of Ser3 (13), Pro4 (12), and Val13 (4), consistently increased binding to all monosaccharides compared to the native OL (1), suggesting that these residues are not important for binding and substitution with other amino acid combinations may further improve the carbohydrate-binding specificity of OL. ITC titration of Ser3Ala (13), Pro4Ala (12), Phe7Ala ( 10), Val13Ala (4), and OL (1) with ASF and MUC1-Tn 19 showed binding in mid to high μM range, respectively, and with fucoidan in low μM range. The binding interactions with ASF were entropy driven, possibly resulting from the conformational or solvent entropy contributions.[47,48] In contrast, the interactions with fucoidan were enthalpy driven. Binding of the small ligands is usually penalized by entropy due to the rigidity of the interacting moieties, as well as some of their torsional degrees of freedom.[49,50] For the fucoidan, this entropic penalty is not compensated by the release of water molecules to the bulk. Therefore, the process is enthalpydriven, typically observed for sugar–lectin interactions. Our findings demonstrate the feasibility of designing novel OL-based molecular probes to study the specific glycosylation changes occurring with the progression of a variety of diseases, including cancer and inflammation. Experimental Section Reagents. TentaGel XV RAM resin was obtained from Rapp Polymer (Tuebingen, Germany). Fmoc-protected amino acids, 5(6)-carboxyfluorescein (5,6-FAM) and coupling reagents, 1-hydroxybenzotriazole (HOBt), 2-(6-chloro-1H-benzotriazol-1-yl)-1,1,3,3-tetramethylaminium hexafluorophosphate (HCTU), and benzotriazole-1-yloxytripyrrolidinophosphonium hexafluorophosphate (PYBOP), for peptide synthesis were purchased from Chem-Impex (Wood Dale, IL). N, N’-Diisopropylethylamine (DIPEA) and N, N’-Diisopropylcarbodiimide (DIC) were purchased from Acros Organics (Thermo Fisher Scientific, Waltham, MA). Fmoc-NH-(PEG)2-COOH (20 atoms) was purchased from EMD Millipore (Billerica, MA). Trifluoroacetic acid (TFA), thioanisole, and HPLC grade solvents (DCM, DMF, acetonitrile, and water) were purchased from Fisher Scientific (Atlanta, GA) or Sigma-Aldrich (St. Louis, MO). Asialofetuin (ASF) and iodine were purchased from Sigma-Aldrich (St. Louis, MO). Fucoidan was purchased from Toronto Research Chemicals (Ontario, Canada). Fluorescein-labelled lectins (AAL and SBA), SBA-Alexa Fluor (488 conjugate), and BSA-conjugated monosaccharides (L-fucose, D-galactose, and N-acetyl-D-galactosamine) were purchased from Vector Labs (Burlingame, CA). The monosaccharides were modified by vinyl sulfone at the anomeric carbon and subsequently conjugated to BSA.[51,52] Synthesis of Cyclic Ala scan Analogues. The linear peptidyl-resin precursors for naturally occurring OL peptide 1 and its Ala scan analogues (2-17) were synthesized using TentaGel XV RAM resin (substitution 0.27 mmol/g, 0.1 mmol scale) and standard Fmoc-SPPS on a PS3 automated peptide synthesizer (Protein Technologies Inc., Tucson, AZ).[17] The amino acid couplings were performed using 4-fold excess of amino acids, HOBt, and HCTU, in the presence of 0.4 M N-methylmorpholine (NMM) in DMF. Fmoc deprotection was carried out using 20% piperidine in DMF. Cyclization on the resin via disulfide bridge was achieved using iodine (10 eq) and 2% anisole in CH2Cl2 for 1 hour.[17] The N-terminal Fmoc was deprotected and peptides were cleaved from the resin using a TFA/thioanisole/water mixture in 95:2.5:2.5 ratio for 3 hours. The cleaved crude peptides were then precipitated with cold methyl-tert-butyl-ether (MTBE). Synthesis of Fluorescent-labelled Cyclic Ala scan Analogues. Fluorescent-labelled peptides were synthesized using standard Fmoc-SPPS approach as mentioned above. Following cyclization and Fmoc deprotection of the N-terminal residue, Fmoc-NH(PEG)2-COOH was coupled to peptidyl-resin using HOBt and PyBOP (2 eq) for 16 h. In the next step, the Fmoc protecting group was removed using standard protocol followed by coupling of 5(6)-carboxyfluorescein (5,6-FAM) using DIC and HOBt (10 eq) in DMF for 16 h.[17,53] The fluorescent-labelled peptides were cleaved from the resin using a TFA/thioanisole/water mixture in 95:2.5:2.5 ratio for 3 hours. The cleaved crude peptides were then precipitated with cold MTBE. Purification and Characterization of Cyclic Ala Scan Analogues. Peptide purification and analysis was performed on an Agilent 1260 Infinity system. The analytical RP-HPLC uses a Phenomenex Aeris Peptide C18 column (150 × 4.6 mm, 3.6 μm, 100Å) at 1 mL/min or 0.8 mL/min flow rate, with 0.1% TFA in water (A) and 0.1%TFA in acetonitrile (B) as the eluents. The elution gradient for analytical RP-HPLC purification was 0 to 60% B over 30 min. The preparative RP-HPLC uses the Grace Vydac monomeric C18 column (250 × 22mm, 15–20μm, 300Å) at 10 mL/min flow rate, with 0.1% TFA in water (A) and 0.1%TFA in acetonitrile (B) as the eluents. The elution gradients for preparative RP-HPLC purification was 0 to 50% B over 110 min. The peptide analogues were detected at 214 nm by the UV-Vis detector (Agilent 1260 Infinity DAD). Purified peptides were characterized by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) by Bruker™ Microflex using α-cyano-4-hydroxycinnamic acid as matrix. Fluorescence-based Binding Assay. The ligand binding assays, enzyme-linked lectin assay (ELLA) and Enzyme-linked immunosorbent assay (ELISA) have been described by our group previously.[17,30] Briefly, the assay was performed in black 96-well immune plates (Thermo Scientific™ 437111) coated with either 1 mg/mL of BSA-conjugated monosaccharide (L-fucose, D-galactose, or N-acetyl-D-galactosamine) or fucoidan in phosphate-buffered saline (PBS) (0.01 M, pH 7.4). MUC1-Tn (glyco)peptides were coated at two concentrations (100 and 250 μg/mL). Each well contained 50 μL and 3 replicates of each ligand. The plate was incubated at 37°C overnight to dry, and then blocked using 3% bovine serum albumin (BSA) in PBS (300 μL/well) overnight to account for any non-specific interactions with the immobilized glycoprotein or (glyco)peptide. The wells were then incubated for 2 h with 60 μg/mL of AAL-FITC or SBA-Alexa Fluor or SBA-Fluorescein or (1-17)-FAM at room temperature in the dark, with gentle mixing on a 3D shaker. After incubation, the wells were washed with PBS buffer (4x), the fluorescence signal was measured at 517 nm using Cytation5 (Biotek), and the mean relative fluorescence unit (RFU) after background subtraction for each lectin or lectinomimic was plotted against BSA-conjugated monosaccharide, fucoidan or MUC1-Tn (glyco)peptide ligands. Isothermal Titration Calorimetry (ITC) Experiments. The thermodynamic parameters were determined using a MicroCal PEAQ-ITC calorimeter (Malvern Panalytical) at 25°C in 20 mM HEPES at pH 7.0. MUC1 glycopeptides and ASF were dialyzed and prepared in the same buffer. At least 18 consecutive injections of 2 μL were applied every 180 s intervals at a constant stir speed of 500 rpm. For binding with ASF, the reaction cell contained the ASF solution (200–210 μM, 300 μL) and it was titrated with cyclic peptide solution at a concentration 10-fold greater (2500 μM). For binding with MUC1-Tn glycopeptides, the reaction cell contained the cyclic peptide solution (180–280 μM, 300 μL) and it was titrated with MUC1-Tn (glyco)peptide solutions at a concentration 2-fold greater (315–500 μM). For binding with fucoidan, the reaction cell contained the cyclic peptide solution (140 μM, 300 μL) and it was titrated with fucoidan at a concentration 4-fold greater (500 μM). The concentration of ASF was confirmed by measurements using Epoch (Biotek) microplate spectrophotometer and that of peptides was determined using analytical RP-HPLC. The raw integrated heat plots were analyzed using the MicroCal PEAQ-ITC software (Malvern) under the one-set-of-sites model and control parameter as fitted offset was applied to each titration as per the company’s guidelines and previous applications.[39,54–56] In all cases, thermodynamic parameters were derived from at least two independent runs that were averaged. Thermograms with the exact concentration of ligand and glycoprotein for each run, integrated heat values, and signature plots of binding interactions are provided in the Supporting Information, Pages S41–43. Circular Dichroism (CD) Analysis. The CD spectra were recorded in phosphate buffer (0.01 M, pH 7.4) on a JASCO J-810 spectropolarimeter (Jasco, Easton, MD). A quartz cell of 1 mm optical path length was used, and the spectra were measured over a wavelength range of 180–250 nm with scanning speed of 100 nm/min, and a response time of 4s, at 25°C. The concentration of the peptides was 0.2 mg/mL and confirmed using RP-HPLC. All spectra were baseline-corrected to account for the signal contribution from buffer, and then converted into molar ellipticity (deg cm2 dmol-1).[57] The percentages of secondary structures were calculated for all spectra using the BeStSel method (see the Supporting Information, Pages S29–34).[27] Supplementary Material supinfo Acknowledgements This research was supported by start-up funds (FAU) to M.C., and National Institute of Health (NIH) Grant R15CA242351 to M.C. Figure 1. (a) Screening of Ala scan analogues (2-14-FAM), OL (1-FAM), and control lectins, AAL and SBA (60 μg/mL) against BSA-conjugated monosaccharides (1 mg/mL). (b) Adjusted y-scale to display low binding analogues. Fluorescence signal (RFU) at 517 nm is the average of triplicate measurements for each BSA-conjugated monosaccharide after background subtraction, which is stacked for each lectin/lectinomimic. The amino acid positions shown were substituted with alanine for each analogue. Figure 2. (a) Screening of Ala scan analogues 4-, 9-, 10-, 13-, 15-, and 16-17-FAM, OL (1-FAM), and control lectins, AAL and SBA (60 μg/mL) against BSA-conjugated monosaccharides (1 mg/mL). (b) Adjusted y-scale to display low binding analogues. Fluorescence signal (RFU) at 517 nm is the average of triplicate measurements for each BSA-conjugated monosaccharide after background subtraction, which is stacked for each lectin/lectinomimic. The amino acid positions shown were substituted with alanine for each analogue. Figure 3. Binding profile of mono- and tetrasubstituted MUC1 glycopeptides, 18 and 19, respectively (250 μg/mL), with select alanine scanning analogues (4-FAM, 10-FAM, 12-FAM, and 13-FAM), OL (1-FAM) and control lectin SBA (60 μg/mL) using fluorescence-based binding assay. Fluorescence signal (RFU) at 517 nm is the average of triplicate measurements for each lectin/lectinomimic after background subtraction, including subtraction of the binding between lectins and non-glycosylated MUC1 peptide 20. Figure 4. ITC titration profile of (a) ASF (205 μM) with OL, 1 (2.5 mM), (b) ASF (210 μM) with Val13Ala, 4 (2.5 mM), (c) ASF (204 μM) with Phe7Ala, 10 (2.5 mM), (d) ASF (208 μM) with Pro4Ala, 12 (2.5 mM), and (e) ASF (210 μM) with Ser3Ala, 13 (1.67 mM) in buffer containing 20 mM HEPES (pH 7.0). Injections of ligand were performed every 180 s at 298 K. The resulting values for stoichiometry (n), binding affinity (Ka), dissociation constant (Kd), enthalpy (ΔH), and change in entropy with respect to temperature (-TΔS) are obtained using the one-set-of-sites model in the MicroCal PEAQ-ITC analysis software as shown in Table 2. Figure 5. ITC titration profile of fucoidan (500 μM) with (a) OL, 1 (140 μM), (b) Val13Ala, 4 (140 μM), (c) Pro4Ala, 12 (140 μM), and (d) Ser3Ala, 13 (140 μM) in buffer containing 20 mM HEPES (pH 7.0). Injections of ligand were performed every 180 s at 298 K. The resulting values for stoichiometry (n), binding affinity (Ka), dissociation constant (Kd), enthalpy (ΔH), and change in entropy with respect to temperature (-TΔS) are obtained using the one-set-of-sites model in the MicroCal PEAQ-ITC analysis software as shown in Table 3. Scheme 1. Schematic presentation of OL Ala scanning library for screening assays with cancer-related glycans. Table 1. Characterization of fluorescein-labelled OL (1-FAM) and its Ala scan analogues (2–17)-FAM. Entry Amino acid sequence[a] RP-HPLC[b] MALDI-TOF MS (M+H)+ tR (min) Calculated (Da) Observed (Da) 1 YASPKC*FRYPNGVLAC*T 20.3 2563.12 2563.75 2 YASPKC*FRYPNGVLAC*A 20.0 2533.12 2531.04 3 YASPKC*FRYPNGVAAC*T 19.7 2523.13 2522.54 4 YASPKC*FRYPNGALAC*T 19.5 2535.87 2534.62 5 YASPKC*FRYPNAVLAC*T 20.6 2577.92 2575.61 6 YASPKC*FRYPAGVLAC*T 20.9 2520.90 2518.70 7 YASPKC*FRYANGVLAC*T 19.7 2541.89 2543.70 8 YASPKC*FRAPNGVLAC*T 20.0 2471.87 2471.73 9 YASPKC*FAYPNGVLAC*T 21.4 2478.84 2479.43 10 YASPKC*ARYPNGVLAC*T 19.8 2487.87 2489.55 11 YASPAC*FRYPNGVLAC*T 21.6 2506.84 2507.52 12 YASAKC*FRYPNGVLAC*T 20.4 2534.89 2532.47 13 YAAPKC*FRYPNGVLAC*T 20.1 2547.91 2545.58 14 AASPKC*FRYPNGVLAC*T 19.9 2471.87 2470.69 15 YAAAKC*FRYPNGALAC*T 20.0 2493.86 2496.83 16 YAAPKC*FRYPNGALAC*T 20.1 2519.87 2496.83 17 YAAAKC*FRYPNGVLAC*T 20.3 2521.89 2524.33 C* = disulfide bond between Cys6-Cys16. [a] N-terminus labelled with a PEG linker [Fmoc-NH(PEG)2-COOH, 20 atoms] and a fluorescence FAM tag [5(6)-carboxyfluorescein]. [b] HPLC analysis conditions can be found in the Supplementary Information, Pages S3–19. Table 2. Thermodynamic parameters for the binding of asialofetuin (ASF) to select alanine scan analogues (4, 9, 10, 12 and 13) and native OL (1) by ITC.[a] Ka (x104 M−1) −ΔG (kcal/mol) −ΔH (kcal/mol) −TΔS (kcal/mol) n Kd (μM) 1 0.675 5.23 3.58 −1.64 0.129 148 4 0.990 5.45 5.16 −0.295 0.135 101 9 No binding - - - - - 10 0.495 5.04 6.16 1.12 0.127 202 12 1.53 5.71 2.38 −3.33 0.146 65.5 13 1.63 5.75 1.78 −3.97 0.115 61.3 [a] The binding isotherms are presented in Figure 4. Errors in ΔH ranged between ±0.02 and 0.04 kcal mol−1 and between ±0.9 and 2.1 μM/experiment for Kd. The error values and exact concentrations of the ligand and lectin for each experiment are provided in the Supporting Information, Page S41 (Figure S4), along with the corresponding thermograms. Values obtained using the one-set-of-sites fitting model with MicroCal PEAQ-ITC analysis software. Table 3. Thermodynamic parameters for the binding of fucoidan to OL (1), Val13Ala (4), Pro12Ala (12), and Ser3Ala analogue (13) by ITC.[a] Ka (x104 M−1) −ΔG (kcal/mol) −ΔH (kcal/mol) −TΔS (kcal/mol) n Kd (μM) 1 135 8.37 21.2 12.9 0.135 0.74 4 128 8.34 21.7 13.3 0.135 0.78 12 227 8.68 20.8 12.1 0.135 0.44 13 321 8.88 12.8 3.9 0.135 0.31 [a] The binding isotherms are presented in Figure 5. Errors in ΔH ranged between ±0.02 and 0.09 kcal mol−1 and between ±0.005 and 0.02 μM/experiment for Kd. The error values and concentrations of the ligand and lectin for each experiment are provided in the Supporting Information, Page S43 (Figure S6), along with the corresponding thermograms. Values obtained using the one-set-of-sites fitting model with MicroCal PEAQ-ITC analysis software. [1] Munkley J , Elliott DJ , Oncotarget 2016, 7 , 35478–35489.27007155 [2] Stowell SR , Ju T , Cummings RD , Annu. Rev. of Pathol.: Mech. of Dis 2015, 10 , 473–510. [3] Magalhães A , Duarte HO , Reis CA , Cancer Cell 2017, 31 , 733–735.28609653 [4] Varki A , Kannagi R , Toole BP , Glycosylation Changes in Cancer, Cold Spring Harbor (NY): Cold Spring Harbor Laboratory Press, 2009. [5] Pinho SS , Reis CA , Nat. Rev. Cancer. 2015, 15 , 540–555.26289314 [6] Cazet A , Julien S , Bobowski M , Burchell J , Delannoy P , Breast Cancer Res. 2010, 12 , 204.20550729 [7] Beckwith DM , Cudic M , Sem. Immunol. 2020, 47 , 101389. [8] Miyoshi E , Moriwaki K , Terao N , Tan C-C , Terao M , Nakagawa T , Matsumoto H , Shinzaki S , Kamada Y , Biomolecules 2012, 2 , 34–45.24970126 [9] Häuselmann I , Borsig L , Front. Oncol. 2014, 4 , 1–15.24478982 [10] Dimitroff J , Proc. Natl. Acad. Sci. U. S. A 2019, 116 , 13729–13737.31213534 [11] Lastovickova M , Strouhalova D , Bobalova J , J. Proteome Res 2020, 19 , 1885–1899.32181666 [12] Totten SM , Adusumilli R , Kullolli M , Tanimoto C , Brooks JD , Mallick P , Pitteri SJ , Sci. Rep. 2018, 8 , 6509.29695737 [13] Liu T , Shang S , Li W , Qin X , Sun L , Zhang S , Liu Y , Front. Physiol. 2017, 8.28167916 [14] Bies C , Lehr C-M , Woodley JF , Adv. Drug Deliv. Rev 2004, 56 , 425–435.14969751 [15] Robinson MA , Charlton ST , Garnier P , Wang X , Davis SS , Perkins AC , Frier M , Duncan R , Savage TJ , Wyatt DA , Watson SA , Davis BG , Proc. Natl. Acad. Sci U. S. A. 2004, 101 , 14527–14532.15448212 [16] Gorelik E , in Lectin Methods and Protocols (Eds.: Rhodes JM , Milton JD ), Humana Press, Totowa, NJ, 1998, pp. 453–459. [17] Rodriguez MC , Yongye AB , Cudic M , Martinez Mayorga K , Liu E , Mueller BM , Ainsley J , Karabencheva-Christova T , Christov CZ , Cudic M , Cudic P , Amino Acids. 2017, 49 , 1867–1883.28894966 [18] Li J , Wu H , Hong J , Xu X , Yang H , Wu B , Wang Y , Zhu J , Lai R , Jiang X , Lin D , Prescott MC , Rees HH , PLOS ONE. 2008, 3 , e2381.18584053 [19] Sajja RK , Cudic P , Cucullo L , BMC Neurosci. 2019, 20 , 22.31068126 [20] Wen Z , Yan Z , He R , Pang Z , Guo L , Qian Y , Jiang X , Fang L , Drug Deliv. 2011, 18 , 555–561.21812752 [21] Wen Z , Yan Z , Hu K , Pang Z , Cheng X , Guo L , Zhang Q , Jiang X , Fang L , Lai R , J. Control. Release. 2011, 151 , 131–138.21362449 [22] Gao X , Wu B , Zhang Q , Chen J , Zhu J , Zhang W , Rong Z , Chen H , Jiang X , J. Control. Release. 2007, 121 , 156–167.17628165 [23] Madhankumar AB , Mintz A , Debinski W , J. Biol. Chem. 2002, 277 , 43194–43205.12189139 [24] Reimer A , Maffenbeier V , Dubey M , Sentchilo V , Tavares D , Gil MH , Beggah S , van der Meer JR , Sci. Rep. 2017, 7 , 8245.28811596 [25] Migoń D , Jaśkiewicz M , Neubauer D , Bauer M , Sikorska E , Kamysz E , Kamysz W , Probiotics Antimicrob. Proteins. 2019, 11 , 1042–1054.30569430 [26] Kavianinia I , Stubbing LA , Abbattista MR , Harris PWR , Smaill JB , v Patterson A , Brimble MA , Bioorg. Med. Chem. Lett. 2020, 30 , 127135.32229061 [27] Micsonai A , Wien F , Bulyáki É , Kun J , Moussong É , Lee YH , Goto Y , Réfrégiers M , Kardos J , Nucleic Acids Res. 2018, 46 , W315–W322.29893907 [28] Micsonai A , Wien F , Kernya L , Lee YH , Goto Y , Réfrégiers M , Kardos J , Proc. Natl. Acad. Sci. U. S. A. 2015, 112 , E3095–E3103.26038575 [29] Jakas A , Perc M , Suć J , Rodriguez MC , Cudic M , Cudic P , J. Carbohydr. Chem. 2016, 35 , 69–85. [30] Singh Y , Rodriguez Benavente MC , Al-Huniti MH , Beckwith D , Ayyalasomayajula R , Patino E , Miranda WS , Wade A , Cudic M , J. Org. Chem. 2020, 85 , 1434–1445.31799848 [31] Mason PE , Lerbret A , Saboungi M-L , Neilson GW , Dempsey CE , Brady JW , Proteins: Struct., Funct., Bioinf 2011, 79 , 2224–2232. [32] Digby HR , Roberts JA , Sutcliffe MJ , Evans RJ , J. Neurochem 2005, 95 , 1746–1754.16236030 [33] Zou J , Song B , Simmerling C , Raleigh D , J. Am. Chem. Soc. 2016, 138 , 15682–15689.27934019 [34] Bigman LS , Levy Y , Isr. J. Chem. 2020, 60 , 705–712. [35] Knotts TA IV , Rathore N , de Pablo JJ , Biophys. J. 2008, 94 , 4473–4483.18326646 [36] Green ED , Adelt G , Baenziger JU , Wilson S , van Halbeek H , J. Biol. Chem. 1988, 263 , 18253–18268.2461366 [37] Sharrow SD , Edmonds KA , Goodman MA , v Novotny M , Stone MJ , Protein Sci. 2005, 14 , 249–256.15608125 [38] Rodriguez MC , Yegorova S , Pitteloud JP , Chavaroche AE , André S , Ardá A , Minond D , Jiménez-Barbero J , Gabius HJ , Cudic M , Biochemistry 2015, 54 , 4462–4474.26129647 [39] Beckwith DM , FitzGerald FG , Rodriguez Benavente MC , Mercer ER , Ludwig AK , Michalak M , Kaltner H , Kopitz J , Gabius HJ , Cudic M , Biochemistry 2021, 60 , 547–558.33560106 [40] Dam TK , Talaga ML , Fan N , Brewer CF , in Meth. Enzymol. (Ed.: Feig AL ), Academic Press, 2016, pp. 71–95. [41] Chen A , Lan Y , Liu J , Zhang F , Zhang L , Li B , Zhao X , Intl. J. Biol. Macromol. 2017, 105 , 1421–1429. [42] Lin Z , Tan X , Zhang Y , Li F , Luo P , Liu H , Mar. Drugs. 2020, 18 , 376. [43] Black WAP , Sci J Food Agric. 1954, 5 , 445–448. [44] Mabeau S , Kloareg B , Joseleau JP , Phytochemistry. 1990, 29 , 2441–2445. [45] Ahmad N , Gabius H-J , Sabesan S , Oscarson S , Brewer CF , Glycobiology. 2004, 14 , 817–825.15148296 [46] Dam TK , Brewer CF , Chem. Rev. 2002, 102 , 387–430.11841248 [47] Caro JA , Harpole KW , Kasinath V , Lim J , Granja J , Valentine KG , Sharp KA , Wand AJ , Proc. Natl. Acad. Sci. USA 2017, 114 , 6563–6568.28584100 [48] Verteramo ML , Stenström O , Ignjatovic MM , Caldararu O , Olsson MA , Manzoni F , Leffler H , Oksanen E , Logan DT , Nilsson UJ , Ryde U , Akke M , J. Am. Chem. Soc 2019, 141 , 2012–2026.30618244 [49] Chan L , Morris GM , Hutchison GR , J. Chem. Theory Comput 2021, 17 , 2099–2106.33759518 [50] Chang CA , Chen W , Gilson MK , Proc. Natl. Acad. Sci. USA 2007, 104 , 1534–1539.17242351 [51] Santoyo-González F , Hernandez-Mateo F , Boutureira O , Carbohydrate Chemistry. CRC Press, 2015, p. 39. [52] Sarkar B , Jayaraman N , Front. Chem 2020, 8 , 2296–2646. [53] Fischer R , Mader O , Jung G , Brock R , Bioconjug. Chem. 2003, 14 , 653–660.12757391 [54] ITC Data Analysis in Origin. Tutorial Guide., MicroCal, 2004. [55] García Caballero G , Beckwith D , v Shilova N , Gabba A , Kutzner TJ , Ludwig AK , Manning JC , Kaltner H , Sinowatz F , Cudic M , v Bovin N , v Murphy P , Gabius HJ , Histochem. Cell Biol. 2020, 154 , 135–153.32335744 [56] Kutzner TJ , Gabba A , FitzGerald FG , v Shilova N , García Caballero G , Ludwig AK , Manning JC , Knospe C , Kaltner H , Sinowatz F , v Murphy P , Cudic M , v Bovin N , Gabius HJ , Glycobiology 2019, 29 , 593–607.31091305 [57] Sreerama N , Woody RW , Anal. Biochem. 2000, 287 , 252–260.11112271

PMC009xxxxxx/PMC9504827.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0410462 6011 Nature Nature Nature 0028-0836 1476-4687 35296864 9504827 10.1038/s41586-022-04518-2 NIHMS1825546 Article A single cell atlas of human and mouse white adipose tissue Emont Margo P. 12 Jacobs Christopher 12 Essene Adam L. 1 Pant Deepti 1 Tenen Danielle 12 Colleluori Georgia 3 Di Vincenzo Angelica 3 Jørgensen Anja M. 4 Dashti Hesam 2 Stefek Adam 2 McGonagle Elizabeth 2 Strobel Sophie 2 Laber Samantha 2 Agrawal Saaket 25 Westcott Gregory P. 1 Kar Amrita 12 Veregge Molly L. 1 Gulko Anton 1 Srinivasan Harini 12 Kramer Zachary 1 De Filippis Eleanna 1 Merkel Erin 1 Ducie Jennifer 6 Boyd Christopher G. 7 Gourash William 8 Courcoulas Anita 8 Lin Samuel J. 9 Lee Bernard T. 9 Morris Donald 9 Tobias Adam 9 Khera Amit V. 2514 Claussnitzer Melina 210 Pers Tune H. 4 Giordano Antonio 3 Ashenberg Orr 11 Regev Aviv 111213 Tsai Linus T. 1214 Rosen Evan D. 1214 1. Division of Endocrinology, Diabetes and Metabolism, Beth Israel Deaconess Medical Center, Boston, MA, USA 2. Broad Institute of MIT and Harvard, Cambridge, MA, USA 3. Department of Experimental and Clinical Medicine, Center of Obesity, Marche Polytechnic University, Ancona, Italy. 4. Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen, Copenhagen, Denmark 5. Center for Genomic Medicine, Department of Medicine, Massachusetts General Hospital, Boston, MA, USA 6. Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, Beth Israel Deaconess Medical Center, Boston, MA, USA 7. Department of Surgery, Beth Israel Deaconess Medical Center, Boston, MA, USA 8. Department of Surgery, University of Pittsburgh Medical Center, Pittsburgh, PA 9. Division of Plastic Surgery, Department of Surgery, Beth Israel Deaconess Medical Center, Boston, MA 10. Diabetes Unit and Center for Genomic Medicine, Massachusetts General Hospital, Boston, Massachusetts, 02114, USA. 11. Klarman Cell Observatory, Broad Institute of MIT and Harvard, Cambridge, MA, USA 12. Howard Hughes Medical Institute, Koch Institute of Integrative Cancer Research, Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, USA 13. Genentech, South San Francisco, CA, USA 14. Harvard Medical School, Boston, MA Correspondence and requests for materials should be addressed to Evan D. Rosen, MD PhD, Division of Endocrinology, Diabetes, and Metabolism, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, Boston, MA 02215, erosen@bidmc.harvard.edu AUTHOR CONTRIBUTIONS MPE, LTT, and EDR conceived of the project. MPE and EDR wrote the manuscript with assistance from LTT, CJ, OA, and AR. MPE, ALE, DP, DT, GC, ADV, AS, EM, SS, SL, GPW, MLV, and AGu performed experiments. GPW, AGu, ZK, JD, CGB, WG, AC, SJL, BTL, DM, and AT collected samples. MPE, CJ, AMJ, HD, SA, AK, and HS performed computational analysis. AVK, MC, THP, AGi, OA, and AR provided additional intellectual input. 3 8 2022 3 2022 16 3 2022 23 9 2022 603 7903 926933 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. White adipose tissue (WAT), once regarded as morphologically and functionally bland, is now recognized to be dynamic, plastic, heterogenous, and involved in a wide array of biological processes including energy homeostasis, glucose and lipid handling, blood pressure control, and host defense1. High fat feeding and other metabolic stressors cause dramatic changes in adipose morphology, physiology, and cellular composition1, and alterations in adiposity are associated with insulin resistance, dyslipidemia, and type 2 diabetes (T2D)2. Here, we provide detailed cellular atlases of human and murine subcutaneous and visceral white fat at single cell resolution across a range of body weight. We identify subpopulations of adipocytes, adipose stem and progenitor cells (ASPCs), vascular, and immune cells and demonstrate commonalities and differences across species and dietary conditions. We link specific cell types to increased risk of metabolic disease, and we provide an initial blueprint for a comprehensive set of interactions between individual cell types in the adipose niche in leanness and obesity. These data comprise an extensive resource for the exploration of genes, traits, and cell types in the function of WAT across species, depots, and nutritional conditions. pmcAn atlas of human white adipose tissue Mature adipocytes are too large and fragile to withstand traditional single cell approaches; as a result, several groups have focused on the non-adipocyte stromal-vascular fraction (SVF) of mouse3-6 and human7 adipose tissue. An alternative strategy involves single nucleus (sNuc) sequencing, which can capture adipocytes, and has been used to describe murine epididymal8,9 and human brown adipose tissue10. To compare these approaches in the context of human WAT, we pursued experiments on two cohorts of subjects. In the first, we collected subcutaneous WAT from 9 women, isolated single cells from the SVF using collagenase digestion, and then performed whole cell Drop-seq [hereafter referred to as single cell (sc)RNA-seq]. Because different depots have been differentially linked to metabolic disease11, for the second cohort we collected paired subcutaneous (SAT) and visceral (VAT) adipose tissue from 10 individuals, and SAT alone from three additional individuals (10 women, 3 men), and performed sNuc-seq (Figures 1a, b; Extended Data Table 1). Doublet and low-quality filtering left 166,149 total cells (28,465 single cells and 137,684 single nuclei). The data from both approaches were integrated, enabling the identification of the canonical cell types found in WAT, including adipocytes, ASPCs, vascular cells, and immune cells (Figures 1c, d; Supplementary Table 1). As expected, adipocytes were found only in the sNuc-seq dataset. The sNuc-seq data was also enriched for vascular cells and macrophages, likely because collagenase digestion did not fully dissociate these cell types. Mesothelial cells were not seen in the scRNA-seq dataset, which did not include visceral tissue. Some of the visceral samples included cells that appeared to be endometrial in origin (PRLR+), likely due to endometriosis. Overall proportions of adipocytes and ASPCs did not differ between depots, but depot clearly affects the distribution of cells within these populations (Supplementary Figure 1, Extended Data Figure 1a, b, Extended Data Table 2). In our limited cohort, we could not detect major effects of BMI on cell type proportions. To assess this finding at larger scale, we utilized our dataset as a reference to estimate cell type proportions in bulk-RNA sequencing data12 obtained from the SAT of 331 men in the METSIM cohort13. This deconvolution analysis found that the relative abundance of adipocytes in that cohort was negatively correlated with BMI, while ASPCs and myeloid cells were positively correlated (Extended Data Figure 1c). An atlas of mouse white adipose tissue Murine models are commonly used to study adipose tissue biology14. We thus sought to compare mouse and human WAT at the single cell level by performing sNuc-seq on inguinal (ING, corresponding to human SAT) and perigonadal [PG, epididymal (EPI) in males, periovarian (POV) in females, corresponding to human VAT] adipose tissue of mice fed either a chow or high fat diet for 13 weeks (Figure 2a, b). After doublet removal and quality filtering, we considered a total of 197,721 cells (106,469 from PG and 91,252 from ING), identifying all cell types observed in human WAT (Figure 2c, d; Supplementary Table 2) with the addition of distinct male and female epithelial populations (Dcdc2a+ and Erbb4+, respectively). The female population is largely found in ING samples and resembles mammary epithelial cells, while the male population is almost exclusively found in PG samples, and as noted by others9 may represent contaminants from the epididymis and other reproductive structures that are tightly apposed to fat15. In contrast to the human data, cell type abundance in mouse WAT are highly dependent on body weight with relatively little variation between depots (Figure 2c and Extended Data Figure 2a, b, Extended Data Table 2). The proportions of cell types in mouse adipose tissue after HFD were notably different between male and female mice, which might reflect a true sex difference, or may reflect higher weight gain in males (Extended Data Figure 2b). To compare across species, we used a reference mapping algorithm to assign each mouse cell to a human cluster and noted a high degree of overall similarity between annotated mouse clusters and mapped human clusters (Extended Data Figure 2c). Cells of human and mouse SVF Vascular Cells Subclustering of human vascular cells revealed expected cell types including blood endothelial clusters that represent arteriolar, stalk, and venular cells, as well as lymphatic endothelial cells (LECs), pericytes, and smooth muscle cells (SMCs) (Extended Data Figure 3a, b). Mouse vascular cells formed similar clusters (Extended Data Figure 3c, d). As expected, reference mapping demonstrated high similarity between human and mouse vascular subclusters (Extended Data Figure 3e) and proportions of subclusters were similar across species (Extended Data Figure 3f, g). Immune Cells Analysis of human immune cells from scRNA-seq and sNuc-seq samples again revealed expected cell types, including multiple subpopulations of monocytes, macrophages (CD14+), dendritic cells (DCs), B and T lymphocytes, and NK cells (CD96+), mast cells (CPA3+), and neutrophils (CSF3R+) (Extended Data Figure 4a, b). Monocyte subpopulations 1 and 2 resemble classical and non-classical monocytes and DC subpopulations 1 and 2 resemble previously reported CLEC9A+ and CD1C+ populations from blood, respectively16. Lymphocytes also resemble previously reported B cell, T Cell, and NK cell populations from human WAT, including CTLA4+ hTregs17. Examination of the mouse WAT immune compartment revealed most of the same cell types, although there were notable differences in the relative abundance of myeloid and lymphoid cells between species (Extended Data Figure 4c, d). Human WAT contains somewhat fewer T/NK cells than macrophages/monocytes (~30% vs. ~60% of recovered immune cells); this imbalance was greatly exaggerated in murine WAT (macrophages ~90% of recovered immune cells vs. 3% T/NK cells). Because a wealth of data supports a key role for macrophages/monocytes in adipose biology18,19, we separated these cell types from other immune cells in silico for subsequent analysis. Mouse clusters of non-monocytes/macrophages mapped relatively well to their human counterparts, with some mixing of T and NK populations (Extended Data Figure 4e). Macrophages and monocytes also mapped well to their general class, but this association often broke down when considering macrophage subpopulations (Extended Data Figure 4f). The proportion of immune cell populations was similar in human SAT and VAT, with a few exceptions (Extended Data Figure 5a, e). In mice, small depot-dependent differences were eclipsed by relatively huge shifts in response to diet in male mice (Extended Data Figure 5b, d, f). Most notably, HFD resulted in a massive increase in macrophage numbers, primarily in PG, consistent with a large body of prior data18,20 (Extended Data Figure 2b, 5f). Reductions in the proportion of most other immune cell types (e.g., NK cells, T and B lymphocytes, DCs, and neutrophils) are likely due to the large increase in macrophages, rather than to intrinsic loss of those specific cell types following HFD (Extended Data Figure 5b, d, f). Mast cells increase proportionally after HFD, as previously reported21. (Extended Data Figure 5f). Accumulation of adipose tissue macrophages in obesity has also been shown in human WAT, using a combination of histomorphometry and flow sorting19,22. Our data are in general support of this conclusion, though the magnitude of the effect is significantly less prominent than that seen in mouse WAT (Extended Data 1b, 5c, e, f). The largest change involves hMac3, which is induced in visceral fat with higher BMI (Extended Data 5c, e). Mesothelial cells Subclustering of mesothelial cells revealed three populations in both human VAT and mouse PG (Extended Data Figure 6a-d). When mouse mesothelial clusters were mapped to human clusters, cells were split between human clusters hMes1 and hMes2, with no cells mapping to hMes3 (Extended Data Figure 6e). The proportions of most mesothelial subpopulations did not vary with obesity or high fat diet, with the exception of hMes1 and hMes2, which were reduced and increased in higher BMIs, respectively. (Extended Data Figure 6f, g). ASPCs (see Supplementary Note 1) We identified six distinct subpopulations of human ASPCs in subclustered scRNA-seq and sNuc-seq samples, all of which express the common marker gene PDGFRA (Extended Data Figure 7a, b). Similarly, we noted six subpopulations in the mouse ASPC data, all of which were also Pdgfra+ and some of which correspond well with a particular human subpopulation (Extended Data Figure 7c-e). For example, mASPC2 and hASPC2 are both characterized by high expression of Aldh1a3/ALDH1A3, and strongly resemble previously identified early multipotent progenitor cells that reside in the reticular interstitium of the fat pad5. Similarly, mASPC4 and hASPC4 express Epha3/EPHA3 and likely represent the anti-adipogenic Areg population reported by Schwalie et. al.3. Seeking to better place our mouse ASPC data into the overall context of the published literature, we integrated our data with that reported by others3-6,9 and found that ASPC populations identified by individual studies were generally preserved after integration, suggesting robustness of these clusters. Specifically, the previously described Icam1+ committed progenitor, Dpp4+ early progenitor, and Cd142+ Areg populations23 cluster together across studies (Extended Data Figure 7f). Many human and mouse ASPC subclusters showed dependency on diet, depot, or both. hASPC1, hASPC4, and hASPC5 were more prevalent in SAT than VAT, with increases in SAT hASPC4 and hASPC5 proportion in subjects with higher BMI (Extended Data Figure 8a, c, e). Conversely, hASPC3 and hASPC6 were more prevalent in VAT. In male mice, early progenitor cells (mASPC2) were notably more abundant in ING than PG and mASPC5 and mASPC6 were more prevalent in EPI vs ING, although this varied with obesity (Extended Data Figure 8b, d, f). Many of these observations are consistent with previous findings. For example, HFD has been shown to increase adipogenesis specifically in PG in mice24,25. Our data indicates that pre-adipocyte subclusters like mASPC6 increase dramatically in response to HFD in PG only. The loss of early progenitors (mASPC2) in PG with HFD is consistent with conversion of these cells along the differentiative pathway, i.e., toward mASPC6 (Extended Data Figure 8b, d, f). Unique populations of human adipocytes White adipocytes are generally considered to be monotypic and essentially uniform in function, although some recent studies have begun to challenge this assumption8-10,26 The high resolution of our data enabled us to find that human white adipocytes cluster into seven subpopulations with distinct markers (Figure 3a-b). We noted strong depot-specific associations of adipocyte subtypes, with hAd1, hAd3, hAd4, and hAd7 localized primarily to SAT, while hAd2 and hAd6 were almost exclusively found in VAT. hAd5 represents a smaller population that is roughly equally distributed between SAT and VAT (Extended Data Figure 9a-c). We also noted a BMI-dependent shift in adipocyte subtype within both depots (Extended Data Figure 9b, c). Importantly, all adipocyte subpopulations are present in the majority of subjects, indicating that these subtype designations are generalizable and do not reflect sample-specific variation (Extended Data Figure 9c). Immunohistochemistry (IHC) and/or immunofluorescence of markers for hAd4, hAd5, hAd6, and hAd7 in human SAT or VAT identified specific subpopulations of adipocytes at proportions similar to those seen in the single cell data (Figure 3c and Extended Data Figure 9 d, e). To examine whether SAT subtype proportion was influenced by BMI in a larger dataset, we estimated individual subtype proportions by deconvolution analysis of bulk RNA-seq data from purified isolated subcutaneous human adipocytes from 43 women (Figure 3d). This analysis showed that clusters hAd4 and hAd7 trend to negative correlation with BMI, aligning with our IHC findings, while hAd5 proportion is positively correlated with BMI. Visceral adipocytes are absent from this dataset and so we were unable to assess the prevalence of hAd2 or hAd6 in this cohort, although IHC of hAd6 marker EBF2 also suggests its prevalence may be positively correlated with BMI (Figure 3c). A critical question is whether individual adipocyte subpopulations have specific functions. To assess this, we first looked at genes that participate in the major metabolic activities of adipocytes. All subpopulations expressed these genes, although their relative amounts differed. Thus, the adipokines adiponectin and adipsin (CFD) are most highly expressed in hAd3, and insulin signaling components like INSR, IRS1 and IRS2 are most highly expressed in hAd5 (Extended Data Figure 9f). We next looked more holistically at the data by performing pathway analysis for markers of each subpopulation (Supplementary Table 3, Extended Data Figure 9g-m). Subpopulations hAd1, which accounts for ~40% of SAT adipocyte nuclei, and hAd2, which accounts for ~60% of VAT adipocyte nuclei, have relatively few specific markers, and the pathways that emerged were similarly bland (Extended Data Figure 9g, h). These populations likely represent “basal” subcutaneous or visceral adipocytes, so we therefore focused on subpopulations hAd3-hAd7 for more detailed analysis. hAd3 was associated with “triglyceride biosynthesis” and included higher expression of DGAT2, SREBF1, and PNPLA3 (Extended Data Figure 9i). The hAd4 cluster expresses the highest amounts of several fatty acid desaturases, including ELOVL5 and FADS3 (Extended Data Figure 9j), which is particularly interesting in light of the insulin-sensitizing role of unsaturated lipokines such as palmitoleate27. hAd5 adipocytes, besides having the highest expression of several insulin signaling genes, were also characterized by expression of “sphingolipid signaling genes” (Extended Data Figure 9k). Both hAd3 and hAd4 exhibit high expression of lipogenic genes, while lipolytic gene expression is higher in hAd5 (Extended Data Figure 9f). We next asked whether cultured human adipocytes retain evidence of subpopulation diversity. To that end, we utilized 57 RNA-seq datasets from human subcutaneous and visceral adipocyte progenitors differentiated ex vivo over a 14 day timecourse28. Deconvolution analysis revealed that many subpopulations identified in vivo were retained in the dish. Furthermore, much of the previously noted depot selectivity was recapitulated, such that the visceral subpopulations hAd2 and hAd6 were significantly more likely to appear in cultured visceral cells and the subcutaneous subpopulation hAd4 was overrepresented in cultured subcutaneous cells (Extended Data Figure 10a). Furthermore, because these cultured samples were also subjected to high-content image-based profiling using LipocyteProfiler28, we were able to correlate individual subpopulations with image-based features representing morphological and cellular phenotypes including lipid and mitochondrial content. Thus, ex vivo differentiated adipocyte cultures predicted to have high amounts of hAd3, associated with high lipogenic gene expression and lower lipolytic gene expression, have more overall lipid and larger lipid droplets (Figure 3e, f). Conversely, ex vivo differentiated adipocyte cultures with high predicted hAd5 content have less overall lipid and smaller lipid droplets, consistent with a higher ratio of lipolytic to lipogenic gene expression (Extended Data Figure 10b-d). One particularly interesting adipocyte subpopulation is hAd6, which selectively expresses genes typically associated with thermogenesis, such as EBF2, ESRRG, and PPARGC1A (Extended Data Figure 9l), a surprising finding given that this population is almost exclusively visceral (Figure 3c, Extended Data Figure 9a, c). To better understand the relationship between this subpopulation and visceral adiposity, we looked further into the hAd6 marker EBF2, which has previously been identified as a pro-thermogenic transcription factor29. SNPs at the EBF2 locus are associated with waist-hip ratio (WHR)30, which could involve actions in either SAT or VAT. Interestingly, however, a recent study of GWAS loci associated with adiposity in specific depots31 found a common variant 15 kb upstream of EBF2 associated specifically with VAT (Extended Data Figure 11a). Further analysis revealed that the minor allele of this SNP (MAF = 0.23) was associated with VAT adjusted for BMI and height (VATadj: beta = 0.062 SD per allele, p = 1.0 x 10−12), but not abdominal subcutaneous (ASAT) or gluteofemoral (GFAT) depots (ASATadj: beta = −0.018 SD per allele, p = 0.03), GFATadj: beta = −0.020 SD per allele, p = 0.02, Extended Data Figure 11b). We additionally stratified individuals into either 0, 1, or 2 carriers of the minor allele and observed an additive trend (G/G median VATadj −0.10 SD, G/A median VATadj = −0.04 SD, A/A median VATadj 0.04 SD; Extended Data Figure 11c). Next, we returned to the visceral human adipocytes differentiated ex vivo, and found that samples predicted to have a higher proportion of hAd6 adipocytes were characterized by higher mitochondrial intensity and increased expression of mitochondrial and thermogenic genes (Extended Data Figure 11d-f). Finally, our analysis of hAd6 markers suggested other pathways associated with thermogenesis, including one for “axon guidance” (Extended Data Figure 11g). We could not measure innervation directly using our data, because the nuclei of innervating sympathetic neurons are located in the spinal ganglia and not the fat depot itself. Nonetheless, we estimated the degree of innervation using the presence of neuron-specific gene expression in the ambient RNA of our visceral sNuc-seq samples. Indeed, the amount of pan-neuronal markers like TUBB3 and UCHL132 strongly correlate with hAd6 proportion (Extended Data Figure 11h), further supporting a role for hAd6 as a novel visceral adipocyte subtype with thermogenic potential. Mouse adipocyte subpopulations Subclustering mouse adipocytes revealed six subpopulations (Figure 3g, h). Unlike human adipocytes, mouse adipocyte subtypes exhibit little depot enrichment (Extended Data Figure 12a-c). There was strong diet-dependency, however, as relative proportions of mAd1 and mAd3 were reduced after HFD, while the opposite was noted for mAd4 and mAd5 (Extended Data Figure 12b, c). In contrast to the relatively good cross-species concordance between immune cells, vascular cells, and ASPCs, mouse adipocytes do not map cleanly onto human adipocyte subpopulations (Extended Data Figure 12d-f). As in humans, genes associated with major adipocyte functions showed some subpopulation selectivity in mice. For example, lipogenesis genes were highest in HFD-induced population mAd5 (Extended Data Figure 12c, g). More detailed pathway analysis on mouse adipocyte subpopulations (Supplementary Table 3) showed that the chow-associated clusters mAd1-3 were notably enriched in metabolic pathways, particularly those involved in lipid handling (Extended Data Figure 12h-j). The HFD-associated clusters mAd4-6, on the other hand, were linked to pathways like “HIF-1 signaling”, “actin cytoskeleton”, and “NF-kB signaling” (Extended Data Figure 12k-n), consistent with the known roles of hypoxia, cytoskeletal remodeling, and inflammation in HFD-induced adipose dysfunction and insulin resistance23-25. Our data allows us to address an important question: are diet-induced changes in gene expression at the population level shared among subpopulations or do they reflect a change in the relative proportion of these subpopulations? To assess this, we examined the twenty most positively and negatively regulated genes from a TRAP-based RNA-seq experiment in white adipocytes from mice fed chow or high fat diet34 (Supplementary Figure 2a). We noted that some genes (e.g., Cyp2e1 and Fam13a) exhibit elevated expression in chow adipocytes in virtually all subpopulations, while for others (e.g., Cfd), expression is largely driven by the mAd3 population which decreases in abundance with HFD (Extended Data Figure 12b,c, Supplementary Figure 2b). Similarly, Sept9, Cdkn1a, and Fgf13 show increased gene expression after HFD across almost all subpopulations while other HFD-induced genes (e.g., Slc5a7 and Dclk1) increase their expression after HFD in the chow-associated clusters mAd1-4 but not in the HFD-associated clusters mAd5-7 (Supplementary Figure 2b). Thus, diet-dependent expression changes reflect both alterations across all clusters and the emergence or disappearance of distinct populations. Finally, we were somewhat surprised that we did not see a murine population that could be clearly delineated as thermogenic. Such cells have been noted by others in WAT, even at room temperature36. However, when we considered the chow adipocytes independently, mAd1 split into three clusters (Supplementary Figure 3a, b). Two of these clusters, mAd1B and mAd1C, were recognizable as thermogenic beige adipocytes, with relatively high expression of Prdm16 and Ppargc1a in mAd1B and even higher expression of these genes, as well as expression of Ucp1 and Cidea in mAd1C (Supplementary Figure 3c). As expected, the thermogenic mAd1B and mAd1C subpopulations were enriched in ING vs. PG samples (Supplementary Figure 3d, e). Cell-cell interactions in adipose tissue The functions of WAT are known to be coordinated by neural and hormonal cues from outside the fat pad37. There is growing appreciation, however, that intercellular communication within the depot is also critical for the WAT response to overnutrition and other stressors38. In particular, attention has focused on cross-talk between adipocytes and immune cells (especially macrophages) in the context of obesity39. To assess potential interactions between all identified cell types in different depots and at different body mass, we utilized CellPhoneDB40, which uses the expression of ligand-receptor pairs as a proxy for intercellular communication (Supplementary Table 4, 5). As expected, we detected increased potential communication between human adipocytes and macrophages in high BMI vs. low BMI subjects; of 84 potential interactions identified between human adipocytes and macrophages, 40 (48%) were specific for high BMI subjects, while only 3 (4%) were specific for low BMI subjects (Figure 4a, Extended Data Figure 13a, d). Notably, obesity was also associated with robustly increased expression of genes encoding ligand-receptor pairs between adipocytes and many non-immune cell types, including blood and lymphatic endothelial cells, vascular SMCs, pericytes, and ASPCs (Figure 4a, b, Extended Data Figure 13a, d). For example, of 145 potential interactions identified between human adipocytes and endothelial cells, 65 (45%) were specific for high BMI subjects, while only 6 (4%) were specific for low BMI subjects (Extended Data Figure 13d). Potential interactions between these cell types are frequently bidirectional, and receptors are often expressed on multiple cell types, suggesting networks of communication (Figure 4b, Extended Data Figure 13e). We also noted differential expression of ligands and receptors within human adipocyte subpopulations, lending further support to the idea that they carry out distinct functions (Extended Data Figure 13b). The specific interactions upregulated during obesity suggest that adipocytes play a significant role in obesity-related adipose tissue remodeling. For example, adipocyte expression of angiogenic factors like JAG1 and VEGFC is increased in the obese state, as is true of the expression of their receptors (e.g., NOTCH3 and KDR) on endothelial cells, consistant with obesity-associated induction of angiogenesis by adipocytes41 (Figure 4b, Supplementary Table 6). Analysis of the mouse data yielded similar results, as HFD increased the intensity of ligand-receptor pair expression, with the most prominent interactions again occurring between non-immune cell types, especially between ASPCs and adipocytes, pericytes, and SMCs (Extended Data Figure 13c). Interestingly, adipose niche interactions were only modestly conserved between mouse and human. (Extended Data Figure 13d). Interactions between WAT cell types include several that have been studied, such as the effect of the adipokine leptin on endothelial cells via LEPR42, and the actions of TGFB1 on adipose fibrosis via TGFBR134. The majority of these interactions, however, are unstudied in the context of WAT biology. WAT cell types and human disease Adiposity is associated with a wide range of metabolic diseases and traits, and GWAS studies have suggested a specific link between WAT and coronary artery disease (CAD), BMI-adjusted T2D, dyslipidemia, and BMI-adjusted waist-hip ratio (WHR, a measure of body fat distribution)43-45. To determine which specific cell types in WAT are likely to mediate these associations, we employed CELLECT, a method for integrating scRNA-seq and sNuc-seq data with GWAS46. As expected, Type 1 Diabetes (T1D) was significantly associated with B and T lymphocytes and NK cells, consistent with the known autoimmune basis of that disease (Figure 4c). No WAT cell type associated with BMI, as expected given the strong neuronal basis of body weight regulation47. The strongest phenotypic association for white adipocytes was with BMI-adjusted WHR, and associations approaching significance were also noted between adipocytes and HDL and BMI-adjusted T2D (Figure 4c, Supplementary Table 7). All adipocyte subpopulations were significantly associated with WHR (Figure 4d), so we looked for genes responsible for the association with WHR that lack specificity for any particular adipocyte subpopulation. One such gene is PPARG, which is highly expressed in all adipocytes (Extended Data Figure 14a). Data from the METSIM cohort indicates a strong inverse relationship between WHR and PPARG in whole WAT (Extended Data Figure 14b). Unfortunately, WHR was not recorded in the cohort used to generate our floated human adipocytes. WHR is, however, highly correlated with HOMA-IR11, and we found that PPARG expression showed a strong inverse relationship with HOMA-IR in both the METSIM cohort and in our floated adipocytes (Extended Data Figure 14c, d). Furthermore, SNPs in the PPARG gene that are associated with BMI-adjusted WHR30 are also significantly associated with PPARG mRNA and HOMA-IR in our floated adipocyte cohort (Extended Data Figure 14e-h). Adipocytes were also the cell type most likely to mediate the association of WAT with T2D, with the strongest association specifically with hAd7 (Figure 4d). To further investigate the association between hAd7 and T2D, we plotted the abundance of hAd7 as a function of HOMA-IR in our deconvolved floated adipocyte data. This revealed that hAd7 shows significant inverse correlation with insulin resistance (Figure 4e). We then searched for specific hAd7 marker genes that exhibit this same relationship with HOMA-IR, and identified several (Figure 4f, g). Of note, AGMO (also called TMEM195) has emerged as a candidate locus in T2D GWAS48,49. Taken together, our data suggest that hAd7 may have an outsized influence on the risk of T2D, despite representing only ~1% of human adipocytes. Additionally, although adipocytes did not meet genome-wide significance for an association with LDL, we were struck by the near significant relationship between LDL and hAd1, and to a lesser extent, hAd4 (Figure 4c, d). Several hAd1 and hAd4 selective genes showed a strong positive relationship with LDL in our floated adipocyte cohort (Extended Data Figure 14i, j) We also performed CELLECT using the mouse data and noted associations between BMI-adjusted WHR and murine adipocytes and pre-adipocytes (Extended Data Figure 14k-m). This suggests that WHR may be determined in large part by alterations in adipocyte differentiation, a hypothesis consistent with the PPARG data above, and with independent studies of different WHR genes50. HDL and TG are also associated with mouse white adipocyte gene expression (Extended Data Figure 14k-m). Discussion Here, we present a comprehensive atlas of human and mouse WAT across depot and body mass. Our analysis reveals a rich array of cell types, including blood and lymphatic vascular cells, immune cells, and ASPCs, in addition to adipocytes. These cell types are grossly similar across species, but differ more profoundly when cellular subpopulations are explored. It is tempting to attribute these subpopulation differences to divergence across 65 million years of evolution, but other factors also need to be considered. For example, the human samples were collected after a fast, while the mice were harvested after ad libitum feeding, which might be expected to cause some differences in cell state related to insulin signaling or related pathways. Overall, our data highlight a central role for adipocytes in the local regulation of the adipose depot as well as in systemic physiology. The single cell resolution of our dataset enables the identification of heterogeneity that cannot be appreciated by bulk RNA sequencing, such as a potentially visceral thermogenic subpopulation (hAd6), and a rare subpopulation associated with T2DM (hAd7). We additionally provide a framework for mouse-human comparison in studies of adipose tissue that will be an important resource for groups hoping to translate murine findings to human treatments. These data provide a lens of unprecedented acuity that better informs our understanding of WAT biology and enables a deeper exploration of the role of adipose tissue in health and disease. METHODS Collection of human adipose tissue samples. Drop-Seq and Floated adipocyte bulk RNA-seq Subcutaneous adipose tissue was collected under Beth Israel Deaconess Medical Center Committee on Clinical Investigations IRB 2011P000079. Potential subjects were recruited in a consecutive fashion, as scheduling permitted, from the plastic surgery operating room rosters at Beth Israel Deaconess Medical Center. Male and female subjects over the age of 18 undergoing elective plastic surgery procedures and free of other acute medical conditions were included and provided written informed consent preoperatively. Excess adipose tissue from the surgical site was collected at the discretion of the surgeon during the normal course of the procedure. Subjects with a diagnosis of diabetes, or taking insulin-sensitizing medications such as thiazolidinediones or metformin, chromatin-modifying enzymes such as valproic acid, anti-retroviral medications, or drugs known to induce insulin resistance such as mTOR inhibitors or systemic steroid medications, were excluded. sNuc-Seq Subcutaneous and visceral adipose tissue was collected under BIDMC Committee on Clinical Investigations IRB 2011P000079 and University of Pittsburgh Medical Center STUDY 19010309. At BIDMC, potential subjects were recruited in a consecutive fashion, as scheduling permitted, from the gynecological, vascular, and general surgery rosters. Male and female subjects over the age of 18 undergoing plastic surgery (panniculectomy, thighplasty or deep inferior epigastric perforators), gynecological surgery (total abdominal hysterectomy and bilateral salpingo-oophorectomy) or general surgery (cholecystectomy (CCY) or colin polyp surgery) and free of other acute medical conditions were included and provided written informed consent preoperatively. Excess adipose tissue from the surgical site was collected at the discretion of the surgeon during the normal course of the procedure. The exclusion criteria were any subjects taking thiazolidinediones, chromatin-modifying enzymes such as valproic acid, anti-retroviral medications, and drugs known to induce insulin resistance such as mTOR inhibitors or systemic steroid medications. At UPMC, inclusion criteria were patients receiving bariatric surgery (Vertical Sleeve Gastrectomy or Roux en Y Gastric Bypass) or lean controls (hernia or CCY surgeries) ages 21-60, exclusion criteria were diagnosis of diabetes (Type 1 or Type 2), pregnancy, alcohol or drug addiction, bleeding or clotting abnormality, or inflammatory abdominal disease. All patients provided written informed consent preoperatively. Excess adipose tissue from the surgical site was collected at the discretion of the surgeon during the normal course of the procedure. 200-500 mg samples were flash frozen immediately after collection for downstream processing. Mouse adipose tissue samples All animal experiments were performed under a protocol approved by the BIDMC Institutional Animal Care and Use Committee. Male C57Bl/6J 16-week-old high fat diet fed (JAX 380050) and chow fed (JAX 380056) mice were obtained from The Jackson Laboratory and maintained on 60% high fat diet (Research Diets, D12492) or chow diet (8664 Harlan Teklad, 6.4% wt/wt fat), respectively, for three weeks before sacrifice at 19 weeks. Female 6-week-old chow fed C57Bl/6J mice (JAX 380056) were maintained on 60% high fat diet for 13 weeks before sacrifice at 19 weeks. Mice were maintained under a 12 hr light/12hr dark cycle at constant temperature (23°C) and humidity in the range of 30%-70% with free access to food and water. There were no calculations performed to determine sample size. Animals were not randomized and researchers were not blinded to the diet of the animal due to the nature of the study. During dissection, to avoid contamination by cells from the inguinal lymph node, we excised the node with a fairly wide margin, possibly de-enriching beige adipocytes from our samples51. Mature human adipocyte sample preparation Purification of mature human adipocytes. Whole tissue subcutaneous adipose specimens were freshly collected from the operating room. Skin was removed, and adipose tissue was cut into 1- to 2-inch pieces and rinsed thoroughly with 37°C PBS to remove blood. Cleaned adipose tissue pieces were quickly minced in an electric grinder with 3/16-inch hole plate, and 400 ml of sample was placed in a 2-l wide-mouthed Erlenmeyer culture flask with 100 ml of freshly prepared blendzyme (Roche Liberase TM, research grade, cat. no. 05401127001, in PBS, at a ratio of 6.25 mg per 50 ml) and shaken in a 37 °C shaking incubator at 120 r.p.m. for 15–20 min to digest until the sample appeared uniform. Digestion was stopped with 100 ml of freshly made KRB (5.5 mM glucose, 137 mM NaCl, 15 mM HEPES, 5 mM KCl, 1.25 mM CaCl2, 0.44 mM KH2PO4, 0.34 mM Na2HPO4 and 0.8 mM MgSO4), supplemented with 2% BSA. Digested tissue was filtered through a 300 μM sieve and washed with KRB/albumin and flow through until only connective tissue remained. Samples were centrifuged at 233g for 5 min at room temperature, clear lipid was later removed, and floated adipocyte supernatant was collected, divided into aliquots and flash-frozen in liquid nitrogen. Sample selection and Bulk-RNA-seq library construction Fasting serum was collected and insulin, glucose, free fatty acids, and a lipid panel were measured by Labcorp. BMI measures were derived from electronic medical records and confirmed by self-reporting, and measures of insulin resistance, the homeostasis model assessment-estimated insulin resistance index (HOMA-IR) and revised quantitative insulin sensitivity check index (QUICKI) were calculated52,53. Female subjects in the first and fourth quartiles for either HOMA-IR or QUICKI and matched for age and BMI were processed for RNA-seq. Total RNA from ~400 μl of thawed floated adipocytes was isolated in TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. For RNA-seq library construction, mRNA was purified from 100 ng of total RNA by using a Ribo-Zero rRNA removal kit (Epicentre) to deplete ribosomal RNA and convert into double-stranded complementary DNA by using an NEBNext mRNA Second Strand Synthesis Module (E6111L). cDNA was subsequently tagmented and amplified for 12 cycles by using a Nextera XT DNA Library Preparation Kit (Illumina FC-131). Sequencing libraries were analyzed with Qubit and Agilent Bioanalyzer, pooled at a final loading concentration of 1.8 pM and sequenced on a NextSeq500. Single Cell and Single Nucleus sample preparation and processing SVF isolation and Drop-seq. Adipose tissue samples were collected and processed as above. After removal of floated adipocytes, remaining supernatant was aspirated and the remaining pelleted stromal vascular fraction (SVF)was combined from multiple tubes. The combined SVF was washed 2 times with 50ml cold PBS with 233g for 5 min centrifugation between washes. Erythrocytes were depleted with two rounds of 25 ml. ACK lysing buffer (Gibco™ A1049201) exposure (5 minutes at RT followed by 233g x 5 min centrifugation). Remaining SVF pellet was further washed x 2 with 50ml cold PBS prior to counting on hematocytometer and loading onto Drop-seq microfluidic devices. Drop-seq was performed as described54, with the following modifications: first, flow rates of 2.1 mL/h were used for each aqueous suspension and 12 mL/h for the oil. Second, libraries were sequenced on the Illumina NextSeq500, using between 1.6-1.7 pM in a volume of 1.2 mL HT1 and 3 mL of 0.3 μM Read1CustSeqB (GCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGTAC) using 20 x 8 x 60 read structure. sNuc-Seq Nuclei were isolated from frozen mouse and human adipose tissue samples for 10x snRNA-seq using a slightly modified approach to what was previously described55-57. Samples were kept frozen on dry ice until immediately before nuclei isolation, and all sample handling steps were performed on ice. Each flash-frozen adipose tissue sample was placed into a gentleMACS C tube (Miltenyi Biotec) with 2 mL freshly prepared TST buffer (0.03% Tween 20 [Bio-Rad], 0.01% Molecular Grade BSA [New England Biolabs], 146 mM NaCl [ThermoFisher Scientific], 1 mM CaCl2 [VWR International], 21 mM MgCl2 [Sigma Aldrich], and 10 mM Tris-Hcl pH 7.5 [ThermoFisher Scientific] in Ultrapure water [ThermoFisher Scientific]) with or without 0.2 U/μL of Protector RNase Inhibitor (Sigma Aldrich). gentleMACS C tubes were then placed on the gentleMACS Dissociator (Miltenyi Biotec) and tissue was dissociated by running the program “mr_adipose_01” twice, and then incubated on ice for 10 minutes. Lysate was passed through a 40 μm nylon filter (CellTreat) and collected into a 50 mL conical tube (Corning). Filter was rinsed with 3 mL of freshly prepared ST buffer buffer (146 mM NaCl, 1 mM CaCl2, 21 mM MgCl2; 10 mM Tris-Hcl pH 7.5) with or without 0.2 U/μL RNase Inhibitor, and collected into the same tube. Flow-through was centrifuged at 500 x g for 5 minutes at 4°C with brake set to low. Following centrifugation, supernatant was removed, and the nuclear pellet was resuspended in 50 - 200 μl PBS pH 7.4 (ThermoFisher Scientific) with 0.02% BSA, with or without 0.2U/μL RNase Inhibitor. In order to reduce ambient mRNA, the nuclear pellets of some samples were washed 1-3 times with 5 mL of PBS-0.02% BSA before final resuspension. An aliquot of nuclei from each sample was stained with NucBlue (Thermofisher Scientific), counted in a hemocytometer using fluorescence to identify intact nuclei, and then immediately loaded on the 10x Chromium controller (10x Genomics) according to the manufacturer’s protocol. For each sample, 10,000-16,500 nuclei were loaded in one channel of a Chromium Chip (10x Genomics). The Single Cell 3’ v3.1 chemistry was used to process all samples. cDNA and gene expression libraries were generated according to the manufacturer's instructions (10x Genomics). cDNA and gene expression library fragment sizes were assessed with a DNA High Sensitivity Bioanalyzer Chip (Agilent). cDNA and gene expression libraries were quantified using the Qubit dsDNA High Sensitivity assay kit (ThermoFisher Scientific). Gene expression libraries were multiplexed and sequenced on the Nextseq 500 (Illumina) with a 75-cycle kit and the following read structure: Read 1: 28 cycles, Read 2: 55 cycles, Index Read 1: 8 cycles. Sequencing, read alignments, and quality control Single-cell/nucleus RNA-seq data analysis. Raw sequencing reads were demultiplexed to FASTQ format files using bcl2fastq (Illumina; version 2.20.0). Digital expression matrices were generated from the FASTQ files using the Drop-Seq tools (https://github.com/broadinstitute/Drop-seq, version 2.4.0) pipeline, with appropriate adjustments made to the default program parameters to account for the different read-structures in the scRNA Drop-Seq data and sNuc 10X data. Reads from mouse and human were aligned with STAR58 (version 2.7.3) against the GRCm38 and GRCh38 genome assemblies, respectively. Gene counts were obtained, per-droplet, by summarizing the unique read alignments across exons and introns in appropriate GENCODE annotations (release 16 of the mouse annotation and release 27 of the human annotation). In order to adjust for downstream effects of ambient RNA expression within mouse nuclei (hereafter “cells”), we used CellBender59 (version 0.2.0) to remove counts due to ambient RNA molecules from the count matrices and to estimate the true cells. We also used CellBender to distinguish droplets containing cells from droplets containing only ambient RNA, by selecting droplets with >50% posterior probability of containing a cell. We compared the true cell estimation obtained using CellBender against the same using the DropletUtils software package60, which estimates ambient RNA expression but does not remove any ambient counts, keeping only the cells that were marked as not ambient by both algorithms. To address ambient RNA in the human sNuc data, we calculated spliced and unspliced RNA content in each cell, because nuclei have a high unspliced RNA content, a high percentage of spliced RNA indicates a high ambient RNA content. We therefore removed sNuc-seq cells containing over 75% spliced RNA. All samples were assessed for doublet content using scrublet61 version 0.2.1, and cells called as doublets were removed before further analysis. All cells were further filtered to have greater than 400 UMIs with <10% of UMIs from mitochondrial genes. Genes were filtered such that only genes detected in two or more cells were retained. For the human data, the median number of UMIs detected per cell was 2559 and the median number of genes detected per cell was 1524. For the mouse data, the median number of UMIs detected per cell was 2291 and the median number of genes detected per cell was 1369. Bulk RNA-seq Analysis. Raw sequencing reads were demultiplexed by using bcl2fastq (Illumina). Salmon62 (version 1.1.0) was used to simultaneously map and quantify transcript abundances of hg19 genes annotated by release 19 of the GENCODE project’s human reference. Salmon was run using “full” selective alignment (SAF) with mapping validation as described previously63. Gene counts were summarized from transcript abundances using the “tximport” package for R64. Integration, clustering, subclustering, and annotation Integration, clustering and subclustering analysis were performed using Seurat 3.9.965. The gene counts were normalized using SCTransform66, and regressed on mitochondrial read percentage, ribosomal read percentage, and cell cycle score as determined by Seurat. In order to avoid smoothing over depot differences, for integration human and mouse data were grouped by ‘individual’, i.e., if both subcutaneous and visceral adipose tissue for an individual human or mouse were available, they were pooled together during this step. Individuals were integrated with reciprocal PCA, using individuals that had both subcutaneous and visceral samples as references. As a result, the human and mouse references were comprised exclusively from the sNuc seq cohort. To integrate, references were integrated together, then the remaining samples—sNuc seq individuals with only subcutaneous data as well as all Drop-seq samples—were mapped to the reference. For clustering, 5000 variable genes were used, and ribosomal and mitochondrial genes were removed from the variable gene set before running PCA and calculating clusters using a Louvain algorithm, 40 PCs, and a resolution of 0.5. Clusters were identified as adipocytes, preadipocytes, mesothelial cells, vascular cells, or immune cells using marker genes, subset into individual objects, and re-integrated using the above method. Samples with fewer than 50 cells in the subset were removed before re-integration. This led to samples having artificially fewer cells in some instances—for example some Drop-seq samples had cells that clustered with adipocytes, but these cells were removed in subclustering because the small numbers of cells introduced too much variability into the integration. Subclustering was performed using a range of variable genes (1000-2000), PCs (10-40) and resolutions (0.2-0.6). Markers were calculated using a non-parametric Wilcoxon rank sum test with p values adjusted using Bonferroni correction (Supplementary Tables 1, 2), and clusters were evaluated based on the distinctness of called markers to determine the final subclustering conditions. In the subclustered objects, we removed clusters that appeared to represent doublets based on the score assigned by scrublet61, or that appeared to be driven by high ambient RNA content as determined by percentage of mitochondrial genes and spliced/unspliced RNA ratio. The remaining clusters were annotated based on marker gene expression. In some cases, smaller subclusters (T and NK cells, B cells, monocytes/neutrophils) were further subset and PCA and clustering analysis but not integration was re-run in order to assign clusters. After subcluster annotation, identities were mapped back onto the original object and cells that were removed from the subclustered objects were similarly removed from the all-cell object. Deconvolution of bulk RNA-seq data Bulk RNA sequencing data for subcutaneous adipose tissue from the METSIM cohort were obtained as described previously13. Only individuals with available metabolic phenotyping data were used for the deconvolution analysis. Bulk RNA sequencing data for floated human adipocytes were obtained described above. Deconvolution analysis was performed using MuSiC12 (version 0.1.1) with human sNuc subcutaneous all cell or adipocyte data as reference. Marker genes used for deconvolution can be found in Supplementary Table 1. Comparison between mouse and human datasets Mapping of mouse cells onto human clusters was performed using Seurat multimodal reference mapping67. To run, for the all-cell and each subset, the mouse data was prepared by extracting the counts matrix from the mouse sNuc object and mapping the mouse gene names to their human orthologs using a database of ortholog mappings from Mouse Genome Informatics (http://www.informatics.jax.org/homology.shtml). In the case of multi-mapping, the first ortholog pair was used. The mouse object was then split by sample and mapped onto the sNuc-seq data from the matching human all-cell or subset object using the RNA assay and PCA reduction. Integration of ASPCs from this and other studies Data was obtained from Burl et. al.4 (SRP145475), Hepler et. al.6 (GSE111588), Merrick et. al.5 (GSE128889), Sárvári et. al.9 (GSE160729), and Schwalie et. al.3 (E-MTAB-6677); processed data, metadata, and/or cell type designations were obtained from the authors when necessary. Datasets were subset to contain only ASPCs, grouped by individual animal/experiment when possible or by lab when not, and integrated with the data from this paper (grouped by animal) using RPCA integration without references with 1500 variable genes. The UMAP reduction was calculated using the top 20 PCs and ASPCs were grouped by cell type annotations from their original papers for analysis. Immunohistochemistry Subcutaneous (abdominal) and omental adipose tissue biopsies belonging to lean and obese women (GRIA4: subcutaneous, 5 lean and 5 obese individuals; PGAP1: subcutaneous, 5 lean, 4 obese, visceral 3 lean, 4 obese; EBF2: omental, 3 lean and 4 obese individuals; AGMO: subcutaneous, 4 lean and 4 obese individuals, for all experiments two slides per individual for GRIA4, EBF2, AGMO, one slide per individual for PGAP1) were fixed (overnight in 4% paraformaldehyde at 4°C, dehydrated, paraffin embedded and sectioned (4μm thick). The following primary antibodies and respective dilution were used: GRIA4, 1:200, Cat #23350-1-AP, Proteintech; PGAP1, 1:400, Cat. #55392-1-AP, Proteintech EBF2, 1:1000, Cat. #AF7006, R&D systems; AGMO (TMEM195) 1:100, Cat #orb395684, Biorbyt. In brief, after rinsing in PBS, tissue slices were blocked with 3% normal goat serum and incubated with the primary antibody in PBS, overnight at 4°C. After a thorough rinse in PBS, sections were incubated in 1:200 v/v biotinylated secondary antibody solution for 30 minutes (Invitrogen), rinsed in PBS and incubated in avidin-biotin-peroxidase complex (ABC Standard, Vector Laboratories), washed several times in PBS and lastly incubated in 3,3′-diaminobenzidine tetrahydrochloride (0.05% in 0.05 M Tris with 0.03% H2O2; 5 min). After immunohistochemical staining, sections were counterstained with hematoxylin, dehydrated in ethanol, cleared in xylene and covered with coverslip using Eukitt (Merck). All observations were performed using Nikon Eclipse E800 light microscope. Immunofluorescence microscopy of mature human adipocytes Adipocyte immunofluorescence protocol was adapted from Sárvári et al9. Abdominal subcutaneous adipose tissue was collected from two adult female human subjects (BMI 24.9 and 40.3) as above and placed on ice. Tissue was minced and digested with 1 mg/mL type II collagenase (Sigma-Aldrich, C6885) in Hanks’ balanced salt solution supplemented with 0.5% fatty acid-free BSA (Sigma-Aldrich, A6003) at 37° in a water bath with constant shaking at 250 rpm. The cell suspension was filtered through a 250 μM nylon mesh strainer (Thermo, 87791) and washed three times with Krebs-Ringer bicarbonate buffer containing 1% fatty acid-free BSA. All washes throughout this protocol were performed without centrifugation to minimize adipocyte damage and loss; cell suspension was maintained upright for at least 5 minutes to allow mature adipocytes to float, and infranatant was removed with a needle and syringe. The floating adipocytes were fixed with 2% PFA and 1% sucrose in PBS for 30 minutes with constant rotation followed by three washes with 2% fatty acid-free BSA in PBS. Adipocytes were subsequently permeabilized with 0.5% Triton-X (Thermo, 28314) in PBS for five minutes, and incubated with 2.5 μg/mL trypsin (Corning, 25053CI) in PBS for 10 minutes at 37° in a water bath with constant shaking. Adipocytes were then blocked with 2% fatty acid-free BSA in PBS for 30 minutes, and incubated overnight at room temperature with rabbit polyclonal anti-GRIA4 (Proteintech, 23350-1-AP) diluted 1:100 in 500 μL 2% fatty acid-free BSA in PBS with constant rotation. The adipocytes were then washed twice for 10 minutes each with 0.1% fatty acid-free BSA and 0.05% Tween-20 (Sigma-Aldrich, P9416) in PBS, followed by incubation with goat anti-rabbit Alexa Fluor 546 (Thermo, A-11035) secondary antibody diluted 1:500 in 2% fatty acid-free BSA for 2 hours with rotation. For the final 30 minutes of incubation, Hoechst 33342 (Thermo, 62249) and BODIPY 493/503 (Invitrogen, D3922) were added at 1:500 dilutions. Adipocytes were washed twice and resuspended in 300 μL Fluoromount G (Southern Biotech, 0100-01) and mounted on glass slides with 1.4-1.6 mm concavity wells (Electron Microscopy Sciences, 71878-03). A sample of adipocytes was also incubated as above but without primary antibody to verify the specificity of the secondary antibody. Fluorescence images were acquired using Zeiss LSM 880 Upright Laser Scanning Confocal Microscope with filter cubes for DAPI, GFP, and Rhodamine in parallel using the 20X objective and processed using Zen Black 2.3 software. Images were analyzed and counted with ImageJ v. 1.53k. Ex vivo differentiation and transcriptional and high-content image-based characterization of differentiating primary human adipocyte progenitors We obtained adipocyte progenitors from subcutaneous and visceral adipose tissue from patients undergoing a range of abdominal laparoscopic surgeries (sleeve gastrectomy, fundoplication or appendectomy). The visceral adipose tissue is derived from the proximity of the angle of His and subcutaneous adipose tissue obtained from beneath the skin at the site of surgical incision. Additionally, human liposuction material was obtained. Each participant gave written informed consent before inclusion and the study protocol was approved by the ethics committee of the Technical University of Munich (Study № 5716/13). Isolation of AMSCs was performed as previously described28, and cells were differentiated in culture over 14 days. Ex vivo differentiated adipocytes were stained and imaged, and features were extracted using LipocyteProfiler as described in Laber et al. RNA-sequencing libraries were prepared and sequenced and QC’ed as previously described28. Bulk-RNA sequencing counts from subcutaneous and visceral samples differentiated for 14 days were deconvoluted using both subcutaneous and visceral adipocytes as reference as described above. Raw images collected during LipocyteProfiler analysis were randomly selected from samples predicted to have high or low content of hAd3, hAd5, or hAd6 adipocytes, and pseudocolored and combined using Adobe Photoshop. Gene Pathway Analysis Analysis of enriched pathways in adipocyte markers was performed using clusterProfiler68 (version 3.16.1). Adipocyte cluster markers were filtered to a Benjamini-Hochberg adjusted p-value < .05, then evaluated for enrichment in GO biological pathways or KEGG pathways containing under 300 genes. All pathways and p values can be found in Supplementary Table 3. Identification and analysis of EBF2 SNP association with visceral adiposity VAT, ASAT, and GFAT volumes in 40,032 individuals from the UK Biobank69,70 who underwent MRI imaging were quantified as described elsewhere71. Variant rs4872393 was identified as a lead SNP associated with VATadjBMI and waist-to-hip ratio from summary statistics of two prior studies31,72. Among the cohort who underwent MRI imaging, all variants at this locus (± 250 kb around rs4872393) with MAF >= 0.005 and imputation quality (INFO) score >= 0.3 were analyzed. For all 554 nominally significant (P < 0.05) variants associated with VATadjBMI in this region, a secondary conditional analysis testing for association with VATadjBMI was performed controlling for rs4872393 carrier status (P < 0.05/554 = 9 x 10−5). Participants were excluded from analysis if they met any of the following criteria: (1) mismatch between self-reported sex and sex chromosome count, (2) sex chromosome aneuploidy, (3) genotyping call rate < 0.95, or (4) were outliers for heterozygosity. Up to 37,641 participants were available for analysis. Fat depot volumes adjusted for BMI and height (“adj” traits) were calculated by taking the residuals of the fat depot in sex-specific linear regressions against age at the time of MRI, age squared, BMI, and height31. Each trait was scaled to mean 0 and variance 1 in sex-specific groups before being combined for analysis. Linear regressions between a given trait-variant pair were adjusted for age at the time of imaging, age squared, sex, the first 10 principal components of genetic ancestry, genotyping array, and MRI imaging center. Analyses were performed using R 3.6.0 (R Project for Statistical Computing). EBF2 regional visualization plot was made with the LocusZoom online tool73. Calculation of pseudobulk datasets to estimate adipose innervation Approximate bulk RNA-seq datasets (pseudobulk) were obtained for visceral sNuc-seq samples by summing the total expression per-gene across all droplets containing a valid 10X cell barcode. This includes all cells that would normally have been removed in the single-nuclei studies by any of the filtering criteria (above): doublet score, splicing content, droplets with fewer than 400 UMIs, etc, in order to preserve the ambient RNA present in otherwise empty droplets. Repeated UMIs were still collapsed into single counts (per-droplet) before summing. Amounts of pan-neuronal markers were calculated using this pesudobulk dataset and plotted against the proportion of visceral populations hAd2 and hAd6 relative to total adipocytes in each sample. Prediction of cell-cell interactions Analysis of cell-cell interactions was performed using CellphoneDB40 (version 2.0.0). For human data, sNuc-seq counts data was split into files containing cells from subcutaneous and visceral fat from individuals with BMI lower than 30 or higher than 40. CellphoneDB with statistical analysis was run on each file separately to evaluate interactions in each condition. For mouse data, counts data was split into files containing cells from the inguinal and perigonadal fat of chow and high fat diet fed mice. Mouse gene names were converted to human gene names, as above, before running CellphoneDB with statistical analysis on each file. Identification of candidate etiologic cell types using CELLEX and CELLECT CELLECT (https://github.com/perslab/CELLECT) and CELLEX (https://github.com/perslab/CELLEX) were used to identify candidate etiological cell types for a total of 23 traits. The input data for CELLECT is GWAS summary statistics for a given trait and cell type expression specificity (ES) estimates derived from single-cell RNA-seq data. The output is a list of prioritized candidate etiologic cell types for a given trait. ES estimates were calculated using CELLEX (version 1.1), which computes robust estimates of ES by relying on multiple expression specificity measures (for further details see Timshel et. al.74). CELLEX was run separately on the raw mouse and human (sNuc) gene expression matricies to compute gene expression specificities for each cluster based on the clustering assignment reported above. The resulting cell type specificity matrix was used along with multiple GWAS studies30,75-79 (Extended Data Table 3) as input for CELLECT74 (version 1.1), which was run with default parameters. Significant cell types were identified using a by-trait and by-species Bonferroni p-value threshold of p<0.05. SNP analysis for bulk mRNA-seq cohort The raw GTC SNP expression data from Infinium OmniExpress-24 Kit was converted to VCF format using Picard version 2.21.6. The pre-processing of the SNP data before phasing and imputation was performed using plink2 (https://www.cog-genomics.org/plink/2.0/). The SNP genotype was then phased and imputed using the Eagle v2.3.580 and Minimac381 packages, respectively. SNPs were mapped to the NCBI database using the rsnps package (https://CRAN.R-project.org/package=rsnps) and filtered to keep only SNPs that had a minor allele frequency > 0.05. For plotting gene expression against genotype, bulk RNA sequencing data was TMM normalized using edgeR82. Statistical validation for significance was done using the Wilcoxon rank-sum Test which is a non-parametric test assuming independent samples. Statistics p-values for scatterplots were calculated using GraphPad Prism version 8.0 and represent the results of an extra sum-of-squares F-test with the null hypothesis that the slope equals zero. All error bars on bar graphs represent standard error. Statistics on proportional composition graphs were calculated using scCODA83 (version 0.1.2) using the Hamiltonian Monte Carlo sampling method. The model formula used was “Depot + BMI” (human) or “Depot + Diet) (mouse) for all objects in for which both of these covariates were present, or the individual covariate when only a single condition was present. Extended Data Extended Data Fig. 1. Additional analysis of the effects of depot and BMI on human WAT populations. a, UMAP projections of cells from the lowest and highest BMI ranges in the dataset, split by depot. To facilitate comparison, samples were randomly subset to contain the same number of cells in each plot (n = 20,339). b, Graph showing the proportion of sNuc-seq cells in each cluster per sample, split by depot and BMI, n = 4 SAT < 30, 6 SAT > 40, 3 VAT < 30, 5 VAT > 40. C, Estimated cell type proportions in bulk RNA sequencing data of subcutaneous adipose tissue from 331 individuals from the METSIM cohort calculated using sNuc-seq data as reference. Vascular cells include endothelial, lymphatic endothelial, pericytes, and smooth muscle cells. Myeloid immune includes macrophages, monocytes, dendritic cells, mast cells and neutrophils, and lymphoid immune includes B cells, NK cells, and T cells. For lines of best fit: Adipocytes R2 = 0.031, ASPCs R2 = 0.034, Vascular R2 = 0.076, Myeloid Immune R2 = 0.13, Lymphoid Immune R2 = 0.0049. For scatterplots, error bands represent a confidence level of 0.95 and p values were calculated using an F-test with the null hypothesis that the slope = 0. For bar graphs, error bars represent SEM, * indicates credible depot effect and # indicates credible BMI effect, calculated using dendritic cells as reference. Extended Data Fig. 2. Additional analysis of the effects of depot and diet on mouse WAT populations and association with human WAT populations. a, UMAP projection of all mouse WAT cells split by depot. b, Proportion of cells in each cluster per sample, split by sex as well as by depot and diet, for male mice n = 4 ING Chow, 4 ING HFD, 3 EPI Chow, and 5 EPI HFD. For female mice, n = 2 per condition. c, Riverplot showing the relationship between mouse and human clusters. Mouse cells were mapped onto human sNuc-seq cells using multimodal reference mapping. The riverplot represents the relationship between manually assigned mouse cluster and mapped human cluster for every mouse cell. For bar graphs, error bars represent SEM, * indicates credible depot effect and # indicates credible diet effect, calculated using dendritic cells as reference. Extended Data Fig. 3. Highly similar vascular cells in human and mouse WAT. a, UMAP projection of 22,734 human vascular cells. b, Marker genes for 11 distinct clusters of human WAT vascular cells. c, UMAP projection of 7,632 mouse vascular cells. d, Marker genes for 9 distinct clusters of mouse WAT vascular cells. e, Riverplot showing the correlation between annotated mouse and human vascular clusters based on multimodal reference mapping for each mouse cell. f-g, Bar graphs showing the proportion of cells in each cluster per sample split by depot and BMI for human (f) and depot, diet, and sex for mouse (g). For humans, n = 9 SAT < 30, 6 SAT > 40, 3 VAT < 30, and 5 VAT > 40. For male mice n = 4 ING Chow, 4 ING HFD, 3 EPI Chow, and 5 EPI HFD. For female mice, n = 2 per condition. For bar graphs, error bars represent SEM, * indicates credible depot effect and # indicates credible BMI/diet effect, calculated using hEndoA2 (human) and mEndoA2 (mouse) as reference. Extended Data Fig. 4. Comparison of immune cells in human and mouse WAT. a, UMAP projection of 34,268 immune cells from human WAT. b, Marker genes for human immune cell clusters. c, UMAP projection of 70,547 immune cells from mouse WAT. d, Marker genes for mouse immune cell clusters. e-f, Riverplots showing the correlation between annotated mouse cluster and mapped human cluster for mouse (e) dendritic cells, mast cells, neutrophils, B cells, NK cells, and T cells and (f) monocytes and macrophages. Extended Data Fig. 5. Human and mouse immune cells are differentially regulated by depot and BMI/diet. a-b, UMAP projections of human (a) and mouse (b) WAT immune cells split by depot. c-d, UMAP projections of human (c) and mouse (d) WAT immune cells split by BMI (c) and diet (d). e-f, Bar graphs showing the proportion of cells in each cluster per sample split by depot and BMI for human (e) and depot, diet, and sex for mouse (f). For humans, n = 10 SAT < 30, 6 SAT > 40, 3 VAT < 30, and 5 VAT > 40. For male mice n = 4 ING Chow, 4 ING HFD, 3 EPI Chow, and 5 EPI HFD. For female mice, n = 2 per condition. For bar graphs, error bars represent SEM, * indicates credible depot effect and # indicates credible BMI/diet effect, calculated using hMono2 (human) and mcDC1 (mouse) as reference. Extended Data Fig. 6. Subpopulations of human and mouse mesothelial cells. a, UMAP projection of 30,482 human mesothelial cells. b, Marker genes for distinct human mesothelial populations. c, UMAP projection of 14,947 mouse mesothelial cells. d Marker genes for distinct mouse mesothelial populations. e, Riverplots showing relationship of mouse and human mesothelial clusters. f-g, Proportion of cells in each cluster per sample, split by BMI for human (f) and diet and sex for mouse (g). For humans, n = 3 VAT < 30, and 5 VAT > 40. For male mice n = 3 EPI Chow, and 5 EPI HFD. For female mice, n = 2 per condition. Error bars represent SEM, # indicates credible BMI/diet effect, calculated using hMes3 (human) and mMes1 (mouse) as reference. Extended Data Fig. 7. Human and mouse ASPCs share commonalities with previously reported subtypes. a, UMAP projection of 52,482 human ASPCs. b, Marker genes for distinct ASPC subpopulations. c, UMAP projection of 51,227 mouse ASPCs. d, Marker genes for distinct ASPC subpopulations. e, Riverplot depicting the relationship between mouse and human ASPC clusters. f, Integration of ASPCs from this paper with ASPCs from other groups. Extended Data Fig. 8. Human ASPCs exhibit strong depot dependency while mouse ASPCs are dependent on both depot and diet. a-b, UMAP projections of human (a) and mouse (b) ASPCs split by depot. c-d, UMAP projections of human (c) and mouse (d) ASPCs split by BMI/diet. e-f, Proportion of ASPC cells in each cluster per sample split by depot and BMI for human (e) and depot, diet, and sex for mouse (f). For humans, n = 11 SAT < 30, 6 SAT > 40, 3 VAT < 30, and 5 VAT > 40. For male mice n = 4 ING Chow, 4 ING HFD, 3 EPI Chow, and 5 EPI HFD. For female mice, n = 2 per condition. For bar graphs, error bars represent SEM, * indicates credible depot effect and # indicates credible BMI/diet effect, calculated using hASPC2 (human) and mASPC4 (mouse) as reference. Extended Data Fig. 9. Human adipocyte subtypes are highly dependent on depot and may be responsible for distinct functions. a-b, UMAP projections of human white adipocytes split by depot (a) and BMI (b). c, Proportion of cells in each human cluster by sample split by depot and BMI, n = 4 SAT < 30, 6 SAT > 40, 3 VAT < 30, and 5 VAT > 40. D, Quantification of immunofluorescence analysis of GRIA4+ cells in mature human adipocytes from two individuals. Each dot represents an image, n = 12 images from individual 1 and 9 images from individual 2 with a total of 704 counted cells. Only cells with visible nuclei were included in the quantification. e, Representative image of GRIA4+ cells, white arrows represent positive cells, grey represent negative, scale bar = 100 μm. In total, there were 21 images from samples taken from two individuals. f, Expression of genes associated with adipokine secretion, insulin signaling, lipid handling, and thermogenesis across human adipocyte subclusters. g-m, Expression of genes associated with GO or KEGG pathways indicative of individual human adipocyte subclusters. For bar graph, error bars represent SEM, * indicates credible depot effect and # indicates credible BMI effect, calculated using hAd5 as reference. Extended Data Fig. 10. Human adipocytes differentiated ex vivo recapitulate many of the adipocyte subclusters found in vivo. a, Plot of estimated cell type proportion in ex vivo adipocyte cultures differentiated from subcutaneous or visceral preadipocytes for 14 days, ordered by estimated proportion. b-c, Scatterplots showing the relationship between estimated cell type proportion and the LipocyteProfiler-calculated features Large BODIPY objects (b) and Median BODIPY Intensity (c). p values were calculated using an F-test with the null hypothesis that the slope = 0. d, Representative images of hAd3 low/hAd5 or hAd3 high hAd5 low ex vivo differentiated cultures. Green represents BODIPY staining, blue represents Hoechst staining. Scale bars are 100 μm, in total, 3 randomly selected images/sample were analyzed from 3 SAT samples and 3 VAT samples with the lowest and highest predicted proportions of hAd3 and hAd5. Extended Data Fig. 11. Visceral-specific adipocyte subpopulation hAd6 is associated with thermogenic traits. a, Regional visualization of associations of common genetic variants near EBF2 with VATadj. b, Effect size of association of rs4872393 with VATadj, ASATadj, GFATadj, and BMI per minor allele A; n = 37,641. Error bars reflect a 95% confidence interval around the effect size estimate from regression. c, VATadj raw data plotted according to rs4872393 carrier status; n = 36,185. For box plots, horizontal line = median, lower and upper bounds of the box = 1st and 3rd quartile respectively, lower and upper whisker = 1st quartile – 1.5 x interquartile range (IQR) and 3rd quartile + 1.5 x IQR respectively, outliers are plotted as points. d, Scatterplot showing the relationship between estimated cell type proportion and the LipocyteProfiler calculated feature Mitochondrial Intensity in visceral samples. e, Expression of mitochondrial and thermogenic genes in visceral ex vivo differentiated adipocytes stratified by estimated hAd6 proportion and matched for amount of differentiation using PPARG expression, n = 7 mAd6 low and 5 mAd6 high. Error bars represent SEM, p values were calculated using two tailed t-tests with no adjustments for multiple comparison, *, p < .05, **, p < .01. Exact p values: EBF2 = 0.027, TFAM = 0.019, CKMT1A = 0.049, CKMT1B = 0.005. f, Representative images of hAd6 low and high visceral in vitro differentiated cultures. Green represents BODIPY staining, red represents MitoTracker staining, and blue represents Hoechst staining. Scale bars are 100 μm, in total 3 random images/sample were analyzed from 5 hAd6 low and 5 hAd6 high samples. g, Violin plot of sNuc-seq data showing axon guidance genes in adipocyte subclusters. h, Scatterplots showing the relationship between calculated proportion of visceral subpopulations hAd2 and hAd6 and expression of pan-neuronal markers on the ambient RNA of individual visceral sNuc-seq samples. For scatterplots, p values were calculated using an F-test with the null hypothesis that the slope = 0. Extended Data Fig. 12. Mouse adipocytes appear to have distinct functionality but are not analogous to human adipocyte subpopulations. a-b, UMAP projections of mouse adipocytes split by depot (a) and diet (b). c, Proportion of cells in each mouse cluster per sample split by depot, diet, and sex. For male mice n = 4 ING Chow, 4 ING HFD, 3 EPI Chow, and 5 EPI HFD. For female mice, n = 2 per condition. d, Expression of genes associated with known adipocyte functions in mouse adipocyte subclusters. e-k, Expression of genes associated with GO or KEGG pathways indicative of individual mouse adipocyte subclusters. l-n, Riverplots of mouse cells showing the association between mouse and human adipocyte clusters from both subcutaneous and visceral depots (l), subcutaneous (ING and SAT) adipocytes only (m) or visceral (PG and VAT) adipocytes only (n). For depot comparisons, both mouse query objects and human reference objects were subset to the respective depot before mapping. For bar graph, error bars represent SEM, * indicates credible depot effect and # indicates credible diet effect, calculated using mAd6 as reference. Extended Data Fig. 13. CellphoneDB identifies increasing numbers of cell-cell interactions within WAT during obesity. a, Heatmap showing number of significant interactions identified between cell types in SAT of low (<30) and high (>40) BMI individuals as determined by CellphoneDB. b, Expression of ligand and receptor genes from Figure 4b in human adipocyte subclusters. c, Heatmaps showing number of significant interactions identified between cell types in ING and PG WAT of chow and HFD fed mice. d, Venn diagrams showing the overlap of significant interactions between adipocytes and endothelial cells, ASPCs, and macrophages between depot, BMI/diet, and species. e, Jitter plots of the relationship between number of WAT cell types expressing a ligand (y axis) vs. the number of cell types expressing the receptor (x axis) for all significant interactions in high BMI human VAT (left) and mouse HFD PG (right). Extended Data Fig. 14. Association with GWAS data provides further insight into the contribution of white adipocytes to human traits. a-c, Expression of PPARG in human adipocyte subclusters (a), and in METSIM SAT bulk RNA-seq plotted against WHR (b) or HOMA-IR (c). d, Expression of PPARG in isolated subcutaneous adipocyte bulk RNA-seq plotted against HOMA-IR. e-h, SNPs in the PPARG gene identified by DEPICT as associated with BMI-adjusted WHR plotted against PPARG gene expression (e, g) and HOMA-IR (f, h) in isolated subcutaneous adipocyte bulk RNA-seq data and cohort. For rs17819328 n = 7 for T/T, 30 for T/G, and 6 for G/G. For rs1797912 n = 7 for A/A, 31 for A/C, and 5 for C/C. For box plots, horizontal line = median, lower and upper bounds of the box = 1st and 3rd quartile respectively, lower and upper whisker = 1st quartile – 1.5 x interquartile range (IQR) and 3rd quartile + 1.5 x IQR respectively. p values were calculated using a Wilcoxin test. i-j, Expression of genes in human adipocyte subtypes from sNuc-seq data (i) and from isolated subcutaneous adipocyte bulk RNA-seq plotted against LDL (j). k, p values of the association between mouse cell types and GWAS studies. l-m, p values of the association between mouse adipocyte (l) or ASPC (m) subclusters with GWAS studies. For all graphs, the grey line represents p = 0.05 and the orange line represents significant p value after Bonferroni adjustment (p = 0.003 for all cell, p = 0.001 for subclusters), calculated based on number of cell types queried. For scatterplots, p values were calculated using an F-test with the null hypothesis that the slope = 0. Extended Data Table 1. Subject information for Drop-Seq, sNuc-seq, and bulk RNA-seq of isolated subcutaneous human adipocytes. BMI, age, sex, race/ethnicity, depot, fat location, and surgery information for all Drop-Seq and sNuc-seq subjects as well as information for insulin sensitive and insulin resistant bulk RNA-seq adipocyte cohort. p values were calculated using two tailed t-tests with no adjustment for multiple comparisons. Subjects for Drop-Seq Subject BMI Age Sex Race/Ethnicity SAT Surgery Institution Hs235 36.04 53 F Caucasian Pannus Panniculectomy BIDMC Hs236 25.74 35 F Caucasian Thigh Thighplasty BIDMC Hs237 22.59 53 F Caucasian Pannus DIEP BIDMC Hs238 19.57 49 F Caucasian Pannus Abdominoplasty BIDMC Hs239 24.8 71 F Caucasian Pannus DIEP BIDMC Hs240 25.82 59 F Caucasian Pannus Panniculectomy BIDMC Hs242 22.88 59 F Caucasian Pannus DIEP BIDMC Hs248 32.28 68 F Caucasian Pannus Panniculectomy BIDMC Hs249 26.46 54 F Caucasian Pannus DIEP BIDMC DIEP: Deep inferior epigastric perforators Subjects for sNuc-seq Subject BMI Age Sex Race/ Ethnicity SAT VAT Surgery Institution Hs001 49.3 29 F Caucasian Periumbilical Omental VSG UPitt Hs002 33.1 57 F Caucasian Periumbilical NA Hernia UPitt Hs004 25.4 51 F Caucasian Periumbilical NA CCY UPitt Hs009 45.7 41 F Black Periumbilical Omental VSG UPitt Hs010 43.1 35 F Caucasian Periumbilical Omental RYGB UPitt Hs011 42.8 58 F Black Periumbilical NA VSG UPitt Hs012 48.7 36 M Caucasian Periumbilical Omental VSG UPitt Hs013 43.2 24 M Caucasian Periumbilical Omental VSG UPitt Hs253 30.04 53 F Caucasian Periumbilical Preperitoneal TAH BSO BIDMC Hs254 23.96 41 F Caucasian/Hispanic Periumbilical Preperitoneal TAH BSO BIDMC Hs255 24.27 73 F Caucasian Periumbilical Preperitoneal TAH BSO BIDMC Hs256 34.53 41 F Black Periumbilical Omental CCY BIDMC Hs266 22.15 68 M Caucasian Periumbilical Omental Colon polyp BIDMC Bulk RNA-seq of floated adipocytes Insulin Sensitive average(min-max) Insulin Resistant average(min-max) p Value N 16 27 AGE 47.3 (36-63) 50.6 (33-71) 0.289 BMI 27.2 (21-33) 30.1 (21-42) 0.042 HOMA-IR 0.70 (0.46-0.88) 5.8 (2.1-24.5) 0.00012 HDL 70.5 (42-154) 54.1 (26-100) 0.022 LDL 93.2 (54-133) 97.9 (51-169) 0.651 Abbreviations: VSG: Vertical sleeve gastrectomy; CCY: Cholecystectomy; RYGB: Roux en Y gastric bypass; TAH BSO: Total abdominal hysterectomy and bilateral salpingo-oophorectomy Extended Data Table 2. Numbers of cells in human and mouse single cell experiments broken down by cluster, depot, BMI/diet, and technology. Cell counts per cluster for human and mouse data broken down by technology, depot, and BMI/diet. Human Cell Numbers VAT SAT VAT total SAT total Total sNuc sNuc Drop BMI < 30 30-40 > 40 < 30 30-40 > 40 < 30 > 30 Adipocyte 5211 1011 5253 7611 2847 3938 0 0 11475 14396 25871 ASPCs 5938 1404 7304 6848 2703 7329 15195 5761 14646 37836 52482 Mesothelium 7773 1927 20782 0 0 0 0 0 30482 0 30482 Endothelial 2351 1030 2345 4231 2783 2059 577 107 5726 9757 15483 Lymphatic Endo 677 240 1138 195 130 305 168 48 2055 846 2901 Pericyte 381 109 254 353 132 172 60 3 744 720 1464 Smooth Muscle 448 360 423 709 621 237 83 5 1231 1655 2886 Macrophage 1908 630 6328 3121 1795 2871 1256 403 8866 9446 18312 Monocyte 98 41 173 187 155 549 359 387 312 1637 1949 Dendritic Cell 125 30 340 169 119 188 756 714 495 1946 2441 Mast Cell 111 27 139 210 294 298 66 23 277 891 1168 Neutrophil 7 9 4 98 12 14 0 2 20 126 146 B Cell 28 12 39 57 49 188 30 26 79 350 429 NK Cell 229 92 242 375 279 669 297 446 563 2066 2629 T Cell 762 382 1661 667 510 1522 977 713 2805 4389 7194 Endometrium 45 150 114 0 0 0 2 1 309 3 312 Total 26092 7454 46539 24831 12429 20339 19826 8639 80085 86064 166149 Mouse Cell Numbers PG Ing Chow HFD Chow HFD PG Total Ing Total Total Adipocyte 12874 5139 8645 13276 18013 21921 39934 ASPCs 9928 10194 16308 14797 20122 31105 51227 Mesothelium 10074 4873 0 0 14947 0 14947 Endothelial 1521 673 1141 2261 2194 3402 5596 Lymphatic Endo 678 101 224 173 779 397 1176 Pericyte 62 170 56 309 232 365 597 Smooth Muscle 56 52 30 125 108 155 263 Macrophage 3788 35673 9370 9017 39461 18387 57848 Monocyte 975 2801 1286 2545 3776 3831 7607 Dendritic Cell 268 688 237 379 956 616 1572 Mast Cell 4 267 13 27 271 40 311 Neutrophil 23 9 8 7 32 15 47 B Cell 301 594 28 279 895 307 1202 NK Cell 110 215 67 282 325 349 674 T Cell 266 472 69 479 738 548 1286 Male Epithelial 3463 36 19 329 3499 348 3847 Female Epithelial 76 45 6331 3135 121 9466 9587 Total 44467 62002 43832 47420 106469 91252 197721 Extended Data Table 3. GWAS studies used for CELLECT analysis. List of sources for GWAS datasets used in the CELLECT analysis. Trait Study/collection BMI Pulit, S. L. et al. Meta-analysis of genome-wide association studies for body fat distribution in 694 649 individuals of European ancestry. HDL https://alkesgroup.broadinstitute.org/sumstats_formatted/ LDL https://alkesgroup.broadinstitute.org/sumstats_formatted/ T1D https://alkesgroup.broadinstitute.org/sumstats_formatted/ T2D (BMI adjusted) Mahajan, A. et al. Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps. Triglycerides https://alkesgroup.broadinstitute.org/sumstats_formatted/ WHR (BMI adjusted) Loh, P.-R., Kichaev, G., Gazal, S., Schoech, A. P. & Price, A. L. Mixed-model association for biobank-scale datasets Supplementary Material 1825546_SI_Guide 1825546_Sup_Info 1825546_Sup_Tab_1 1825546_Sup_Tab_2 1825546_Sup_Tab_3 1825546_Sup_Tab_4 1825546_Sup_Tab_5 1825546_Sup_Tab_6 1825546_Sup_Tab_7 1825546_SD_Fig_2 1825546_SD_Fig_3 1825546_SD_Sup_Fig_3 1825546_SD_ED_Fig_1 1825546_SD_ED_Fig_2 1825546_SD_ED_Fig_3 1825546_SD_ED_Fig_5 1825546_SD_ED_Fig_6 1825546_SD_ED_Fig_8 1825546_SD_ED_Fig_9 1825546_SD_ED_Fig_10 1825546_SD_ED_Fig_11 1825546_SD_ED_Fig_12 ACKNOWLEDGEMENTS This work was supported by NIH grants RC2 DK116691 to EDR, LTT, AC, OA, and AR, AHA POST14540015 and DoD PRMRP-DAW81XWH to LTT, Broad-BADERC Collaboration Initiative Award (NIH 5P30DK057521) to LTT and EDR, and R01 DK102173 to EDR. MPE is supported by NIH grant F32DK124914. Additional support includes PRIN 2017 (Italian Ministry of University, #2017L8Z2EM) to AG, THP acknowledges the Novo Nordisk Foundation (unconditional donation to the Novo Nordisk Foundation Center for Basic Metabolic Research; grant number NNF18CC0034900) and the Lundbeck Foundation (Grant number R190-2014-3904), grants AMP-T2D RFB8b (FNIH) and UM1DK126185 (NIDDK) to MC, Sarnoff Cardiovascular Research Foundation Fellowship to S.A., grants 1K08HG010155 and 1U01HG011719 to A.V.K. from the National Human Genome Research Institute, and a sponsored research agreement from IBM Research to the Broad Institute of MIT and Harvard to A.V.K. All single cell library construction and sequencing was performed through the Boston Nutrition Obesity Research Center Functional Genomics and Bioinformatics Core (NIH P30DK046200). We thank Christina Usher for artistic support and Miriam Udler for helpful discussions. DATA AVAILABILITY Single cell RNA expression and count data is deposited in the Single Cell Portal (Study #SCP1376). Processed count data for bulk RNA-seq and dge matrices for single cell and single nucleus RNA-seq have been deposited in GEO (Bulk-seq Accession #GSE174475, sc-RNA-seq Accession # GSE176067, sNuc-sec Acession #GSE176171), raw sequencing reads for mouse data are available in SRA, study #SRP322736. FASTQ and SNP array files for human samples are deposited in dbGaP, Accession #phs002766.v1.p1. Publicaly available datasets and databases used were the following: METSIM RNA-seq data from Raulerson et. al.13 (GSE135134); single cell ASPC data from Burl et. al.4 (SRP145475), Hepler et. al.6 (GSE111588), Merrick et. al.5 (GSE128889), Sárvári et. al.9 (GSE160729), and Schwalie et. al.3 (E-MTAB-6677); human assembly GRCh38 and GENCODE annotation 27 (https://www.gencodegenes.org/human/release_27.html); mouse assembly GRCm38 and GENCODE annotation M16 (https://www.gencodegenes.org/mouse/release_M16.html). Fig. 1. A single cell atlas of human white adipose tissue. a, Schematic of workflows for scRNA-seq and sNuc-seq of human WAT. b, Graphical representation of the cohorts for both studies. Only the sNuc-seq cohort contains VAT. c, UMAP projection of all 166,129 sequenced human cells split by cohort. d, Marker genes for each cell population in the human WAT dataset. Fig. 2. A single cell atlas of mouse white adipose tissue. a, Schematic of workflow for sNuc-seq of mouse ING and PG adipose tissue. a, Body weight of chow and high fat fed animals used for sNuc-seq (n = 5 chow and 5 HFD male mice, 2 chow and 2 HFD female mice). Error bars represent standard error of the mean (SEM). c, UMAP projection of all 197,721 sequenced mouse cells split by diet. d, Marker genes for each cell population in the mouse WAT dataset. Fig. 3. Subclustering of human and mouse adipocytes reveals multiple distinct populations that vary across depot and diet. a, UMAP projection of clusters formed by 25,871 human white adipocytes. b, Expression of adipocyte marker ADIPOQ and specific marker genes for each adipocyte subpopulation. c, IHC for marker genes of adipocyte subpopulations hAd4, hAd5, hAd6, and hAd7 in human adipose tissue and percentage of positive adipocytes per slide in lean and obese individuals (GRIA4: 5 lean, 5 obese, 2 slides each; PGAP1: 5 lean SAT, 4 obese SAT, 3 lean VAT, 4 obese VAT, 1 slide each; EBF2: 3 lean, 4 obese, 2 slides each; AGMO: 4 lean, 4 obese, 2 slides each). Scale bars are 25 μm for GRIA4, EBF2, and AGMO, 20 μm for PGAP1. d, Estimated proportions of adipocyte subpopulations in bulk RNA sequencing data of enzymatically isolated subcutaneous adipocytes from 43 individuals plotted against BMI. p values were calculated using an F-test (with null hypothesis slope = 0), error bands represent a confidence level of 0.95. e, Representative images of ex vivo differentiated human subcutaneous adipocytes predicted to have low or high hAd3 content based on deconvolution of bulk RNA sequencing data. Green represents BODIPY staining, blue represents Hoechst staining. Scale bars are 100 μm. f, Normalized count of BODIPY-related features in ex vivo differentiated adipocytes stratified into hAd3 low and high populations. Points represent normalized feature for cultures derived from individual subjects, n = 8 hAd3 low SAT, 4 hAd3 high SAT, 19 hAd3 low VAT, 4 hAd3 high VAT. g, UMAP projection 39,934 mouse white adipocytes. h, Expression of Adipoq and marker genes for each mouse adipocyte subpopulation. For bar graphs, error bars represent SEM, p values were calculated using two tailed t-tests with no correction for multiple comparisons. Fig. 4. Extensive cell-cell interactions in WAT and associations with human disease traits. a, Heatmap showing number of significant interactions identified between cell types in VAT of low (<30) and high (>40) BMI individuals as determined by CellphoneDB. b, Selected interactions between adipocytes and ASPCs, endothelial cells, and macrophages identified using CellphoneDB; orange and green indicate interactions that are significant only in BMI > 40 or only in BMI >30, respectively. c, CELLECT p values of the association between cell types in the human adipose sNuc-seq dataset with GWAS studies. The grey line represents p = 0.05 and the orange line represents significant p value after Bonferroni adjustment (p = 0.003), based on number of cell types queried. Both T2D and WHR were BMI-adjusted. d, CELLECT p values for adipocyte subpopulations. The grey line represents p = 0.05 and the orange line represents significant p value after Bonferroni adjustment (p = 0.001), based on all cell subtypes queried. e, Estimated cell type proportion of hAd7 in bulk RNA-seq data of enzymatically isolated subcutaneous adipocytes from 43 individuals plotted against HOMA-IR. For line of best fit, R2 = 0.11, the error band represents a confidence level of 0.95. f-g, Expression of hAd7 marker genes negatively correlated with HOMA-IR in human adipocyte subpopulations (f) and bulk RNA sequencing data of human adipocytes (g). For scatterplots, p values were calculated using an F-test with the null hypothesis that the slope = 0. CODE AVAILABILITY Data analysis pipelines used in this study for processing of raw sequencing data, integration, and clustering can be obtained from https://gitlab.com/rosen-lab/white-adipose-atlas. COMPETING INTEREST DECLARATION S.A. has served as a scientific consultant to Third Rock Ventures. A.V.K. has served as a scientific advisor to Sanofi, Amgen, Maze Therapeutics, Navitor Pharmaceuticals, Sarepta Therapeutics, Novartis, Verve Therapeutics, Silence Therapeutics, Veritas International, Color Health, Third Rock Ventures, and Columbia University (NIH); received speaking fees from Illumina, MedGenome, Amgen, and the Novartis Institute for Biomedical Research; and received a sponsored research agreement from the Novartis Institute for Biomedical Research. M.C. holds equity in Waypoint Bio and is a member of the Nestle Scientific Advisory Board. A.R. is a co-founder and equity holder of Celsius Therapeutics, an equity holder in Immunitas Therapeutics and a scientific advisory board member of Thermo Fisher Scientific, Syros Pharmaceuticals, Asimov and Neogene Therapeutics. A.R. is also an employee of Genentech. All other authors declare no competing interests. REFERENCES 1. Rosen ED & Spiegelman BM What We Talk About When We Talk About Fat. Cell 156 , 20–44 (2014).24439368 2. Kahn SE , Hull RL & Utzschneider KM Mechanisms linking obesity to insulin resistance and type 2 diabetes. Nature 444 , 840–846 (2006).17167471 3. Schwalie PC A stromal cell population that inhibits adipogenesis in mammalian fat depots. Nature 559 , 103–108 (2018).29925944 4. Burl RB Deconstructing Adipogenesis Induced by β3-Adrenergic Receptor Activation with Single-Cell Expression Profiling. Cell Metab. 28 , 300–309.e4 (2018).29937373 5. Merrick D Identification of a mesenchymal progenitor cell hierarchy in adipose tissue. Science 364 , (2019). 6. Hepler C Identification of functionally distinct fibro-inflammatory and adipogenic stromal subpopulations in visceral adipose tissue of adult mice. eLife 7 , e39636 (2018).30265241 7. Vijay J Single-cell analysis of human adipose tissue identifies depot- and disease-specific cell types. Nat. Metab 2 , 97–109 (2020).32066997 8. Rajbhandari P Single cell analysis reveals immune cell–adipocyte crosstalk regulating the transcription of thermogenic adipocytes. eLife 8 , e49501 (2019).31644425 9. Sárvári AK Plasticity of Epididymal Adipose Tissue in Response to Diet-Induced Obesity at Single-Nucleus Resolution. Cell Metab. 33 , 437–453.e5 (2021).33378646 10. Sun W snRNA-seq reveals a subpopulation of adipocytes that regulates thermogenesis. Nature 587 , 98–102 (2020).33116305 11. Benites-Zapata VA High waist-to-hip ratio levels are associated with insulin resistance markers in normal-weight women. Diabetes Metab. Syndr. Clin. Res. Rev 13 , 636–642 (2019). 12. Wang X , Park J , Susztak K , Zhang NR & Li M Bulk tissue cell type deconvolution with multi-subject single-cell expression reference. Nat. Commun 10 , 380 (2019).30670690 13. Raulerson CK Adipose Tissue Gene Expression Associations Reveal Hundreds of Candidate Genes for Cardiometabolic Traits. Am. J. Hum. Genet 105 , 773–787 (2019).31564431 14. Blüher M Transgenic animal models for the study of adipose tissue biology. Best Pract. Res. Clin. Endocrinol. Metab 19 , 605–623 (2005).16311220 15. An atlas of cell types in the mouse epididymis and vas deferens ∣ eLife. https://elifesciences.org/articles/55474. 16. Villani A-C Single-cell RNA-seq reveals new types of human blood dendritic cells, monocytes, and progenitors. Science 356 , (2017). 17. Hildreth AD Single-cell sequencing of human white adipose tissue identifies new cell states in health and obesity. Nat. Immunol 1–15 (2021) doi:10.1038/s41590-021-00922-4.33335328 18. Suganami T & Ogawa Y Adipose tissue macrophages: their role in adipose tissue remodeling. J. Leukoc. Biol 88 , 33–39 (2010).20360405 19. Weisberg SP Obesity is associated with macrophage accumulation in adipose tissue. J. Clin. Invest 112 , 1796–1808 (2003).14679176 20. Reilly SM & Saltiel AR Adapting to obesity with adipose tissue inflammation. Nat. Rev. Endocrinol 13 , 633–643 (2017).28799554 21. Shi M & Shi G-P Different Roles of Mast Cells in Obesity and Diabetes: Lessons from Experimental Animals and Humans. Front. Immunol 3 , 7 (2012).22566893 22. Xu H Chronic inflammation in fat plays a crucial role in the development of obesity-related insulin resistance. J. Clin. Invest 112 , 1821–1830 (2003).14679177 23. Ferrero R , Rainer P & Deplancke B Toward a Consensus View of Mammalian Adipocyte Stem and Progenitor Cell Heterogeneity. Trends Cell Biol. 30 , 937 (2020).33148396 24. Wang QA , Tao C , Gupta RK & Scherer PE Tracking adipogenesis during white adipose tissue development, expansion and regeneration. Nat. Med 19 , 1338–1344 (2013).23995282 25. Jeffery E , Church CD , Holtrup B , Colman L & Rodeheffer MS Rapid Depot-Specific Activation of Adipocyte Precursor Cells at the Onset of Obesity. Nat. Cell Biol 17 , 376–385 (2015).25730471 26. Bäckdahl J Spatial mapping reveals human adipocyte subpopulations with distinct sensitivities to insulin. Cell Metab. 33 , 1869–1882.e6 (2021).34380013 27. Stefan N Circulating Palmitoleate Strongly and Independently Predicts Insulin Sensitivity in Humans. Diabetes Care 33 , 405–407 (2010).19889804 28. Laber S Discovering cellular programs of intrinsic and extrinsic drivers of metabolic traits using LipocyteProfiler. 2021.07.17.452050 https://www.biorxiv.org/content/10.1101/2021.07.17.452050v1 (2021) doi:10.1101/2021.07.17.452050. 29. Rajakumari S EBF2 determines and maintains brown adipocyte identity. Cell Metab. 17 , 562–574 (2013).23499423 30. Pulit SL Meta-analysis of genome-wide association studies for body fat distribution in 694 649 individuals of European ancestry. Hum. Mol. Genet 28 , 166–174 (2019).30239722 31. Agrawal S Inherited basis of visceral, abdominal subcutaneous and gluteofemoral fat depots. 2021.08.24.21262564 https://www.medrxiv.org/content/10.1101/2021.08.24.21262564v1 (2021) doi:10.1101/2021.08.24.21262564. 32. Willows JW Visualization and analysis of whole depot adipose tissue neural innervation. iScience 24 , 103127 (2021).34622172 34. Roh HC Adipocytes fail to maintain cellular identity during obesity due to reduced PPARγ activity and elevated TGFβ-SMAD signaling. Mol. Metab 42 , 101086 (2020).32992037 36. Park J Progenitor-like characteristics in a subgroup of UCP1+ cells within white adipose tissue. Dev. Cell 56 , 985–999.e4 (2021).33711247 37. Priest C & Tontonoz P Inter-organ cross-talk in metabolic syndrome. Nat. Metab 1 , 1177–1188 (2019).32694672 38. Schling P & Löffler G Cross talk between adipose tissue cells: impact on pathophysiology. News Physiol. Sci. Int. J. Physiol. Prod. Jointly Int. Union Physiol. Sci. Am. Physiol. Soc 17 , 99–104 (2002). 39. Kane H & Lynch L Innate Immune Control of Adipose Tissue Homeostasis. Trends Immunol. 40 , 857–872 (2019).31399336 40. Efremova M , Vento-Tormo M , Teichmann SA & Vento-Tormo R CellPhoneDB: inferring cell-cell communication from combined expression of multi-subunit ligand-receptor complexes. Nat. Protoc 15 , 1484–1506 (2020).32103204 41. Cao Y Angiogenesis and vascular functions in modulation of obesity, adipose metabolism, and insulin sensitivity. Cell Metab. 18 , 478–489 (2013).24035587 42. Hubert A Selective Deletion of Leptin Signaling in Endothelial Cells Enhances Neointima Formation and Phenocopies the Vascular Effects of Diet-Induced Obesity in Mice. Arterioscler. Thromb. Vasc. Biol 37 , 1683–1697 (2017).28705795 43. Scott RA An Expanded Genome-Wide Association Study of Type 2 Diabetes in Europeans. Diabetes 66 , 2888–2902 (2017).28566273 44. Shungin D New genetic loci link adipose and insulin biology to body fat distribution. Nature 518 , 187–196 (2015).25673412 45. Huang LO Genome-wide discovery of genetic loci that uncouple excess adiposity from its comorbidities. Nat. Metab 3 , 228–243 (2021).33619380 46. Timshel PN , Thompson JJ & Pers TH Genetic mapping of etiologic brain cell types for obesity. eLife 9 , e55851 (2020).32955435 47. O’Rahilly S & Farooqi IS Human obesity as a heritable disorder of the central control of energy balance ∣ International Journal of Obesity. https://www.nature.com/articles/ijo2008239. 48. Sailer S , Keller MA , Werner ER & Watschinger K The Emerging Physiological Role of AGMO 10 Years after Its Gene Identification. Life Basel Switz. 11 , (2021). 49. Dupuis J New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nat. Genet 42 , 105–116 (2010).20081858 50. Loh NY RSPO3 impacts body fat distribution and regulates adipose cell biology in vitro. Nat. Commun 11 , 2797 (2020).32493999 METHODS REFERENCES 51. Chi J Three-Dimensional Adipose Tissue Imaging Reveals Regional Variation in Beige Fat Biogenesis and PRDM16-Dependent Sympathetic Neurite Density. Cell Metab. 27 , 226–236.e3 (2018).29320703 52. Katz A Quantitative Insulin Sensitivity Check Index: A Simple, Accurate Method for Assessing Insulin Sensitivity In Humans. J. Clin. Endocrinol. Metab 85 , 2402–2410 (2000).10902785 53. Matthews DR Homeostasis model assessment: insulin resistance and β-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28 , 412–419 (1985).3899825 54. Macosko EZ Highly Parallel Genome-wide Expression Profiling of Individual Cells Using Nanoliter Droplets. Cell 161 , 1202–1214 (2015).26000488 55. Drokhlyansky E The Human and Mouse Enteric Nervous System at Single-Cell Resolution. Cell 182 , 1606–1622.e23 (2020).32888429 56. Slyper M A single-cell and single-nucleus RNA-Seq toolbox for fresh and frozen human tumors. Nat. Med 26 , 792–802 (2020).32405060 57. Delorey TM A single-cell and spatial atlas of autopsy tissues reveals pathology and cellular targets of SARS-CoV-2. bioRxiv (2021) doi:10.1101/2021.02.25.430130. 58. Dobin A STAR: ultrafast universal RNA-seq aligner. Bioinforma. Oxf. Engl 29 , 15–21 (2013). 59. CellBender remove-background: a deep generative model for unsupervised removal of background noise from scRNA-seq datasets ∣ bioRxiv. https://www.biorxiv.org/content/10.1101/791699v1. 60. Lun ATL EmptyDrops: distinguishing cells from empty droplets in droplet-based single-cell RNA sequencing data. Genome Biol. 20 , 63 (2019).30902100 61. Wolock SL , Lopez R & Klein AM Scrublet: Computational Identification of Cell Doublets in Single-Cell Transcriptomic Data. Cell Syst. 8 , 281–291.e9 (2019).30954476 62. Patro R , Duggal G , Love MI , Irizarry RA & Kingsford C Salmon provides fast and bias-aware quantification of transcript expression. Nat. Methods 14 , 417–419 (2017).28263959 63. Srivastava A Alignment and mapping methodology influence transcript abundance estimation. Genome Biol. 21 , 239 (2020).32894187 64. Soneson C , Love MI & Robinson MD Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences. F1000Research 4 , 1521 (2015).26925227 65. Stuart T Comprehensive integration of single-cell data. Cell 177 , 1888–1902.e21 (2019).31178118 66. Hafemeister C & Satija R Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression. Genome Biol. 20 , 296 (2019).31870423 67. Hao Y Integrated analysis of multimodal single-cell data. bioRxiv 2020.10.12.335331 (2020) doi:10.1101/2020.10.12.335331. 68. Yu G , Wang L-G , Han Y & He Q-Y clusterProfiler: an R Package for Comparing Biological Themes Among Gene Clusters. OMICS J. Integr. Biol 16 , 284–287 (2012). 69. Littlejohns TJ The UK Biobank imaging enhancement of 100,000 participants: rationale, data collection, management and future directions. Nat. Commun 11 , 2624 (2020).32457287 70. Sudlow C UK Biobank: An Open Access Resource for Identifying the Causes of a Wide Range of Complex Diseases of Middle and Old Age. PLOS Med. 12 , e1001779 (2015).25826379 71. Agrawal S Association of machine learning-derived measures of body fat distribution in >40,000 individuals with cardiometabolic diseases. 2021.05.07.21256854 https://www.medrxiv.org/content/10.1101/2021.05.07.21256854v1 (2021) doi:10.1101/2021.05.07.21256854. 72. Kichaev G Leveraging Polygenic Functional Enrichment to Improve GWAS Power. Am. J. Hum. Genet 104 , 65–75 (2019).30595370 73. Pruim RJ LocusZoom: regional visualization of genome-wide association scan results. Bioinforma. Oxf. Engl 26 , 2336–2337 (2010). 74. Timshel PN , Thompson JJ & Pers TH Genetic mapping of etiologic brain cell types for obesity. eLife 9 , e55851 (2020).32955435 75. Mahajan A Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps. Nat. Genet 50 , 1505–1513 (2018).30297969 76. Loh P-R , Kichaev G , Gazal S , Schoech AP & Price AL Mixed-model association for biobank-scale datasets. Nat. Genet 50 , 906–908 (2018).29892013 77. Finucane HK Partitioning heritability by functional annotation using genome-wide association summary statistics. Nat. Genet 47 , 1228–1235 (2015).26414678 78. Teslovich TM Biological, clinical and population relevance of 95 loci for blood lipids. Nature 466 , 707–713 (2010).20686565 79. Bradfield JP A Genome-Wide Meta-Analysis of Six Type 1 Diabetes Cohorts Identifies Multiple Associated Loci. PLOS Genet. 7 , e1002293 (2011).21980299 80. Loh P-R Reference-based phasing using the Haplotype Reference Consortium panel. Nat. Genet 48 , 1443–1448 (2016).27694958 81. Das S Next-generation genotype imputation service and methods. Nat. Genet 48 , 1284–1287 (2016).27571263 82. Robinson MD , McCarthy DJ & Smyth GK edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26 , 139–140 (2010).19910308 83. Büttner M , Ostner J , Müller C , Theis F & Schubert B scCODA: A Bayesian model for compositional single-cell data analysis. bioRxiv 2020.12.14.422688 (2020) doi:10.1101/2020.12.14.422688. 84. Cawthorn WP , Scheller EL & MacDougald OA Adipose tissue stem cells meet preadipocyte commitment: going back to the future[S]. J. Lipid Res 53 , 227–246 (2012).22140268

PMC009xxxxxx/PMC9511123.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101088070 22479 Nano Lett Nano Lett Nano letters 1530-6984 1530-6992 35786891 9511123 10.1021/acs.nanolett.2c02019 NIHMS1835298 Article Locking and Unlocking Thrombin Function Using Immunoquiescent Nucleic Acid Nanoparticles with Regulated Retention In Vivo Ke Weina Nanoscale Science Program, Department of Chemistry, University of North Carolina at Charlotte, Charlotte, North Carolina 28223, United States Chandler Morgan http://orcid.org/0000-0003-3078-6000 Nanoscale Science Program, Department of Chemistry, University of North Carolina at Charlotte, Charlotte, North Carolina 28223, United States Cedrone Edward Nanotechnology Characterization Lab., Cancer Research Technology Program, Frederick National Laboratory for Cancer Research sponsored by the National Cancer Institute, Frederick, Maryland 21702, United States Saito Renata F. Centro de Investigação Translacional em Oncologia (LIM24), Departamento de Radiologia e Oncologia, Faculdade de Medicina da Universidade de São Paulo and Instituto do Câncer do Estado de São Paulo, São Paulo, SP 01246-903, Brazil Rangel Maria Cristina Centro de Investigação Translacional em Oncologia (LIM24), Departamento de Radiologia e Oncologia, Faculdade de Medicina da Universidade de São Paulo and Instituto do Câncer do Estado de São Paulo, São Paulo, SP 01246-903, Brazil Junqueira Mara de Souza Centro de Investigação Translacional em Oncologia (LIM24), Departamento de Radiologia e Oncologia, Faculdade de Medicina da Universidade de São Paulo and Instituto do Câncer do Estado de São Paulo, São Paulo, SP 01246-903, Brazil Wang Jian Department of Pharmacology, Department of Biochemistry & Molecular Biology, Penn State College of Medicine, Hershey, Pennsylvania 17033, United States Shi Da Nanotechnology Characterization Lab., Cancer Research Technology Program, Frederick National Laboratory for Cancer Research sponsored by the National Cancer Institute, Frederick, Maryland 21702, United States Truong Nguyen Nanoscale Science Program, Department of Chemistry, University of North Carolina at Charlotte, Charlotte, North Carolina 28223, United States Richardson Melina Nanoscale Science Program, Department of Chemistry, University of North Carolina at Charlotte, Charlotte, North Carolina 28223, United States Rolband Lewis A. http://orcid.org/0000-0001-9085-8753 Nanoscale Science Program, Department of Chemistry, University of North Carolina at Charlotte, Charlotte, North Carolina 28223, United States Dréau Didier Department of Biological Sciences, University of North Carolina at Charlotte, Charlotte, North Carolina 28223, United States Bedocs Peter Department of Anesthesiology, School of Medicine, Uniformed Services University of the Health Sciences, Bethesda, Maryland 20814, United States; Henry M. Jackson Foundation for the Advancement of Military Medicine, Bethesda, Maryland 20817, United States Chammas Roger Nanoscale Science Program, Department of Chemistry, University of North Carolina at Charlotte, Charlotte, North Carolina 28223, United States; Centro de Investigação Translacional em Oncologia (LIM24), Departamento de Radiologia e Oncologia, Faculdade de Medicina da Universidade de São Paulo and Instituto do Câncer do Estado de São Paulo, São Paulo, SP 01246-903, Brazil Dokholyan Nikolay V. http://orcid.org/0000-0002-8225-4025 Department of Pharmacology, Department of Biochemistry & Molecular Biology, Penn State College of Medicine, Hershey, Pennsylvania 17033, United States; Department of Chemistry, Department of Biomedical Engineering, Penn State University, University Park, Pennsylvania 16802, United States Dobrovolskaia Marina A. http://orcid.org/0000-0002-4233-9227 Nanotechnology Characterization Lab., Cancer Research Technology Program, Frederick National Laboratory for Cancer Research sponsored by the National Cancer Institute, Frederick, Maryland 21702, United States Afonin Kirill A. http://orcid.org/0000-0002-6917-3183 Nanoscale Science Program, Department of Chemistry, University of North Carolina at Charlotte, Charlotte, North Carolina 28223, United States Corresponding Author: Kirill A. Afonin – Nanoscale Science Program, Department of Chemistry, University of North Carolina at Charlotte, Charlotte, North Carolina 28223, United States; Phone: +1 704 687 0685; kafonin@uncc.edu 10 9 2022 27 7 2022 05 7 2022 27 7 2023 22 14 59615972 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. The unbalanced coagulation of blood is a life-threatening event that requires accurate and timely treatment. We introduce a user-friendly biomolecular platform based on modular RNA-DNA anticoagulant fibers programmed for reversible extra-cellular communication with thrombin and subsequent control of anticoagulation via a “kill-switch” mechanism that restores hemostasis. To demonstrate the potential of this reconfigurable technology, we designed and tested a set of anticoagulant fibers that carry different thrombin-binding aptamers. All fibers are immunoquiescent, as confirmed in freshly collected human peripheral blood mononuclear cells. To assess interindividual variability, the anticoagulation is confirmed in the blood of human donors from the U.S. and Brazil. The anticoagulant fibers reveal superior anticoagulant activity and prolonged renal clearance in vivo in comparison to free aptamers. Finally, we confirm the efficacy of the “kill-switch” mechanism in vivo in murine and porcine models. Graphical Abstract anticoagulation RNA-DNA fibers aptamers immunoquiescent kill-switch in vivo pmcThe major function of the blood coagulation system is to maintain hemostasis by preventing bleeding. In addition, the coagulation system plays an essential role in innate immunity.1 Coagulation triggered by damage to blood vessels or initiated by pathogens helps prevent blood loss or the spread of infection, respectively. When a balance between the pro- and anticoagulant arms of the coagulation system is altered, its initially protective function can have harmful consequences to the host. For example, consumptive coagulopathy, also known as disseminated intravascular coagulation, is common in sepsis, allergic and autoimmune responses, tissue injury, and cancer.2,3 Pathologic or excessive coagulation can lead to thrombosis and result in myriad downstream symptoms.4 Complications depend on where thrombosis occurs and are of increasing likelihood with age, which in combination with other lifestyle risk factors5 make it a leading cause of global mortality.4 Also, coagulopathy and thromboembolic events are the main causes of death in severe and critical COVID-19 patients.6,7 Anticoagulants or blood thinners are routinely used in short-term surgical procedures and long-term treatments of thrombosis. Currently, there are several commercially available anticoagulants, including vitamin K antagonists (e.g., Coumadin), heparins (e.g., Fragmin),8 Factor Xa inhibitors (e.g., Xarelto), and direct thrombin inhibitors (e.g., Angiomax).4,9 Side effects with anticoagulant therapies include excessive bleeding, passage of blood in urine, severe bruising, and bleeding gums, among others.5 The efficacy of these anticoagulants is also dependent on the specific thrombosis indication as well as drug−drug interactions and patient conditions, and large-scale trials are needed to explore efficacy with diversely represented patient cofactors.4 Moreover, the patient’s immune system often develops antibodies that neutralize anticoagulants, rendering them ineffective.10 Therefore, novel, effective, safe, and low-cost treatments of thrombosis with reversible control over the coagulation process are in high demand. Aptamers are single-stranded oligonucleotides selected in vitro for high affinity binding to target molecules.11 The high specificity, low cost, batch-to-batch consistency, and biodegradability of aptamers make them a promising class of therapeutics.12 Moreover, unlike protein and polysaccharide-based anticoagulants, aptamers have been shown to be nonimmunogenic, likely due to the natural immune tolerance to nucleic acids.13 Several aptamers selected to target coagulation cascade proteins have been reported,14 and the use of their reverse complement sequences has been shown as a reliable strategy for controlled restoration of thrombin activity.15–22 All three pathways that contribute to the blood coagulation cascade23 share thrombin as a central protein, making it a promising target for the continued development of anticoagulants. Thrombin is a globular enzyme with three functional binding sites: the protease catalytic site, the anionic fibrinogen recognition site (exosite I), and the anionic heparin binding site (exosite II).24 Exosites I and II are located on opposite sides of thrombin and mediate its interactions with cofactors and substrates, while the catalytic site enables the serine protease activity of thrombin.25 The first and the most studied thrombin-binding aptamer (HD1/ARC183) was selected to inhibit coagulation by blocking the function of exosite I on thrombin.26,27 Despite its medicinal promise, the suboptimal dosing and rapid clearance of ARC183 have halted further clinical trials of this formulation.28 To improve the potency, numerous modifications of ARC183 have been introduced (e.g., RA-3629,30 and NU17231–33) but their short postinfusion half-lives with quickly restorable coagulation34 hamper broader biomedical applications. To improve the operation of current anticoagulants, we introduce a dynamic platform based on RNA-DNA fibers35 designed for the efficient and reversible control of blood coagulation. These nanoassemblies contain multiple thrombin-binding aptamers to substantially increase their molecular weight, prolong blood stability, and increase retention time in vivo. Another unique feature of this molecular system stems from its ability to be conditionally deactivated via a “kill-switch” mechanism that reverses its anticoagulant function and produces low-molecular-weight assemblies that undergo rapid renal excretion (Figure 1). With a minimal set of short, chemically synthesized oligos, we engineered 12 distinct anticoagulant fibers that carry NU172 or RA-36 or combinations thereof (Figure S1). We demonstrated the effective inhibition of human plasma coagulation using the constructs and reversal of this effect through the introduction of kill-switches. To address potential demographic and interdonor variability, all coagulation experiments were carried out with fresh blood samples from human donors in the United States or in Brazil. To address safety concerns, we analyzed immunological profiles and toxicities of nanodevices in human peripheral blood mononuclear cells (PBMCs) freshly collected from healthy donors. The biodistribution, retention time, and anticoagulant function of fibers and kill-switches were compared in murine and porcine models. We conclude that anticoagulant fibers offer (i) simple design and assembly protocols, (ii) excellent batch-to-batch consistency and shelf life, (iii) prolonged anticoagulation time, (iv) prolonged blood circulation time, and (v) a kill-switch mechanism to control anticoagulation and excretion of functionally inactive metabolites. DESIGN, ASSEMBLY, AND CHARACTERIZATION OF ANTICOAGULANT FIBERS We previously reported RNA-DNA hybrids35 to conditionally activate various functions in cancer cells. Here, we explore the potential of this technology for the conditional regulation of blood coagulation. We designed anticoagulant fibers (Figure 1a,b and Figure S1) to carry multiple copies of antithrombin aptamers. Because of their small size, free aptamers are prone to fast renal clearance, while their assembly into larger structures is expected to enhance the circulation time in vivo (Figure 1c). To exert greater control over thrombin activity, we introduce kill-switches, which are fibers fully complementary to their anticoagulant counterparts (Figure 1d), to undergo isothermal reassociation and release inactive short duplexes for accelerated clearance. Three representative anticoagulant fibers, decorated with NU172 (NU fibers), RA-36 (RA fibers), or both aptamers (NU/RA fibers), were chosen for extensive characterization in silico, in vitro, and in vivo. The anticoagulation activity was assessed for all 12 different fibers and compared to nonfunctionalized analogues and free aptamers (Table S1). Anticoagulant fibers and kill-switches were analyzed by AFM and native-PAGE (Figure 2 and Figure S2) to confirm their efficient transformation into short duplexes upon reassociation. Kinetic studies show that RA and NU/RA fibers need less than 30 min of incubation with kill-switches to complete the reassociation, whereas NU fibers require a longer incubation time. The differences in the fibers’ morphology and kinetics stem from the three-dimensional structures of aptamers, which show the presence of two symmetric G-quadruplexes in RA-36 and one G-quadruplex in NU172 (Figure 2c). To gain further details, we employed molecular dynamic simulations using DMD (discrete molecular dynamics) and calculated the RMSF (root-mean-square fluctuation). NU/RA fibers have a higher RMSF value (3.65 Å) in comparison RA (3.37 Å) and NU (3.21 Å) fibers. High RMSF regions mainly consist of 5′- and 3′-ends of DNA strands, G-quadruplexes, and single-stranded regions (Figure 2b). The number of G-quadruplexes correlates with fiber flexibility and length. NU/RA fibers contain six G-quadruplex regions, RA fibers contain four G-quadruplex regions, and NU fibers contain two G-quadruplex regions. Thus, in agreement with AFM imaging, NU/RA fibers have greater flexibility and shorter length in comparison to RA fibers, while RA fibers are more flexible and shorter than NU fibers. Due to their high flexibility, NU/RA fibers are also most prone to disassociation,consistently with native-PAGE analysis. The durability of anticoagulant fibers and kill-switches against nucleases was assessed in blood stability assays (Figure S3) that show the detectable presence of all constructs after 5 h of incubation, with all structures being mostly digested (∼90%) after 24 h. Also, fibers maintain their integrity in different replica serum buffers (Figure S4). LOCKING AND UNLOCKING OF THROMBIN ACTIVITY BY ANTICOAGULANT FIBERS AND KILL-SWITCHES Three coagulation pathways (intrinsic, extrinsic, and common) have been described.36 In vitro/ex vivo coagulation assays assess blood clot formation via the activated partial thromboplastin time (APTT) that measures the functionality of the intrinsic pathway, prothrombin time (PT) that assesses the extrinsic pathway, and thrombin time (TT) that evaluates the common pathway (Table 1 and Table S2). Prolongation of the blood coagulation in all assays is commonly considered to assess thrombin function due to its key role in three pathways; for this reason APTT, PT, and TT are used in clinics to diagnose blood coagulation deficiencies and monitor the efficacy of antithrombotic therapies. To evaluate the effectiveness of anticoagulant fibers, we performed APTT, PT, and TT tests using blood of healthy donors from both the US and Brazil. Coagulation assays were conducted according to current clinical standards utilizing World Health Organization (WHO) certified human plasma as controls and WHO-qualified plasma coagulation reagents with known time limits for normal plasma coagulation. In the US, these standards qualified any time measurements below 13.4, 37, and 21 s as normal times for PT, APTT, and TT assays, respectively. Coagulation times above these limits were considered as prolongation. In Brazil, 12.1, 36.5, and 16.6 s were normal times for PT, APTT, and TT assays, respectively. In both studies, aptamers alone resulted in a slight prolongation of the plasma coagulation time in the APTT assay, but not in PT or TT assays (Table 1a, plasma versus NU172 and RA-36 aptamers). The nonfunctional fibers and free aptamers at the tested concentrations did not affect the plasma coagulation time (Table 1). However, the anticoagulant potency of the same aptamers was significantly improved by their codelivery on fiber backbones. To verify that the action of anticoagulant fibers can be controlled, the plasma coagulation was assessed after kill-switches were added to the same set of specimens; the coagulation times in all three assays returned to normal, consistent with the expected mechanism of action (Table 1b). IMMUNORECOGNITION OF ANTICOAGULANT FIBERS Since anticoagulant fibers are intended for intravenous administration, it is crucial to determine whether they can induce cytokine responses or complement activation (Figure 3), both of which represent immunostimulatory reactions commonly reported as dose-limiting toxicity of therapeutic oligonucleotides.37,38 To assess the magnitude of associated pro-inflammatory cytokine production and complement activation, human PBMCs were treated with the fibers (all tested prior for endotoxin contamination). PBMCs were chosen as a model system that provides more accurate predictions of cytokine storm toxicity in humans among all preclinical models. The complement system plays an essential role in innate immunity.39 This system is composed of over 30 plasma proteins that trigger a proteolytic cascade upon activation, resulting in the production of opsonins, anaphylatoxins, and the terminal membrane-attack complex whose coordinated function leads to immune cell activation and the destruction of invading pathogens.39 There are three pathways of complement activation: lectin, classical, and alternative. Activation of any of these pathways involves a series of cleavage reactions culminating in the formation of C3 convertase.40 C3 convertase cleaves the C3 complement component into several split products, some of which, e.g., C3a, act as anaphylatoxins, whereas others, e.g., C3b, act as opsonins.40 The C3b fragment is unstable and quickly degrades to iC3b, which can be quantified and therefore serves as a biomarker of complement activation. Here, we used an immunoassay to detect the presence of iC3b (Figure 3b). Cobra venom factor (CVF), a known complement-activating protein, and clinically used PEGylated liposomal doxorubicin formulation (Doxil), known to cause complement activation-related pseudoallergy in sensitive patients, were used as positive controls.41 Our data show that activation by any of the fibers is less than that seen with Doxil. Although these results do not completely rule out potential complement activation at higher concentrations or in particularly sensitive individuals, the outcomes suggest negligible stimulation of the complement system by anticoagulant fibers in vitro for comparable safety with regard to the complement activation in vivo. The immunostimulatory response in PBMCs was assessed with a multiplex panel including 15 cytokines of different families produced by various cells composing PBMCs and activated by various inflammatory pathways (Figure 3c). A combination of lipopolysaccharide (LPS), ODN2216, and phytohemagglutinin (PHA-M) was used as a positive control due to their abilities to activate inflammatory cytokines and type I and II IFNs in PBMCs.42–44 Relative to the positive controls, the production of cytokines by cells exposed to anticoagulant fibers was negligible. These data are in agreement with earlier studies reporting that carrier-free nucleic acid constructs are immunoquiescent due to their inability to enter the endosomal compartment of cells35,45 where many nucleic acid-specific receptors are localized. These data suggested that the possibility of endothelial activation and thrombogenicity stemming from a material-activated immune response is negligible46 and warranted further in vivo studies. BIODISTRIBUTION AND CONDITIONAL RETENTION OF ANTICOAGULANT FIBERS Endotoxin-free fluorescently labeled anticoagulant fibers and kill-switches were administered to BALB/c mice via retro-orbital injection (Figure 4a). Biodistributions were evaluated using an in vivo optical imaging system (IVIS) at various time points postinjection (Figure 4b). A whole-body fluorescence analysis revealed a strong signal in the bladder region (Figure 4b). In comparison to free aptamers, anticoagulant fibers exhibited prolonged accumulation in the bladder, indicating delayed renal excretion. Following the administration of the corresponding kill-switches, we observed rapid excretion due to reassociation and the concomitant generation of short DNA and RNA duplexes. Since the whole-body imaging may not be accurate in providing a sufficient level of detail about the biodistribution, additional ex vivo imaging was carried out. The liver, kidneys, heart, and lungs were harvested and imaged 2 h postinjection (Figure S5). Liver tissues displayed the highest level of fluorescence in most animals. To estimate the concentrations of the constructs, liver and kidney lysates were processed and their fluorescent signal intensities were measured, which in turn reflected the concentrations of the constructs (Figure 4c). On the basis of this analysis, the concentrations of anticoagulation fibers were found to be much higher than those of free aptamers. However, the animals that were also injected with kill-switches displayed no obvious difference due to the excretion of constructs via urine throughout the duration of the experiments. Next, the fluorescence intensity of the bladder from the in vivo images was recorded and converted to concentration (Figure 4d) using a calibration curve (y = 0.00000005485x + 3.01). The three pairs of anticoagulant fibers and corresponding kill-switches display similar trends: ∼30 min postinjection, anticoagulant fibers show peak accumulation in the bladder and then the concentration decreased. NU/RA fibers have a smoother reduction in comparison NU and RA fibers. NU/RA fibers also show the lowest concentration at the end point of this study (2 h postinjection), which indicates that they are excreted from the system the fastest. With the addition of kill-switches, NU/RA fibers also show a delayed accumulation in the bladder at ∼60 min postinjection in comparison to a slight increase seen for the other two constructs. In agreement with in vitro tests, the fibers with different aptamer combinations displayed different kinetics of reassociation. Finally, we quantified the concentration of fibers with and without kill-switches in mouse urine samples (Figure 4e). Urine was collected immediately after each mouse urinated, and the time duration from injection to urination was recorded. According to our results, kill-switches significantly increase the excretion rate of fiber components in a way that is different from that in earlier studies with other types of nanomaterial-formulated aptamers.47 Our data are consistent with earlier studies reporting a rapid clearance of free aptamers from the bloodstream and accumulation in the liver and kidneys.48 Prolonged circulation of PEGylated aptamers and aptamers delivered by nanoparticles in comparison to the free aptamers have also been reported.49 IN VIVO FUNCTION OF ANTICOAGULANT FIBERS AND KILL-SWITCHES To determine whether anticoagulant fibers interfere in normal hemostasis in vivo, we performed a murine tail-bleeding experiment50 (Figure 5). To comply with internationally recognized 3R principles of research animal care and use,51 we narrowed down our study to NU fibers and free aptamers as representative materials. The duration of bleeding was similar among the groups, and the majority of animals bled during the 30 min of the assay. The overall prolonged bleeding time in vivo for all samples could be explained by the effect of anesthetics on the coagulation system. Xylazine and isoflurane can cause vasodilation and increase basal blood flow velocity, inhibiting or reducing the effect of anticoagulants and promoting variable results due to unstable thrombi events.52 Additionally, we observed episodes of intermittent bleeding and variation in bleeding flow. Therefore, the bleeding time alone was insufficient to determine the hemostatic effect of the anticoagulant fibers, and so we estimated the total volume of bleeding by assessing hemoglobin concentration (Figure S6). The treatment with control enoxaparin resulted in an increased bleeding volume in comparison with animals treated with a negative control (p < 0.05). On comparing NU fibers with nonfunctional fibers and NU aptamers, we observed an increase in the bleeding volume, which was partially reversed by the kill-switch fibers. Importantly, bleeding samples from animals treated with NU fibers exhibited a decreased formation of blood clots (33.3%) in comparison to controls (83.3%) and was completely recovered after the injection of kill-switches (100%) (Figure 5a,b). As would be expected, animals treated with NU fibers, which showed less clot formation, lost a higher blood volume over the course of the experiment, while animals treated with free aptamers, which exhibited a higher degree of clotting, lost lower volumes of blood (Figure 5a). This relationship between clotting and blood volume lost was evident in the comparison of clots from blood samples of control animals with those of animals treated with enoxaparin (Figure S6b). To quantify the bleeding flow variation over time, we annotated all bleeding flow episodes for each animal along the experiment and categorized bleeding flow intensity as low, medium, or high. Then we determined a bleeding score by assigning factors 1–3 to the bleeding times corresponding to each flow intensity (Figure S6c). Events of progressive reduction of bleeding until cessation, followed by an abrupt restart of bleeding, termed rebleeding episodes, were found decreased in mice treated with NU fibers (11%), while controls treated with NU aptamers and fibers exhibited 50% and 33%, respectively (Figure S6d). Notably, this phenomenon was reversed by the kill-switch treatment of NU fibers (42.8%). The rebleeding occurrence suggests a partial inhibition of the platelets’ hemostatic action, which generates unstable thrombin that can be detached under blood flow. These results point toward an increased antithrombotic activity of anticoagulant fibers, despite the possible interference of the anesthetic drugs used in the experiments. These results were confirmed in Yorkshire swine models demonstrating prolongation of blood coagulation, measured by a TT assessment, after the injection of anticoagulant fibers and restoration of the normal blood coagulation after the injection of kill-switches (Figure 5c). During these experiments, we also assessed arterial pressure, heart rate, and ETCO2, which did not change significantly (Table S3). Blood cell counts and blood chemistry analysis results were normal at all time points. Overall, using in vitro assays and two animal models, we demonstrated the successful inhibition of blood coagulation using anticoagulant fibers and the restoration of normal blood coagulation time by their kill-switches. The use of low-cost materials makes this strategy attractive for clinical use, especially in comparison to currently available antidotes for direct thrombin inhibitors (e.g., idarucizumab, a monoclonal antibody53). Moreover, we demonstrated that the new anticoagulant fibers and kill-switches are biocompatible and do not induce overt complement and cytokine activation. In vivo biodistribution studies revealed that anticoagulant fibers mostly accumulate in the liver and kidneys 2 h postinjection. Short duplexes resulting from the isothermal reassociation process between fibers and kill-switches increased the rate of excretion in urine. This work introduces a new concept that allows for the construction of biocompatible reconfigurable nucleic acid nanoassemblies to enable controlled blood coagulation both in vitro and in vivo. We envision that the design principles of regulated anticoagulation described in this work would help to address current global public health challenges related to cardiovascular diseases and thrombosis-driven complications from infectious diseases such as COVID-19. The regulated anticoagulation may also help overcome the issue of anticoagulant drug overdose and improve the overall safety of anticoagulants. Supplementary Material SI 2 ACKNOWLEDGMENTS Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award Numbers R01GM120487 and R35GM139587 (to K.A.A.). The study was also funded in part by federal funds from the National Cancer Institute, National Institutes of Health, under contract 75N91019D00024 (M.A.D. and E.C.). The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does the mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. We also acknowledge support from the National Institutes of Health (1R35 GM134864) and the Passan Foundation (to N.V.D.). The authors thank Dr. Chandra Williams and the Vivarium staff at UNC Charlotte for their assistance with animal studies and Dr. Erdem Tabdanov for assistance with data representation. Figure 1. Mechanism of action for anticoagulant fibers and kill-switches. (a, b) Design of anticoagulant fibers carrying NU172 and RA-36 aptamers with three possible aptamer locations within the fibers being indicated. (c) Binding of anticoagulant fibers to thrombin, preventing the blood clotting cascade. (d) Binding of kill-switches to anticoagulant fibers, causing reinstatement of thrombin function and producing smaller assemblies for accelerated renal excretion. Figure 2. Characterization of anticoagulant fibers. (a) Predicted 3D structures and AFM images of fibers, kill-switches, and their reassociation products. On the basis of the models, the distances between the aptamers in each structure were estimated (Table S1). (b) Root-mean-square fluctuation (RMSF) of NU, RA, and NU/RA fibers and (c) modeled interactions of NU fiber and thrombin. The numbered residues indicate where the interactions occur. Figure 3. Immunostimulation by anticoagulant fibers. (a) Schematic of the experimental flow. (b) Complement activation and (c) cytokines produced in response to anticoagulant fibers and aptamers assessed in human PBMCs freshly isolated from the blood of healthy donors. Data are shown as mean ± SD, N = 2 repeats for N = 3 donors. The statistical significance of NU fibers in comparison to untreated cells (NC) is denoted by an asterisk (p < 0.05). Figure 4. In vivo and ex vivo analysis of anticoagulant fibers’ biodistribution and effect of kill-switches. (a) Schematic presentation of the experimental flow and motivation. (b) Animal (N = 3/group) IVIS images at different time points post retro-orbital (RO) injections with fluorescently labeled aptamers, anticoagulant fibers, and kill-switches. (c) Further imaging of liver and kidney lysates to assess the estimated concentration of fibers. (d) Recording and assessment of the overtime fluorescence of the bladder of the treated mice as estimated concentrations. (e) Increase in the antithrombin fiber excretion rate by the presence of kill-switch fibers. Figure 5. In vivo function of anticoagulant fibers and effect of kill-switches. (a) Blood clot formation in saline from freshly bled animals treated with various constructs. (b) Hemostasis analysis of anticoagulant fibers and kill-switch activity. Bleeding tails from animals treated with NU fiber or NU fiber and kill-switch and the total blood volume collected in saline after 30 min of bleeding showed respectively a decreased blood flow and volume collected from the animals treated both with NU fibers and kill-switch. (c) Thrombin coagulation time measured in blood samples collected from Yorkshire swine treated with NU fibers and kill-switch. Table 1. Plasma Coagulation Assessment: (Top) Results of Prothrombin Time (PT), Activated Partial Thromboplastin Time (APTT), and Thrombin Time (TT) of Anti-Thrombin Fibers for Their Abilities in Delaying the Coagulation in Donors from the United States and Brazil, Displaying Some Minor Regional Variations; (Bottom) Restoration of Normal Coagulation Time by Addition of Kill-Switch Fibersa PT APTT TT sample U.S. data (13.4 sb) Brazil data (12.1 sb) U.S. data (37.0 sb) Brazil data (36.5 sb) U.S. data (21.0 sb) Brazil data (16.6 sb) NU fiber 17.3 ± 0.2 16.6 ± 0.3 >120.00 89.4 ± 1.4 >60.00 26.8 ± 0.2 RA fiber 16.6 ± 0.4 19.7 ± 0.2 70.9 ± 5.0 104.7 ± 1.9 >60.00 31.6 ± 0.6 NU/RA fiber 16.8 ± 0.4 17.7 ± 0.0 81.3 ± 4.3 100.4 ± 1.5 >60.00 28.6 ± 0.0 DNA-RNA fiber 11.8 ± 0.3 13.5 ± 0.2 36.7 ± 0.8 40.2 ± 0.1 15.8 ± 0.7 19.3 ± 0.1 plasma 10.8 ± 0.2 13.6 ± 0.2 33.4 ± 1.2 28.6 ± 0.3 16.6 ± 0.3 19.4 ± 0.1 NU172 aptamer 10.9 ± 0.2 14.1 ± 0.2 38.0 ± 1.5 39.5 ± 0.1 15.3 ± 0.8 19.0 ± 0.2 RA-36 aptamer 11.3 ± 0.4 15.3 ± 0.5 41.5 ± 3.2 45.5 ± 0.8 16.6 ± 0.8 26.5 ± 0.5 sample PT (13.4 sb) APTT (37.0 sb) TT (21.0 sb) control (normal) 12.7 ± 0.2 36.8 ± 1.5 21.4 ± 0.6 control (abnormal) 20.7 ± 0.4 71.8 ± 1.4 41.3 ± 1.4 plasma 9.7 ± 0.0 29.4 ± 0.2 16.1 ± 0.3 NU172 aptamer 16.1 ± 0.5 37.9 ± 0.7 25.3 ± 1.8 RA-36 aptamer 17.3 ± 0.3 44.2 ± 0.6 45.0 ± 1.4 NU fiber 18.6 ± 1.7 101.7 ± 2.0 59.5 ± 0.5 NU fiber + kill-switch 10.9 ± 0.1 36.7 ± 0.2 17.5 ± 0.4 RA fiber 20.8 ± 0.1 89.9 ± 1.1 >60.00 RA fiber + kill-switch 13.7 ± 0.1 42.9 ± 0.3 44.7 ± 0.1 NU/RA Fiber 16.8 ± 0.9 95.3 ± 3.0 >60.00 NU/RA Fiber + Kill-Switch 10.6 ± 0.3 38.5 ± 0.4 18.4 ± 0.6 a In this study, whole blood was collected only from donors from the United States. Data are shown as mean ± SD, N = 3. b Normal coagulation times of PT, APTT, and TT are adapted from recomendations for these methods in the US and Brazil, in compliance with the manufacturers' specifications. Control (normal) and control (abnormal) are WHO-certified plasma samples used to qualify instrument and reagent performance. The authors declare no competing financial interest. REFERENCES (1) Esmon CT ; Xu J ; Lupu F Innate immunity and coagulation. J. Thromb Haemost 2011, 9 , 182–188.21781254 (2) Prandoni P ; Falanga A ; Piccioli A Cancer, thrombosis and heparin-induced thrombocytopenia. Thromb Res 2007, 120 , S137–S140.18023709 (3) ten Cate H ; Falanga A Overview of the postulated mechanisms linking cancer and thrombosis. Pathophysiol Haemost Thromb 2009, 36 (3–4 ), 122–30. (4) Mackman N ; Bergmeier W ; Stouffer GA ; Weitz JI Therapeutic strategies for thrombosis: new targets and approaches. Nat. Rev. Drug Discovery 2020, 19 (5 ), 333–352.32132678 (5) Kushner A ; West WP ; Pillarisetty LS Virchow Triad In StatPearls, 2022. (6) Araya S ; Mamo MA ; Tsegay YG ; Atlaw A ; Aytenew A ; Hordofa A ; Negeso AE ; Wordofa M ; Niguse T ; Cheru M ; Tamir Z Blood coagulation parameter abnormalities in hospitalized patients with confirmed COVID-19 in Ethiopia. PLoS One 2021, 16 (6 ), e0252939.34153056 (7) Rajabto W ; Priantono D ; Mulyadi R Pulmonary Embolism in Hospitalized Patient with Coronavirus Disease 2019 (COVID-19). Acta Med. Indones 2021, 53 (4 ), 493–496.35027499 (8) Cook BW Anticoagulation management. Semin Intervent Radiol 2010, 27 (4 ), 360–7.22550377 (9) Nutescu EA ; Burnett A ; Fanikos J ; Spinler S ; Wittkowsky A Pharmacology of anticoagulants used in the treatment of venous thromboembolism. J. Thromb. Thrombolysis 2016, 41 (1 ), 15–31.26780737 (10) Hong E ; Halman JR ; Shah A ; Cedrone E ; Truong N ; Afonin KA ; Dobrovolskaia MA Toll-like receptor-mediated recognition of nucleic acid nanoparticles (NANPs) in human primary blood cells. Molecules 2019, 24 (6 ), 1094. (11) Panigaj M ; Johnson MB ; Ke W ; McMillan J ; Goncharova EA ; Chandler M ; Afonin KA Aptamers as modular components of therapeutic nucleic acid nanotechnology. ACS Nano 2019, 13 (11 ), 12301–12321.31664817 (12) Johnson MB ; Chandler M ; Afonin KA Nucleic acid nanoparticles (NANPs) as molecular tools to direct desirable and avoid undesirable immunological effects. Adv. Drug Deliv Rev 2021, 173 , 427–438.33857556 (13) Bouchard PR ; Hutabarat RM ; Thompson KM Discovery and development of therapeutic aptamers. Annu. Rev. Pharmacol Toxicol 2010, 50 , 237–57.20055704 (14) Woodruff RS ; Sullenger BA Modulation of the coagulation cascade using aptamers. Arterioscler Thromb Vasc Biol 2015, 35 (10 ), 2083–91.26315404 (15) Chan MY ; Cohen MG ; Dyke CK ; Myles SK ; Aberle LG ; Lin M ; Walder J ; Steinhubl SR ; Gilchrist IC ; Kleiman NS ; Vorchheimer DA ; Chronos N ; Melloni C ; Alexander JH ; Harrington RA ; Tonkens RM ; Becker RC ; Rusconi CP Phase 1b randomized study of antidote-controlled modulation of factor IXa activity in patients with stable coronary artery disease. Circulation 2008, 117 (22 ), 2865–74.18506005 (16) Chan MY ; Rusconi CP ; Alexander JH ; Tonkens RM ; Harrington RA ; Becker RC A randomized, repeat-dose, pharmacodynamic and safety study of an antidote-controlled factor IXa inhibitor. J. Thromb Haemost 2008, 6 (5 ), 789–96.18284597 (17) Dyke CK ; Steinhubl SR ; Kleiman NS ; Cannon RO ; Aberle LG ; Lin M ; Myles SK ; Melloni C ; Harrington RA ; Alexander JH ; Becker RC ; Rusconi CP First-in-human experience of an antidote-controlled anticoagulant using RNA aptamer technology: a phase 1a pharmacodynamic evaluation of a drug-antidote pair for the controlled regulation of factor IXa activity. Circulation 2006, 114 (23 ), 2490–7.17101847 (18) Krissanaprasit A ; Key CM ; Froehlich K ; Pontula S ; Mihalko E ; Dupont DM ; Andersen ES ; Kjems J ; Brown AC ; LaBean TH Multivalent aptamer-functionalized single-strand RNA origami as effective, target-specific anticoagulants with corresponding reversal agents. Adv. Healthc Mater 2021, 10 (11 ), 2001826. (19) Rusconi CP ; Scardino E ; Layzer J ; Pitoc GA ; Ortel TL ; Monroe D ; Sullenger BA RNA aptamers as reversible antagonists of coagulation factor IXa. Nature 2002, 419 (6902 ), 90–4.12214238 (20) Zhao S ; Tian R ; Wu J ; Liu S ; Wang Y ; Wen M ; Shang Y ; Liu Q ; Li Y ; Guo Y ; Wang Z ; Wang T ; Zhao Y ; Zhao H ; Cao H ; Su Y ; Sun J ; Jiang Q ; Ding B A DNA origami-based aptamer nanoarray for potent and reversible anticoagulation in hemodialysis. Nat. Commun 2021, 12 (1 ), 358.33441565 (21) Rusconi CP ; Roberts JD ; Pitoc GA ; Nimjee SM ; White RR ; Quick G ; Scardino E ; Fay WP ; Sullenger BA Antidote-mediated control of an anticoagulant aptamer in vivo. Nat. Biotechnol 2004, 22 (11 ), 1423–1428.15502817 (22) Rangnekar A ; Zhang AM ; Li SS ; Bompiani KM ; Hansen MN ; Gothelf KV ; Sullenger BA ; LaBean TH Increased anticoagulant activity of thrombin-binding DNA aptamers by nanoscale organization on DNA nanostructures. Nanomedicine-Nanotechnology Biology and Medicine 2012, 8 (5 ), 673–681.21889476 (23) Raber MN ; Walker HK ; Hall WD ; Hurst JW , Coagulation Tests. Clinical Methods: The History, Physical, and Laboratory Examinations, 3rd ed.; Butterworths: 1990; Chapter 157. (24) De Cristofaro R ; De Candia E Thrombin domains: Structure, function and interaction with platelet receptors. J. Thromb. Thrombolysis 2003, 15 (3 ), 151–163.14739624 (25) Di Cera E Thrombin. Mol. Aspects Med 2008, 29 (4 ), 203–54.18329094 (26) Bock LC ; Griffin LC ; Latham JA ; Vermaas EH ; Toole JJ Selection of single-stranded DNA molecules that bind and inhibit human thrombin. Nature 1992, 355 (6360 ), 564–6.1741036 (27) Kretz CA ; Stafford AR ; Fredenburgh JC ; Weitz JI HD1, a thrombin-directed aptamer, binds exosite 1 on prothrombin with high affinity and inhibits its activation by prothrombinase. J. Biol. Chem 2006, 281 (49 ), 37477–85.17046833 (28) Mayer G ; Rohrbach F ; Potzsch B ; Muller J Aptamer-based modulation of blood coagulation. Hamostaseologie 2011, 31 (4 ), 258–63.22065102 (29) Zavyalova E ; Golovin A ; Reshetnikov R ; Mudrik N ; Panteleyev D ; Pavlova G ; Kopylov A Novel modular DNA aptamer for human thrombin with high anticoagulant activity. Curr. Med. Chem 2011, 18 (22 ), 3343–50.21728967 (30) Zavyalova E ; Samoylenkova N ; Revishchin A ; Turashev A ; Gordeychuk I ; Golovin A ; Kopylov A ; Pavlova G The Evaluation of Pharmacodynamics and Pharmacokinetics of Anti-thrombin DNA Aptamer RA-36. Frontiers in Pharmacology 2017, 8 , 922.29311929 (31) Roxo C ; Kotkowiak W ; Pasternak A G-Quadruplex-Forming Aptamers-Characteristics, Applications, and Perspectives. Molecules 2019, 24 (20 ), 3781. (32) Zavyalova E ; Golovin A ; Pavlova G ; Kopylov A Module-Activity Relationship of G-quadruplex Based DNA Aptamers for Human Thrombin. Curr. Med. Chem 2013, 20 (38 ), 4836–4843.24083606 (33) Zavyalova E ; Samoylenkova N ; Revishchin A ; Golovin A ; Pavlova G ; Kopylov A Evaluation of Antithrombotic Activity of Thrombin DNA Aptamers by a Murine Thrombosis Model. PLoS One 2014, 9 (9 ), e107113.25192011 (34) Becker RC ; Povsic T ; Cohen MG ; Rusconi CP ; Sullenger B Nucleic acid aptamers as antithrombotic agents: Opportunities in extracellular therapeutics. Thromb Haemost 2010, 103 (3 ), 586–95.20135076 (35) Ke W ; Hong E ; Saito RF ; Rangel MC ; Wang J ; Viard M ; Richardson M ; Khisamutdinov EF ; Panigaj M ; Dokholyan NV ; Chammas R ; Dobrovolskaia MA ; Afonin KA RNA-DNA fibers and polygons with controlled immunorecognition activate RNAi, FRET and transcriptional regulation of NF-kappaB in human cells. Nucleic Acids Res 2019, 47 (3 ), 1350–1361.30517685 (36) Palta S ; Saroa R ; Palta A Overview of the coagulation system. Indian J. Anaesth 2014, 58 (5 ), 515–523.25535411 (37) Henry SP ; Jagels MA ; Hugli TE ; Manalili S ; Geary RS ; Giclas PC ; Levin AA Mechanism of alternative complement pathway dysregulation by a phosphorothioate oligonucleotide in monkey and human serum. Nucleic Acid Ther 2014, 24 (5 ), 326–35.25093529 (38) Paz S ; Hsiao J ; Cauntay P ; Soriano A ; Bai L ; Machemer T ; Xiao X ; Guo S ; Hung G ; Younis H ; Bennett CF ; Henry S ; Yun TJ ; Burel S The Distinct and Cooperative Roles of Toll-Like Receptor 9 and Receptor for Advanced Glycation End Products in Modulating In Vivo Inflammatory Responses to Select CpG and Non-CpG Oligonucleotides. Nucleic Acid Ther 2017, 27 (5 ), 272–284.28605247 (39) Ilinskaya AN ; Shah A ; Enciso AE ; Chan KC ; Kaczmarczyk JA ; Blonder J ; Simanek EE ; Dobrovolskaia MA Nanoparticle physicochemical properties determine the activation of intracellular complement. Nanomedicine: Nanotechnology, Biology and Medicine 2019, 17 , 266–275.30794962 (40) Janeway CA Jr ; Travers P ; Walport M ; Shlomchik MJ , The complement system and innate immunity. In Immunobiology: The Immune System in Health and Disease 5th ed.; Garland Science: 2001. (41) Vogel CW ; Fritzinger DC ; Hew BE ; Thorne M ; Bammert H Recombinant cobra venom factor. Molecular Immunology 2004, 41 (2–3 ), 191–199.15159065 (42) Janovec V ; Hodek J ; Clarova K ; Hofman T ; Dostalik P ; Fronek J ; Chlupac J ; Chaperot L ; Durand S ; Baumert TF ; Pichova I ; Lubyova B ; Hirsch I ; Weber J Toll-like receptor dual-acting agonists are potent inducers of PBMC-produced cytokines that inhibit hepatitis B virus production in primary human hepatocytes. Sci. Rep 2020, 10 (1 ), 12767.32728070 (43) Tahtinen S ; Tong AJ ; Himmels P ; Oh J ; Paler-Martinez A ; Kim L ; Wichner S ; Oei Y ; McCarron MJ ; Freund EC ; Amir ZA ; de la Cruz CC ; Haley B ; Blanchette C ; Schartner JM ; Ye W ; Yadav M ; Sahin U ; Delamarre L ; Mellman I IL-1 and IL-1ra are key regulators of the inflammatory response to RNA vaccines. Nat. Immunol 2022, 23 (4 ), 532–542.35332327 (44) Franzese O ; Palermo B ; Di Donna C ; Sperduti I ; Ferraresi V ; Stabile H ; Gismondi A ; Santoni A ; Nisticò P Polyfunctional Melan-A-specific tumor-reactive CD8(+) T cells elicited by dacarbazine treatment before peptide-vaccination depends on AKT activation sustained by ICOS. Oncoimmunology 2016, 5 (5 ), e1114203.27467927 (45) Hong E ; Halman JR ; Shah AB ; Khisamutdinov EF ; Dobrovolskaia MA ; Afonin KA Structure and Composition Define Immunorecognition of Nucleic Acid Nanoparticles. Nano Lett 2018, 18 (7 ), 4309–4321.29894623 (46) Markiewski MM ; Lambris JD The role of complement in inflammatory diseases from behind the scenes into the spotlight. Am. J. Pathol 2007, 171 (3 ), 715–727.17640961 (47) Guo JW ; Gao XL ; Su LN ; Xia HM ; Gu GZ ; Pang ZQ ; Jiang XG ; Yao L ; Chen J ; Chen HZ Aptamer-functionalized PEG-PLGA nanoparticles for enhanced anti-glioma drug delivery. Biomaterials 2011, 32 (31 ), 8010–8020.21788069 (48) Kryza D ; Debordeaux F ; Azema L ; Hassan A ; Paurelle O ; Schulz J ; Savona-Baron C ; Charignon E ; Bonazza P ; Taleb J ; Fernandez P ; Janier M ; Toulme JJ Ex Vivo and In Vivo Imaging and Biodistribution of Aptamers Targeting the Human Matrix MetalloProtease-9 in Melanomas. PLoS One 2016, 11 (2 ), e0149387.26901393 (49) Healy JM ; Lewis SD ; Kurz M ; Boomer RM ; Thompson KM ; Wilson C ; McCauley TG Pharmacokinetics and biodistribution of novel aptamer compositions. Pharm. Res-Dordr 2004, 21 (12 ), 2234–2246. (50) Liu Y ; Jennings NL ; Dart AM ; Du XJ Standardizing a simpler, more sensitive and accurate tail bleeding assay in mice. World J. Exp Med 2012, 2 (2 ), 30–6.24520531 (51) Russell WMS ; Burch RL The Principles of Humane Experimental Technique; Methuen: 1959. (52) Sashindranath M ; Sturgeon SA ; French S ; Craenmehr DDD ; Selan C ; Freddi S ; Johnson C ; Cody SH ; Nesbitt WS ; Hamilton JR ; Nandurkar HH The mode of anesthesia influences outcome in mouse models of arterial thrombosis. Res. Pract Thromb Haemost 2019, 3 (2 ), 197–206.31011704 (53) Pollack CV Jr ; Reilly PA ; Eikelboom J ; Glund S ; Verhamme P ; Bernstein RA ; Dubiel R ; Huisman MV ; Hylek EM ; Kamphuisen PW ; Idarucizumab for dabigatran reversal. New England Journal of Medicine 2015, 373 (6 ), 511–520.26095746

PMC009xxxxxx/PMC9515481.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8803460 3682 Eur Respir J Eur Respir J The European respiratory journal 0903-1936 1399-3003 36137595 9515481 10.1183/13993003.01117-2022 NIHMS1838595 Article Intrapulmonary shunt and alveolar dead space in a cohort of patients with acute COVID-19 pneumonitis and early recovery Harbut Piotr 1 Prisk G. Kim 2 Lindwall Robert 1 Hamzei Sarah 1 Palmgren Jenny 1 Farrow Catherine E. 345 Hedenstierna Goran 6† Amis Terence C. 345 Malhotra Atul 2 Wagner Peter D. 2 Kairaitis Kristina 345 1 Karolinska Institutet, Danderyd Hospital, Stockholm, Sweden. 2 Department of Medicine, University of California, San Diego, CA, USA. 3 Ludwig Engel Centre for Respiratory Research, Westmead Institute for Medical Research, Sydney, Australia. 4 Department of Respiratory and Sleep Medicine, Westmead Hospital, Sydney, Australia. 5 Sydney Medical School, Faculty of Medicine and Health, University of Sydney, Sydney, Australia. 6 Department of Medical Sciences, University of Uppsala, Uppsala, Sweden. † Deceased. Author contributions: The majority of the authors (P. Harbut, G.K. Prisk, C.E. Farrow, G. Hedenstierna, T.C. Amis, A. Malhotra, P.D. Wagner and K. Kairaitis) met on at least a fortnightly basis remotely. The original idea and study design were conceived in these meetings (P. Harbut, G.K. Prisk, C.E. Farrow, G. Hedenstierna, T.C. Amis, A. Malhotra, P.D. Wagner and K. Kairaitis). Ethics approval, subject identification, data collection, including clinical data, and data analysis occurred in Sweden (P. Harbut, R. Lindwall, S. Hamzei and J. Palmgren). Analysis of the raw data signals was primarily performed by G.K. Prisk, with discussion of signal analysis by P. Harbut, G.K. Prisk, C.E. Farrow, G. Hedenstierna, T.C. Amis, A. Malhotra, P.D. Wagner and K. Kairaitis, and further data analysis, including statistical analysis, in Sydney (K. Kairaitis, C.E. Farrow and T.C. Amis). All stages of the study were discussed at fortnightly meetings, including troubleshooting of protocols and quality control of the data. All authors contributed to the preparation of the manuscript and approved the final version. Corresponding author: Kristina Kairaitis (kristina.kairaitis@sydney.edu.au) 6 3 2023 1 2023 27 1 2023 27 7 2023 61 1 2201117This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Background Pathological evidence suggests that coronavirus disease 2019 (COVID-19) pulmonary infection involves both alveolar damage (causing shunt) and diffuse microvascular thrombus formation (causing alveolar dead space). We propose that measuring respiratory gas exchange enables detection and quantification of these abnormalities. We aimed to measure shunt and alveolar dead space in moderate COVID-19 during acute illness and recovery. Methods We studied 30 patients (22 males; mean±SD age 49.9±13.5 years) 3–15 days from symptom onset and again during recovery, 55±10 days later (n=17). Arterial blood (breathing ambient air) was collected while exhaled oxygen and carbon dioxide concentrations were measured, yielding alveolar–arterial differences for each gas (PA−aO2 and Pa−ACO2, respectively) from which shunt and alveolar dead space were computed. Results For acute COVID-19 patients, group mean (range) for PA−aO2 was 41.4 (−3.5–69.3) mmHg and for Pa−ACO2 was 6.0 (−2.3–13.4) mmHg. Both shunt (% cardiac output) at 10.4% (0–22.0%) and alveolar dead space (% tidal volume) at 14.9% (0–32.3%) were elevated (normal: <5% and <10%, respectively), but not correlated (p=0.27). At recovery, shunt was 2.4% (0–6.1%) and alveolar dead space was 8.5% (0–22.4%) (both p<0.05 versus acute). Shunt was marginally elevated for two patients; however, five patients (30%) had elevated alveolar dead space. Conclusions We speculate impaired pulmonary gas exchange in early COVID-19 pneumonitis arises from two concurrent, independent and variable processes (alveolar filling and pulmonary vascular obstruction). For most patients these resolve within weeks; however, high alveolar dead space in ~30% of recovered patients suggests persistent pulmonary vascular pathology. Shareable abstract (@ERSpublications) Hypoxaemia in COVID-19 is due to a variable combination of intrapulmonary shunt and increased dead space, likely from both airspace and vascular pathology. Increased dead space present up to 2 months later suggests persistent pulmonary vascular pathology. https://bit.ly/3ROtboc pmcIntroduction Since the beginning of 2020, over half a billion people have been diagnosed with coronavirus disease 2019 (COVID-19), a disease caused by infection with the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Many have required hospitalisation, with 10–20% of those requiring intensive care, mainly due to respiratory compromise [1]. Worldwide there have been more than 6 million deaths. Patients with early COVID-19 respiratory failure present with hypoxaemia and hyperventilation [2–6]. The hypoxaemia is likely related to ventilation/perfusion (V′A/Q′) mismatch and in particular to increased intrapulmonary shunt arising from alveolar filling with fluid or cellular debris. However, pathology reports from COVID-19 infected lungs also frequently demonstrate pulmonary vasculature involvement, including severe endothelial injury, widespread thrombosis with microangiopathy and new vessel growth [7–9]. Pulmonary vessel microembolism reduces capillary blood flow, thus generating areas of high V′A/Q′ and promoting increased alveolar dead space. We hypothesised that in early COVID-19 pneumonitis, hypoxaemia is associated with both increased intrapulmonary shunt and increased alveolar dead space. Using bedside measurement of exhaled partial pressure of oxygen (PO2) and carbon dioxide (PCO2) combined with arterial blood gas measurements, we measured alveolar–arterial partial pressure differences for both oxygen (PA−aO2) and carbon dioxide (Pa−ACO2), from which both intrapulmonary shunt and alveolar dead space values were then determined using a novel computational model [10]. Methods Subjects 30 patients admitted to Danderyd Hospital in Stockholm, Sweden who were ≥18 years old and PCR-positive for SARS-CoV-2 were recruited between November and December 2020. Patients were excluded if they were in immediate need of intubation or mechanical ventilation, had advanced pulmonary or cardiac disease, current malignancy, previous thromboembolic disease, current pregnancy, or were unable to tolerate the study protocol, which required spontaneous breathing of ambient air for several minutes. Data were collected within 24 h of presentation to hospital and again 55±10 days after discharge in early recovery (n=17). 13 patients did not return for follow-up. A subset of the acute data has been reported previously [6]. 13 healthy volunteers, negative for SARS-CoV-2 on PCR testing and with normal pulmonary function (Vyaire-Vyntus Spiro PC-spirometer with SentrySuite Software; Vyaire, Mettawa, IL, USA), were recruited from hospital staff as methodological controls. All participants gave written informed consent and the study protocol was approved by the Swedish Ethical Review Authority (diary number 2020-02966). Clinical data Anthropometric and demographic data were collected, along with body temperature for the acutely ill. Pathology test results, pharmacological interventions, respiratory support (including oxygen therapy) and duration of hospitalisation data were also collected. Study protocol Subjects were studied in a semirecumbent position, wearing a noseclip and breathing ambient air via a mouthpiece with an antivirus filter (MicroGard II; Vyaire Medical, Hoechberg, Germany; apparatus dead space 75 mL) attached to an inspired/expired gas measurement system (Oxycon Pro; Vyaire Medical). Collected data included PO2, PCO2 and respiratory gas flow at high sampling frequency (100 Hz). After a few minutes of adaptation, participants maintained steady-state breathing (end-tidal PCO2 within ±2 mmHg across several breaths) at a (metronome-facilitated) frequency and tidal volume of their choice. Data were recorded over a steady-state 3–5 min period, during which radial arterial blood was collected over several breaths and then processed immediately (ABL800 Flex Plus; Radiometer Medical, Bronshoj, Denmark) to obtain values for arterial PO2 (PaO2) and PCO2 (PaCO2). For the acute patient studies, arterial blood gas values were expressed at body temperature [11]. Recovered patients and healthy subjects were assumed to have a body temperature of 37°C. Exhaled gas analysis Three separate breaths, preceding the arterial blood sample by 35±10 s, were selected, independently analysed and resulting parameters averaged. Following alignment of gas and volume signals, PO2 and PCO2 values for each breath were plotted as a function of expired volume and a linear least-squares fit applied to the alveolar plateau (phase III; see figure 1). Mean alveolar gas values, PAO2 and PACO2 [10], at the mid-point (by volume) of the expired breath were determined from the fitted lines. Exhaled gas measurements in the acute patient studies were corrected for water vapour pressure at body temperature using Antoine’s formula [12]. Alveolar–arterial partial pressure differences for both O2 and CO2 were then calculated as PAO2−PaO2 (PA−aO2) and PaCO2−PACO2 (Pa−ACO2). Computational analysis We employed the seven-decade-old RILEY and COURNAND [13] three-compartment lung model to calculate shunt and alveolar dead space from PA−aO2 and Pa−ACO2. In this model, the lung is considered to have an intrapulmonary shunt compartment with a V′A/Q′ of zero, also encompassing regions of very low V′A/Q′, and an alveolar dead space compartment with V′A/Q′ of infinity, also encompassing regions of very high V′A/Q′. The remainder of the lung is assumed to be normal, with a V′A/Q′ ratio given by non-dead space ventilation divided by non-shunt blood flow. A critical difference exists between the Riley and Cournand model and the current approach: Riley and Cournand used the venous admixture equation for oxygen to calculate shunt and the Bohr equation for carbon dioxide to calculate dead space. However, shunt may increase arterial PCO2 and contribute to calculated alveolar dead space, while dead space (without compensatory hyperventilation) lowers arterial PO2 and contributes to calculated shunt. Our approach [10] recognises this complexity and resolves it, still within the three-compartment framework, by determining the values of shunt and dead space that, present together, predict the measured arterial and expired alveolar PO2 and PCO2 values, and hence their partial pressure differences. In addition, the Bohr equation usually includes anatomical dead space, which commonly dominates total dead space numbers. Our approach eliminates anatomical dead space by using mean alveolar partial pressures as in figure 1 and not those of mixed exhaled gas. Using the measured V′CO2 (respiratory frequency×volume CO2 per breath), PAO2, fractional alveolar oxygen concentration (FAO2), PACO2 and fractional alveolar carbon dioxide concentration (FACO2), we calculated: 1) total expired alveolar ventilation V′A=V′CO2/FACO2, 2) inspired alveolar ventilation V′I=V′A×((1−FAO2 −FACO2)/(1−FIO2)) and 3) V′O2=V′I×FIO2−V′A×FAO2, where FIO2 is the inspiratory oxygen fraction. Cardiac output (Q′T) was then estimated as Q′T=V′O2+5, with both Q′T and V′O2 expressed in L·min−1 [14, 15]. The oxygen tension at which haemoglobin is 50% saturated (HbP50) was assumed to be 26.8 mmHg. Using these values and each patient’s own data for V′O2, V′CO2, V′A, Q′T, base excess, haemoglobin concentration and body temperature, we applied the algorithm first published in 1969 by WEST [16] to estimate the values of intrapulmonary shunt and alveolar dead space that would result in the measured PaO2 and PaCO2 in each individual to within 0.2 mmHg. Intrapulmonary shunt was expressed as the percentage of pulmonary blood flow perfusing unventilated regions (shunt %), while alveolar dead space was expressed as the percentage of alveolar ventilation associated with unperfused regions (alveolar dead space %). Based on published measurements using the multiple inert gas elimination technique in normal subjects [17], the 95% upper confidence limit for physiological shunt is 5% and for alveolar dead space is 10%. Further methodological details can be found elsewhere [10]. Statistical analysis This was an observational study and a sample size was not calculated. Individual data were pooled and reported as group mean with standard deviation or interquartile range. Comparisons were made using paired t-tests. Relationships between variables were examined using Spearman’s rank correlation coefficients. p<0.05 was considered significant. Results Subject characteristics See table 1. Acute COVID-19 patient data All patients had moderate disease and were studied at hospital presentation and within 3–15 days of symptom onset. 24 were admitted (mean 4.0±2.8 days, range 1–12 days); of these, 18 required supplemental oxygen and one was subsequently admitted to the intensive care unit. There were no deaths. Pharmacological interventions 22 patients received prophylactic low-molecular-weight heparin (LMWH), 12 of these patients <24 h before the study (tinzaparin 4500 U per day, subcutaneously), while therapeutic LMWH (tinzaparin 175 U·kg−1 twice daily, daily, s.c.) was administered to one patient. Post-study, 14 patients received corticosteroids (betamethasone 6 mg orally), three received remdesivir and convalescent plasma was administered to two patients. Pathology findings C-reactive protein levels were elevated in all patients, while almost 40% had elevated levels for D-dimer (table 2). Arterial blood gas, oxyhaemoglobin saturation and exhaled alveolar gas data Arterial blood gas data reflected an acute respiratory alkalosis with hypoxaemia (table 3). PAO2 values were all >102 mmHg, PACO2 values were all <38 mmHg and arterial oxygen saturation (SaO2) values were all >89%. Alveolar–arterial differences There was wide interpatient variance in PA−aO2 and Pa−ACO2 (table 3 and figure 2a). Any negative values (expected from experimental noise) were considered to be zero when calculating shunt and dead space. Intrapulmonary shunt and alveolar dead space Values for intrapulmonary shunt and alveolar dead space varied considerably between patients (table 4 and figure 2b); however, importantly, there was no significant correlation between the shunt and dead space values (p=0.27). Significant positive correlations were detected between shunt and both D-dimer (r=0.36, p=0.03) and CRP (r=0.42, p=0.01), but not for alveolar dead space (p>0.15). Overall, 23 patients had elevated shunt (defined as >5%), while 22 patients had elevated alveolar dead space (defined as >10%). Simultaneously increased shunt and dead space were present in 19 patients, four patients had elevated shunt but normal alveolar dead space and three had elevated alveolar dead space but normal shunt (figure 2b). Healthy subject data Healthy subjects were all normoxaemic with normal acid–base status (see table 3). No healthy subject had a shunt value >5% (table 4 and figure 2b); two had slightly elevated values for dead space (table 4 and figure 2b). Overall, shunt and dead space values from the healthy subjects conformed to the previously established upper limits of normal [17]. Transition from acute illness to recovery Shunt and alveolar dead space fell significantly following recovery (figures 3 and 4). However, shunt was slightly elevated in two patients, while five patients (29.4%) had elevated dead space. There was no correlation with duration after recovery (both p>0.15). Discussion Using simultaneous measurements of arterial and exhaled oxygen and carbon dioxide tensions, interpreted using a three-compartment computational lung model, we developed a bedside methodology for quantifying if there is both intrapulmonary shunt and alveolar dead space [10], and then applied it here in the highly infectious disease setting of COVID-19 pneumonitis. Measurements in a healthy subject cohort (technical controls) conformed to historical normal limits previously established by more sophisticated techniques [17]. Pulmonary gas exchange in early COVID-19 pneumonitis The main finding from the present study is that in early acute COVID-19 pneumonitis intrapulmonary shunt is elevated for most patients; however, many also have elevated alveolar dead space. Furthermore, intrapulmonary shunt and dead space values were not correlated in moderate COVID-19 pneumonitis. Hence, dead space cannot be predicted from the magnitude of shunt and vice versa. This suggests that in early SARS-CoV-2 pulmonary infection two pathophysiologically distinct process, either separately or together, are in play: 1) patchy alveolar filling/oedema/atelectasis (pneumonia) resulting in lung regions with low or zero V′A/Q′ (i.e. shunt) and 2) patchy pulmonary vascular occlusion (emboli) resulting in high V′A/Q′ regions (i.e. alveolar dead space). However, the relative contribution of these two processes is highly variable from patient to patient. This is similar to findings in intubated patients with severe acute respiratory distress syndrome, where hypoxaemia is due to increased intrapulmonary shunt and increased dead space [18]. This is the first time that this has been demonstrated physiologically in mild/moderate COVID pneumonitis. While intrapulmonary shunt is perhaps an expected contributor to hypoxaemia in a pulmonary viral infection [18], previous authors have suggested that for COVID-19 patients shunt values can be so elevated as to be considered “excessive” for the degree of lung injury present [2]. This observation has led to the suggestion that SARS-CoV-2 may specifically impair hypoxic pulmonary vasoconstriction, thus preventing hypoxic pulmonary vasoconstriction-driven restriction of pulmonary capillary blood flow to lung regions with poor or no ventilation, increasing shunt [2, 19]. Whether intrapulmonary shunt values detected in the present study are “excessive” remains undetermined. A major feature of our results was the high dead space values occurring in 73% of patients, with 63% having both elevated intrapulmonary shunt and dead space and 10% having elevated dead space with no evidence of shunt. Increased alveolar dead space is a consequence of ventilation of underperfused alveoli, and we speculate is consistent with reported diffuse and widespread small pulmonary arterial obstruction from thrombus formation, as detected in pulmonary pathological specimens from deceased COVID-19 patients [7, 20]. This interpretation is further supported by the absence of large vessel emboli on contrast computed tomography imaging (performed for clinical purposes) in a subgroup (n=18) of the present patient cohort, although there was no correlation between dead space and D-dimer. There is extensive pathological evidence of pulmonary microvascular involvement in COVID-19 pneumonia [20]. Histopathology of patients who died from COVID-19, compared with those who died from influenza, showed microvascular changes present in COVID-19 infected lungs that were not present with influenza, including endothelial injury, alveolar capillary microthrombi and new vessel growth [7]. In a study which correlated radiological findings with histopathological inflammation in eight patients who died of COVID-19, there was evidence of vascular damage and thrombosis in regions without concurrent airspace involvement [21]. It has been proposed that the vascular injury may precede the development of frank pneumonia and alveolar filling [22]. High V′A/Q′ regions (dead space) may also be a consequence of redistribution of ventilation from unventilated alveoli (shunt) resulting in relative overventilation of normally perfused alveoli. Our analysis adjusts for any effect of intrapulmonary shunt on Pa−ACO2, so that redistribution of ventilation from obstructed ventilation zones is unlikely to contribute significantly to our measured dead space values. Pulmonary gas exchange in early recovery from COVID-19 pneumonitis In recovery, intrapulmonary shunt decreased in all patients and values were within normal expected limits for all but two patients (figures 3 and 4). Consequently, ventilation was restored to lung zones previously identified as having no or very low alveolar ventilation. Dead space also decreased in most patients to within, or close to, normal expected limits (figures 3 and 4). However, in five patients, dead space was >10%, ranging from 10.7% to 22.4%. Consequently, for most patients perfusion was restored to lung zones previously identified as having no or very low perfusion. However, for ~30% of those studied in recovery (figures 3d and 4), persistent, or even increasing, dead space values suggest persistent or evolving damage to the pulmonary vasculature. This may be a consequence of pre-morbid disease (e.g. COPD or interstitial lung disease). However, if a consequence of COVID-19 infection, this finding is of particular interest, since recently published data suggest that COVID-19 infection poses increased risk for deep vein thrombosis, pulmonary embolism and bleeding episodes, at 3, 6 and 2 months, respectively, after an acute infection [23]. Clinical implications The finding that there is likely both acute and chronic pulmonary microvascular disease in COVID-19 pneumonitis has implications for the clinical management of patients, both in the acute and recovery phases of the disease process. For example, anticoagulation with LMWH has had mixed outcomes, with benefit in moderate disease but no real improvement found in severe disease [24, 25]. Perhaps a clearer picture might emerge if outcomes were stratified by presence or absence of high V′A/Q′ regions. Targeted therapeutic approaches may then be deployed for patients with increased dead space, who may need anticoagulation, versus those with shunt but low dead space, where anticoagulants may, in theory, have risk without major benefits. Application of this methodology could be used in patients with persistent COVID-19 symptoms to better delineate pulmonary pathology. Limitations This study was performed in a small sample of patients with early and mild/moderate disease, early in the pandemic, when few therapeutic interventions were available, and was restricted to those able to tolerate breathing room air. Also, recovery data were collected at only one time-point and we do not have follow-up data on all patients. Consequently, it may not be generalisable to more severe disease. Other (pre-existing) comorbidities such as anaemia may have impacted on our findings, although there was no clinical evidence of these abnormalities. The logistical limitations imposed by an acute airborne infection meant that we were unable to use more sophisticated techniques such as multiple inert gas elimination [26] or imaging tools [27]. We estimated cardiac output and HbP50, required for the calculation of intrapulmonary shunt and dead space. Cardiac output was determined from its well-documented relationship to V′O2 [14, 15], which was measured; HbP50 was taken to be 26.8 mmHg for all participants. Sensitivity analysis for these two uncertain variables shows a minor effect over a wide range of determined intrapulmonary shunt values, with no effect on dead space [10]. Conclusions Our study shows that in early mild/moderate COVID-19 pneumonitis, there is both increased intrapulmonary shunt and increased alveolar dead space, with marked interpatient variability and lack of correlation. These findings suggest that alveolar filling, resulting in little or no ventilation to some alveoli, and pulmonary microvascular compromise, resulting in little or no blood flow to other alveoli, are both present to varying, but separate, degrees in the early phases of mild/moderate COVID-19 pneumonitis. For essentially all patients, shunt resolved in early recovery; however, for ~30% of patients, elevated dead space persisted, at least in the early recovery phase. Characterising individual patient pulmonary pathophysiological responses may help inform more personalised approaches to effective treatments in both the acute and recovery phases of infection with SARS-CoV-2. Support statement: We would like to thank Djurgården IF Hockey in Stockholm, Sweden for the loan of the equipment used in this study and Vyaire Medical Pty Ltd for data collection software support. A. Malhotra is funded by the National Institutes of Health. A. Malhotra has received grants from the National Institutes of Health, and consulting fees for medical education from Livanova, Equilium, Corvus and Jazz. P.D. Wagner has received consulting fees from SMS Biotechnology and Third Pole Inc., and has participated on the data safety monitoring board on Dr Levine’s (UT Southwestern, Dallas, TX, USA) PPG. P. Harbut received an unconditional loan of the equipment used in this study from the Djurgården Hockey Club, Stockholm. Data availability: Individual de-identified data and the protocol used in this study will be available to investigators whose proposed use of the data has been approved by an independent review committee identified for this purpose. Data will be available for meta-analysis from 9 to 36 months of publication. After 36 months the data will be available in the Karolinska Institutet (Stockholm, Sweden) data repository but without investigator support other than the deposited data. FIGURE 1 Expired gas tracings from one patient: exhaled partial pressure of oxygen (PO2; top black line, red symbols and lines) and carbon dioxide (PCO2; bottom black line, blue symbols and lines). The second dotted line indicates the beginning of alveolar emptying (phase III), sloping dashed lines indicate the linear regression fits to phase III. Alveolar partial pressures (PACO2 and PAO2) were measured from the mid-volume of the linear regression fit to phase III (~225 mL in this example). Filled symbols indicate mean alveolar values; open symbols indicate contemporaneous arterial values; arterial–alveolar differences are indicated by the solid vertical lines connecting the symbols. FIGURE 2 a) Alveolar–arterial partial pressure differences for oxygen (PA−aO2) and carbon dioxide (Pa−ACO2). b) Intrapulmonary shunt and alveolar dead space. Acute COVID-19 patients (n=30) and healthy subjects (n=13). Dotted lines in (b) show 95% upper limits for normal [17]. FIGURE 3 Individual (symbols and lines) and grouped data (box-and-whisker plots showing mean, interquartile range and range) for 17 patients during acute COVID-19 infection and again in early recovery, together with 13 healthy individuals: alveolar–arterial partial pressure differences for a) oxygen (PA−aO2) and b) carbon dioxide (Pa−ACO2), and c) shunt and d) alveolar dead space. Horizontal dotted lines in (c) and (d) show the 95% upper confidence intervals for normal. *: p<0.05; **: p<0.0001. FIGURE 4 Shunt and dead space trajectories from acute COVID-19 illness to recovery. Blue dashed lines connect paired acute and recovery data (n=17). Black dotted lines show the 95% upper confidence limits of normal. TABLE 1 Anthropometric, demographic and body temperature data for acutely ill COVID-19 patients, recovered COVID-19 patients and healthy subjects Acute (n=30) Recovered (n=17) Healthy (n=13) Age (years) 49.9±13.5 (23–78) 50.6±12.0 (51–78) 51.1±17.0 (34–68) Sex  Female 8 6 5  Male 22 11 8 Weight (kg) 87.5±17.0 (46–113) 87.5±18.2 (46–110) 80.2±26.1 (58–110) Height (cm) 177.0±11.3 (150–195) 176.8±13.2 (150–195) 177.2±19.5 (159–198) BMI (kg·m−2) 27.8±4.3 (20.2–38.5) 27.9±4.7 (20.4–38.5) 25.5±6.0 (19.6–31.5) Body temperature (°C) 38.0±1.0 (36.5–40.0) 37.0# 37.0# Data are presented as mean±SD (range) or n. #: assumed. TABLE 2 Pathology data at the time of acute COVID-19 infection Pathology Mean±SD Range Normal range Abnormal (n (%)) Haemoglobin (g·L−1) (n=29) 139.5±11.7 110–165 Female: 117–153; male: 134–170 2 (6.8) D-dimer (mg·L−1) (n=28) 0.8±0.6 0.3–3.2 Female: <0.54; male: <0.50 11 (39.2) C-reactive protein (mg·L−1) (n=30) 79.1±65.7 3–256 <3 30 (100) TABLE 3 Arterial blood gases and exhaled alveolar gas measurements Acute (n=30)# Recovered (n=17) Healthy (n=13) pH 7.48±0.04 (7.42–7.60) 7.44±0.03 (7.39–7.51) 7.43±0.02 (7.41–7.49) PaCO2 (mmHg) 34.8±4.2 (26.5–43.0) 34.4±3.7 (26.7–40.6) 37.0±2.8 (31.4–40.8) PaO2 (mmHg) 68.3±12.6 (52.3–105.8) 100.6±13.6 (84.0–131.3) 105.0±8.3 (94.5–120.0) PAO2 (mmHg) 113.3±5.7 (102.2–122.3) 113.6±6.3 (100.2–125.3) 119.5±5.1 (112.7–127.3) PACO2 (mmHg) 30.3±4.0 (23.4–37.3) 31.4±3.1 (24.9–38.7) 35.2±2.9 (29.1–39.9) PA−aO2 (mmHg) 41.4±16.3 (−3.5–69.3) 13.0±9.7 (−6.0–30.2) 6.5±6.9 (−5.8–21.4) Pa−ACO2 (mmHg) 6.0±4.2 (−2.3–13.4) 3.0±2.3 (−2.2–8.9) 1.8±1.9 (−1.4–6.0) SaO2 (%) 94.3±0.2 (89.4–98.7) 98.7±0.5 (97.7–99.5) 98.9±0.2 (98.4–99.3) Data are presented as mean±SD (range). #: all acute patient gas partial pressures are expressed at body temperature. PaCO2: arterial partial pressure of carbon dioxide; PaO2: arterial partial pressure of oxygen; PAO2: alveolar partial pressure of oxygen; PACO2: alveolar partial pressure of carbon dioxide; PA−aO2: alveolar–arterial partial pressure difference for oxygen; Pa−ACO2: alveolar–arterial partial pressure difference for carbon dioxide; SaO2: arterial oxygen saturation. TABLE 4 Shunt and alveolar dead space in acute COVID-19 patients (including the subgroup of 17 patients studied later in recovery), recovered patients and healthy subjects. Acute (n=30) Acute# (n=17) Recovered (n=17) Healthy (n=13) Shunt (%) 10.4±5.4 (0–22.0) 11.7±5.1 (4.1–22.0) 2.4±1.9** (0–6.1) 1.2±1.1 (0–4.1) Alveolar dead space (%) 14.9±9.6 (0–32.3) 14.4±9.7 (0–29.7) 8.6±5.5* (0–22.4) 4.7±4.3 (0–11.8) Data are presented as mean±SD (range). #: subgroup studied in recovery. *: p<0.05; **: p<0.0001 (compared with acute subgroup). Conflict of interest: No other author has a conflict of interest to declare. References 1 Chew MS , Blixt PJ , Ahman R , National outcomes and characteristics of patients admitted to Swedish intensive care units for COVID-19: a registry-based cohort study. Eur J Anaesthesiol 2021; 38 : 335–343.33534266 2 Gattinoni L , Coppola S , Cressoni M , COVID-19 does not lead to a “typical” acute respiratory distress syndrome. Am J Respir Crit Care Med 2020; 201 : 1299–1300.32228035 3 Ottestad W , Sovik S . COVID-19 patients with respiratory failure: what can we learn from aviation medicine? Br J Anaesth 2020; 125 : e280–e281.32362340 4 Meng H , Xiong R , He R , CT imaging and clinical course of asymptomatic cases with COVID-19 pneumonia at admission in Wuhan, China. J Infect 2020; 81 : e33–e39. 5 Gattinoni L , Chiumello D , Caironi P , COVID-19 pneumonia: different respiratory treatments for different phenotypes? Intensive Care Med 2020; 46 : 1099–1102.32291463 6 Kairaitis K , Harbut P , Hedenstierna G , Ventilation is not depressed in patients with hypoxemia and acute COVID-19 infection. Am J Respir Crit Care Med 2022; 205 : 1119–1120.35130468 7 Ackermann M , Verleden SE , Kuehnel M , Pulmonary vascular endothelialitis, thrombosis, and angiogenesis in Covid-19. N Engl J Med 2020; 383 : 120–128.32437596 8 Leisman DE , Deutschman CS , Legrand M . Facing COVID-19 in the ICU: vascular dysfunction, thrombosis, and dysregulated inflammation. Intensive Care Med 2020; 46 : 1105–1108.32347323 9 Magro C , Mulvey JJ , Berlin D , Complement associated microvascular injury and thrombosis in the pathogenesis of severe COVID-19 infection: a report of five cases. Transl Res 2020; 220 : 1–13.32299776 10 Wagner PD , Malhotra A , Prisk GK . Using pulmonary gas exchange to estimate shunt and deadspace in lung disease: theoretical approach and practical basis. J Appl Physiol 2022; 132 : 1104–1113.35323050 11 Bradley AF , Severinghaus JW , Stupfel M . Effect of temperature on PCO2 and PO2 of blood in vitro. J Appl Physiol 1956; 9 : 201–204.13376428 12 Antoine C . Tension des vapeurs: nouvelle relation entre les tension et les temperatures. Comptes Rendus 1888; 107 : 681–684, 778–780, 836–837. 13 Riley RL , Cournand A . ‘Ideal’ alveolar air and the analysis of ventilation–perfusion relationships in the lungs. J Appl Physiol 1949; 1 : 825–847.18145478 14 Astrand PO , Cuddy TE , Saltin B , Cardiac output during submaximal and maximal work. J Appl Physiol 1964; 19 : 268–274.14155294 15 Rowell LB . Human Cardiovascular Control. New York, Oxford University Press, 1993. 16 West JB . Ventilation–perfusion inequality and overall gas exchange in computer models of the lung. Respir Physiol 1969; 7 : 88–110.5809098 17 Wagner PD , Hedenstierna G , Bylin G . Ventilation–perfusion inequality in chronic asthma. Am Rev Respir Dis 1987; 136 : 605–612.3631733 18 Dantzker DR , Brook CJ , Dehart P , Ventilation–perfusion distributions in the adult respiratory distress syndrome. Am Rev Respir Dis 1979; 120 : 1039–1052.389116 19 Herrmann J , Mori V , Bates JHT , Modeling lung perfusion abnormalities to explain early COVID-19 hypoxemia. Nat Commun 2020; 11 : 4883.32985528 20 Tian S , Hu W , Niu L , Pulmonary pathology of early-phase 2019 novel coronavirus (COVID-19) pneumonia in two patients with lung cancer. J Thorac Oncol 2020; 15 : 700–704. 21 Kianzad A , Meijboom LJ , Nossent EJ , COVID-19: histopathological correlates of imaging patterns on chest computed tomography. Respirology 2021; 26 : 869–877.34159661 22 Price LC , Ridge C , Wells AU . Pulmonary vascular involvement in COVID-19 pneumonitis: is this the first and final insult? Respirology 2021; 26 : 832–834.34322959 23 Katsoularis I , Fonseca-Rodriguez O , Farrington P , Risks of deep vein thrombosis, pulmonary embolism, and bleeding after Covid-19: nationwide self-controlled cases series and matched cohort study. BMJ 2022; 377 : e069590. 24 The ATTACC ACTIV -4a and REMAP-CAP Investigators. Therapeutic anticoagulation with heparin in noncritically ill patients with Covid-19. N Engl J Med 2021; 385 : 790–802.34351721 25 The REMAP-CAP ACTIV -4a and ATTACC Investigators. Therapeutic anticoagulation with heparin in critically ill patients with Covid-19. N Engl J Med 2021: 385 : 777–789.34351722 26 Radermacher P , Herigault R , Teisseire B , Low VA/Q areas: arterial-alveolar N2 difference and multiple inert gas elimination technique. J Appl Physiol 1988; 64 : 2224–2229.3391922 27 Hopkins SR . Functional magnetic resonance imaging of the lung: a physiological perspective. J Thorac Imaging 2004; 19 : 228–234.15502609

PMC009xxxxxx/PMC9527096.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8608542 6627 Pancreas Pancreas Pancreas 0885-3177 1536-4828 35881699 9527096 10.1097/MPA.0000000000002063 NIHMS1814660 Article Urine Proteomics Reveals Sex-Specific Response to Total Pancreatectomy With Islet Autotransplantation Bennike Tue Bjerg PhD 12 Templeton Kate MD 3 Fujimura Kimino MD, PhD 4 Bellin Melena D. MD 56 Ahmed Saima BS 1 Schlaffner Christoph N. PhD 1478 Arora Rohit PhD 9 Cruz-Monserrate Zobeida PhD 10 Arnaout Ramy PhD 9 Beilman Gregory J. MD 6 Grover Amit S. MD 311 Conwell Darwin L. MD 10 Steen Hanno PhD 1 1 Department of Pathology, Boston Children’s Hospital and Harvard Medical School, Boston, MA 2 Department of Health Science and Technology, Aalborg University, Aalborg, Denmark 3 Division of Gastroenterology, Hepatology and Nutrition, Boston Children’s Hospital, Boston, MA 4 F.M. Kirby Neurobiology Center, Boston Children’s Hospital and Harvard Medical School, Boston, MA 5 Department of Pediatrics, University of Minnesota Medical Center and Masonic Children’s Hospital, Minneapolis, MN 6 Department of Surgery, University of Minnesota Medical School, Minneapolis, MN 7 Data Analytics and Computational Statistics, Hasso Plattner Institute for Digital Engineering, Potsdam, Germany 8 Digital Engineering Faculty, University of Potsdam, Potsdam, Brandenburg, Germany 9 Departments of Pathology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA 10 Division of Gastroenterology, Hepatology and Nutrition, Comprehensive Cancer Center, The Ohio State University Wexner Medical Center, Columbus, OH 11 Department of Pediatrics, Harvard Medical School, Boston, MA T.B.B. and K.T. are co-first authors. R.H. currently with Iktos Inc., Boston, MA. Address correspondence to: Hanno Steen, Department of Pathology, Boston Children’s Hospital, 300 Longwood Ave, Boston, MA 02115 (hanno.steen@childrens.harvard.edu) 7 8 2022 01 5 2022 27 7 2022 27 7 2023 51 5 435444 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Objectives: Total pancreatectomy with islet autotransplantation (TPIAT) is a surgical option for refractory chronic pancreatitis related pain. Despite the known clinical implications of TPIAT, the molecular effects remain poorly investigated. We performed the first hypothesis-generating study of the urinary proteome before and after TPIAT. Methods: Twenty-two patients eligible for TPIAT were prospectively enrolled. Urine samples were collected the week prior to and 12–18 months after TPIAT. The urine samples were prepared for bottom-up label-free quantitative proteomics using the ‘MStern’ protocol. Results: Using 17 paired samples, we identified 2477 urinary proteins, of which 301 were significantly changed post- vs pre-TPIAT. Our quantitative analysis revealed that the molecular response to TPIAT was highly sex specific, with pronounced sex differences pre-TPIAT but minimal differences afterwards. Comparing post- vs pre-TPIAT, we found changes in cell-cell adhesion, intracellular vacuoles, and immune response proteins. After surgery, immunoglobulins, complement proteins, and cathepsins were increased, findings that may reflect glomerular damage. Finally, we identified both known and novel markers for immunoglobulin A nephropathy after one patient developed the disease 2 years after TPIAT. Conclusions: We found distinct changes in the urinary proteomic profile following TPIAT and the response to TPIAT is highly sex specific. TPIAT pancreatectomy diabetes autoislet pancreatitis nephropathy pmcINTRODUCTION Chronic pancreatitis (CP) is a progressive fibroinflammatory disease of the pancreas, in which there is a persistent injury to pancreatic parenchyma and eventual destruction of acinar and islet cells, leading to exocrine and endocrine pancreatic insufficiency.1 There are various genetic, environmental, and other risk factors that are involved in the pathogenesis as well as variations in the natural history of CP.1 Historically, alcohol use was thought to be the primary etiology of CP, however it is now clear that underlying genetic mutations and smoking are also important factors in the development of CP.2–6 Chronic pancreatitis is estimated to affect approximately 50 per 100,000 individuals, with recent studies indicating an increasing incidence over the past decade.7,8 Additionally, CP also represents a significant cause of morbidity and financial burden in the United States, with estimated annual costs of $638 million.9,10 Patients with CP often suffer severe abdominal pain that can be difficult to treat and ultimately impact patients’ quality of life.5,11 The management of pancreatitis-related pain includes pancreatic enzyme replacement therapy, analgesia, and more invasive options for those with refractory pain, such as endoscopic retrograde cholangiopancreatography (ERCP), nerve block procedures, and surgical decompression via pancreaticojejunostomy.12 However, for 25% of patients undergoing such surgery, pain relief is temporary at best.13,14 For patients with refractory pain, total pancreatectomy (TP) may be an option. Even though the procedure removes the fundamental cause of the pain, TP immediately results in iatrogenic type 1 diabetes due to a lack of endocrine pancreatic insulin secretion. The resulting diabetes can be quite difficult to manage and can be associated with significant morbidity.15 Autologous islet cell transplantation has become a promising treatment for those patients undergoing TP. Total pancreatectomy with islet autotransplantation (TPIAT) involves isolating the islets from the resected pancreas and infusing them back into the liver where they remain while maintaining their functionality, ie, the natural responsiveness to blood sugar levels and insulin secretory capacity, which can reverse or reduce the risk for post-operative diabetes associated with TP. The first TPIAT surgery was performed at the University of Minnesota in 1977, and as of 2018, approximately 1000 patients have received the treatment.16,17 A study of TPIAT patients from the period 1977 to 2011 reported that patient-survival one (five) year post-TPIAT was 96% (89%) in adults and 98% (98%) in children. Additionally, 90% of the patients had post-TPIAT C-peptide levels greater than 0.6 ng/mL blood, indicating a successful procedure and preserved insulin-producing capacities. Three years post-TPIAT, 30% were still insulin-independent (25% in adults and 55% in children) and an additional 33% retained partial insulin production function (35% in adults and 25% in children), strongly complementing the treatment of type-1 diabetes. Additionally, post-TPIAT, 85% of the patients had an improvement or absence of pain, making the procedure highly attractive as a last-resort treatment option for CP.17 Most TPIAT-related publications so far have focused on the clinical aspects, likely due to the scarcity and nature of TPIAT. The systemic molecular response to TPIAT thereby remains poorly investigated. We recently published the first hypothesis-generating study of the effect of TPIAT on the serum proteome, where we demonstrated that several serum proteins that are known to be altered in various pancreatic diseases, including CP, return to normal levels post-TPIAT.18 In the present study, we sought to investigate the effect of TPIAT on the urine proteome. Urine is an ultrafiltrate of blood from the kidneys and reflects changes in the entire body. Metabolites, proteins, and peptides are concentrated in urine by the kidneys, yet the urine proteome is significantly more manageable compared to the serum- and plasma proteome due to a reduced concentration range.18,19 Our urine proteomics approach captured 1) sex-specific changes following TPIAT, 2) increased urinary concentrations of immunoglobulins, complement proteins, and cathepsins in post-TPIAT samples indicating glomerular damage, and 3) decreased proinflammatory protein abundances in post-TPIAT indicating reduced inflammation. Taken together, the urinary proteomics approach enabled us to capture the overall landscape of the biological mechanisms affected in CP and the impact of TPIAT. MATERIALS AND METHODS Study Cohort and TPIAT Urine Samples At the University of Minnesota Medical Center, samples were collected from 22 patients with CP prospectively enrolled in a TPIAT follow up study between September 2011 and February 2014.20 The study protocol was reviewed and approved by the University of Minnesota Institutional Review Board (IRB) (protocol #0609M91887) and patient consent or parental consent and patient assent was obtained as age appropriate for all participants. Urine samples were collected within the week prior to TPIAT and 12–18 months post-surgery, yielding a total of 44 samples. Clinical information for each participant was obtained, including baseline demographics, pre- and post-operative kidney function, any history of diagnosed renal disease including chronic kidney disease or renal stones, islet cell yield, and islet cell graft function at 1-year post-TPIAT. Graft functions were defined as: 1) full graft (insulin independent at 1-year post-TPIAT), 2) partial graft (insulin dependence at 1-year post-TPIAT with C-peptide >0.6 ng/dl), and 3) failed graft (insulin dependence at 1-year post-TPIAT with C-peptide <0.6 ng/dl). In addition, creatinine samples were obtained at time of morning fasting metabolic blood work before and after TPIAT. Using this creatinine measurement, we retrospectively defined chronic kidney disease (CKD) as estimated glomerular filtration rate (eGFR) <90 mL/min/1.73m2. Urine samples and clinical history were collected under an IRB approved study at the University of Minnesota (IRB# 0609M91887). Patient consent or parental consent as age appropriate was obtained from all participants. Deidentified data and urine samples were then analyzed at Harvard Medical School under an IRB exempt protocol (#2018H0429). Sample Preparation for Proteomics The urine samples were prepared for proteome analysis using the in-house developed MStern blotting approach as previously described.18,21 Briefly explained, 150 μL urine (nominally ~15 μg protein) was mixed with dry urea and ammonium bicarbonate (ABC) to a final concentration of 8 M urea, 50 mM ABC pH 8.5 in a 96-well plate. Protein cysteine disulfide bonds were reduced by addition of dithiothreitol to a final concentration of 10 mM in 50 mM ABC and incubation for 20 min. Reduced disulfide bonds were alkylated by addition of iodoacetamide to a final concentration of 50 mM and incubation for 20 min. From each sample, a volume corresponding to nominally 15 μg of reduced and alkylated protein was transferred to a 96-well hydrophobic polyvinylidene difluoride (PVDF) membrane plate (MSIPS4510, Millipore, Burlington, Mass), which had been primed with 70% ethanol and conditioned with 8 M urea in 50 mM ABC pH 8.5. A 96-well plate adaptable vacuum manifold (Millipore) was used to apply a vacuum and facilitate liquid transfer through the PVDF membrane plate. After adsorption of the proteins onto the membrane, the membranes were washed with 50 mM ABC. Protein digestion was performed by adding 100 μg of 10 μg/mL sequencing grade modified trypsin (V5111, Promega, Madison, Wis) diluted in 5% acetonitrile, 50 mM ABC (nominal trypsin to protein ratio 1:15) and incubated for 2 hours at 37°C in a humidified incubator. The peptides were eluted from the PVDF membrane plate with 40% acetonitrile (ACN), 0.1% formic acid (FA). Subsequently, the elution solutions were pooled and dried in a vacuum concentrator. The dry urine product was stored at −20°C until liquid chromatography–mass spectrometry (LC-MS) analysis. Proteome Analysis The samples were analyzed using a microfluidic LC chip system coupled online to a high resolution/high accuracy Q Exactive mass spectrometer (Thermo Fisher Scientific, Waltham, Mass). The samples were resuspended in 50 μL 5% ACN, 5% FA, and 3 μL was loaded onto a reversed-phase C18 analytical column (PicoChip 150 μm x 15 cm [New Objective, Littleton, Mass]) with 2 μL solvent A (0.1% FA). The peptides were eluted from the column by increasing the ratio of solvent B (0.1% FA in ACN) on a 70 min linear gradient from 7% buffer B (0.1% FA in ACN) to 25% buffer B, at a flow rate of 1000 nL/min a flowrate. The PicoChip with an integrated emitter was kept at 50°C and mounted directly in front of the inlet of the heated capillary of the Q Exactive mass spectrometer, which was operated in positive mode. The mass spectrometer was operated in data-dependent TOP10 (DDA) mode with the following settings: mass range 400–1000 Th; resolution for MS1 scan 70,000 @ 200 Th; lock mass: 445.120025 Th; resolution for MS2 scan 17,500 @ 200 Th; isolation width 1.6 Th; NCE 27; underfill ratio 1%; charge state exclusion: unassigned, 1, >6; dynamic exclusion 30 s. Data Analysis The urine raw files were searched with MaxQuant (v1.5.5.1, Max-Planck Institute for Biochemistry, Computational Systems Biochemistry, Martinsried, Germany) against a database containing the reviewed UniProt human reference proteome (downloaded: 2016-08-21; # of protein entries: 20,209).22,23 Standard settings were employed in MaxQuant, with the match-between-runs feature enabled. A maximum of three tryptic missed cleavages were allowed, as recommended for the MStern protocol. Additionally, the following modifications were found to be abundant with the applied protocol, and were included in the search: carbamidomethylated cysteine residues (fixed), acetylation of proteins N-terminals (variable), oxidation of methionine (variable), and deamidation of asparagine and glutamine residues (variable).24,25 Identified proteins and peptides were filtered to <1% false discovery rate (FDR) using the forward/reverse database search strategy in MaxQuant. Data Sharing The mass spectrometry raw data and protein databases have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository with data set identifier PXD028570.26,27 Statistical Analysis To increase the quantitation accuracy, we employed additional filtering of the ≤1% FDR identified proteins in Perseus (v1.5.5.3, Max-Planck Institute for Biochemistry).28,29 We removed samples (and the corresponding paired sample) with less than 30% of the cumulated list of proteins identified (<743 proteins) from further analysis. Additionally, proteins quantified with only one peptide unique to a protein group, and proteins with less than 70% valid values in either the pre or post condition were removed. Of note, the applied filtering strategy could potentially remove proteins with a specific abundance between males and female subjects. Therefore, we conducted an additional analysis of proteins with a sex-specific abundance, using a dataset filtered for proteins with less than 70% valid values in the samples from males or females. However, the lists of significant proteins between males and females were nearly identical regardless of the applied filtering scheme. Proteins with a statistically significant change of abundance were identified by paired two-tailed t-tests, corrected for multiple hypothesis testing by permutation-based FDR control using standard parameters in Perseus (250 randomizations, FDR <0.05, s0 = 0.1). Protein changes with FDR (q-value) <0.05 were considered statistically significant. For the purpose of conducting Principal Component Analysis, missing values were replaced with values from a normal distribution (width 0.3 and down shift 1.8), to simulate signals from low-abundance proteins.30 To identify underlying biological themes and pathways of differentially expressed proteins, we performed a pathway enrichment analysis using both the Database for Annotation, Visualization and Integrated Discovery (DAVID) database and ClueGo plugin of Cytoscape (v3.8.2, The Cytoscape Consortium, San Diego Calif).31–34 The urinary dataset constitutes by itself a sub-proteome, enriched in selected categories and pathways. To correct for this bias, we used all quantifiable urinary proteins in the present study as a background in the calculations. Additionally, we performed Pearson’s correlation analyses of the LFQ value differences post- vs pre-TPIAT in Rstudio (v1.3, RStudio PBC, Boston, Mass), using a previously developed script.35,36 To ensure reliable correlations we required >50% valid values in both post- and pre-TPIAT samples ensuring at least 7 data points for the correlations, in addition to the mentioned filtering of quantifiable proteins. RESULTS Patient Selection and Proteomics Overview Urine samples were obtained from 22 participants pre- and post-TPIAT, resulting in 44 paired samples. The samples were analyzed by LC-MS/MS. Five samples were found to have less than 30% of the cumulative list of proteins (<743 proteins) and so these samples and their respective matching sample were thus removed from our dataset. This resulted in a complete dataset with 17 matched pre- and post-TPIAT sample pairs (Table 1). Of these 17 participants, following surgery seven were classified as having full graft function, nine were classified as partial graft function, and one had islet graft failure. Additionally, 13 of the 17 participants were female and 4 were male, reflective of the fact that nearly 75% of the adults TPIAT patients are females.14 Subjecting these samples to quantitative mass-spectrometry based proteomics identified over 2400 urinary proteins. To ensure high quality protein data, we applied additional filtering and included only proteins that were identified in at least 12 of the 17 the paired urine samples. This filtering resulted in 1106 quantified proteins (Supplemental Table 1). Proteomic Changes Post vs Pre-TPIAT To exclude potential batching and to investigate the largest variances in the dataset, we conducted an unsupervised principal component analysis (PCA) (Fig. 1A). Approximately half of the variance observed in all the samples were characterized by principal components (PC) 1 and 2. The urine samples all showed a clear pre-post TPIAT trajectory towards lower PC1 values except for two samples (see * on Fig. 1A), with significantly larger differences for the samples from the female than the male participants. Additionally, the samples from all but one male (see + on Fig. 1A) demonstrated distinct clustering with minor changes post-TPIAT. Sex-specific Changes The scores plot showed a distinct clustering of samples based on sex, with a clear separation of the female samples from male samples pre-TPIAT which disappears post-TPIAT with the female samples more closely resembling the male samples (Figs 1B, C). Additionally, protein correlation matrices investigating protein change before and after TPIAT in the female and male cohorts (Supplemental Fig. 1) show that the TPIAT related changes are more pronounced and distinct within the female cohort, compared to the male cohort which shows minimal changes of the urinary proteome following TPIAT. Focusing our analysis on proteins with a specific abundance in the samples from female or male participants, we identified 60 proteins with changed abundance comparing males to females in the pre-TPIAT samples (Supplemental Fig. 2 and Supplemental Table 2A–C), with pro-inflammatory and cell-cell adhesion molecules increased in females and anti-inflammatory proteins increased in males (Fig. 1D). However, we found only 9 proteins with a significant changed abundance between males and females in the post-TPIAT samples. These findings are consistent with the PCA in Figure 1 and the protein-protein correlation analysis (Supplemental Fig. 1), demonstrating that the urinary proteomes from males and females are much more similar post-TPIAT. Analyzing the effect of TPIAT with respect to sex, we plotted the abundances of the 60 sex-specific proteins pre- and post-TPIAT for males and females, respectively (Supplemental Fig. 3). The linear regression of the sex-specific proteins in the male cohort (male y = 0.92x + 1.8) demonstrated that the proteins are relatively stable for the samples post- vs pre-TPIAT with a slope close to a 1 and a minimal intercept. In contrast, the female cohort showed distinct differences post- vs pre-TPIAT as apparent from a slope and intercept deviating from 1 and 0, respectively (female y = 0.76x + 7.8). Effectively, this demonstrates a distinct impact of TPIAT on males and females. Urinary Protein With a TPIAT Specific Abundance To identify urinary proteins with altered abundance post- vs. pre-TPIAT, we performed a sex-independent statistical analysis encompassing all 13 sample pairs (excluding samples with CKD). We identified 301 urinary proteins with a statistically significant induced change of abundance post- vs pre-TPIAT (FDR <0.05). Most of these proteins were related to cell-cell adhesion, stress response, glycolysis, adaptive immune response, and intracellular vacuoles (Fig. 2D). Proteins related to cell-cell adhesion, stress response, and glycolysis were mostly increased in the pre-TPIAT samples whereas proteins related to upregulation of the adaptive immune response and intracellular vacuoles were increased post-TPIAT. Additionally, due to the observed sex differences and independent trajectories post- vs pre-TPIAT, we decided to analyze the male and female TPIAT patients independently (Fig. 2). In the urine proteomics dataset from the male cohort, not a single protein passed the significance threshold for post- vs pre-TIPAT (Fig. 2A). However, in the female cohort 434 urinary proteins showed statistically significant TPIAT-induced abundance differences (Fig. 2B, Supplemental Table 3): 169 with increased and 265 with decreased abundances post-TPIAT. To investigate whether the lack of significant proteins in the smaller male cohort was due to lower statistical power, we correlated the fold-changes of all the significant proteins in the female samples to the fold-changes of the corresponding (non-significant) proteins in the male samples to determine if the fold changes correlated irrespective of the statistical significance. We found poor correlation (R = 0.22) between the fold changes of all the significant proteins in the female cohort to the corresponding fold changes in the male cohort (Fig. 2C), suggesting that the lower number of significant proteins in the male cohort was not explained by the smaller cohort size. Effect of Kidney Function on the Urinary Proteome Upon studying the clinical information for the two outlier samples from the scores plot (see * on Fig. 1), we found that these participants suffered from CKD based on eGFR prior to TPIAT. We found two additional subjects with CKD based on eGFR (n = 4). To investigate the effect of early-stage CKD on the urinary proteome, we identified urinary proteins with altered abundance in CKD samples compared to samples without CKD using the post- vs pre-TPIAT fold change of all urinary proteins. In the CKD samples, we identified 22 proteins with statistically significant change of abundance (FDR <0.05) in comparison to samples without CKD (Fig. 3A). Specifically, we found that the CKD samples had upregulation of proteins related to oxidative stress, glycolysis, and exocytosis (Fig. 3B). Additionally, the scores plot identified another marked outlier (see + on Fig. 1), who was the only urine sample from the male cohort that showed vast differences pre- versus post-TPIAT. Upon further investigation, this patient was found to have developed severe immunoglobulin A (IgA) nephropathy two years after TPIAT. Accordingly, the post-TPIAT sample from this patient had the highest normalized intensity, which is a proxy for urinary concentration, of IgA1 protein compared to the rest of the male samples. To investigate any potential early markers of IgA nephropathy, we included data from a study by Prikryl et al37 comparing urine from IgA nephropathy patients to healthy controls. In the study, Prikryl et al identified 30 urinary proteins with significantly altered abundances (21 more and 9 less abundant proteins) in patients with IgA nephropathy compared to healthy controls,37 with all 30 proteins detected in our study. We then compared the effect of these proteins on the variance observed in our loadings plot, investigating if the variation of the IgA nephropathy patient observed in PCA score (* in Fig. 4A) was due to the 30 biomarkers found by Prikryl et al. Comparing the PC1 PCA loading plot, we found that the 21 more and 9 less abundant proteins had statistically significant different average PC1-coordinate value (combined P = 7.79 × 10−3; more abundant proteins PC1 average coordinate: P = 2.83 × 10−2; less abundant proteins PC1 average coordinate: P = −9.89 × 10−3) with the proteins clustering into two unique groups based on directionality as determined by Prikryl et al (Fig. 4B blue and red circles). Thus, suggesting that the 30 IgA nephropathy biomarkers were significantly contributing to the variance observed within our dataset from the patient with IgA nephropathy. Additionally, we found that those proteins upregulated in IgA nephropathy (including the 21 proteins from Prikryl et al combined with our proposed 10 biomarker candidates) were found to be involved in the humoral immune response, acute inflammatory response, glomerular filtration, and phagocytosis (Fig. 4C). DISCUSSION Over the past two decades, TPIAT has been increasingly performed for refractory pain in the setting of CP.38 However, the systemic molecular response to TPIAT remain poorly understood. We performed the first hypothesis-generating proteomics study of the molecular impact of TPIAT on the urinary proteome. We identified 434 urinary proteins with significantly changed protein abundance in the female cohort, of which 265 proteins were upregulated pre-TPIAT and 169 proteins were upregulated post-TPIAT. Our analysis demonstrated several novel findings, including (1) the response to TPIAT is highly sex-specific and these sex differences were diminished post-TPIAT, (2) the urine proteomic changes observed are dependent on kidney function, and (3) TPIAT specific changes of the urine proteome includes decreased abundance of urinary proinflammatory proteins. Sex-specific Changes According to our analysis, the female and male samples showed clear separation based on principal component analysis (Fig. 1), indicating i) clear sex-dependent differences in the urinary proteomes prior to TPIAT, and ii) clear changes between the female and male urinary composition in response to TPIAT. When examining sex differences with respect to TPIAT status, we found significant sex differences in the baseline pre-surgical samples. Our finding of the significant differences in the urinary proteome of females and males pre-TPIAT could represent sex specific responses to CP or normal variations of the proteomic profile present at baseline. Prior studies have demonstrated sex-specific differences in the proteomic profiles of healthy females compared to males.39,40 In this context, it is most likely that the significant sex specific differences that we observed were due to normal variations of the proteomic profiles between sexes. However, it does appear that the response to TPIAT is highly sex dependent as we found that females had significant changes in their urinary proteome upon TPIAT, however males showed almost no TPIAT-induced change in their urinary proteome. While our study cohort was skewed, containing three times more female participants than males, these sex-specific changes did not appear to be due to smaller number of male participants as there was poor correlation in the TPIAT-induced protein fold change between the females and males (Fig. 2C). Interestingly, we identified that these sex differences were diminished following TPIAT. Based on the PCA plot as well as our linear regression analysis of the abundance of sex-specific proteins (Fig. 1 and Supplemental Fig. 3), we identified a distinct difference in the females with TPIAT and that the abundance of sex-specific proteins decreased post-TPIAT to more closely resemble the urinary proteome of males. While several studies have identified sex differences in the proteomic profile of various bodily fluids,39–41 our study identifies that these sex differences can diminish following TPIAT (Fig. 1). These novel findings are not clearly understood, highlighting the need for future studies to investigate potential causes of why these changes occur. TPIAT-specific Changes From a clinical perspective, one of the major differences after TPIAT is the removal of the chronically inflamed pancreas. To better understand the effect of removing the source of inflammation on the urinary proteome, we analyzed the effect that TPIAT has on different immune system- and inflammation-related proteins. We found that there were clear distinctions in the proteomic profile before compared to after TPIAT. Prior to surgery, there was significant upregulation of proteins related to cell-cell adhesion and proinflammatory proteins. Specifically, there were increased concentrations of actin proteins (actin-1, actin-related protein 2 and 3, and alpha-actinin-4), desmosome proteins (desmoglein 3, desmoplakin, envoplakin, junction plakoglobin, and plakophilin 1 and 3), S100 proteins, interleukin-18, annexin 1, extracellular matrix protein 1, and toll interacting protein in the pre-TPIAT samples. However, following surgery there were increased concentrations of proteins involved in the adaptive immune response and intracellular vacuoles, including cathepsin proteins, immunoglobulins, and complement proteins. The reduced concentrations of proinflammatory proteins after TPIAT likely reflects the removal of the inflamed pancreas following the surgery. Conversely, we observed decreased concentrations in cell-cell adhesion modules as well as the increased concentrations of immunoglobulins and cathepsin proteins in the post-TPIAT samples. We have previously reported a similar increase of most immunoglobulins in the serum post TPIAT.18 While the mechanism of this is unclear, one hypothesis is that this may represent glomerular damage that occurs after the surgery. These post-TPIAT changes are significantly more pronounced in female patients. Prior studies have demonstrated increased urinary concentrations of immunoglobulins complement proteins, and cathepsins in diabetic nephropathy, raising concern that the observed findings could be from early stage diabetic renal disease.42–46 While diabetic nephropathy was historically defined as proteinuria >300 mg/day, this definition is limited by the fact that significant glomerular damage has occurred by the time a patient develops overt albuminuria. Indeed, studies have demonstrated that some maladaptive changes have occurred within the kidney even at the time of diagnosis.47 However, it should be noted that our findings of upregulation of proteins suggestive of glomerular damage post-TPIAT have several limitations. These include the fact that these samples were collected at only 12–18 months after surgery, that all participants were non-diabetic before TPIAT, and that the post-TPIAT HbA1c was <7% in all but one participant which indicates normoglycemia or near normoglycemia. Renal Function and Urinary Proteome The application of urinary proteomics for kidney diseases has long been established, as it has been shown to improve diagnostic accuracy and lead to early diagnosis for various kidney diseases.48–50 Most available studies have focused on case-control analysis of the urinary proteome of patients with CKD to healthy controls or cohort studies in patients with CKD or diabetes mellitus.51 Additionally, prior studies have limited their analysis of the urinary proteome on patient’s who carry a clinical diagnosis of CKD, potentially restricting the discovery of early biomarkers of kidney disease.52 Given our finding of increased concentrations of proteins involved with glomerular damage post-TPIAT, we decided to compare the urinary proteome in CP patients with and without CKD based on pre-operative eGFR. Although many patients in our cohort had related renal disorders (including renal stones, renal cysts, and history of acute kidney injury), none of our study participants carried an official diagnosis of CKD. This may be because these participants had such mild alterations in their creatinine that were not clinically apparent. Despite lacking a formal diagnosis of CKD, we identified four participants who met criteria for early-stage CKD based on pre-operative eGFR. Comparing the urine proteome of the patients with early-stage CKD to those without, we demonstrated that the proteomic profile of patients with CKD is readily distinguishable from patients without CKD. Patients with CKD were found to have increased concentrations of proteins related to oxidative stress and glycolysis. Specific examples include peroxiredoxin (PRDX-1, PRDX-2, and PRDX-6), glutathione S-transferase pi 1 (GSTP1), and phosphoglycerate mutase 1 (PGAM1). These findings align with and expand upon prior studies, which have shown elevation of these proteins in different bodily fluids.53–55 Upregulation of oxidative stress mediators as well as glycolysis proteins in patients with CKD may provide additional insight into the molecular changes that occur within the early stage of the disease. Through our proteomic analysis we also discovered one patient who was diagnosed with IgA nephropathy approximately 2 years after sample collection. Analyzing the concentration of previously identified biomarkers for IgA nephropathy within our data, we found that these biomarkers contributed significantly to the variance observed within our dataset from the patient with IgA nephropathy. This finding suggests the ability to detect IgA nephropathy in urine samples obtained at least two years prior to clinical diagnosis. Using this knowledge, we constructed a list of possible very early onset IgA nephropathy urinary biomarkers, including pancreatic secretory trypsin inhibitor, CD27 antigen, neutrophil defensin, and ganglioside GM2 activator, to mention a few (Supplemental Table 4). As this list is based on one patient only, additional studies are needed to validate these markers, but our findings clearly demonstrate the feasibility of early detection of IgA nephropathy. Limitations There are several limitations to the present study that must be acknowledged. First, the study took place at a single center and as such samples were collected from a single institution, so results may not be generalizable. Second, given the relatively small sample size, the etiology of CP varied among study participants and may have impacted our results. Lastly, the sample size of the present study was relatively small and further studies with larger cohorts of patients to validate our findings may be needed. CONCLUSION In this study, we performed a hypothesis-generating proteomic analysis on urine from participants before and after TPIAT. We report novel findings including highly sex-specific changes that occur with TPIAT, showing pronounced sex differences in the urinary proteome before TPIAT but minimal differences following TPIAT, where after surgery the proteins with a sex-specific abundance in the female urine more closely resembled that of the male urine. We also found that following TPIAT, there was downregulation of inflammatory and adhesion proteins but upregulation of immunoglobulins and cathepsins. Finally, we were able to identify several tentative biomarkers which may be useful for early detection of IgA nephropathy. Overall, our data can provide a basis of the pathophysiology of sex differences observed in the patients undergoing TPIAT and potentially extended to the development of a more personalized selection of the TPIAT applicants, leading to the improved outcomes of the graft function. Supplementary Material Supplemental Data File (doc, pdf, etc.)_1 Supplemental Figure S1. Correlation matrices for the urinary proteins, ordered based on the clustering of the pre-TPIAT females. The strong correlations seen in the pre-TPIAT female samples is not preserved in the post-TPIAT samples, nor found in the samples from the males. The figure highlights that the protein-protein correlations in the female samples pre-TPIAT are diminished post-TPIAT, and that the clustering is distinct to the females. Supplemental Figure S2. Analysis of sex-specific proteins in TPIAT participants. The log-transformed student’s t-test p value of each protein is plotted against the log-transformed fold change. Statistically significant proteins (FDR <0.05) highlighted with proteins significantly increased in females labeled in red () and proteins significantly increased in males in blue (). Supplemental Figure S3. Scatterplot with linear regression of the average abundance of all sex-specific proteins post- vs. pre-TPIAT for males (blue) and females (red), respectively. Ribbon indicates regression confidence interval (0.95). Supplemental Data File (doc, pdf, etc.)_2 Supplementary Table S1: List of quantifiable urinary proteins. Supplemental Data File (doc, pdf, etc.)_3 Supplementary Table S2A-C: List of proteins with a sex-specific abundance in the (A) pre-, (B) post-TPIAT, and (C) healthy samples. Supplemental Data File (doc, pdf, etc.)_4 Supplementary Table S3: List of proteins with a significant change of abundance (q-value<0.05) post- vs. pre-TPIAT in Females. Supplemental Data File (doc, pdf, etc.)_5 Supplementary Table S4: List of tentative very early onset IgA nephropathy urinary biomarkers. Funding source: Research reported in this publication was supported by the National Cancer Institute and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) under award numbers related to The Consortium for the Study of Chronic Pancreatitis, Diabetes, and Pancreatic Cancer (CPDPC): U01DK126300 (M.D.B and G.J.B.) and U01DK108327 (H.S. and D.C.). FIGURE 1. Principal component analysis (PCA) pre- and post-TPIAT. A, Unsupervised PCA scores plot demonstrating grouping of samples based on post/pre TPIAT and sex, with a trajectory of female samples pre- (circles) vs. post-TPIAT (squares) toward the male samples, except for 1 participant (*). In males, post-TPIAT samples showed distinct clustering except for 1 participant (+). Comparison of proteomic signature between males (blue) and females (red) from pre-TPIAT (B) and post-TPIAT (C) samples, with shaded area indicating the 95% confidence region. D, Pathway enrichment of the top gene ontology (GO) terms associated with the proteins significantly increased in females compared to males pre-TPIAT. Labels: , female pre-TPIAT; , male pre-TPIAT; , female post-TPIAT; , male post-TPIAT. FIGURE 2. Comparison of the urinary proteome before and after TPIAT. A and B, Volcano plots of all identified proteins from the urinary samples of males (A) and females (B), with the green and blue dots representing significantly up regulated proteins pre-TPIAT and post-TPIAT, respectively. C, Correlation of TPIAT protein fold-changes between females and males (R = 0.22), for TPIAT proteins with a significant change in female with green dots representing proteins increased pre-TPIAT in both males and females, blue dots representing proteins increased post-TPIAT in both males and females, and grey dots with no correlation between males and females. D, Top GO terms associated with the proteins significantly increased pre- (green) and post-TPIAT (blue). Represented are the three GO terms with the highest FDR from three 3 categories (from top to bottom): biological process, cellular component, and molecular function. FIGURE 3. Comparison of the proteomic profile in CKD patients compared patients without CKD. A, Analysis of CKD specific proteins using the post- vs pre-TPIAT fold-change of all urinary proteins with orange dots representing proteins that significantly increased in CKD samples. B, Pathway enrichment of the proteins significantly increased in CKD with the pink circles representing proteins in involved in oxidative stress and glycolysis and blue circles representing those involved in exocytosis. FIGURE 4. A, Unsupervised PCA of male post- and pre-TPIAT samples with single outlier (*) who developed IgA nephropathy post TPIAT. B, PCA loadings plot of male samples where red (blue) indicates increased (decreased) protein concentration during IgA as reported by Prikryl et al37 and orange indicates top ten potential IgA nephropathy biomarkers. Loadings plot shows clustering of the proteins reported by Prikyl et al into two distinct groupings, those found to be more abundant in IgA nephropathy (red) and those less abundant in IgA nephropathy (blue). C, Pathway enrichment of the top (GO) terms associated with the proteins increased in IgA nephropathy, including the 21 proteins from Prikryl et al37 and the top 10 proposed biomarkers from this present study. TABLE 1. Clinical Inormation of the 17 Included Participants Pre-TPIAT (n = 17) Post-TPIAT (n = 17) Age, mean (SD), y 38 (12.7) 40 (12.6) Sex ratio, F:M, n 13:4 13:4 Weight, mean (SD), kg 70 (20.5) 63 (16.3) Smoking, n (%)  Never 9 (53)  Prior 4 (23.5)  Current 4 (23.5) Alcohol use, n (%)  No 12 (71)  Yes 5 (29) Steatorrhea, n (%) 5 (29) Genetic Abnormality, n (%)  CFTR 2 (15)  SPINK-1 3 (23)  PRSS-1 1 (7) eGFR, Mean (SD), mL/min/1.73m2 124 (53.0) 120 (53.1) HbA1c, mean (SD), % 6.2 (1.7) Insulin, mean (SD), U/d 11.6 (147) Beta-Score, mean (SD) 5.8 (24) Islet graft function, n (%)  Full graft 7 (41)  Partial graft 9 (53)  Failed graft 1 (6) Associated kidney disease, n (%)  Renal stones 4 (23.5) 3 (23)  Renal cyst 1 (6) 1 (6) eGFR indicates estimated glomerular filtration rate; SD, standard deviation. Conflict of interest: The authors declare no conflict of interest. Financial disclosure: T.B. was supported by the Lundbeck Foundation, the Carlsberg Foundation, and A.P. Møller Foundation. K.F. was supported by Overseas Research Fellowship Grant from the Japan Society for the Promotion of Science. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. REFERENCES 1. Whitcomb DC , Frulloni L , Garg P , Chronic pancreatitis: An international draft consensus proposal for a new mechanistic definition. Pancreatology. 2016;16 :218–224.26924663 2. Conwell DL , Banks PA , Sandhu BS , Validation of demographics, etiology, and risk factors for chronic pancreatitis in the USA: A report of the North American Pancreas Study (NAPS) group. Dig Dis Sci. 2017;62 :2133–2140.28600657 3. Singh VK , Yadav D , Garg PK . Diagnosis and management of chronic pancreatitis: A review. JAMA. 2019;322 :2422–2434.31860051 4. Seicean A , Tantau M , Grigorescu M , Mortality risk factors in chronic pancreatitis. J Gastrointestin Liver Dis. 2006;15 :21–26.16680228 5. Hart PA , Conwell DL . Chronic pancreatitis: Managing a difficult disease. Am J Gastroenterol. 2020;115 :49–55.31764092 6. Gupte A , Goede D , Tuite R , Chronic pancreatitis. BMJ. 2018;361 :k2126.29880587 7. Kleeff J , Whitcomb DC , Shimosegawa T , Chronic pancreatitis. Nat Rev Dis Primers. 2017;3 :17060.28880010 8. Muniraj T , Aslanian HR , Farrell J , Chronic pancreatitis, a comprehensive review and update. Part I: epidemiology, etiology, risk factors, genetics, pathophysiology, and clinical features. Dis Mon. 2014;60 :530–550.25510320 9. Peery AF , Dellon ES , Lund J , Burden of gastrointestinal disease in the United States: 2012 update. Gastroenterology. 2012;143 :1179–1187.e3.22885331 10. Mullady DK , Yadav D , Amann ST , Type of pain, pain-associated complications, quality of life, disability and resource utilisation in chronic pancreatitis: a prospective cohort study. Gut. 2011;60 :77–84.21148579 11. Machicado JD , Amann ST , Anderson MA , Quality of life in chronic pancreatitis is determined by constant pain, disability/unemployment, current smoking, and associated co-morbidities. Am J Gastroenterol. 2017;112 :633–642.28244497 12. Steer ML , Waxman I , Freedman S . Chronic pancreatitis. N Engl J Med. Jun 1 1995;332 :1482–1490.7739686 13. Cahen DL , Gouma DJ , Nio Y , Endoscopic versus surgical drainage of the pancreatic duct in chronic pancreatitis. N Engl J Med. 2007;356 :676–684.17301298 14. Friess H , Michalski CW . Diagnosing exocrine pancreatic insufficiency after surgery: when and which patients to treat. HPB (Oxford). 2009;11 Suppl 3 :7–10.20495626 15. Hart PA , Bellin MD , Andersen DK , Type 3c (pancreatogenic) diabetes mellitus secondary to chronic pancreatitis and pancreatic cancer. Lancet Gastroenterol Hepatol. 2016;1 :226–237.28404095 16. Najarian JS , Sutherland DE , Matas AJ , Human islet autotransplantation following pancreatectomy. Transplant Proc. 1979;11 :336–340.109963 17. Sutherland DE , Radosevich DM , Bellin MD , Total pancreatectomy and islet autotransplantation for chronic pancreatitis. J Am Coll Surg. 2012;214 :409–424; discussion 424–406. 22397977 18. Bennike TB , Bellin MD , Xuan Y , A cost-effective high-throughput plasma and serum proteomics workflow enables mapping of the molecular impact of total pancreatectomy with islet autotransplantation. J Proteome Res. 2018;17 :1983–1992.29641209 19. Muntel J , Xuan Y , Berger ST , Advancing urinary protein biomarker discovery by data-independent acquisition on a quadrupole-orbitrap mass spectrometer. J Proteome Res. 2015;14 :4752–4762.26423119 20. Chinnakotla S , Beilman GJ , Dunn TB , Factors predicting outcomes after a total pancreatectomy and islet autotransplantation lessons learned from over 500 cases. Ann Surg. 2015;262 :610–622.26366540 21. Berger ST , Ahmed S , Muntel J , MStern blotting-high throughput polyvinylidene fluoride (PVDF) membrane-based proteomic sample preparation for 96-well plates. Mol Cell Proteomics. 2015;14 :2814–2823.26223766 22. Cox J , Neuhauser N , Michalski A , Andromeda: a peptide search engine integrated into the MaxQuant environment. J Proteome Res. 2011;10 :1794–1805.21254760 23. Cox J , Hein MY , Luber CA , Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol Cell Proteomics. 2014;13 :2513–2526.24942700 24. Bennike TB , Kastaniegaard K , Padurariu S , Comparing the proteome of snap frozen, RNAlater preserved, and formalin-fixed paraffin-embedded human tissue samples. EuPA Open Proteom. 2016;10 :9–18.29900094 25. Bennike T , Ayturk U , Haslauer CM , A normative study of the synovial fluid proteome from healthy porcine knee joints. J Proteome Res. 2014;13 :4377–4387.25160569 26. Vizcaino JA , Deutsch EW , Wang R , ProteomeXchange provides globally coordinated proteomics data submission and dissemination. Nat Biotechnol. 2014;32 :223–226.24727771 27. Vizcaino JA , Csordas A , Del-Toro N , 2016 update of the PRIDE database and its related tools. Nucleic Acids Res. 2016;44 :11033.27683222 28. Tyanova S , Temu T , Sinitcyn P , The Perseus computational platform for comprehensive analysis of (prote)omics data. Nat Methods. 2016;13 :731–740.27348712 29. Cox J , Mann M . 1D and 2D annotation enrichment: a statistical method integrating quantitative proteomics with complementary high-throughput data. BMC Bioinformatics. 2012;13 Suppl 16 :S12. 30. Deeb SJ , D’Souza RC , Cox J , Super-SILAC allows classification of diffuse large B-cell lymphoma subtypes by their protein expression profiles. Mol Cell Proteomics. 2012;11 :77–89.22442255 31. Huang da W , Sherman BT , Lempicki RA . Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4 :44–57.19131956 32. Huang da W , Sherman BT , Lempicki RA . Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009;37 :1–13.19033363 33. Shannon P , Markiel A , Ozier O , Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13 :2498–2504.14597658 34. Bindea G , Mlecnik B , Hackl H , ClueGO: a Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics. 2009;25 :1091–1093.19237447 35. RStudio: Integrated Development Environment for R. 2020. Boston, MA; RStudio, Inc.: 2015 . 36. The R Core Team. R: A language and environment for statistical computing. Vienna, Austria; R Foundation: 2013. 37. Prikryl P , Vojtova L , Maixnerova D , Proteomic Approach for Identification of IgA Nephropathy-Related Biomarkers in Urine. Physiological research. 2017;66 :621–632.28406696 38. Sutherland DE , Radosevich DM , Bellin MD , Total pancreatectomy and islet autotransplantation for chronic pancreatitis. J Am Coll Surg. 2012;214 :409–424; discussion 424–426. 22397977 39. Shao C , Zhao M , Chen X , Comprehensive analysis of individual variation in the urinary proteome revealed significant gender differences. Mol Cell Proteomics. 2019;18 :1110–1122.30894400 40. Guo Z , Zhang Y , Zou L , A Proteomic analysis of individual and gender variations in normal human urine and cerebrospinal fluid using iTRAQ quantification. PLoS One. 2015;10 :e0133270.26222143 41. Xiao X , Liu Y , Guo Z , Comparative proteomic analysis of the influence of gender and acid stimulation on normal human saliva using LC/MS/MS. Proteomics Clin Appl. 2017;11 42. Garsen M , Rops AL , Dijkman H , Cathepsin L is crucial for the development of early experimental diabetic nephropathy. Kidney Int. 2016;90 :1012–1022.27575559 43. Musante L , Tataruch D , Gu D , Proteases and protease inhibitors of urinary extracellular vesicles in diabetic nephropathy. J Diabetes Res. 2015;2015 :289734.25874235 44. Lewandowicz A , Bakun M , Kohutnicki R , Changes in urine proteome accompanying diabetic nephropathy progression. Pol Arch Med Wewn. 2015;125 :27–38.25578432 45. Narita T , Sasaki H , Hosoba M , Parallel Increase in urinary excretion rates of immunoglobulin G, ceruloplasmin, transferrin, and orosomucoid in normoalbuminuric type 2 diabetic patients. Diabetes Care. 2004;27 :1176–1181.15111541 46. Guillén-Gómez E , Bardají-de-Quixano B , Ferrer S , Urinary proteome analysis identified neprilysin and VCAM as proteins involved in diabetic nephropathy. J Diabetes Res. 2018;2018 47. Mogensen CE , Christensen CK , Vittinghus E . The stages in diabetic renal disease: with emphasis on the stage of incipient diabetic nephropathy. Diabetes. 1983;32 :64–78.6400670 48. Gao Y , ed. Urine Proteomics in Kidney Disease Biomarker Discovery. Dordrecht, Netherlands; Springer: 2015. 49. Magalhães P , Mischak H , Zürbig P . Urinary proteomics using capillary electrophoresis coupled to mass spectrometry for diagnosis and prognosis in kidney diseases. Curr Opin Nephrol Hypertens. 2016;25 :494–501.27584928 50. Siwy J , Zürbig P , Argiles A , Noninvasive diagnosis of chronic kidney diseases using urinary proteome analysis. Nephrol Dial Transplant. 2017;32 :2079–2089.27984204 51. Jiang S , Wang Y , Liu Z . The application of urinary proteomics for the detection of biomarkers of kidney diseases. In: Gao Y , ed. Urine Proteomics in Kidney Disease Biomarker Discovery. Dordrecth, Netherlands; Springer: 2015:151–165. 52. Spasovski G , Ortiz A , Vanholder R , Proteomics in chronic kidney disease: The issues clinical nephrologists need an answer for. Proteomics Clin Appl. 2011;5 :233–240.21538916 53. Antonelou MH , Georgatzakou HT , Tzounakas VL , Blood modifications associated with end stage renal disease duration, progression and cardiovascular mortality: a 3-year follow-up pilot study. J Proteomics. 2014;101 :88–101.24549005 54. Bocedi A , Noce A , Fabrini R , Erythrocyte glutathione transferase as a biomarker in kidney health and disease. In: Patel V , Preedy V , eds. Biomarkers in Kidney Disease. Biomarkers in Disease: Methods, Discoveries and Applications (series). Dordrecht, Netherlands; Springer: 2016:577–598. 55. Dessì M , Noce A , Dawood KF , Erythrocyte glutathione transferase: a potential new biomarker in chronic kidney diseases which correlates with plasma homocysteine. Amino Acids. 2012;43 :347–354.21984376

PMC009xxxxxx/PMC9530410.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8914275 1426 Crit Rev Toxicol Crit Rev Toxicol Critical reviews in toxicology 1040-8444 1547-6898 35894754 9530410 10.1080/10408444.2022.2091423 NIHMS1838398 Article Critical review and analysis of literature on low dose exposure to chemical mixtures in mammalian in vivo systems Elcombe Chris S. http://orcid.org/0000-0002-7869-0123 abc Evans Neil P. http://orcid.org/0000-0001-7395-3222 a Bellingham Michelle http://orcid.org/0000-0002-3646-8989 b a Institute of Biodiversity Animal Health and Comparative Medicine, College of Medical, Veterinary and Life Sciences, University of Glasgow, UK b School of Veterinary Medicine, College of Medical, Veterinary and Life Sciences, University of Glasgow, UK Authors’ contributions CSE – Conceptualization, Data curation, Investigation, Methodology, Writing – original draft, Writing – review & editing NPE and MB – Conceptualization, Writing – review & editing c Corresponding author – Chris S. Elcombe: chris.elcombe@glasgow.ac.uk 28 9 2022 3 2022 27 7 2022 27 7 2023 52 3 221238 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Anthropogenic chemicals are ubiquitous throughout the environment. Consequentially, humans are exposed to hundreds of anthropogenic chemicals daily. Current chemical risk assessments are primarily based on testing individual chemicals in rodents at doses which are orders of magnitude higher than that of human exposure. The potential risk from exposure to mixtures of chemicals is calculated using mathematical models of mixture toxicity based on these analyses. These calculations, however, do not account for synergistic or antagonistic interactions between co-exposed chemicals. While proven examples of chemical synergy in mixtures at low doses are rare, there is increasing evidence which, through non-conformance to current mixture toxicity models, suggests synergy. This review examined the published studies that have investigated exposure to mixtures of chemicals at low doses in mammalian in vivo systems. Only seven identified studies were sufficient in design to directly examine the appropriateness of current mixture toxicity models, of which three showed responses significantly greater than additivity model predictions. While the remaining identified studies were unable to provide evidence of synergistic toxicity, it became apparent that many results of such studies were not always explicable by current mixture toxicity models. Additionally, two data gaps were identified. Firstly, there is a lack of studies where individual chemical components of a complex mixture (>10 components) are tested in parallel to the chemical mixture. Secondly, there is a lack of dose-response data for mixtures of chemicals at low doses. Such data is essential to address the appropriateness and validity of future chemical mixture toxicity models. Mammalian Low dose In vivo Exposure Mixture Model Review pmc1. Introduction A wide variety of natural and anthropogenic chemicals are found throughout the environment in air, water, food, soil, and dust. Sources of such environmental chemicals (ECs) include agrochemical food residues, consumer products, industrial chemical effluent, occupational use, and pharmaceutical or recreational drugs, to name but a few. Biomonitoring of EC exposure is routinely performed by various agencies across the world, which commonly see a range of ECs in blood, plasma, and urine samples from across the general population. These include heterocyclic amines, organochlorides, polychlorinated biphenyls (PCBs), polychlorinated dibenzofurans, polycyclic aromatic hydrocarbons (PAHs), metals and metalloids, and volatile organic compounds (VOC) (U.S. HHS 2019). These components, at high doses, have a diverse range of toxicological profiles; many are known to be carcinogenic, hormonally active, hepatotoxic, nephrotoxic, and/or neurotoxic. Co-exposure to ECs from multiple sources, or accumulation of EC exposure over time, is of increasing concern to both regulators and the public, however the exposome (the totality of EC exposure over a lifespan) is particularly difficult to risk assess (Rappaport 2011; Sarigiannis and Karakitsios 2018) and may be having an inpact on increasingly prevailant chronic diseases and viral susceptibility (Tsatsakis et al. 2020). Within traditional chemical risk-assessment, points of departure (PODs) such as the no observable adverse effect level (NOAEL), the highest administered dose which did not produce a statistically significant toxicological effect, are crucial values to determine for every chemical. PODs are defined relative to toxicological dose-response curves for individual chemicals and have been cornerstones of toxicology for decades (reviewed by Dorato and Engelhardt, 2005). Indeed, the idea of a NOAEL was first conceived in the 16th century by the “father of toxicology” – Paracelsus (Temkin 1941). PODs are used to determine the tolerable daily intake (TDI) for a chemical, by dividing PODs by safety factors for inter- and intra-species variance (usually around 100 – 1000). While invaluable in the context of individual chemical toxicological assessments, the toxicological profile of a chemical mixture is not just the sum of the effects of the component chemicals but may be altered by additive and/or synergistic interactions between chemicals in the mixture, even where each component chemical is below accepted individual PODs (here described as “low dose” mixtures). This has led to calls for an urgent assessment of the effects of chemical mixtures on human and environmental health (Bergman et al. 2012; Ribeiro et al. 2017; Drakvik et al. 2020). Assessing risk from exposure to chemical mixtures poses many unique problems, including the near infinite number of possible chemical combinations, interactions between chemicals within mixtures, and the attribution of responses to component chemicals (Bopp et al. 2018). There are two main approaches for evaluating risk from exposure to chemical mixtures: whole-mixture and component-based. Whole-mixture methodologies provide a comprehensive assessment of a specific mixture where unidentified chemical components and inter-component reactions are maintained. This “top-down” approach is analogous to the assessment methods used for individual chemicals (Hernández et al. 2017). Whole-mixture methodologies, however, are not always feasible, due to variation between chemical mixtures, changes due to chemical degradation or metabolism, and factors that affect environmental load, over time. In addition, whole-mixture methodologies do not identify which chemicals within the mixture are responsible for a response(s), nor any interactions between chemicals within the mixture. Component-based methodologies use a “bottom up” approach, where a limited number of chemicals with defined concentrations are mixed and tested, or toxicological information for defined chemical components within a mixture are analysed with additivity models, to assess the risk from mixed chemical exposure (Heys et al. 2016; Hernández et al. 2017). Additivity models, however, assume no compounding toxicodynamics or toxicokinetics, and the required toxicological information for such models can be inconsistent, or absent (Boobis et al. 2011) in particular regarding in vivo data pertaining to chemical interactions (synergy/antagonism) (Heys et al. 2016). Current models to define and predict mixture toxicity fall into four categories: dose addition (DA), response addition (RA, also known as independent action), effect summation (ES), and integrated addition (IA) (Rider et al. 2018). Appropriate choice of model is critical for accurate predictions of mixture toxicity, however equally important is which chemicals within a mixture are grouped together for mixture risk assessment – a subject deeply debated (Kortenkamp 2020). Indeed, these two considerations are connected, and while similarities exist between toxicological models for chemical mixtures, the model which a chemical mixture is thought to follow most closely depends on the individual mixture components. Toxicological similarities between mixture components may occur at one or more levels (Rider et al. 2018). The highest form of similarity is chemicals which share a common active metabolite, as is the case of benzyl butyl phthalate and dibutyl phthalate (which assumes an inactive parent molecule). Next similar are compounds which share a molecular initiating event, and whose toxic events therefore occur via identical pathways, as with parathion and chlorpyrifos which are both acetylcholinesterase inhibitors. At the next level of similarity, which is less defined, chemicals share adverse outcome pathways (AOPs), but convergence of different initiating events may occur at several key events in that pathway. For example, perchlorates and dioxin both reduce circulating thyroid hormone concentrations, but they act to reduce the production of, and increase the elimination of, thyroid hormones, respectively (Boas et al. 2006). Less similar still are chemicals which induce toxicity within the same organ system but via different mechanisms. For example, caffeine and ephedrine can individually produce cardiotoxicity through separate mechanisms, but when co-administered act synergistically to produce enhanced toxicity (Dunnick et al. 2007). Toxic additivity of chemicals at this level of similarity within a mixture is a topic of debate, as they may or may not compound each other. Finally, toxicologically least similar are chemicals which share a common disease outcome. For example, diethylstilbestrol and cyclophosphamide are both recognised carcinogens but act via different mechanisms of action and can affect different organ systems. Here, the nature of the disease is of greater importance. Reports indicate advances in terms of hazard identification, exposure and risk assessment, and subsequent risk management relative to co-exposure to chemicals in mixtures (reviewed by Bopp et al. (2019)) as well as harmonisation of methodologies for mixture toxicity assessment (EFSA Scientific Committee, 2019). However, much of the focus of this work has been towards intentional mixtures, such as formulations within pesticides or cosmetic products, and does not cover doses below PODs (Rotter et al. 2018). Furthermore, even less attention has been paid to ‘real-life’ chemical exposure, which is characterised as being to vastly more complex and varied mixtures, at doses far below PODs. This pattern of exposure is that to which the wider human and wildlife populations are exposed throughout their lifetimes. Regulatory agencies often review literature on mixture toxicity. Although varying conclusions by different bodies, the general approach is to separate chemicals into groups with distinct modes of action before individually applying additive models to each group for risk assessment, noting that in the case of knowledge gaps dose additivity should be assumed and that interactions between chemical components are rare, and generally only occur at mid to high doses (NAS, 2017; OECD, 2018; SCHER/SCCS/SCENIHR, 2013; U.S. EPA, 2007; WHO, 2009). Few publications, however, focus on EC mixtures around or below POD doses, or chemical interactions at these doses. A review of literature concerning low dose mixtures of pesticides from 1985–1998 concluded that exposure to such mixtures is not a source of concern to human health (Carpy et al. 2000). An EU commissioned report on the state of the art of mixture toxicity risk assessment concluded that generally mixture toxicity risk assessments by dose addition models were reliable. However, they also noted several important examples of synergy which highlights the need for clarity as to which chemicals to group in cumulative risk assessments (Kortenkamp et al. 2009). A review of low EC dose synergy noted synergy as rare and that the effects of synergy did not exceed additivity models by more than a factor of 4 (Boobis et al. 2011). The most recent and most extensive review of low dose mixtures was performed by the European Centre for Ecotoxicology and Toxicology of Chemicals (ECETOC). Assessing mixture toxicity studies where component concentrations were at or below NOAELs and examining toxicity reported as more than expected based on current models of mixture exposure (i.e., more than additivity in the case of similar chemicals, and more than independent action in the case of dissimilar chemicals), ECETOC concluded that there is no evidence of a risk to human health from exposure to complex mixtures of components where each is below regulated levels (ECETOC 2012). However, to only consider effects provably greater than model predictions, rather than effects unsatisfactorily explained by model predictions, is tantamount to requiring proof of synergistic toxicity – a considerably larger endeavour. To place the burden of proof on the literature rather than the toxicity model might also be considered contrary to the traditionally conservative nature applied in toxicology. Still, despite this requirement, over 5% of studies reviewed by the ECETOC showed a deviation from additivity that suggested synergistic chemical effects (ECETOC 2012). Here we critically review the literature which has investigated the effects of exposure to chemical mixtures in mammalian in vivo systems, with each component at or below respective POD doses. The review covers the period of 2000 to 2020, inclusive, as a resource collating the most recent literature for those interested in mixtures research and to identify and highlight areas in need of additional research. 2. Methodology A literature search was conducted across multiple journal databases (PubMed, Google Scholar, and SCIRUS) for peer reviewed studies published between 2000 and 2020 relating to low dose mixtures of chemicals in mammalian in vivo systems. Low dose mixtures were defined as those in which all components were at or below their POD. Identified literature with at least one experimental group fulfilling these criteria were included for discussion. Figure 1 illustrates the logic underpinning the literature search and subsequent evaluation for inclusion. To identify any missed literature a wider search of non-scientific sources (Google) was performed, and publication histories of relevant authors or research groups were interrogated. For each identified study which met all inclusion criteria, detailed experimental information was extracted (mixture components, dosing levels and regime, route, species, duration, and timing). Reported significant outcomes relevant to exposure were captured and summarised. Results from authors’ original analyses were used, and no statistical re-analysis was performed. 3. Results 3.1. General findings Literature searches identified sixty-eight research articles which met the inclusion criteria. Fifty-four reported findings from studies that used component-based methodologies, and fourteen from studies that used whole-mixture methodologies. Detailed experimental information can be found in supplementary tables. Thirty research articles that used component-based methodologies tested mixtures with at least one component at or close to its NOAEL. In nineteen of these research articles this was the lowest dose tested. Twenty-one research articles that used component-based methodologies had at least one group exposed to a mixture with all components at or below TDI values. Research articles with component-based methodologies were divided into groups, firstly on the number of components (simple mixtures that contained ≤10 individual chemicals, complex mixtures that contained >10), and secondly simple mixtures were divided based on commonality between the characteristics of individual mixture components. These secondary groups were antiandrogen mixtures, antiandrogen plus xenoestrogen mixtures, mixed endocrine disrupting mixtures, neurotoxic mixtures, and mixed mode of action mixtures. Identified research articles that used a whole-mixture methodology were from studies that used one of two experimental models: the biosolid treated pasture (BTP) sheep model, and a disinfection by-product model. As such, that literature was grouped by model type, and then for BTP articles also as a function of age; effects in the fetus, juveniles, and adult offspring. 3.2. Component-based methodologies 3.2.1. Antiandrogen mixtures Six research articles tested mixtures of known antiandrogens and chemicals with antiandrogenic action. Phthalates were included as while not strictly antiandrogens, phthalates induce antiandrogenic effects through non-androgen receptor mediated mechanisms. Effects on sexual development were reported in male rat offspring following in utero exposure via parental oral dosing with antiandrogen mixtures with individual chemicals at/below NOAEL. Exposure to a mixture of vinclozolin, procymidone, linuron, prochloraz, BBP, DBP, and DEHP at NOAEL resulted in increased areola numbers at post-natal day (PND) 13 (Rider et al. 2008). Nipple retention (NR) was observed following exposure to a mixture of vinclozolin, procymidone, linuron, prochloraz, BBP, DBP, and DEHP, DiBP, DiHP, and DPP at NOAEL (Rider et al. 2010). Exposure to a mixture of vinclozolin, finasteride, prochloraz, and DEHP at NOAEL resulted in reduced anogenital distances (AGD) at birth and increased nipple retention (NR) at PND13 (Christiansen et al. 2009). However, a study by Schneider et al. (2017) reported no effects of chemical exposure on male or female offspring at PND1, 12, 20, or 83, following gestational and lactational exposure to a mixture of flutamide, prochloraz, and vinclozolin at TDI and NOAEL (parental oral and direct oral dosing after weaning). The apparent lack of an effect of exposure to the chemical mixture was despite observation of multiple statistically significant physiological effects, both on the mothers, such as increased gestation length and lower oestradiol concentrations, and male offspring, which exhibited decreased progesterone concentrations and decreased absolute cauda, epididymis, and relative bulbourethral gland weights. These effects were not attributed to chemical exposure due to the absence of significant effects in other organs which are known to be more sensitive to androgen disruption, and the limited magnitude of the observed effects. In the same study, however, rats were exposed to flutamide, prochloraz, and vinclozolin at LOAEL, both individually and as a mixture (Schneider et al. 2017). In these groups they observed greater than additivity (i.e., potentially synergistic) in the effects of chemical mixture exposure with regards to increased NR and delayed sexual maturation in male offspring, and increased androstenedione in mothers (Schneider et al. 2017). This would indicate that the lack of effects seen at or below NOAEL may be due to the limited number of chemical components within the mixture, rather than a lack of synergy between mixture components. As noted above, authors reported no effects from co-exposure to flutamide, prochloraz, and vinclozolin at NOAEL or TDI, but greater than additive effects when the chemicals were administered at LOAEL (Schneider et al. 2017). More in-depth comparison of chemical effects to additivity models were reported in the previous three studies (Rider et al. 2008, 2010; Christiansen et al. 2009), which investigated more complex mixtures of antiandrogens. Rider et al. (2008) reported logistically regressed ED50 values for hypospadias and undescended testes from co-exposure to vinclozolin, procymidone, linuron, prochloraz, BBP, DBP, and DEHP, which differed with statistical significance to those from additivity models (DA, TEF, RA, IA), with each model predicating higher ED50 values than those observed, while models based on additivity (DA and/or TEF) were accurate for predicting AGD shortening, epididymal agenesis, ventral prostate weight reductions, epididymal and seminal vesicle weight, and gubernacular agenesis. On balance, Rider et al. (2008) reported that additive toxicity models were generally accurate despite mechanistically distinct chemicals examined, which would have been risk assessed by RA or IA. Similarly, with regards to vinclozolin, finasteride, prochloraz, and DEHP co-exposure, the differences between effective doses for producing AGD shortening, cleft phallus, and malformations in general derived from observations and from additivity models (DA, RA) were statistically significant, with models predicting higher effective doses than those observed (Christiansen et al. 2009). However, Rider et al. (2010) showed DA additivity model predictions to be accurate for all studied endpoints (AGD reduction, various malformations, NR, organ weights) except ventral prostate weight, where the effect magnitude was under-predicted. In Rider et al. (2010) non-phthalate components were identically dosed as in Rider et al. (2008), however, 6 phthalates were used instead of 3 while total phthalate dosage was maintained. That DA model predictions were accurate in Rider et al. (2010) yet inaccurate in Rider et al. (2008) suggests unknown interactions between phthalates and other mixture components, yet current guidelines would use an RA approach to these mixtures, which gave inaccurate predictions in both Rider et al. (2008) and Rider et al. (2010). Discrepancies between model predictions and observations despite dissimilarities between antiandrogenic mechanisms across mixture components supports concerns regarding compounding and/or synergistic effects of antiandrogens on the fetus, and of improper considerations of additivity/synergy between mechanistically distinct antiandrogens by currently employed mixture toxicity models. Two studies were identified that investigated mixtures containing only phthalate plasticisers. Exposure of juvenile male rats to a mixture of six phthalates (DMP, DEP, DBP, BBP, DEHP, and DnOP), at doses equivalent to 1x or 0.1x NOAEL, via dietary exposure for 15 weeks, resulted in reduced body weights and altered the weight of several vital organs (Gao et al. 2017). The exposed mice had lower testosterone and higher luteinizing hormone concentrations compared to the controls and the testes showed increased deciduous spermatids (at both dose levels), and reduced concentrations of several proteins involved in testicular development and spermatogenesis (at 1x NOAEL) (Gao et al. 2017). Effects on body weight and the reproductive system were also seen when male mice were exposed to a mixture of 3 phthalates with 2 alkyl-aromatics (DEHP, DBP, BBP, NP, and 4-tert-octylphenol) at doses equivalent to individual TDI values from conception to PND60 via drinking water. In this instance, the mixture exposed mice exhibited increased body weights and reduced relative testes weights compared to the controls (Buñay et al. 2018). Within the testes of exposed mice, the seminiferous tubules showed multiple signs of damage: tubule widths were reduced, germ cell exfoliation was increased, a greater proportion of tubules were without lumen, and steroidogenic gene expression was altered. Germ cells also showed signs of disruption: an increase in germ cell death, apoptotic signalling, and a greater proportion at an undeterminable stage of spermatogenesis due to the high level of degeneration/atrophy (Buñay et al. 2018). Of these two studies, only that of Buñay et al. (2018) also tested for the effects of phthalates individually in parallel to the mixture (at TDI doses), where each effect observed in response to the mixture was also observed in response to an individual chemical, however the number of animals affected was, in general, greater when exposed to the mixture of chemicals. 3.2.2. Antiandrogen plus xenoestrogen mixtures Seven research articles tested mixtures of known antiandrogens and/or xenoestrogens. Six of these investigated offspring effects following gestational and lactational exposure to a mixture of genistein and vinclozolin via parental oral dosing, in rats. Both chemicals were administered at 1 mg/kg/day, which is below the NOAEL for vinclozolin and at the levels of genistein expected for animals on a soya-bean diet, which is accepted as below the TDI for genistein. Following exposure to this mixture, litter sizes were significantly smaller, and pups were significantly heavier than unexposed controls (Eustache et al. 2009). While litter sizes were also smaller, and with heavier pups, following exposure to genistein alone, the effect was greater when mothers were exposed to the mixture of genistein and vinclozolin. Mixture exposed male offspring exhibited reduced testosterone secretion upon ex vivo stimulation at PND3 (Lehraiki et al. 2011), AGD shortening and more frequent immature penile development at weaning, and decreased epididymis and seminal vesicle weights, sperm quality and quantity, and increased FSH and oestradiol at PND80 (Eustache et al. 2009). However, of these effects, all except reduced seminal vesicle weight and lower sperm quantity were also seen in animals dosed with either genistein or vinclozolin alone. Interestingly, microarray analysis of testicular gene expression in animals at PND80 also demonstrated a correlation between the effects of the low dose mixture and individual exposure to a 30x higher vinclozolin dose, which would suggest a synergistic action between the two chemicals when present in the mixture (Eustache et al. 2009). Mixture exposed female offspring exhibited earlier vaginal opening, and multiple histopathological changes in mammary gland structure at PND35, including increased duct sizes, thicknesses, and cellular proliferation, however some of these effects were also noted in animals dosed with either genistein or vinclozolin alone (Saad et al. 2011). Female rats exposed to genistein and vinclozolin from conception through weaning via parental oral dosing also exhibited fewer striated submandibular salivary ducts, an increased area of striated salivary ducts, and decreased proliferation with lower expression of growth factors EGF, NGF, and TGFα in submandibular salivary glands, however most of these findings were also seen in animals exposed to just vinclozolin, although the effect size was greater when exposed to the mixture (Kouidhi et al. 2012). Direct and multigenerational effects on bone formation in the paw and spine following gestational exposure to genistein and vinclozolin alone or in combination with BPA have been reported. Specifically, alterations in forepaw digit lengths of F1 (exposed in utero) and F2 (not directly exposed) males towards a more feminine digit length ratio have been reported following F1 gestational exposure to mixtures of various combinations of genistein at sub-NOAEL, vinclozolin at NOAEL or TDI, and BPA at TDI (Auger et al. 2013). Histopathological alterations in the vertebral growth plates (intertransverse apophysis width increases and v5 to v8 vertebral length shortening) has been reported in F1 females, but not males, at PND30–110 following similar exposure to genistein, vinclozolin and BPA, however these effects were also seen when F1 animals were exposed to individual chemicals and did not continue into the F2 generation (Auxietre et al. 2014). Interestingly, F1 males exposed to vinclozolin and/or genistein at sub-NOAEL, alone or as a mixture, were unaffected, while co-exposure of either vinclozolin at the lower TDI dose, or genistein, at sub-NOAEL, with BPA at TDI resulted in histopathological alterations in the vertebral growth plates, but these findings were not present following exposure to genistein, vinclozolin and BPA, at these levels (Auxietre et al. 2014). In all the above studies, where individual chemicals were tested in parallel to mixtures of the chemicals, some effects were also noted from exposure to the individual chemicals, in some cases effects were only noted from exposure to an individual chemical. However, mostly, effects were more numerous and/or greater in size when animals were exposed to the mixture. While the above studies noted compounding effects from chemicals when administered as a mixture, compared to when administered individually, the remaining study concluded that co-exposure to certain chemicals could, through counteraction measures, result in fewer physiological effects. This conclusion was based on the observation that, compared to controls, mice exposed (via parental oral dosing during gestation and lactation) to sub-LOAEL doses of DEHP (daily) or BPA (daily), or TCDD (single treatment) had lower body weights (PND14 and PND21), increased relative brain weights at PND14, and lower absolute brain weights at PND42, with changes in midbrain dopaminergic nuclei at both PND14 and PND42. However, when animals were dosed with the chemicals as a mixture, body weight was unaffected, brain weight (relative and absolute) was increased at PND14, and there were no signs of changes in midbrain dopaminergic nuclei (Tanida et al. 2009). 3.2.3. Mixed endocrine disrupting chemical mixtures Fourteen research articles tested mixtures of known EDCs. One study investigated the effects of drinking water disinfection by-products at higher levels than maximum contaminant levels (MCLs), but with two groups (500x and 1000x MCL) ≤NOAEL (0,5x and 1x, respectively). Pregnant rats and their offspring were exposed from F1 GD0 to F2 PND6 via drinking water. Water consumption was reduced in treated dams intermittently during gestation and consistently throughout lactation at 1000x MCL, and in F1 females at 500x and 1000x MCL. F1 pup weights were reduced in males at PND26 and in females at PNDs 21 and 26, and the onset of puberty was delayed in both F1 sexes at 1000x. No effects were seen in the F2 generation (Narotsky et al. 2015). Multiple research articles were identified that had investigated mixtures of endocrine disrupting pesticides at low doses. When juvenile rats were exposed, via a standard diet, to three azole fungicides (cyproconazole, epoxiconazole, and prochloraz) at individual doses equivalent to 1x NOAEL and 0.01x NOAEL (TDI) either no adverse effects were seen (Schmidt et al. 2016), or effects were noted on organ weights and circulating hormone levels, but these effects were not attributed to exposure to the chemical mixture as they were only seen at the 0.01x NOAEL dose and not at 1x NOAEL (Rieke et al. 2017). However, gestational, or gestational and lactational exposure to 5 fungicides of different classes (procymidone, mancozeb, epoxiconazole, tebuconazole, and prochloraz), via parental oral gavage, at individual doses equivalent to 0.083x – 1x NOAEL, did result in adverse reproductive effects. In fact, increased gestation duration, pup mortality and birthing complications were so severe at 0.75x and 1x NOAEL doses that this part of the experiment had to be discontinued (Jacobsen et al. 2010; Hass et al. 2012). Against the lower doses, the pups exhibited reduced birth weights, females had increased AGD, whereas males had decreased AGD, and increased incidence of genital dysgenesis and NR (Jacobsen et al. 2010; Hass et al. 2012). At weaning, relative liver weights were reduced in both sexes, and in males prostate and epididymis weights were reduced (Jacobsen et al. 2010). When allowed to mature, the relative weights of sexual organs were reduced (PND16 and 280), and reduced sperm counts, and increased prostate hyperplasia were observed (PND280) (Jacobsen et al. 2012). When the mixture dose response data was compared to mixture toxicity models (DA and IA) using dose response data from individual chemicals, a synergistic effect was reported for the mixture on gestation length and NR, and potentially also towards genital malformations however synergy could not be distinguished from a marked joint effect as no significant genital malformations were observed in response to chemicals individually (Hass et al. 2012). Mice that received, from conception to PND98, dietary exposure to a mixture of the pesticides atrazine, chlorpyrifos, and endosulfan, at doses equivalent to individual TDI values, expressed increased circulating glucose concentrations at PND21 and PND98, and by PND98 a significant reduction in bodyweight was observed (Demur et al. 2013). At PND21, the EC exposed animals, of both sexes, were characterised by lower levels of many amino-acids and nutrients compared to the controls, however, by PND98 this profile was reversed in the males (Demur et al. 2013). Effects on cognition from gestational exposure to atrazine in combination with 3 different EDCs (PFOA, BPA, and TCDD), all at doses equivalent to TDI values or below NOAEL values, have also been reported. The results indicated reduced environmental habituation in exposed mice of both sexes, and males were found to display decreased exploratory behaviour, short-term memory, and attention to task, relative to the controls (Sobolewski et al. 2014). In both above cases which looked at the effects of chemical mixtures including atrazine, effects from exposure to individual chemicals were also noted, however the observations were of greater effect sizes when chemical exposure was to a mixture, which indicates at least additivity despite separate toxicological mechanisms. Administration of a mixture of TCDD, PCB-153, DEHP, and BPA through dietary supplementation equivalent to TDI induced oestrogeno-mimetic and metabolic effects in mice, effects which were also shown to be influenced by diet. At 7–12 weeks old, various metabolic changes were observed in male and female offspring of parents exposed prior to mating, through gestation and lactation, and which received direct exposure after weaning, via supplementation of a standard (Labaronne et al. 2017) or high-fat-high-sucrose (HFHS) diet (Naville et al. 2013, 2015), with the EC mixture. These include changes in glucose tolerance, circulating and hepatic cholesterol, triglyceride, fatty acid levels, and adipose tissue mass. In the HFHS diet groups, changes in hepatic metabolic gene expression were reported in response to the mixed chemical exposure, which included increased arylhydrocarbon receptor (Ahr) and peroxisome proliferator-activated receptor α (Pparα) gene expression, and alterations in the expression patterns of various xenobiotic transforming and metabolic enzymes (Naville et al. 2013, 2015). In animals which received the chemical mixture in the standard diet, enriched hepatic gene expression was found relative to drug and xenobiotic metabolism, steroid biosynthesis, and fatty acid metabolism (Labaronne et al. 2017). A follow up study in juvenile male mice compared the effect of standard or HFHS with and without EC supplementation (Naville et al. 2019). Metabolic changes like those previously recorded against a HFHS diet were reported in response to EC exposure, however, several effects reversed in direction between standard and HFHS diets or were only present when administered in one type of diet (Naville et al. 2019). Studies in ovariectomised female mice receiving the same chemical exposure (TCDD, PCB-153, DEHP, and BPA at TDI equivalent doses in HFHS diet) demonstrated oestrogeno-mimetic effects from exposure to the chemical mixture from prior to conception to a juvenile age, with additional metabolic alterations including insulin and glucose tolerance (worsened with EC exposure in sham operated but alleviated with EC exposure in ovariectomised mice) and increased hepatic triglycerides (Julien et al. 2018, 2019). EC mixture exposure after ovariectomy, with or without oestrogen replacement, also resulted in tissue and steroid replacement specific effects on steroid hormone production and several oestrogen pathways (Julien et al. 2019). Consistent across these studies was the observation of an increase in hepatic oestrogen receptor expression, which was also observed in male mice exposed to the chemical mixture via standard and HFHS diets (Julien et al. 2018, 2019; Naville et al. 2019). None of the above studies which tested mixtures of TCDD, PCD153, DEHP, and BPA concurrently investigated the physiological effects of individual chemical exposure. 3.2.4. Neurotoxic mixtures Five research articles tested mixtures of chemicals which alone can produce neurotoxic effects. Although the primary toxicity of organophosphorus insecticides is that of neurotoxicity, the identified research which investigated the effects of organophosphorus insecticide mixtures demonstrated hepatic and nephrotoxic effects. Adult male rats, exposed to a mixture of diazinon, chlorpyrifos, malathion, and profenofos for 28 days at doses equivalent to individual NOAEL values, via oral gavage, exhibited reduced body weight and increased relative liver weight (Mossa et al. 2011). Upon closer examination, liver damage was evident, as indicated by histopathological changes (dilatation and congestion of central veins sinusoids, dilated cystic bile ducts, oedemas, and hepatocyte degradation) as well as serum clinical chemistry indicative of liver damage (increased ALT, AST, ALP, and ChE) (Mossa et al. 2011). Whereas exposure of rats to a mixture of dichlorvos, dimethoate, acephate, and phorate, at doses equivalent to respective individual NOAEL values, for 24 weeks, via drinking water, found no effect on hepatic antioxidative defence mechanisms or lipid peroxidation (Yang et al. 2012). However, increased circulating triglycerides, lipoproteins, and cholesterol concentrations, and a greater incidence of renal tubular epithelial cell swelling and granular degeneration have been reported following the same EC mixture exposure (Du et al. 2014). Organophosphorus insecticides were tested only individually in parallel with the chemical mixtures in the study by Yang et al. (2012), where no effects were seen at NOAEL following exposure to individual chemicals or the mixture, and they did not compare observations to mixture toxicity models. Despite pyrethroids being the most widely used commercial insecticide, only three research articles were identified that investigated low dose pyrethroid mixtures. Mild signs of neurotoxicity were seen in adult rats orally dosed with a mixture of 11 pyrethroid pesticides (S-bioallethrin, permethrin, cypermethrin, deltamethrin, esfenvalerate, β-cyfluthrin, fenpropathrin, tefluthrin, λ-cyhalothrin, bifenthrin, resmethrin) at doses below individual NOAEL values. For 8 hours after exposure to the pyrethroid mixture, at both 43% and 85% individual NOAEL values, animals displayed mild whole-body tremors and reduced motor activity (Wolansky et al. 2009). When rats were dosed with all pesticides simultaneously, the mixture produced effects which conformed well to DA model predictions based on the effects of each individual pesticide, but when the chemicals were dosed in three sets to align the effect maxima of individual chemicals (at 1, 2 or 4 hours) the threshold was 3.7x lower than dose addition model predictions. As this reduction in threshold was not statistically significant (p=0.07) the study concluded that the data supported the suitability of DA models for predicting the effects of this pesticide mixture (Wolansky et al. 2009) as opposed to there being a synergistic mixture effect. A similar conclusion was reached when it was shown that a mixture of tefluthrin, bifenthrin, cypermethrin, and deltamethrin dosed to juvenile rats at doses determined in a preliminary study to be below that which individually cause a lowering of body temperature, did not influence body temperature as a mixture (Ortega et al. 2018). Similarly, no significant effects were found on ventilation efficiency, pulmonary perfusion, cardiac output, or metabolic rate in mice following inhalation of prallethrin and phenothrin, each at NOAEL, individually or as a mixture (Santiasih et al. 2020). 3.2.5. Mixed mode of action mixtures Ten research articles studied mixtures of chemicals which alone can produce adverse effects through various modes of action. Nine of these investigated the physiological effects of exposure to mixtures of pesticides with mechanistically distinct modes of action. Two pesticide mixtures induced haematological changes. Dietary exposure to alphacypermethrin, bromopropylate, carbendazim and mancozeb (neurotoxic, hepatotoxic, EDC) at NOAEL equivalent doses, and chlorpyrifos ranging from 0.04x NOAEL to 1x NOAEL, lowered haematocrit, haemoglobin, and red blood counts, with liver and thyroid weights increased, and thymus weight reduced (Jacobsen et al. 2004). While exposure to a mixture of alachlor, captan, diazinon, endosulfan, maneb, and mancozeb (neurotoxic, hepatotoxic, EDC), at TDI doses via thrice weekly oral gavage, altered bone marrow colony compositions in terms of cell type proportions and haematopoietic protein levels, reduced liver weight and increased spleen weight, with distinct and distinguishable metabolic profiles for both exposure and sex (Merhi et al. 2010). Neither study investigated exposure to individual chemicals in parallel to the mixture. Reproductive effects were reported in male and female rats following dietary exposure to a mixture of dicofol, dichlorvos, permethrin, endosulfan, and dieldrin (hepatotoxic, neurotoxic, EDC) at NOAEL equivalent doses. Specifically, male rats had reduced sperm motility and increased numbers of immotile sperm (Perobelli et al. 2010), while various effects were seen in female rats which appeared to be dependent on strain: decreased oestrous cycle and diestrus period lengths in Lewis rats, decreased ovarian primordial and primary follicles, antral follicles, and corpora lutea in Sprague-Dawley rats, and interestingly, no statistically significant effects in Wistar rats, the most frequently used experimental strain (Pascotto et al. 2015). Individual chemicals were only tested in parallel to the mixture in males, where effects were also observed for some individual chemicals, but the effect size was greater when animals were exposed to the pesticide mixture (Perobelli et al. 2010). This could indicate synergistic effects between chemicals within the mixture, however, the study design did not provide dose response relationships appropriate to compare observations to mixture toxicity models. Reproductive effects have also been reported following gestational exposure via maternal dietary supplementation with cyromazine, MCPB, pirimicarb, quinoclamine, thiram, and ziram (reprotoxic, nephrotoxic, neurotoxic, teratogenic), at doses related to 0.05x – 0.375x of the benchmark doses for a 5% birthweight reduction, which were derived from regulatory draft assessment reports (DARs) (Hass et al. 2017; Svingen et al. 2018). Body weight was reduced in both dams and pups through gestation and from birth to PND16 (Hass et al. 2017) following EC mixture exposure, comnpared to untreated controls. EC exposed male offspring had reduced relative liver and retroperitoneal fat pad weights at PND16, but not at 5–6 months of age (Svingen et al. 2018), relative to untreated controls. Overall, at 5–6 months old, no statistically significant differences were observed in the male offspring, whereas the females showed significantly elevated plasma leptin concentrations relative to the controls (Svingen et al. 2018). Although pesticides were not tested individually in these studies, the empirically derived chemical mixture LOAEL for birthweight was used to determine the dose level to compare to DAR derived LOAEL values. As each chemical was at between 0.2x – 0.09x respective LOAELs, a mixture of 6 chemicals producing the reported effects is in approximate agreement with DA model predictions, which supports the appropriateness of the model (Hass et al. 2017). Hepatic effects were reported following 12 months dietary exposure to a mixture of ziram, chlorpyrifos, thiacloprid, boscalid, thiofanate, and captan (neurotoxic, EDC, hepatotoxic) at TDI equivalent doses in wild type (WT) and Car knockout (KO) C57Bl/6J mice (Lukowicz et al. 2018). Exposure to the chemical mixture was associated with increased body weight gain in both strains of mice. WT mice exhibited increased hepatic steatosis and triglycerides, with decreased fasting glucose and glucose tolerance, and altered glutathione redox states. KO mice did not exhibit the same hepatic responses, and predictably displayed differential hepatic expression of many detoxifying enzymes and circulating levels of metabolites. Interestingly, subsequent microarray analysis of WT livers identified over 500 genes were upregulated, and over 500 down regulated, in both sexes in response to chemical exposure but less than 10% of the gene changes were common between the sexes. Pathway analysis also highlighted enrichment in differing pathways between sexes (Lukowicz et al. 2018). These results highlight that the effects of exposure to chemical mixtures can be sex specific or sexually differentiated - an important element when considering studies in which only one sex was examined or where sex is either unconsidered or unreported. Neurobehavioral effects were reported in male and female rats exposed to diquat, imazamox, bentazone, imazethapyr, tepraloxydim, and acifluorfen (hepatotoxic, teratogenic, genotoxic) via drinking water (Tsatsakis et al. 2019b; Sergievich et al. 2020). Exposure to the mixture at 0.25x and 1x NOAEL doses resulted in decreased anxiety at 3, 6 and 12 months, with additional changes in researching activity and problem solving (Sergievich et al. 2020). Rats exposed to the same mixture but at TDI values for 9 months, either with 100% or 25% RDI of essential vitamins, saw reduced locomotor activity resulting from chemical mixture exposure or vitamin deficiency, but not both (i.e., in vitamin deficient controls and vitamin sufficient treated). Locomotor activity and spatial orientation activity were reported to be increased in exposed animals that received insufficient vitamins, which interestingly also did not present increased anxiety that was seen in control animals that received insufficient vitamins (Tsatsakis et al. 2019b). Neurobehavioral effects have also been observed in mice gestationally exposed to a mixture of the flame retardant DecaBDE and lead (EDC, neurotoxic) via subcutaneous osmotic pumps, at doses far below (<0.017x) TDI and reference doses, respectively (Chen et al. 2019). In this study, chemical exposure was associated with increased repetitive stereotyped behaviours and impaired spatial learning ability and these behavioural effects were accompanied by increased circulating levels of many cytokines and a reduction in the number of hippocampal neuronal cells. Chen et al. (2019) also noted effects from exposure to DecaBDE or lead alone, especially lead which resulted in equally lowered hippocampal neuronal cell numbers as the mixture exposure group. While the effect size with regards to serum levels of pro-inflammation cytokines was greater when animals were exposed to the mixture, which suggests a synergistic action, the study design did not permit comparison of observations to mixture toxicity model predictions. These results may be of concern as the levels of lead were considerably lower than the accepted TDI. 3.2.6. Complex chemical mixtures Eleven research articles tested complex chemical mixtures with various toxicological mechanisms. These ten research articles reported findings from five separate experiments, none of which tested mixture chemicals individually. Exposure to 16 organochloride pesticides and 2 heavy metals, for 70 consecutive days, via oral gavage, at the lowest investigated dose level (equivalent to individual TDI, reference dose (RfD), maximum residue limit (MRL), or NOAEL values) resulted in increased epididymal sperm counts and greater natural killer cell activity from ex vivo splenocytes from exposed male rats (Wade et al. 2002a). This was accompanied by multiple signs of thyroid toxicity: increased thyroid stimulating hormone (TSH), larger median thyroid follicle areas, and reduced hepatic thyroxine outer-ring deiodinase activity (Wade et al. 2002b). Hepatic effects have also been reported in rats following exposure to a mixture of 27 chemicals (8 Heavy metals, 4 pollutants, 3 pesticides, 2 food-derived carcinogens, 2 plasticisers, 2 preservatives, 2 surfactants, 1 disinfectant, 1 food additive, 1 photostabilizer, and 1 polycyclic musk) at values lower than TDI (Hadrup et al. 2016). In this study, chemical ratios were based on comparisons to measured concentrations in human samples, and certain chemical classes were grouped into one of that class (e.g., PCBs were represented by PCB153 at a level equal to the sum of all PCBs). As a result, the comparability to individual PODs was reduced. The results indicated that chemical mixture exposure was associated with increased liver weight, macrovesicular changes, and vacuolisation. Hepatic lipid metabolomics also indicated separate and distinguishable metabolic profiles for chemical mixture exposed and control animals, however, direct quantification indicated that only one, unidentified, lipid was significantly changed in response to chemical mixture exposure (Hadrup et al. 2016). While these studies support an effect of low dose chemical mixtures on hepatic function, a study that used a Solt-Farber model of hepatocellular carcinoma demonstrated that a mixture of 12 pesticides of various chemical classes at doses equivalent to TDI had no effect on hepatic carcinogenesis (Perez-Carreon et al. 2009). Two studies investigated reproductive effects of gestational exposure to a complex mixture of chemicals with a diverse range of toxicological mechanisms. Gestational exposure of rats to 13 chemicals (3 plasticisers, 6 various pesticides, 2 pollutants, 1 preservative, and 1 analgesic) at doses equivalent to individual NOAEL values for AGD or NR, or where NOAEL values were not available LOAEL values divided by a conservative value, indicated specific effects on the reproductive system suggestive of at least additive, potentially synergistic, chemical action. Despite exposure to such low concentrations of mechanistically dissimilar chemicals, male offspring exhibited an increased number of nipples at birth, and greater NR at PND13 (Christiansen et al. 2012). Administration of a mixture of 18 chemicals (5 pesticides, 3 pharmaceuticals, 9 phthalates, and 1 pollutant) to rats during gestation resulted in male offspring with multiple signs of developmental/reproductive toxicity. Reduced testes, epididymis, and levator ani plus bulbocavernosus muscles (LABC) weights (PND21) were seen in offspring exposed to the mixture from ≤0.125x individual NOAEL, reduced AGD (PND2) and glans penis weights (PND21) from ≤0.25x NOAEL, and increased NR (PND13), seminal vesicle weights, and total malformation rates (PND21) at ≤0.5x and ≤1x NOAEL, (Conley et al. 2018). The remaining five research articles all reported findings from an 18-month rat study which exposed rats via drinking water to a mixture of 13 chemicals (4 preservatives, 4 insecticides, 1 herbicide, 1 fungicide, 1 plasticiser, 1 food additive, and 1 chelating agent) at doses equivalent to 0.25x, 1x and 5x TDI values. The reports detailed clinical observations and serum clinical chemistry at 6 and 12 months of exposure (Docea et al. 2018, 2019), behavioural tests (Tsatsakis et al. 2019c), oxidative stress findings in serum at 12 months of exposure and in organs at 18 months of exposure (Fountoucidou et al. 2019), and genotoxic, cytotoxic, and histopathological findings at 18 months of expoure (Tsatsakis et al. 2019a). Over the first 12 months, mixture treated animals had increased weight gain, yet reduced food and water consumption (Docea et al. 2018, 2019). By 12 months, animals that received the lower two doses of the chemical mixture exhibited increased exploratory behaviour (Tsatsakis et al. 2019c). ALT and ALP were increased in animals of each exposure groups at 6 and 12 months of exposure, which is indicative of liver damage (Docea et al. 2018, 2019). At 18 months of exposure deleterious histopathological findings were found in the liver and stomach of animals that received the chemical mixture at 1x and 5x TDI, and in testes, kidneys, lungs, and brains at all dose levels (Tsatsakis et al. 2019a). At 12 months of exposure, signs of oxidative stress or adaptive redox capacity were mixed within the chemical mixture exposed groups, but these animals had generally lower protein carbonyl levels than controls (Fountoucidou et al. 2019). By 18 months of exposure to the chemical mixture at 0.25x and 1x TDI most tissues showed lower protein carbonyls and thiobarbituric acid reactive substances and higher redox capacity, which suggests that compensatory mechanisms may have become operative in the exposed relative to control animals. However, at 5x TDI, protein and/or lipid oxidation biomarkers were generally increased relative to controls (Fountoucidou et al. 2019), indicating sufficient oxidative stress to overcome compensatory responses, and suggesting greater than additivity (potential synergy), as 13 chemicals at 5x TDI (approximately 0.05x NOAEL) would not be expected to elicit a response even by conservative DA modelling. 3.3. Whole-mixture methodologies Fourteen research articles used whole-mixtures methodologies, and these were all associated with one of two experimental models: the biosolid treated pasture (BTP) sheep model and a concentrated drinking water model. 3.3.1. BTP sheep model Eleven research articles used sheep reared on pastures fertilised using biosolids, a by-product of wastewater treatment. Due to the origins of biosolids they contain a diverse range of anthropogenic chemicals which encompass the human exposome (Rigby et al. 2020), which result in organ chemical loads of 0.5 – 200 μg/kg dry matter (Rhind et al. 2005, 2009, 2010; Bellingham et al. 2012; Filis et al. 2019). Due to a lack of toxicological studies in sheep, empirically determined PODs are not available. A series of studies have quantified the chemicals in biosolids, soil and herbage from BTP and blood and tissue samples from animals grazed on BTP. Chemical levels are small and not significantly different from control pastures; however, it must be noted that due to their ubiquitous nature many chemicals are also detectable in control pastures (Rhind et al. 2002, 2010, 2013; Evans et al. 2014). Chemical quantification and oral dosage estimations are limited to dioctyl phthalate, octyl phenol, and nonyl phenol, which were concluded to be below TDI (Rhind et al. 2002). As where residual chemical levels are monitored in crops and animals grazed on BTPs, they remain below human TDI values, the model was included in this review. Articles are presented further grouped as a function of age, i.e., effects in the fetus, juveniles, and adult offspring. 3.3.1.1. Effects in the fetus Seven research articles which utilised the BTP sheep model investigated the effects on fetuses. Gestational exposure to BTPs was shown to cause lower body weights in exposed male (Paul et al. 2005) and female (Fowler et al. 2008) fetuses at gestation day (GD) 110. These studies also documented multiple reproductive effects of exposure, in both sexes. Exposed male fetuses showed lower circulating testosterone concentrations, and had testis with reduced weights, fewer Sertoli cells, Leydig cells, and gonocytes, and less androgen receptor expression (Paul et al. 2005). Exposed female fetuses showed decreased oocyte numbers and altered oocyte type ratios. Proteomic analysis identified differentially expressed ovarian proteins related to core cellular processes (Fowler et al. 2008). Alterations of the hypothalamic-pituitary axis have also been reported in both sexes at GD110, with reduced kisspeptin gene expression and protein levels in the hypothalamus and pituitary and altered pituitary cellular composition (Bellingham et al. 2009) in lambs exposed to the chemical mixture. In two separate studies the timing of exposure to BTP and therefore gestational chemical exposure, prior to and/or during gestation, was investigated: prior to conception only (TC), gestation only (CT), and prior to conception and throughout gestation (TT), compared to controls (CC) where the mothers were never grazed on BTP. These studies reported that at GD110 foetal ovary weight was increased from TT exposure, yet effects on ovarian cell type counts and ratios were only seen in TC and CT (acute exposure) fetuses (Bellingham et al. 2013). All groups exposed to chemical mixtures, regardless of when exposure occurred, showed an increase in ovarian proteins involved in stress, oocyte maturation, and apoptosis, relative to the controls (CC). Ovarian proteomic pathway analysis identified enrichment in two functional networks: 1) cancer, gastrointestinal disease, and cellular movement, and 2) cancer, genetic disorder, and respiratory disease (Bellingham et al. 2013). The timing of exposure to the chemical mixture also alters the reported effects on the thyroid and the hypothalamic-pituitary axis (Hombach-Klonisch et al. 2013; Bellingham et al. 2016). The most severe thyroid effects of exposure to a mixture of chemicals were seen in CT and TC (acute exposure) groups, and males (Hombach-Klonisch et al. 2013). Thyroid weight was significantly increased, displayed reduced blood vessel area, reduced follicle numbers and areas, and increased cell proliferation in these acute BTP exposure groups. In the hypothalamic-pituitary axis the most notable observations from BTP exposure were altered expression of gonadotropin releasing hormone (GnRH) and its receptor (GnRHR), KISS1, the gene that encodes kisspeptin, and its receptor (KISS1R), oestrogen receptor, and androgen receptor. These expression changes were seen in most BTP exposed groups, for both sexes, although effects of exposure were not consistent between exposure group or sex (Bellingham et al. 2016). The timing of maternal exposure to BTPs in early, middle, late, or the entirety of gestation, has also been demonstrated to cause differential effects in female fetuses (GD140) (Lea et al. 2016), including body weight reductions, altered relative weights for the uterus, thyroid, and liver, increases in AGD, and changes to circulating testosterone, and free T3 and T4 concentrations (Lea et al. 2016). Despite differences between the effects of the timing of mixed chemical exposure, females in all the exposed groups had a lower proportion of healthy type 1a follicles relative to the controls, with a concordant increase in atretic type 1 and 1a follicles (Lea et al. 2016). Transcriptomic and proteomic analyses of ovaries identified many differentially expressed genes and proteins, but with very little overlap between groups. Pathway analysis of transcriptomic data identified enrichment within cellular growth and differentiation, cell cycle regulation, cell death, cellular development, and cell movement functions. Pathway analysis of proteomic data identified enrichment within free radical scavenging, cell-to-cell signalling and interaction, small molecule biochemistry, drug metabolism, and protein synthesis functions (Lea et al. 2016). 3.3.1.2. Effects in juveniles One paper that utilised the BTP sheep model investigated 5-month-old lambs. This study reported that following gestational and direct exposure to a chemical mixture through maternal and experimental subjects grazing on BTPs, lambs of both sexes had increased body weights at weaning and showed increased vocalisation and lower maximal activity levels while restrained compared to controls. Males that had been exposed to the chemical mixture also exhibited increased exploratory behaviours relative to the controls, suggestive of the ability of chemical mixtures in this model to affect cognitive ability (Erhard and Rhind 2004). 3.3.1.3. Effects in adults Three research articles that utilised the BTP sheep model investigated the effects in adult sheep of gestational exposure to a chemical mixture. Homeostasis of bone tissue was disrupted by exposure, with bone mineral content, thickness, circumference, cross-sectional area, and cavity size all affected (Lind et al. 2009). Interestingly, the adult studies that have used the BTP model also allow for the observation of effects of exposure to a chemical mixture following an extended period of non-exposure. Testicular morphology was altered in adult males (18 months old) following gestational exposure from conception, lactational exposure through weaning, and then direct exposure on BTPs until 7 months old. These rams exhibited a higher occurrence of Sertoli-cell-only seminiferous tubules, and lower numbers and volumes of germ cells (Bellingham et al. 2012). Interestingly these effects were not consistent in all animals, with only a subset of animals (5 of 12) showing a markedly altered phenotype, which may reflect the effects of mixed chemical exposure against the diverse genetic background in this outbred study population. Proteomic analysis of livers from these males and their female counterparts (maintained on treated pastures until 18 months) also identified differentially expressed proteins involved in detoxification and fatty-acid β-oxidation, as well as albumin and transferrin, in both sexes, with pathway analysis indicating dysregulation of cancer-related and lipid-related pathways (Filis et al. 2019). 3.3.2. Drinking water disinfection by-products A series of whole-mixture studies were conducted by the U.S. Environmental Protection Agency which examined the effects of drinking water disinfection by-products (DBPs) in pregnant rats. Only a small percentage of the >600 identified DBPs have been toxicologically evaluated, but of those that have, many are cytotoxic and genotoxic at concentrations achievable in the disinfection process. For example, epidemiological studies indicate trihalomethanes (THM4; chloroform, bromodichloromethane, dibromochloromethane, and bromoform) at concentrations >50 μg/L are associated with an increased risk of bladder cancer (Costet et al. 2011). Published research on DBPs have focussed on some of the chemicals currently regulated in the US; specifically total THM4, HAA5 (5 haloacetic acids: chloroacetic acid, bromoacetic acid, dichloroacetic acid, dibromoacetic acid, trichloroacetic acid), and bromate, at 80, 60, and 10 μg/L, respectively. It is of note that in the EU only THM4 and bromate are regulated on this regard, and that THM4 is allowed at levels up to 100 μg/L (Li and Mitch 2018; Andersson et al. 2019). Three research articles investigated the effects of DBP mixtures from various disinfection methods. The model used in these articles provides a mixture of DBPs at realistic ratios, at concentrations higher than maximum contaminant levels but lower than determined NOAELs for monitored chemicals. While not all the chemicals have determined PODs, these studies were included as regulatory assumed structure-activity relationships and chemical groupings for regulatory conditions can be applied. Narotsky et al. (2008) supplied water approximately 130x concentrated in DBPs, disinfected by either chlorination or ozonation, to pregnant rats, for 10 days during gestation (GD6 – 16) while controls received boiled, distilled, deionized water. While water consumption was increased and gestation lengths were reduced in dams supplied concentrated water, no adverse developmental effects were reported in pups. In a similar experiment conducted by Narotsky et al. (2012), two strains of rat (F344 and Sprague-Dawley) were exposed to water that had undergone chlorination and then concentration (chlor/conc), or concentration and then chlorination (conc/chlor), over gestation and lactation. Both concentration methods increasing DBPs by around 120x. Dams of both strains given chlor/conc water from early gestation to weaning experienced diarrhoea and polyuria. Sprague-Dawley dams that received concentrated drinking water also had lower body weights at GD20 - PND6 and F344 dams exhibited increased gestation lengths. There were fewer live Sprague-Dawley offspring, with greater perinatal loss/mortality, by PND6, and both strains had reduced bodyweights at PND6. The authors concluded, however, that some of the effects noted may be related to the concentration of inorganic materials such as sodium and sulphate in the concentrated water. However, when dams were administered conc/chlor water, which was similar in DBP composition but with greatly reduced sodium and sulphate levels to the chlor/conc water, only increased water consumption by dams was noted. Finally, using the chlor/conc method, Narotsky et al. (2013) performed a multi-generational study using water concentrated around 130x, with exposure starting at GD2 for the F1 generation and lasting until termination of the F2 generation. In this study, concentrated water consumption was associated with reduced caput sperm counts in adult F1 males, delayed puberty in juvenile F1 females, thyroid follicular hypertrophy in parental females as well as adult F1 females, and increased birthweights in F2 offspring. While these studies tested concentrated drinking water containing broadly similar concentrations and ratios of DBPs, there are also considerable differences between batches of concentrated drinking water used between studies, which may alter the toxic potential (Li and Mitch 2018). 4. Discussion A key finding of this review was that in studies where low dose chemical mixtures were tested in parallel with individual component chemicals, most studies reported that the response to mixture exposures were more numerous and/or greater in severity than the responses to individual chemicals. While it is not uncommon to see occasional mild signs of toxicity when individual chemicals are tested at, or close to, NOAEL levels, eight studies summarised in this review reported significant effects from individual chemicals at or below TDI values, theoretically two or three orders of magnitude away from an effective dose. The observation of physiological effects at these very low ‘safe’ doses calls into question the validity of the POD values used in the calculation of TDI values. Alternatively, these results could indicate significant variation between experiments/laboratories, genomic drift between animal breeders that purport to supply the “same” strain of animals, strain differences in susceptibility, or differences in detection sensitivity. However, all the above still expose potential shortcomings in toxicity assessment. Observations where toxicological effects were more numerous and/or more severe following exposure to a mixture of chemicals rather than the chemical components individually, suggests at least additivity between components of a chemical mixture. However, most studies did not employ experimental designs which allowed for direct comparisons to, and assessments of, mixture toxicity models, which can definitively distinguish additivity and synergy. Of the seven which provided data appropriate for such comparisons, three reported responses significantly greater than all investigated additivity model predictions, strongly indicating synergy. Although the remaining studies were not appropriate for similar direct critiques of mixture toxicity models, indications to the appropriateness of the mixture toxicity models can be inferred, for example, where there are effects following TDI exposure, which current applications of mixture toxicity models cannot explain. While this work has shown some commonality between various studies, there were also disparities between others. The focus of this review on the most relevant research for translational considerations (mammalian, in vivo) also posed the largest limitation, as data and designs between papers were too disparate from each other, and the quantity of literature too small, for true comparative re-evaluations. In a recent systematic review and quantitative reappraisal by Martin et al. (2021), which covered all living organisms, in vitro and in vivo, most mixtures were found to conform to dose additivity models. This agrees with many reviews which showed dose additivity-based models to be the most accurate across the whole dose-response curve, even when chemical components of a mixture are mechanistically distinct. A most notable example is the additive nature of antiandrogenic chemicals with phthalates, provided by Howdeshell et al. 2017, where additive models accurately predict exposure outcomes, although many of the studies involved did not meet inclusion criteria for this review. However, in twenty percent of literature identified by Martin et al. (2021) effects exceeded dose additivity models substantially, with synergistic interactions more than two-fold. These interactions were attributed to groupings already suspected of synergy (combinations of triazine, azole and pyrethroid pesticides), while also indicating new, potentially synergistic groupings (EDCs within metallic compounds). Signs of toxicity may be expected from co-exposure to mixtures containing chemicals at or close to NOAEL, regardless of additivity type. At this dose level an important consideration is the endpoint used to derive a NOAEL. Where NOAELs have been derived from endpoints dissimilar to those being investigated, study endpoints may have greater or lesser sensitivity to disruption. In these cases, compounds could be dosed either above or below NOAEL for study endpoints. In the latter case, greater additivity would be needed to cross the effect threshold, prohibitive of detecting effects. Similarly, experiments often used mixtures of too few components. As deviations from additivity are commonly small, simpler mixtures may not have the power to elicit an observable effect. Nearly half of the identified component-based literature used ≤5 mixture components. Additionally, some studies have used pilot data to generate dose-response curves for the specific endpoints being investigated, whereas others have used values derived in some form from regulatory studies or determinations. This could lead to contradictory findings; for example, vinclozolin NOAEL determined at 4 and 5 mg/kg/d by Schneider et al. (2017) and Christiansen et al. (2009) respectively. It is impossible to know if this difference could go towards explaining the differences between the studies (greater than additivity at NOAEL in Christiansen et al. (2009) and no effect at NOAEL in Schneider et al. (2017)), although this was not considered a major issue as PODs were broadly similar for the same chemicals across most studies. Twenty-one of the thirty studies which examined mixtures at NOAEL values reported physiological effects that were attributed to chemical exposure. However, no toxicological or physiological effect should be expected from co-exposure to chemical mixtures where the components are present at or close to TDI. This review identified that in eighteen of the twenty-one studies that tested mixtures at or below TDI values toxicological or physiological effects were reported. The greatest number of chemicals tested at doses equivalent to TDI values was twenty-seven chemicals. In this example, even using dose addition for all components, this level of exposure would still be anticipated to be more than three-fold lower than a dose expected to be able to elicit an effect. This is an indication of interactions between chemical components within the chemical mixtures, unaccounted for by mixture toxicity models. Of the literature identified which used a whole-mixture methodology, the BTP sheep model is the only which reflects actual human exposure as the drinking water by-products model is orders of magnitude away from realistic exposure. This lack of variation means that inter-species variance remains unaccounted for. While the BTP sheep model could be criticised for a lack of empirically determined PODs, limited quantification of individual chemicals, and no normalisation of dosages, it represents a real-world situation, with a chemical mixture used according to regulatory guidelines. Such use is deemed appropriate to ensure contaminant levels are below conservative calculations for acceptable human exposure, and thus also for other species for which there is a lack of empirically determined PODs. The BTP sheep model is also representative of human exposure in that many chemicals to which humans are exposed have little or no toxicological data in any species. This is a recognised problem which can not feasibly be solved by the traditional route of testing of each chemical individually but will most likely rely on new and future methodologies, including read-across, quantitative structure-activity relationship (QSAR) analysis, machine learning, and artificial intelligence (Aschner et al., 2022; NAS, 2017). The most examined endpoints in identified research articles were reproductive and/or developmental (nineteen), endocrine disruption (fifteen), hepatic (twelve), and behavioural (eight). There is concern that current guidelines do not sufficiently account for the multitude and ubiquity that characterises human exposure, especially fetal EDC exposure, which could be contributing to current global health problems, including the decline in male reproductive health (Skakkebæk et al. 2001; Skakkebæk 2002; Bergman et al. 2012). It has been suggested that current inclusion criteria for chemicals in mixture risk assessments, based on shared mechanisms of action, may be too restrictive in terms of mixture risk assessments for male reproductive health (Kortenkamp 2020). Of the nineteen research articles that reported effects on reproduction and/or development after in utero exposure to chemical mixtures, ten used mixtures with individual chemicals at TDI or ≤0.25x NOAEL values. Common responses observed were gonadal dysgenesis, deleterious germ cell alterations, and altered AGD, and specifically in males increased areola number, NR, and genital malformations. The relevance of systems biology approaches in a toxicological context without additional confirmation of biological effect has previously been questioned (Schneider et al. 2015). The use of omics data was seen in six component-based research articles, and four whole-mixture research articles. However, of the research articles which used omics technologies, four of the six component-based research articles, and three of the four whole-mixture research articles, also had strong supporting morphological data, therefore this was not considered as a factor for exclusion. 5. Conclusions The basis of compounding toxicity from chemical mixtures at low doses, especially at or below TDI values, remains a subject of debate. While there have been great advancements in mixture toxicity assessment, with some acceptance within regulatory bodies, there remains a lack of harmonisation as well as a lack of dose coverage to those far below individually determined NOAEL values. The present work represents a collation and analysis of research articles reporting experiments testing chemical mixtures with individual components at doses believed unable to elicit effects alone. This review, however, did not address other (controversial) aspects of low-dose chemical mixture exposure, such as hormetic, non-monotonic, or biphasic responses. While the extensive ECETOC literature review concluded no substantial evidence of mixture toxicity not already accounted for (ECETOC 2012), this review includes many studies which tested mixtures at TDI levels and were published after the ECETOC review. In addition, it should be noted that the ECETOC review focused on the identification of studies that provided evidence of effects greater than additivity model predictions, rather than evidence of effects unaccounted for by additivity model predictions. With this view, most of the literature identified here also falls short, as experimental designs which would have made this possible were typically not employed. However, of those which could, around half found additivity model predictions were inaccurate, and responses significantly greater than additivity model predictions were reported. Additionally, at doses around TDI, this fact is somewhat immaterial, especially when applied towards real-world situations, where co-exposure occurs to thousands of chemicals. It is in this respect that current methodological approaches for cumulative risk assessment fall short, and as such a novel paradigm has been suggested focussing on complex mixtures of chemicals at individual doses around TDI (Tsatsakis et al. 2016, 2017) and to assess risk using a real-life risk simulation (RLRS) approach (Hernández et al. 2020). However, it is logistically impossible to truly simulate real-life exposure by component-based methodologies. For this the BTP sheep model is most realistic, however it was not accepted by ECETOC due to the lack of empirically determined chemical concentrations and dose calculations, despite being common practice on fields growing crops for human consumption where chemical loads within crops remain below human TDIs. Additionally, due to the expansive chemical nature of biosolids and the limitations of current analytical techniques for such mass chemical quantification, such empirical determination is impractical. Finally, whole-mixture studies cannot give answers to the question of synergy, which can only be addressed through carefully designed component-based studies, but rather give snapshots of an extremely complex and dynamic exposure. Thus, there is little extra to be gained from precise quantification of individual chemicals and exact calculations of dosages resulting from BTP exposure without a greater understanding of biological and chemical interactions between mixture components. With these considerations in mind, it can be concluded that although no unequivocal evidence to refute conclusions from previous reviews, there is strong evidence to support additional research in this area. However, there were two data gaps identified which would facilitate a greater understanding of low dose chemical mixture toxicity, especially for complex chemical mixtures. Firstly, within the literature there was a lack of data on complex mixtures (with >10 components) at TDI values with data on individual components generated in parallel, likely because of greatly increased study requirements to generate such data. This data gap has been previously identified as impeding improvements to mixture risk assessments (Evans et al. 2015). Secondly, there is a lack of mixture dose-response data at very low-doses, with very few research articles reporting more than two concentrations less than or equal to NOAEL. This data gap deprives analyses of response curves and PODs for mixtures with which to compare to individual components, and thus leaves no basis to assert as evidence for or against current additivity models. Without these gaps being addressed, results cannot be interpreted regarding the nature of any combination effects, nor on the type of interactions or non-interactions occurring. Thus, the appropriateness of current mixture toxicity assessments cannot be critically examined further. Supplementary Material Summarising tables Acknowledgements The authors have no further acknowledgements to declare. The manuscript was prepared solely by the authors, whose contributions are declared herein. We would like to thank the anonymous reviewers for their insight and valuable input which has improved the manuscript to the current form. Funding This research received funding from the National Institute of Environmental Health Sciences (grant number R01 ES030374). Declaration of interest The authors have no conflicts of interest to declare. The author’s affiliations are as shown on the cover page. The motivation for the preparation of this manuscript was as part of the completion of doctoral training for CSE, which was paid for by the University of Glasgow, and utilised funding from NIEHS detailed herein. The authors have not participated in, nor anticipate participation in, any legal, regulatory, or advocacy proceedings related to the contents of the paper. The authors had sole responsibility for the preparation of this manuscript. Opinions and conclusions expressed within are those of the authors and are not necessarily those of any sponsoring entity. Abbreviations AGD Anogenital distances Ahr Arylhydrocarbon receptor ALP Alkaline phosphatase ALT Alanine transaminase AOP Adverse outcome pathway AST Aspartate transaminase BBP Benzyl butyl phthalate BMD Benchmark dose BPA Bisphenol A BTP Biosolid treated pasture Car Constitutive androstane receptor ChE Cholinesterase DA Dose addition DAR Draft assessment report DBP Dibutyl phthalate DecaBDE Decabromodiphenyl ether DEHP Diethylhexyl phthalate DEP Diethyl phthalate DMP Dimethyl phthalate DnOP Di-n-octyl phthalate EC Environmental chemical ECETOC European centre for ecotoxicology and toxicology of chemicals EDC Endocrine disrupting chemical EGF Epidermal growth factor ES Effect summation FSH Follicle-stimulating hormone GD Gestation day GnRH Gonadotropin releasing hormone GnRHR Gonadotropin releasing hormone receptor HFHS High-fat-high-sucrose HPT axis Hypothalamic–pituitary–thyroid axis IA Integrated addition KISS1 Kisspeptin 1 KISS1R Kisspeptin 1 receptor KO Knockout LOAEL Lowest observable adverse effect level MCPB 4-(4-chloro-2-methylphenoxy)butanoic acid MRL Maximum residue limit NGF Nerve growth factor NOAEL No observable adverse effect level NP Nonylphenol NR Nipple retention PAH Polycyclic aromatic hydrocarbon PCB Polychlorinated biphenyl PFOA Perfluorooctanoic acid PND Post-natal day PO Per os (by mouth) POD Point of departure Pparα Peroxisome proliferator-activated receptor alpha QSAR Quantitative structure-activity relationship RA Response addition RDI Recommended daily intake RfD Reference dose SC Subcutaneous TCDD Tetrachlorodibenzo-p-dioxin TDI Tolerable daily intake TEF Toxic equivalency factor TGFα Transforming growth factor alpha TSH Thyroid-stimulating hormone VOC Volatile organic compound WT Wild type Figure 1. Decision logic tree for the inclusion of studies. Bibliography Aschner M , Mesnage R , Docea AO , Paoliello MMB , Tsatsakis A , Giannakakis G , Papadakis GZ , Vinceti SR , Santamaria A , Skalny AV , Tinkov AA , 2022. Leveraging artificial intelligence to advance the understanding of chemical neurotoxicity. Neurotoxicology 89 , 9–11. 10.1016/j.neuro.2021.12.007 34968636 Andersson A , Ashiq MJ , Shoeb M , (2019) Evaluating gas chromatography with a halogen-specific detector for the determination of disinfection by-products in drinking water. Environ Sci Pollut Res 26 :7305–7314. 10.1007/s11356-018-1419-2 Bellingham M , Mckinnell C , Fowler PA , (2012) Foetal and post-natal exposure of sheep to sewage sludge chemicals disrupts sperm production in adulthood in a subset of animals. Int J Androl 35 :317–329. 10.1111/j.1365-2605.2011.01234.x 22150464 Bergman Å , Heindel JJ , Jobling S , (2012) State of the Science of Endocrine Disrupting Chemicals 2012. Geneva, Switzerland Bliss CI (1939) the Toxicity of Poisons Applied Jointly. Ann Appl Biol 26 :585–615. 10.1111/j.1744-7348.1939.tb06990.x Boas M , Feldt-Rasmussen U , Skakkebæk NE , Main KM (2006) Environmental chemicals and thyroid function. Eur J Endocrinol 154 :599–611. 10.1530/eje.1.02128 16645005 Boobis A , Budinsky R , Collie S , (2011) Critical analysis of literature on low-dose synergy for use in screening chemical mixtures for risk assessment. Crit Rev Toxicol 41 :369–383. 10.3109/10408444.2010.543655 21309635 Bopp SK , Barouki R , Brack W , (2018) Current EU research activities on combined exposure to multiple chemicals. Environ Int 120 :544–562. 10.1016/j.envint.2018.07.037 30170309 Bopp SK , Kienzler A , Richarz AN , (2019) Regulatory assessment and risk management of chemical mixtures: challenges and ways forward. Crit Rev Toxicol 49 :174–189. 10.1080/10408444.2019.1579169 30931677 Carpy SA , Kobel W , Doe J (2000) Health Risk of Low-Dose Pesticides Mixtures: A Review of the 1985–1998 Literature On Combination Toxicology and Health Risk Assessment. J Toxicol Environ Heal Part B 3 :1–25. 10.1080/109374000281122 Christiansen S , Scholze M , Dalgaard M , (2009) Synergistic disruption of external male sex organ development by a mixture of four antiandrogens. Environ Health Perspect 117 :1839–1846. 10.1289/ehp.0900689 20049201 Costet N , Villanueva CM , Jaakkola JJK , (2011) Water disinfection by-products and bladder cancer: Is there a European specificity? A pooled and meta-analysis of European caseecontrol studies. Occup Environ Med 68 :379–385. 10.1136/oem.2010.062703 21389011 Dorato MA , Engelhardt JA (2005) The no-observed-adverse-effect-level in drug safety evaluations: Use, issues, and definition(s). Regul Toxicol Pharmacol 42 :265–274. 10.1016/j.yrtph.2005.05.004 15979222 Drakvik E , Altenburger R , Aoki Y , (2020) Statement on advancing the assessment of chemical mixtures and their risks for human health and the environment. Environ Int 134 :. 10.1016/j.envint.2019.105267 Dunnick JK , Kissling G , Gerken DK , (2007) Cardiotoxicity of Ma Huang/caffeine or ephedrine/caffeine in a Rodent model system. Toxicol Pathol 35 :657–664. 10.1080/01926230701459978 17676524 ECETOC (2012) Effects of Chemical Co-exposures at Doses Relevant for Human Safety Assessments. Eur Cent Ecotoxicol Toxicol Chem EFSA Scientific Committee, More SJ , Hardy A , (2019) Guidance on harmonised methodologies for human health, animal health and ecological risk assessment of combined exposure to multiple chemicals. EFSA J 17 :. 10.2903/j.efsa.2019.5634 Evans NP , Bellingham M , Sharpe RM , (2014) Reproduction Symposium: Does grazing on biosolids-treated pasture pose a Pathophysiological risk associated with increased exposure to endocrine disrupting compounds? J Anim Sci 92 :3185–3198. 10.2527/jas.2014-7763 24948646 Evans RM , Scholze M , Kortenkamp A (2015) Examining the feasibility of mixture risk assessment: A case study using a tiered approach with data of 67 pesticides from the Joint FAO/WHO Meeting on Pesticide Residues (JMPR). Food Chem Toxicol 84 :260–269. 10.1016/j.fct.2015.08.015 26344759 Filis P , Walker N , Robertson L , (2019) Long-term exposure to chemicals in sewage sludge fertilizer alters liver lipid content in females and cancer marker expression in males. Environ Int 124 :98–108. 10.1016/j.envint.2019.01.003 30641261 Fraser TR (1872) On the antagonism between the actions of active substances. Br Med J 2 :457–459. 10.1136/bmj.2.617.457 Howdeshell KL , Hotchkiss AK , Earl Gray L Jr (2017) Cumulative effects of antiandrogenic chemical mixtures and their relevance to human health risk assessment. International Journal of Hygiene and Environmental Health 220 :179–188. 10.1016/j.ijheh.2016.11.007 27923611 Hernández AF , Docea AO , Goumenou M , Sarigiannis D , Aschner M , Tsatsakis A , (2020) Application of novel technologies and mechanistic data for risk assessment under the real-life risk simulation (RLRS) approach. Food Chem. Toxicol 137 :111123. 10.1016/j.fct.2020.111123 31926207 Hernández AF , Gil F , Lacasaña M (2017) Toxicological interactions of pesticide mixtures: an update. Arch Toxicol 91 :3211–3223. 10.1007/s00204-017-2043-5 28845507 Heys KA , Shore RF , Pereira MG , (2016) Risk assessment of environmental mixture effects. RSC Adv 6 :47844–47857. 10.1039/c6ra05406d Kortenkamp A (2020) Which chemicals should be grouped together for mixture risk assessments of male reproductive disorders? Mol Cell Endocrinol 499 :110581. 10.1016/j.mce.2019.110581 31525431 Kortenkamp A , Backhaus T , Faust M (2009) State of the Art Report on Mixture Toxicity - Final Report. ECHA 070307/2007/485103/ETU/D.1 Li X-F , Mitch WA (2018) Drinking Water Disinfection Byproducts (DBPs) and Human Health Effects: Multidisciplinary Challenges and Opportunities. Environ Sci Technol 52 :1681–1689. 10.1021/acs.est.7b05440 29283253 Loewe S , Muischnek H (1926) Über Kombinationswirkungen. Arch Exp Pathol und Pharmakologie 114 :313–326 Martin O , Scholze M , Ermler S , (2021) Ten years of research on synergisms and antagonisms in chemical mixtures: A systematic review and quantitative reappraisal of mixture studies. Environ Int 146 :106206. 10.1016/j.envint.2020.106206 33120228 NAS, 2017. Using 21st Century Science to Improve Risk-Related Evaluations, Using 21st Century Science to Improve Risk-Related Evaluations. National Academies Press, Washington, D.C. 10.17226/24635 OECD, 2018. Considerations for Assessing the Risks of Combined Exposure to Multiple Chemicals, Series on Testing and Assessment No. 296. Environ. Heal. Saf. Div. Environ. Dir Rappaport SM (2011) Implications of the exposome for exposure science. J Expo Sci Environ Epidemiol 21 :5–9. 10.1038/jes.2010.50 21081972 Rhind SM , Kyle CE , MacKie C , (2010) Maternal and fetal tissue accumulation of selected endocrine disrupting compounds (EDCs) following exposure to sewage sludge-treated pastures before or after conception. J Environ Monit 12 :1582–1593. 10.1039/c0em00009d 20676422 Rhind SM , Kyle CE , MacKie C , McDonald L (2009) Accumulation of endocrine disrupting compounds in sheep fetal and maternal liver tissue following exposure to pastures treated with sewage sludge. J Environ Monit 11 :1469–1476. 10.1039/b902085c 19657530 Rhind SM , Kyle CE , Telfer G , (2005) Alkyl phenols and diethylhexyl phthalate in tissues of sheep grazing pastures fertilized with sewage sludge or inorganic fertilizer. Environ Health Perspect 113 :447–453. 10.1289/ehp.7469 15811823 Rhind SM , Smith A , Kyle CE , (2002) Phthalate and alkyl phenol concentrations in soil following applications of inorganic fertiliser or sewage sludge to pasture and potential rates of ingestion by grazing ruminants. J Environ Monit 4 :142–148. 10.1039/b107539j 11871695 Rhind SMM , Kyle CEE , Ruffie H , (2013) Short- and long-term temporal changes in soil concentrations of selected endocrine disrupting compounds (EDCs) following single or multiple applications of sewage sludge to pastures. Environ Pollut 181 :262–270. 10.1016/j.envpol.2013.06.011 23896644 Ribeiro E , Ladeira C , Viegas S (2017) EDCs mixtures: A stealthy hazard for human health? Toxics 5 :1–17. 10.3390/toxics5010005 Rider CV , Dinse GE , Umbach DM , (2018) Predicting Mixture Toxicity with Models of Additivity. In: Chemical Mixtures and Combined Chemical and Nonchemical Stressors. Springer International Publishing, Cham, pp 235–270 Rider CV , LeBlanc GA (2005) An integrated addition and interaction model for assessing toxicity of chemical mixtures. Toxicol Sci 87 :520–528. 10.1093/toxsci/kfi247 16002478 Rigby H , Dowding A , Fernandes A , (2020) Concentrations of organic contaminants in industrial and municipal bioresources recycled in agriculture in the UK. Sci Total Environ 142787. 10.1016/j.scitotenv.2020.142787 Rotter S , Beronius A , Boobis AR , (2018) Overview on legislation and scientific approaches for risk assessment of combined exposure to multiple chemicals: the potential EuroMix contribution. Crit Rev Toxicol 48 :796–814. 10.1080/10408444.2018.1541964 30632445 Safe S (1990) Polychlorinated biphenyls (PCBs), dibenzo-p-dioxins (PCDDs), dibenzofurans (PCDFs), and related compounds: Environmental and mechanistic considerations which support the development of toxic equivalency factors (TEFs). Crit Rev Toxicol 21 :51–88. 10.3109/10408449009089873 2124811 Sarigiannis DA , Karakitsios SP (2018) Modeling Complex Exposures. In: Chemical Mixtures and Combined Chemical and Nonchemical Stressors. Springer International Publishing, Cham, pp 81–125 SCHER/SCCS/SCENIHR, 2013. Toxicity and Assessment of Chemical Mixtures. Eur. Comm 10.2772/21444 Schneider S , Fussell KC , Melching-Kollmuss S , (2015) Low Dose/Dose Response Relationship of Hormonally Active Substances and their Mixture – Testing Endocrine Disruptors in Classical and Molecular Endpoints at Human-Relevant Exposure Levels. Toxicol Lett 238 :S52. 10.1016/j.toxlet.2015.08.145 Schneider S , Fussell KC , Melching-Kollmuss S , (2017) Investigations on the dose–response relationship of combined exposure to low doses of three anti-androgens in Wistar rats. Arch Toxicol 91 :3961–3989. 10.1007/s00204-017-2053-3 28879601 Skakkebæk NE (2002) Endocrine Disrupters and Testicular Dysgenesis Syndrome. Horm Res Paediatr 57 :43–43. 10.1159/000058100 Skakkebæk NE , Rajpert-De Meyts E , Main KM (2001) Testicular dysgenesis syndrome: An increasingly common developmental disorder with environmental aspects. Hum Reprod 16 :972–978. 10.1093/humrep/16.5.972 11331648 Temkin CL (1941) Seven defences, the reply to certain calumniations of his enemies. In: Sigerist HE (ed) Four Treatises of Theophrastus Von Hohenheim Called Paracelsus. The Johns Hopkins Press Tsatsakis AM , Docea AO , Tsitsimpikou C (2016) New challenges in risk assessment of chemicals when simulating real exposure scenarios; simultaneous multi-chemicals’ low dose exposure. Food Chem Toxicol 96 :174–176. 10.1016/j.fct.2016.08.011 27515866 Tsatsakis AM , Kouretas D , Tzatzarakis MN , (2017) Simulating real-life exposures to uncover possible risks to human health: A proposed consensus for a novel methodological approach. Hum Exp Toxicol 36 :554–564. 10.1177/0960327116681652 28539089 Tsatsakis A , Petrakis D , Nikolouzakis TK , Docea AO , Calina D , Vinceti M , Goumenou M , Kostoff RN , Mamoulakis C , Aschner M , Hernández AF , (2020) COVID-19, an opportunity to reevaluate the correlation between long-term effects of anthropogenic pollutants on viral epidemic/pandemic events and prevalence. Food Chem. Toxicol 141 :111418. 10.1016/j.fct.2020.111418 32437891 U.S. EPA (1986) Guidelines for the Health Risk Assessment of Chemical Mixtures. Fed Regist 51 :34014–34025 U.S. EPA, 2007. Concepts, Methods, and Data Sources for Cumulative Health Risk Assessment of Multiple Chemicals, Exposures and Effects: A Resource Document (Final Report). U.S. Environ. Prot. Agency, Natl. Cent. Environ. Assessment, Cincinnati, OH EPA/600/R-06/013F U.S. HHS (2019) Fourth National Report on Human Exposure to Environmental Chemicals. Updated Tables, January 2019, Volume One. Fourth Natl Rep Hum Expo to Environ Chem 1–529 WHO, 2009. Assessment of Combined Exposures to Multiple Chemicals: Report of a WHO/IPCS International Workshop. IPCS Harmon. Proj. Doc. no. 7 Primary References Auger J , Le Denmat D , Berges R , (2013) Environmental levels of oestrogenic and antiandrogenic compounds feminize digit ratios in male rats and their unexposed male progeny. Proc R Soc B Biol Sci 280 :20131532. 10.1098/rspb.2013.1532 Auxietre T-A , Dumontier M-F , Balguy I , (2014) Sub-NOAEL amounts of vinclozolin and xenoestrogens target rat chondrogenesis in vivo. Biochimie 99 :169–177. 10.1016/j.biochi.2013.12.001 24333988 Bellingham M , Amezaga MR , Mandon-Pepin B , (2013) Exposure to chemical cocktails before or after conception - The effect of timing on ovarian development. Mol Cell Endocrinol 376 :156–172. 10.1016/j.mce.2013.06.016 23791816 Bellingham M , Fowler PA , Amezaga MR , (2009) Exposure to a Complex Cocktail of Environmental Endocrine-Disrupting Compounds Disturbs the Kisspeptin/GPR54 System in Ovine Hypothalamus and Pituitary Gland. Environ Health Perspect 117 :1556–1562. 10.1289/ehp.0900699 20019906 Bellingham M , Fowler PA , MacDonald ES , (2016) Timing of maternal exposure and foetal sex determine the effects of low-level chemical mixture exposure on the foetal neuroendocrine system in sheep. J Neuroendocrinol 28 :jne.12444. 10.1111/jne.12444 Bellingham M , Mckinnell C , Fowler PA , (2012) Foetal and post-natal exposure of sheep to sewage sludge chemicals disrupts sperm production in adulthood in a subset of animals. Int J Androl 35 :317–329. 10.1111/j.1365-2605.2011.01234.x 22150464 Buñay J , Larriba E , Patiño-Garcia D , (2018) Differential effects of exposure to single versus a mixture of endocrine-disrupting chemicals on steroidogenesis pathway in mouse testes. Toxicol Sci 161 :76–86. 10.1093/toxsci/kfx200 29029336 Chen Y , Liu S , Xu H , (2019) Maternal exposure to low dose BDE209 and Pb mixture induced neurobehavioral anomalies in C57BL/6 male offspring. Toxicology 418 :70–80. 10.1016/j.tox.2019.02.016 30836164 Christiansen S , Kortenkamp A , Axelstad M , (2012) Mixtures of endocrine disrupting contaminants modelled on human high end exposures: an exploratory study in rats. Int J Androl 35 :303–316. 10.1111/j.1365-2605.2011.01242.x 22372636 Christiansen S , Scholze M , Dalgaard M , (2009) Synergistic disruption of external male sex organ development by a mixture of four antiandrogens. Environ Health Perspect 117 :1839–1846. 10.1289/ehp.0900689 20049201 Conley JM , Lambright CS , Evans N , (2018) Mixed “antiandrogenic” chemicals at low individual doses produce reproductive tract malformations in the male rat. Toxicol Sci 164 :166–178. 10.1093/toxsci/kfy069 29945228 Demur C , Métais B , Canlet C , (2013) Dietary exposure to a low dose of pesticides alone or as a mixture: The biological metabolic fingerprint and impact on hematopoiesis. Toxicology 308 :74–87. 10.1016/j.tox.2013.03.004 23528616 Docea AO , Gofita E , Goumenou M , (2018) Six months exposure to a real life mixture of 13 chemicals’ below individual NOAELs induced non monotonic sex-dependent biochemical and redox status changes in rats. Food Chem Toxicol 115 :470–481. 10.1016/j.fct.2018.03.052 29621577 Docea AO , Goumenou M , Calina D , (2019) Adverse and hormetic effects in rats exposed for 12 months to low dose mixture of 13 chemicals: RLRS part III. Toxicol Lett 310 :70–91. 10.1016/j.toxlet.2019.04.005 30999039 Du L , Li S , Qi L , (2014) Metabonomic analysis of the joint toxic action of long-term low-level exposure to a mixture of four organophosphate pesticides in rat plasma. Mol BioSyst 10 :1153–1161. 10.1039/C4MB00044G 24626741 Erhard HW , Rhind SM (2004) Prenatal and postnatal exposure to environmental pollutants in sewage sludge alters emotional reactivity and exploratory behaviour in sheep. Sci Total Environ 332 :101–108. 10.1016/j.scitotenv.2004.04.012 15336895 Eustache F , Mondon F , Canivenc-Lavier MC , (2009) Chronic Dietary Exposure to a Low-Dose Mixture of Genistein and Vinclozolin Modifies the Reproductive Axis, Testis Transcriptome, and Fertility. Environ Health Perspect 117 :1272–1279. 10.1289/ehp.0800158 19672408 Filis P , Walker N , Robertson L , (2019) Long-term exposure to chemicals in sewage sludge fertilizer alters liver lipid content in females and cancer marker expression in males. Environ Int 124 :98–108. 10.1016/j.envint.2019.01.003 30641261 Fountoucidou P , Veskoukis AS , Kerasioti E , (2019) A mixture of routinely encountered xenobiotics induces both redox adaptations and perturbations in blood and tissues of rats after a long-term low-dose exposure regimen: The time and dose issue. Toxicol Lett 317 :24–44. 10.1016/j.toxlet.2019.09.015 31541690 Fowler PA , Dorà NJ , McFerran H , (2008) In utero exposure to low doses of environmental pollutants disrupts fetal ovarian development in sheep. Mol Hum Reprod 14 :269–280. 10.1093/molehr/gan020 18436539 Gao HT , Xu R , Cao WX , (2017) Effects of six priority controlled phthalate esters with long-term low-dose integrated exposure on male reproductive toxicity in rats. Food Chem Toxicol 101 :94–104. 10.1016/j.fct.2017.01.011 28089693 Hadrup N , Svingen T , Mandrup K , (2016) Juvenile male rats exposed to a low-dose mixture of twenty-seven environmental chemicals display adverse health effects. PLoS One 11 :1–22. 10.1371/journal.pone.0162027 Hass U , Boberg J , Christiansen S , (2012) Adverse effects on sexual development in rat offspring after low dose exposure to a mixture of endocrine disrupting pesticides. Reprod Toxicol 34 :261–274. 10.1016/j.reprotox.2012.05.090 22659286 Hass U , Christiansen S , Axelstad M , (2017) Combined exposure to low doses of pesticides causes decreased birth weights in rats. Reprod Toxicol 72 :97–105. 10.1016/j.reprotox.2017.05.004 28526456 Hombach-Klonisch S , Danescu A , Begum F , (2013) Peri-conceptional changes in maternal exposure to sewage sludge chemicals disturbs fetal thyroid gland development in sheep. Mol Cell Endocrinol 367 :98–108. 10.1016/j.mce.2012.12.022 23291342 Jacobsen H , Østergaard G , Lam H ., (2004) Repeated dose 28-day oral toxicity study in Wistar rats with a mixture of five pesticides often found as residues in food: alphacypermethrin, bromopropylate, carbendazim, chlorpyrifos and mancozeb. Food Chem Toxicol 42 :1269–1277. 10.1016/j.fct.2004.03.003 15207377 Jacobsen PR , Axelstad M , Boberg J , (2012) Persistent developmental toxicity in rat offspring after low dose exposure to a mixture of endocrine disrupting pesticides. Reprod Toxicol 34 :237–250. 10.1016/j.reprotox.2012.05.099 22677472 Jacobsen PR , Christiansen S , Boberg J , (2010) Combined exposure to endocrine disrupting pesticides impairs parturition, causes pup mortality and affects sexual differentiation in rats. Int J Androl 33 :434–442. 10.1111/j.1365-2605.2009.01046.x 20487043 Julien B , Pinteur C , Vega N , (2018) Evidence for estrogeno-mimetic effects of a mixture of low-dose pollutants in a model of ovariectomized mice. Environ Toxicol Pharmacol 57 :34–40. 10.1016/j.etap.2017.11.008 29175711 Julien B , Pinteur C , Vega N , (2019) Estrogen withdrawal and replacement differentially target liver and adipose tissues in female mice fed a high-fat high-sucrose diet: impact of a chronic exposure to a low-dose pollutant mixture☆. J Nutr Biochem 72 :108211. 10.1016/j.jnutbio.2019.07.002 31473509 Kouidhi W , Desmetz C , Nahdi A , (2012) In Utero and Lactational Exposure to Low-Dose Genistein-Vinclozolin Mixture Affects the Development and Growth Factor mRNA Expression of the Submandibular Salivary Gland in Immature Female Rats. Toxicol Pathol 40 :593–604. 10.1177/0192623311436183 22317923 Labaronne E , Pinteur C , Vega N , (2017) Low-dose pollutant mixture triggers metabolic disturbances in female mice leading to common and specific features as compared to a high-fat diet. J Nutr Biochem 45 :83–93. 10.1016/j.jnutbio.2017.04.001 28433925 Lea RG , Amezaga MR , Loup B , (2016) The fetal ovary exhibits temporal sensitivity to a ‘real-life’ mixture of environmental chemicals. Sci Rep 6 :22279. 10.1038/srep22279 26931299 Lehraiki A , Messiaen S , Berges R , (2011) Antagonistic effects of gestational dietary exposure to low-dose vinclozolin and genistein on rat fetal germ cell development. Reprod Toxicol 31 :424–430. 10.1016/j.reprotox.2010.12.005 21172421 Lind PM , Gustafsson M , Hermsen SAB , (2009) Exposure to pastures fertilised with sewage sludge disrupts bone tissue homeostasis in sheep. Sci Total Environ 407 :2200–2208. 10.1016/j.scitotenv.2008.12.035 19162300 Lukowicz C , Ellero-Simatos S , Régnier M , (2018) Metabolic Effects of a Chronic Dietary Exposure to a Low-Dose Pesticide Cocktail in Mice: Sexual Dimorphism and Role of the Constitutive Androstane Receptor. Environ Health Perspect 126 :067007. 10.1289/EHP2877 29950287 Merhi M , Demur C , Racaud-Sultan C , (2010) Gender-linked haematopoietic and metabolic disturbances induced by a pesticide mixture administered at low dose to mice. Toxicology 267 :80–90. 10.1016/j.tox.2009.10.024 19883720 Mossa A-TH , Refaie AA , Ramadan A (2011) Effect of Exposure to Mixture of Four Organophosphate Insecticides at No Observed Adverse Effect Level Dose on Rat Liver: The Protective Role of Vitamin C. Res J Environ Toxicol 5 :323–335. 10.3923/rjet.2011.323.335 Narotsky MG , Best DS , Rogers EH , (2008) Integrated disinfection by-products mixtures research: assessment of developmental toxicity in Sprague-Dawley rats exposed to concentrates of water disinfected by chlorination and ozonation/postchlorination. J Toxicol Environ Health A 71 :1216–21. 10.1080/15287390802182623 18636393 Narotsky MG , Klinefelter GR , Goldman JM , (2015) Reproductive toxicity of a mixture of regulated drinking-water disinfection by-products in a multigenerational rat bioassay. Environ Health Perspect 123 :564–570. 10.1289/ehp.1408579 25695961 Narotsky MG , Klinefelter GR , Goldman JM , (2013) Comprehensive assessment of a chlorinated drinking water concentrate in a rat multigenerational reproductive toxicity study. Environ Sci Technol 47 :10653–10659. 10.1021/es402646c 23909560 Narotsky MG , Pressman JG , Miltner RJ , (2012) Developmental Toxicity Evaluations of Whole Mixtures of Disinfection By-products using Concentrated Drinking Water in Rats: Gestational and Lactational Effects of Sulfate and Sodium. Birth Defects Res Part B - Dev Reprod Toxicol 95 :202–212. 10.1002/bdrb.21004 Naville D , Gaillard G , Julien B , (2019) Chronic exposure to a pollutant mixture at low doses led to tissue-specific metabolic alterations in male mice fed standard and high-fat high-sucrose diet. Chemosphere 220 :1187–1199. 10.1016/j.chemosphere.2018.12.177 30722647 Naville D , Labaronne E , Vega N , (2015) Metabolic Outcome of Female Mice Exposed to a Mixture of Low-Dose Pollutants in a Diet-Induced Obesity Model. PLoS One 10 :e0124015. 10.1371/journal.pone.0124015 25909471 Naville D , Pinteur C , Vega N , (2013) Low‐dose food contaminants trigger sex‐specific, hepatic metabolic changes in the progeny of obese mice. FASEB J 27 :3860–3870. 10.1096/fj.13-231670 23756648 Ortega MEM , Romero DM , Pato AM , (2018) Relationship between exposure, body burden and target tissue concentration after oral administration of a low-dose mixture of pyrethroid insecticides in young adult rats. Toxicology 409 :53–62. 10.1016/j.tox.2018.07.006 30009846 Pascotto VM , Guerra MT , Franci JAA , (2015) Effects of a Mixture of Pesticides on the Adult Female Reproductive System of Sprague-Dawley, Wistar, and Lewis Rats. J Toxicol Environ Heal Part A 78 :602–616. 10.1080/15287394.2015.1010467 Paul C , Rhind SM , Kyle CE , (2005) Cellular and hormonal disruption of fetal testis development in sheep reared on pasture treated with sewage sludge. Environ Health Perspect 113 :1580–1587. 10.1289/ehp.8028 16263515 Perez-Carreon JI , Dargent C , Merhi M , (2009) Tumor promoting and co-carcinogenic effects in medium-term rat hepatocarcinogenesis are not modified by co-administration of 12 pesticides in mixture at acceptable daily intake. Food Chem Toxicol 47 :540–546. 10.1016/j.fct.2008.12.008 19126423 Perobelli JE , Martinez MF , da Silva Franchi CA , (2010) Decreased Sperm Motility in Rats Orally Exposed to Single or Mixed Pesticides. J Toxicol Environ Heal Part A 73 :991–1002. 10.1080/15287391003751802 Rider CV , Furr J , Wilson VS , Gray LE (2008) A mixture of seven antiandrogens induces reproductive malformations in rats. Int J Androl 31 :249–262. 10.1111/j.1365-2605.2007.00859.x 18205796 Rider CV , Furr JR , Wilson VS , Gray LE Jr (2010) Cumulative effects of in utero administration of mixtures of reproductive toxicants that disrupt common target tissues via diverse mechanisms of toxicity. Int J Androl 33 :443–462. 10.1111/j.1365-2605.2009.01049.x 20487044 Rieke S , Heise T , Schmidt F , (2017) Mixture effects of azole fungicides on the adrenal gland in a broad dose range. Toxicology 385 :28–37. 10.1016/j.tox.2017.04.012 28454766 Saad HES , Meduri G , Phrakonkham P , (2011) Abnormal peripubertal development of the rat mammary gland following exposure in utero and during lactation to a mixture of genistein and the food contaminant vinclozolin. Reprod Toxicol 32 :15–25. 10.1016/j.reprotox.2011.03.001 21539910 Santiasih I , Titah HS , Hermana J (2020) The effects of particulate matters inhalation exposures of prallethrin and d-phenothrin mixture in mice (Mus musculus) against exhaled carbon dioxide concentration. Toxicol Res 36 :59–67. 10.1007/s43188-019-00002-8 31998626 Schmidt F , Marx-Stoelting P , Haider W , (2016) Combination effects of azole fungicides in male rats in a broad dose range. Toxicology 355–356 :54–63. 10.1016/j.tox.2016.05.018 Schneider S , Fussell KC , Melching-Kollmuss S , (2017) Investigations on the dose–response relationship of combined exposure to low doses of three anti-androgens in Wistar rats. Arch Toxicol 91 :3961–3989. 10.1007/s00204-017-2053-3 28879601 Sergievich AA , Khoroshikh PP , Artemenko AF , (2020) Behavioral impacts of a mixture of six pesticides on rats. Sci Total Environ 727 :138491. 10.1016/j.scitotenv.2020.138491 32335449 Sobolewski M , Conrad K , Allen JL , (2014) Sex-specific enhanced behavioral toxicity induced by maternal exposure to a mixture of low dose endocrine-disrupting chemicals. Neurotoxicology 45 :121–130. 10.1016/j.neuro.2014.09.008 25454719 Svingen T , Ramhøj L , Mandrup K , (2018) Effects on metabolic parameters in young rats born with low birth weight after exposure to a mixture of pesticides. Sci Rep 8 :305. 10.1038/s41598-017-18626-x 29321614 Tanida T , Warita K , Ishihara K , (2009) Fetal and neonatal exposure to three typical environmental chemicals with different mechanisms of action: Mixed exposure to phenol, phthalate, and dioxin cancels the effects of sole exposure on mouse midbrain dopaminergic nuclei. Toxicol Lett 189 :40–47. 10.1016/j.toxlet.2009.04.005 19481886 Tsatsakis A , Docea AO , Constantin C , (2019a) Genotoxic, cytotoxic, and cytopathological effects in rats exposed for 18 months to a mixture of 13 chemicals in doses below NOAEL levels. Toxicol Lett 316 :154–170. 10.1016/j.toxlet.2019.09.004 31521832 Tsatsakis A , Tyshko NV ., Docea AO , (2019b) The effect of chronic vitamin deficiency and long term very low dose exposure to 6 pesticides mixture on neurological outcomes – A real-life risk simulation approach. Toxicol Lett 315 :96–106. 10.1016/j.toxlet.2019.07.026 31386889 Tsatsakis AM , Docea AO , Calina D , (2019c) Hormetic Neurobehavioral effects of low dose toxic chemical mixtures in real-life risk simulation (RLRS) in rats. Food Chem Toxicol 125 :141–149. 10.1016/j.fct.2018.12.043 30594548 Wade MG , Foster WG , Younglai EV ., (2002a) Effects of Subchronic Exposure to a Complex Mixture of Persistent Contaminants in Male Rats: Systemic, Immune, and Reproductive Effects. Toxicol Sci 67 :131–143. 10.1093/toxsci/67.1.131 11961226 Wade MG , Parent S , Finnson KW , (2002b) Thyroid Toxicity Due to Subchronic Exposure to a Complex Mixture of 16 Organochlorines, Lead, and Cadmium. Toxicol Sci 67 :207–218. 10.1093/toxsci/67.2.207 12011480 Wolansky MJ , Gennings C , DeVito MJ , Crofton KM (2009) Evidence for dose-additive effects of pyrethroids on motor activity in rats. Environ Health Perspect 117 :1563–1570. 10.1289/ehp.0900667 20019907 Yang J , Cao J , Sun X , (2012) Effects of long-term exposure to low levels of organophosphorous pesticides and their mixture on altered antioxidative defense mechanisms and lipid peroxidation in rat liver. Cell Biochem Funct 30 :122–128. 10.1002/cbf.1825 22034174

PMC009xxxxxx/PMC9549741.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 2985134R 2913 Chem Rev Chem Rev Chemical reviews 0009-2665 1520-6890 35849738 9549741 10.1021/acs.chemrev.1c00914 NIHMS1838846 Article Second and Outer Coordination Sphere Effects in Nitrogenase, Hydrogenase, Formate Dehydrogenase, and CO Dehydrogenase Stripp Sven T. http://orcid.org/0000-0002-8412-0258 Freie Universität Berlin, Experimental Molecular Biophysics, Berlin 14195, Germany Duffus Benjamin R. http://orcid.org/0000-0002-0343-4328 University of Potsdam, Molecular Enzymology, Potsdam 14476, Germany Fourmond Vincent http://orcid.org/0000-0001-9837-6214 Laboratoire de Bioénergétique et Ingénierie des Protéines, Institut de Microbiologie de la Méditerranée, Institut Microbiologie, Bioénergies et Biotechnologie, CNRS, Aix Marseille Université, Marseille 13402, France Léger Christophe http://orcid.org/0000-0002-8871-6059 Laboratoire de Bioénergétique et Ingénierie des Protéines, Institut de Microbiologie de la Méditerranée, Institut Microbiologie, Bioénergies et Biotechnologie, CNRS, Aix Marseille Université, Marseille 13402, France Leimkühler Silke http://orcid.org/0000-0003-3238-2122 University of Potsdam, Molecular Enzymology, Potsdam 14476, Germany Hirota Shun http://orcid.org/0000-0003-3227-8376 Nara Institute of Science and Technology, Division of Materials Science, Graduate School of Science and Technology, Nara 630-0192, Japan Hu Yilin http://orcid.org/0000-0002-9088-2865 Department of Molecular Biology & Biochemistry, University of California, Irvine, California 92697-3900, United States Jasniewski Andrew http://orcid.org/0000-0001-7614-0796 Department of Molecular Biology & Biochemistry, University of California, Irvine, California 92697-3900, United States Ogata Hideaki http://orcid.org/0000-0002-2894-2417 Nara Institute of Science and Technology, Division of Materials Science, Graduate School of Science and Technology, Nara 630-0192, Japan; Hokkaido University, Institute of Low Temperature Science, Sapporo 060-0819, Japan; Graduate School of Science, University of Hyogo, Hyogo 678-1297, Japan Ribbe Markus W. http://orcid.org/0000-0002-7366-1526 Department of Molecular Biology & Biochemistry, University of California, Irvine, California 92697-3900, United States; Department of Chemistry, University of California, Irvine, California 92697-2025, United States Corresponding Author: Sven T. Stripp – Freie Universität Berlin, Experimental Molecular Biophysics, Berlin 14195, Germany; sven.stripp@gmail.com 27 9 2022 27 7 2022 18 7 2022 27 7 2023 122 14 1190011973 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Gases like H2, N2, CO2, and CO are increasingly recognized as critical feedstock in “green” energy conversion and as sources of nitrogen and carbon for the agricultural and chemical sectors. However, the industrial transformation of N2, CO2, and CO and the production of H2 require significant energy input, rendering routines like steam reformation or the Haber process economic and environmental dead ends. Nature, on the other hand, performs similar tasks efficiently at ambient temperature and pressure, exploiting gas-processing metalloenzymes (GPMs) that bind low-valent metal cofactors based on iron, nickel, molybdenum, tungsten, and sulfur. Such systems are studied to understand the biocatalytic principles of gas conversion including N2 fixation by nitrogenase and H2 production by hydrogenase as well as CO2 and CO conversion by formate dehydrogenase, carbon monoxide dehydrogenase, and nitrogenase. In this review, we emphasize the importance of the cofactor/protein interface, discussing how second and outer coordination sphere effects determine, modulate, and optimize the catalytic activity of GPMs. These may comprise ionic interactions in the second coordination sphere that shape the electron density distribution across the cofactor, hydrogen bonding changes, and allosteric effects. In the outer coordination sphere, proton transfer and electron transfer are discussed, alongside the role of hydrophobic substrate channels and protein structural changes. Combining the information gained from structural biology, enzyme kinetics, and various spectroscopic techniques, we aim toward a comprehensive understanding of catalysis beyond the first coordination sphere. Graphical Abstract pmc1. INTRODUCTION Enzymes enhance the probability of a chemical reaction to proceed on biologically relevant time scales. Kinetic rates may differ by a factor of up to 106, from the exclusively diffusionlimited enzymes carbonic anhydrase, catalase, or superoxide dismutase to average performers like RuBisCO that produces only three equivalents of product per second. To catalyze a given reaction, the enzyme must provide a specific binding pocket for reactants, for example, offering electrostatic interactions with polar and charged amino acid side chains, the protein backbone, and water molecules. Additionally, hydrophobic interactions play an important role (e.g., in the binding pockets of ATP synthase the separation of reactants and water drives ATP formation). Many enzymes rely on essential cofactors that are not formed by the protein fold. These may be organic “coenzymes” (covalently bound prosthetic groups and loosely bound “cosubstrates”) or inorganic clusters of metal ions. The later defines the class of metalloenzymes.1–3 Metalloenzymes catalyze important energy conversion reactions throughout all kingdoms of life. The fast-performing enzymes carbonic anhydrase, catalase, and superoxide dismutase are metalloenzymes as well. Introducing two prominent examples, photosystem II (PSII, oxygenic photosynthesis) and cytochrome c oxidase (CcO, aerobic respiration) are high-valent, gas-processing metalloenzymes (GPMs). At the “oxygen evolving complex” of PSII, a high-potential heterometallic manganese cluster (MnIII−MnV, E° ≈ 1.0 V vs SHE) catalyzes water splitting,4–6 whereas CcO catalyzes the reverse reaction at a binuclear heme-copper center (FeII–FeIV, E° ≈ 0.8 V vs SHE).7–9 In contrast, low-valent GPMs operate under reducing conditions, exploiting soft, electron-rich metal ions in the catalytic activation of N2, H2, CO2, and CO (E° ≈ −0.4 V vs SHE).10 Such systems carry cofactors reminiscent of abiotic metal clusters like pyrite, mackinawite, and pentlandite. These minerals show residual activity catalyzing CO2, N2, and proton reduction,11–13 but efficient turnover can only be achieved as “molecularly tuned derivates”14 within the active sites of GPMs like nitrogenase, hydrogenase, or CO dehydrogenase. From an evolutionary perspective, it is debated whether mineral cofactors were a prerequisite for catalysis or optimized the activity of protoenzymes. Russell and co-workers argue that metalloenzymes seem to be older than all-organic enzymes,15 suggesting atmospheric CO2 and hydrothermal H2 and CH4 as earliest sources of electrons and carbon.16 Similarly, Martin and co-workers argue that life could have evolved from gases that reacted with the help of transition metals.17 Embedding metal ions or mineral cofactors, simple peptide “nests” may represent the onset of metalloenzyme evolution18–20 that produced the GPMs discussed in this review including nitrogenase (Section 2), hydrogenase (Section 3), formate dehydrogenase, and carbon monoxide dehydrogenase (FDH and CODH, Section 4 and Section 5). All of these enzymes rely on iron, either as part of the active site cofactor (hydrogenase and nitrogenase) or in iron–sulfur clusters that primarily serve in long-range electron transfer.21 Further metal ions include nickel (CODH and [NiFe]-hydrogenase) and molybdenum (FDH and nitrogenase). In certain isoenzymes, molybdenum is replaced with tungsten (FDH), vanadium, or iron (nitrogenase). Such systems are studied to understand the biocatalytic principles of gas conversion that may inspire the design of biomimetic catalysts for the activation of N2, CO2, and CO as well as the production of H2 as a climate-neutral fuel. Second and outer coordination sphere effects critically impact the catalytic performance of GPMs. Although the influences are manifold and difficult to predict a priori, comparing the reaction stoichiometries (eqs 1–4) certain similarities become evident: (1) N2+8H++8e−→2NH3+H2 (nitrogenase) (2) H2⇌2H++2e− (hydrogenase) (3) CO2+H++2e−⇌HCOO− (FDH) (4) CO+H2O⇌CO2+2H++2e− (CODH) Each enzyme catalyzes a redox reaction that involves one proton per electron, which hints at proton-coupled electron transfer (PCET)22–24 as a common principle in GPMs. This demands (i) electron transfer to or from a redox site or redox partner in the outer coordination sphere of the metal ion coupled to (ii) proton transfer via polar amino acid residues or functional water molecules in the second coordination sphere (Figure 1). Proton transfer pathways and (iii) hydrophobic gas channels further modulate the exchange of reactants between active site and bulk solution, sometimes across several nanometers.25 [FeFe]-hydrogenase, shown in Figure 1, illustrates how individual second coordination sphere effects may comprise (iv) hydrophobic interactions and (v) hydrogen-bonding contacts.26 Moreover, the protonation state and polarity of the active site niche influences the (vi) charge distribution across the cofactor and redox centers in channeling distance.27 Long-distance effects play a crucial role in enzyme catalysis. Within each enzyme family, defined by a common active site structure, homologous enzymes may have very different catalytic properties in terms of turnover rates, catalytic bias (defined here as the ratio of the maximal rates in both directions28), reversibility (the requirement for a thermodynamic driving force to trigger catalysis29), and resistance to “stress” (e.g., sensitivity to O2 or visible light). The immediate environment of the active sites in each family is very much conserved, but these enzymes differ by their protein sequences, cofactor composition, and quaternary structures. Nature therefore provides a quasi-infinite playground for examining outer coordination sphere effects in catalysis. For example, certain microorganisms produce many homologous enzymes to catalyze the same reaction (isoenzymes), which suggests that distinct catalytic properties are needed for different physiological functions. In this review, we highlight and compare the role of second and outer coordination sphere effects on the activation of gaseous substrates such N2, H2, CO2, or CO by various low-valent GPMs. We provide a comprehensive overview on protein variants that were found to affect the catalytic properties of nitrogenase, hydrogenase, FDH, and CODH. To understand or mimic GPMs, we emphasize that second and outer coordination sphere effects must not be neglected. 2. NITROGENASE Nitrogen is an essential element for life, necessary for the synthesis of biomolecules such as deoxyribonucleic acid (DNA), ribonucleic acid (RNA), and amino acids. While nitrogen can be found in many different chemical forms throughout atmospheric and geological sources, the largest reservoir of nitrogen on the surface of the Earth is found in the form of dinitrogen (N2): a small, gaseous, nonpolar, diatomic molecule with a strong triple bond.30 The strength of this bond is on the order of ~940 kJ/mol, which makes its scission into bioavailable forms, also called “N2 fixation”, one of the most challenging transformations in Nature. Two main processes are recognized as natural engines for N2 fixation within the global nitrogen cycle: lighting strikes and biological N2 fixation (BNF).31–33 The high temperature environment in the air surrounding lightning causes N2 to react with other components in the atmosphere, such as dioxygen (O2), ozone (O3), or oxygen-based radical species, forming various nitrogen oxides (NOx).31,34,35 The NOx species are then converted to nitric acid (HNO3) in the atmosphere, and brought down to the ground through rainfall or surface deposition, where the fixed nitrogen species can be utilized.30,36 This lightning-based process requires extreme conditions such as high temperature, and is completely unregulated which causes many chemical pathways to converge and ultimately produce the highly oxidized HNO3. In contrast, biological N2 fixation occurs in diazotrophic microorganisms that span across multiple phyla between bacteria and archaea.37–39 These organisms are found in many environments, such as surface soils, sediments, marine and fresh waters, and geothermal sources, in both the presence or absence of oxygen.40–43 Some organisms can live freely within microbial communities, such as the obligate aerobe Azotobacter vinelandii,44 or the thermophilic methanogen Methanococcus thermolithotrophicus,45,46 while others, such as Rhizobium leguminosarum, form symbiotic relationships with specific plant species.47,48 In all cases, N2 is converted into the highly reduced nitrogen species ammonia (NH3), although more often ammonium (NH4+) is the observed product.30 Despite the wide variation in living conditions, each of the diazotrophs have genes that encode for the common enzyme nitrogenase. Nitrogenase is a gas-processing metalloenzyme composed of two proteins, the reductase component NifH (also called the “Fe protein”), as well as the catalytic component NifDK (also called the “MoFe protein”) that contains a unique molybdenum-containing iron sulfur cluster called the M-cluster (or “FeMoco”).49,50 There are also so-called “alternative” nitrogenases that replace the Mo ion in FeMoco for either vanadium (“FeVco” in the VFe protein) or iron (“FeFeco” in the FeFe protein).51 All nitrogenases are capable of reducing small molecule substrates, such as acetylene (C2H2) and cyanide (CN−), under ambient temperature and pressure, but the native reaction catalyzed by the enzyme is the reduction of N2 to NH3.51–53 In comparison, humans have developed the Haber-Bosch process that takes N2 and dihydrogen (H2) under high temperature and intense pressure conditions to yield NH3.54 Nitrogenase also catalyzes the reduction of protons (H+) to H2, and in all reactions with other substrates (including N2), some portion of the electron equivalents (or electron flux) is diverted to produce H2.50,55 During the native nitrogenase reaction, H2 is generated as an additional product (eq 1), but the mechanism of its synthesis is distinct from simple proton reduction.52,56–60 The complexity of nitrogenase has captivated researchers for decades as they seek out answers to the question, how does nitrogenase carry out these difficult chemical transformations? Many scientific approaches have been utilized in its study, each providing complementary information that can add to the larger picture. Since nitrogenase has been extensively reviewed,50–52,61–63 this section shall primarily focus on secondary and outer sphere effects that modulate the function of the Mo-dependent enzyme. 2.1. Structural Features of Nitrogenase The Mo-nitrogenase from A. vinelandii is the most extensively studied nitrogenase protein, and under N2 fixing conditions up to 10% of all cell proteins are nitrogenase related.44,64 The MoFe protein is an α2β2 heterotetramer of the nif D and nif K gene products that form NifDK (~220 kDa), where two αβ-dimer units pair together.61 Each αβ-dimer houses two different metalloclusters that are critical for enzyme function, the P- and M-clusters, such that each NifDK tetramer binds a total of four metalloclusters (Figure 2).61,65,66 The nif H gene product, NifH, serves as the reductase partner of NifDK, and is composed of a homodimeric protein (γ2) with a molecular weight of ~60 kDa that binds an [4Fe-4S] cluster at the subunit interface.67 Additionally, each subunit can bind one molecule of MgATP.63,67,68 One NifH protein can bind to each αβ-dimer of NifDK, forming an Fe:MoFe protein stoichiometry of 2:1.69,70 The P-cluster mediates electron transfer between the reductase component, NifH, and the M-cluster during catalysis.71 The cofactor is composed of an [8Fe-7S] cluster positioned at the α/β interface of NifDK, ~ 10 Å below the surface of the protein.65,72 It appears as the fusion of two [4Fe-4S] clusters that share a μ6-sulfide vertex, with ligation from six cysteine residues, α-C62, α-C88, α-C154, β-C70, β-C95, β-C153, three of which come from each subunit (Figure 3). Interestingly, the cysteine residues bind terminally to some of the Fe centers, as is typically found with canonical [4Fe-4S] clusters, but two of them bind in bridging modes between the two cubane halves of the P-cluster. Structurally, this fused cubane geometry of the P-cluster is unique and has not been identified elsewhere in biology. The P-cluster has three interconvertible oxidation states that have been identified as relevant for catalysis: the as-isolated PN state, the one-electronoxidized P1+ state and the two-electron oxidized POX (or P2+) state.62,73 The PN-cluster has been characterized as an all ferrous (Fe2+) cluster,74–77 and subsequent oxidations result in reversible structural changes that physically open the cofactor.78–80 This structural fluctuation of the P-cluster during redox changes has been proposed as a means of regulating or “gating” electron transfer within NifDK.62,81–83 The M-cluster is an asymmetric [Mo-7Fe-9S–C-R-homocitrate] species, with the Mo-capped end ligated by the organic acid R-homocitrate through the 2-hydroxy and carboxyl groups, as well as by the α−442His residue, and the opposite, Fe-capped end ligated by α-C275 residue (Figure 4A).65,66,85 The cluster is buried within the α subunit of NifDK and is positioned ~19 Å from the P-cluster and over 60 Å from the clusters of the partner αβ-dimer. The structure of the M-cluster is similar to the P-cluster, as both are a fusion of two different cubane species with a common vertex, but the two partial cubane units of the M-cluster, [Mo-3Fe–3S] and [4Fe-3S], are not identical, and the common vertex is an interstitial μ6-carbide as opposed to a sulfide.66,86–88 Additionally, there are three “belt” μ2-sulfido ligands that bridge between the Mo- and Fe-capped halves of the cofactor. These belt sulfide ligands have been shown to exhibit lability in both Mo- and V-nitrogenases, and this apparent flexibility has been implicated to be important during catalysis.89–95 In V-nitrogenase (or VnfDGK), the V-cluster is a structural analog of the M-cluster (Figure 4B), but with a V ion in place of the Mo ion, and one of the μ2-sulfide ligands replaced with a carbonate (CO32−) ligand of unknown origin.92–95 There are physical metrics that vary between the M- and V-clusters, but they are still remarkably similar.51 The electronic description of the M-cluster has been of interest for decades, as the M-cluster is capable of supporting many redox states,56 and so the cluster has been experimentally studied using combinations of Mössbauer, X-ray absorption (XAS), magnetic circular dichroism (MCD), and electron paramagnetic resonance (EPR) spectroscopic techniques.62,75,96–98 Many of these techniques probe the sample in an ill-defined state, and as such, there can be great difficulty in deconvoluting the specific contributions of individual clusters. The most commonly observed state of the M-cluster is the resting, dithionite (S2O42−) reduced MN state that is associated with the diagnostic S = 3/2 rhombic signal from EPR spectroscopy (g = 4.3, 3.7 and 2.0).73,96,99,100 Computations and high-energy resolution fluorescence detected (HERFD) Mo K-edge XAS experiments identified that the Mo center of the MN-cluster is best described as an S = 1/2 Mo3+ center, as opposed to the classical assignment of Mo4+.98 This indicates that the Mo exists in an atypical “non-Hund” configuration, and additional spectroscopy and density functional theory (DFT) calculations support a [3Fe2+:4Fe3+:Mo3+]1− assignment.97,98,101,102 Further, a crystallographic spatially resolved anomalous dispersion (SpReAD) analysis compared electron densities to Fe K-edge XAS experiments to determine the oxidation states of specific Fe atoms of the M-cluster, and this study agreed with the previously described oxidation assignment.103 The MN-cluster can also be oxidized by one electron to the MOX form, which coincides with a loss of the S = 3/2 EPR signal, and is generally not believed to be relevant for catalysis.73,104,105 While NifH binds a smaller, more prevalent type of biological [4Fe-4S] cluster, the Fe protein is far from simple. NifH has three recognized physiological functions: (1) Mo and homocitrate insertase in the biosynthesis of the M-cluster, (2) reductase involved in P-cluster biosynthesis on NifDK, (3) obligate reductase for nitrogenase catalysis that couples electron transfer to MgATP hydrolysis.68 The [4Fe-4S] cluster binds between the two subunits of NifH and is positioned on a C2 rotation axis with cysteine coordination by two residues (γ-C97 and γ-C132) from each subunit.67 This cluster is located in a solvent-exposed position, which differs from [4Fe-4S] ferredoxins that have clusters buried within the protein.63 Further, MgATP molecules that bind to the Walker motif A protein fold have been shown to affect the position of the [4Fe-4S] cluster despite being bound ~20 Å away from the cluster (Figure 2).68,106,107 This apparent cluster movement has been studied by various structural and spectroscopic methods and is associated with modulation of the reduction potential of NifH.63,108–112 The protein can support [4Fe-4S]2+, [4Fe-4S]1+, and [4Fe-4S]0 oxidation states, which is abnormal compared to other iron–sulfur cluster containing proteins that only support two redox states.113 In NifH, the [4Fe-4S]1+ state is favored in the presence of dithionite51,100 and interestingly exists as a mixture of two different spin states, S = 1/2 and S = 3/2, the composition of which changes with additives such as glycerol or urea.96,114 During catalysis, the [4Fe-4S]2+/1+ couple is proposed to be operative, transferring one electron to NifDK for every two molecules of MgATP cleaved by NifH, and NifH becomes reduced in vivo through physiological reductants such as ferredoxins or flavodoxins.63,83,115–118 The [4Fe-4S]0 “all ferrous” or “super-reduced” state has been implicated in reactivity of NifH with carbon substrates like CO and CO2, yet has also been proposed to play a role in nitrogenase catalysis, but the characterization of this species is fraught with complications that make understanding a physiological role difficult.51 2.2. Assembly of Nitrogenase Cofactors Nitrogenase is a complicated system, not only because of the multimeric metalloprotein composition and unique cofactors, but also because of the requirement of many genes for the regulation and biosynthesis of nitrogenase.50,51,119 For the NifDK protein from the model organism A. vinelandii, a minimum set of nitrogenase proteins that includes NifU, NifS, NifB, NifEN, NifH, NifV, and NifZ are required for in vitro activation of N2 fixation.49,120 These proteins are encoded by nif or “nitrogen fixation” genes, and the nif operon is specific to the Mo-nitrogenase system. Organisms that encode for V- or Fe-only nitrogenase (vnf and anf genes, respectively) still require some of the nif operon for proper function.51 For Mo-nitrogenase, the biochemistry and assembly of the MoFe protein and the cofactors bound therein have been extensively studied over the past two decades, yielding crucial insights into the process. The biosynthesis of NifDK involves the assembly of both the P- and M-clusters in addition to the polypeptide, but cofactor generation occurs through different routes (Figure 5). While the P-cluster is assembled in situ on the NifDK polypeptide scaffold, the M-cluster is assembled ex situ on a series of different proteins before terminally reaching NifDK.49,50,120 The P-cluster originates from a pair of [4Fe-4S] clusters, called the P*-cluster, that are synthesized at the α/β subunit boundary of apo NifDK.84,121–124 The iron–sulfur fragments are assembled and delivered to apo-NifDK through the action of NifU, a scaffold protein that binds iron–sulfur clusters, and NifS, a cysteine desulfurase that provides sulfide from cysteine.125,126 The P*-cluster is then fused together with concomitant loss of sulfur by the stepwise action of NifH and NifZ in a MgATP- and electron-dependent process. The result is an apo-NifDK protein with two equivalents of P-cluster bound but lacks M-cluster. In contrast, the assembly of the M-cluster starts with the construction of [4Fe-4S] clusters by NifU and NifS that are subsequently delivered to NifB, a radical S-adenosyl-l-methionine (SAM)-dependent enzyme.127,128 NifB binds three [4Fe-4S] clusters; one is specifically for catalyzing SAM reactivity, while the other two are the substrate clusters (K1 and K2) known collectively as the K-cluster.129–131 Under reducing conditions, two equivalents of SAM are cleaved to first transfer a methyl group to the K2 cluster on NifB, and the second equivalent of SAM initiates a hydrogen atom abstraction of the transferred methyl group.87,88,131 In a series of uncharacterized subsequent steps, carbide (C4−) is generated upon removal of the remaining hydrogen atoms. Concurrently, the two K-cluster cubane units are fused yielding a [8Fe-8S–C] cluster, called the L*-cluster, that contains an interstitial carbide bound to the six central Fe atoms of the cluster.132,133 A final “9th sulfur” is then incorporated into the L*-cluster through the reduction of a sulfite (SO32−) ion to produce the [8Fe-9S–C] L-cluster. The L-cluster on NifB is transferred to another scaffold protein, NifEN, that facilitates the conversion of the L-cluster into the M-cluster through the loss of an Fe atom and the incorporation of Mo and R-homocitrate.134–137 In vitro experiments demonstrate that M-cluster-loaded NifEN can then interact directly with apo NifDK (containing only P-clusters) to transfer the M-cluster and produce active NifDK, although additional chaperone proteins have been suggested to assist with this process in vivo.49,50 2.3. Nitrogen Reduction Mechanism N2 reduction requires a high degree of regulation to successfully carry out the reaction, and one of the key factors to this process is the shuttling of electrons. Electrons are transferred from NifH to NifDK through direct complexation of both proteins and the hydrolysis of two MgATP molecules for each electron delivered.63 This creates a complex system that has been the subject of study for decades. Extensive investigation of the mechanism of N2 reduction in the 1970s and 1980s led to a kinetic framework known as the Lowe-Thorneley model or cycle.51,52,63 This model can be viewed from two different perspectives; one is an Fe protein-centric view known as the “Fe protein cycle” and the other, more common view is that of the MoFe protein, or the “MoFe protein cycle”. The Fe protein cycle begins with the NifH protein being in the reduced [4Fe-4S]1+ state (NifHR), binding two equivalents of MgATP.63,83 The reduction to NifHR has been facilitated in vitro by dithionite, but other reducing agents such as Eu(II) chelates have been used as well. The nucleotide-bound Fe protein then interacts with NifDK, forming a 2:1 complex that has been crystallographically characterized.69,70,138 Subsequently, there are a series of steps that involve conformational changes, electron transfer, ATP hydrolysis, and phosphate release, resulting in oxidized NifH (NifHOX, [4Fe-4S]2+) with two equivalents of MgADP bound in complex with NifDK.63,83 NifHOX and NifDK undergo dissociation, releasing the reductase before NifHOX is reduced back to the NifHR state, which becomes capable of exchanging MgADP for MgATP, thereby restarting the cycle. This cycle is presumably active during the reduction of most substrates by NifDK, though in the absence of NifDK, NifH has also been shown to carry out C1 substrate reduction through alternative mechanisms.51,68 The Lowe-Thorneley model for N2 reduction by NifDK, or the MoFe protein cycle, describes the catalytic intermediates in the reaction, each designated by an En notation where n is the number of proton and electron equivalents that have accumulated on one αβ-dimer of NifDK (Figure 6).56 However, the value of n does not indicate where proton and electron equivalents are located, only that they have been delivered to the system, which has sparked considerable research efforts in the field.52 The first state, E0, reflects the resting, as-isolated state of NifDK associated with the well-characterized S = 3/2 EPR signal of the M-cluster.62 As each proton and electron pair is accumulated through the early (n = 2–4) steps of the cycle, it is possible for a “relaxation” to occur, releasing H2 with concomitant transition to an n − 2 state.52,60 N2 has been proposed to bind to the M-cluster in the E3 and E4 states, but reduction of N2 only occurs in the E4 (or E4(4H)) state.52,56,139 This makes the E4 state one of the most studied of the cycle, because either E4 can relax backward to lower En with release of H2, or the cycle can proceed forward with the cleavage of the N2 bond. Intriguingly, when N2 binds to E4(4H), one equivalent of H2 is produced with concomitant formation of the E4(2N2H) state through a proposed mechanism of reversible reductive elimination or oxidative addition.52,60,140 The E4(4H) to E4(2N2H) reaction represents the first lowering of the N2 bond order, though several possibilities have been put forward as to the identity of the resultant E4(2N2H) species.51,52 The subsequent steps of the MoFe protein cycle E5–E8 promote the further reduction of the N2 bond. Initially, there were two competing proposals for this latter half of the cycle that differed in the location where proton and electron equivalents are delivered to the reduced nitrogen species: a “distal” and an “alternating” pathway. The distal pathway posits that protons are delivered to the unbound N atom of N2 first, forming an M = N–NH2 hydrazido species that undergoes further proton/electron addition to yield one equivalent of NH3 and a terminal M ≡ N metal nitrido species by the E5 state.141–146 The E5 species is then sequentially protonated to release the second equivalent of NH3 after the E8 state. The alternating pathway, based on synthetic Fe-catalyzed versus Mo-catalyzed N2 reduction,147,148 proposes that protons are distributed between both N atoms of N2, starting with diazene (HN=NH), proceeding through hydrazine (H2N–NH2) and releasing NH3 after E6 and E8. While there is still uncertainty in distinguishing between these two mechanistic possibilities, mounting evidence supports the alternating pathway for biological N2 reduction,51,52 including crystallographic evidence of a putative N2-bound species.91 However, evidence from the V-dependent nitrogenase potentially suggests a distal pathway may be operative in that system, but kinetic analysis and calculations propose that the Mo-, V-, and Fe-only nitrogenases all share a common mechanism.51,93,149,150 Clearly, further investigation will be necessary. 2.4. Nitrogenase Active Site Pocket To understand the outer sphere environment surrounding the M-cluster, it is necessary to take a critical look at the residues that make up the active site pocket. However, there is no “best” way to discuss the location of amino acid residues with relation to the catalytic cofactor of nitrogenase. High resolution crystallography has provided three-dimensional representations of nitrogenase and the cofactors with precision, but it can be difficult to gain a sense orientation of the cofactor with respect to its protein environment. The 3-fold rotation axis that runs from the Mo-capped end to the Fe-capped end of the cluster does not inherently provide a landmark around the M-cluster by which relative position can be compared other than the Mo and Fe subcubanes. The crystallographic labeling of the cofactor atoms provides some basis of orientation, but the labeling convention is also not particularly intuitive; for instance, the three belt sulfur positions are given labels of S2B, S3A, and S5A, whereas other SXA and SXB labeled sulfur atoms are bound to the Fe-capped (or Fe1) and Mo-capped subcubane structures, respectively (Figure 7). In this discussion, a helpful means to navigate the M-cluster site is to establish an origin at the central carbide atom, followed by an “x-axis” that incorporates the C3 symmetry axis running between Fe1 and Mo. The Fe1 end of the M-cluster is pointed toward the surface of the α-subunit adjacent to the bulk solvent (designated as the “negative” x direction), whereas the Mo end of the cluster is directed toward the αβ-subunit interface (designated as the “positive” x direction). The S2B sulfur site has been established as a labile position on the catalytic cofactor,89,90 and is adjacent to several residues critical for reactivity, so a “z-axis” would bisect S2B as well as the central carbide atom of the cluster, with the direction of the S2B atom being positive. There is no convenient atom to position a “y-axis” relative to the carbide, but the S3A side of the xz plane can be the positive direction of y and the S5A side can be negative. Within this framework of (x, y, z) coordinates, α-H442 would be in the (+, +, −) direction relative to carbide, α-C275 would have a (−, 0, 0) position, and the nearest P-cluster would have a bearing of (+, −, +). While also an imperfect system due to constraints, such as α-R359 (Figure 8) and the R-homocitrate ligand (HC) running through positive and negative values of y, the coordinates should generally provide a spatial navigation within the active site. The catalytic M-cluster of nitrogenase is one of the largest, most complicated biological cofactors known so far but it is only bound to the protein scaffold through two amino acid residues, α-C275 and α-H442.61,65,66,85,151 While R-homocitrate does bind to the Mo-end of the cluster, the lack of additional ligation to the M-cluster indicates that second coordination sphere effects play a crucial role in catalysis. Many amino acids have side chains with nitrogen-containing functional groups such as histidine, arginine and glutamine that are capable of hydrogen-bonding interactions. Nonpolar or otherwise bulky residues such as valine, phenylalanine and tryptophan have been found to play roles in the steric environment for substrate access and binding, as well as appropriate positioning of the M-cluster within the active site. All of these residues are connected to each other in a tightly regulated way, and disruptions in this network tend to be detrimental to the function of the enzyme. Figure 8 shows the positions of some of the important residues discussed within this review relative to the M-cluster. In 1991, before crystal structures of NifDK were solved, a series of point mutations of conserved histidine residues were reported, hoping to identify those interacting with the M-cluster. Making use of electron spin echo envelop modulation (ESEEM) spectroscopy,152 microwave pulses are used in a specific sequence within an EPR experiment to probe nuclei (such as 14N or1H) that are coupled to a paramagnetic species.62 Specifically, the intense S = 3/2 EPR signal (Figure 9) associated with the M-cluster was probed to search for 14N atoms from histidine residues that may be hyperfine-coupled with or covalently bound to the cofactor.152 Residues α-H83 (+, −, +), α-H195 (−, 0, +), α-H196 (−, 0, +), α-H274 (−, −, −), and β-H90 (+, +, +) were targeted for single point mutations. ESEEM studies on whole cells were carried out and signals assigned to N atoms from histidines were observed in all variants, only the variant α-H195 NifDK proteins showed no ESEEM signal, and it was concluded that α-H195 must bind to the M-cluster. However, once the crystal structure of NifDK became available in 1992, it was clear that α-H195 was too far from the M-cluster to bind directly, but instead was likely involved with a hydrogen-bonding interaction to the S2B sulfur of the cofactor. In 1995, Dean, Hoffman, and co-workers studied the EPR and ESEEM signature of partially purified NifDK variants, namely α-H195N and α-H195Q.153 Consistent with the previous observations, it was found that the ESEEM signal was dramatically attenuated in the α-H195N NifDK variant but α-H195Q NifDK showed nearly identical signals in both EPR and ESEEM experiments to those of the wild-type enzyme. If the hydrogen bond of α-H195 was responsible for the ESEEM signals, then both variants should display similar spectra, but they did not. This suggested that a nitrogen-containing residue that is sensitive to the position of the α-H195 residue was responsible for the signal. Subsequently, ESEEM studies were expanded to include purified variant proteins targeting residues α-R96 (+, −, 0), α-H195 (−, 0, +), α-R359 (0, +, −), α-F381 (0, +, +), and α-H442 (+, +, −).154 The spectra of α-H195N NifDK showed that the ESEEM signal (labeled as “N1”) was indeed eliminated as previously reported,152,153 but a weaker signal “N2” that was buried underneath was also observed.154 Mutation at α-R96 only showed the N1 signal, eliminating this residue as the source of N1, and substitution of α-H442 resulted in a lack of observable EPR signal, consistent with loss of the M-cluster entirely. However, mutation of α-R359 or α-F381 resulted in a loss of the N1 signal while the N2 signal remained intact. Together, these experiments were used to assign the N1 signal to α-R359 and the N2 signal to the peptide NH groups of one or both of the residues α-G356 (+, +, 0) and α-G357 (−, +, 0).154 However, the catalytic activity of the variant NifDK proteins could not be correlated to the ESEEM signals. Despite this, the study demonstrates that changes in the second coordination sphere on one side of the M-cluster can perturb the interactions occurring on the other side of the cluster, as seen by the loss of the N1 signal at α-G359 (0, +, −) when α-H195 (−, 0, +) is modified. In fact, changing the steric environment as seen with α-F381L NifDK variants (0, +, +) was enough to disrupt hydrogen-bonding interactions around the cluster even though α-F381 is not directly involved in such an interaction, reinforcing that the active site residues are highly interconnected. Another salient observation about the active site of nitrogenase comes from the crystal structure of an M-cluster depleted NifDK, called Δnif B NifDK.155 In 2002, Burgess, Rees, and co-workers managed crystallized the NifDK protein without the catalytic cofactor, which resulted in changes to the tertiary structure as compared to wild-type enzyme. Rearrangements of the amino acids left a “funnel” on the surface of the protein that was positively charged, with a loop of residues from α−353 to α−364 positioned at the entrance to the funnel. This loop also contained residues that had positively charged, nitrogen-containing side chains, and the authors proposed that the electrostatic charge of the funnel would help direct the negatively charged M-cluster into the appropriate position so it could be locked into the active site through movement of the funnel loop. Interestingly, a similar concept was discussed for the insertion of the active site cofactor of [FeFe]-hydrogenase (Section 3.2.2). Subsequent point mutation studies further explored this proposal by targeting specific residues that may assist in guiding the M-cluster into NifDK.156–158 Thus, the side chain selection within the nitrogenase active site has the demonstrated capability to support a robust network of hydrogen bonds, while at the same time functioning in a biosynthetic capacity to orient the M-cluster optimally within the protein for catalysis. 2.5. Assessing the Nitrogenase Second Coordination Sphere Before the publication of the crystal structures in the 1990s,65,67,69,80,85,159 the study of nitrogenase was based on genetic, spectroscopic, and reactivity experiments that were used to gain insight into the structure of the protein as well as the bound metallocofactors. At that time, direct information regarding the properties of the individual P- and M-clusters was scarce, and their three-dimensional relationship. Even after the appearance of high-quality crystallographic data, the study of structure–function relationships in nitrogenase was difficult due, in no small part, to the multiple, unique iron–sulfur clusters that are bound to the enzyme. Some of the early efforts to characterize and understand the nitrogenase cofactors involved chemical extraction into organic solvents.160,161 The P-cluster is integral to the nitrogenase structure, so the acid quenching required for extraction destroys the P-cluster and disrupts the protein fold such that the M-cluster can be removed. This allowed for new avenues of study and analysis, but to date, there have been no successful reports of recrystallized M-cluster outside of the nitrogenase protein.162 To better assess the nitrogenase protein structure, amino acid sequencing was carried out through genetic analysis and protein digestion in the 1970s and 1980s to compare nitrogenases from different species.163–168 Conserved residues were identified and targeted for single-point mutations as a way to connect information about the cofactors to the protein structure and function. Initially, conserved cysteine residues were investigated for their widespread use in biology as ligands for iron–sulfur clusters, as it was understood that nitrogenase likely contained multiple [4Fe-4S] clusters. These cysteine residues were changed into serine or alanine, and the resulting bacterial strains (of A. vinelandii or Klebsiella pneumoniae) were studied.169–172 Growth rates in the absence of a fixed nitrogen source were used to establish the N2 fixation phenotype, and whole cell and/or crude extract activity assays were carried out in an acetylene atmosphere to compare diazotrophic growth rates with nitrogenase activity.169,172 In addition, EPR spectra of whole cell variants or crude extracts of these cells would be collected to search for the intense S = 3/2 signal associated with the M-cluster as a means to cross reference the observed reactivity.172 Generally, when the cysteine residues (that would later be found to bind to the P- and M-clusters) were mutated to alanine, diazotrophic growth and acetylene-based nitrogenase activity would be eliminated. Serine subsitutions of cysteine residues bound to cofactors would also eliminate activity, but in one case, Newton and co-workers found that substitution of C183 (+, −, +) on the NifD subunit (or α-C183) from A. vinelandii to serine did not arrest activity but resulted in diminished diazotrophic growth rates.169 Conserved residues such as α-Q191 (+, +, +) were mutated to change the polarity or charge; for instance, changing the polar, amide-bearing amino acid side chain of glutamine to the carboxyl group of glutamate (α-Q191E).169 In this case, the activity was eliminated, despite the fact that glutamine residues are not typically found as ligands to iron–sulfur clusters in biology.113,173 These findings were unanticipated, but provided indications of the active site of nitrogenase in the absence of structural data, and marked the beginning of targeted efforts to understand the secondary sphere effects within nitrogenase. In the years that followed, many mutagenic studies were reported with a focus on determining effects to nitrogenase activity as well as the number and location of substrate binding sites on the M-cluster. The residues that have been targeted for mutagenic studies related to the M-cluster site in NifDK from A. vinelandii have been summarized in Table 1 and Table 2. One complicating factor when discussing the results of site-directed mutagenic studies is that often the modification substantially reduces or eliminates nitrogenase activity, but it is not always clear why this occurs. In studies of the MoFe protein from K. pneumoniae, Buck and co-workers tested the activity of cell extracts from several cysteine to alanine variants and found almost no acetylene reduction, implying their mutations have made nitrogenase nonfunctional.172 They employed a parallel experiment to assess the stability of the variant proteins where wild-type apo NifDK (Δnif B NifDK) was added to their cell extracts and would show activity only if free M-clusters were present in the extracts to enable reconstitution of the Δnif B NifDK protein. Low acetylene reduction activity was observed for many variants following such a treatment, but the α-C275A NifDK protein that targeted the M-cluster-binding residue showed an increased activity by an order of magnitude.172 This indicated that some variant proteins are unable to retain the M-cluster, a conclusion that was validated by adding additional M-cluster to cell extracts of the NifDK variants alone with no observed increase in activity. Indeed, extracts were commonly used because of the difficulty associated with purification, and this problem becomes exacerbated when the stability of the protein is affected. Several reports from Newton and Dean acknowledge that certain proteins could only be partially purified because they would not survive heat treatment steps.139,153,183,185,191 In some instances where purification was successful, the Mo content of the NifDK variant would be lowered in the wild-type enzyme (<50%), implying that cofactor loading was impaired resulting in variable mixtures of apo and holo NifDK proteins.139,183,191 These studies still provide valuable insight, but it is important to recognize that it can be difficult to make definitive assessment of outer sphere effects when activities can be attenuated from simple loss of the M-cluster or degradation of the protein. However, there are investigations targeting the assembly of NifDK that use the retention or loss of the M-cluster as means to probe the function of certain amino acids.156–158 Ribbe and co-workers have carried out site-directed mutagenesis on several outer coordination sphere residues ≥6 Å from the M-cluster (Figure 10): α-W444 (+, −, −), α-H274 (−, −, −), α-H362 (−, +, −), and α-H451 (−, −, −). A crystal structure of apo NifDK from a Δnif B background revealed that α-H442 in the absence of the M-cluster shifts ~5 Å within the active site.155,158 The α-H442 residue is rearranged and placed nearby two other histidine residues, α-H274 and α-H451, forming a “triad” while another residue, α-W444, is found to take up the same space as α-H442 does in the holo enzyme (Figure 10). The apparent position swapping of α-H442 and α-W444 was intriguing, so mutagenesis was carried out to replace α-W444. In the α-W444Y and α-W444F NifDK variants, the M-cluster loading was ~90% of the wild-type enzyme, whereas α-W444A and α-W444G contained only 14% and 3% of wild-type M-cluster, respectively.158 As the side chain of the α−444 residue was changed to smaller, less sterically bulky groups, the Mo concentration decreased. The reactivity assays of the variant proteins with N2, C2H2, and protons, as well as the intensity of the EPR signals all roughly decreased by the same magnitude as the Mo concentration. Paired with heat treatment studies, it was suggested that the variant proteins themselves were stable, leading to the conclusion that α-W444 must be involved in a steric interaction that keeps the M-cluster secure within the active site pocket.158 It was also observed that in the α-W444A and α-W444G variants, the EPR signal of the P-cluster appeared to distort relative to the larger substituent-containing variants, implying that α-W444 may also have longer range effects in NifDK than immediately around the M-cluster site. The other two residues of the histidine triad, α-H274 and α-H451 (Figure 10), were also targets for point mutations.156 A. vinelandii cells separately expressing NifDK variants α-H274A, α-H451A and a double mutant α-H274A/α-H451A grew more slowly under N2-fixing conditions, indicating issues with nitrogenase activity, with the double mutation having the most deleterious effect on the growth rate. Assessment of the standard substrate reduction assays showed decreased activity of 67%, 56%, and 43% for the α-H274A, α-H451A, and α-H274A/α-H451A NifDK variants, respectively. The decreased activity also correlated well to the Mo concentration, and to the intensity of EPR signals for each of the variants. The α-H451A NifDK protein in the dithionite-reduced state had a very similar EPR spectrum for the S = 3/2 M-cluster as wild-type NifDK, but the introduction of α-H274A as either the single or double mutant yielded a spectrum with large distortions in g-values as well as line shapes. The rationalization for this observation was that α-H274 is adjacent to α-C275, which binds the M-cluster and may have more of a direct impact on the M-cluster environment than the more distant α-H451.156 Additionally, none of the studied mutations affected the EPR signals of the P-cluster, indicating that these residues have less of a long-range steric effect than α-W444. Introduction of isolated M-cluster into solutions of M-cluster deficient α-H274A, α-H451A, and α-H274A/α-H451A Δnif B NifDK variants did not allow for full activation of the protein to wild-type levels of reactivity. Thus, it was concluded that both α-H274A and α-H451A are required during the biosynthesis of NifDK through sequential binding events that draw the M-cluster into the protein.156 These residues may also play roles in the proton shuttling within the active site, as seen from shifts from N2 reduction toward proton reduction in the variants, but one factor that was not explored is what steric effect the size of the side chain played in going from histidine to alanine. Alanine substitutions were certainly deleterious, but it would be interesting to see how the protein would retain the M-cluster if uncharged residues larger in size than alanine were studied. Along similar lines, a final outer coordination sphere residue, α-H362 (−, +, −), had been studied with respect to M-cluster insertion into the NifDK protein (Figure 10).157 In the crystal structure of the apo NifDK protein, α-H362 is surface exposed and is at the entrance to the insertion funnel of positively charged residues that have been proposed to guide a negatively charged M-cluster into the protein.155,157 After M-cluster incorporation, the residue moves with the protein secondary structure to pin the cofactor in place. Single point mutants were generated and two variant NifDK proteins were studied, α-H362D and α-H362A.157 Much like the previous mutations described, the α−362 variants grew more slowly under N2-fixing conditions, expressed a decreased reactivity profile that correlated with Mo content, and had perturbations to the line shape and intensity of EPR signals, relative to wild-type enzyme. The decreased activities of the α-H362D and α-H362A NifDK variants are ~80% and ~40%, respectively. While modification of this residue did not cause a cessation of activity like that observed for α-W444,158 substitution of α-H362 for aspartate induces a ~ 20% reduction in activity while alanine substitution results in a more drastic reduction of ~60%. This suggests that either α-H362 is involved in directing the M-cluster into the NifDK active site (supported by the decreased Mo concentration), or the residue is structurally important for adequately positioning the M-cluster within the active site (supported by the observed perturbations of the M- and P-cluster signals). The best answer is likely a combination of both, but further structural characterization of these NifDK variants may shed light on the M-cluster insertion process. 2.5.1. Reduction of Nitrogen, Acetylene, and Protons. As mentioned earlier (eq 1), the reduction of N2 by nitrogenase is a challenging transformation that requires the delivery of 8 electrons, 8 protons, and 16 equiv of MgATP to produce NH3, H2, and MgADP.51,52,100 Not only is NH3 produced, but H2 is generated as an integral part of the N2 reduction mechanism.57,58,192 As discussed in Section 2.2, electron and proton equivalents are sequentially delivered to the catalytic cofactor by way of NifH and the P-cluster, leading to distinct En steps along the Lowe-Thorneley model reaction pathway.56,60 At the E4(4H) level, N2 has been proposed to bind through a reductive elimination/oxidative addition mechanism, forming a reduced nitrogen species E4(2N2H) and liberating one equivalent of H2 (Figure 6). It is important to emphasize that while the bond order of N2 is lowered when binding to E4, the conversion between E4(4H) and E4(2N2H) technically requires no additional energy input, and the process is reversible in the presence of H2. In fact, this reversibility is what allows H2 to act as an inhibitor of N2 reduction. This phenomenon has also been exploited to investigate N2 binding. If isotopically labeled D2 (2H2) is mixed with N2, the wild-type NifDK can produce HD gas, and, if separately, T2 (3H2) is mixed with N2, NifDK does not significantly accumulate mixed label TOH.56–60 This means that the reversible H2 production through N2 binding is not a simple proton exchange process with solvent, which supports the notion that hydrides (H−) are generated in the Lowe-Thorneley model.52,60 Moreover, the relative rates between the forward and reverse reactions have been proposed to favor N2 reduction over N2 release,52,149,150 but these rates change when moving to the alternative nitrogenases. Protons are also reduced by nitrogenase in a hydrogenase-like process (hydrogenases are discussed in Section 3) that competes with other substrate reduction reactions, but also occurs in the absence of substrates such as N2, C2H2, or CO.55 Additionally, there is no known additive that can increase the proton reduction of the wild-type enzyme, and it is unclear what the site(s) of proton reduction is. This can complicate analysis of nitrogenase catalysis, as often enough N2 or acetylene reduction is reported without proton reduction under the same conditions. Consequently, it becomes difficult to rigorously compare systems, particularly from single-point mutagenic studies. For instance, if N2 reduction decreases between two NifDK variants, the proton reduction under those same conditions could either increase or decrease, and each scenario would lead to a different conclusion. There is also difficulty in the accurate measurement of NH3/NH4+ produced from catalysis compared to gaseous H2, particularly at low concentrations. For this reason, acetylene is often used as an N2 surrogate due to the analogous triple bond and gaseous nature of both the substrate and product states. Acetylene was first discovered as a competent substrate for nitrogenase in 1966 by Michael J. Dilworth,193 and since then it has become crucial to the study of the nitrogenase mechanism despite the nonphysiological role the substrate plays.51,52,100,194 Wild-type Mo-nitrogenase reacts with C2H2 such that >90% of electron equivalents are funneled to produce exclusively ethylene (C2H4), substantially attenuating the H+ reduction pathway.100,195 The alternative substrate is thought to bind to the nitrogenase E1 or E2 states, although, C2H4 is not released from the protein until three proton/electron equivalents are delivered to the M-cluster so that the E1 state can be reformed.195 This is in contrast to the proposed N2 binding states of E3 or E4, the latter of which is proposed to be the only productive binding state that results in N2 reduction (Figure 6). Further, the analysis of acetylene reduction in the presence of the nitrogenase inhibitor carbon monoxide (CO) has led to the proposal that there are multiple substrate binding sites on the M-cluster, and that N2, C2H2 and CO occupy different positions.195 Finding where the substrate binding sites are located as well as the properties of each has been a driving force behind the analysis of single-point mutations of NifDK. There have been many amino acid residues that have been targeted for single-point mutagenesis, and herein, only examples pertinent to the outer sphere effects on catalysis will be discussed. Table 3 summarizes some of the available assay data for NifDK variant proteins. In the 1990s, NifDK variants at the α-Q191 (+, +, +) and α-H195 (−, 0, +) residues were reported based on sequence conservation among NifDK analogs, and between NifDK and the biosynthetic protein NifEN.182,185,187,188 The cells of variant α-Q191K and α-H195N NifDK expressing strains had a Nif− phenotype where the cells did not appear to grow without the addition of a nitrogen source (such as ammonium chloride or urea).182 This indicated that these NifDK variants were unable to reduce N2 under standard conditions. Crude extracts of cells expressing α-Q191K and α-H195N NifDK variants were found to yield no N2 fixation products, however, reaction with C2H2 produced C2H4 with <10% of the activity of the wild-type enzyme. Furthermore, ethane (C2H6) was found as a product from the variant protein strains, with 12.5% and 35% of the electron equivalents diverting toward ethane production for the α-Q191K and α-H195N NifDK proteins, respectively.182 This was an interesting result at the time because it demonstrated that the M-cluster site was likely nearby the targeted residues prior to the publication of crystallographic data. The α-Q191K NifDK protein was purified in 1992 and the reactivity of the enzyme was found to be ~30% of the activity of the wild-type NifDK. It produced no N2 reduction products, and diverted 90% of electron equivalents to the production of H2 in the presence of C2H2, compared to wild-type enzyme that uses 90% of electron flux to generate C2H4.185 Subsequent characterization has shown that the Michaelis constant (Km) for C2H2 to produce C2H4 is 35 kPa, 2 orders of magnitude higher than that for the wild-type enzyme (Km = 0.5 kPa), indicating that acetylene is not a great substrate for α-Q191K NifDK.183 It has been further demonstrated that α-Q191K NifDK does not react with C2H4 to produce C2H6, in contrast to wild-type NifDK that will produce only a small amount (<1% of electron flux) of C2H6 from C2H4 when C2H2 is absent. This and isotope labeling experiments led to the assertion that there must be a longer-lived, common ethylenic intermediate that can be differentiated to release ethylene, or ethane.183 The lack of N2 reduction was also explored through the introduction of N2 gas into C2H2 and proton reduction assays.139 Under standard conditions, wild-type NifDK will divert electron flux from proton reduction to N2 reduction if N2 is introduced into the reaction vessel, indicating that N2 binds to or otherwise interacts with the M-cluster. HD will also be observed as a product in the presence of both N2 and D2. In contrast, α-Q191K NifDK does not appear to even bind N2 because the catalytic performance is unaffected when N2 is added into the typical acetylene or proton reduction assays.139 When mixtures of N2 and D2 are added to reactions of α-Q191K NifDK, the product HD is absent, consistent with a lack of N2 binding as well as an inability to access the E4 state. A rationalization provided for these observations is that the α-Q191K variant is unable to even reach the E3 level of the Lowe-Thorneley cycle (Figure 6).139 From crystallography, it is known that α-Q191 is in proximity to the R-homocitrate ligand, and both the protein residue and the ligand are involved in hydrogen-bonding interactions,65,85,151 potentially causing deleterious modulation of the redox properties of the M-cluster. Interestingly, when R-homocitrate is exchanged for citrate in a Δnif V variant of NifDK from K. pneumoniae (see below), N2 reduction decreases to 40% of the wild-type activity, diverting more electron flux toward the production of H2 that exceeds that for the wild-type NifDK; yet, the Δnif V NifDK variant is still capable of binding N2.196 Clearly, α-Q191 plays a key role in regulating the energetics in the active site, but the mechanism by which this is accomplished remains elusive. In parallel, α-H195 (−, 0, +) variants have also been purified and studied.139,183,187,188 The initial report of α-H195N showed that the enzyme was relatively unstable, did not have a high reactivity toward C2H2 and proton reduction compared to wild-type, and was unable to form N2 reduction products.182,185 EPR spectra of whole cells indicated that while α-H195N NifDK showed a low intensity S = 3/2 M-cluster signal, the α-H195Q variant had an EPR signal of comparable intensity to wild-type enzyme with some perturbations of the g-values, suggesting that the electronic properties of the cluster had not substantially changed.187 The α-H195Q variant showed a modest decrease in activity toward C2H2 compared to wild-type enzyme, but there was no observable products of N2 reduction, even under acid-quenched conditions that would yield N2H4 in wild-type NifDK.198,199 However, in the presence of N2, the α-H195Q protein proton reduction was reversibly inhibited by slowing the reaction rate, but did not divert electron flux toward other observable products.187 This indicated that N2 would bind to the M-cluster of the variant, but would not become reduced. To further explore this notion, Fisher and co-workers increased the pressure of N2 used in reduction assays of α-H195Q NifDK and saw that under hyperbaric conditions, NH3/NH4+ was observed at a rate <2% of the wild-type enzyme.188 Additionally, in the presence of a mixture of N2 and D2, α-H195Q NifDK produced measurable HD gas, indicating that the protein variant could access the same E4 state as the wild-type enzyme, but could just not reduce N2 well. A crystal structure of the α-H195E NifDK protein was later published, showing that in agreement with EPR spectra, the N–H–S2B hydrogen bond stayed intact in the α-H195Q variant like that in the wild-type protein.184 Thus, the hydrogen bond does not sufficiently control N2 reduction capabilities, and additional factors, such as proton donors and redox potentials, should be considered. The α-H195N variant was also tested for the ability to bind N2, and it was found that N2 would inhibit proton reduction much like in the α-H195Q variant, but no HD evolution was measured. This suggests that the α-H195N variant could bind N2 (that is, achieve the E3 state) but was unable to reduce N2 (that is, unable to access the E4 state).139 In this case, it is unlikely that any hydrogen-bonding interaction between α-N195 and the M-cluster exists, but the protein is still able to bind, but not reduce, N2. In comparison, α-Q191K NifDK likely maintains the hydrogen bond between α-H195 and S2B of the cofactor, but the protein variant is unable to bind N2 at all. Acetylene reduction was also investigated with the variant α-H195 proteins, and it was found that N2 would inhibit the production of both C2H4 and C2H6, but this inhibition could be reversed with H2.139,183,188 In the presence of 10% acetylene, the α-H195N and α-H195Q proteins used 69% and 55% of electron equivalents to generate reduced products, but the former protein yields both C2H4 and C2H6 while the latter yields C2H4 exclusively.183 The α-H195Q protein behaves similarly to wild-type enzyme, with a Km for C2H2 of 0.5 kPa for both enzymes, and neither producing C2H6 from C2H2; however, the α-H195Q NifDK variant is less efficient than the wild-type enzyme in C2H2 reduction. In comparison, the α-H195N variant has an affinity for C2H2 (Km = 1.0 kPa) similar to that of the wild-type enzyme but produces ethane which accounts for 23% of the total electron flux. Ethylene was also tested as a substrate for the variant proteins in comparison to the wild-type enzyme, with a Km value of 120 kPa measured for both wild-type and α-H195Q proteins (though the latter uses less than 1% of total electron flux toward ethane production) and a Km value of 48 kPa measured for the α-H195N variant (alongside a 10-fold increase in activity). These experiments revealed that the high affinity for acetylene in the wild-type enzyme and α-H195Q variants likely prevents ethane production, because once ethylene is produced, it is rapidly replaced by a new molecule of acetylene.183 The slightly lowered affinity for the substrate in α-H195N NifDK may explain why ethane can be observed as a product, but it is possible that the steric environment also contributes, as asparagine has a shorter side chain than glutamine. These acetylene reduction experiments, in comparison to N2 reduction, illustrate that more rigorous conditions are required for the cleavage of the triple bond of N2 compared to alkyne or alkane substrates. To better identify and understand substrate binding sites on the M-cluster, additional mutations of nitrogenase were studied.174–178,180 In wild-type NifDK, several observations have been made with respect to the competition between N2 and acetylene binding to the catalytic cofactor; specifically, it has been shown that two binding sites with different affinities for acetylene exist, and that acetylene is a noncompetitive inhibitor of N2 reduction while N2 is a competitive inhibitor of acetylene reduction.193,200–203 Additionally, C2H2 has also been found to be a more potent inhibitor of N2 reduction under low electron flux conditions, where the rate of electron delivery is decreased, compared to those under high flux conditions.204 In 2000, a study was reported that used an Azotobacter vinelandii strain expressing a NifDK variant with a β-Y98H (+, −, +) mutation, which was shown to have a lower electron flux, as a progenitor for a genetic study.174 The β-Y98H variant would not grow well under diazotrophic conditions in the presence of 2.5% C2H2, because N2 reduction by the low flux NifDK protein would be strongly inhibited by C2H2. The goal was to screen for a mutation that would remove the acetylene sensitivity while maintaining similar levels of N2 reduction activity. After analysis, it was found that in addition to the β-Y98H mutation, an α-G69S (+, −, +) mutation would allow for cells to grow under N2 fixing condition in the presence of C2H2. A separate protein was then generated that only possessed the α-G69S mutation. The Michaelis–Menten parameters of the α-G69S NifDK protein for C2H2, N2, and proton reduction were similar to those of the wild-type enzyme, indicating that this mutation did not affect nitrogenase activity. In contrast, the affinity for C2H2 decreased for α-G69S NifDK (Km = 14.2 kPa) compared to wild-type enzyme (Km = ~0.71 kPa), and the variant protein now showed a competitive inhibition of N2 reduction in the presence of C2H2, instead of a noncompetitive inhibition.174 The rationale for these observations was that the mutation of α-G69 removed the high-affinity C2H2 binding site through steric changes nearby the M-cluster. However, the authors proposed that α-V70 was the actual residue that blocked the substrate binding site, due to the closer proximity to the cofactor. They also identified that α-R96 is positioned nearby α-V70 and is likely to be involved in determining access to the binding site. A separate study concluded that using the α-G69S NifDK protein, two substrate binding sites could be better distinguished on the α-V70 (0, −, +) face of the cluster.175 One site is denoted as a “high-affinity site” for acetylene binding that may also accommodate CO but not N2, N2O, and N3−. The other site is designated the “low-affinity” acetylene-binding site that also binds CO, N2, N2O, and N3−. Substitutions of the α-R96 (+, −, 0) and α-V70 (0, −, +) positions in NifDK were then explored for their effects on the access and binding of substrates and inhibitors.176,180 In 2001, multiple mutations of α-R96 were reported, and it was shown that the EPR spectra of all of the NifDK variants were essentially the same as the spectrum of the wild-type enzyme.180 This result was used to conclude that the substitution had little to no effect on the electron properties of the M-cluster, and as such, the authors selected the α-R96L NifDK variant as a representative example to further study. When high concentrations of C2H2 were added to the protein, a new EPR signal appeared in a 70% acetylene atmosphere with g-values of 4.5, 3.5, and <2.0, constituting 40% of the total spin (Figure 11). The addition of N2 or CO did not affect the appearance of this signal, which is totally absent in the wild-type enzyme. It was found that addition of CN− to α-R96L NifDK would yield a new paramagnetic species with a g-value at 4.5, that also was absent from the wild-type system. Pulsed electron nuclear double resonance (ENDOR) experiments with 13C-enriched acetylene and cyanide showed that CN− definitively binds to the M-cluster, while it was inconclusive if C2H2 did as well, but in both cases the new EPR signals were derived from the M- as opposed to the P-cluster.180 These results led to the conclusion that α-R96 must play a steric “gatekeeping” role during catalysis. The change in size of the α-R96 residue presumably opened the active site allowing the small molecules to bind on or near the M-cluster. However, this finding should be taken with some scrutiny: the concentrations required to affect a change in the EPR spectrum are far beyond what is required under usual catalytic conditions, and other molecules, such as N2, CO, and N3− do not cause perturbations in the EPR spectrum of α-R96L NifDK. Interestingly, the α-R96 variants do not product ethane, in contrast to α-Q191 and α-H195 variants.205 The α-R96 variants do all have different affinities and rates for the reduction of C2H2, some similar to wild-type enzyme while α-R96L NifDK has a 3-fold higher affinity, so it is unclear to what extent physiologically relevant conclusions can be made. In 2002, a follow up to the mutagenesis study of α-G69 NifDK was reported with the substitution of α-V70 (0, −, +) for alanine.176 The α-V70A NifDK protein was able to grow under diazotrophic conditions, indicating that N2 reduction was unaffected by the substitution. The active site access of the variant compared to wild-type enzyme was assessed through the introduction of the water-soluble acetylene derivate propargyl alcohol (HC≡C–CH2–OH). When added to the growth media, the wild-type enzyme was unaffected while the α-V70A protein was unable to grow, presumably due to a larger substrate pocket that allows the larger alcohol in. Propargyl alcohol was a potent inhibitor of α-V70A NifDK with respect to proton and N2 reduction, but direct reduction of propargyl alcohol resulted in propene (H2C = CH–CH3; 4-electron product) as a minor product, as well as allyl alcohol (HC = C–CH2–OH; 2-electron product) as the major product. Propyne (HC≡C–CH3) was also found to serve as a substrate for α-V70A NifDK, producing the 2-electron product propene, in contrast to the wild-type enzyme where propyne is a poor substrate.55,59 Additionally, α-V70A NifDK was unable to produce ethane from the reduction of acetylene.176 Dean, Seefeldt, and co-workers then published a study of the NifDK protein, using the smaller (α-V70A) and larger (α-V70I) side chain variants to study the effect the substitutions at α-V70 had on N2 reduction.177 The ability of the α-V70I variant to reduce N2 was decreased by ~70% compared to wild-type enzyme while electron flux was diverted toward proton reduction, with a concomitant increase in the Km values for N2 reduction from Km = 0.1 atm to >1.5 atm. Acetylene reduction was analogously affected with a substantial decrease in C2H4 production, and an increase to both proton reduction and the Km value compared to wild-type NifDK. These results were rationalized by stating that the longer side chain of isoleucine must impinge on the “catalytic face” of the M-cluster where N2 and C2H2 bind, including Fe2, Fe3, Fe6, and Fe7 from the crystallographic labeling scheme (Figure 7). Substitution of glycine at position α−70 resulted in a nitrogenase protein (α-V70G) that is unable to react with N2 at all, but maintains activity with acetylene and larger carbon-based substrates.178 The lack of N2 reduction with the smallest side chain variant was attributed to the inability to access higher redox states of the M-cluster, but this hypothesis was not further tested. A double mutation consisting of α-V70A and α-Q191A was also generated, and the enlarged substrate pocket allowed for longer chain alkynes, such as 2-butyne (H3C–C≡C–CH3), to be reduced near the proposed reactive site at Fe6 on the M-cluster.178 Undoubtedly, the second coordination sphere environment is important for determining substrate access through steric effects, but with the drastic changes that sometimes occur “globally” in variant proteins, it is logical to be cautious in extending insights gained from the variants to the native system. The α-V70I and α-V70A NifDK proteins have been used to trap and study a variety of intermediate states with a battery of spectroscopic and computational techniques, the like of which have been described at length elsewhere52,60,206 and will not be further discussed in this review. Recently, a crystal structure was published by Ribbe, Hu, and co-workers that provides some structural insight into substrate binding on wild-type NifDK.91 Previous attempts at trapping intermediates of nitrogenase have employed single-point mutations in combination with rapid freeze-quench techniques that can capture transient species.52,60,206 The strategy used in the recent crystallographic study involved isolation of NifDK in an anaerobic environment that did not contain the exogenous reductant dithionite (S2O42).91 Previous work has shown that when dithionite and other sulfur sources are stringently excluded from in vitro maturation assays, an M-cluster precursor called the L*-cluster ([8Fe-8S–C]) can be isolated on NifB that lacks a μ2-sulfide ligand that it otherwise binds in the presence of dithionite.133 This sulfur-deficient cluster cannot be matured into the active M-cluster unless a source of SO32− is added, which is usually provided by the decomposition of dithionite. Drawing from this inspiration, it was reported that when NifDK was anaerobically purified under a N2 atmosphere in the absence of dithionite (denoted NifDK*), the resultant protein was initially unreactive toward substrate reduction.91 However, reactivity could be restored upon the addition of dithionite. When cells expressing NifDK* were grown in the presence of15N2, the subsequent protein was reported to have15N2 bound, as acid quenching samples of NifDK* yielded labeled N2 by gas chromatography mass spectrometry (GC-MS) analysis. The labeled N2 ligand was not observed when dithionite was included as part of the purification protocol. Isolated NifDK* was then crystallized, and a structure was solved that contained M-clusters with additional electron density between central Fe ions of the cluster (Figure 12).91 Not only that, the two M-cluster sites within the heterotetrametric NifDK each had different sets of electron density; at ‘Site 1’, this density appeared in the usual place of S2B (0, 0, +), and at ‘Site 2’, density was observed in the location of S3A (0, +, −) and S5A (0, −, −). On the basis of anomalous sulfur density, it was concluded that Site 1 lacked a sulfur atom at the S2B position, and Site 2 lacked sulfur atoms at both S3A and S5A. During refinement, best fits of the data resulted in N2 molecules displacing the respective sulfur atoms at the two different M-cluster sites. At Site 1 (Figure 12A), the N2 unit appears in a pseudo μ1,2-bridging mode with N6A sitting 1.8 Å from Fe2 and N6B being 2.3 Å from Fe6. N6A is also within 2.9 Å of α-H195 (−, 0, +), indicative of a potential hydrogen-bonding interaction. A stabilizing interaction between the putative N2 unit and α-H195 would be consistent with observations from the mutagenesis studies described in this review.139,182–185,187 At Site 2 (Figure 12B), the N2 units are bound in asymmetric μ1,1-bridging modes where the proximal N atom is within 1.8 Å of one Fe ion, and 2.1 Å from the other Fe (with the Fe pairs Fe4/Fe5 for the “S3A position” and Fe7/Fe3 for the “S5A position”, respectively).91 The distal N atom at the S3A position is 2.9–3.4 Å from the backbone amide groups of α-G356 (+, +, 0) and α-G357, and at the S5A position, the distal N is within 3.2 Å of the side chain of α-R96 (+, −, 0) and a water molecule, also suggesting potential hydrogen-bonding interactions. These positions have not been previously proposed as substrate binding sites, which is an interesting finding, although in crystal structures of V-nitrogenase, an unusual carbonate (CO32−) ligand is found to displace the equivalent S3A atom.92–95 Additionally, in both Site 1 and 2 of sulfur-deficient NifDK*, the R-homocitrate ligand appears to be bound to the Mo center through one O atom exclusively; at Site 1 the O atom from the −OH group is 2.7 Å from Mo and the −COO− group is closer, while at Site 2 the opposite is observed.91 This could potentially indicate protonation, which may elongate the refined Mo–O distances, consistent with the proposed role R-homocitrate may play in proton transfer, but additional evidence is required to validate the protonation states. Strikingly, when NifDK* is put under turnover conditions in the presence of dithionite and then recrystallized (PDB ID 6VXT), the cofactor returns to the normal resting state of NifDK, indicating that the displacement of sulfur is not deleterious but rather reversible.91 The resolution of the crystallographic data for NifDK* does not allow for the absolute determination of the N2 unit as a reduced species or otherwise;207–209 however, the assignment of the N2 unit is supported by available structural data, and it would further validate the importance of residues such as α-H195 and α-R96, as well as the “catalytic face” of the M-cluster, while putting forward sulfur displacement as a critical mechanistic necessity for substrate reduction by Mo-nitrogenase. A recent study by Hu, Ribbe and coworkers723 confirmed binding of N2 to NifDK* in a catalytically competent conformation, showing formation of C2H3D in the C2H2 reduction assay (an equivalent to HD formation in the absence of C2H2) of NifDK* where only D2 was supplied instead of a D2/N2 mixture, appearance of new S = 1/2 features in the EPR spectrum of NifDK* concomitant with a decreased intensity of the M-cluster-associated S = 3/2 signal, and release of15NH3/NH4+ upon turnover of the15N2-bound NifDK*. In addition, this report revealed that product release only occurred via displacement by sulfite-derived belt-sulfide, and that catalysis involved all belt sulfur locations that were shown to be labeled with selenite-derived selenides in a XAS pulse chase experiment. These observations provide support to a previously proposed mechanism involving a stepwise reduction of N2 at position S3A, S2B, and S5A that occurs asynchronously in the two αβ-dimers of NifDK.91 Further crystallographic, biochemical, spectroscopic, and computational work will be required to explore and validate these unique findings. 2.5.2. Inhibition by and Reduction of Carbon Monoxide. Carbon monoxide has long been characterized as an inhibitor of nitrogenase reactivity in the Mo-dependent systems that affects the reduction of all substrates apart from protons.52,100,200,210 Mo-nitrogenase has multiple CO-related states that have been spectroscopically characterized under different conditions and concentrations of CO.52,62 Under low electron flux conditions (1:5 Fe:MoFe protein) and in the presence of low concentration of CO (<1% in argon), the typical S = 3/2 EPR signal associated with the M-cluster converts to an S = 1/2 signal with g values of 2.09, 1.97, and 1.93 (designated the “lo-CO state”), whereas under high concentrations of CO (>50% in argon) a new S = 1/2 signal arises (designated the “hi-CO state”) with g = 2.17 and 2.06 (Figure 13).184,202,203,211–213 Under higher electron flux conditions (1:1 Fe:MoFe protein) and high concentrations of CO, a third state known as “hi(5)-CO” was observed in wild-type NifDK, with g values of its EPR signature at 5.78, 5.15, and 2.7 (Figure 13).181,184,211 A low intensity EPR signal was observed for VnfDGK in the presence of 1 atm of CO with g values at 2.09, 1.99, and 1.91, and the intensity of this signal could be increased with more potent reducing agents.214,215 By comparison to NifDK, the EPR signal in VnfDGK was assigned to the lo-CO form of the enzyme. In the presence of strong Europium-based reducing agents and 2.6 atm of CO, a new set of EPR signals were observed that could be described as the superposition of the lo-CO state, and a newly identified species with g-values at 2.13, 2.01, and 1.97.215,216 Although not identical, this new signal was assumed to be analogous to the hi-CO state of NifDK. No equivalent of the hi(5)-CO state has been reported for the VFe protein. As the crystallographic characterization of CO-bound states was not reported until 2014,89 many of the earlier studies employed competitive substrate reduction experiments, where separate N2, C2H2, and proton reduction assays were compared in the presence and absence of CO.51,52 The behavior of nitrogenase in the presence of multiple substrates could then be analyzed and extrapolated to provide insight into CO-bound states of the M-cluster. Additionally, a series of spectroscopic and computational studies revealed EPR signals associated with CO binding to Mo-nitrogenase. As described above, an S = 1/2 spin species trapped under low concentrations of CO was designated “lo-CO”, and at higher CO concentrations, another S = 1/2 signal would appear that was assigned “hi-CO”.202,211,217–221 When the initial characterization of V-dependent nitrogenase was undertaken, it was found that CO inhibited reactivity in a similar manner to Mo-nitrogenase, so it was reasonable to anticipate analogous EPR signals in V-nitrogenase. Hales and co-workers constructed a hybrid nitrogenase that used the polypeptide of V-nitrogenase from A. vinelandii and the M-cluster from the A. vindelandii Mo-nitrogenase (hereafter labeled M-VFe protein).222 It was found that the hybrid M-VFe protein had an activity profile more like the native V-nitrogenase. However, in the presence of CO, no EPR signals analogous to the lo-CO or hi-CO signals of Mo-nitrogenase could be observed, although the M-VFe protein carried the same cofactor as the native Mo-nitrogenase. This result was puzzling, but was a strong indication that second and outer coordination sphere effects played a pivotal role in determining the properties of nitrogenase.51,222 Later, it was shown by Ribbe and co-workers that Av V-nitrogenase (and to a lesser degree, wild-type Av Mo-nitrogenase) could catalyze the coupling of gaseous CO into longer chain hydrocarbon products following Fischer–Tropsch-like chemistry, but with an observed decrease of ~75% in the specific activity for proton reduction by the V-nitrogenase.223–225 This finding further expanded the known capabilities of the Mo- and V-nitrogenases, but still lacked a comprehensive structural understanding of CO activity. With the advent of crystallographically characterized, CO-bound intermediates, insight into structural features of CO binding to nitrogenase could be obtained.89,94,95,226 In 2014, Spatzal, Rees, and co-workers published the first structure of CO binding to the M-cluster of NifDK (designated NifDK-CO), which corresponded to an inhibited state of the enzyme (Figure 14A).89 The CO molecule is ligated between Fe2 and Fe6 of the M-cluster in a μ2-binding mode, apparently displacing the μ2-S2B sulfur atom that is usually found in resting state structures of NifDK.61,66,151 The CO ligand is bound symmetrically between the two Fe ions, with Fe–CO distances of 1.86 Å to both metal centers, and the oxygen atom of the CO moiety within close proximity to the side chains of residues α-H195 (2.8 Å) and α-V70 (3.4 Å). The interaction between α-H195 and CO is within the range of a hydrogen bond, which would affect the stability of the CO moiety, while nonbonding interactions from α-V70 may provide stability. When crystals of NifDK-CO were dissolved into buffer in the absence of exogenous CO, the S2B atom and acetylene reduction activity were quantitatively recovered.89 EPR spectra of NifDK-CO in the solution state, obtained under analogous conditions to crystallography, demonstrated the appearance of an S = 1/2 signal previously assigned to the lo-CO species, connecting observation derived from spectroscopy and structural biology.214 Recently, Spatzal, Rees, and co-workers reported another CO-bound crystal structure of NifDK (Figure 14B), but with two CO ligands (designated NifDK-(2CO)) as opposed to one.226 This species was isolated when crystals of NifDK-CO were pressurized with CO at 5.4 atm. The first CO ligand (“μCO”) is found in a bridging position between Fe2 and Fe6 of the M-cluster in a μ2-binding mode analogous to that in NifDK-CO, with Fe–CO distances of 1.93 and 1.92 Å, respectively. The second CO ligand (“tCO”) is bound in a terminal mode to Fe6, with an Fe–CO distance of 2.03 Å. Binding of tCO causes the Fe6–C distance to slightly elongate from 2.01 Å in NifDK-CO to 2.06 Å in NifDK-(2CO).89,226 The oxygen atom of tCO may interact with the amide N of the α-Q191 residue (~3.3 Å) and a carboxylic acid group of R-homocitrate (~2.7 Å), while μCO maintains analogous interactions with the side chains of α-V70 and α-H195, in a similar manner as in NifDK-CO. Compared to μCO that is modeled with 100% occupancy, tCO is modeled with only 50% occupancy, likely because the binding of this ligand is relatively weak. Parallel EPR spectra of the solution state and crystal slurry of NifDK-(2CO) were subsequently used to assign μCO and tCO to the lo- and hi-CO states, respectively.226 Analogous crystal structures were solved by Einsle and co-workers of the V-dependent nitrogenase with one (VnfDGK-CO) and two (VnfDGK-(2CO) CO molecules bound (Figure 15).94,95 In the 1.0-Å resolution structure of VnfDGK-CO, the CO molecule is bound in a μ2-bridging mode between Fe2 and Fe6, much like that found in the lo-CO state of NifDK-CO, but with a slight asymmetry (Fe2–CO = 2.03 Å, Fe6–CO = 1.94 Å).94 The μCO ligand has similar interactions with local amino acid side chains, being within 2.9 Å of α-H180 and 3.6 Å of α-V57 (analogous to α-H195 and α-V70 in NifDK), respectively. Unlike NifDK-CO, when VnfDGK-CO crystals were used for activity assays in the absence of exogenous CO, the typical S2B ligand was not found between Fe2 and Fe6. Instead, a monatomic ligand interacting with the nearby α-Q176 residue was observed and assigned as a putative μ2–OH species. The authors proposed that this oxygen-based ligand might result from reduction of a previously bound CO ligand.94 When crystals of VnfDGK-CO were pressurized with 1.5 atm of CO for 1 min, a 1.05 Å crystal structure of the VnfDGK-(2CO) was obtained (Figure 15B).95 The μCO ligand in VnfDGK-(2CO) is positioned in the same manner as that in VnfDGK-CO, and the terminal CO ligand is bound to Fe6 with an occupancy of 50%, analogous to that observed in the structure of NifDK-(2CO); however, the tCO in VnfDGK-(2CO) has an Fe6–tCO distance of 1.89 Å, compared to 2.03 Å in NifDK.95,226 The oxygen atom of tCO in VnfDGK-(2CO) is also in close proximity to the R-homocitrate ligand (~2.9 Å) and the α−176Gln residue (~3.1 Å). Despite the high degree of similarity in active site for the two structures, the tCO ligand appears to be more strongly bound in VnfDGK-(2CO) versus NifDK-(2CO), which may align with the difference in reactivity observed for both systems.51 However, there are strong indications that the lo-CO and not the hi-CO state of V-nitrogenase is catalytically competent for CO reduction.215,216 A parallel set of EPR measurements using solution and crystal slurry samples of VnfDGK-(2CO) was not reported by Einsle and co-workers,95 so the crystallography could not be directly connected to the available spectroscopic characterization of the lo- and hi-CO states. Instead, in situ Fourier-transform infrared (FTIR) difference spectroscopy was conducted using a film of VnfDGK in the presence of N2 and CO isotopes.95 Supporting the crystallographic assignment, isotopically sensitive features at 1931 and 1888 cm–1 were assigned to t12CO and t13CO, respectively. Because of strong background signals at lower frequencies, a vibration for the μ12CO could not be assigned unambiguously, but a putative signal was identified at 1720 cm–1. More work will be needed in the future to fully understand the structure–function relationships gleaned from these structural studies. On the basis of the crystallographic information, what can be said is that α-Q191, α-H195, and α-V70 in NifDK (and equivalent analogs in VnfDGK) are residues that can interact with the catalytic cofactor when CO molecules are bound. The α-H195 residue is in hydrogen-bonding distance to the μCO ligand and has been proposed to be generally involved in proton delivery during substrate turnover.227–229 The α-Q191 residue is within an appropriate distance from the tCO ligand to potentially interact with this ligand; additionally it helps stabilize the R-homocitrate ligand that binds to the Mo or V ion of the catalytic cofactor.95,226 As stated previously, the hi-CO state does not appear to be competent for CO catalysis,215,216 thus binding to tCO may not be relevant, although the equivalent residue in VnfDGK, α-Q176, has been shown to stabilize a putative N2 reduction intermediate through a hydrogen-bonding interaction, which further reinforces the importance of this residue.93 These details regarding CO-bound species provide a measure of perspective that was previously inaccessible and allows for a retrospective look on experiments conducted to assess the structural basis of CO inhibition within nitrogenase. 2.5.2.1. Influence of Point Mutations on the Reaction with CO. The research groups of Dean and Newton have generated a multitude of single-point mutants of NifDK that have been used to study substrate interactions with the enzyme, including CO. Many of these studies yield overlapping information regarding the properties of NifDK variants, including insights gleaned from experiments with crude extracts of A. vinelandii cells and purified variant proteins, so it can be difficult to comprehensively describe the results. Instead of addressing every possible variant reported, the effects of mutations to the α-H195 and α-Q191 residues on CO reactivity will be the primary focus of our discussion below, as well as relevant comparisons to other variants, as necessary. As described earlier, the site-directed mutagenesis in NifDK of α-H195 (−, 0, +) and α-Q191 (+, +, +) to α-H195N and α-Q191K were reported for A. vinelandii.182 The crude extracts of variant cells were individually assayed in an atmosphere of 10% acetylene (C2H2), demonstrating that the variant NifDK proteins were much less active than wild-type enzyme for the generation of C2H4, but they also produced some amount of ethane (C2H6), whereas wild-type NifDK does not. This raised interest in these variants, as crystallographic evidence would not be available for several years, and the pattern of reactivity was similar to that found for V-nitrogenases.51 When C2H2 reduction assays were carried out, the α-H195N variant was less sensitive to inhibition by CO, showing 50% activity in the presence of 0.5% CO, compared to wild-type NifDK that showed the same decrease in activity already at 0.02% CO. It is important to note that these effects were observed with only the crude extracts, as the protein could not be purified at that time. EPR spectra of whole cells indicated that the S = 3/2 M-cluster signal was still present in the variant, albeit with axial perturbations, and at a low intensity (~6% compared to wild-type enzyme).182,185 Several years later, five additional variants of α-H195 were reported.187 Similar to the previous reports of α-H195N,182,185 the crude extracts of the tyrosine, threonine, and glycine variants were not particularly active, but the glutamine and leucine variants expressed slightly higher activity, with α-H195Q NifDK maintaining ~70% the activity of crude extracts of wild-type enzyme.187 Whole cell EPR of all variants had similar line shapes to wild-type enzyme, but the glycine and leucine variants had much lower intensities of the S = 3/2 signal (≤10%), whereas α-H195Q NifDK had comparable intensity to native enzyme, with slight perturbations of the g-values. The proton reduction activities in the presence and absence of CO were assayed, and it was demonstrated that the asparagine, glutamine, and leucine variants were relatively insensitive to CO, but α-H195T NifDK was sensitive to the presence of CO, with ~50% reduction of activity. The α-H195Q NifDK protein was further purified for more accurate comparisons to wild-type enzyme. It was found that CO was a noncompetitive inhibitor of C2H2 and N2 reduction for both proteins, but CO had a more potent inhibition of α-H195Q NifDK compared to wild-type enzyme. The authors suggested that α-H195 could be involved in steric interactions, as the much smaller glutamine residue would be unable to fill the same space as a histidine side chain, leaving room within the active site. Additionally, it was noted that the glutamine residue could still provide a hydrogen-bonding interaction to the S2B sulfur and, although the differential electronic effect of the histidine- and glutamine-derived interactions could modulate the redox potential of the M-cluster, it was not sufficient to completely change the activity profile. We now know that the S2B site is the location of the μCO ligand in the lo-CO state,89 and by comparing a 2.5 Å crystal structure of α-H195Q NifDK184 to the structure of the wild-type NifDK, we know that glutamine and histidine residues at position 195 have a similar distance to the S2B sulfur (Figure 16). This could rationalize why some of the other α-H195 variants do not generally exhibit any sensitivity to CO, that is, the responsible residues are simply too far from the μCO site to strongly interact with CO. The crystal structure also shows that the substitution of histidine with glutamine greatly destabilizes the local hydrogen-bonding network surrounding the M-cluster, complicating the interpretation. Moreover, different substitutions at α-H195 result in protein variants with CO-insensitive proton reduction and CO-inhibited acetylene reduction,187 leading to the proposal of multiple CO binding site at the M-cluster. In pursuit of this idea, Newton, Fisher, and co-workers used the α-H195Q and α-H195N NifDK proteins in reactivity experiments with HCN, CN−, and N3− in combination with CO.183,188,191 For the variants, CO was found to remove the inhibitory effects of CN− and N3−, implying that these classic nitrogenase inhibitors all bind to the same sites, albeit with different affinities. In this context, it should be noted that other GPMs like formate dehydrogenase (Section 4) and CO dehydrogenase (Section 5), CN− and N3− inhibit activity, which highlights a functional connection between the low-valent metalloenzymes discussed herein. With respect to the binding site(s) of CO in nitrogenase, recent structural analysis shows that a second CO binds terminally to Fe6 of the M-cluster, and α-H195 is located near Fe2 and is generally unperturbed going from the NifDK-CO structure to the NifDK-(2CO) structure.89,226 It may be possible for CO molecules (or other inhibitors) to bind to the Fe2 site on the M-cluster in α-H195 variants due to the change in the relative steric protection provided by the imidazole ring, underlining the importance of second coordination sphere effects. Additional structural information with other inhibitory molecules would be important for better understanding these complex effects. The α-Q191 (+, +, +) residue of NifDK has been a target of mutagenic studies for as long as α-H195.182,185 Crude extracts of α-Q191E and α-Q191K NifDK were assayed with C2H2 and in the presence of 0.02% CO, activities were inhibited by 50% and 75%, respectively.185 The α-Q191K NifDK was then further purified, but the α-Q191E variant was not stable enough to sustain such a treatment. The purified α-Q191K NifDK protein exhibited an EPR spectrum that had similar line shape to that of the wild-type NifDK, but with perturbations to the g-values, suggesting that the M-cluster was bound to the protein similarly in both proteins. The introduction of 3% CO into substrate reduction assays showed that electron flux for reactions with α-Q191K NifDK decreased by about 70%, 40%, and 50% for N2, C2H2, and proton reduction, respectively, compared to in the absence of CO.185 This is an unusual finding considering that wild-type NifDK generally redistributes electron equivalents from the reduction of N2 and C2H2 toward proton reduction in the presence of CO.51,185 Much like for the α-H195 variants, the purified α-Q191K NifDK was also assayed for proton reduction in the presence of both CN− and CO.191 The α-Q191K variant showed a mild inhibition of proton reduction in the presence of CN−, and subsequent introduction of a 10% CO atmosphere further decreased the proton reduction activity. In contrast, the wild-type and α-H195 variant proteins show inhibition of total electron flux by CN− but this activity is restored in the presence of CO. The result agrees with the previous observation of CO-derived inhibition of the electron flux for α-Q191K NifDK,185 and additionally supports that α-Q191 likely can interact with both substrates and inhibitors. It is now possible to glean additional insight from the crystal structures of NifDK-CO and NifDK-(2CO). Figure 14 shows how the α-Q191 residue of NifDK in both structures is bent away from the M-cluster, but the residue is in proximity to several molecules, 3.6 Å from an H2O molecule, 3.3 Å from the tCO ligand and 2.9 Å from a terminal carboxylate of the R-homocitrate ligand.89,226 Similar distances are also found for the analogous α-Q176 residue in the CO-bound structures of the V-nitrogenase (Figure 15).94,95 In addition, it has been noted that the α-Q176 residue can swing in to interact with the V-cluster at the S2B position, so the residue is less static.93 The closest and likely strongest interaction is between α-Q191 and R-homocitrate, so when glutamine is mutated to lysine, the side chain is extended by an additional carbon atom, which may introduce steric misalignment of the Lys residue in the active site. Lysine may clash with the tCO binding site, preventing substrate molecules from properly accessing the M-cluster. This would be consistent with an observation made by Hales, Newton, and co-workers that α-Q191K NifDK does not generate a hi-CO EPR signal and, by inference, does not bind tCO; instead, it only shows the lo-CO signal even under 100% CO atmosphere.181 2.5.2.2. Role of the Homocitrate Ligand. While a physical blockage by lysine may be the source of the aberrant properties of α-Q191K NifDK, another important factor is that a disruption of the α-Q191–homocitrate interaction may destabilize the hydrogen-bonding network within the active site M-cluster, which is thought to be critical for proton delivery during catalysis.102,181,229–232 When the nif V gene is deleted in a nitrogenase-expressing organism, the resulting Δnif V NifDK protein has an M-cluster with citrate in place of R-homocitrate.196,197 In the presence of 3% CO, crude extracts of Δnif V NifDK from K. pneumoniae (Kp) demonstrated a ~50% decrease of proton reduction, similar to what was observed for α-Q191K NifDK.185,196,233 Further investigation showed that CO would decrease the electron flow through Kp Δnif V NifDK, and uncouple MgATP hydrolysis from electron transfer with a corresponding decrease to the observed proton reduction.233 This analysis by Dixon and co-workers also indicated that CO was unlikely to be reduced by the variant Mo-nitrogenase from K. pneumoniae. Analogous Δnif V NifDK variants were also generated in A. vinelandii (Av), showing a range of activities in the presence of CO.181,234,235 Newton, Hales, and co-workers showed that proton reduction by Av Δnif V NifDK was inhibited by ~40% in the presence of CO, a similar magnitude to that of Kp Δnif V NfiDK, and that Av Δnif V NifDK was capable of adopting the lo-CO, hi-CO, and hi(5)-CO states.181,234 However, CO reduction by a Δnif V NifDK protein was not reported until 2021 by Ribbe, Hu, and coworkers.235 This protein, compared to wild-type NifDK, expressed lower activities for all substrates, including CO. It was reported that the Mo content of the in vivo isolated Av Δnif V NifDK was only ~30% of that in the wild-type protein, which may in part explain the observed low reactivity of the protein. To improve cofactor incorporation, an in vitro reconstitution of Δnif V NifDK was carried out, which increased the M-cluster content and rendered the in vitro reconstituted variant higher in activity but similar in reactivity pattern as compared to its in vivo isolated counterpart. In the presence of CO, Δnif V NifDK diverts more electron flux from proton reduction toward hydrocarbon formation while still maintaining lower specific activity than the wild-type enzyme. These observations were not made previously for the A. vinelandii variant because it was not reported that Av NifDK could reduce CO to hydrocarbons until 2011 when scaled up reaction conditions were used to detect low-yielding products for V-nitrogenase.223 An analogous Δnif V variant was also generated for V-nitrogenase, showing that citrate also binds to the V-cluster.236 However, unlike the scenario in the Mo-dependent system, the Δnif V VnfDGK protein produces a decreased amount of longer chain hydrocarbon products compared to wild-type VnfDGK, favoring a slight increase in proton reduction. Initially, it was unclear what role the nif V gene product played in nitrogenase biosynthesis, but in 1990, Burris and co-workers reported that the organic ligand bound to the M-cluster was identified as citrate, which was subsequently confirmed in 2002 by Lawson and co-workers, who reported a 1.9 Å resolution crystal structure of the Kp Δnif V NifDK.197,237 The structure of the Δnif V protein is highly similar to wild-type NifDK, but while R-homocitrate can form hydrogen-bonding interactions with the backbone amide of α-I423 (~2.8 Å), the citrate ligand in Δnif V NifDK is further away from α-I423 (~3.7 Å) and requires H2O molecules to facilitate an interaction.237 This would potentially leave the citrate ligand loosely bound to the protein scaffold, disrupting the hydrogen-bonding network necessary for proton delivery to the M-cluster. An additional ramification would be that the loss of second coordination sphere interactions may also modulate the redox properties of the catalytic cofactor, similar to what was proposed for certain α-H195 NifDK variants (vide infra). This also means that it might be rather difficult to deconvolute the effects of α-Q191 substitutions from those of Δnif V knockout proteins, because the Gln residue is involved in stabilizing interactions with the citrate ligand. However, there are distinct differences. The α-Q191K NifDK protein only supports the lo-CO EPR signal, while Av Δnif V NifDK supports all three CO-based EPR signals. In fact, Newton and co-workers expanded investigations of CO on the reactivity of NifDK by introducing mutations to residues α-R96 (+, −, 0), α-R277 (−, +, +), and α-R359 (0, +, −).181,186 In this study, CO-bound states were generated for a series of NifDK variants and EPR spectra were collected along with data from proton reduction assays to correlate the spectroscopic signals to catalytic activity. Wild-type NifDK, as well as α-R96Q NifDK, and Av Δnif V NifDK all showed the lo-CO, hi-CO, and hi(5)-CO states. However, the variants α-R96K, α-R277C, and α-R359K adopted only the lo-CO and hi(5)-CO states, but not the hi-CO state. Interestingly, α-R277H only showed the hi(5)-CO signal. Inhibition of proton reduction by CO was also equally dynamic, with instances of no inhibition in α-R96Q but ~50% inhibition with the α-R96K substitution, among others.181 Based on these observations, the authors concluded that the CO-related EPR signals and proton reduction activity could not be correlated, noting that the single-point mutations did not introduce additional CO-binding sites.181 The α-R277C NifDK variant was slightly different, in that it demonstrated a 70% increase in proton reduction in the presence of 10% CO compared to the wild-type enzyme.186 With higher electron flux conditions, the proton reduction activity was enhanced, but dropping tchhe flux resulted in a lack of inhibition of proton reduction, although this reaction occurred at slower rates, similar to that observed in the case of the wild-type NifDK. It was proposed that CO binding to α-R277C NifDK affected catalytic proton transfer and additionally prevented the M-cluster from accessing more reduced states (i.e., E3 and E4 from the Lowe-Thorneley cycle). Instead, the variant may divert all the electron equivalents into the E2 state which subsequently generates H2.186 It is unclear why CO binding would have such a different effect on the reactivity of the α-R277C variant as compared to the others, and more work is needed to gain further understanding of CO reactivity. 2.5.2.3. Reduction of Carbon Monoxide. Attempts to stimulate CO reduction in Mo-nitrogenase have been carried out using combinations of point mutations to α-V70 (0, −, +), α-R96 (+, −, 0), α-H195 (−, 0, +), and α-Q191 (+, +, +).179 Modifications of α-V70 in Av NifDK have been shown to increase the substrate scope of the enzyme by changing the steric environment around the M-cluster.52,194 The α-V70G and α-V70A NifDK variants, were reported to react with CO, producing hydrocarbons ethylene (C2H4), propene (C3H6), ethane (C2H6), and propane (C3H8) in a ratio of 7:4:2:1.179 These rates are much lower than those reported for the V-nitrogenase223 and not directly compared with those for the wild-type Mo-nitrogenase under the same experimental conditions. Double mutations were then incorporated to assess the effect of α-R96, α-H195, and α-Q191 on the α-V70-mediated CO reactivity.179 The α-V70A/α-R96H and α-V70A/α-Q191A NifDK variants show a different product pattern than the single substitutions, with similar values for all C2 and C3 products, as well as the inclusion of trace amounts of methane (CH4). The α-V70A/α-Q191A NifDK, though, produces twice as much C3H6 than the α-V70A/α-R96H variant. Additionally, α-V70A/α-H195Q was shown to drastically decrease product formation compared to the single α-V70A NifDK variant. It is difficult to make definitive statements about the exact effect of point mutations in this case because of the interconnected nature of the second coordination sphere effects. What can be said is that the α-V70 residue seems important for regulating the size of substrates and products, consistent with previous observations.52 The size reduction of the α−70 residue side chain presumably enlarged the active site to accommodate multiple CO molecules, and by extension, additional modification of α-R96 and α-Q191 further expanded the active site to favor the generation of longer chain hydrocarbons. However, the loss of activity in the α-V70A/α-H195Q NifDK protein shows that the size of the cavity is not the only factor governing reactivity toward CO. The α-Q191 and α-H195 residues are known to be involved in proton transfer,102,181,229–232 but the present data may reflect a more direct role of α-H195 in protonation of reduced, CO-coupled species whereas α-Q191 may play more of a structural role. It may also be possible that substitution of α-Q191 could affect the rate of protonation and allow for intermediate species to accumulate, increasing the hydrocarbon length. Rigorous studies involving additional mutations that vary the size and proton donor ability of the side chains may be necessary to shed further light on the “steric vs protonation” arguments for the reaction of CO with nitrogenase. It is also possible that yet unidentified factors, such as electron transfer rate or allosteric changes, may play a role. 2.5.3. Reduction of Carbon Dioxide. The triatomic carbon dioxide (CO2) is a highly oxidized small molecule that was found to react with Mo-nitrogenase in 1995.238 CO was initially found as the product of the reaction, as the concentration of CO would increase only in the presence of NifH and MgATP, indicating it was a nitrogenase-dependent phenomenon. However, the exact substrate form, CO2, CO32−, or HCO3−, was unclear because of the equilibrium between CO2 and bicarbonate (HCO3−) in solution. It was later reported by Ribbe and coworkers that NifDK was capable of generating CH4 and C2H4 from from CO2, although the rate of hydrocarbon formation was an order of magnitude slower than CO formation.239 Interestingly, V-nitrogenase was found to reduce CO2 to CO in H2O buffers, but in deuterated buffers, deuterated hydrocarbons (CD4, C2D4, C2D6) were observed while NifDK was unable to generate deuterated products under these same conditions. For both Mo- and V-nitrogenase, the specific activity of CO evolution from CO2 increased in the deuterated buffer system resulting in an inverse kinetic isotope effect (KIE).239 This could indicate that the formation of CO is carried out through similar mechanisms in both nitrogenases, but the change in hydrocarbon product profile between the two nitrogenases is reflective of something more complex that requires further investigation. Subsequently, Mo-nitrogenase variants α-V70A and α-H195Q were shown to convert CO2 into methane (CH4), confirmed by use of13C-labeled bicarbonate.240 Under the reported conditions, C2 and C3 hydrocarbon products were not observed. As the NifH:NifDK ratio in the reaction was increased, the rate of the competing proton reduction reaction decreased, diverting electron flux to the formation of CH4. Even with a high electron flux (50:1), the primary product was H2, which makes sense considering that proton reduction is a two-electron process, while the conversion of CO2 to CH4 requires eight electrons. To separate the reaction of the α-V70A and α-H195Q NifDK variants between CO and CO2, deoxyhemoglo-bin was added to the activity assays so that any released CO molecule would be rapidly removed from the analysis.240 The resultant reactions demonstrated a ~25% decrease in the production of CH4, indicating that two different processes were being observed: one where CO2 remains bound to nitrogenase throughout the reduction reaction and the other where CO dissociates and presumably rebinds to the M-cluster before reduction to CH4. CO2 reduction was further explored when the NifDK variant was incubated in an atmosphere of 45% CO and Ar, including 1–3% acetylene. Propylene (C3H6) was observed as the major product with propane (C3H8) as a minor component, and isotopically labeled13C atom from bicarbonate was found to incorporate into propylene. This indicates that one molecule each of CO2 and acetylene can couple to form the product. Adjustments of the electron flux would favor CH4 formation at a higher nitrogenase ratio and C3H6 formation at lower ratios. Additionally, introduction of ethylene to the CO2 reaction did not yield propylene, suggesting that the mechanism requires that both molecules must bind and be activated by the cluster before they can be coupled. It is interesting that in the double mutant α-V70A/α-H195Q NifDK, the reactivity with CO is inhibited compared to α-V70A NifDK, but the same α-H195Q mutation enhances the reactivity of the variant NifDK with CO2. Electron transfer rates seem to be important factors for determining the result of CO2 reduction. A slower rate would favor reduced intermediate species that are long-lived, encouraging C–C bond formation, whereas faster electron delivery appears to more quickly generate CH4, discouraging carbon coupling.240 However, electron transfer in nitrogenase must also be understood in the context of proton transfer, because the two events are closely linked to each other.63 In the α-V70A NifDK variant, CO molecules are readily coupled to form longer chain hydrocarbons, but this activity becomes disrupted when the α-H195 residue is exchanged for one less capable of proton transfer. The timing of proton-coupled electron transfer to the substrate is affected for CO reduction but happens to function well for the eight-electron reduction of CO2. What should be emphasized is that a new activity was generated by controlling the second coordination sphere via the steric environment to house larger substrates in the active site niche and by tuning protonation of one of the oxygen atoms of CO2 to generate CO and H2O. Recently, CO2 has also been converted to CH4 by the wild-type Fe-only nitrogenase from Rhodopseudomonas palustris,241 but without additional mutational or structural information about the Fe-only nitrogenase it is difficult to gain further insight into the factors that control CO2 reduction. The structure and catalytic mechanism of other low-valent GPMs active in CO2 reduction like formate dehydrogenase and CO dehydrogenase are discussed in Section 4 and Section 5, respectively. 3. HYDROGENASE Hydrogen turnover (eq 2) is a important reaction in many microorganisms that thrive under reducing, anoxic conditions, and also plays a role in certain aerobes.242–244 Trace amounts of H2 provide electrons to power the anabolism of numerous archaea and bacteria in the soil, aqueous environments, or host tissue (H2 oxidation or “H2 uptake”).245–247 Under fermentative or “microaerobic” conditions, proton reduction and H2 release have been shown to be a key in the redox regulation of autotrophs like photosynthetic bacteria and algae.248–250 Additionally, H2 contributes to the cellular redox equilibrium as a side product of N2 fixation by nitrogenase, as discussed in Section 2. The GPMs responsible for proton reduction and H2 oxidation are referred to as hydrogenases.251 Three classes of phylogenetically unrelated hydrogenases have been defined.252 In archaea, [Fe]-hydrogenase is involved in the conversion of CO2 to CH4 (methanogenesis), catalyzing H2 splitting and hydride transfer at a cofactor with a monometallic iron center.253–255 Not much is known about the influence of second or outer coordination sphere effects, thus we will focus on the remaining two classes in this article. Unlike [Fe]-hydrogenase, [NiFe]- and [FeFe]-hydrogenase are iron–sulfur enzymes and carry a bimetallic active site cofactor (Figure 17). [NiFe]-hydrogenases have been found in archaea and bacteria, and are typically involved in heterotrophic H2 uptake and the “recycling” of H2 during methanogenesis or N2 fixation (Section 2).255–257 A classification into Groups 1–4 has been proposed,246 in which the water-soluble “standard” [NiFe]-hydrogenases are distinguished from membrane-bound, multimeric, as well as O2-tolerant and bidirectional [NiFe]-hydrogenases.257–259 Although highly diverse, all [NiFe]-hydrogenases share the same active site cofactor. It is composed of a nickel and an iron ion, covalently attached to the enzyme by either four cysteine residues, or three cysteines and one selenocysteine (Figure 17A). The iron ion is coordinated by two CN− and one CO ligand, the latter being a common feature of hydrogenases.260 [FeFe]-hydrogenases have been found in bacteria, lower eukaryotes, and green algae, representing the phylogenetically most recent class.261–263 The physiological diversity among [NiFe]-hydrogenases is less pronounced than in [FeFe]-hydrogenases:246 often, [FeFe]-hydrogenases are involved in redox regulation (H2 release upon proton reduction) but in complex with other metalloproteins, additional roles in CO2 reduction (Section 4) or bifurcation have been suggested.264–266 “Standard” [FeFe]-hydrogenases belong to group A and have been investigated intensively whereas knowledge about group B–C [FeFe]-hydrogenase is only beginning to emerge.267,268 Figure 17B shows the active site cofactor including the covalently attached [4Fe-4S] cluster. The metal ions of the diiron site are modified with a bridging carbonyl ligand (μCO) and two terminal CN− and a CO ligand. Here, we shall elaborate on the active site architecture, catalytic mechanism, and second/outer coordination sphere effects in [NiFe]-hydrogenase (Section 3.1) and [FeFe]-hydrogenase (Section 3.2). 3.1. [NiFe]-Hydrogenase 3.1.1. Structural Features of [NiFe]-Hydrogenase. [NiFe]-hydrogenases are divided into Groups 1–4, according to the classification by Morales, Greening, and co-workers.246 In these groups, [NiFe]-hydrogenases are further distinguished into “H2 uptake” (Groups 1 and 2), “bidirectional” (Group 3), and “H2 evolving” (Group 4) clades. Each group is divided into subgroups, for example, the O2-sensitive prototypical or standard-type hydrogenases in Group 1b and the O2-tolerant membrane-bound hydrogenases in Group 1d. The complete list is available in the “hydrogenase database”, HydDB.269 [NiFe]-hydrogenases are multisubunit proteins, which contain at least the catalytic subunit and an iron–sulfur cluster subunit. For example, standard [NiFe]-hydrogenases (Group 1b) are composed of two subunits, large and small, with molecular weights of approximately 60 kDa and 30 kDa, respectively (Figure 18A).270 The large subunit binds the catalytic cofactor and the small subunit embeds the accessory iron–sulfur cluster. The first crystal structure of a standard [NiFe]-hydrogenase was determined for the enzyme from Desulfovibrio gigas (Dg) in 1995 at 2.85 Å resolution,271 followed by the structures of [NiFe]-hydrogenases from Desulfovibrio vulgaris Miyazaki F (DvMF)272–276 and other species.277–290 The Ni–Fe active site cofactor is located inside the large subunit, where the Ni and Fe ions are bridged with two cysteine thiolates, and two other cysteine residues are bound to the Ni ion in a terminal fashion.273,291,292 In addition to two CN− ligands and one CO ligand coordinated to the Fe ion, a bridging hydride ligand (μH–) may exist between the metal ions (Figure 18B).273 The Ni site changes its oxidation state (i.e., Ni3+, Ni2+, and Ni+) among various intermediate states, some of which (Ni3+ and Ni+) have been characterized by EPR spectroscopy. On the other hand, the Fe site maintains the low oxidation, low spin state (Fe2+, S = 0).251,293–298 In the O2-tolerant NAD+-reducing [NiFe]-hydrogenase (Group 3d) from Hydrogenophilus thermoluteolus TH-1, three cysteine thiolates bridge the Ni and Fe ions while one cysteine residue terminally ligates to the Ni ion in the oxidized state.284 In the case of [NiFeSe]-hydrogenase, a terminal cysteine is replaced by a selenocysteine.299–303 In standard [NiFe]-hydrogenases, three iron–sulfur clusters are located in the small subunit and mediate the electron transfer between the Ni–Fe active site and physiological redox partners.304–307 In the crystal structure of the O2-tolerant membrane-bound [NiFe]-hydrogenase (Group 1d, see Figure 18C), a proximal [3Fe-4S] cluster harbored by six cysteine residues has been identified.286,288,289 The substrate (H2) or inhibitors (O2 and CO) are transferred via hydrophobic channels from the molecular surface to the Ni–Fe active site.308–311 After splitting H2 into electrons and protons (eq 2), proton transfer pathways provide an exit route to the molecular surface where protons are release into the solvent. The crystal structures of [NiFe]-hydrogenases revealed that highly conserved amino acid residues like arginine, serine, glutamate, and histidine are located in the second coordination sphere of the Ni–Fe active site (Figure 19). The CO ligand on the Fe ion interacts with the hydrophobic side chains of valine and leucine (V500 and L482 in DvMF). On the other hand, the two CN− ligands accept hydrogen bonds from arginine R479 and serine S502, respectively, in DvMF.273 One of the bridging cysteine thiolates is hydrogen-bonded to histidine H88.312 These interactions have been confirmed in the electron density map of the hydrogen atoms in the subatomic resolution structure of DvMF [NiFe]-hydrogenase.273 3.1.2. Ni–Fe Cofactor and Catalytic Cycle. Standard [NiFe]-hydrogenases form two inactive oxidized states, Ni-A and Ni-B (Ni3+, S = 1/2).259,313,314 In the presence of H2 or other reductants, activation of Ni-A requires long times, whereas the Ni-B is readily activated.251,296 A hydroxide is present at the bridging position in the Ni-B state (μOH–, Figure 20C);275 however, the identification of the bridging ligand in the Ni-A state remains controversial.312,315,316 The crystal structure of the Ni-A state in DvMF was interpreted to represent a peroxide ligand (μOOH–, Figure 20A), while a μOH− ligand was found in Allochromatium vinosum (Av, Figure 20B).275,277,278 The later is in agreement with DFT calculations and single-crystal ENDOR spectroscopy that indicated a μOH− ligand in both Ni-A and Ni-B.315 It was proposed that the differences in the EPR spectra of the two states are caused by a ligand rotation of 7–10° involving C546 and C549 in DvMF. Accordingly, the bridging μOH− ligand of the active site could be same in Ni-A and Ni-B, and the differences of the activation rates between these states may result from the slight distortion in the coordination of the cysteine residues. Further investigation is needed to clarify the difference of these states. One-electron reduction of the Ni-A and Ni-B states leads to the EPR-silent inactive states, Ni-SU and Ni-SIr (Ni2+), respectively (Figure 21).317 The midpoint redox potential for the interconversion between the Ni-A and Ni-SU states was shown to decrease by ~60 mV per pH unit in various [NiFe]-hydrogenases, indicating that the one-electron reduction of Ni-A is coupled to the uptake of one proton.296,308–311,318–321 While the crystal structure of the Ni-SU state is not yet available, DFT calculations proposed coordination of a water molecule to the Fe ion and modification of the Ni coordination structure.322 A glutamate residue in the second coordination sphere of the Ni–Fe cofactor plays an important role for the inactivation process. This highly conserved glutamate residue (E34 in DvMF) is located next to the Cys residue which is terminally bound to the Ni ion (C546 in DvMF, see Figure 19). When the catalytically active [NiFe]-hydrogenase is oxidized under anaerobic conditions, the enzyme converts from the Ni-SIa state to the Ni-B state via the Ni-SIr state (Figure 21). The first order rate constant of the anaerobic inactivation process in the E25Q (equivalent E34 in DvMF) variant of [NiFe]-hydrogenase from Desulfovibrio fructosovorans (Df) was 5.6 times smaller than that of the wild-type enzyme, although the activation rate constant was hardly affected by the replacement.323,324 Thus, the glutamate residue is important for the incorporation of the bridging OH− ligand to the Ni–Fe cofactor. Under anaerobic conditions, formation of the Ni-B state with a bridging OH− ligand from the Ni-SIa state requires not only the transfer of a water molecule into the active site but also the extraction of a proton from a water molecule and oxidation of the Ni ion (Ni2+ to Ni3+). The carboxylic group of the glutamate residue may deprotonate the bound water molecule at the active site, producing the bridging OH− ligand of the Ni–Fe cofactor.323 Interestingly, the Ni-B state was not produced when the E28Q variant (E34 in DvMF and E25 in Df) of the [NiFe]-hydrogenase 1 from Escherichia coli (EcHyd-1) was oxidized on an electrode in the presence of H2, indicating that the interaction of H2 with the Ni–Fe cofactor is maintained in the E28Q variant and that the early stages of H2 activation outpace electrochemical oxidation.325 It may become much easier to start another cycle by activating H2 and generating the Ni-R state in the E28Q variant, since protons may not easily leave from the Ni–Fe cofactor in the variant and thus it becomes difficult to produce OH− from water. During the catalytic cycle, the Ni–Fe cofactor changes the oxidation states in the following order: from Ni-SIa (Ni2+), Ni-R (Ni2+), and Ni-C (Ni3+) to the Ni-L (Ni+) state (Figure 21).251,293–297,308 These states convert among each other by addition or release of H2, protons, and/or electrons. In the first step of the cycle, the Ni-SIa state (with the bridging ligand position vacant) reacts with H2 (Figure 20E). H2 is cleaved heterolytically to a proton (H+) and a hydride (H–), which initiates the transition of the Ni-SIa state to the fully reduced Ni-R state. The high-resolution crystal structure of DvMF revealed that the Ni-R state possesses a μH− ligand and a protonated cysteine (C546-SH) at the Ni–Fe cofactor (Figure 20F).273 The μH− ligand and the protonated cysteine residue in the Ni-R state was also elucidated by nuclear resonance vibrational spectroscopy (NRVS), which is particularly suited to monitor 57Fe ligand vibrations, in combination with DFT calculations.326–328 The hydride ligand was located closer to the Ni ion in the Ni-R state of the enzyme, whereas a short Fe–H− bond and a long Ni–H− bond were found in nearly all synthetic Ni–Fe models.329–332 The protonated cysteine residue is considered as a part of the proton transfer pathways (see below). The H2 binding site at the first step of the cycle is still unclear, and either of the two metal ions may be involved in the formation of a (side-on) H2 σ-bond complex. Although experimental data on H2 binding to the Ni–Fe cofactor are not yet available due to the transient nature of intermediate, theoretical studies suggest the Ni ion as the initial site of H2 binding.333–337 In addition, complementary studies showed that the competitive inhibitor CO binds to the Ni ion at the Ni–Fe active site (Figure 20D).274,338 The coordination geometry of the Ni ion is key to the thermodynamically favorable interaction of H2 with the Ni–Fe cofactor. Theoretical calculations predicted that a peculiar seesaw-shaped geometry in the Ni-SIa state with trans S–Ni–S angles near 120° and 180° is necessary for favorable binding of H2 to the Ni site.335 In line with this proposal, the crystal structure of the Ni–Fe cofactor of F420-reducing [NiFe]-hydrogenase from Methanosarcina barkeri (Mb) exhibited no ligand at the bridging position with unusual trans S–Ni–S angles of 107° and 171° in the Ni-SIa state (Figure 20E).339 Similar angles have also been suggested by resonance Raman studies of the O2-tolerant, membrane-bound [NiFe]-hydrogenase from Cupriavidus necator, ReMBH (noted by its former name Ralstonia eutropha).340 One-electron oxidation of the Ni-R state leads to the paramagnetic Ni-C state (Ni3+, S = 1/2) with a μH− ligand and deprotonated C546-S− at the Ni–Fe active site.341,342 The Ni-C state is then converted to the paramagnetic Ni-L state (Ni+), with a vacant bridging position and a (re-)protonated cysteine C546-SH (see below).296,297,342–345 In the last step of the catalytic cycle, one-electron oxidation coupled with removal of the proton from C546-SH results in conversion of the Ni-L state to the Ni-SIa state.295,346–351 3.1.3. Proton Transfer and Proton-Coupled Electron Transfer. The proton acceptor during the catalytic H2 oxidation cycle has been proposed by theoretical352 and Raman spectroscopic studies.340,353 Apparently, the proton is transferred from the Ni–Fe cofactor to the Ni-coordinating terminal cysteine thiolate (C546 in DvMF), which is the first step in the proton transfer pathway from the Ni–Fe cofactor to the outer coordination sphere (Figure 19).273,297,346,354–356 The H2 oxidation activity of the E25Q variant (E34 in DvMF) in Df [NiFe]-hydrogenase decreased to less than 0.1% compared to the wild-type enzyme.357 These results indicated that the glutamate residue located close to the C546 ligand plays an important role for the proton transfer during the catalytic reaction, which was supported by theoretical studies.358,359 Time-resolved infrared spectroscopy addressing the E17Q variant of O2-tolerant Pyrococcus furiosus [NiFe]-hydrogenase (Pf SH-1) showed that glutamic acid E17 (E34 in DvMF) is a proton relay for the interconversion between the Ni-C and Ni-SIa states.351 The replacement E17Q did not interfere with the μH− photolysis of the Ni-C state but it disrupted PCET from the Ni-L state to the Ni-SIa state, preventing formation of the Ni-SIa state. Alternatively, the highly conserved arginine (R479 in DvMF and R509 in EcHyd-1) has been proposed as a proton acceptor, forming a frustrated Lewis pair for the oxidation of H2 in EcHyd-1 from studies using variants R509K, D574N, D118A, P508A, and D118N/D574N (R479, D544, D123, P478 and D123/D544 in DvMF, see Figure 19).360,361 The turnover rate of the R509K variant decreased by a factor of 100 compared to that of the wild-type enzyme. The H2 oxidation efficiency of the D118A variant, in which aspartate D118 forms a salt bridge to arginine R509, also decreased compared to native EcHyd-1. In the case of the soluble [NiFe]-hydrogenase 1 from Pf SH-1, the R355K (R479 in DvMF) variant altered the ligand binding environment at the Ni–Fe cofactor and destabilized the Ni-C state, resulting in a Ni-C/Ni-L tautomeric equilibrium.362 The SH stretching frequency at 2505 cm−1 has been detected by highly sensitive Fourier-transform infrared (FTIR) difference spectra utilizing the photoconversion of the Ni-C state to the Ni-L and Ni-SIa states in DvMF [NiFe]-hydrogenase. The data showed that C546 is protonated in the Ni-L state (Cys-SH) and deprotonated in the Ni-C and Ni-SIa states (Cys-S−).297,346 Glutamic acid E34-COOH was found to be double hydrogen-bonded in the Ni-L state and single hydrogen-bonded in the Ni-C and Ni-SIa states (Figure 22) according to the COOH stretching frequency between 1700–1730 cm–1. Additionally, a stretching mode of a “dangling” water molecule was observed in the Ni-L and Ni-C states. These results suggest that the Ni–Fe cofactor and its surrounding amino acids function as an optimized proton transfer system in standard [NiFe]-hydrogenases, and the direction of the proton transfer is regulated by the rearrangement of the hydrogen bond network around C546, E34, and a dangling water molecule (Figure 22). The Ni-L and Ni-R states of DvMF [NiFe]-hydrogenase are likely to construct a similar hydrogen bond network between C546-SH, E34, T18, a backbone contact including A548-NH, and a dangling water molecule (Figure 22).346 Several Ni-L states have been identified by light irradiation of the Ni-C state under anaerobic conditions at T < 100 K.342,363–366 From temperature-dependent FTIR studies of DvMF, the ΔH and ΔS values for the equilibrium between the protonated/deprotonated forms (i.e., Cys-SH and Cys-S−) of two Ni-L states were obtained as 6.4 ± 0.8 kJ mol−1 and 25.5 ± 10.3 J mol−1 K−1, respectively.348 The small ΔH and ΔS values indicate efficient proton transfer at the cysteine residue C546 of the Ni–Fe cofactor between the two Ni-L states. The Ni-SIa and Ni-C states may form a similar hydrogen bond network between C546-S–, E34, A548-NH, and a dangling water molecule (Figure 22). Theoretical studies have indicated involvement of a threonine residue in the proton transfer of the transmembrane proton pump bacteriorhodopsin;367 however, T18 seems to be nonessential for proton transfer in Df [NiFe]-hydrogenase and rather stabilizes the local protein structure. This observation was suggested by structural and spectroscopic studies of variants in which threonine was replaced with serine, valine, glutamine, glycine, and asparagine.368 3.1.4. Accessory Iron–Sulfur Clusters. In standard [NiFe]-hydrogenases, three accessory iron–sulfur clusters (denoted proximal [4Fe-4S]2+/+, medial [3Fe-4S]+/0, and distal [4Fe-4S]2+/+) are located almost linearly (each ~12 Å apart) in the small subunit (Figure 23),273 and mediate electron transfer between the Ni–Fe active site and physiological redox partner, such as cytochrome c3.304–307 The proximal [4Fe-4S]2+/+, medial [3Fe-4S]+/0, and distal [4Fe-4S]2+/+ clusters of Dg [NiFe]-hydrogenase exhibit redox potentials of −315, −80, and −445 mV versus SHE, respectively.369 The accessory iron–sulfur clusters are not only important for electron transfer but may also play a role with respect to hydrogen turnover in the presence of O2: some [NiFe]-hydrogenases are “O2-sensitive” (i.e., inhibited by O2) while the “O2-tolerant” [NiFe]-hydrogenases maintain activity under aerobic conditions. The latter have been shown to quickly reactivate from the Ni-B state in a process that appears to be very dependent on the nature of the proximal cluster.370–372 The reactivity toward O2 clearly distinguishes the bimetallic hydrogenases; as discussed in Section 3.2.2, [FeFe]-hydrogenases are irreversibly destroyed by O2. 3.1.4.1. Proximal Cluster. When the proximal [4Fe-4S]2+/+ cluster is reduced in standard [NiFe]-hydrogenases, the [4Fe-4S]+ cluster interacts magnetically with the Ni3+/+ center, which causes splitting of the Ni-C/Ni-L EPR signals below 10 K.373 Spectroscopic studies revealed that the conversion of Ni-C to Ni-SIa via the Ni-L state is controlled by the redox state of the proximal [4Fe-4S]2+/+ cluster in DvMF [NiFe]-hydrogenase.347 The transition of the Ni-L state to the Ni-SIa state may occur when the proximal [4Fe-4S]2+/+ cluster is oxidized but not when it is reduced. These results suggest that the catalytic cycle is controlled by the redox state of the proximal [4Fe-4S]2+/+ cluster, which may act as a gate for the catalytic electron transfer. Most of the standard [NiFe]-hydrogenases possess a glutamate near the proximal [4Fe-4S] cluster (Figure 24A), while some [NiFe]-hydrogenases possess an aspartate instead. The crystal structure of membrane-bound Av [NiFe]-hydrogenase in the Ni-A state showed that there are two different forms for the proximal [4Fe-4S] cluster; one is a standard cubane, while in the other form the proximal [4Fe-4S] cluster is distorted with one of the Fe ions bound to aspartate D75, which is located nearby the cluster (Figure 24B).278 It was proposed that the negatively charged side chain of aspartate stabilizes the distorted form of the [4Fe-4S] cluster. A similar coordination was found in the O2-tolerant [NiFe]-hydrogenase from Citrobacter sp. S-77.285 Additionally, the water molecules near the proximal cluster relocated upon reduction, emphasizing the importance of the water network in this hydrogenase.285 Contrary to the standard [NiFe]-hydrogenases, some of the O2-tolerant, membrane-bound [NiFe]-hydrogenases, such as EcHyd-1, ReMBH, Hydrogenovibrio marinus (HmMBH), and Aquifex aeolicus (AeMBH), possess a unique proximal [4Fe-3S]5+/4+/3+ cluster ligated by six cysteine residues (Figure 24CD).286,288–290 This cluster can be oxidized twice within a very small potential range.374 Its electronic structure has been characterized and debated.375–378 The importance of the coordination by six cysteines to the proximal iron–sulfur cluster has been shown by site-directed mutagenesis studies in ReMBH and EcHyd-1.379–382 Replacement of the two supernumerary cysteine residues C19 and C120 in ReMBH with glycine residues altered the electronic structure of the proximal cluster, and the double variant was unable to sustain activity under prolonged O2 exposure.379 Similar mutagenesis studies in EcHyd-1 showed that O2 tolerance depends on C19 and not on the two-electron oxidation at the proximal cluster, which is detected with the O2-sensitive C19G variant but not with the O2-tolerant C120G variant.380 For another example, the O2-tolerant NAD+-reducing soluble [NiFe]-hydrogenase from C. necator, ReSH (noted by its former name R. eutropha), can produce trace amounts of superoxide in H2-driven NAD+ reduction with O2.381 The C41S variant of ReSH displayed up to 10% of wild-type activity, suggesting that the coordinating role of C41 at the proximal [4Fe-4S] cluster might be partly substituted by the nearby C39 residue, which is present only in O2-tolerant pyridine nucleotide-dependent [NiFe]-hydrogenases, whereas the C39G, C39A, and C39S variants increased the O2 sensitivity compared to wild-type ReSH.382 EPR and 57Fe Mössbauer spectroscopic studies have shown that the proximal [4Fe-3S]3+ cluster donates two electrons to the Ni–Fe active site, resulting in the “superoxidized” state [4Fe-3S]5+.374 A charged glutamate near the cluster (COO−, e.g. in HmMBH and EcHyd-1) or a deprotonated water molecule (OH−, e.g. in ReMBH) stabilizes the [4Fe-3S]5+ cluster.286,288,289,307,383 The deprotonated amide nitrogen of the backbone formed by cysteine C26 (in HmMBH) may bind to one of the Fe ions of the cluster, also stabilizing the highly oxidized state. In AeMBH, the proximal [4Fe-3S]5+/4+/3+ and medial [3Fe-4S]+/0 clusters show midpoint potentials that are higher than potential necessary value for the conversion of the Ni-SIa to Ni-L/Ni-C states in standard [NiFe]-hydrogenase.345,374,384 Additionally, a kinetic argument favoring the appearance of the Ni-L state in EcHyd-1 is that the proximal [4Fe-3S]3+ cluster is fully reduced at the potentials that are required for both the Ni-C and Ni-L states, impeding the elementary electron transfer step that converts the Ni-L to Ni-SIa state. The conserved glutamic acid residue (E34 in DvMF) plays an important role for the mechanism of O2 protection in the O2-tolerant NAD+-reducing [NiFe]-hydrogenase from H. thermoluteolus TH-1, which is revealed by the crystal structures (Figure 25).284 The Ni–Fe cofactor of HtTH-1 in the H2-reduced state exhibits a similar coordination structure as those of standard [NiFe]-hydrogenases, whereas in the air-oxidized state, an octahedral Ni geometry was found; three bridging thiolate ligands between the Ni and Fe ions, one terminally bound cysteine thiolate, and an unprecedented bidentate ligation of the glutamate side chain (E32 in Ht and E34 in DvMF) (Figure 25). The fully occupied octahedral Ni geometry protects the Ni–Fe cofactor from direct attack by O2. Higuchi, Shomura, and coworkers suggested that the conformational change of the Ni–Fe active site was triggered by the reduction of the proximal [4Fe-4S] cluster (Y1 in Figure 25). The coordination structure of the amino acid residues located in between the Ni–Fe cofactor and the proximal [4Fe-4S] cluster also changed upon reduction. In the reduced state, E32 does not bind to the Ni ion and forms a hydrogen bond network including S464 and E56, notably unrelated to the catalytic proton transfer pathway as discussed above (Figure 22). The side chain of arginine R58 flips between E56 and the proximal [4Fe-4S] cluster (Figure 25), which may affect the efficiency of catalytic electron transfer. 3.1.4.2. Medial and Distal Clusters. In standard [NiFe]-hydrogenases, the medial [3Fe-4S]+/0 cluster possesses a rather high midpoint potential, which may play a role in effectively “trapping” electrons from the Ni–Fe cofactor, facilitating the reaction of the Ni-SIr state with H2 via the acid–base equilibrium between the Ni-SIr and Ni-SIa states.373,385,386 Substitution of a proLine, a potential ligand to the [3Fe-4S] cluster, to cysteine (P238C in Df [NiFe]-hydrogenase, equivalent P242 in DvMF) triggered the conversion of the medial [3Fe-4S] cluster to a [4Fe-4S] cluster decreasing the midpoint potential from +65 mV to −250 mV versus SHE. This potential decrease caused a 30% reduction in H2 oxidation activity and a 2-fold increase in H2 evolution activity without significantly altering the spectroscopic properties of the Ni–Fe active site and the proximal and distal [4Fe-4S] clusters.387 The distal [4Fe-4S] cluster is located very close to the solvent and interacts with electron acceptors.251 Three cysteines and one histidine coordinate this cluster (Figure 23). The H184C and H184G variants (distal [4Fe-4S] cluster ligand variants) of Df [NiFe]-hydrogenase (H188 in DvMF) showed only 1.5% and 3% oxidative activity compare to the wild-type enzyme, respectively.388 The activity of the H184G variants decreases or increases upon addition of mercaptoethanol or imidazole to the assay, respectively: these ligands modulate electron transfer rates upon binding to a free coordination site on the distal cluster.388,389 The affinity of the variants H184G, P238C/H184G, and P238C/H184C for the electron acceptors methyl viologen and cytochrome c3 was similar or greater than that of the wild-type enzyme, suggesting that H184 does not govern the partner recognition.389 DFT calculations demonstrated that substitution of the histidine ligand to a cysteine in Df [NiFe]-hydrogenase does not change the reorganization energy of the distal [4Fe-4S] cluster.390 The calculated rate of electron transfer was, however, reduced by three orders of magnitude, resulting from a change in electronic donor–acceptor coupling including histidine H184 and phenylalanine F193 (F197 in DvMF). These results indicated that the protein environment is tuned for efficient electron transfer. A systematic survey of the substitution of amino acid residues related to the medial and distal iron–sulfur clusters has been reported for the [NiFe]-hydrogenase from Alteromonas macleodii (which belongs to Group 1e but shows modest O2 tolerance).391–393 For example, the double substitution of P285C at the medial [3Fe-4S] cluster (P242 in DvMF) and H230C at the distal [4Fe-4S] cluster (H188 in DvMF) increased the H2 evolution activity three- to four-fold compared to the wild-type enzyme.393 Electrochemistry measurements showed that in the R193L variant of O2-tolerant EcHyd-1, in which R193 is located near the histidine ligand (Figure 26, L194 in DvMF), the midpoint potentials of the medial [3Fe-4S]1+/0 and distal [4Fe-4S]2+/+ clusters are more negative than those of the wild-type enzyme.394 The R193L variant also enhanced bias toward H2 evolution and slightly diminished the O2 tolerance, whereas the catalytic activity of the K189N and Y191E variants (equivalent Q190 and P192 in DvMF) did not change compared to that of wild-type EcHyd-1.394 The electron transfer route between the Ni–Fe cofactor and the protein surface is complicated, and theoretical calculations proposed that multiple routes coexist between the metal centers during catalysis.395 Thus, enzymatic activity is tuned by not only in the second coordination sphere of the Ni–Fe cofactor but also via the electron transfer trajectory in the outer coordination sphere. Furthermore, in the case of O2-tolerant [NiFe]-hydrogenases, most variants at the proximal iron–sulfur cluster became O2 sensitive.379 In addition to controlling electron transfer, the accessory iron sulfur clusters play an important role in the O2 tolerance. 3.1.5. Gas Channels in [NiFe]-Hydrogenase. The Ni–Fe cofactor is deeply buried in the center of the large subunit (Figure 27).251 Both substrate and inhibitory gas molecules, such as H2, O2, and CO, can access the Ni–Fe active site from the solvent region through hydrophobic gas channels predicte from the crystal structure of the protein.309,354,396–398 Cavity calculations of these cavities using a 1 Å radius probe showed that there are four gas access points at the protein surface, combining into one channel that leads to the Ni–Fe active site.396,397 In vicinity of the Ni–Fe cofactor, the end of the channel becomes narrower due to a “bottleneck” composed of two hydrophobic amino acid residues, typically valine and leucine (Figure 27).397,398 Noble gases like Xe and Kr bind to the hydrophobic area of the protein and are well visible in electron density maps; thus, protein crystal structures pressurized with Xe or Kr provide useful information for hydrophobic gas channels.354,396,399,400 The crystal structures of the Xe-bound standard [NiFe]-hydrogenases from Df and DvMF have been determined at 6.0 and 1.8 Å resolution, respectively, where the Xe atoms were observed in the hydrophobic cavities of these enzymes.354,396 Controversially, molecular dynamics (MD) simulations suggested that gas molecules like H2 and O2 could reach the Ni–Fe active site via several pathways.398,401 The O2 transfer pathways of O2-tolerant ReMBH has also been investigated by X-ray crystallography together with computational studies and has been compared with those of O2-sensitive [NiFe]-hydrogenases.399,400 The channels of [NiFe]-hydrogenases span distances of 28–76 Å between the protein surface and the Ni–Fe active site, wherein O2-sensitive [NiFe]-hydrogenases have a wider bottleneck and a more complex channel network than those of their O2-tolerant counterparts. Several O2 binding sites have been revealed within the hydrophobic channels that have two entrances and extend to the Ni–Fe active site (Figure 27). Various site-directed mutagenesis studies have tested the suggestion that the presence of bulky residues at the end of the gas channel in the so-called O2-tolerant, regulatory hydrogenases (RH) may prevent O2 access to the active site (e.g., isoleucine and phenylalanine instead of valine and leucine).397 The H2 oxidation activity of the I62V and F110L variants of the regulatory [NiFe]-hydrogenase from C. necator, ReRH (noted by its former name R. eutropha), decreased under aerobic conditions 25-fold and 5.5-fold, respectively, compared to the O2-tolerant wild-type enzyme.402 These results suggest that Re RH exhibits O2 and CO tolerance owing to interruption of the gas channel at the bottleneck to the Ni–Fe active site with the bulky amino acid residues isoleucine I62 and phenylalanine F110. Furthermore, the double variant I62V/F110L was completely inactive, indicating that the shape of the gas channels may play a key role in O2 tolerance.402 The I65V/F113L double variant of the regulatory hydrogenase from Rhodobacter capsulatus is also more active and more O2-sensitive than the native enzyme.403 The hypothesis is that narrower gas channels and fewer gas entrances limit the mass transport to the catalytic active site, presumably contributing to the net O2 tolerance of these enzymes. However, site-directed mutagenesis studied aimed at narrowing the end of the gas channel of the O2-sensitive Df [NiFe]-hydrogenase showed that steric hindrance slows down intramolecular diffusion and O2 inhibition, but without making the variant O2-tolerant. These gas diffusion processes have been studied intensively using many residue replacements at the bottleneck position, namely V74 and L122 in Df [NiFe]-hydrogenase.309,310,404–406 The rates of intramolecular gas diffusion were measured by using the bimolecular rate constant of inhibition by CO as a proxy of the rate of ligand access to the active site, by analyzing the results of isotope exchange assays, and by measuring the Michaelis constant for H2.309–311 The results show that the variations have the same relative effects on the transport of all ligands, CO, O2, and H2, and that not only the size but also the charge of the residues in the bottleneck have a very strong effect (up to nearly four orders of magnitude) on the kinetics of ligand access. Slowing down H2 egress has the direct consequence of selectively slowing H2 production and increasing the catalytic bias of the enzyme in the direction of H2 oxidation.311 Some of the observed effects of the variations on the rates of diffusion could be explained by the results of MD simulations, according to which two distinct gates determine the kinetics, namely V74/R476 and V74/L122.401,407 Although site-directed mutagenesis slows down the intramolecular diffusion rates, the resulting effect on the kinetics of inhibition by O2 is 10-fold or less, because the transport toward the active site limits the reaction with O2 only when it is very slow (e.g., in the V74 M and V74Q variants).406,408 However, some V74 substitutions (in particular V74C and V74H) have a very beneficial effect on O2 inhibition that is not related to the kinetics of diffusion: for reasons that have not been clarified, these variations strongly increase the rate of reactivation after O2 inhibition. The rate of reactivation of Df [NiFe]-hydrogenase V74H is close to that observed with the O2-tolerant enzyme from A. aeolicus,404,409 and mutation to cysteine further increases the O2 tolerance of EcHyd-1.410 According to the type of [NiFe]-hydrogenase, the gas channels differ in size, shape, and branching, with respect to the physiological requirement of each [NiFe]-hydrogenase. The crystal structure of the Mb F420-reducing [NiFe]-hydrogenase in the Xe-bound form revealed an additional gas channel (Figure 28).339 Interestingly, the [NiFeSe]-hydrogenase from Desulfovibrio vulgaris Hildenborough (DvH) has a similar hydrophobic channel,339,411 although this channel is not observable in the crystal structures of standard [NiFe]-hydrogenases.354,396,399,400 In the case of DvH [NiFeSe]-hydrogenase site-directed mutagenesis experiments suggest that O2 reaches the active site through a hydrophilic water channel.412 3.2. [FeFe]-Hydrogenase 3.2.1. Structural Features of [FeFe]-Hydrogenase. Standard [FeFe]-hydrogenases are in most cases monomeric enzymes with a molecular weight ranging between 50–80 kDa. The [FeFe]-hydrogenase from Clostridium pasteurianum, CpI, whose X-ray crystal structure has been published in 1998, shows a mushroom-like shape with four iron–sulfur clusters binding to the “stipe” (Figure 29A).413–415 Because of the sequence similarity with ferredoxin, this fold is referred to as “F-domain”. The [FeFe]-hydrogenase from C. acetobutylicum, CaI, albeit not crystallized, is likely of similar architecture.416 In both enzymes, the iron–sulfur clusters form a conductive chain that serves in electron exchange between soluble redox partners and the catalytic active site. This chain consists of two buried [4Fe-4S] clusters, one surface-exposed [2Fe-2S] cluster, and one surface-exposed [4Fe-4S] cluster, the later which is bound by a three cysteine and one histidine ligand.724 From a comparison of CaI variants, it was concluded that the [2Fe-2S] cluster is the main electron transfer conduit, at least when the enzyme is wired to an electrode.417 On the other hand, docking simulations with CpI suggest that the physiological electron donor, ferredoxin, interacts predominantly with the surface-exposed [4Fe-4S] cluster. According to EPR titrations this cluster has the lowest potential.418 The authors investigated a catalytically inactive cofactor variant to facilitate steady-state conditions, notably below the H+/H2 Nernstian potential, where CpI spontaneously reoxidizes as a result of proton reduction, which prevents equilibrium conditions during the redox titration at low potential.418,419 The [FeFe]-hydrogenase from Desulfovibrio desulfuricans, DdH, whose X-ray crystal structure was published in 1999, features a smaller F-domain with only two [4Fe-4S] clusters and is surrounded by a peptide “belt” (Figure 29B).420,421 The [FeFe]-hydrogenase from Megasphaera elsdenii, MeHydA, lacks this subunit but displays an equally small F-domain.422 Voltammetry experiments suggested that intramolecular electron transfer in MeHydA is rate limiting in H2 oxidation, which primarily determines the catalytic bias of this enzyme.423 The electron transfer chain of the recently crystallized hydrogenase from Clostridium beijeirincki, CbA5H, consists of only two [4Fe-4S] clusters and an additional N-terminal cluster located more than 20 Å away from any other cluster.424 Because of significant disorder, these clusters are not visible in PDB entry 6TTL. A crystal structure of the functional [FeFe]-hydrogenase from the green alga Chlamydomonas reinhardtii, CrHydA1, has yet to be obtained; however, based on several lines of evidence,425–430 it can be concluded that CrHydA1 completely lacks the F-domain and any additional metal centers. The homology model shown in Figure 29C highlights an insertion or “loop” that is typical for [FeFe]-hydrogenases of the Chlorophyta-type.26 The above-mentioned hydrogenases have been crystallized and thoroughly studied using biophysical methods; however, they are not representative of the diversity of the [FeFe]-hydrogenases, some of which are much larger and associated in multienzyme complexes.266 As an example, the hydrogen-dependent carbon dioxide reductase from Acetobacterium woodii has four subunits and 11 predicted iron–sulfur clusters, with an apparent mass around 169 kDa431,432 while the respiratory formate hydrogen lyase complex (FHL) of Thermococcus onnurineus consists of 18 subunits, with a molecular weight of approximately 600 kDa.433 The coupling of CO2 and H2 turnover is discussed in Section 4. The small number of [FeFe]-hydrogenases that have been crystallized so far is certainly an obstacle to learning about how diverse structures and outer sphere effects modulate activity; this is actually true for all GPMs discussed in this review. Unlike the F-domain, the catalytic “H-domain” shows a high degree of similarity in all [FeFe]-hydrogenases.246 Four cysteine residues bind a [4Fe-4S] cluster that is part of the active site cofactor and facilitates electron exchange with the accessory iron–sulfur clusters (or soluble redox partners, in case of CrHydA1). These residues are part of at least three protein loops 1–3 (Figure 30).26 A strictly conserved cysteine residue in loop 3 additionally coordinates the diiron site, which is equipped with CO and CN− ligands and an azadithiolate group (ADT, compare Figure 17B).434–436 Together, the [4Fe-4S] cluster and the diiron site form the active site “H-cluster”. Figure 30 depicts the H-cluster, including a number of functionally relevant amino acid residues and water molecules in the second and outer coordination sphere. The standard or group A [FeFe]-hydrogenases CpI, CaI, DdH, and CrHydA1 have been studied intensively and most of what is known about the influence of second and outer coordination sphere is based on the biochemical and biophysical analysis of these enzymes. The diversity of [FeFe]-hydrogenases has been emphasized;437–439 advances in understanding the second and outer coordination sphere effects in the lesser studied or more complex [FeFe]-hydrogenase groups is currently an emerging area of study. For example, the sensory or bifurcating [FeFe]-hydrogenases from Thermotoga maritime (TmHydS, TmHydABC) and Thermoanaerobacter mathranii (TamHydS) have been characterized.440–443 A cryo-EM structure of TmHydABC has recently been published as a preprint.444 Table 4 shows a comparison of important amino acid residues in the H-domain, highlighting the level of similarity among group A [FeFe]-hydrogenases and the differences to group C and D [FeFe]-hydrogenases. In addition to the above-mentioned enzymes, the table also includes CpII and CpIII, MeHydA1, and CrHydA2.424 This comparison is not comprehensive and restricted to experimentally characterized [FeFe]-hydrogenases. As such, Table 4 allows correlating amino acid composition and catalytic function and will serve as a guideline for the analysis of second and outer coordination sphere effects. Additionally, Table 5 provides a compilation of all amino acid and H-cluster variants that have been published in the last 20 years. We will discuss the influence of natural and artificial variations of the H-domain in Section 3.2.2. Next to the biochemical and electrochemical analysis of turnover activity, the spectroscopic signatures of the active site cofactor have been a major source of information. We define the various H-cluster states hereafter. 3.2.2. H-Cluster, Catalytic Cycle, and Artificial Maturation. 3.2.2.1. Oxidized States. In the absence of H2 or other reductants, the H-cluster in standard [FeFe]-hydrogenases adopts the “active-ready”, oxidized state, Hox. Moderate spin exchange coupling between [4Fe-4S] cluster and diiron site gives rise to the characteristic, rhombic EPR signal (g = 2.105, 2.044, 2.001, see Figure 31), indicating that the [4Fe-4S] cluster is oxidized (+2 state, a pair of Fe(II)–Fe(III)) while the diiron site adopts a mixed-valence +3 state (Fe(II)–Fe(I)).481–483 The distribution of charges among Fep and Fed is debated.267 [FeFe]-hydrogenase in the Hox-state was analyzed by EPR spectroscopy within cells, isolated in solution, and in crystallized form.484 X-ray diffraction structures of such crystals indicated a Fe–Fe bridging CO ligand,413,420 which reflects in a low-frequency IR absorbance band at 1802 cm−1, accompanied by two terminal CO and CN− bands at higher frequencies (Table 6).485–487 A small upshift in frequency relative to Hox has been explained by a protonation of a cysteine residue that coordinates the [4Fe-4S] cluster (see below);488 the EPR signal shifts accordingly (g = 2.106, 2.048, 2.001, see Figure 31).267 This state is referred to as HoxH. In the presence of CO, no H2 oxidation and proton reduction activity is observed489 as the oxidized H-cluster reacts with exogenous CO to form the CO-inhibited state, Hox-CO.414,415 The rhombic EPR signal of Hox converts into an axial signature (g = 2.05, 2.01, see Figure 31), indicative of pronounced spin exchange coupling.490–492 While the crystal structure of Hox shows an open, apical binding site at Fed, the crystal structure of Hox-CO hints at a labile, inhibiting ligand in apical position. Spin polarization favors an apical CO ligand,492 which contrasts from IR spectroscopy that has suggested ligand rotation and the stabilization of an apical CN− ligand.479,493,494 The IR spectrum shows four CO ligands instead of three (Table 6), but due to significant vibrational energy transfer495–497 the assignment of normal modes is less intuitive than in Hox.493 Under acidic conditions, HoxH-CO is formed.488 The EPR spectrum of this state has not yet been reported. 3.2.2.2. One-Electron Reduced States. The H-cluster may adopt two one-electron reduced states, both of which are largely diamagnetic. Reduction of the [4Fe-4S] cluster from +2 to +1 gives rise to the characteristic IR signature of Hred′ (sometimes referred to as just “Hred”), which looks similar to Hox but shifted to lower frequencies (Figure 32A).470,498,499 In Hox and Hred′ the H-cluster binds a μCO ligand and maintains the “rotated’ structure” with an open, apical biding site at Fed (Figure 33A).500 As such, the H-cluster readily interacts with exogenous CO, forming the reduced, CO-inhibited state Hred′-CO.494 As in HoxH, protonation of a cysteine has been proposed, stabilizing the excess electron of the [4Fe-4S] cluster in Hred′.488,501 This may also explain the small upshift of frequencies under acidic conditions that has been assigned to the Hred′H state (Figure 32A). Reduction of the diiron site from +3 to +2 ((Fe(I)–Fe(I)) gives rise to the characteristic IR signature of Hred. Reduction and protonation of the diiron site under ambient conditions triggers the release of the μCO ligand into a terminal position421 and the formation of a Fe–Fe bridging hydride ligand, μH.502 Accordingly, Hred shows three terminal CO ligands and the apical position is occupied by a CO ligand, most likely (Figure 32A).479 In variance to Hox and Hred′, the square-pyramidal geometry of the diiron site is symmetrical (Figure 33B). Under cryogenic conditions, such changes in geometry are precluded and result in an IR signature that clearly shows a μCO ligand.503–505 The latter is likely to be protonated at the ADT ligand and should be referred to as ‘HredH+’. Whether Hred or HredH+ are catalytically relevant cofactor intermediates or representative of an H2-inhbited H-cluster geometry has been debated.506–509 3.2.2.3. Two-Electron Reduced States. Reduction of the [4Fe-4S] cluster starting from Hred results in the formation of Hsred.510 This redox state was described for CrHydA1, where it accumulates under H2, presumably due to the lack of accessory iron–sulfur clusters that distribute reducing equivalents between H-domain and F-domain.511 The H-cluster is diamagnetic in this “super-reduced” state; however, the assignment of the rhombic EPR spectrum with g = 2.08, 1.94, 1.87 has recently been questioned by Zebger, Horch, and coworkers who assigned an axial signal to Hsred (g = 2.15, 1.86).504 The IR difference between Hred and Hsred is similar to Hox and Hred′, that is, a small shift to lower energies (Figure 32B).512 Just like Hred, the rotation of the μCO ligand into terminal position is precluded under cryogenic conditions (‘HsredH+’).503–505 A catalytic intermediate with a terminal hydride ligand was first suggested by Krasna and Rittenberg in 1954, based on H2 and D2 exchange experiments with the [FeFe]-hydrogenase from Proteus vulgaris.513 60 years later, the FTIR, Mössbauer, and EPR signatures (g = 2.07, 1.93, 1.88) of the so-called hydride state, Hhyd, were identified in protein and cofactor variants (Table 5).456–458 In wild-type enzyme, accumulation of Hhyd was achieved upon H2 oxidation starting from HoxH, which confirmed the hydride state as a “natural”, catalytically relevant H-cluster intermediate (Table 6).446 Subsequent NRVS and NMR studies further established our knowledge about Hhyd.459,474,514 The diiron site resides in a formally “superoxidized” state (Fe(II)–Fe(II)), binding the terminal hydride in apical position at the distal iron ion. This position trans to μCO has been exploited to demonstrate the hydride character of the ligand in FTIR H/D exchange experiments.446,457 Similar to Hred′, the [4Fe-4S] cluster is reduced (+1) and likely to be protonated.506 The geometry of Hhyd resembles Hox and Hred′ (Figure 33A). 3.2.2.4. Catalytic Mechanism of Hydrogen Turnover. The ongoing debate about the catalytic mechanism of [FeFe]-hydrogenase506,509 is inseparable from questions regarding the influence of second and outer coordination sphere effects that are discussed in Section 3.2.3 and Section 3.2.4. In the following, the two main models will be introduced. The key difference between the models is the assumption of a protonated and reduced diiron site in the “5-step model” or a protonation event at the [4Fe-4S] cluster in the “3-step model”. In the absence of H2 or other reducing components, all standard [FeFe]-hydrogenases adopt the Hox state,481–483 which is widely accepted as the starting and end point of a catalytic cycle that accounts for hydrogen turnover (eq 2). Figure 34 shows a simplified proposal for the succession of redox intermediates in the direction of H2 evolution. Under the assumption that the H-cluster maintains the μCO geometry upon reduction of the diiron site,503–505 HredH+ and HsredH+ may be catalytic intermediates. The 5-step model starts with reduction of the [4Fe-4S] cluster and the formation of Hred′ over Hox (transition A in Figure 34). It is conceivable that the ADT ligand of the diiron site is protonated via the catalytic proton transfer pathway (Section 3.2.5), which may induce intramolecular electron transfer from the [4Fe-4S] cluster to the diiron site, resulting in the formation of HredH+ over Hred′ (transition B). From here, HsredH+ is formed upon reduction of the [4Fe-4S] cluster (transition C).510 Although there is no experimental evidence for a protonation of the ADT ligand in any H-cluster state,506 the formation of a terminal hydride at Fed upon intramolecular proton transfer and oxidation of the diiron site seems a reasonable assumption (Hhyd over HsredH+, transition D).498 In the final step of the 5-step model (transition E), protonation of the terminal hydride would lead to H2 evolution and the recovery of the oxidized resting state, Hox.514 On the basis of the observation that the H-cluster undergoes geometry changes upon reduction of the diiron site at ambient temperatures (Figure 32), Hred and Hsred have been excluded as catalytic intermediates.505–508 In the alternative “3-step model”, Hred′ is formed over Hox upon proton-coupled electron transfer to the [4Fe-4S] cluster (transition A*).501 Here, the protons may reach the H-cluster via the so-called regulatory proton transfer pathway (Section 3.2.4) while in the 5-step model, transition A is a simple reduction process. Hred′ may convert directly into Hhyd in a second step of proton-coupled electron transfer (transition B* in Figure 34) before protonation of the terminal hydride ligand leads to H2 evolution, analogous to the 5-step model (transition E). It is debated whether Hox or the protonated resting state HoxH is recovered upon H2 evolution.488 Other H-cluster species are also described in the literature. Electrochemistry experiments have shown that various inactive states are formed under reducing conditions; whether these forms correspond to any spectroscopically identified species is unclear, although it can be said that not all species that have been isolated and characterized are necessarily catalytic intermediates.515 Inactive states also accumulate under certain oxidizing conditions. Hinact (or ‘Hoxair’) and Htrans represent unready, “superoxidized” H-cluster states that have been observed in some [FeFe]-hydrogenases.419,516,517 These states differ in the redox state of the [4Fe-4S] cluster, and the IR difference between Hinact and Htrans is similar to Hox and Hred′ (Table 6). Htrans is paramagnetic and gives a substoichiometric rhombic EPR spectrum with g = 2.06, 1.96, 1.89.516 Remarkably, [FeFe]-hydrogenases in the Hinact state do not react with CO or O2.518 The reductive activation of unready DdH was long believed to be irreversible but it has recently been demonstrated that Hox can be converted into Htrans and Hinact in the presence of exogenous sulfide, which binds to Fed.519–521 The formation of Hinact in the absence of sulfide was considered to cause the reversible high potential inactivation of the enzyme as detected in electrochemistry experiments.522 Later, it became clear that sulfide-independent oxidative inactivation results from the inhibition of the enzyme by traces of halide ions like chloride or bromide.523 Recently, the formation of Hinact was observed in the [FeFe]-hydrogenase CbA5H, notably in the absence of either sulfide or halides.524 The structure of the air-oxidized H-cluster suggests that Fed is coordinated by the sulfur atom of the conserved cysteine residue C367 (Figure 35). The formation of Hinact in CbA5H and the resulting resistance to O2 are lost when this cysteine is replaced with aspartate (Table 5).424 This mechanism explains why the same Hinact IR signature is observed in air-oxidized CbA5H,524 in DdH natively produced in the sulfate-reducing bacterium D. desulfuricans,487 and in certain standard [FeFe]-hydrogenases exposed to sulfide.520 However, it is unclear why CpI is reportedly insensitive toward exogenous sulfide, unlike other standard [FeFe]-hydrogenases like CrHydA1 and DdH.521 Moreover, direct coordination of Fed by the cysteine side chain was observed exclusively in CbA5H and not in standard [FeFe]-hydrogenases. It has been shown that C367 in CbA5H is located on a loop whose unique flexibility depends on residues up to 13 Å away from the H-cluster (Figure 35).424 The investigation of how the conversion between active and inactive states depends on the nature of these residues provided a clear demonstration of long-range outer coordination sphere effects on active site reactivity. Overall, the mechanism of aerobic inactivation is not well understood. While O2 inhibition of DdH and CbA5H is reminiscent of the reaction of [NiFe]-hydrogenase with O2 (Section 3.1.4), most [FeFe]-hydrogenases are irreversibly destroyed under aerobic conditions. It is believed that O2 binds to the H-cluster like a ligand525–528 and becomes reduced to superoxide, H2O2, or other reactive oxygen species (ROS) that subsequently damage the H-cluster.529–533 Outer coordination sphere effects are likely to play a role, for example, considering O2 diffusion through the enzyme465 and the influence of electron transfer423 or proton transfer532 on ROS formation. The comparatively sluggish reaction of sensory [FeFe]-hydrogenases with O2 had been explained with inefficient proton transfer.443 Additionally, second coordination sphere effects may influence the formation of the Hox-O2 state where the superoxide ligand is stabilized by hydrogen-bonding interactions with the ADT headgroup, not unlike Hox-CO and Hhyd.479 As the hydride state is a key intermediate in the catalytic cycle (Figure 34) truly O2-tolerant [FeFe]-hydrogenases may not be feasible. 3.2.2.5. Artificial Maturation of the H-Cluster. In vivo, the iron–sulfur clusters of [FeFe]-hydrogenase are inserted by the regular iron–sulfur machinery, namely the isc system.534 This includes the [4Fe-4S] cluster of the H-domain that serves are “nucleation core” for the synthesis of the H-cluster.535 This task is performed by three auxiliary enzymes, HydEFG.536 A precursor of the diiron site is assembled on HydF537,538,539 by the help of radical S-adenosylmethionine (rSAM) enzyme HydG.540–544 Then, HydF delivers the active site precursor to the hydrogenase apoprotein where it fuses with the bridging cysteine upon CO release to form the H-cluster. A mechanism has been proposed that includes guidance by charged amino acids,445 not unlike the channel proposal for the maturation of nitrogenase (Section 2.2). It has been suggested that the role of the rSAM enzyme HydE is the synthesis of the ADT ligand.545–547 In vitro, the rather intricate maturation process can be hijacked by loading HydF with diiron compounds and mixing HydF with apo-hydrogenase.468 This allowed for inserting different diiron site mimics into otherwise unaltered enzyme, creating “cofactor variants” or semisynthetic enzymes that are new to nature.480,548,549 Further, it has been shown that even HydF can be avoided in the maturation process: incubation of apo-hydrogenase with diiron compounds lead to functional enzyme.469 The protocol established by Happe and co-workers led to a great variety of cofactor variants, including different dithiolate and diatomic ligands,418,468–474 chalcogenides,475,476 and metal ions (Table 5).478 For example, replacing the central atom X of the dithiolate ligand from nitrogen (ADT) with carbon (PDT), suppresses any redox change at the diiron site.468–470 This allowed for studying the protonation and redox chemistry of the [4Fe-4S] cluster in detail (Figure 32).488 3.2.3. Direct Interactions between H-Cluster and Protein. The presence of CO and CN− ligand at the active site cofactor of [FeFe]-hydrogenase was proven by FTIR spectroscopy in 1996.517 When the H-cluster was modeled into the first electron density of [FeFe]-hydrogenases CpI and DdH, the position of the diatomic ligands was assigned according to second coordination sphere interactions.413,420 The chemical identity of the CO and CN− ligands could not be deduced from the electron densities alone. In CpI, the proximal CN− ligand may accept a hydrogen bond from the side chain of serine S232 and the secondary amine of the backbone formed by S232 and P231 (Figure 36).26 The latter is strictly conserved in [FeFe]-hydrogenase and most likely is responsible for the correct orientation of the backbone amine toward Fep-CN−; in CpI and DdH the angle between ligand and backbone is ~170°, indicative of a strong hydrogen bond. Serine S232 of CpI is replaced with an alanine in DdH (A109), which cannot form a bond with the H-cluster. Phylogenetically, the low level of conservation (Table 4) suggests a minor role in the second coordination sphere; however, the effect can be demonstrated. Winkler, Happe, Rüdiger, and co-workers identified siteselective shifts of the Fep-CN− IR band by analyzing CpI–S232A and the corresponding variant of CrHydA1, A92S.464 The authors even introduced an additional hydrogen bond by changing alanine A94 to serine, which doubled the IR shift observed for A92S (Figure 36). Affecting the electron density at Fep and the coupling with the [4Fe-4S] cluster, these variants had a clear influence on the directionality of hydrogen turnover,464 which may explain the difference in catalytic performance between wild-type CpI and CrHydA1. The hydrogen-bonding environment of the distal CN− ligand is difficult to characterize experimentally. Early on, the strictly conserved lysine K358 (CpI) and K237 (DdH) was suggested as hydrogen-bonding donor to Fed-CN− (Figure 36).550 As the chemical nature of the ADT ligand was still unknown in 1999, Fontecilla-Camps and co-workers even proposed a catalytic proton transfer pathway based on K237 in DdH.420 Site-directed mutagenesis of this lysine, for example, toward alanine450 or asparagine,445 precluded any cofactor insertion and hinted at a key role in the activation of apo-hydrogenase.430 This renders functional studies of hydrogen bonding difficult. Investigating 14N-labeled DdH by hyperfine sublevel correlation spectroscopy (HYSCORE),483 Lubitz, Silakov, and co-workers assigned the large quadrupole couplings to a hydrogen bond between Fed-CN− and K237.436 On the contrary, similar experiments on the CrHydA1 cofactor variant PDT by the same group resulted in substantially simplified 14N HYSCORE spectra that showed no coupling with K228 (equivalent K358 or K237).551 While the role of K358 in hydrogen-bonding remains ambiguous, the sum of weak interactions around Fed-CN− results in a certain level of structural rigidity in the oxidized state;552 however, this might not preclude ligand rotation upon H-cluster state transitions. For example, DFT calculations based on 16 different13CO isotopomers of the H-cluster suggested an apical CN− ligand in the Hox-CO state,493 stabilized by an “internal” hydrogen bond with the ADT ligand. While this interpretation is not without controversy,492 the stability of Hox-CO was found to be largely reduced upon modification of the dithiolate ligand, following a trend, which cannot be explained with an apical CO ligand.479 In contrast to CN–, carbonyl ligands are poor hydrogen bond acceptors.553 A similar concept of internal hydrogen bonding may play a role in the formation and stabilization of Hhyd.458 Moreover, CO/CN− ligand rotation was discussed in the context of the temperature-dependent interconversion of Hox and Hred/Hsred.505,506 While the role of K357 in hydrogen-bonding remains ambiguous, other conserved residues in the vicinity of Fed have been proposed to interact with the CN− ligand, including S323 as well as a backbone contact involving P234 (Figure 36).26 Steric or hydrophobic interactions of the protein fold with the terminal CO ligand have not yet been explored experimentally, but the influence of methionine M353 on the bridging carbonyl was subject to dedicated analysis. Triggering the formation of Hox from Hox-CO by visible light irradiation, Lemon and Peters observed an elongation of the distance between μCO and the sulfur atom of the M353 side chain in CpI.415 Although both Hox-CO and Hox carry a μCO ligand,485–487 the increased intensity of the μCO band in the CO-inhibited state may be related to the lack of steric pressure and a greater change in dipole moment. Twenty years later, Peters, King, and co-workers revisited CpI by cryogenic XFEL crystallography, assigning similar differences to an oxidized “conformation A” and a reduced “conformation B” (Figure 37).554 This includes reorientation of both M353 and sereine S357, the latter which is located close to the [4Fe-4S] cluster in CpI (potential effect are discussed in Section 3.2.4). The presence of a μCO ligand has been recognized as a key requirement for hydrogen turnover by various authors.503,506,555 Precluding the formation of bridging hydride intermediates,502 the orientation of M353 may play a key role in maintaining the CO-bridged, rotated H-cluster geometry. Note that M353 and K358 are part of Loop 2 (Figure 30), which may indicate coupling between Fed-CN− and μCO. On the contrary, Table 4 indicates that M353 is moderately conserved at best. [FeFe]-hydrogenases with a glycine or serine residues instead of methionine feature altered catalytic properties (CpIII or TamHydS) or pronounced downshifts of the μCO frequency (CpII and TmHydS).440,443,554 No such shift was observed in MeHydA, despite the unique variation from methionine to threonine.422 In the case of the sensory hydrogenases TmHydS and TamHydS, multiple variations seem to contribute to the catalytic phenotype, which makes it difficult to pinpoint individual effects. Replacement of methionine with leucine in the standard hydrogenases CpI and CrHydA1 greatly diminished catalytic activity.450 Methionine M353 may interact with the μCO ligand while M497 and cysteine C299 are hydrogen-bonded to the ADT ligand (Figure 37). These residues are strictly conserved in Group A [FeFe]-hydrogenases but replaced with glycine, leucine, or sereine in TmHydS and TamHydS (Table 4). The cysteine plays a key role in proton transfer and will be discussed in the next paragraph. Methionine M497 is part of Loop 3, which also bears C499 and C503 (Figure 30). Its function is unclear; it has been discussed as an alternative proton transfer relay or hydrogen-bonding partner that keeps the ADT ligand in position.267 Site-directed mutagenesis of CpI (M497L) and CrHydA1 (M415L) demonstrated the importance of this residue.450 3.2.4. Proton Transfer and Proton-Coupled Electron Transfer. In all hydrogenases, proton transfer is a key determinant of catalytic efficiency.297 Protons enter the enzyme to be consumed in the evolution of H2 or leave the enzyme upon H2 oxidation, following a Grotthuß-type mechanism. This demands a network of hydrogen-bonded amino acid residues, water molecules, or hydrophilic cofactors. Proton transfer across the second and outer coordination sphere has been recognized as one of the major differences between biomimetic catalysts and enzymes.556–558 On the basis of the crystal structure of [FeFe]-hydrogenases CpI and DdH various proton transfer pathways have been proposed.559–562 As an experimental marker for an involvement of individual amino acids, the analysis was initially restricted to catalytic activity.454,563 However, when it was discovered that the H-cluster accumulates the Hhyd in variants specifically impaired in proton transfer (Table 5),456–458 spectroscopic methods could be used to investigate the minutiae of proton uptake and release. X-ray crystallography448 and FTIR difference spectra564 uncovered a tight hydrogen-bonding network in CrHydA1 and CpI, stretching across >20 Å including a small water cluster W1, and C299, the cysteine residue close to the distal iron ion (Figure 38). On the basis of the first crystal structure of CpI, this pathway was proposed early on.413 The reduction and protonation of the H-cluster upon formation of Hred over Hox has been exploited to trigger proton transfer. The changes in protonation and hydrogen bonding gave rise to distinct FTIR difference spectra, suggesting only subtle changes, for example, the formation of a hydrogen bond between E279 and S319 as well as changes with respect to W1. The lack of difference signals in the SH regime hinted toward a largely preserved hydrogen-bonding environment around C299 and did not support protonation of the ADT ligand, that is, in Hred.564 While the residues involved in proton transfer are mostly conserved within Group A, sensory [FeFe]-hydrogenases show drastic variations (Table 4).440,443 It remains to be understood how protons are transferred in such enzymes; the influence of amino acids replacements on catalytic activity, bias and reversibility29 certainly emphasize the influence of the outer coordination sphere. Next to the “catalytic” proton transfer pathway, a comparison of several [FeFe]-hydrogenase crystal structures suggested the existence of a second, water-based proton transfer pathway.488 Because of the lack of “acidic” moieties in this trajectory (e.g., glutamic acid residues), the difference in apparent pKa compared the catalytic pathway allows controlling proton transfer to the diiron site and/or the [4Fe-4S] cluster.512 Figure 39 shows how a chain of water molecules permeates toward the intersection of the F- and H-domains, that is, C193, which coordinates the most proximal [4Fe-4S] cluster and C499, the later which in turn coordinates the H-cluster. Protonation of a cysteine ligand is rather uncommon in nature; however, quantum mechanical calculations of the IR spectra of HoxH, Hred′, and Hhyd provided excellent agreement with the experimental CO/CN− signature when C499 was protonated in silico.565 As demonstrated in Figure 32, redox- and protonation changes in the second coordination sphere of the diiron site (e.g., at the [4Fe-4S] cluster) result in small yet discernible IR band shifts. The spectroscopic characterization is in agreement with the pH-dependent population of HoxH,488 the pH-dependent transition potential of the Hox/Hred′ couple,501 and the observation that Hhyd is formed upon H2 oxidation via HoxH directly.446 Influencing the electron density distribution across the H-cluster, the second proton transfer path has been named “regulatory”.488 In general, electrons and protons tend to move in a concerted fashion, minimizing energetic barriers22–24 and facilitating long-range coupling between electron transfer and proton transfer, an extreme case of outer coordination sphere effects.447,566,567 In the case of [FeFe]-hydrogenase, PCET shapes the electron density distribution across the H-cluster by locking reducing equivalents at the diiron site upon protonation of the ADT ligand498 or formation of a bridging hydride (μH).502 Similar ideas have been put forward to explain the pH-dependence of the [4Fe-4S] cluster,488,501 although this has been disputed.568 The existence of two separate proton transfer pathways (Figure 39) complicates the analysis of the pH-dependence and led to conflicting results.512 These differences have important implications for the understanding of [FeFe]-hydrogenase catalysis, as discussed elsewhere.506 3.2.5. Gas Channels in [FeFe]-Hydrogenase. The mass transport of reactants like H2, O2, and CO within [FeFe]-hydrogenase CpI has been studied using molecular dynamics.465,569–571 Schulten and co-workers calculated how “the protein’s natural equilibrium dynamic motion on the nano-second time scale can define predetermined pathways for hydrophobic gas transport”, arguing that permanent channels may not be needed.569 Figure 40A depicts that this applies to small gases like H2 in particular. But while CO and O2 may not access the protein as easily as H2, Mohammadi and Vashisth suggested defining a “network of pathways” that predicts gas diffusion more accurately570 (Figure 40B). This complexity must be met with a statistical approach to site-directed mutagenesis, therefore automated screening protocols572,573 were employed understanding the influence of “gas filters”, for example, selectively precluding access of O2 and CO to the active site (Table 5).452,453 Currently, the role of hydrophobic gas channels in the outer coordination sphere is not well understood. With respect to CO inhibition, there is a remarkable variety that may help identifying certain trends. The rate of CO inhibition among different standard [FeFe]-hydrogenases was found to vary over three orders of magnitude423,522 while sensory [FeFe]-hydrogenases440,443 and [FeFe]-hydrogenase complexes show even lower CO affinities.432 Apparently, nonconserved residues remote from the active site tune the reactivty; however, the identity of the responsible amino acids and the underlying mechanistic contributions have yet to be explored. In the second coordination sphere of the H-cluster, the replacement of the V296 and F290 in CrHydA1 decreased the rate of CO inhibition ten-fold, but the effects on O2 deactivation were not proportional.465 The simultaneous variation of two residues located between the H-cluster and the proximal accessory cluster in CpI moderately reduced the rate of inactivation.452 As the diiron site has been shown to be the initial target of oxidation, this is an interesting observation that may hint at the role of the [4Fe-4S] cluster in the reaction with “activated” O2.529,531,533 4. FORMATE DEHYDROGENASE Formate dehydrogenases (FDHs) are a diverse group of enzymes in bacteria, archaea, and eukaryotes that catalyze the reversible two electron and one proton abstraction of formate (HCOO−) to produce CO2 (eq 3).574,575 In general, these enzymes are involved in diverse metabolic pathways but mostly in the production of CO2 from formate.576 On the basis of the low standard potential of formate oxidation (E° = −420 mV vs SHE) prokaryotes derive energy by coupling it to the reduction to one of several terminal electron acceptors.577 Some enzymes, however, were described to act as CO2 reductases, with a preference for the reaction of reducing CO2 to formate.578 FDHs are divided into two major classes, into the metal-dependent and metal-independent FDH enzymes.576 The metal-independent FDHs generally are NAD+-dependent and distributed from bacteria to yeast, fungi, and plants.579 These enzymes have no redox-active metal centers involved in catalysis, but instead NAD+ is used as a cosubstrate and serves as a hydride acceptor for converting formate directly to CO2.580 In these proteins, formate is positioned in proximity to the NAD+ cosubstrate to facilitate the hydride transfer step. Since this review is focused on GPMs, only the metal-dependent FDHs will be described in detail. The metal-dependent FDH enzymes are restricted to bacteria and archaea and contain either molybdenum (Mo) or tungsten (W) as metal ions in the active site.575,581,582 They have a higher complexity by being composed of several subunits, which contain in addition to the active site bis-metal-binding pterin (MPT) guanine dinucleotide (bis-MGD) cofactor (Figure 41) various additional cofactors, such as diverse Fe–S clusters, FMN, or hemes (Figure 42).575 4.1. Metal-Containing Cofactor of FDH The metal-containing FDHs are members of the dimethyl sulfoxide (DMSO) reductase family of mononuclear molybdenum (Moco) or tungsten cofactor (Wco)-containing enzymes.583,584 In these enzymes, the active site in the oxidized state comprises a Mo or W ion present in the bis-MGD, which is coordinated by the two dithiolene groups from the two MGD moieties, a protein derived selenocysteine or cysteine ligand and a sixth ligand that is accepted to be a sulfido ligand (Figure 41B). Selenocysteine-containing enzymes have generally a higher turnover number than the cysteine-containing ones.585 The analogous chemical properties of W and Mo, the similar active sites of W- and Mo-containing enzymes, and the fact that W can replace Mo in some enzymes have led to the conclusion that both Mo- and W-containing FDHs have the same reaction mechanism.581 The question of why in FDH enzymes both metal ions can be found while other enzymes of the DMSO reductase family exclusively function with only one type of metal ion is still not completely understood.584,586,587 Generally, it was found that W-containing enzymes catalyze reactions with low redox potentials (E° = −420 mV vs SHE), and it was shown that WIV is a more reducing ion than MoIV.587 Possibly for these reasons, the conversion of CO2 to formate was demonstrated to occur preferentially for W-FDH enzymes.588–590 However, the purified Mo-containing FDHs were also shown to perform CO2 reduction in the presence of excess reducing equivalents.578 In the cell, however, this reaction would thermodynamically not be favored.577 4.2. Structural Features of FDH Overall, there is a great variety in the subunit architecture of metal-containing FDHs. The α-subunit, or the Moco/Wco containing domain, is highly conserved and minimally comprises the Moco/Wco binding site and a [4Fe-4S] cluster proximal to the Moco/Wco cofactor (Figure 42).575,578 These α-subunits or domains are typically between 80–95 kDa and the overall folds are highly identical. In most cases, the Moco/Wco cofactor and the proximal [4Fe-4S] cluster interface with additional Fe–S clusters, either within the α-subunit or close by in a neighboring subunit. Metal-dependent FDHs are found either in the cytosol, in the periplasm or are membrane-bound, facing the periplasm (Figure 42).575,582 The enzymes are generally involved in diverse biochemical pathways and use different electron acceptors/mediators like hemes, ferredoxins, NAD+, coenzyme F420, or membrane quinols for the reaction of formate oxidation or CO2 reduction.591 The formate oxidation activity of FDH is typically followed via the reduction of methyl viologen (MV), benzyl viologen (BV), or NAD+. Additionally, the analysis of “catalytic” currents can be followed via protein film electrochemistry (Table 7).592,593 In organisms that catalyze the oxidation of formate, this reaction is a key process for obtaining energy and reducing equivalents.577 The CO2 formed is usually incorporated predominantly into the Calvin-Benson cycle, as in autotrophic organisms.576 Formate is a key metabolite for bacteria, arising as a metabolic product of bacterial fermentations and functioning as a growth substrate for many microorganisms (e.g., methanogens and sulfate-reducing bacteria).599–601 Formate is also an intermediate in the energy metabolism of several prokaryotes, and formate is used as electron donor during anaerobic respiration in these organisms. Because of the low redox potential of the CO2/formate couple, formate can be oxidized not only through the aerobic respiratory chain, but it also can serve as an electron donor for the anaerobic reduction of fumarate (E°fumarate/succinate = +30 mV vs SHE) and nitrate (E°nitrate/nitrite = +420 mV vs SHE) or nitrite (E°nitrite/NO = +375 mV vs SHE).581 Some FDHs are in complex with hydrogenases, like the formate-hydrogen-lyase (FHL) systems from E. coli, Pectobacterium atrosepticum, or Clostridium carboxidovorans.602–605 Physiologically, the E. coli FHL is a membrane-bound system involved in formate oxidation and H2 evolution under fermentative growth conditions.606 This system contains the formate dehydrogenase subunit FdhF, and a membrane-bound, cytoplasm-faced [NiFe]-hydrogenase (Section 3.1).607 Similar coupling of FDH and hydrogenase subunits are observed in some CO2 reductases that catalyze the reduction of CO2 to formate with the simultaneous and direct oxidation of H2. For example, the enzyme from Acetobacter woodii is a tetramer, with one Moco/[4Fe-4S] cluster subunit, two subunits with four [4Fe-4S] clusters each, and one [FeFe]-hydrogenase subunit (Section 3.2).431,608 A W-containing homologue has been identified in Thermoanaerobacter kivui.266 4.2.1. Carbon Dioxide Reduction and Coupled Catalysis. A distinct class of FDH-associated enzymes are those found in methanogenic archaea that are involved in hydrogenotrophic CO2 fixation, en route toward the generation of methane (CH4).609,610 The first enzyme involved in reductive CO2 fixation is the enzyme formylmethanofuran dehydrogenase, which is biased toward the reduction of CO2 and the coupling of the intermediate product formate to the C1-carrier methanofuran without the requirement of ATP.611,612 Furthermore, some organisms that undergo methylotrophic or aceticlastic methanogenesis have an ability to produce CO2 as a final step, also employing a formylmethanofuran dehydrogenase.613,614 The basis for this biochemical transformation, until recently has been largely speculative. Similar to FDHs described above, formylmethanofuran dehydrogenases are able to coordinate a bis-MGD cofactor with either Mo or W;597,609,615–620 the enzyme also exhibits a distinct dinuclear Zn amidohydrolase motif that is involved in the condensation of formate and methanofuran.612 The Methanothermobacter strains M. marburgensis and M. wolfeii harbor distinct operons of formylmethanofuran dehydrogenase that depend on the presence of tungstate (WO42−) or molybdate (MoO42−).597,615–618,621 The former is constitutively expressed in both organisms, whereby a Mo-substituted isoenzyme has been also isolated.620,622,623 Similarly, the genome of the hyperthermophilic methanogen Methanopyrus kandleri encodes for two W-containing formylmethanofuran dehydrogenases, with one that is also only expressed in the presence of Se.624 Interestingly, distinct molybdopterin dinucleotides have been characterized for formylmethanofuran dehydrogenases from M. marburgensis, such as coordinating a molybdopterin adenine or hypoxanthine dinucleotide.625 4.2.2. Oxygen Sensitivity. Most FDH enzymes characterized so far are described to be inactivated in the presence of O2 (referred to as oxygen-sensitive enzymes).626–629 To prevent damage by O2 and to stabilize the enzyme, inhibitors like azide (N3−) or nitrate (NO3−) are added during the purification of the enzyme, inhibitors that are known to significantly increase the stability of most FDH enzymes.630 FDHs that exhibit activity in the presence of O2 with inhibitor present (referred to as oxygen-tolerant enzymes) were described for the Mo-containing enzymes from R. capsulatus and C. necator and the W-containing enzyme from D. vulgaris Hildenborough.595,596,631,632 What contributes to the higher stability of some enzymes (e.g., from various Desulfovibrio specimen590,633–636 and Methylobacterium extorquens609,637) in the presence of O2 without inhibitor present, while other enzymes are extremely sensitive to the exposure of O2 (e.g., E. coli FdhF626) is not known so far. While hydrogenases act as oxygenases reducing O2 to reactive oxygen species (Section 3.2) or water (Section 3.1), the O2 sensitivity of FDH seems to follow other molecular principles. Recently, it was shown the enzyme from R. capsulatus is mainly sensitive to O2 in the absence of N3–.638,639 By extended X-ray absorption fine structure (EXAFS) spectroscopy, it was revealed that the exposure of the enzyme to O2 results in a heterogeneous ligandation at the Mo ion and to the exchange of the terminal sulfido ligand by an oxo ligand.638,639 Azide stabilizes the sulfido ligand, suggesting that N3− binds in vicinity to the sulfido ligand and hinders thereby the access of O2 to the active site sulfido group. 4.2.3. Cofactor Biosynthesis and Moco Sulfuration. The biosynthetic maturation process of FDHs, in conjunction with the number of redox cofactors, is critical to obtain a bis-MGD cofactor with the correct set of ligands at the active site.640 FDHs, in conjunction with most Mo- and W-containing enzymes coordinate multiple redox centers in addition to the prosthetic Moco or Wco.575 The maturation of bacterial bis-MGD-containing enzymes is a complex process leading to the insertion of the bulky bis-MGD cofactor into the apoenzyme. Most of these enzymes were shown to contain a specific chaperone for the insertion of the bis-MGD cofactor.640 FDH together with its molecular chaperone seems to display an exception to this specificity rule, since the E. coli chaperone FdhD has been proven to be involved in the maturation of all three FDH enzymes present in the cell.641–643 For the FDH FdsGBAD from R. capsulatus the two proteins FdsC and FdsD were identified to be essential for enzyme activity, but FdsC is not a subunit of the mature enzyme.596 While FdsD is a subunit of the (FdsGBAD)2 heterodimer,591 it has only counterparts in some O2-tolerant FDHs.575 While an enzyme purified in the absence of FdsD was devoid of Moco, the role of this subunit is not clear so far.596,644 On the other hand, FdsC shares high amino acid sequence identity to E. coli FdhD, the chaperone for the cytosolic E. coli formate dehydrogenase FdhF and the two membrane-bound FDHs FdnGHI and FdoGHI from E. coli.644,645 FdhD was reported to be involved in the formation of the essential terminal sulfido ligand at the Mo ion facilitating the reaction between the L-cysteine desulfurase IscS642 and FdhF-bound bis-MGD, a mechanism which is essential to yield active FdhF.642 Initially, it was proposed that the two cysteine residues C121 and C124 in FdhD are involved in the sulfur transfer reaction from IscS further onto the bis-MGD in FdhF. For sulfuration of bis-MGD, FdhD specifically interacts with IscS in E. coli. It has been suggested that IscS transfers the sulfur from l-cysteine to FdhD in form of a persulfide. Located in a conserved CXXC motif of FdhD, cysteine C121 and C124 were proposed to be involved in the sulfur transfer process from IscS to bis-MGD by Magalon and co-workers642 as well as Walburger, Arnoux, and co-workers.643 FdhD was cocrystallized in complex with GDP suggesting direct binding of bis-MGD to FdhD; however, it was not possible to prove that the cofactor is bound in an active form.643 A study using R. capsulatus FdsC provided evidence that the cofactor is indeed bound in the active form containing the terminal sulfido ligand, which was revealed by the insertion of MGD cofactor into TMAO reductase,644 producing an active enzyme.645 Further, a phylogenetic study revealed that the CXXC motif is not conserved among FdhD-like chaperones and by site-directed mutagenesis it was proven that the two cysteines in the CXXC motif are not essential, putting the mechanism of the persulfide-transfer involving FdhD as a sulfurtransferase into question.645 It therefore remains possible that IscS directly transfers the persulfide sulfur to the Mo ion in FdhD-bound bis-MGD, notably without the involvement of additional cysteine residues. The source of electrons for the reductive cleavage of the persulfide is also unknown. Directly comparing the chaperones FdhD from E. coli and FdsC from R. capsulatus in the same study revealed that their roles in the maturation of FDH enzymes from different subgroups can be exchanged.645 In summary, the binding of sulfido-containing bis-MGD to the chaperone is a common characteristic of FdhD-like proteins and ensures insertion of a competent active site cofactor for catalysis. 4.3. Role of Sulfido- and Amino Acid Ligands The first coordination sphere at the active site cofactor of FDHs includes the sulfido ligand and the cysteine or selenocysteine ligand (Figure 41A).627,646 The sulfido group was shown to act as a hydride acceptor from formate in the reaction of formate oxidation.647 This reaction step was mainly established by EPR spectroscopy, showing that the Cα hydrogen atom from formate is transferred to a ligand in the first coordination sphere, which resulted in the binding of a strongly coupled and solvent exchangeable proton, with a hyperfine constant of 20–30 MHz in the MoV state.648–650 Similar hyperfine constant values were determined in the Mo-containing enzyme xanthine oxidase, where a hydrogen is transferred from the C8 hydrogen atom of xanthine to the active site sulfido ligand at the Mo ion.651 Similar to xanthine oxidase, previous studies showed that FDHs exhibited sensitivity to cyanide (CN−).652 In a recent XAS study on R. capsulatus FDH it was shown that the sulfido ligand was lost forming thiocyanate (SCN−), which represents another similarity with xanthine oxidase.638 The resulting desulfo enzyme harbors an oxo group in place of the sulfido group, which is not able to act as hydride acceptor. However, examples of FDHs that exhibit apparent or partial CN− insensitivity are known, including W-FDHs from D. desulfuricans648,653 and D. gigas,636 as well as the W-containing isoenzyme of formylmethanofuran dehydrogenase from M. marburgensis.617 Formate dehydrogenases catalyze formate oxidation and CO2 reduction at the bis-MGD with the involvement of amino acids in the first and second coordination sphere. Of these, the first amino acid of importance is the cysteine or selenocysteine residue that serves as a covalent linkage between the bis-MGD cofactor and the protein (Figure 41). Given the diversity of FDHs, the preference to encode for a selenocysteine residue has been discussed to reflect an evolutionary advantage. Selenium as being essential for bacteria was discovered by Pinsent in 1954,654 whereby the activity of FDH was dependent on the presence of Se, and was later discovered to be in the form of a single SeCys residue encoded by the unique codon UGA.655 Selenocysteine offers a lowered pKa and improved performance as a nucleophile relative to cysteine that results in enzyme exhibiting an increased catalytic reactivity.656 These considerations compelled Axley and co-workers in the early 1990s to investigate the essentiality of the residue in E. coli FdhF, by performing site-directed mutagenesis.585 They showed that exchange of the active site selenocysteine to cysteine (U140C) in E. coli FdhF resulted in a reduction of kcat by a factor of 300, and a decrease in KM formate relative to a constant KM with BV as an electron acceptor (Tables 7 and 8). While the reduction in kcat is consistent with differences in chemical properties of Se vs S, it also affected the relative affinity of the substrate formate at the active site.585 Despite the preliminary studies investigating the role of selenocysteine U140 in E. coli FdhF, the role of the amino acid ligand in catalysis is a matter of debate and not clear to date. As discussed in detail below, mainly two models exist, one in which the proteinaceous ligand remains bound to the Mo ion during catalysis and therefore has no direct role in the catalytic mechanism.632 In the other model, the amino acid ligand is displaced from the Mo ion, thereby providing a coordination site for the substrate and it temporarily serving as a second coordination sphere ligand.646 While in the MoV state of E. coli FdhF, it was shown that the Mo ion is hexacoordinated with the amino acid ligand,629 different data exist on the reduced MoIV state. In a reinterpretation of the crystallographic data of the formate-reduced E. coli FdhF enzyme it was suggested that the selenocysteine residue is dissociated from the Mo ion and is shifted 12 Å away.646 However, a recent crystal structure of the formate-reduced D. vulgaris Hildenborough W-FDH showed a hexacoordinated metal ion.599 It needs to be considered that in these structures the oxidation state of the Mo/W ion is speculative (given the presence of multiple redox cofactors) and that none of the structures was solved in a high enough resolution to firmly establish the coordination of the substrate formate. On the contrary, different results have been obtained in XAS studies using different enzyme sources. In a recent study of the R. capsulatus enzyme it was suggested that in the MoV and the MoVI state the cysteine is a ligand to the Mo ion, while in the formate-reduced MoIV state (in the absence of N3–), the cysteine was displaced from the metal and instead a bond reflecting a Mo–O distance was present.638 Contrary to this study, EXAFS data of the E. coli and D. desulfuricans enzymes in the oxidized and reduced states were interpreted with a hexacoordinated Mo ion.657,658 Here, it needs to be pointed out that the data of the E. coli were derived in the presence of high concentrations of the inhibitor N3−, which influence the binding of formate (discussed in more detail below). As for the D. desulfuricans enzyme the presence of the sulfido ligand is not clear, since it was not assigned by the EXAFS study. To probe the displacement of the amino acid ligand, inhibition studies with iodoacetamide were performed, an alkylating agent that reacts with “free” thiols. For both the E. coli and the R. capsulatus enzymes an alkylation of the selenocysteine or cysteine ligand was identified, notably only in the formate-reduced and NO3−/N3− inhibited enzymes, but not in the oxidized enzyme.585,598 However, no alkylation of the selenocysteine residue could be identified in the W-containing D. vulgaris enzyme.599 The E. coli FdnGHI enzyme likewise is not alkylated when incubated with formate.630 Heider and Böck speculated that this difference might be correlated to electron transfer from the reduced metal center to the [4Fe-4S] cluster and further to the final electron acceptor, since in some enzymes the MoV state is more stable than for other enzymes.659 A fast reoxidation of the enzyme from the MoIV to the MoVI state might protect the enzyme from alkylation, while in enzymes with a long-lived reduced state, the amino acid ligand has a higher probability to be alkylated. In conclusion, while hexacoordination of the Mo ion in its MoV and MoVI states is established now, the coordination of the MoIV state needs to be clarified in future studies. 4.3.1. Amino Acids in the Second Coordination Sphere. In addition to the active site selenocysteine or cysteine residue, primary components of the second coordination sphere of all metal-containing FDHs include a highly conserved arginine and a histidine residue (Figure 41B) that have been proposed to ensure optimal substrate binding to and proton transfer away from the active site.575 Site directed mutagenesis studies on R. capsulatus FDH revealed that in the H387M variant (equivalent to H141M in Figure 41B), the kcat remained similar to that of wild-type enzyme, suggesting a similar orientation of H387 and M387 (Tables 7 and 8).598 However, in this variant, the pH optimum was lowered to 7.5 with a 19-fold increase in KMformate. The replacement of H387 with a lysine reduced the activity to 3% of that of the wild-type enzyme, accompanied by an increase in KMformate and a shift of the pH optimum to 8.0. When R587 (equivalent to R333 in Figure 41B) was replaced by a lysine, retaining the positive charge of the side chain, ~ 30% of wild-type activity was obtained, however, accompanied by an increase in KMformate and a shift in the pH optimum to 8.0 (Tables 7 and 8). These results show that H387 influences the pH optimum of the reaction, while the drastic increase in KMformate reveals a role for R587 in substrate binding at the active site. 4.3.2. Substrate Channels in the Outer Coordination Sphere. The first FDH structure characterized by X-ray diffraction was the monomeric FdhF protein from E. coli, which was solved both in the oxidized (2.8 Å resolution) and formate-reduced (2.3 Å resolution) state (Figure 43A and Figure 44A).627 The active site Mo ion was initially interpreted to carry a hydroxyl ligand; however, later the Mo–OH bond was reinterpreted as being a Mo-SH one.646 Two other catalytically important amino acids in the active site were identified to be H141 and R333 (Figure 41B). These two residues are strictly conserved in all Mo- and W-containing FDHs, and their involvement in substrate binding and/or proton abstraction from the substrate has been proposed.598 After substrate oxidation, electrons are transferred from MoIV to the nearby [4Fe-4S] cluster. Here, a conserved lysine K44 is located between one pterin ring of bis-MGD and the [4Fe-4S] cluster (Figure 41B), and an involvement of this residue in the electron transfer reaction has been suggested (Figure 43A and Figure 44A).646 The second structure of an E. coli FDH enzyme is the one of FdnGHI (EcFDH-N) which was solved as a “mushroom shaped” (αβγ)3 heterotrimer of trimers, with subunits of 113, 32, and 21 kDa, respectively.660 The crystal structure of FDH-N showed that the catalytic subunit of FdnG, composed of 982 amino acids, has an overall fold similar to the structure of FdhF. The substrate access funnel is opening down to the active site Mo ion, constructed of predominantly positively charged residues thus facilitating substrate binding, and likewise appears similar to FdhF (Figure 44B). The Mo ion of oxidized FdnG binds the bis-MGD cofactor, while additionally coordinated by U196. A sixth ligand in the oxidized enzyme was initially modeled as a hydroxy species (at 2.2 Å), but in light of the reinterpretation of the FdhF structure, the sixth ligand is most likely a terminal sulfido species.646 Overall the Mo coordination sphere closely resembles that seen in oxidized FdhF (Figure 44A), with an arginine and a histidine present in the active site, thus implying a conserved mechanism involving the same amino acids for formate oxidation. At the active site of D. gigas FDH, the W ion is bound to two MPT moieties, to a selenocysteine, and to an inorganic S or O ligand (Figure 44C).634,635 The crystallographic data favor a S atom for the sixth ligand, although the resolution of the data did not permit unambiguous discrimination between O and S. As has been shown for other enzymes from the same family, the bis-MGD cofactor is buried in the protein, stabilized by an extensive network of hydrogen bonding interactions. The conserved lysine K56 bridges the pterin of the bis-MGD cofactor and the proximal [4Fe-4S] cluster. In vicinity of the buried W ion H159 and R407 are found (Figure 44C). Formate access to the active site could be facilitated by a channel of charged residues, including H159, R156, H423, H963, H376, K91, K444, K445, A587, and A407. CO2 release from the active site might be facilitated by a hydrophobic channel, including V412, H159, W730, and Y428. The crystal structure of the tungsten and SeCys containing FdhAB from D. vulgaris Hildenborough was solved in the oxidized and reduced form (Figure 43B and Figure 44D).599 In the oxidized form, the W ion is hexacoordinated to four S atoms from the two dithiolenes, a Se atom from U192 and an indeterminate sulfur ligand (W-SH or W = S). The catalytic pocket and respective entrance channel are positively charged, and three glycerol molecules haven been modeled in this cleft. In the vicinity of the W ion there are two strictly conserved amino acids: H193, establishing a π-interaction with the Se atom of U192, and R441, which establishes a hydrogen bond with a cocrystallized glycerol molecule. A similar π-interaction is observed in the FDH from D. gigas, among respective residues H159, U158, and R407 (Figure 44C).634 The glycerol closest to the W ion is hydrogen-bonded by S194 and H457, the second glycerol is oriented by R441, T450, and a water molecule oriented by Y462. A third glycerol is found at the channel entrance, oriented by a backbone contact including G207 and water molecules that connect with W492 and R206. These glycerol molecules probably occupy formate binding sites and may shed light on the formate shuttling mechanism. In the formate-reduced form of FdhAB from D. vulgaris Hildenborough, a conformational rearrangement was observed, involving I191, U192, and H193 (Figure 43B).599 The U192 Cα atom shifts up to 1.00 Å from the position occupied in the oxidized form while the Se atom remains coordinated to the W ion. Further, the catalytically relevant H193 moves away from the active site and a water molecule occupies the position of the imidazole ring, establishing a hydrogen bond with G442. Together, H193 and R441 are oriented in a cis conformation, relative to the trans orientation observed in the oxidized structure and other FDH structures. This shift also resulted in assignment of a hydrophobic channel distinct from the hydrophilic one identified in the oxidized state (Figure 44D). The U192 loop shift is propagated to neighboring amino acids from different domains that promotes the tilt of the side chains of W533, F537, and F160. Overall, neither formate nor CO2 were identified in the reduced structure, leaving the possibility that the reduced structure was solved after CO2 release from the enzyme. The cryogenic electron microscopy (cryo-EM) structure of R. capsulatus FDH revealed a 360 kDa dimer of FdsABGD heterotetramers in which the FdsD subunit is bound to the FdsA subunit (Figure 43C and Figure 44E).591 In the oxidized structure, the active site structurally resembles that of E. coli FdhF with six ligands coordinated to the Mo ion represented by the two dithiolene groups of the bis-MGD molecule, the C386 ligand, and a sulfido ligand that is oriented toward V592 (Figure 44E). Additionally, the NADH-reduced cryo-EM structure was also solved; however, the protocol did not result in a full reduction to the MoIV state, instead the enzyme was only partially reduced to the MoV state.591 In the MoV state, no structural changes at the bis-MGD pterin and dithiolenes were observed and C386 was present as a ligand to the Mo ion (Figure 43C). The inability of NADH to completely reduce the enzyme might either be dependent on unfavorable redox potentials of the cofactors for the back reaction or is based on the presence of 10 mM N3− that prevent complete reduction of the enzyme. Interestingly, in none of the structures clear evidence for stochiometric binding of N3− in any particular location of the EM map could be assigned. Two reactant channel could be assigned in the cryo-EM structures, starting at the Mo ion and separating at the active site residue R587 into different exits (Figure 44E). The pore of the shorter channel is mainly formed by polar and charged residues suggesting channeling of hydrophilic substrate (i.e., formate) from an entry site near FdsD to the active site. The channel-forming residues are similar to those in oxidized FdhF. The second channel bears predominantly hydrophobic residues suggesting the possibility of CO2 transport to or away from the active site. FdhF contains a similar hydrophobic channel, which is however blocked by V145 and M157 in place of glycine residues in R. capsulatus FDH. On the basis of the positioning of R587 at the cross section of both channels, it has been suggested that this residue might control gate opening to each channel thereby facilitating efficient catalysis.591,598 Intriguingly, the glycine residues are conserved in NAD+-dependent FDHs while other FDHs display larger hydrophobic residues at this position. The structural characterization of the W-containing formylmethanofuran dehydrogenase from M. wolfeii has provided considerable insight toward the importance of the second and outer coordination spheres in the FDH reaction (Figure 44F).612 The enzyme, crystallized as a dimer and a tetramer of the FwdABCDFG heterohexamer, respectively, showed a modular composition of the catalytic subunits and 46 iron–sulfur clusters bound in the tetrameric supercomplex. Interestingly, an internal 43 Å-long hydrophilic water cavity separates the bis-MGD and the amidohydrolase active sites (Figure 44F). The interface of the subunits FwdB (housing the bis-MGD) and FwdA (housing the Zn2-amidohydrolase site) that comprises the hydrophilic channel is distinct from the FdsD–FdsA interface characterized in the cryo-EM structure of RcFDH.591 However, several of the second coordination sphere residues that constitute this channel are similar, including H119, R288, N113, and N297 (MwFwdA numbering). Interestingly, a H2O molecule was found in proximity to the active site residue R288 at the interface of the hydrophobic and hydrophilic channels (Figure 44F). The lack of access by which exogenous formate might be expected to bind is consistent with the substantial differences in the substrate affinity among the isoenzymes from M. wolfeii and the Mo-containing enzyme from Methanosarcina barkeri.616,617 Despite being an enzyme that is biased toward the reduction of CO2, features of the hydrophobic channel associated with CO2 binding are not so different relative to structurally characterized FDH counterparts (Figure 44). Similar to RcFDH, DvhFDH, and DgFDH, the hydrophobic channel principally begins at the interface of the active site H387 and R587 residues (RcFDH numbering) as an elaborate cavity characterized crystallographically without H2O molecules bound.612 While the cryo-EM map of the RcFDH structure591 does not allow for a direct comparison to assess the hydrophobicity of the channel, the predicted channel is nevertheless similar to that of the crystallographically characterized DvH FDH enzyme.599 The hydrophilic channel was proposed to represent a conduit by which protons could be delivered to and from the bis-MGD cofactor in formylmethanofuran dehydrogenase.612 This is insightful, with respect to clarifying the distinct, requisite pathways to deliver protons, electrons, and substrates. The proton transfer pathway starts at the active site H119 and C118 residue in vicinity of the distal pterin, whereby the protons are aided by numerous H2O molecules toward H290 and C91 (MwFwdA numbering, Figure 44F).612 By comparison, other structurally characterized FDHs also exhibit residues at these positions that could in principle aid in proton shuttling to the active site.591,599,627 4.4. Inhibition of FDH in the First and Second Coordination Sphere Inhibition of different FDH enzymes has been studied intensively using various inhibitors, and among them the most prominent one is N3–, which is an isoelectronic molecule to CO2. Azide has been thought to be a transition analogue for the FDH reaction, and is most often used during the purification of FDHs to protect the enzyme from oxidative damage and loss of the sulfido ligand.638 Inhibition of FDH using pseudohalides of varying electron donor strengths (e.g., N3−, OCN−, NO2−, and NO3−) have been precursorily characterized in solution-based, in vitro assays.594,626,652,661–664 While all inhibitors were more potent to inhibit formate oxidation, N3− and OCN− are generally reported to be mixed-type inhibitors, revealing two binding sites with one being competitive and the other binding site being noncompetitive. By comparison, NO2− and NO3− have been shown to be competitive inhibitors using formate as a substrate.661 In the crystal structure of E. coli FdhF, NO2− was proposed to be coordinated to the Mo ion.627 Recently, in a chronoamperometry study of E. coli FdhF these pseudohalide inhibitors were used to study their binding to the reduced and oxidized enzyme.665 The inhibition studies revealed that inhibitor binding is oxidation state-dependent, since potential- and formate-dependent results could be ascribed to differences in population of the oxidation state Hirst, Reisner, and co-workers concluded that N3–, OCN–, and NO3− bind directly to a vacant coordination site of the Mo ion and that these inhibitors bind more tightly to MoVI than to MoIV.665 In addition, it further was suggested that the selenocysteine residue in E. coli FdhF has to dissociate from the Mo ion to generate the vacant position to which the inhibitors can bind. 4.5. Catalytic Mechanism of Formate Oxidation While metal-dependent FDH enzymes have been studied for several decades and the E. coli FdhF enzyme was among the first molybdoenzymes to be crystallized,627 details of the reaction mechanism, involving the first and second coordination sphere remain poorly understood, and the catalytic mechanism of formate oxidation is still unclear. As will be described below, the numerous mechanisms proposed for FDH reflect a lack in clarity of the coordination environment and oxidation state of the bis-MGD cofactor. Overall, the formate oxidation mechanism is believed to be similar between Mo-containing and W-containing FDHs. To form CO2 from formate, FDH enzymes have to catalyze the transfer of one proton and two electrons (eq 3). Currently, it is assumed that this reaction is not an O atom transfer reaction, as characteristic of many Mo/W-bis-MGD family enzymes.583 It has been proposed that the reaction product is CO2 rather than bicarbonate (HCO3−), the product expected in an O atom transfer mechanism. This assumption was first proposed by Thauer and co-workers in 1975 in the characterization of the FDH from Clostridium pasteurianum666 and has been supported in 1998 by Khangulov and co-workers on EcFDH–H by determining the product formation using of 13C-labeled formate in 18O-enriched water.647 A recent study by Fourmond and co-workers applied an electrochemical approach on the DvhFDH enzyme to confirm that the substrate of FDHs for the backreaction of CO2 reduction is CO2 rather than HCO3−.592 The method consisted in monitoring the changes in activity that occur during the slow relaxation of the equilibrium between CO2 and bicarbonate.592 However, as already pointed out in 1968 by Benedict, Cooper, and co-workers,667 it still remains possible that these observations are rather a reflection of the binding of the preferred molecule to the active site, rather than what is happening directly at the active site (i.e., catalysis). It therefore might be possible that a charged HCO3− molecule is hindered from entering the active site via the hydrophobic channel, while CO2 could enter easily. At the active site cofactor, CO2 then could react with H2O to form HCO3–, which would provide the substrate for the reaction of CO2 reduction. This option has so far been neglected in most studies and should be considered, since in some structures a water molecule has been identified in the active site bound in vicinity to the conserved arginine residue, as discussed above (Figure 44). It also needs to be considered that H218O exchange in the active site of the enzyme might be slow and therefore the reason why Khangulov and coworkers did not find18O-labeled CO2 in their approach.647 In this respect, approaches that factor second and outer coordination sphere effects regarding CO2 production or formate diffusion would provide significant insight toward clarifying the formate oxidation/CO2 reduction mechanism. Overall, on the basis of different experimental data and theoretical calculations, different catalytic mechanisms for FDHs have been proposed, which we want to describe in more detail (Figure 45). As explained above, a central question is whether the amino acid ligand (selenocysteine or cysteine) remains coordinated to the Mo- or W metal during catalysis or whether it dissociates to provide a vacant site for substrate binding. Further questions relate to the involvement of the second coordination sphere residues that also serve a critical role in the reaction mechanism. The first mechanism was proposed by Heider and Böck in 1993, before the crystal structure of E. coli FdhF was described (Figure 45A).659 In this mechanism, the binding of only one pterin ligand is proposed (the existence of two dithiolene MPT groups as ligands to the metal ion were first observed in the crystal structure of the Pyrococcus furiosus W-containing aldehyde-ferredoxin-oxidoreductase in 1995),668 one oxo ligand, and the fifth ligand being the selenocysteine residue. The authors suggested that binding of formate to the active site displaces the selenocysteine ligand from the Mo ion and the selenide then initiates formate oxidation by a nucleophilic attack to the Mo-bound formyl group. Carboxyl transfer to the selenocysteine residue would releases CO2 in the next step. Following the first crystal structure of E. coli FdhF solved by Sun and co-workers in 1997627 and concurrent XAS studies,657 it was suggested that the Mo ion is coordinated by a hydroxyl group instead of a terminal sulfur atom (Figure 45B). The subsequent identification of a terminal sulfido group in the first coordination sphere of Mo and the observation that the loop containing U140 was shifted away (12 Å) from the Mo ion in the formate-reduced E. coli FdhF led to a reformulation of that mechanism (Figure 45C).646 Romao and Raaijmakers suggested that upon formate binding to the active site, U140 is displaced from the Mo ion. After formate binding to the Mo ion through an O atom, The U140 selenol anion (Se−, with R333 as counterion) would deprotonate formate at the Cα atom and two electrons are transferred to the metal center simultaneously to form MoIV. Subsequently, the proton is transferred to H141, which changes its conformation in the reduced enzyme. After CO2 release and rebinding of the selenocysteine ligand, the Mo ion is in a hexacoordinated form, with the sulfido group as the axial ligand.657 The catalytic cycle would be completed with the oxidation of MoIV to MoVI, via intramolecular electron transfer to the [4Fe-4S] center. Theoretical calculations of the activation energy for C–H bond cleavage showed that the proton abstraction is much more favored with an unbound selenocysteine (19 kcal/mol compared to 36 kcal/mol for bound selenocysteine). In this mechanistic proposal, the conserved arginine R133 facilitates formate binding and histidine H141 acts as the final proton acceptor.646 The terminal sulfido ligand of the Mo ion, on the contrary, would have no active role in formate oxidation. Another mechanism was proposed based on theoretical calculations involving selenocysteine dissociation from the Mo ion through a “sulfur shift” (Figure 45D).653 In this mechanism, Cerqueira, Gonzales, Moura, and co-workers proposed that the oxidized enzyme is hexacoordinated by the two MGD molecules plus the selenocysteine and the terminal sulfido group. When formate enters the active site, oriented by the positively charged arginine (e.g., R333 in E. coli FdhF), the repulsive environment generated would trigger a ligand shift to yield a Mo–S–SeCys moiety. This would create a vacant position to which the formate could bind. Subsequently, the S-SeCys bond is cleaved and the selenol anion abstracts the formate Cα proton (analogous to the model in Figure 45C). Carbon dioxide is subsequently released. The catalytic cycle would be completed with the oxidation of MoIV to MoVI via intramolecular electron transfer to the [4Fe-4S] center and deprotonation of the selenocysteine residue. In the absence of formate, the selenocysteine-containing loop (e.g., V139, U140, and H141 in E. coli FdhF, see Figure 43A) is reoriented, and the Mo–S–SeCys bond reformed. This proposal653 is in agreement with the increased acidity of the selenocysteine side chain (pKa ≈ 5.2), which allows the existence of a selenol anion at physiological pH values. The alkaline pKa value of a cysteine side chain (pKa ≈ 8.2) would prevent a cysteine residue having such a function. The pKa values of both amino acids have been evoked to explain why the U140C variant of E. coli FdhF shows a 300-fold activity decrease.585 In addition, the high activation energy calculated for the proton transfer from formate to selenol (SeH) is in agreement with the isotopic effect studies647 that showed that the formate C–H bond cleavage is the rate-limiting step of the catalytic cycle. The role of the sulfido ligand, however, is also unclear in this mechanism. An additional mechanism was proposed by Zampella and coworkers in a computational study (Figure 45E).669 In this mechanism formate first binds to the Mo ion by displacing the selenocysteine group, then Mo abstracts a hydride ion from formate, giving a Mo–H species and releasing CO2. Upon recovery of the pentacoordinated Mo ion, the sulfido group is protonated at the expense of the hydride species. In this mechanism, also the role of the sulfido group is unclear and a Mo–H bond is very unusual. Following theoretical assumptions, Dong and Ryde recently proposed an additional mechanism (Figure 45F).670 They proposed that the sulfido ligand abstracts a hydride ion from the substrate, yielding MoIV–SH; however, distinct from any other mechanism, carboxylate was considered binding to the cysteine ligand, forming a thiocarbonate/cysteine zwitterion. Upon charge compensation, CO2 leaves the enzyme and the Mo ion is oxidized back to the MoVI state. Finally, Hille and Niks proposed a mechanism that is based on experimental data582,632 and considers binding of formate to the second coordination sphere without contacting the Mo ion (Figure 45G).671 In their proposal, the sulfido ligand abstracts a hydride ion from formate, resulting in a two electron-reduced intermediate MoIV–SH, and CO2 is released. In this mechanism, however, the (seleno-) cysteine would not be involved and the role of the metal ion is only to provide a hydride acceptor by coordinating the sulfido ligand. 4.5.1. Conversion of FDH to a Nitrate Reductase. The sulfur shift-based mechanism to oxidize formate, described above and depicted in Figure 45A, has also been evoked to explain the nitrate reduction catalyzed by periplasmatic nitrate reductases (NR).653 The FDH and nitrate reductase active sites are surprisingly superimposable,598 with the later harboring a Mo ion coordinated by the two characteristic MGD molecules, one terminal sulfido group, and a cysteine sulfur atom.672 In addition, the nitrate reductase active site also comprises conserved threonine and methionine residues (arginine and histidine, in FDH). In the oxidized active site of both enzymes, the Mo/W ion is hexacoordinated and a sulfur shift is needed to displace the selenocysteine or cysteine residue to create a vacant position for substrate binding (i.e., formate or nitrate).589,653 The similarities between FDH and nitrate reductases were highlighted by converting RcFDH to an enzyme with nitrate reductase activity.598 This was achieved by exchanging the histidine H387 to a methionine residue and the arginine R587 to a threonine residue, both that are conserved in nitrate reductases. In addition, an additional arginine residue was inserted to the active site of RcFDH and this enzyme variant showed bis-MGD dependent nitrate reductase activity. However, the involvement of the sulfido ligand of this enzyme variant for nitrate reductase activity is not clear, since also an enzyme form containing an oxo ligand instead of the sulfido ligand was able to reduce nitrate.598 Nitrate reductases catalyzes the typical O atom transfer mechanism characteristic for all other molybdoenzymes.672 4.5.2. Second and Outer Coordination Sphere Effects on Electron Transfer. In addition to the active site Mo/W ion, the sulfido ligand, and active site residues involved in CO2 reduction and formate oxidation of FDH and formylmethanofuran dehydrogenases, additional components are critical toward the catalytic activity. For example, a feature common to all FDHs and formylmethanofuran dehydrogenases is the coordination of a [4Fe-4S] cluster that is within electron channeling distance to the bis-MGD cofactor (Figure 41). For all FDH enzymes identified to date, a [4Fe-4S] cluster is within 12 Å of the Mo/W ion, and within 6 Å of the proximal pterin of the bis-MGD (Figure 44).591,599,612,627,634,660 This iron–sulfur cluster is proposed to be the entry point of the electron transfer pathway by which electrons that are generated from formate oxidation or are needed for CO2 reduction are passed along, respectively. Reduction of this [4Fe-4S]2+/+ cluster has been observed by EPR spectroscopy, using either formate or dithionite as a reductant. These are typically conditions by which the bis-MGD cofactor can also become reduced to a MoV/WV oxidation state, which complicated the spectral analysis. Early characterization work of FDH from Methanobacterium formicicum to differentiate the [4Fe-4S] cluster from the MoV/WV signal showed that this cluster had g-values of 2.047, 1.948, and 1.911;649,650 however, it should be noted that this FDH also can bind a [2Fe-2S] cluster whose g-values would overlap.575 By comparison, reduction of the proximal [4Fe-4S] cluster in EcFDH–H (an FDH that coordinates only one [4Fe-4S] cluster) gave a similar assignment (g-values of 2.045, 1.957, and 1.840). Similar assignments have been made for the corresponding iron–sulfur clusters for several FDHs by which EPR spectral data is available.647 As a participatory component in the redox events catalyzed by the bis-MGD cofactor, perhaps it is not surprising that communication with the proximal [4Fe-4S] cluster is essential for the catalytic reaction. However, there are few examples of FDHs by which concomitant catalytic activity and loss of the proximal [4Fe-4S] cluster has been shown. For example, a EPR spectral characterization of the FDH from C. necator FDH has shown that under conditions by which the [4Fe-4S] cluster is reduced and the bis-MGD is poised in the paramagnetic MoV state, electron–electron spin coupling between the two paramagnetic centers occurs, primarily reflected in the loss of sharpness of the Mo(V)-based 1H hyperfine splitting.632 As EPR spectroscopy cannot characterize the diamagnetic oxidation states of the bis-MGD cofactor and iron–sulfur cluster, a parallel, multifaceted spectroscopic approach is required to identify all oxidation states. In addition to participation of the proximal [4Fe-4S] cluster in the catalytic activity of the bis-MGD cofactor, an additional, undercharacterized aspect of FDH catalysis relates to the two molybdopterin guanine dinucleotide ligands that coordinate the Mo/W ion. These relatively large, pterin-based ligands are noninnocent and redox active, meaning that they partake in the electron transfer reactions between the Mo/W ion and the proximal [4Fe-4S] cluster immediately preceding or following catalysis. By comparison, the longer [4Fe-4S] cluster–Mo/W distance (13–16 Å) observed relative to the shorter proximal [4Fe-4S]–(proximal) MGD distance (6–10 Å) supports the presumption that the proximal MGD ligand links the FDH redox events at the Mo/W ion with the iron–sulfur cluster electron conduit.591,599,612,627,634,660 So far, all bacterial FDHs coordinate two MGD ligands at the active site; however, distinct nucleotides were identified to be bound in place of the guanine in FDH-like enzymes from archaea (e.g., formylmethanofuran dehydrogenase from M. marburgensis).625 The reason for this remains unclear so far. 5. CO DEHYDROGENASE 5.1. Physiological Function and Structure of CODH CO dehydrogenases (CODHs) are gas-processing metalloenzymes that catalyze the reversible oxidation of CO to CO2 (eq 4). There are two unrelated classes: one is a Mo-based enzyme exclusively found in aerobic carboxydotrophs.673,674 These CODHs catalyze the reaction in the direction of CO oxidation (CO2 release) and will not be discussed here any further. The other is a class of metalloenzymes found mainly in anaerobic bacteria and archaea hosting a Ni- and Fe-containing cofactor, which catalyze both CO oxidation and CO2 reduction.675 Organisms such as Rhodospirillum rubrum (Rr) and Carboxydothermus hydrogenoformans (Ch) use the enzyme CODH to oxidize CO to CO2.676 The reaction occurs at the so-called “C-cluster”, a unique [Ni-3Fe-4S] cluster with an additional Fe ion. The enzyme is a homodimer that houses two identical C-clusters and three electron transferring metal centers: two “B-clusters” of the [4Fe-4S] type and one “D-cluster” located at the interface of the two monomers (Figure 46A). The latter is either a [4Fe-4S] cluster (e.g., in RrCODH) or a [2Fe-2S] cluster (e.g., in CODH from Desulvovibrio vulgaris, Dv).677 In Figure 46B, the proposed proton transfer pathway toward the C-cluster is shown (see Section 5.3). In acetogenic bacteria such as M. thermoacetica (Mt) and methanogenic archaea such as Methanosarcina barkeri (Mb), a homodimeric CODH is present as part of bifunctional enzyme complexes. This may be in association with the enzyme acetyl-CoA synthase (ACS) in the ACS/CODH complex, which allows acetyl-CoA formation using a CO molecule produced by the reduction of CO2 as part of the Wood–Ljungdahl carbon fixation pathway,678 or in an acetyl-CoA decarbonylase/synthase (ACDS) complex that degrades acetyl-CoA to form methane and CO2.679 The ACS/CODH complex is a large α2β2 tetramer that associates two β CODH subunits and two α ACS subunits. The CO produced from CO2 at the C-cluster is combined with a methyl group and coenzyme A (CoA) to form acetyl-CoA at the A-cluster of the ACS (α) subunit; in vitro, acetyl-CoA synthesis can also occur using CO as a substrate.680 C. hydrogenoformans illustrates the diversity of possible functions of CODHs, since its genome encodes four different isoenzymes:676 ChCODH I, which is likely involved on respiration on CO; ChCODH II, proposed to provide electrons to form NADPH; ChCODH III, which is incorporated in a bifunctional ACS/CODH complex; and ChCODH IV, proposed to be involved in resistance against O2.681 Eisen and co-workers originally proposed that the genome encodes a fifth CODH in C. hydrogenoformans,676 but the latter is more likely to be a hybrid cluster protein. 5.2. Structure and Placidity of the C-Cluster The structure of the C-cluster has been heavily debated, but the current consensus is that it comprises a [Ni-3Fe-4S] cluster coordinated to a unique fourth iron ion (Feu); the latter was also referred to as the “dangling Fe” or “ferrous component II” (Figure 47A).682 The [Ni-3Fe-4S] cluster is coordinated by four cysteines as in classical [4Fe-4S] clusters including C333, C446, C476, and C526 (DvCODH II numbering). The dangling Fe ion is coordinated by cysteine C295 and histidine H261. The catalytic active site is completed by histidine H93 and lysine K563 that are within hydrogen-bonding distances from the O atoms of the bound reactant, CO2.682 Neighboring cysteines C295 and C294 are conserved among CODHs. They are found in all clades except D, C1.3, and A4 of the classification proposed by Sako and co-workers;683 however, no CODH with verified activity lacks this cysteine residue. Mutating C294 into a serine prevented the assembly of the C cluster in the case of Mt ACS/CODH,684 and yielded a C-cluster devoid of Ni in the case of DvCODH.685 Exposure to air of DvCODH crystals yielded an alternative form of the C-cluster in which the Ni ion occupies the position of the dangling iron, which moved a little further (Figure 47B). In this configuration, cysteine C295 becomes a bridging ligand between the Fe and Ni ions and C294 coordinates the dangler ion. This state is formed reversibly, as reducing the crystals gives back the “classical” coordination of the C-cluster, and may be involved in the high resistance of DvCODH to O2 damage.686,687 Additionally, C294 may be involved in the insertion of the active site, which would explain the lack of C-cluster when it is mutated into a serine. This hypothesis is also consistent with the fact that mutation of H261 and C295 yield an enzyme devoid of Ni.688 Histidine H261 and C295 are ligands to the dangling Fe ion in the usual form of the C-cluster (Figure 47B), but bind the Ni ion in the oxidized C-cluster. The role of the cysteine ligands to the C-cluster was investigated by site-directed mutagenesis on RrCODH. Cysteines C338, C451, C481, and C531 (corresponding to C333, C446, C476, and C526 in ChCODH II) were mutated into serine or alanine residues.689 None of the mutations affected the incorporation of Ni; however, most of the mutants lost their activity, except the C451S and C531A variant, which retained ~1.4% and 0.1% of CO oxidation activity, respectively. The exact reason for the decreased activity is not known yet, but redox effects may be involved. While the EPR signature of Cred2 (see below) was observed in the reduced C451S variant, no Cred1 could be detected under the redox conditions in which the wild-type RrCODH usually gives Cred1.689 The C-cluster of CODHs may exist in at least three redox states: the fully oxidized, inactive, and EPR-silent “Cox” state, which is reduced at potentials of −100 mV to the paramagnetic “Cred1” state (g = 2.01, 1.81, and 1.65). Below approximately −500 mV, the “Cred2” state is formed (g = 1.97, 1.87, and 1.75), which is two electrons more reduced than Cred1. An EPR-silent, singly reduced state “Cint” has been postulated to exist as a transition between Cred1 and Cred2, but it was never isolated.690 The consensus is that Cred1, Cint, and Cred2 are catalytic intermediates.675,691 The proposed mechanisms all assume that the structure of the CO2-bound enzyme (Figure 47A) corresponds to a catalytic intermediate and that the binding of CO2 is followed by the breaking of the bond between the C and the O coordinated to the dangling Fe ion, yielding a CO-bound Ni ion and a hydroxo ligand on the dangling Fe (Figure 48). The main dispute for the time being is the nature of the Cred2 state. According to the model by Jeoung and Dobbek,682 Cred2 corresponds to a Ni0 state with a hydroxo ligand on the dangling Fe and no extraneous ligand on the nickel (this state is indicated by Cred2* in Figure 48).682 On the basis of DFT computations, Fontecilla-Camps, Amara, and co-workers have proposed instead that the Cred2 state is a Ni2+ hydride (this state is indicated by Cred2† in Figure 48), and that the CO2-bound structure is the result of the unusual insertion of CO2 into the Ni–H bond,692 even though insertion of CO2 into a M-H bond usually results in the production of formate.693 Bruschi and coworkers have recently used density functional theory (DFT) calculations to investigate the binding of CO2 to different redox states of the active site C-cluster; their results appear to be incompatible with the presence of a Ni–H species in Cred2, thus favoring Cred2* over Cred2†.694 They calculated that the binding is governed by the protonation states of histidine H93 and lysine K563 (equivalent H93 and K563 in ChCODH II, Figure 47), and opened the possibility that CO2 may also bind to the Cint redox state. 5.3. Substrate Access and Product Egress The diffusion of CO and CO2 has been investigated both in monofunctional CODH and in the bifunctional ACS/CODH complex. In ACS/CODH, the site of CO production and the site of CO utilization are over 70 Å apart. As CO does not leak to the solvent when the ACS/CODH complex produces acetyl-CoA using CO2 as a substrate, a gastight channel was proposed that would guide the CO molecule from the C-cluster toward the A-cluster, where CO is condensed with a methyl group and coenzyme A.695,696 The search for cavities in the X-ray structure of ACS/CODH and the identification of Xe binding sites680,697,698 defines the long hydrophobic channel that connects the four active sites in the tetrameric ACS/CODH complex (Figure 49). This channel is blocked but the A-cluster is accessible to the solvent for methyl transfer in the so-called open conformation of ACS/CODH.698,699 Support for these channels transporting CO from the C-cluster to the A-cluster comes from the observation that mutations of the conserved alanine residues that line the putative channel in the ACS subunit of Mt ACS/CODH (A110C, A222L, and A265M) only have an effect of the rate of acetyl-CoA production when CO2 is used as a substrate.700,701 The mutations have a no significant effect on the CO oxidation activity, and only a moderate effect on the rate of acetyl-CoA production from CO. The acetyl-coA synthesis from CO is strongly inhibited by CO in wild-type ACS/CODH, but not in the three channel variants; this is explained by the possibility that part of the activity in wild-type ACS/CODH is due to CO molecules entering at the C-cluster and crossing the channel toward the A-cluster, combined with CO crowding in the channel regulating the flow of CO from the C-cluster.700 The question of how CO2 reaches the C-cluster of ACS/CODH is less clear. It has been suggested that CO2 may enter the enzyme at the level of the A-cluster (which would involve transport against the opposite flux of CO),699 or via the water channel along the ββ subunit interface,703 or along dynamically formed pathways that are not detected as clear cavities in the crystal structure.704 The latter hypothesis has been investigated in molecular dynamics (MD) calculations, according to which CO2 transport is controlled by two strictly conserved histidine residues, H113 and H116 (H93 and H96 in ChCODH II, Figure 47). Upon reduction of CO2, a hydrogen-bond network in the active site pocket becomes rigid enough to prevent CO from exiting the protein via the CO2 access pathway, and directs CO from the C-cluster to the A-cluster.704 In the CODH subunit of Mt ACS/CODH, histidines H113 and H116, along with H119 and H122, asparagine N284 and lysine K587 were proposed to be involved in proton transfer between the solvent and the C-cluster (H93, H96, H99, H102, N262, and K563 in ChCODH II, see Figure 46B). Variants H116A and H122A did not show catalytic turnover but activity could be restored by introducing a histidine as a neighboring amino acid.684 On the basis of a series of single and multiple site-directed mutagenesis replacements, Kim and co-workers proposed that protons first go to histidine H113 or lysine K587, then necessarily H116, then either H119, asparagine N284 or a water molecule, and finally to the solvent-exposed histidine H122. Additional hydrophobic cavities have been identified through calculations in the monofunctional CODH from C. hydrogenoformans and R. rubrum.705,706 Structure comparison suggests that bulky residues that are present in ACS/CODH block some of the pathways that are functional in reactant diffusion in ChCODH and RrCODH, and whose presence in the ACS/CODH complex would prevent effective channeling of CO from the C-cluster to the A-cluster.698 5.4. Diversity of CO Dehydrogenase Figure 50 represents a sequence alignment of the main CODHs that have been studied so far, with the exception of the ACS/CODH from M. barkeri. All amino acid residues that have been modified by mutagenesis so far are indicated using red backgrounds, and the amino acids within 8 Å of the Ni ion of the active site are framed in red. Figure 50 shows that the environment of the active site is highly conserved, with only a few positions with nonconserved amino acids, and a few more with small variations, notably at position 293 (with isoleucine, leucine, or methionine), 311 (with asparagine, histidine, or serine), 312 (with tyrosine, serine, histidine or phenylalanine), 331 (with valine or tyrosine), 444 (with valine, alanine, or cysteine), 478 (with alanine, serine, or threonine), 529 (with asparagine or cysteine), 560 (with methionine, glutamine, or tyrosine), 561 (with histidine, serine, or threonine), and 561 with serine, histidine, or threonine). Positions 477, 479, and 559 are not conserved. In spite of this high sequence identity, the properties of the CODH appear to change greatly from one CODH to another. Regarding the Km for CO, the values range from 20 nM for ChCODH IV to 30 μM for Rr CODH, with other values between 0.5–3 μM for DvCODH, ChCODH I, and TcCODH I and II from Thermococcus sp. AM4 (Tc).681,707–710 To date, no clear pattern has been noticed relating the sequence and the values of Km, and no mutation has been published that affects the Km for CO. Km values for CO2 reduction are much higher than those for CO oxidation, and in some cases they could not be determined because it was not possible to saturate the enzyme with CO2. They range from 0.2 mM for Rr, 0.47 mM for TcCODH I, to values undetermined but larger than 2 mM (ChCODH I) or 7 mM (TcCODH II).709–711 Similarly to CO, no mutation yet is known to affect the Km for CO2. CODHs have been said to be highly O2-sensitive,675 but in fact the reactivity of CODHs with O2 varies greatly, from RrCODH being entirely destroyed by a 10 s exposure to O2712 to DvCODH, losing only 70% of its activity upon aerobic purification,687 and reactivating almost fully upon reduction after exposure to O2,686 a property that is reminiscent of the well-known reductive activation of [NiFe]-hydrogenases (Section 3.1) and certain [FeFe]-hydrogenases (Section 3.2). ChCODH IV requires much higher concentrations of O2 than the other CODHs to lose 50% of its activity, but unlike DvCODH it does not reactivate upon reduction.681 This has been attributed to the presence of a bulky phenylalanine residue on the backside of the C-cluster (F307), in the position corresponding to S312 in ChCODH II (Figure 50). However, TcCODH I and II, which also feature a phenylalanine in the same position (F346 and F322, respectively), do not share the exceptionally high resistance of ChCODH IV against O2.709 Perhaps the best understood reactivity with O2 is that of DvCODH, which harbors a [2Fe-2S] cluster as the D-cluster, unlike the other CODHs characterized so far, which hold a [4Fe-4S] cluster. The [2Fe-2S] type D-cluster of DvCODH is at least partially responsible for the high O2 resistance of DvCODH, since upon mutation to a “classical” [4Fe-4S] cluster, the enzyme loses the ability to be purified aerobically.687 Deletion of the D-cluster also prevents C-cluster assembly,685 which shows that the maturation of the active site is also sensitive to either long-distance effects, or long-range electron transfer. 6. CONCLUDING REMARKS Throughout Sections 2–5, we have seen how the activity of gas-processing metalloenzymes (GPMs) is defined by second and outer coordination sphere effects. As noted in Section 1, the optimized catalysis of GPMs has been afforded through tailored evolution of the active site cofactors from abiotic minerals and peptide “nests” coordinating these inorganic microcompartments.15–20 Several factors that drive second and outer coordination sphere effects in nitrogenase, hydrogenase, FDH and CODH have been described herein. It will be difficult to understand GPMs without knowledge about the structure and electronic properties of the metallic cofactors; however, we emphasize that bioinorganic chemistry must explore beyond the first coordination sphere. Concluding this article, we will highlight a number of common motifs including mass transport (Section 6.1), redox leveling (Section 6.2), and the stabilization of reactive geometries (Section 6.3). The developed concepts may inspire the design of biomimetic catalysts for the transformation of N2, H2, CO2, and CO as critical feedstock in “green” energy conversion. 6.1. Mass Transport Under ambient conditions, N2, H2, CO2, and CO are small volatile molecules with mostly hydrophobic properties. Such reactants approach the active site cofactor by molecular diffusion. Wolfenden and co-workers discussed how enzymatic catalysis is occasionally limited by mass transport, demonstrating the efficiency of GPMs like superoxide dismutase and carbonic anhydrase.713 At the example of hydrogenase, FDH, and CODH we saw that proteinaceous “gas filters” influence the catalytic rates and substrate selectivity (Sections 3–5). While hydrophobic channels and pockets are routinely identified in xenon- or krypton-pressurized protein crystals,714 the collective motion of the protein fold and packing defects715 complicate the identification of individual “gate keepers”, that is, amino acid residues of key importance. From the viewpoint of synthetic chemistry, we note that it may not be necessary to install molecular gas filters. In recent years, redox hydrogels have been demonstrated to facilitate enzymatic turnover in the presence of gaseous inhibitors, for example, by reducing O2 before it reaches the enzyme.716–718 Proton transfer is another mass transport phenomenon critical to the performance of GPMs. In Section 1, we compared the basic reactions catalyzed by nitrogenase, hydrogenase, FDH and CODH (eq 1–4). Each reaction includes at least one proton. Protons are transferred via water molecules and the hydrophilic groups of a protein when hydrogen-bonded or in hydrogen-bonding distance (2.7–3.3 Å).719 These criteria make proton transfer pathways extremely reliant on protein structural dynamics. In AvNifDK of the Mo-nitrogenase, second coordination sphere residues R96, Q191, and H195 (analogous to K83, Q176, and H180 in AvVnfDGK of the V-nitrogenase) are potentially involved in proton transfer and hydrogen bonding to catalytic intermediates that replace the bridging sulfide ligand S2B (Figure 14). Additionally, the homocitrate ligand may be involved,232 but any outer coordination sphere effects on proton transfer have yet to be unraveled. The cysteine residue that coordinates the active site cofactor of nitrogenase (C275 in AvNifDK, C257 in AvVnfDGK) is not involved in proton transfer, which is a marked difference from hydrogenase where cysteines of the first and second coordination sphere play key roles in proton transfer.297 In [NiFe]-hydrogenase DvMF, nickel-coordinating C546 accepts a proton after heterolytic H2 splitting. Glutamic acid residue E34 represents a proton relay before the proton is released into bulk water (Figure 22). In [FeFe]-hydrogenase Cp1, cysteine C299 close to the distal iron ion (Fed) of the H-cluster functions as a relay site between cofactor and a ~ 20 Å proton transfer pathway including two glutamic acid residues (E279, E282), a serine residue (S319), and a small water cluster (Figure 39). In the primary sequence, C299 is followed by C300, which binds to the [4Fe-4S] part of the H-cluster. A similar motif can be found in EcFDH where Mo-binding selenocysteine U140 is followed by histidine H141 (Figure 42), the later which has been suggested to play a key role in proton transfer. In the outer coordination sphere, R333 may be involved in proton transfer, but the molecular details of this event are yet to be understood.575 Similar to FDH and nitrogenase, histidines are the most likely proton relay sites in CODH. On the basis of biochemical experiments with various amino acid variants of ChCODH II, a trajectory between the catalytic C-cluster and bulk solvent including H93, H96, H99, H102, N262, and K563 has been proposed (Figure 48). Hydrogenases, on the contrary, rely on cysteine and glutamic acid residues primarily,297 resulting in an overall pKa decrease of 2–3 compared to histidine-based proton transfer pathways. This may reflect the necessity for a tightly controlled proton transfer in systems that exclusively catalyze H2 oxidation and proton reduction. 6.2. Redox Leveling While proton transfer demands short distances, the probability for electrons to tunnel across distances as far as 18 Å is high enough to enable electron transfer within biologically relevant time scales.720 Conductive wires are formed by iron–sulfur clusters in nitrogenase, hydrogenase, FDH, and CODH, guiding electrons back and forth between the active site cofactor and redox proteins like ferredoxin, flavodoxin, or cytochrome. The “accessory” clusters of GPMs are determined by first and second coordination sphere effects to ensure directional electron transfer. For example, the replacement of a cysteine by a histidine ligand at the distal [4Fe-4S] cluster in [FeFe]-hydrogenase has a strong effect on activity.418 In special cases, the electronic properties are flexible and controlled by the enzyme. Examples include the P-cluster in nitrogenase that changes its geometry upon reduction (Figure 3) or the proximal iron–sulfur cluster in O2-tolerant [NiFe]-hydrogenases (Figure 24). In [FeFe]-hydrogenase, a protonation event at the backend of the H-cluster has been shown to affect the redox potential (Figure 39). These special cases demonstrate how enzymes modulate electron transfer in a flexible way. Compared to the branching chains of metal clusters found in electron bifurcating/confurcating enzymes,265,266,721 the GPMs discussed in this review are wired in a straightforward fashion. The sophistication, however, lies in the spatiotemporal coupling of redox and protonation events at the active site cofactor, which allows for an accumulation of charges at mild potentials (“redox leveling”). This is clearly illustrated in the Lowe-Thorneley model for N2 reduction (Figure 6) and holds true for H2 turnover as well (Figure 21 and Figure 34). The mechanistic details in FDH and CODH are less well understood, but similar proposals have been considered (Figure 45 and Figure 48). How do GPMs facilitate proton-coupled electron transfer? While the electron transfer chains are hard-wired, the distance constraints of proton transfer allow for flexible fine-tuning, for example, by small changes in protein structure as demonstrated in [NiFe]- and [FeFe]-hydrogenase.297 The data discussed in this review now suggest that metalloenzymes initiate PCET by guiding protons to the cofactor. Here, the catalytic redox reaction is induced only upon arrival of the proton (and other reactants like N2, CO2, or CO). As tunneling is a quantum mechanical phenomenon, the formal requirements of PCET are fulfilled.22–24 From the viewpoint of synthetic chemistry, we note that molecular wiring may not be necessary. For example, when biomimetic Ni-complexes were deposited in a conductive hydrogel, redox events could be triggered by external electric fields.722 At variance, the design of proton transfer pathways seems to be imperative for efficient electron transfer and catalysis.556 6.3. Stabilization of Reactive Geometries Second coordination sphere effects are not limited to mass transport and redox leveling. In nitrogenase, hydrogen-bonding between H195 and R96 with bridging ligands in position S2B or between Q191 and terminal ligands (Figure 14) stabilizes the reactive geometry of the active site cofactor. Ligands include N2, CO, or H. Positions S3A and S5A could be catalytically relevant as well; in AvNifDK of the Mo-nitrogenase, ligands likely derived from the bound N2 entities have been identified as being hydrogen-bonded to backbone amines of G356 and G357 at position S3A, and the side chain of R96 at position S5A (Figure 12). In hydrogenases, potential hydrogen-bonding contacts inspired the assignment of CO and CN− ligands. This includes S502, P478 and R479 in [NiFe]-hydrogenase DvMF (Figure 19). Arginine 479 may also form a frustrated Lewis pair (FLP) with the Ni–Fe cofactor and has been suggested to be part of the “outer-shell canopy”, an alternative proton transfer pathway.360 The mechanism shown in Figure 22 favors C546 as FLP but the role of R479 in polarizing the bridging ligand cannot be excluded.361 Arginine 96 and H195 at the nitrogenase cofactor may have a similar function (Figure 14). In [FeFe]-hydrogenase, the ADT ligand serves as “internal” FLP; however, hydrogen-bonding with the adjacent C299 and M497 residues (Figure 38) was found to be of key importance for efficient proton transfer and catalysis. Another methionine, M353, may destabilize the μCO ligand and contribute to the flexible geometry of the diiron site (Figure 37). Around the proximal CN− ligand, strong hydrogen bonding with P231 and S232 was confirmed whereas any contacts around the distal CN− ligand remain speculation (Figure 36). The catalytic mechanism of FDH is controversial and various proposals coexist (Figure 45). The involvement of R333 and H141 seems to be accepted, which represents a similar mechanistic element to those employed by [NiFe]-hydrogenase and nitrogenase. Histidines play important roles in CODH, across all coordination spheres. In the reduced, CO2-bound state of ChCODH II, H93 and K563 are in hydrogen-bonding distance to the substrate, stabilizing the reactive geometry (Figure 47). Lysine K563 is turned into a nickel ligand in the oxidized state, switching between the first and second coordination sphere.677 This behavior is reminiscent of C367 in [FeFe]-hydrogenase CbA5H (C299 in CpI) binding to Fed in the O2-inhibited state (Figure 35). From the viewpoint of synthetic chemistry, hydrophilic groups in the second coordination sphere should have dual functions. Our data emphasize that histidines, arginines, and cysteines serve as hydrogen-bonding partners to substrates like N2, H, CO2, and CO, stabilizing reactive geometries. However, without contact to residues involved in proton transfer in the outer coordination sphere (Section 6.1), efficient catalysis cannot be achieved. ACKNOWLEDGMENTS S.T.S. was supported by the Deutsche Forschungsgemeinschaft (DFG) via the SPP 1927 priority program “Iron–Sulfur for Life” Grant No. STR1554/5-1. Jifu Duan is acknowledged for helpful comments on Table 5. Guodong Rao is acknowledged for the simulation of EPR spectra in Figure 32. The work of C.L. and V.F. was supported by Centre National de la Recherche Scientifique (CNRS), Aix Marseille Université, Agence Nationale de la Recherche (ANR-11-BSV5-0005, ANR-12-BS08-0014, ANR-14-CE05-0010, ANR-15-CE05-0020, ANR-16-CE29-0010, ANR-17-CE11-0027, ANR-18-CE05-0029), and the Excellence Initiative of Aix-Marseille University – A*MIDEX, a French “Investissements d’Avenir” programme (ANR-11-IDEX-0001-02). The work of S.L. was supported by the Deutsche Forschungsgemeinschaft (DFG) via the SPP 1927 priority program “Iron–Sulfur for Life” Grant No. E1171-15-2 and by the EXC 2008/1 (UniSysCat), funded by Germany ′s Excellence Strategy – EXC 2008-390540038 – UniSysCat. The work of S.H. and H.O. was supported by the Grant-in-Aid from the Japan Society for the Promotion of Science (JSPS Category B, No. JP21H02060 to S.H. and No. JP20H03215 to H.O.). M.W.R. and Y.H. were supported by NIH-NIGMS grants GM67626 (to M.W.R. and Y.H.) and GM141046 (to Y.H. and M.W.R.), which funded research related to nitrogenase assembly and catalysis, respectively. The authors were also supported by the Department of Energy grants DOE (BES) DE-SC0016510 (to Y.H. and M.W.R.) and DE-SC0014470 (to M.W.R. and Y.H.), which funded work related to the mechanistic investigation of ammonia formation by nitrogenase engineering and hydrocarbon formation by nitrogenase hybrid systems, respectively. In addition, the authors were supported by NSF grants CHE-1904131 (to M.W.R. and Y.H.) and CHE-1651398 (to Y.H.) that supported work related to CO activation by nitrogenase and CO2 reduction by nitrogenase Fe proteins, respectively. Sven T. Stripp graduated in Biology and Biotechnology and received his Ph.D. degree from Ruhr-Universität Bochum, Germany. Afterwards, he moved to Freie Universität Berlin for postdoctoral studies on surface-enhanced vibrational spectroscopy. Since 2015 he has been a group leader in the Department of Physics where he develops in situ FTIR, Raman, and UV–vis spectroscopic approaches to investigate metalloenzymes under biologically relevant conditions. He habilitated in Physical Chemistry in 2021. Benjamin R. Duffus received his B.A. in chemistry (cum laude) from Concordia College (Moorhead, MN) in 2005. Following postgraduate study in synthetic inorganic chemistry at Texas A&M University, he completed his Ph.D. at Montana State University in 2014, focusing on the biosynthesis of complex iron–sulfur clusters. Since 2014, he is a postdoctoral research associate in Molecular Enzymology at the University of Potsdam, Germany. His interests blend biochemical and spectroscopic approaches toward mechanistic and biosynthetic studies of metalloenzymes, with a focus toward complex, iron–sulfur cluster molybdoenzymes. Vincent Fourmond received his Ph.D. at the Université Paris Descartes in 2007. After two postdocs in the groups of Christophe Léger and Vincent Artero, he became permanent CNRS research associate in 2011. His interests lie in using kinetics to understand the mechanism of redox metalloenzymes and in the development of the general purpose data analysis software QSoas. Christophe Léger received his Ph.D. from the University of Bordeaux and was a postdoc in the group of Fraser Armstrong from 1999 to 2002. He is “Directeur de Recherche” at CNRS. His interests lie in kinetic and mechanistic studies of complex metalloenzymes. Silke Leimkühler received her Ph.D. in 1998 from Ruhr-Universität Bochum, Germany. After postdoctoral studies at Duke University, USA, she returned to Germany as a research group leader to the Technical-University of Braunschweig. In 2004 she accepted a Junior-Professor position at the University of Potsdam, Germany, and since 2009 she is a full professor in Molecular Enzymology at the same institute. Her research interests focus on molybdenum cofactor biosynthesis, molybdoenzyme enzymology, and iron–sulfur cluster assembly. Shun Hirota received his Ph.D. from the Graduate University for Advanced Studies in Japan. After postdoctoral studies in Japan and the US, he joined Nagoya University as an assistant professor in 1996 and became an associate professor at Kyoto Pharmaceutical University in 2002. He was invited to Nara Institute of Science and Technology as a full professor in 2007. His research interests include structures, functions, and reaction mechanisms of metalloproteins. Yilin Hu received her Ph.D. in Biochemistry from Loma Linda University, USA. She was a postdoctoral fellow at the University of California, Irvine, and is currently Professor at the same institute. She focuses on studies related to nitrogenase mechanism and assembly, with an emphasis on the genetic manipulation of nitrogenase systems. Andrew J. Jasniewski received his B.S. degree from the University of Wisconsin—Madison working with Thomas Brunold on functional models of the Mn-dependent superoxide dismutase. He then moved to the University of Minnesota to study the structures and spectroscopy of nonheme diiron enzymes and related model complexes with Lawrence Que Jr., receiving his Ph.D. degree in 2017. He currently works at the University of California, Irvine with Markus Ribbe and Yilin Hu on the biochemistry and spectroscopy of nitrogenase. Hideaki Ogata received his Ph.D. from Kyoto University, Japan in 2003. After a short period of the postdoctoral research at Himeji Institute of Technology, he joined the Max Planck Institute for Chemical Energy Conversion, Mülheim an der Ruhr, Germany where he worked as a group leader until 2017. He then moved to Hokkaido University and later to Nara Institute of Science and Technology. He is now a professor at University of Hyogo, Japan. His research interests are structural and spectroscopic studies of metalloenzymes. Markus W. Ribbe received his Ph.D. in Microbiology from the University of Bayreuth, Germany. He was a postdoctoral fellow at the University of California, Irvine, and is now Chancellor’s Professor at the same institute. He focuses on the mechanistic investigation of nitrogenase catalysis and assembly by combined biochemical, spectroscopic, and structural approaches. Figure 1. Definitions. The figure shows the active site of [FeFe]-hydrogenase (PDB ID 4XDC). For this enzyme, the first coordination sphere (light magenta) is defined by the two iron ions of the catalytic cofactor and its ligands. The second coordination sphere (white) comprises a [4Fe-4S] cluster as well as (iv) hydrophobic interactions and (v) hydrogen-bonding contacts with the protein fold. Outer coordination sphere effects include (i) electron transfer, (ii) proton transfer, (iii) hydrophobic gas channels, and (vi) electric field effects. Figure 2. Crystal structure of the NifH:NifDK complex (PDB ID 1N2C). The relevant metalloclusters are highlighted and the proposed direction of electron flow through the enzyme is depicted as a yellow arrow. The coloring scheme is as follows: NifD (green cartoon), NifK (blue cartoon), NifH (orange and gray cartoon), Fe, orange; S, yellow; C, gray. Reprinted (adapted or reprinted in part) with permission from ref 50. Copyright 2020 American Chemical Society. Figure 3. Structure of the P-cluster. (a) Dithionite-reduced state (PN, PDB ID 3U7Q) and (b) indigo disulfonate-oxidized state (POX, PDB ID 3MIN). Atomic coloring: Fe, orange; S, yellow; C, gray; N, blue. Reprinted (adapted or reprinted in part) with permission from ref 84. Copyright 2014 American Chemical Society. Figure 4. Catalytic cofactors from the crystal structures of (a) NifDK and (b) VnfDGK. The atoms of the clusters are shown as ball-and-stick models with the amino acid residues, R-homocitrate and carbonate presented as sticks. Atomic coloring: Fe, orange; S, yellow; Mo, teal; V, dark gray; C, light gray; N, blue; O, red (PDB ID 3U7Q, NifDK; 5N6Y, VnfDGK). Figure 5. Biosynthetic diagram for the assembly of nitrogenase metalloclusters. The red portion represents the in situ pathway for the synthesis of the P-cluster on NifDK. The blue portion reflects the ex situ synthesis of the M-cluster. Proteins and cluster species are labeled accordingly. Atoms are represented as ball-and-stick models and with atomic coloring as follows: Fe, orange; S, yellow; C, gray; O, red; Mo, teal. Reprinted (adapted or reprinted in part) with permission from ref 51. Copyright 2020 American Chemical Society. Figure 6. Modified Lowe-Thorneley model for N2 reduction by Mo-dependent nitrogenase. The En labeling indicates n number of proton and electron pairs that have been loaded on one of the αβ-dimers of NifDK. Reprinted (adapted or reprinted in part) with permission from ref 51. Copyright 2020 American Chemical Society. Figure 7. A relative coordinate system is (blue arrows) overlaid with the crystal structure of M-cluster. The coordinate system assumes CX as the origin (0, 0, 0), the x-axis passes through Fe1, CX and Mo1 with positive values toward Mo1, the y-axis passes through CX into the page with positive values pointing out of the page, and the z-axis passes through S2B and CX with positive values toward S2B. The cluster is presented as a ball-and-stick model with atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 8. Display of amino acid residues targeted for mutagenesis within the active site M-cluster of NifDK from A. vinelandii. (a) Side view and (b) view along the Mo-C-Fe1 axis of the M-cluster site are shown (PDB ID 3U7Q). The cluster is presented as a ball-and-stick model while the homocitrate ligand (HC) and amino acid side chains are shown as sticks. Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 9. EPR characterization of the M-cluster. The figure depicts wild-type (black), α-H195Q (red), and α-H195N (blue) NifDK proteins from A. vinelandii. The g ≈ 2 region is not shown for α-H195N due to contamination. Adapted with permission from ref 153. Copyright 1995 American Chemical Society. Data digitalized via WebPlotDigitizer v4.5. Figure 10. Crystal structure of Δnif B NifDK (in gray; PDB ID 1L5H), overlaid with the M-cluster and selected residues from the crystal structure of wild-type AvNifDK (in pink; PDB ID 3U7Q). The two structures were aligned, and the M-cluster and specific residues from the wild-type protein were superimposed on top of the Δnif B NifDK structure. The residues α-H274, α-H442, and α-H451 represent the identified “histidine triad”. The cluster atoms are shown as ball-and-stick models, and the amino acid residues are presented as sticks. Atomic coloring: Fe, orange; S, yellow; C, gray (in DnifB NifDK) or pink (in wild-type NfDK); O, red; N, blue. Figure 11. EPR spectra of α-R96L NifDK under nonturnover conditions with acetylene. The α-R96L NifDK protein was prepared under 1 atm of Ar (top spectrum, black) or under 1 atm of acetylene (bottom spectrum, red). Reprinted (adapted or reprinted in part) with permission from ref 180. Copyright 2001 American Chemical Society. Data digitalized via WebPlotDigitizer v4.5. Figure 12. Crystal structure of NifDK*. The M-cluster and key interacting residues at Site 1 (a) and Site 2 (b) are depicted (PDB ID 6UGO). The cluster and N molecules are presented as ball-and-stick models while the residues are shown as sticks. The M-clusters are superimposed with the F0 − Fc omit map of the N2 ligands contoured at 10σ (mesh). Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Reprinted (adapted or reprinted in part) with permission from ref 91. Copyright 2020 American Association for the Advancement of Science. Figure 13. EPR characterization of A. vinelandii NifDK during enzymatic turnover in the presence of CO. (a) α-H195Q NifDK in the presence of 1.0 atm CO (top, black) and 0.001 atm CO (bottom, red) showing the hi-CO and lo-CO signals, respectively. The small inflections are associated with the S = 3/2 signal of the resting state of the M-cluster. (b) EPR spectra of α-H195Q NifDK (top, black) and wild-type NifDK (bottom, blue) during turnover with 1 atm of CO, with the three lowest inflections (1)–(3) of the hi(5)-CO signal. The CO-related spectra of α-H195Q and wild-type NifDK are nearly identical. Reprinted (adapted or reprinted in part) with permission from ref 184. Copyright 2001 American Chemical Society. Data digitalized via WebPlotDigitizer v4.5. Figure 14. Crystal structure of NifDK-CO (a) and NifDK-(2CO) (b) from A. vinelandii. The cluster atoms are shown as ball-and-stick models, the CO, R-homoctirate (HC), and amino acid residues are presented as sticks. Black dotted lines represent a bonding interaction, blue dotted lines represent potential hydrogen bonding interactions. Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. PDB IDs: NifDK-CO, 4TKV; NifDH-(2CO), 7JRF. Figure 15. Crystal structure of VnfDGK-CO (a) and VnfDGK-(2CO) (b) from A. vinelandii. The cluster atoms are shown as ball-and-stick models, the CO, R-homoctirate (HC), carbonate, and amino acid residues are presented as sticks. Black dotted lines represent a bonding interaction, blue dotted lines represent potential hydrogen bonding interactions. Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. PDB IDs: VnfDGK-CO, 7ADR; VnfGDH-(2CO), 7AIZ. Figure 16. Crystal structure of α-H195Q NifDK (in gray; PDB ID 1FP4), overlaid with the crystal structure of wild-type NifDK (in cyan; PDB ID 3U7Q) from A. vinelandii. The cluster atoms are shown as ball-and-stick models, the R-homocitrate (HC), and amino acid residues are presented as sticks. Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 17. Schematic representation of hydrogenase cofactors. (a) [NiFe]-hydrogenase active site cofactor and (b) [FeFe]-hydrogenase active site cofactor. The [4Fe-4S] cluster is coordinated by three further cysteines (not shown). Note the presence of an azadithiolate ligand (ADT, S2(CH2)2NH) at the diiron site. ‘X’ marks the catalytic binding site in both cofactors. Figure 18. Crystal structures of [NiFe]-hydrogenase. (a) Crystal structure of the standard [NiFe]-hydrogenase from D. vulgaris Miyazaki F (PDB ID 1WUJ). (b) Stick representation of the Ni–Fe active site with two CN−, one CO, and a bridging ligand X that may be μOH−, μOOH−, or μH− (PDB ID 1WUJ shows the Ni-B state). The Ni–Fe active site cofactor is located inside the large subunit, whereas three iron–sulfur clusters are located almost linearly (each ~12 Å apart) in the small subunit. (c) Crystal structure of O2-tolerant membrane-bound [NiFe]-hydrogenase from E. coli (PDB ID 4GD3). Atomic coloring: Ni, green; Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 19. Second and outer coordination spheres of the Ni–Fe active site. Crystal structure of the standard [NiFe]-hydrogenase from D. vulgaris Miyazaki F in the Ni-B state (PDB ID 1WUJ). Glutamate E34, histidine H88, arginine R479, and serine S502 are important residues located in the second coordination sphere of the Ni–Fe active site. The dotted lines indicate the hydrogen bond network. Atomic coloring: Ni, green; Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 20. Crystal structures of the Ni–Fe active site in various oxidation states. (a) The Ni-A state of DvMF hydrogenase (PDB ID 1WUI) and (b) of Av hydrogenase (PDB ID 3MYR). (c) Ni-B state (PDB ID 1WUJ). (d) CO-inhibited Ni-SCO state (PDB ID 1UBH). (e) The Ni-SIa state (PDB ID 6QGR). (f) Ni-R state (PDB ID 4U9H). A bridging ligand exists in the Ni-A (μOOH− and μOH−), Ni-B (μOH−), and Ni-R states (μH−), while the bridging position is vacant in the Ni-SIa state. Atomic coloring: Ni, green; Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 21. Proposed catalytic cycle of [NiFe]-hydrogenase. The catalytic cycle comprises four states (Ni-SIa, Ni-R, Ni-C, and Ni-L) that interconvert in direction of H2 oxidation or proton reduction. The EPR-active states are shown in red, diamagnetic states are shown in blue. Among the inactive states, a μOH− ligand is tentatively assigned in the Ni-SU state. Figure 22. Proposed proton transfer mechanism for [NiFe]-hydrogenase. The hydrogen bond network between C546 (DvMF nomenclature, the respective sulfur atom is bold), E34, T18, A548, and a dangling water molecule is rearranged during the catalytic reaction. An involvement of R479 (gray) is possible. Reprinted (adapted or reprinted in part) with permission from ref 346. Copyright 2019 Wiley. Figure 23. Location of the accessory iron–sulfur clusters. The large subunit with the Ni–Fe cofactor is shown in white cartoon, the small subunit with the proximal [4Fe-4S], the medial [3Fe-4S], and the distal [4Fe-4S] cluster is shown in green (DvMF [NiFe]-hydrogenase, PDB ID 1H2R). The three accessory iron–sulfur clusters are located almost linearly. Atomic coloring: Ni, dark green; Fe, orange; S, yellow; C, green; O, red; N, blue. Figure 24. Structural changes of the proximal iron–sulfur cluster in various [NiFe]-hydrogenases. (a) Oxidized form (PDB ID 1UBH) in DvMF and (b) oxidized-distorted form (PDB ID 3MYR) in Av. (c) Reduced and oxidized forms of the [4Fe-3S] cluster in HmMBH (PDB ID 3AYX and 5Y34). (d) Reduced, partially reduced, and oxidized forms of the [4Fe-3S] cluster in ReMBH (PDB ID 3RGW, 4IUD, and 4IUB). The proximal [4Fe-4S] cluster is distorted with one of the Fe ions bound to D75 in one of the oxidized forms (b). Atomic coloring: Ni, green; Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 25. Proposed proton transfer in the O2-tolerant [NiFe]-hydrogenase HtTH-1. The oxidized state is shown in cyan (PDB ID 5XF9) and the H2-reduced state is shown in white (PDB ID 5XFA). Y1 refers to the proximal iron–sulfur cluster. The side chain of E32 binds to the Ni ion in the oxidized state, while it does not bind to the Ni ion and forms a hydrogen bond network including S464 and E56 in the reduced state. The metal centers in the reduced state are omitted for clarity. Atomic coloring: Ni, green; Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 26. Amino acids surrounding the distal [4Fe-4S] cluster. The [NiFe]-hydrogenase EcHyd-1 is shown (PDB ID 3UQY). Arginine R193 in the second sphere affects the electron transfer rate. Atomic coloring: Fe, orange; S, yellow; C, green; O, red; N, blue. Figure 27. Gas transfer channels. (a) Standard Df [NiFe]-hydrogenase. Reprinted (adapted or reprinted in part) with permission from ref 309. Copyright 2008 National Academy of Sciences. (b) O2 pathway in O2-tolerant ReMBH. Two entrances for gas transfer channels have been revealed, which are combined and extend to the Ni–Fe active site. Reprinted (Adapted or Reprinted in part) with permission from ref 400. Copyright 2018 National Academy of Sciences. Figure 28. Gas tunnels of Mb F420-reducing [NiFe]-hydrogenase. Seven Xe atoms (red spheres) were detected within an additional gas tunnel from the surface to the Ni–Fe cofactor (PDB ID 6QII). Reprinted (adapted or reprinted in part) with permission from ref 339. Copyright 2019 Wiley. Figure 29. Crystal structures of various [FeFe]-hydrogenases. (a) C. pasteurianum CpI (PDB ID 4XDC) and (b) D. desulfuricans DdH (PDB ID 1HFE). The characteristic peptide belt of DdH is shown in green. Panel (c) depicts a homology model of the [FeFe]-hydrogenase from C. reinhardtii CrHydA1. Here, the characteristic “loop” that is specific for this enzyme is shown in yellow. Accessory iron–sulfur clusters and the catalytic H-cluster are highlighted as spheres. Note the successive miniaturization of the F-domain from (a) to (c). Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 30. Active site cavity and catalytic cofactor of [FeFe]-hydrogenase CpI (PDB ID 4XDC). The H-cluster is composed of an [4Fe-4S] cluster and the diiron site. Loops 1–3 provide four cysteine residues to bind the [4Fe-4S] cluster and the diiron site. In loop 1 (green), C299 is part of the proton transfer pathway (further including water cluster W1, E279, and other groups). In loop 2 (blue), M352 may interact with the μCO ligand while K358 was suggested to form a hydrogen bond with the CN− ligand at the distal iron ion, Fed. In loop 3 (magenta), M497 may form a hydrogen bond with the ADT group. Cysteine C499 was proposed to be the 1st proton acceptor of the proton transfer pathway. Serine S232 provides a hydrogen bond to the CN− ligand of the proximal iron ion, Fep, presumably involving a backbone contact with P231. Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 31. EPR spectra of the oxidized H-cluster. Simulations of EPR spectra for the oxidized states Hox, HoxH, and Hox-CO as observed in CrHydA1. The g-values are indicated. Reprinted (adapted or reprinted in part) with permission from ref 267. Copyright 2020 American Chemical Society. Figure 32. Influence of the [4Fe-4S] cluster on the IR signature of the H-cluster. (a) ‘H2 − N2’ difference spectra of CrHydA1 cofactor variant PDT at pH 5 (‘Hred′H − HoxH’, blue trace) and pH 8 (‘Hred′ – Hox’, red trace). Negative and positive bands are assigned to H-cluster states with an oxidized and reduced [4Fe-4S] cluster, respectively. The small frequency shifts between the oxidized and reduced states are assigned to a protonation at or near the [4Fe-4S] cluster. Data taken from ref 488. (b) Spectro-electrochemical difference spectrum of CrHydA1 that shows the accumulation of Hsred (magenta labels, −650 mV vs SHE) over Hred (green labels, −450 mV vs SHE) at pH 9. Negative and positive bands are assigned to H-cluster states with an oxidized and reduced [4Fe-4S] cluster, respectively. At ambient temperature, the bridging carbonyl ligand (μCO) converts into a terminal ligand, which results in three tCO bands. Figure 33. Changes in geometry upon reduction. (a) In the paramagnetic II/I state of the diiron site (Hox and Hred′), the μCO ligand stabilizes the “rotated structure” that is characterized by a squarepyramidal/inverted square-pyramidal arragenment and an open coordination site at Fed (X). In Hhyd, the diiron site is “superoxidized” (II/II) and X is a terminal hydride. (b) Upon reduction (I/I), the diiron site adopts the symmetrical square-pyramidal/square-pyramidal arrangement. No μCO ligand is observed in the respective redox states Hred and Hsred. Figure 34. Proposed catalytic cycle. The 5-step model (A–E) includes HredH+ and HsredH+, presuming the H-cluster retains the μCO ligand upon reduction of the diiron site. The 3-step model (A*–B*–E) suggests a “short-cut” from Hred′ to Hhyd as both species share the same geometry. The geometry of the reduced diiron site (I/I) under ambient conditions is debated. Reprinted (adapted or reprinted in part) with permission from ref 267. Copyright 2020 American Chemical Society. Figure 35. Comparison of CpI (white) and CbA5H (green) in the Hox and Hinact state. The distance of the sulfur atom of C299/C367 shrinks from ~6 Å to ~3 Å, making a direct coordination of the distal iron ion (Fed) in the Hinact state of CbA5H likely (dashed line). The annotated amino acid residues have been identified to cause the flexibility of the unique loop in CbA5H. Drawn after PDB IDs 4XDC and 6TTL. Figure 36. Hydrogen-bonding environment of the CN− ligands. In proximal (p) position, polar interactions (v) with a backbone amine, S232, and variant A230S have been experimentally verified in CpI and CrHydA1. The hydrophilic pocket including K358, S323, and a P324 backbone contact inspired the assignment of the CN− ligand in distal (d) position in CpI and DdH. The putative hydrogen-bonding environment (v) is shown. Drawn after PDB ID 6NAC. Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 37. Structural differences between “conformation A” and “conformation B”. The M353 side chain shifts by ~0.4 Å away from μCO and Fep-CO in the “reduced” state (equivalent conformation B), which lowers the steric constraints (iv). Concomitantly, the side chain of S357 switches to an orientation toward the [4Fe-4S] cluster and the cysteine ligand C355. This may influence the electron density distribution across the cluster (vi). Drawn after PDB ID 6NAC. Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 38. The “catalytic” proton transfer pathway in [FeFe]-hydrogenase CpI. The side chain of M497 may stabilizes the orientation of the ADT ligand by a hydrogen bond (v). Dashed blue lines represent the hydrogen-bonding network (ii) of the proton transfer pathway. Note that a bond between E279 and S319 is unlikely in the oxidized state (dashed gray line) but formed upon reduction and protonation of the cofactor. Drawn after PDB ID 4XDC. Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 39. The “regulatory” proton transfer pathway in [FeFe]-hydrogenase CpI. In most crystal structures of CpI, a trajectory of water molecules (red spheres) can be found, putatively connected by hydrogen bonds (ii). This water channel “W2” is located at the intersection of the accessory domain (white) and the catalytic domain (including the most proximal iron–sulfur cluster, [4Fe-4S]p), leading up to cysteine C499 that coordinates the [4Fe-4S] cluster of the catalytic cofactor. Here, protonation and redox changes influence the diiron site (vi). Drawn after PDB ID 4XDC. Atomic coloring: Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 40. Gas diffusion pathways in [FeFe]-hydrogenase. Representative simulations of 1000 copies of (a) H2 diffusing out from the H-cluster and (b) of O2 diffusing out from the H-cluster or from the middle of a previously identified H2 channel. Possible exits, based on the proximity of the external solution, are highlighted with arrows. Contrary to H2 diffusion, the O2 molecules move collectively through the same pathway for a given simulation, though they may employ different pathways for different independent simulations. Reprinted (adapted or reprinted in part) with permission from ref 569. Copyright 2005 Cell Press. Figure 41. Bis-metal binding pterin guanine dinucleotide cofactor (bis-MGD) present in formate dehydrogenases. (a) Members of the DMSO reductase family of molybdoenzymes bind the bis-MGD form of the cofactor in the active site. The bis-MGD in formate dehydrogenases coordinates the metal (depicted as M to be Mo or W) by the dithiolene groups from two pterin ligands with the additional ligands being a sulfido group, and a sixth ligand which is either a selenocysteine (Sec) or a cysteine (Cys) from the amino acid backbone. (b) Representative depiction of the bis-MGD cofactor coordination environment in formate dehydrogenases, of E. coli FdhF in the oxidized state (PDB ID 1FDO). Figure 42. Schematic overview of cytoplasmic and periplasmic FDHs. Localization of FDHs, their subunit organization and cofactor composition. Shown are the membrane bound respiratory FdnGHI from E. coli, the periplasmic FdhAB and FdhABC from the D. gigas and D. vulgaris Hildenborough, the formate hydrogen-lyase complex formed by (FdhF) and Hyd3 (HycBCDEFG) from E. coli, the cytoplasmic FDH from R. capsulatus, the cytoplasmic CO2-reducing dimeric FDH (FdhAB) from the acetogen Morella thermoacetica, the HDCR complex from A. woodii (Mo-containing) or from T. kivui (W-containing), and the formylmethanofuran dehydrogenase complex from M. wolfei. Figure 43. Structural comparison of FDH undergoing redox change. Structures correspond to (a) E. coli FdhF oxidized (PDB ID: 1FDO) and formate-reduced (PDB ID: 2IV2), (b) D. vulgaris Hildenborough FDH oxidized (PDB ID: 6SDR) and formate-reduced (PDB ID: 6SDV), and (c) R. capsulatus FDH oxidized (PDB ID: 6TGA) and NADH-reduced (PDB ID: 6TG9). Participatory amino acids and bound H2O and glycerol molecules (where appropriate) in the substrate-binding cleft of FDH are depicted in dark salmon in the oxidized state, and deep teal in the reduced state. The structural effect of specific amino acids accompanying redox changes are depicted with black arrows. The appearance of highlighted H2O molecules in the reduced state are indicated with red arrows. Figure 44. Structural characterization of the active site of Mo/W–FDHs and formylmethanofuran dehydrogenases. Structures correspond to (a) E. coli FdhF (PDB ID 1FDO), (b) E. coli FdnG (PDB ID 1KQF), (c) D. gigas FdhA (PDB ID 1H0H), (d) D. vulgaris Hildenborough FdhA (PDB ID 6SDR), (e) R. capsulatus FdsA (PDB ID 6TGA), and (f) M. wolfeii formylmethanofuran dehydrogenase FwdABCD (PDB ID 5T5I). Key amino acid residues and bound H2O molecules are depicted, where appropriate. Predicted substrate and gas channels, are noted for each enzyme; the hydrophobic channel predicted for D. vulgaris Hildenborough FDH was predicted from the reduced enzyme (PDB ID 6SDV). Subunits harboring the bis-MGD cofactor are labeled with black text. The hydrophilicity or hydrophobicity of the channel is noted. Channels were constructed via the CAVER plugin in PYMOL. Figure 45. Proposed reaction mechanisms for formate oxidation by FDH enzymes. (a) Mechanism proposed by Heider and Böck in 1993; (b) mechanism proposed by Sun and co-workers in 1997; (c) mechanism proposed by Raaijmakers and Romao in 2006; (d) mechanism proposed by Cerqueira, Gonzales, Moura, and co-workers in 2011; (e) mechanism proposed by Zampella and co-workers in 2012; (f) mechanism proposed by Dong and Ryde in 2018; (g) mechanism proposed by Hille and co-workers in 2016; details are given in the text. Active site residues without a defined role in an individual mechanism are depicted in gray. Figure 46. Structure of CODH. Overall structure of ChCODH II (PDB ID 1SU8). Each monomer of the ChCODH II homodimer is represented with a different color. The spheres represent the inorganic cofactors, labeled cluster B–D. The C-cluster is the catalytic active site cofactor. The purple amino acids correspond to the putative proton transfer pathway. (b) Close up view of the C-cluster and residues putatively involved in proton transfer (PDB ID 3B52). Atomic coloring: Fe, orange; Ni, green; S, yellow; C, gray; O, red; N, blue. Figure 47. Structure of the catalytic C-cluster of CODH. (a) Close up view of the CO2-bound active site (ChCODH II, PDB ID 3B52), with a representation of most of the amino acids discussed here. Feu refers to the “dangling” iron ion. (b) View of the active site of DvCODH (PDB ID 6B6W) in the oxidized, protected state (Cox). The orientation is the same as in the left panel. Metal-coordinating residues are bold. Atomic coloring: Ni, green; Fe, orange; S, yellow; C, gray; O, red; N, blue. Figure 48. Proposed mechanisms for the oxidation of CO to CO2 by CODH. The black scheme is the original proposal by Jeoung and Dobbek.682,691 The blue part is the modification to that scheme proposed by Fontecilla-Camps, Amara, and co-workers.692 For the sake of clarity, we have only indicated the arrows in the direction of CO oxidation, but CO2 reduction is supposed to follow the exact same steps in reverse order. Figure 49. Hydrophobic gas channels in CODH. Structure of the ACS/CODH from M. thermoacetica (PDB ID 2Z8Y), indicating the position of the clusters A–D, and marking the long hydrophobic channel that connects them. Reprinted (adapted or reprinted in part) with permission from ref 702. Copyright 2022 Elsevier. Figure 50. Sequence alignment of the main CODHs studied so far. Conservation is indicated by a blue background. The amino acids with a red background have been modified by site-directed mutagenesis in the literature. The red frames correspond to the amino acids within 8 Å of any atom from inorganic C-cluster. The numbering is that of ChCODH II. Table 1. List of Single Point Mutations Reported for Azotobacter vinelandii NifDK with Proximity to the M-Cluster Sitea,b NifDK residue Arg (R) His (H) Lys (K) Asp (D) Glu (E) Ser (S) Thr (T) Asn (N) Gln (Q) Cys (C) Gly (G) Pro (p) Ala (A) Ile (i) Leu (L) Phe (F) Tyr (Y) Trp (W) ref α-G69 x 174, 175 α-V70 x x x 176, 177, 178, 179 α-H83 x x x x x 152 α-R96 x x x x x 154, 179, 180, 181 α-Q151 x 169 α-C154 x 169 α-D161 x 169 α-C183 x 169 α-S190 x 174 α-Q191 x x x x 139, 169, 174, 178, 181, 182, 183, 184, 185 α-S192 x x x x x x x x 174, 186 α-H195 x x x x x x 139, 152, 154, 179, 182, 183, 184, 187, 188, 189 α-H196 x x x x x 152 α-H274 x x 152, 156, 158 α-C275 x 169 α-R277 x x x x x x x x 181, 186, 190 α-G356c 186 α-R359 x x 154, 174, 181 α-H362 x x 157 α-F381 x x 154, 174 α-H442 x x 154 α-W444 x x x x 158 α-H451 x 156, 158 a NifD = α, NifK = β. b NifDK residues listed have been mutated to the marked amino acids. c No specific mutations are listed. Table 2. List of A. vinelandii NifDK Residues near the M-Cluster That Have Been Investigated through Mutagenesis NifDK residue location near M-clustera coordinateb proposed role(s) α-G69 outer sphere near S2B (+, −, +) Hydrophobic interactions with substrate α-V70 2nd sphere near S2B (0, −, +) Hydrophobic interactions with substrate α-R96 2nd sphere near S5A (+,−, 0) Steric/Hydrogen-bonding interactions with M-cluster, substrates α-S190 outer sphere near S2B (0, +, +) Steric or hydrogen-bonding interactions α-Q191 2nd sphere near S2B, HC (+, +,+) Hydrogen-bonding interactions with R-homocitrate ligand ofM-cluster α-S192 outer sphere near S2B (−, +, +) Hydrogen-bonding interactions with conserved H2O molecules α-H195 2nd sphere near S2B (−, 0, +) Hydrogen-bonding and steric interactions with M-cluster, substrates α-H196 outer sphere near S2B, Fe1 (−, 0, +) Unknown α-H274 outer sphere near Fe1, S4A (−,−,−) Assists with M-cluster incorporation, adjacent to α-C275 α-C275 cofactor bound (−, 0, 0) Ligates Fe-capped end of M-cluster α-R277 2nd sphere near S1A (−, +, +) Hydrogen-bonding interactions with conserved H2O molecules α-G356 2nd sphere near S3A (+, +, 0) Hydrogen-bonding interaction with M-cluster α-R359 2nd sphere near S5A (0, +, −) Hydrogen-bonding interaction with M-cluster α-H362 outer sphere near Fe1 (−, +, −) Assists with M-cluster incorporation α-F381 2nd sphere between S2B, S3A (0, +, +) Hydrophobic interactions with M-cluster α-H442 cofactor bound (+, +, −) Ligates Mo-capped end of M-cluster α-W444 outer sphere near S4B (+,−, −) Hydrophobic interactions with M-cluster α-H451 outer sphere near S4A (−,−,−) Assists with M-cluster incorporation a Locations are based on proximity to the M-cluster atoms from the crystal structure of the resting state NifDK (PDB ID 3U7Q).38 b The general location of the amino acid side chain in the coordinate system defined in Figure 7. Table 3. Reactivity of Select Purified NifDK Variantsa NifDK protein type NH3b (N2) H2b (N2) C2H4c (C2H2) C2H6c (C2H2) H2c (C2H2) H2d (Ar) NH3e (N2/CO) H2e (N2/CO) C2H4f (C2H2/CO) C2H6f (C2H2/CO) H2f (C2H2/CO) H2g (Ar/CO) ref Av wild-type 860 615 1706 0 276 2494 0 2727 1134 0 959 2841 185 α-V70A 800 1800 176 α-V70I 170 1940 130 0 2000 2340 177 α-R96K 373 804 972 154 α-Q191K 0 514 40 7.2 528 540 0 182 14 2.3 340 254 185 α-H195Q - - 1350 0 1115 2860 183 α-H195N - - 535 135 355 1470 183 α-R277C 1400 2400h 186 α-R359K 709 1523 1587 154 α-F381L 816 1656 1844 154 Av Wild-type 989 555 2096 2211 158 α-W444Y 1036 507 2066 2069 158 α-W444F 906 505 1906 1795 158 α-W444A 212 101 364 410 158 α-W444G 0 6 8 11 158 Av ΔnifV 840 181 Kp Wild-type 725 471 1127 - - 1527 196 Kp ΔnifV 318 842 1044 - - 1242 196 Kp ΔnifV - 812 - 202 1369 - 574 - - 585 718 197 a Specific activity is reported with units of nmol product × min−1 × (mg protein)−1. Gas in parentheses reflects the substrate conditions used for the assay. b Assays are reported in an atmosphere of 100% N2. c Assays reported in an atmosphere of 10% C2H2, 90% Ar. d Assays reported in an atmosphere of 100% Ar. e Assays are reported in an atmosphere of 3% CO, 97% N2. f Assays are reported in an atmosphere of 3% CO, 10% C2H2, 87% Ar. g Assays are reported in an atmosphere of 3% CO, 97% Ar. h Assays are reported in an atmosphere of 10% CO, 90% Ar. Table 4. Conservation of Amino Acid Residues in the H-Domain of Experimentally Characterized [FeFe]-Hydrogenasesa CpI CpII CpIII CaI DdH MeHydA CbA5H CrHydA1 CrHydA2 TmHydA TmHydS TamHydS HB S232 S101 A162 S231 A109 A113 A294 A94 A97 A227 S79 S81 PT E279 E148 E202 E278 E156 E151 E341 E141 E144 E274 E118 F119 PT E282 E151 E205 E281 E159 E154 E344 E144 E147 E277 E121 E122 PT R286 R155 R209 R285 R163 R158 R348 R148 R151 R281 G125 K126 PT C299 C169 C222 C298 C178 C171 C367 C169 C172 C294 A131 A137 CC C300 C170 C223 C299 C179 C172 C368 C179 C173 C295 C132 C138 PT S319 S189 S242 S318 S189 S192 S387 S189 S192 S314 A151 L157 HP/HB M353 T223 G268 M352 M232 T227 M421 M223 M226 M348 G177 S191 CC C355 C225 C270 C354 C234 C229 C423 C225 C228 C350 C179 C193 HB S357 A227 A272 D356 A236 A231 A425 R227 R230 A352 A181 A195 HB K358 K228 K273 K357 K237 K232 K426 K228 K231 K353 K182 K196 HB M497 M367 M381 M497 M376 M381 M565 M415 M426 M480 S267 L291 CC C499 C369 C383 C498 C378 C391 C567 C417 C428 C482 C269 C293 CC C503 C382 C387 C502 C382 C395 C571 C421 C432 C486 C273 C297 PDB IDs 6N59 - - - 1FEH - 6TTL 3LX4 - - - - a Representative PDB accession codes for CpI, DdH, CbA5H, and the apoprotein of CrHydA1 are given in the last row. Legend: HB, hydrogen bonding; HP, hydrophobic; PT, proton transfer; CC, cluster coordination. Table 5. List of [FeFe]-Hydrogenase Amino Acid and H-Cluster Variantsa variant PDB ID target ref CpI G412H, G414A/H, G418A/H, G421H, G422H, K358N, R449A/H N/A Maturation 445 E279A 5LA3 Proton transfer, formation of Hhyd 446 C299A 6GLY Proton transfer 447, 448, 449 C299D 6GLZ C299S N/A E279A N/A E279D 6YF4 E279Q 6GM0 E282A 6GM1 E282D 6GM2 E282Q 6GM8 R286A/L 6GM3 S319A 6GM4 C299S, M353L, K358N, M497L N/A Hydrogen bonding 450 S31P, E47G, R86H, C153Y, N160D, D186G, T188A, I197V, K208E, K252R, I253V, A280V, N289D N/A Hydrogen turnover, O2 sensitivity 451 L192G, G194C, N189C, T356C/T/V, S357C/G/P/T, M397C, A498C N/A Hydrogen turnover, O2 sensitivity 452 M70L, M211L, M243L, M277L, M387L, M481L, M485L, M551L N/A O2 sensitivity 453 C48A, C100A, C298D N/A Proton transfer, electron transfer 417, 454, 455 CrHydA1 C169A/S N/A Proton transfer, formation of Hhyd 456, 457, 458, 459 C169A/D/S N/A Proton transfer 448 E141A 6GM5 E141D N/A E141Q 6GM6 E144A 6GM7 E144D/Q N/A R148A N/A S189A N/A C417Hb 6GL6 [4Fe-4S] 460 C170A/D/S, C225A/D/S, C417A/D/S, C421A/D/Sb N/A [4Fe-4S] 461 T226K/V N/A [4Fe-4S] 462 R171D/Wc N/A Electron transfer 463 A92S, A94S, S193C, E231D N/A Hydrogen bonding to CN− 464 C169S, M223L, K228N, M415L N/A Hydrogen bonding 450 C169D, F290Y, V296F N/A O2 diffusion 465 R96Q K179Q K262Q R349Q R353Q R379Q K396Q K397E, K433Q N/A Interaction with ferredoxin 466 Other L364F, C367D/A, P386L, A561F (in CbA5H) N/A O2 inhibition, formation of Hinact 424 C48A, C100A, C298D (in CaI) N/A Proton transfer and electron transfer 417, 454, 455 V229T (in CrHydA2) N/A [4Fe]H 467 c Cofactor Variants PDT, ODT, SDT, and other headgroups N/A Maturation, hydrogen turnover 418, 468, 469, 470, 471, 472, 473, 474 Other modifications N/A Hydrogen turnover 475, 476, 477, 478, 479 ODT 5BYQ Hydrogen turnover, hydrogen bonding 479, 480 PDT 5BYR SDT 5BYS EDT 6H63 a The majority of amino acid variants were produced for CpI and CrHydA1, with most H-cluster variants being produced for CrHydA1. Other enzymes include CaI, CbA5H, and CrHydA2. b In refs 460 and 461, the numbering is different due to the deletion of 55 C-terminal amino acids. c In ref 463, arginine R171 resembles R227 following the nomenclature used for CrHydA1 in this work. Table 6. Vibrational and Electronic Properties of the Main H-Cluster Statesa vCN−/cm−1 vCO/cm−1 [4Fe-4S] [FeFe] μ/t H ox 2088 2070 1964 1940 1802 +2 II/I CO/− H ox H 2092 2074 1970 1946 1812 +2 II/I CO/− H ox -CO 2091 2081 1968 1962c 1808 +2 II/I CO/CN− H ox H–CO 2094 2086 1972 1966c 1816 +2 II/I CO/CN− H red′ 2084 2066 1962 1933 1792 + 1 II/I CO/− H red′ H 2086 2068 1966 1938 1800 +1 II/I CO/− H red′ -CO 2086 2076 1967 1951c 1793 +1 II/I CO/CN− H red 2070 2033 1961 1915 1891 +2 I/I H/COb H hyd 2088 2076 1980 1960 1860 +1 II/II CO/H− H sred 2068 2026 1953 1918 1882 +1 I/I H/COb H inact 2106 2087 2007 1983 1848 +2 II/II CO/? H trans 2100 2075 1983 1977 1836 +1 II/II CO/? a All data stem from DdH (Hinact and Htrans) or CrHydA1 (other H-cluster states). b Under cryogenic conditions, HredH+ and HsredH+ are formed that bind a bridging carbonyl ligand and are likely to feature an open coordination site (μ/t = CO/−). c In the CO-inhibited states, the coupled pCO/dCO stretching vibration gives rise to an additional band at 2012 cm−1 (Hox-CO), 2006 cm−1 (HoxH–CO), or 2002 cm−1 (Hred′-CO). Table 7. Oxidation Kinetic Parameters of Structurally Characterized FDH and Formylmethanofuran Dehydrogenase Enzymesa organism/type E. coli FdhF E. coli FdnGHI D.gigas FdhAB D. vulgaris Hildenborough FdhAB R. capsulatus FdsGBAD M. wolfeii formylmethanofuran dehydrogenase catalytic subunit FdhF FdnG FdhA FdhA FdsA FwdB metal Mo Mo W W Mo W PDB ID 1FDO, 2IV2 1KQF 1H0H 6SDR, 6SDV 6TGA, 6TG9 5T5I formate oxidation kinetic parameters kcat = 2800 s−1 N/A kcat = 138 s−1 kcat = 3684 s−1 kcat = 36 s−1 kcat = 18 s−1 KM = 26 mM KM = 1 μM KM = 281 μM KM = 13 μM Kd = 100 mM KM NAD+ = 173 μM KM MV = 400 μM KM BV = 3 mM ref 594 590 595 596 597 a Reported formate oxidation kinetic parameters for E. coli, D. gigas, and D. vulgaris Hildenborough FDH represent those following the enzymatic reduction of benzyl viologen; formylmethanofuran dehydrogenase follows the enzymatic reduction of methyl viologen, while for R. capsulatus FDH the enzymatic reduction of NAD+ is reported. Table 8. Amino Acid Comparison and Oxidation Kinetic Parameters of Active Site Variants of Structurally Characterized FDH and Formylmethanofuran Dehydrogenase Enzymesa organism/type E. coli FdhF E. coli FdnGHI D. gigas FdhAB D. vulgaris Hildenborough FdhAB R capsulatus FdsGBAD M. wolfeii formylmethanofuran dehydrogenase Function Formate Oxidation Kinetic Parameters ref conserved amino acids U140 U196 U158 U192 C386 C118 substrate binding, catalytic efficiency E. coli (FdhF) U140C: kcat = 9 s−1, KM = 9 mM, KD = 5 mM, KM BV = 2 mM 585 H141 H197 H159 H193 H387 H119 proton shuttling, channel gating R capsulatus H387M: kcat = 34 s−1, KM = 3.6 mM R capsulatus H387K: kcat = 1 s−1, KM = 28 mM 598 R333 R446 R407 R441 R587 R288 substrate binding, channel gating R capsulatus R587K: kcat = 11 s−1, KM = 362 mM 598 M157 M213 M175 M209 G403 I133 hydrophobic channel direction N/A V338 L450 V412 V446 V592 V293 hydrophobic interaction at bis-MGD N/A K44 K94 K56 K90 K295 I37 electron transfer to Fe–S relay N/A Q335 H448 E409 E443 Q589 H290 gating proton shuttling N/A V145 V201 V198 V197 G391 V123 hydrophobic interaction with active site Arg N/A Q339 Q452 Q413 Q447 Q593 N297 hydrophobic interaction with active site Arg N/A a Amino acid residues in the primary or secondary coordination sphere by which a role has been assigned or postulated are noted. Enzymatic activity of reported active site variants for respective organisms follow the same electron acceptor used as reported for the WT enzyme, as reported in Table 7. Complete contact information is available at: https://pubs.acs.org/10.1021/acs.chemrev.1c00914 The authors declare no competing financial interest. REFERENCES (1) Holm RH ; Solomon EI Preface: Bioinorganic Enzymology. Chem. Rev 1996, 96 , 2237–2238.11848827 (2) Holm RH ; Solomon EI Preface: Biomimetic Inorganic Chemistry. Chem. Rev 2004, 104 , 347–348.14871127 (3) Holm RH ; Solomon EI Introduction: Bioinorganic Enzymology II. Chem. Rev 2014, 114 , 4039–4040.24758378 (4) McEvoy JP ; Brudvig GW Water-Splitting Chemistry of Photosystem II. Chem. Rev 2006, 106 , 4455–4483.17091926 (5) Vinyard DJ ; Ananyev GM ; Charles Dismukes G Photosystem II: The Reaction Center of Oxygenic Photosynthesis. Annu. Rev. Biochem 2013, 82 , 577–606.23527694 (6) Vinyard DJ ; Brudvig GW Progress Toward a Molecular Mechanism of Water Oxidation in Photosystem II. Annu. Rev. Phys. Chem 2017, 68 , 101–116.28226223 (7) Babcock GT ; Wikström M Oxygen Activation and the Conversation of Energy in Cell Respiration. Nature 1992, 356 , 301–308.1312679 (8) Yoshikawa S ; Muramoto K ; Shinzawa-Itoh K Proton-Pumping Mechanism of Cytochrome c Oxidase. Annu. Rev. Biophys 2011, 40 , 205–223.21545285 (9) Wikström M ; Krab K ; Sharma V Oxygen Activation and Energy Conservation by Cytochrome c Oxidase. Chem. Rev 2018, 118 , 2469–2490.29350917 (10) Fontecilla-Camps JC ; Amara P ; Cavazza C ; Nicolet Y ; Volbeda A Structure-Function Relationships of Anaerobic Gas-Processing Metalloenzymes. Nature 2009, 460 , 814–822.19675641 (11) Brandes JA ; Boctor NZ ; Cody GD ; Cooper BA ; Hazen RM ; Yoder HS Abiotic Nitrogen Reduction on the Early Earth. Nature 1998, 395 , 365–367.9759725 (12) Russell MJ The Alkaline Solution to the Emergence of Life: Energy, Entropy and Early Evolution. Acta Biotheor. 2007, 55 , 133–179.17704896 (13) Konkena B ; junge Puring K ; Sinev I ; Piontek S ; Khavryuchenko O ; Dürholt JP ; Schmid R ; Tüysüz H ; Muhler M ; Schuhmann W ; Pentlandite Rocks as Sustainable and Stable Efficient Electrocatalysts for Hydrogen Generation. Nat. Commun 2016, 7 , 12269.27461840 (14) McGlynn SE ; Mulder DW ; Shepard EM ; Broderick JB ; Peters JW Hydrogenase Cluster Biosynthesis: Organometallic Chemistry Nature’s Way. Dalt. Trans 2009, No. 22 , 4274–4285. (15) Nitschke W ; McGlynn SE ; Milner-White EJ ; Russell MJ On the Antiquity of Metalloenzymes and Their Substrates in Bioenergetics. Biochim. Biophys. Acta - Bioenerg 2013, 1827 , 871–881. (16) Russell MJ ; Nitschke W ; Branscomb E The Inevitable Journey to Being. Philos. Trans. R. Soc. B Biol. Sci 2013, 368 , 20120254. (17) Sousa FL ; Thiergart T ; Landan G ; Nelson-Sathi S ; Pereira IAC ; Allen JF ; Lane N ; Martin WF Early Bioenergetic Evolution. Philos. Trans. R. Soc. B Biol. Sci 2013, 368 , 20130088. (18) Milner-White EJ ; Russell MJ Sites for Phosphates and Iron-Sulfur Thiolates in the First Membranes: 3 to 6 Residue Anion-Binding Motifs (Nests). Orig. Life Evol. Biosph 2005, 35 , 19–27.15889648 (19) Milner-White EJ ; Russell MJ Predicting the Conformations of Peptides and Proteins in Early Evolution. Biol. Direct 2008, 3 , 3.18226248 (20) Hanscam R ; Shepard EM ; Broderick JB ; Copié V ; Szilagyi RK Secondary Structure Analysis of Peptides with Relevance to Iron-Sulfur Cluster Nesting. J. Comput. Chem 2019, 40 , 515–526.30548652 (21) Beinert H ; Holm RH ; Münck E Iron-Sulfur Clusters: Nature’s Modular, Multipurpose Structures. Science 1997, 277 , 653–659.9235882 (22) Dempsey JL ; Winkler JR ; Gray HB Proton-Coupled Electron Flow in Protein Redox Machines. Chem. Rev 2010, 110 , 7024–7039.21082865 (23) Warren JJ ; Tronic TA ; Mayer JM Thermochemistry of Proton-Coupled Electron Transfer Reagents and Its Implications. Chem. Rev 2010, 110 , 6961–7001.20925411 (24) Hammes-Schiffer S Proton-Coupled Electron Transfer: Moving Together and Charging Forward. J. Am. Chem. Soc 2015, 137 , 8860–8871.26110700 (25) Banerjee R ; Lipscomb JD Small-Molecule Tunnels in Metalloenzymes Viewed as Extensions of the Active Site. Acc. Chem. Res 2021, 54 , 2185–2195.33886257 (26) Winkler M ; Esselborn J ; Happe T Molecular Basis of [FeFe]-Hydrogenase Function: An Insight into the Complex Interplay between Protein and Catalytic Cofactor. Biochim. Biophys. Acta 2013, 1827 , 974–985.23507618 (27) Gray HB ; Winkler JR Electron Transfer in Proteins. Annu. Rev. Biochem 1996, 65 , 537–561.8811189 (28) Mulder DW ; Peters JW ; Raugei S Catalytic Bias in Oxidation-Reduction Catalysis. Chem. Commun 2021, 57 , 713–720. (29) Fourmond V ; Plumeré N ; Léger C Reversible Catalysis. Nat. Rev. Chem 2021, 5 , 348–360. (30) Zhang X ; Ward BB ; Sigman DM Global Nitrogen Cycle: Critical Enzymes, Organisms, and Processes for Nitrogen Budgets and Dynamics. Chem. Rev 2020, 120 , 5308–5351.32530264 (31) Logan JA Nitrogen Oxides in the Troposphere: Global and Regional Budgets. J. Geophys. Res. Ocean 1983, 88 , 10785–10807. (32) Liaw YP ; Sisterson DL ; Miller NL Comparison of Field, Laboratory, and Theoretical Estimates of Global Nitrogen Fixation by Lightning. J. Geophys. Res. Atmos 1990, 95 , 22489–22494. (33) Cleveland CC ; Townsend AR ; Schimel DS ; Fisher H ; Howarth RW ; Hedin LO ; Perakis SS ; Latty EF ; Von Fischer JC ; Elseroad A ; Global Patterns of Terrestrial Biological Nitrogen (N2) Fixation in Natural Ecosystems. Global Biogeochem. Cycles 1999, 13 , 623–645. (34) Noxon JF Atmospheric Nitrogen Fixation by Lightning. Geophys. Res. Lett 1976, 3 , 463–465. (35) Noxon JF Tropospheric NO2. J. Geophys. Res. Ocean 1978, 83 , 3051–3057. (36) Darwent BB Bond Dissociation Energies in Simple Molecules; U.S. National Bureau of Standards, 1970. (37) Raymond J ; Siefert JL ; Staples CR ; Blankenship RE The Natural History of Nitrogen Fixation. Mol. Biol. Evol 2004, 21 , 541–554.14694078 (38) Boyd ES ; Hamilton TL ; Peters JW An Alternative Path for the Evolution of Biological Nitrogen Fixation. Front. Microbiol 2011, 2 , 205.22065963 (39) Mus F ; Colman DR ; Peters JW ; Boyd ES Geobiological Feedbacks, Oxygen, and the Evolution of Nitrogenase. Free Radic. Biol. Med 2019, 140 , 250–259.30735835 (40) Zehr JP ; Capone DG Changing Perspectives in Marine Nitrogen Fixation. Science 2020, 368 , No. eaay9514.32409447 (41) Fan K ; Weisenhorn P ; Gilbert JA ; Shi Y ; Bai Y ; Chu H Soil PH Correlates with the Co-Occurrence and Assemblage Process of Diazotrophic Communities in Rhizosphere and Bulk Soils of Wheat Fields. Soil Biol. Biochem 2018, 121 , 185–192. (42) Wang C ; Zheng M ; Song W ; Wen S ; Wang B ; Zhu C ; Shen R Impact of 25 Years of Inorganic Fertilization on Diazotrophic Abundance and Community Structure in an Acidic Soil in Southern China. Soil Biol. Biochem 2017, 113 , 240–249. (43) Gaby JC ; Buckley DH Assessment of Nitrogenase Diversity in the Environment. In Biological Nitrogen Fixation; Wiley, 2015; pp 209–216. (44) Noar JD ; Bruno-Bárcena JM Azotobacter Vinelandii: The Source of 100 Years of Discoveries and Many More to Come. Microbiology 2018, 164 , 421–436.29533747 (45) Huber H ; Thomm M ; König H ; Thies G ; Stetter KO Methanococcus Thermolithotrophicus, a Novel Thermophilic Lithotrophic Methanogen. Arch. Microbiol 1982, 132 , 47–50. (46) Belay N ; Sparling R ; Daniels L Dinitrogen Fixation by a Thermophilic Methanogenic Bacterium. Nature 1984, 312 , 286–288.6504143 (47) Weidner S ; Pühler A ; Küster H Genomics Insights into Symbiotic Nitrogen Fixation. Curr. Opin. Biotechnol 2003, 14 , 200–205.12732321 (48) Stougaard J Genetics and Genomics of Root Symbiosis. Curr. Opin. Plant Biol 2001, 4 , 328–335.11418343 (49) Ribbe MW ; Hu Y ; Hodgson KO ; Hedman B Biosynthesis of Nitrogenase Metalloclusters. Chem. Rev 2014, 114 , 4063–4080.24328215 (50) Buren S ; Jimenez-Vicente E ; Echavarri-Erasun C ; Rubio LM Biosynthesis of Nitrogenase Cofactors. Chem. Rev 2020, 120 , 4921–4968.31975585 (51) Jasniewski AJ ; Lee CC ; Ribbe MW ; Hu Y Reactivity, Mechanism, and Assembly of the Alternative Nitrogenases. Chem. Rev 2020, 120 , 5107–5157.32129988 (52) Seefeldt LC ; Yang Z-Y ; Lukoyanov DA ; Harris DF ; Dean DR ; Raugei S ; Hoffman BM Reduction of Substrates by Nitrogenases. Chem. Rev 2020, 120 , 5082–5106.32176472 (53) Seefeldt LC ; Yang Z-Y ; Duval S ; Dean DR Nitrogenase Reduction of Carbon-Containing Compounds. Biochim. Biophys. Acta, Bioenerg 2013, 1827 , 1102–1111. (54) Schlögl R Catalytic Synthesis of Ammonia—A “Never-Ending Story”? Angew. Chemie Int. Ed 2003, 42 , 2004–2008. (55) Burgess BK Substrate Reactions of Nitrogenase. In Molybdenum Enzymes; Spiro TG , Ed.; Metal Ions In Biology; Wiley: New York, 1985; pp 161–219. (56) Thorneley RNF ; Lowe DJ Kinetics and Mechanism of the Nitrogenase Enzyme System. In Molybdenum Enzymes; Spiro TG , Ed.; Metal Ions In Biology; Wiley: New York, 1985; pp 221–284. (57) Burgess BK ; Wherland S ; Newton WE ; Stiefel EI Nitrogenase Reactivity: Insight into the Nitrogen-Fixing Process through Hydrogen-Inhibition and HD-Forming Reactions. Biochemistry 1981, 20 , 5140–5146.6945872 (58) Jensen BB ; Burris RH Effect of High PN2 and High PD2 on Ammonia Production, Hydrogen Evolution, and Hydrogen Deuteride Formation by Nitrogenases. Biochemistry 1985, 24 , 1141–1147.3913463 (59) Burgess BK ; Corbin JL ; Rubinson JF ; Li JG ; Dilworth MJ ; Newton WE Nitrogenase Reactivity. In Advances in Nitrogen Fixation Research: Proceedings of the 5th International Symposium on Nitrogen Fixation, Noordwijkerhout, The Netherlands, August 28 - September 3, 1983; Veeger C , Newton WE , Eds.; Springer Netherlands: Dordrecht, 1984; p 146. (60) Hoffman BM ; Lukoyanov D ; Yang Z-Y ; Dean DR ; Seefeldt LC Mechanism of Nitrogen Fixation by Nitrogenase: The Next Stage. Chem. Rev 2014, 114 , 4041–4062.24467365 (61) Einsle O ; Rees DC Structural Enzymology of Nitrogenase Enzymes. Chem. Rev 2020, 120 , 4969–5004.32538623 (62) Van Stappen C ; Decamps L ; Cutsail GE ; Bjornsson R ; Henthorn JT ; Birrell JA ; Debeer S The Spectroscopy of Nitrogenases. Chem. Rev 2020, 120 , 5005–5081.32237739 (63) Rutledge HL ; Tezcan FA Electron Transfer in Nitrogenase. Chem. Rev 2020, 120 , 5158–5193.31999100 (64) Dingler C ; Kuhla J ; Wassink H ; Oelze J Levels and Activities of Nitrogenase Proteins in Azotobacter Vinelandii Grown at Different Dissolved Oxygen Concentrations. J. Bacteriol 1988, 170 , 2148–2152.3162907 (65) Kirn J ; Rees DC Crystallographic Structure and Functional Implications of the Nitrogenase Molybdenum-Iron Protein from Azotobacter Vinelandii. Nature 1992, 360 , 553–560.25989647 (66) Spatzal T ; Aksoyoglu M ; Zhang L ; Andrade SLA ; Schleicher E ; Weber S ; Rees DC ; Einsle O Evidence for Interstitial Carbon in Nitrogenase FeMo Cofactor. Science 2011, 334 , 940.22096190 (67) Georgiadis MM ; Komiya H ; Chakrabarti P ; Woo D ; Kornuc JJ ; Rees DC Crystallographic Structure of the Nitrogenase Iron Protein from Azotobacter Vinelandii. Science 1992, 257 , 1653–1659.1529353 (68) Jasniewski A ; Sickerman N ; Hu Y ; Ribbe M The Fe Protein: An Unsung Hero of Nitrogenase. Inorganics 2018, 6 , 25. (69) Schindelin H ; Kisker C ; Schlessman JL ; Howard JB ; Rees DC Structure of ADP·AlF4-Stabilized Nitrogenase Complex and Its Implications for Signal Transduction. Nature 1997, 387 , 370–376.9163420 (70) Tezcan FA ; Kaiser JT ; Mustafi D ; Walton MY ; Howard JB ; Rees DC Nitrogenase Complexes: Multiple Docking Sites for a Nucleotide Switch Protein. Science 2005, 309 , 1377–1380.16123301 (71) Peters JW ; Fisher K ; Newton WE ; Dean DR Involvement of the P Cluster in Intramolecular Electron Transfer within the Nitrogenase MoFe Protein. J. Biol. Chem 1995, 270 , 27007–27013.7592949 (72) Chan MK ; Kim J ; Rees DC The Nitrogenase FeMo-Cofactor and P-Cluster Pair: 2.2 Å Resolution Structures. Science 1993, 260 , 792–794.8484118 (73) Pierik AJ ; Wassink H ; Haaker H ; Hagen WR Redox Properties and EPR Spectroscopy of the P Clusters of Azotobacter Vinelandii MoFe Protein. Eur. J. Biochem 1993, 212 , 51–61.8383042 (74) Smith BE ; Lang G Mössbauer Spectroscopy of the Nitrogenase Proteins from Klebsiella Pneumoniae. Structural Assignments and Mechanistic Conclusions. Biochem. J 1974, 137 , 169–180.4596139 (75) Huynh BH ; Henzl MT ; Christner JA ; Zimmermann R ; Orme-Johnson WH ; Münck E Nitrogenase XII. Mössbauer Studies of the MoFe Protein from Clostridium Pasteurianum W5. Biochim. Biophys. Acta, Protein Struct 1980, 623 , 124–138. (76) Zimmermann R ; Münck E ; Brill WJ ; Shah VK ; Henzl MT ; Rawlings J ; Orme-Johnson WH Nitrogenase X: Mössbauer and EPR Studies on Reversibly Oxidized MoFe Protein from Azotobacter Vinelandii OP Nature of the Iron Centers. Biochim. Biophys. Acta - Protein Struct 1978, 537 , 185–207. (77) McLean PA ; Papaefthymiou V ; Orme-Johnson WH ; Münck E Isotopic Hybrids of Nitrogenase. Mössbauer Study of MoFe Protein with Selective 57Fe Enrichment of the P-Cluster. J. Biol. Chem 1987, 262 , 12900–12903.2820958 (78) Peters JW ; Stowell MHB ; Soltis SM ; Finnegan MG ; Johnson MK ; Rees DC Redox-Dependent Structural Changes in the Nitrogenase P-Cluster. Biochemistry 1997, 36 , 1181–1187.9063865 (79) Keable SM ; Vertemara J ; Zadvornyy OA ; Eilers BJ ; Danyal K ; Rasmussen AJ ; De Gioia L ; Zampella G ; Seefeldt LC ; Peters JW Structural Characterization of the Nitrogenase Molybdenum-Iron Protein with the Substrate Acetylene Trapped near the Active Site. J. Inorg. Biochem 2018, 180 , 129–134.29275221 (80) Mayer SM ; Lawson DM ; Gormal CA ; Roe SM ; Smith BE New Insights into Structure-Function Relationships in Nitrogenase: A 1.6 Å Resolution X-Ray Crystallographic Study of Klebsiella Pneumoniae MoFe-Protein. J. Mol. Biol 1999, 292 , 871–891.10525412 (81) Danyal K ; Dean DR ; Hoffman BM ; Seefeldt LC Electron Transfer within Nitrogenase: Evidence for a Deficit-Spending Mechanism. Biochemistry 2011, 50 , 9255–9263.21939270 (82) Danyal K ; Mayweather D ; Dean DR ; Seefeldt LC ; Hoffman BM Conformational Gating of Electron Transfer from the Nitrogenase Fe Protein to MoFe Protein. J. Am. Chem. Soc 2010, 132 , 6894–6895.20429505 (83) Seefeldt LC ; Hoffman BM ; Peters JW ; Raugei S ; Beratan DN ; Antony E ; Dean DR Energy Transduction in Nitrogenase. Acc. Chem. Res 2018, 51 , 2179–2186.30095253 (84) Rupnik K ; Lee CC ; Wiig JA ; Hu Y ; Ribbe MW ; Hales BJ Nonenzymatic Synthesis of the P-Cluster in the Nitrogenase MoFe Protein: Evidence of the Involvement of All-Ferrous [Fe4S4] 0 Intermediates. Biochemistry 2014, 53 , 1108–1116.24520862 (85) Kim J ; Rees DC Structural Models for the Metal Centers in the Nitrogenase Molybdenum-Iron Protein. Science 1992, 257 , 1677–1682.1529354 (86) Lancaster KM ; Roemelt M ; Ettenhuber P ; Hu Y ; Ribbe MW ; Neese F ; Bergmann U ; DeBeer S X-Ray Emission Spectroscopy Evidences a Central Carbon in the Nitrogenase Iron-Molybdenum Cofactor. Science 2011, 334 , 974–977.22096198 (87) Wiig JA ; Hu Y ; Lee CC ; Ribbe MW Radical SAM-Dependent Carbon Insertion into the Nitrogenase M-Cluster. Science 2012, 337 , 1672–1675.23019652 (88) Wiig JA ; Hu Y ; Ribbe MW Refining the Pathway of Carbide Insertion into the Nitrogenase M-Cluster. Nat. Commun 2015, 6 , 8034.26259825 (89) Spatzal T ; Perez KA ; Einsle O ; Howard JB ; Rees DC Ligand Binding to the FeMo-Cofactor: Structures of CO-Bound and Reactivated Nitrogenase. Science 2014, 345 , 1620–1623.25258081 (90) Spatzal T ; Perez KA ; Howard JB ; Rees DC Catalysis-Dependent Selenium Incorporation and Migration in the Nitrogenase Active Site Iron-Molybdenum Cofactor. Elife 2015, 4 , No. e11620.26673079 (91) Kang W ; Lee CC ; Jasniewski AJ ; Ribbe MW ; Hu Y Structural Evidence for a Dynamic Metallocofactor during N 2 Reduction by Mo-Nitrogenase. Science 2020, 368 , 1381–1385.32554596 (92) Sippel D ; Einsle O The Structure of Vanadium Nitrogenase Reveals an Unusual Bridging Ligand. Nat. Chem. Biol 2017, 13 , 956–960.28692069 (93) Sippel D ; Rohde M ; Netzer J ; Trncik C ; Gies J ; Grunau K ; Djurdjevic I ; Decamps L ; Andrade SLA ; Einsle O A Bound Reaction Intermediate Sheds Light on the Mechanism of Nitrogenase. Science 2018, 359 , 1484–1489.29599235 (94) Rohde M ; Grunau K ; Einsle O CO Binding to the FeV Cofactor of CO-Reducing Vanadium Nitrogenase at Atomic Resolution. Angew. Chemie Int. Ed 2020, 59 , 23626–23630. (95) Rohde M ; Laun K ; Zebger I ; Stripp ST ; Einsle O Two Ligand-Binding Sites in CO-Reducing V Nitrogenase Reveal a General Mechanistic Principle. Sci. Adv 2021, 7 , No. eabg4474.34049880 (96) Orme-Johnson WH ; Hamilton WD ; Jones TL ; Tso MYW ; Burris RH ; Shah VK ; Brill WJ Electron Paramagnetic Resonance of Nitrogenase and Nitrogenase Components from Clostridium Pasteurianum W5 and Azotobacter Vinelandii OP. Proc. Natl. Acad. Sci. U. S. A 1972, 69 , 3142–3145.4343957 (97) Yoo SJ ; Angove HC ; Papaefthymiou V ; Burgess BK ; Münck E Mössbauer Study of the MoFe Protein of Nitrogenase from Azotobacter Vinelandii Using Selective 57Fe Enrichment of the M-Centers. J. Am. Chem. Soc 2000, 122 , 4926–4936. (98) Bjornsson R ; Lima FA ; Spatzal T ; Weyhermuller T ; Glatzel P ; Bill E ; Einsle O ; Neese F ; DeBeer S Identification of a Spin-Coupled Mo(III) in the Nitrogenase Iron-Molybdenum Cofactor. Chem. Sci 2014, 5 , 3096–3103. (99) Morgan TV ; Mortenson LE ; McDonald JW ; Watt GD Comparison of Redox and EPR Properties of the Molybdenum Iron Proteins of Clostridium Pasteurianum and Azotobacter Vinelandii Nitrogenases. J. Inorg. Biochem 1988, 33 , 111–120.2842451 (100) Burgess BK ; Lowe DJ Mechanism of Molybdenum Nitrogenase. Chem. Rev 1996, 96 , 2983–3012.11848849 (101) Bjornsson R ; Neese F ; DeBeer S Revisiting the Mössbauer Isomer Shifts of the FeMoco Cluster of Nitrogenase and the Cofactor Charge. Inorg. Chem 2017, 56 , 1470–1477.28071903 (102) Benediktsson B ; Bjornsson R QM/MM Study of the Nitrogenase MoFe Protein Resting State: Broken-Symmetry States, Protonation States, and QM Region Convergence in the FeMoco Active Site. Inorg. Chem 2017, 56 , 13417–13429.29053260 (103) Spatzal T ; Schlesier J ; Burger E-M ; Sippel D ; Zhang L ; Andrade SLA ; Rees DC ; Einsle O Nitrogenase FeMoco Investigated by Spatially Resolved Anomalous Dispersion Refinement. Nat. Commun 2016, 7 , 10902.26973151 (104) O’Donnell MJ ; Smith BE Electron-Paramagnetic-Resonance Studies on the Redox Properties of the Molybdenum-Iron Protein of Nitrogenase between + 50 and −450 MV. Biochem. J 1978, 173 , 831–838.30448 (105) Watt GD ; Burns A ; Lough S ; Tennent DL Redox and Spectroscopic Properties of Oxidized MoFe Protein from Azotobacter Vinelandii. Biochemistry 1980, 19 , 4926–4932.6252962 (106) Walker JE ; Saraste M ; Runswick MJ ; Gay NJ Distantly Related Sequences in the Alpha- and Beta-Subunits of ATP Synthase, Myosin, Kinases and Other ATP-Requiring Enzymes and a Common Nucleotide Binding Fold. EMBO J 1982, 1 , 945–951.6329717 (107) Howard JB ; Rees DC Structural Basis of Biological Nitrogen Fixation. Chem. Rev 1996, 96 , 2965–2982.11848848 (108) Jang SB ; Seefeldt LC ; Peters JW Insights into Nucleotide Signal Transduction in Nitrogenase: Structure of an Iron Protein with MgADP Bound. Biochemistry 2000, 39 , 14745–14752.11101289 (109) Chen L ; Gavini N ; Tsuruta H ; Eliezer D ; Burgess BK ; Doniach S ; Hodgson KO MgATP-Induced Conformational Changes in the Iron Protein from Azotobacter Vinelandii, as Studied by Small-Angle X-Ray Scattering. J. Biol. Chem 1994, 269 , 3290–3294.8106367 (110) Ryle MJ ; Seefeldt LC Elucidation of a MgATP Signal Transduction Pathway in the Nitrogenase Iron Protein: Formation of a Conformation Resembling the MgATP-Bound State by Protein Engineering. Biochemistry 1996, 35 , 4766–4775.8664266 (111) Ryle MJ ; Lanzilotta WN ; Seefeldt LC ; Scarrow RC ; Jensen GM Circular Dichroism and X-Ray Spectroscopies of Azotobacter Vinelandii Nitrogenase Iron Protein: MgATP and MgADP Induced Protein Conformation Changes Affecting the [4Fe-4S] Cluster and Characterization of a [2Fe-2S] Form. J. Biol. Chem 1996, 271 , 1551–1557.8576152 (112) Stephens PJ ; McKenna CE ; Smith BE ; Nguyen HT ; McKenna M-C ; Thomson AJ ; Devlin F ; Jones JB Circular Dichroism and Magnetic Circular Dichroism of Nitrogenase Proteins. Proc. Natl. Acad. Sci. U. S. A 1979, 76 , 2585–2589.379860 (113) Liu J ; Chakraborty S ; Hosseinzadeh P ; Yu Y ; Tian S ; Petrik I ; Bhagi A ; Lu Y Metalloproteins Containing Cytochrome, Iron-Sulfur, or Copper Redox Centers. Chem. Rev 2014, 114 , 4366–4369.24758379 (114) Lindahl PA ; Day EP ; Kent TA ; Orme-Johnson WH ; Münck E Mössbauer, EPR, and Magnetization Studies of the Azotobacter Vinelandii Fe Protein. Evidence for a [4Fe-4S]1+ Cluster with Spin S = 3/2. J. Biol. Chem 1985, 260 , 11160–11173.2993304 (115) Yates MG Electron Transport to Nitrogenase in Azotobacter Chroococcum: Azotobacter Flavodoxin Hydroquinone as an Electron Donor. FEBS Lett 1972, 27 , 63–67.11946808 (116) Thorneley RN ; Deistung J Electron-Transfer Studies Involving Flavodoxin and a Natural Redox Partner, the Iron Protein of Nitrogenase. Conformational Constraints on Protein-Protein Interactions and the Kinetics of Electron Transfer within the Protein Complex. Biochem. J 1988, 253 , 587–595.3140782 (117) Bennett LT ; Jacobson MR ; Dean DR Isolation, Sequencing, and Mutagenesis of the NifF Gene Encoding Flavodoxin from Azotobacter Vinelandii. J. Biol. Chem 1988, 263 , 1364–1369.3121629 (118) Duyvis MG ; Wassink H ; Haaker H Nitrogenase of Azotobacter Vinelandii: Kinetic Analysis of the Fe Protein Redox Cycle. Biochemistry 1998, 37 , 17345–17354.9860849 (119) Setubal JC ; dos Santos P ; Goldman BS ; Ertesvåg H ; Espin G ; Rubio LM ; Valla S ; Almeida NF ; Balasubramanian D ; Cromes L ; Genome Sequence of Azotobacter Vinelandii, an Obligate Aerobe Specialized To Support Diverse Anaerobic Metabolic Processes. J. Bacteriol 2009, 191 , 4534–4545.19429624 (120) Sickerman NS ; Rettberg LA ; Lee CC ; Hu Y ; Ribbe MW Cluster Assembly in Nitrogenase. Essays Biochem 2017, 61 , 271–279.28487403 (121) Hu Y ; Corbett MC ; Fay AW ; Webber JA ; Hedman B ; Hodgson KO ; Ribbe MW Nitrogenase Reactivity with P-Cluster Variants. Proc. Natl. Acad. Sci. U. S. A 2005, 102 , 13825–13830.16166259 (122) Hu Y ; Fay AW ; Lee CC ; Ribbe MW P-Cluster Maturation on Nitrogenase MoFe Protein. Proc. Natl. Acad. Sci. U. S. A 2007, 104 , 10424–10429.17563349 (123) Lee CC ; Blank MA ; Fay AW ; Yoshizawa JM ; Hu Y ; Hodgson KO ; Hedman B ; Ribbe MW Stepwise Formation of P-Cluster in Nitrogenase MoFe Protein. Proc. Natl. Acad. Sci. U. S. A 2009, 106 , 18474–18478.19828444 (124) Rupnik K ; Hu Y ; Lee CC ; Wiig JA ; Ribbe MW ; Hales BJ P+ State of Nitrogenase P-Cluster Exhibits Electronic Structure of a [Fe4S4]+ Cluster. J. Am. Chem. Soc 2012, 134 , 13749–13754.22839751 (125) Kennedy C ; Dean D The NifU, NifS and NifV Gene Products Are Required for Activity of All Three Nitrogenases of Azotobacter Vinelandii. Mol. Gen. Genet 1992, 231 , 494–498.1538703 (126) Zheng L ; Dean DR Catalytic Formation of a Nitrogenase Iron-Sulfur Cluster. J. Biol. Chem 1994, 269 , 18723–18726.8034623 (127) Shah VK ; Allen JR ; Spangler NJ ; Ludden PW In Vitro Synthesis of the Iron-Molybdenum Cofactor of Nitrogenase. Purification and Characterization of NifB Cofactor, the Product of NIFB Protein. J. Biol. Chem 1994, 269 , 1154–1158.8288575 (128) Sickerman NS ; Ribbe MW ; Hu Y Nitrogenase Cofactor Assembly: An Elemental Inventory. Acc. Chem. Res 2017, 50 , 2834–2841.29064664 (129) Wiig JA ; Hu Y ; Ribbe MW NifEN-B Complex of Azotobacter Vinelandii Is Fully Functional in Nitrogenase FeMo Cofactor Assembly. Proc. Natl. Acad. Sci. U. S. A 2011, 108 , 8623–8627.21551100 (130) Kang W ; Rettberg LA ; Stiebritz MT ; Jasniewski AJ ; Tanifuji K ; Lee CC ; Ribbe MW ; Hu Y X-Ray Crystallographic Analysis of NifB with a Full Complement of Clusters: Structural Insights into the Radical SAM-Dependent Carbide Insertion During Nitrogenase Cofactor Assembly. Angew. Chemie Int. Ed 2021, 60 , 2364–2370. (131) Rettberg LA ; Wilcoxen J ; Lee CC ; Stiebritz MT ; Tanifuji K ; Britt RD ; Hu Y Probing the Coordination and Function of Fe4S4Modules in Nitrogenase Assembly Protein NifB. Nat. Commun 2018, 9 , 2824.30026506 (132) Jasniewski AJ ; Wilcoxen J ; Tanifuji K ; Hedman B ; Hodgson KO ; Britt RD ; Hu Y ; Ribbe MW Spectroscopic Characterization of an Eight-Iron Nitrogenase Cofactor Precursor That Lacks the “9th Sulfur. Angew. Chemie Int. Ed 2019, 58 , 14703–14707. (133) Tanifuji K ; Lee CC ; Sickerman NS ; Tatsumi K ; Ohki Y ; Hu Y ; Ribbe MW Tracing the ‘Ninth Sulfur’ of the Nitrogenase Cofactor via a Semi-Synthetic Approach. Nat. Chem 2018, 10 , 568–572.29662207 (134) Hu Y ; Corbett MC ; Fay AW ; Webber JA ; Hodgson KO ; Hedman B ; Ribbe MW FeMo Cofactor Maturation on NifEN. Proc. Natl. Acad. Sci. U. S. A 2006, 103 , 17119–17124.17050696 (135) Yoshizawa JM ; Blank MA ; Fay AW ; Lee CC ; Wiig JA ; Hu Y ; Hodgson KO ; Hedman B ; Ribbe MW Optimization of FeMoco Maturation on NifEN. J. Am. Chem. Soc 2009, 131 , 9321–9325.19514721 (136) Hu Y ; Corbett MC ; Fay AW ; Webber JA ; Hodgson KO ; Hedman B ; Ribbe MW Nitrogenase Fe Protein: A Molybdate/Homocitrate Insertase. Proc. Natl. Acad. Sci. U. S. A 2006, 103 , 17125–17130.17062756 (137) Fay AW ; Blank MA ; Rebelein JG ; Lee CC ; Ribbe MW ; Hedman B ; Hodgson KO ; Hu Y Assembly Scaffold NifEN: A Structural and Functional Homolog of the Nitrogenase Catalytic Component. Proc. Natl. Acad. Sci. U. S. A 2016, 113 , 9504–9508.27506795 (138) Tezcan FA ; Kaiser JT ; Howard JB ; Rees DC Structural Evidence for Asymmetrical Nucleotide Interactions in Nitrogenase. J. Am. Chem. Soc 2015, 137 , 146–149.25522159 (139) Fisher K ; Dilworth MJ ; Newton WE Differential Effects on N2 Binding and Reduction, HD Formation, and Azide Reduction with α-195His- and α-191Gln-Substituted MoFe Proteins of Azotobacter Vinelandii Nitrogenase. Biochemistry 2000, 39 , 15570–15577.11112544 (140) Yang Z-Y ; Khadka N ; Lukoyanov D ; Hoffman BM ; Dean DR ; Seefeldt LC On Reversible H2 Loss upon N2 Binding to FeMo-Cofactor of Nitrogenase. Proc. Natl. Acad. Sci. U. S. A 2013, 110 , 16327–16332.24062454 (141) Chatt J ; Pearman AJ ; Richards RL The Reduction of Mono-Coordinated Molecular Nitrogen to Ammonia in a Protic Environment. Nature 1975, 253 , 39–40. (142) Chatt J ; Pearman AJ ; Richards RL Relevance of Oxygen Ligands to Reduction of Ligating Dinitrogen. Nature 1976, 259 , 204. (143) Chatt J ; Pearman AJ ; Richards RL Conversion of Dinitrogen in Its Molybdenum and Tungsten Complexes into Ammonia and Possible Relevance to the Nitrogenase Reaction. J. Chem. Soc. Dalt. Trans 1977, No. 19 , 1852–1860. (144) Chatt J ; Pearman AJ ; Richards RL Diazenido (Iminonitrosyl) (N2H), Hydrazido(2-) (N2H2, and Hydrazido(1-) (N2H3) Ligands as Intermediates in the Reduction of Ligating Dinitrogen to Ammonia. J. Organomet. Chem 1975, 101 , C45–C47. (145) Yandulov DV ; Schrock RR Reduction of Dinitrogen to Ammonia at a Well-Protected Reaction Site in a Molybdenum Triamidoamine Complex. J. Am. Chem. Soc 2002, 124 , 6252–6253.12033849 (146) Yandulov DV ; Schrock RR Catalytic Reduction of Dinitrogen to Ammonia at a Single Molybdenum Center. Science 2003, 301 , 76–78.12843387 (147) Hinnemann B ; Nørskov JK Catalysis by Enzymes: The Biological Ammonia Synthesis. Top. Catal 2006, 37 , 55–70. (148) Tanabe Y ; Nishibayashi Y Developing More Sustainable Processes for Ammonia Synthesis. Coord. Chem. Rev 2013, 257 , 2551–2564. (149) Harris DF ; Lukoyanov DA ; Kallas H ; Trncik C ; Yang Z-Y ; Compton P ; Kelleher N ; Einsle O ; Dean DR ; Hoffman BM ; Mo-, V-, and Fe-Nitrogenases Use a Universal Eight-Electron Reductive-Elimination Mechanism To Achieve N2 Reduction. Biochemistry 2019, 58 , 3293–3301.31283201 (150) Harris DF ; Yang Z-Y ; Dean DR ; Seefeldt LC ; Hoffman BM Kinetic Understanding of N2 Reduction versus H2 Evolution at the E4(4H) Janus State in the Three Nitrogenases. Biochemistry 2018, 57 , 5706–5714.30183278 (151) Einsle O ; Tezcan FA ; Andrade SLA ; Schmid B ; Yoshida M ; Howard JB ; Rees DC Nitrogenase MoFe-Protein at 1.16 Å Resolution: A Central Ligand in the FeMo-Cofactor. Science 2002, 297 , 1696–1700.12215645 (152) Thomann H ; Bernardo M ; Newton WE ; Dean DR N Coordination of FeMo Cofactor Requires His-195 of the MoFe Protein Alpha Subunit and Is Essential for Biological Nitrogen Fixation. Proc. Natl. Acad. Sci. U. S. A 1991, 88 , 6620–6623.11607203 (153) DeRose VJ ; Kim C-H ; Newton WE ; Dean DR ; Hoffman BM Electron Spin Echo Envelope Modulation Spectroscopic Analysis of Altered Nitrogenase MoFe Proteins from Azotobacter Vinelandii. Biochemistry 1995, 34 , 2809–2814.7893692 (154) Lee H-I ; Thrasher KS ; Dean DR ; Newton WE ; Hoffman BM 14N Electron Spin-Echo Envelope Modulation of the S = 3/2 Spin System of the Azotobacter Vinelandii Nitrogenase Iron-Molybdenum Cofactor. Biochemistry 1998, 37 , 13370–13378.9748344 (155) Schmid B ; Ribbe MW ; Einsle O ; Yoshida M ; Thomas LM ; Dean DR ; Rees DC ; Burgess BK Structure of a Cofactor-Deficient Nitrogenase MoFe Protein. Science 2002, 296 , 352–356.11951047 (156) Fay AW ; Hu Y ; Schmid B ; Ribbe MW Molecular Insights into Nitrogenase FeMoco Insertion - The Role of His 274 and His 451 of MoFe Protein α Subunit. J. Inorg. Biochem 2007, 101 , 1630–1641.17521738 (157) Hu Y ; Fay AW ; Ribbe MW Molecular Insights into Nitrogenase FeMo Cofactor Insertion: The Role of His 362 of the MoFe Protein α Subunit in FeMo Cofactor Incorporation. JBIC J. Biol. Inorg. Chem 2007, 12 , 449–460.17203313 (158) Hu Y ; Fay AW ; Schmid B ; Makar B ; Ribbe MW Molecular Insights into Nitrogenase FeMoco Insertion: Trp-444 of MoFe Protein α-Subunit Locks FeMoco in its Binding Site. J. Biol. Chem 2006, 281 , 30534–30541.16893897 (159) Kim J ; Woo D ; Rees DC X-Ray Crystal Structure of the Nitrogenase Molybdenum-Iron Protein from Clostridium Pasteurianum at 3.0-Å Resolution. Biochemistry 1993, 32 , 7104–7115.8393705 (160) Shah VK ; Brill WJ Isolation of an Iron-Molybdenum Cofactor from Nitrogenase. Proc. Natl. Acad. Sci. U. S. A 1977, 74 , 3249–3253.410019 (161) Burgess BK The Iron-Molybdenum Cofactor of Nitrogenase. Chem. Rev 1990, 90 , 1377–1406. (162) Tanifuji K ; Ohki Y Metal-Sulfur Compounds in N2 Reduction and Nitrogenase-Related Chemistry. Chem. Rev 2020, 120 , 5194–5251.32459087 (163) Lundell DJ ; Howard JB Isolation and Partial Characterization of Two Different Subunits from the Molybdenum-Iron Protein of Azotobacter Vinelandii Nitrogenase. J. Biol. Chem 1978, 253 , 3422–3426.649581 (164) Hausinger RP ; Howard JB The Amino Acid Sequence of the Nitrogenase Iron Protein from Azotobacter Vinelandii. J. Biol. Chem 1982, 257 , 2483–2490.6801032 (165) Brigle KE ; Newton WE ; Dean DR Complete Nucleotide Sequence of the Azotobacter Vinelandii Nitrogenase Structural Gene Cluster. Gene 1985, 37 , 37–44.3863780 (166) Wang SZ ; Chen JS ; Johnson JL Distinct Structural Features of the α and β Subunits of Nitrogenase Molybdenum-Iron Protein of Clostridium Pasteurianum: An Analysis of Amino Acid Sequences. Biochemistry 1988, 27 , 2800–2810.2840948 (167) Holland D ; Zilberstein A ; Zamir A ; Sussman JL A Quantitative Approach to Sequence Comparisons of Nitrogenase MoFe Protein α- and β-Subunits Including the Newly Sequenced NifK Gene from Klebsiella Pneumoniae. Biochem. J 1987, 247 , 277–285.3322261 (168) Ioannidis I ; Buck M Nucleotide Sequence of the Klebsiella Pneumoniae NifD Gene and Predicted Amino Acid Sequence of the α-Subunit of Nitrogenase MoFe Protein. Biochem. J 1987, 247 , 287–291.3322262 (169) Brigle KE ; Setterquist RA ; Dean DR ; Cantwell JS ; Weiss MC ; Newton WE Site-Directed Mutagenesis of the Nitrogenase MoFe Protein of Azotobacter Vinelandii. Proc. Natl. Acad. Sci. U. S. A 1987, 84 , 7066–7069.16593879 (170) Dean DR ; Brigle KE ; May HD ; Newton WE Site-Directed Mutagenesis of the Nitrogenase MoFe Protein. In Nitrogen fixation: hundred years after; Bothe H , de Bruijn FJ , Newton WE , Eds.; Gustav Fischer: Stuttgart; New York, 1988; pp 107–113. (171) Kent HM ; Baines M ; Gormal C ; Smith BE ; Buck M Analysis of Site-Directed Mutations in the α-and β-Subunits of Klebsiella Pneumoniae Nitrogenase. Mol. Microbiol 1990, 4 , 1497–1504. (172) Kent HM ; Ioannidis I ; Gormal C ; Smith BE ; Buck M Site-Directed Mutagenesis of the Klebsiella Pneumoniae Nitrogenase. Effects of Modifying Conserved Cysteine Residues in the α- and β-Subunits. Biochem. J 1989, 264 , 257–264.2557830 (173) Meyer J Iron-Sulfur Protein Folds, Iron-Sulfur Chemistry, and Evolution. JBIC J. Biol. Inorg. Chem 2008, 13 , 157–170.17992543 (174) Christiansen J ; Cash VL ; Seefeldt LC ; Dean DR Isolation and Characterization of an Acetylene-Resistant Nitrogenase. J. Biol. Chem 2000, 275 , 11459–11464.10753963 (175) Christiansen J ; Seefeldt LC ; Dean DR Competitive Substrate and Inhibitor Interactions at the Physiologically Relevant Active Site of Nitrogenase*. J. Biol. Chem 2000, 275 , 36104–36107.10948195 (176) Mayer SM ; Niehaus WG ; Dean DR Reduction of Short Chain Alkynes by a Nitrogenase α-70Ala-Substituted MoFe Protein. J. Chem. Soc. Dalt. Trans 2002, No. 5 , 802–807. (177) Barney BM ; Igarashi RY ; Dos Santos PC ; Dean DR ; Seefeldt LC Substrate Interaction at an Iron-Sulfur Face of the FeMo-Cofactor during Nitrogenase Catalysis. J. Biol. Chem 2004, 279 , 53621–53624.15465817 (178) Dos Santos PC ; Mayer SM ; Barney BM ; Seefeldt LC ; Dean DR Alkyne Substrate Interaction within the Nitrogenase MoFe Protein. J. Inorg. Biochem 2007, 101 , 1642–1648.17610955 (179) Yang Z-Y ; Dean DR ; Seefeldt LC Molybdenum Nitrogenase Catalyzes the Reduction and Coupling of CO to Form Hydrocarbons. J. Biol. Chem 2011, 286 , 19417–19421.21454640 (180) Benton PMC ; Mayer SM ; Shao J ; Hoffman BM ; Dean DR ; Seefeldt LC Interaction of Acetylene and Cyanide with the Resting State of Nitrogenase α-96-Substituted MoFe Proteins. Biochemistry 2001, 40 , 13816–13825.11705370 (181) Maskos Z ; Fisher K ; Sørlie M ; Newton WE ; Hales BJ Variant MoFe Proteins of Azotobacter Vinelandii: Effects of Carbon Monoxide on Electron Paramagnetic Resonance Spectra Generated during Enzyme Turnover. JBIC, J. Biol. Inorg. Chem 2005, 10 , 394–406.15887041 (182) Scott DJ ; May HD ; Newton WE ; Brigle KE ; Dean DR Role for the Nitrogenase MoFe Protein α-Subunit in FeMo-Cofactor Binding and Catalysis. Nature 1990, 343 , 188–190.2153269 (183) Fisher K ; Dilworth MJ ; Kim C-H ; Newton WE Azotobacter Vinelandii Nitrogenases Containing Altered MoFe Proteins with Substitutions in the FeMo-Cofactor Environment: Effects on the Catalyzed Reduction of Acetylene and Ethylene. Biochemistry 2000, 39 , 2970–2979.10715117 (184) Sørlie M ; Christiansen J ; Lemon BJ ; Peters JW ; Dean DR ; Hales BJ Mechanistic Features and Structure of the Nitrogenase α-Gln195 MoFe Protein. Biochemistry 2001, 40 , 1540–1549.11327812 (185) Scott DJ ; Dean DR ; Newton WE Nitrogenase-Catalyzed Ethane Production and CO-Sensitive Hydrogen Evolution from MoFe Proteins Having Amino Acid Substitutions in an Alpha-Subunit FeMo Cofactor-Binding Domain. J. Biol. Chem 1992, 267 , 20002–20010.1328190 (186) Fisher K ; Hare ND ; Newton WE Another Role for CO with Nitrogenase? CO Stimulates Hydrogen Evolution Catalyzed by Variant Azotobacter Vinelandii Mo-Nitrogenases. Biochemistry 2014, 53 , 6151–6160.25203280 (187) Kim C-H ; Newton WE ; Dean DR Role of the MoFe Protein α-Subunit Histidine-195 Residue in FeMo-Cofactor Binding and Nitrogenase Catalysis. Biochemistry 1995, 34 , 2798–2808.7893691 (188) Dilworth MJ ; Fisher K ; Kim C-H ; Newton WE Effects on Substrate Reduction of Substitution of Histidine-195 by Glutamine in the α-Subunit of the MoFe Protein of Azotobacter Vinelandii Nitrogenase. Biochemistry 1998, 37 , 17495–17505.9860864 (189) Lee H-I ; Sørlie M ; Christiansen J ; Yang T-C ; Shao J ; Dean DR ; Hales BJ ; Hoffman BM Electron Inventory, Kinetic Assignment (En), Structure, and Bonding of Nitrogenase Turnover Intermediates with C2H2 and CO. J. Am. Chem. Soc 2005, 127 , 15880–15890.16277531 (190) Shen J ; Dean DR ; Newton WE Evidence for Multiple Substrate-Reduction Sites and Distinct Inhibitor-Binding Sites from an Altered Azotobacter Vinelandii Nitrogenase MoFe Protein. Biochemistry 1997, 36 , 4884–4894.9125509 (191) Fisher K ; Dilworth MJ ; Kim C-H ; Newton WE Azotobacter Vinelandii Nitrogenases with Substitutions in the FeMo-Cofactor Environment of the MoFe Protein: Effects of Acetylene or Ethylene on Interactions with H+, HCN, and CN-. Biochemistry 2000, 39 , 10855–10865.10978172 (192) Simpson FB ; Burris RH A Nitrogen Pressure of 50 atm Does Not Prevent Evolution of Hydrogen by Nitrogenase. Science 1984, 224 , 1095–1097.6585956 (193) Dilworth MJ Acetylene Reduction by Nitrogen-Fixing Preparations from Clostridium Pasteurianum. Biochim. Biophys. Acta, Gen. Subj 1966, 127 , 285–294. (194) Seefeldt LC ; Hoffman BM ; Dean DR Mechanism of Mo-Dependent Nitrogenase. Annu. Rev. Biochem 2009, 78 , 701–722.19489731 (195) Lowe DJ ; Fisher K ; Thorneley RN Klebsiella Pneumoniae Nitrogenase. Mechanism of Acetylene Reduction and Its Inhibition by Carbon Monoxide. Biochem. J 1990, 272 , 621–625.2268290 (196) McLean PA ; Dixon RA Requirement of NifV Gene for Production of Wild-Type Nitrogenase Enzyme in Klebsiella Pneumoniae. Nature 1981, 292 , 655–656.7019729 (197) Liang J ; Madden M ; Shah VK ; Burris RH Citrate Substitutes for Homocitrate in Nitrogenase of a NifV Mutant of Klebsiella Pneumoniae. Biochemistry 1990, 29 , 8577–8581.2271541 (198) Dilworth MJ ; Eady RR Hydrazine Is a Product of Dinitrogen Reduction by the Vanadium-Nitrogenase from Azotobacter Chroococcum. Biochem. J 1991, 277 , 465–468.1859374 (199) Thorneley RNF ; Eady RR ; Lowe DJ Biological Nitrogen Fixation by Way of an Enzyme-Bound Dinitrogen-Hydride Intermediate. Nature 1978, 272 , 557–558. (200) Rivera-Ortiz JM ; Burris RH Interactions among Substrates and Inhibitors of Nitrogenase. J. Bacteriol 1975, 123 , 537–545.1150625 (201) Hwang JC ; Chen CH ; Burris RH Inhibition of Nitrogenase-Catalyzed Reductions. Biochim. Biophys. Acta - Bioenerg 1973, 292 , 256–270. (202) Davis LC ; Henzl MT ; Burris RH ; Orme-Johnson WH Iron-Sulfur Clusters in the Molybdenum-Iron Protein Component of Nitrogenase. Electron Paramagnetic Resonance of the Carbon Monoxide Inhibited State. Biochemistry 1979, 18 , 4860–4869.228701 (203) Lowe DJ ; Eady RR ; Thorneley RNF Electron-Paramagnetic-Resonance Studies on Nitrogenase of Klebsiella Pneumoniae. Evidence for Acetylene- and Ethylene-Nitrogenase Transient Complexes. Biochem. J 1978, 173 , 277–290.210766 (204) Davis LC ; Shah VK ; Brill WJ Nitrogenase: VII. Effect of Component Ratio, ATP and H2, on the Distribution of Electrons to Alternative Substrates. Biochim. Biophys. Acta - Enzymol 1975, 403 , 67–78. (205) Benton PMC ; Christiansen J ; Dean DR ; Seefeldt LC Stereospecificity of Acetylene Reduction Catalyzed by Nitrogenase. J. Am. Chem. Soc 2001, 123 , 1822–1827.11456800 (206) Hoffman BM ; Lukoyanov D ; Dean DR ; Seefeldt LC Nitrogenase: A Draft Mechanism. Acc. Chem. Res 2013, 46 , 587–595.23289741 (207) Bergmann J ; Oksanen E ; Ryde U Critical Evaluation of a Crystal Structure of Nitrogenase with Bound N2 Ligands. JBIC J. Biol. Inorg. Chem 2021, 26 , 341–353.33713183 (208) Kang W ; Lee CC ; Jasniewski AJ ; Ribbe MW ; Hu Y Response to Comment on “Structural Evidence for a Dynamic Metallocofactor during N2 Reduction by Mo-Nitrogenase. Science 2021, 371 , No. eabe5856.33574184 (209) Peters JW ; Einsle O ; Dean DR ; DeBeer S ; Hoffman BM ; Holland PL ; Seefeldt LC Comment on “Structural Evidence for a Dynamic Metallocofactor during N2 Reduction by Mo-Nitrogenase. Science 2021, 371 , No. eabe5481.33574183 (210) Lockshin A ; Burris RH Inhibitors of Nitrogen Fixation in Extracts from Clostridium Pasteurianum. Biochim. Biophys. Acta - Gen. Subj 1965, 111 , 1–10. (211) Cameron LM ; Hales BJ Investigation of CO Binding and Release from Mo-Nitrogenase during Catalytic Turnover. Biochemistry 1998, 37 , 9449–9456.9649328 (212) Yates MG ; Lowe DJ Nitrogenase of Azotobacter Chroococcum: A New Electronparamagnetic-Resonance Signal Associated with a Transient Species of the MoFe Protein during Catalysis. FEBS Lett 1976, 72 , 121–126. (213) Christie PD ; Lee H-I ; Cameron LM ; Hales BJ ; Orme-Johnson WH ; Hoffman BM Identification of the CO-Binding Cluster in Nitrogenase MoFe Protein by ENDOR of 57Fe Isotopomers. J. Am. Chem. Soc 1996, 118 , 8707–8709. (214) Lee CC ; Fay AW ; Weng T-C ; Krest CM ; Hedman B ; Hodgson KO ; Hu Y ; Ribbe MW Uncoupling Binding of Substrate CO from Turnover by Vanadium Nitrogenase. Proc. Natl. Acad. Sci. U. S. A 2015, 112 , 13845–13849.26515097 (215) Lee CC ; Wilcoxen J ; Hiller CJ ; Britt RD ; Hu Y Evaluation of the Catalytic Relevance of the CO-Bound States of V-Nitrogenase. Angew. Chemie Int. Ed 2018, 57 , 3411–3414. (216) Hiller CJ ; Lee CC ; Stiebritz MT ; Rettberg LA ; Hu Y Strategies Towards Capturing Nitrogenase Substrates and Intermediates via Controlled Alteration of Electron Fluxes. Chem. Eur. J 2019, 25 , 2389–2395.30225894 (217) Lee HI ; Cameron LM ; Hales BJ ; Hoffman BM CO Binding to the FeMo Cofactor of CO-Inhibited Nitrogenase: 13CO and 1H Q-Band ENDOR Investigation. J. Am. Chem. Soc 1997, 119 , 10121–10126. (218) George SJ ; Ashby GA ; Wharton CW ; Thorneley RNF Time-Resolved Binding of Carbon Monoxide to Nitrogenase Monitored by Stopped-Flow Infrared Spectroscopy. J. Am. Chem. Soc 1997, 119 , 6450–6451. (219) Pollock RC ; Lee H-I ; Cameron LM ; DeRose VJ ; Hales BJ ; Orme-Johnson WH ; Hoffman BM Investigation of CO Bound to Inhibited Forms of Nitrogenase MoFe Protein by 13C ENDOR. J. Am. Chem. Soc 1995, 117 , 8686–8687. (220) Maskos Z ; Hales BJ Photo-Lability of CO Bound to Mo-Nitrogenase from Azotobacter Vinelandii. J. Inorg. Biochem 2003, 93 , 11–17.12538048 (221) Yan L ; Dapper CH ; George SJ ; Wang H ; Mitra D ; Dong W ; Newton WE ; Cramer SP Photolysis of Hi-CO Nitrogenase - Observation of a Plethora of Distinct CO Species Using Infrared Spectroscopy. Eur. J. Inorg. Chem 2011, 2011 , 2064–2074.27630531 (222) Moore VG ; Tittsworth RC ; Hales BJ Construction and Characterization of Hybrid Component 1 from V-Nitrogenase Containing FeMo Cofactor. J. Am. Chem. Soc 1994, 116 , 12101–12102. (223) Hu Y ; Lee CC ; Ribbe MW Extending the Carbon Chain: Hydrocarbon Formation Catalyzed by Vanadium/Molybdenum Nitrogenases. Science 2011, 333 , 753–755.21817053 (224) Lee CC ; Hu Y ; Ribbe MW Vanadium Nitrogenase Reduces CO. Science 2010, 329 , 642.20689010 (225) Lee CC ; Hu Y ; Ribbe MW Unique Features of the Nitrogenase VFe Protein from Azotobacter Vinelandii. Proc. Natl. Acad. Sci. U. S. A 2009, 106 , 9209–9214.19478062 (226) Buscagan TM ; Perez KA ; Maggiolo AO ; Rees DC ; Spatzal T Structural Characterization of Two CO Molecules Bound to the Nitrogenase Active Site. Angew. Chemie Int. Ed 2021, 60 , 5704–5707. (227) Dance I How Does Vanadium Nitrogenase Reduce CO to Hydrocarbons? Dalt. Trans 2011, 40 , 5516–5527. (228) Dance I The Pathway for Serial Proton Supply to the Active Site of Nitrogenase: Enhanced Density Functional Modeling of the Grotthuss Mechanism. Dalt. Trans 2015, 44 , 18167–18186. (229) Dance I Computational Investigations of the Chemical Mechanism of the Enzyme Nitrogenase. ChemBioChem 2020, 21 , 1671–1709.31803989 (230) Cao L ; Caldararu O ; Ryde U Protonation States of Homocitrate and Nearby Residues in Nitrogenase Studied by Computational Methods and Quantum Refinement. J. Phys. Chem. B 2017, 121 , 8242–8262.28783353 (231) Imperial J ; Hoover TR ; Madden MS ; Ludden PW ; Shah VK Substrate Reduction Properties of Dinitrogenase Activated in Vitro Are Dependent upon the Presence of Homocitrate or Its Analogs during Iron-Molybdenum Cofactor Synthesis. Biochemistry 1989, 28 , 7796–7799.2514794 (232) Siegbahn PEM A Major Structural Change of the Homocitrate Ligand of Probable Importance for the Nitrogenase Mechanism. Inorg. Chem 2018, 57 , 1090–1095.29303565 (233) McLean PA ; Smith BE ; Dixon RA Nitrogenase of Klebsiella Pneumoniae NifV Mutants. Investigation of the Novel Carbon Monoxide-Sensitivity of Hydrogen Evolution by the Mutant Enzyme. Biochem. J 1983, 211 , 589–597.6349611 (234) Newton WE ; Vichitphan K ; Fisher K Substrate Reduction and CO Susceptibility of the ΔnifV and α-Q191K MoFe Proteins of Azotobacter Vinelandii Nitrogenase. In 13th International Congress on Nitrogen Fixation; Finan TM , O’Brian MR , Layzell DB , Vassey JK , Newton WE , Eds.; CABI Publishing: Hamilton, Ontario, Canada, 2001; p 370. (235) Liedtke J ; Lee CC ; Tanifuji K ; Jasniewski AJ ; Ribbe MW ; Hu Y Characterization of a Mo-Nitrogenase Variant Containing a Citrate-Substituted Cofactor. ChemBioChem 2021, 22 , 151–155.32918851 (236) Newcomb MP ; Lee CC ; Tanifuji K ; Jasniewski AJ ; Liedtke J ; Ribbe MW ; Hu Y A V-Nitrogenase Variant Containing a Citrate-Substituted Cofactor. ChemBioChem 2020, 21 , 1742–1748.31747483 (237) Mayer SM ; Gormal CA ; Smith BE ; Lawson DM Crystallographic Analysis of the MoFe Protein of Nitrogenase from a NifV Mutant of Klebsiella Pneumoniae Identifies Citrate as a Ligand to the Molybdenum of Iron Molybdenum Cofactor (FeMoco)*. J. Biol. Chem 2002, 277 , 35263–35266.12133839 (238) Seefeldt LC ; Rasche ME ; Ensign SA Carbonyl Sulfide and Carbon Dioxide as New Substrates, and Carbon Disulfide as a New Inhibitor, of Nitrogenase. Biochemistry 1995, 34 , 5382–5389.7727396 (239) Rebelein JG ; Hu Y ; Ribbe MW Differential Reduction of CO2 by Molybdenum and Vanadium Nitrogenases. Angew. Chemie Int. Ed 2014, 53 , 11543–11546. (240) Yang Z-Y ; Moure VR ; Dean DR ; Seefeldt LC Carbon Dioxide Reduction to Methane and Coupling with Acetylene to Form Propylene Catalyzed by Remodeled Nitrogenase. Proc. Natl. Acad. Sci. U. S. A 2012, 109 , 19644–19648.23150564 (241) Zheng Y ; Harris DF ; Yu Z ; Fu Y ; Poudel S ; Ledbetter RN ; Fixen KR ; Yang Z-Y ; Boyd ES ; Lidstrom ME ; A Pathway for Biological Methane Production Using Bacterial Iron-Only Nitrogenase. Nat. Microbiol 2018, 3 , 281–286.29335552 (242) Gaffron H Reduction of Carbon Dioxide with Molecular Hydrogen in Green Algae. Nature 1939, 3614 , 204–205. (243) Kluyver AJ ; Manten A Some Observations on the Metabolism of Bacteria Oxidizing Molecular Hydrogen. Antonie Van Leeuwenhoek 1942, 8 , 71–85. (244) Gest H Oxidation and Evolution of Molecular Hydrogen by Microorganisms. Bacteriol. Rev 1954, 18 , 43–73.13140082 (245) Schwartz E ; Friedrich B The H2-Metabolizing Prokaryotes. In The Prokaryotes; Dworkin M , Falkow S , Rosenberg E , Schleifer K-H , Stackebrandt E , Eds.; Springer: New York, 2006; pp 496–563. (246) Greening C ; Biswas A ; Carere CR ; Jackson CJ ; Taylor MC ; Stott MB ; Cook GM ; Morales SE Genomic and Metagenomic Surveys of Hydrogenase Distribution Indicate H2 Is a Widely Utilised Energy Source for Microbial Growth and Survival. ISME J 2016, 10 , 761–777.26405831 (247) Benoit SL ; Maier RJ ; Sawers RG ; Greening C Molecular Hydrogen Metabolism: A Widespread Trait of Pathogenic Bacteria and Protists. Microbiol. Mol. Biol. Rev 2020, 84 , No. e00092–19.31996394 (248) Das D ; Veziroglu TN Hydrogen Production by Biological Processes: A Survey of Literature. Int. J. Hydrogen Energy 2001, 26 , 13–28. (249) Tamagnini P ; Axelsson R ; Lindberg P ; Oxelfelt F ; Wünschiers R ; Lindblad P Hydrogenases and Hydrogen Metabolism of Cyanobacteria. Microbiol. Mol. Biol. Rev 2002, 66 , 1–20.11875125 (250) Melis A ; Happe T Trails of Green Alga Hydrogen Research - from Hans Gaffron to New Frontiers. Photosynth. Res 2004, 80 , 401–409.16328836 (251) Lubitz W ; Ogata H ; Rüdiger O ; Reijerse E Hydrogenases. Chem. Rev 2014, 114 , 4081–4148.24655035 (252) Vignais PM ; Billoud B Occurrence, Classification, and Biological Function of Hydrogenases: An Overview. Chem. Rev 2007, 107 , 4206–4272.17927159 (253) Shima S ; Pilak O ; Vogt S ; Schick M ; Stagni MS ; Meyer-Klaucke W ; Warkentin E ; Thauer RK ; Ermler U The Crystal Structure of [Fe]-Hydrogenase Reveals the Geometry of the Active Site. Science 2008, 321 , 572–575.18653896 (254) Thauer RK ; Kaster A-K ; Goenrich M ; Schick M ; Hiromoto T ; Shima S Hydrogenases from Methanogenic Archaea, Nickel, a Novel Cofactor, and H2 Storage. Annu. Rev. Biochem 2010, 79 , 507–536.20235826 (255) Huang G ; Wagner T ; Ermler U ; Shima S Methanogenesis Involves Direct Hydride Transfer from H2 to an Organic Substrate. Nat. Rev. Chem 2020, 4 , 213–221. (256) Bothe H ; Schmitz O ; Yates MG ; Newton WE Nitrogen Fixation and Hydrogen Metabolism in Cyanobacteria. Microbiol. Mol. Biol. Rev 2010, 74 , 529–551.21119016 (257) Fritsch J ; Lenz O ; Friedrich B Structure, Function and Biosynthesis of O2-Tolerant Hydrogenases. Nat. Rev. Microbiol 2013, 11 , 106–114.23321533 (258) Parkin A ; Sargent F The Hows and Whys of Aerobic H2 Metabolism. Curr. Opin. Chem. Biol 2012, 16 , 26–34.22366384 (259) Shafaat HS ; Rüdiger O ; Ogata H ; Lubitz W [NiFe] Hydrogenases: A Common Active Site for Hydrogen Metabolism under Diverse Conditions. Biochim. Biophys. Acta 2013, 1827 , 986–1002.23399489 (260) Armstrong FA ; Fontecilla-Camps JC A Natural Choice for Activating Hydrogen. Science 2008, 321 , 498–499.18653870 (261) Horner DS ; Foster PG ; Embley TM Iron Hydrogenases and the Evolution of Anaerobic Eukaryotes. Mol. Biol. Evol 2000, 17 , 1695–1709.11070057 (262) Meyer J [FeFe] Hydrogenases and Their Evolution: A Genomic Perspective. Cell. Mol. Life Sci 2007, 64 , 1063–1084.17353991 (263) Hemschemeier A ; Happe T Alternative Photosynthetic Electron Transport Pathways during Anaerobiosis in the Green Alga Chlamydomonas Reinhardtii. Biochim. Biophys. Acta - Bioenerg 2011, 1807 , 919–926. (264) Melis A ; Zhang L ; Forestier M ; Ghirardi ML ; Seibert M Sustained Photobiological Hydrogen Gas Production upon Reversible Inactivation of Oxygen Evolution in the Green Alga Chlamydomonas Reinhardtii. Plant Physiol 2000, 122 , 127–136.10631256 (265) Buckel W ; Thauer RK Flavin-Based Electron Bifurcation, A New Mechanism of Biological Energy Coupling. Chem. Rev 2018, 118 , 3862–3886.29561602 (266) Schuchmann K ; Chowdhury NP ; Müller V Complex Multimeric [FeFe] Hydrogenases: Biochemistry, Physiology and New Opportunities for the Hydrogen Economy. Front. Microbiol 2018, 9 , 2911.30564206 (267) Land H ; Senger M ; Berggren G ; Stripp ST Current State of [FeFe]-Hydrogenase Research - Biodiversity and Spectroscopic Investigations. ACS Catal 2020, 10 , 7069–7086. (268) Fasano A ; Land H ; Fourmond V ; Berggren G ; Léger C Reversible or Irreversible Catalysis of H+/H2 Conversion by FeFe Hydrogenases. J. Am. Chem. Soc 2021, 143 , 20320–20325.34813699 (269) Søndergaard D ; Pedersen CNS ; Greening C HydDB: A Web Tool for Hydrogenase Classification and Analysis. Sci. Rep 2016, 6 , 34212.27670643 (270) Nishikawa K ; Ogata H ; Higuchi Y Structural Basis of the Function of [NiFe]-Hydrogenases. Chem. Lett 2020, 49 , 164–173. (271) Volbeda A ; Charon M ; Piras C ; Hatchikian EC ; Frey M ; Fontecilla-Camps JC Crystal Structure of the Nickel-Iron Hydrogenase from Desulfovibrio Gigas. Nature 1995, 373 , 580–587.7854413 (272) Higuchi Y ; Yagi T ; Yasuoka N Unusual Ligand Structure in Ni-Fe Active Center and an Additional Mg Site in Hydrogenase Revealed by High Resolution X-Ray Structure Analysis. Structure 1997, 5 , 1671–1680.9438867 (273) Ogata H ; Nishikawa K ; Lubitz W Hydrogens Detected by Subatomic Resolution Protein Crystallography in a [NiFe] Hydrogenase. Nature 2015, 520 , 571–574.25624102 (274) Ogata H ; Mizoguchi Y ; Mizuno N ; Miki K ; Adachi S ; Yasuoka N ; Yagi T ; Yamauchi O ; Hirota S ; Higuchi Y Structural Studies of the Carbon Monoxide Complex of [NiFe] Hydrogenase from Desulfovibrio Vulgaris Miyazaki F: Suggestion for the Initial Activation Site for Dihydrogen. J. Am. Chem. Soc 2002, 124 , 11628–11635.12296727 (275) Ogata H ; Hirota S ; Nakahara A ; Komori H ; Shibata N ; Kato T ; Kano K ; Higuchi Y Activation Process of [NiFe] Hydrogenase Elucidated by High-Resolution X-Ray Analyses: Conversion of the Ready to the Unready State. Structure 2005, 13 , 1635–1642.16271886 (276) Higuchi Y ; Ogata H ; Miki K ; Yasuoka N ; Yagi T Removal of the Bridging Ligand Atom at the Ni-Fe Active Site of [NiFe] Hydrogenase upon Reduction with H2, as Revealed by X-Ray Structure Analysis at 1.4 Å Resolution. Structure 1999, 7 , 549–556.10378274 (277) Volbeda A ; Martin L ; Cavazza C ; Matho M ; Faber BW ; Roseboom W ; Albracht SPJ ; Garcin E ; Rousset M ; Fontecilla-Camps JC Structural Differences between the Ready and Unready Oxidized States of [NiFe] Hydrogenases. J. Biol. Inorg. Chem 2005, 10 , 239–249.15803334 (278) Ogata H ; Kellers P ; Lubitz W The Crystal Structure of the [NiFe] Hydrogenase from the Photosynthetic Bacterium Allochromatium Vinosum: Characterization of the Oxidized Enzyme (Ni-A State). J. Mol. Biol 2010, 402 , 428–444.20673834 (279) Mills DJ ; Vitt S ; Strauss M ; Shima S ; Vonck J De Novo Modeling of the F420-Reducing [NiFe]-Hydrogenase from a Methanogenic Archaeon by Cryo-Electron Microscopy. Elife 2013, 2 , No. e00218.23483797 (280) Vitt S ; Ma K ; Warkentin E ; Moll J ; Pierik AJ ; Shima S ; Ermler U The F-420-Reducing [NiFe]-Hydrogenase Complex from Methanothermobacter Marburgensis, the First X-Ray Structure of a Group 3 Family Member. J. Mol. Biol 2014, 426 , 2813–2826.24887099 (281) Schafer C ; Bommer M ; Hennig SE ; Jeoung JH ; Dobbek H ; Lenz O Structure of an Actinobacterial-Type [NiFe]-Hydrogenase Reveals Insight into O-2-Tolerant H-2 Oxidation. Structure 2016, 24 , 285–292.26749450 (282) Matias PM ; Soares CM ; Saraiva LM ; Coelho R ; Morais J ; Le Gall J ; Carrondo MA [NiFe] Hydrogenase from Desulfovibrio Desulfuricans ATCC 27774: Gene Sequencing, Three-Dimensional Structure Determination and Refinement at 1.8 Å and Modelling Studies of Its Interaction with the Tetrahaem Cytochrome c 3. J. Biol. Inorg. Chem 2001, 6 , 63–81.11191224 (283) Bowman L ; Flanagan L ; Fyfe PK ; Parkin A ; Hunter WN ; Sargent F How the Structure of the Large Subunit Controls Function in an Oxygen-Tolerant [NiFe]-Hydrogenase. Biochem. J 2014, 458 , 449–458.24428762 (284) Shomura Y ; Taketa M ; Nakashima H ; Tai H ; Nakagawa H ; Ikeda Y ; Ishii M ; Igarashi Y ; Nishihara H ; Yoon KS ; Structural Basis of the Redox Switches in the NAD+-Reducing Soluble [NiFe]-Hydrogenase. Science 2017, 357 , 928–932.28860386 (285) Noor NDM ; Matsuura H ; Nishikawa K ; Tai HL ; Hirota S ; Kim J ; Kang JY ; Tateno M ; Yoon KS ; Ogo S ; Redox-Dependent Conformational Changes of a Proximal [4Fe-4S] Cluster in Hyb-Type [NiFe]-Hydrogenase to Protect the Active Site from O2. Chem. Commun 2018, 54 , 12385–12388. (286) Volbeda A ; Amara P ; Darnault C ; Mouesca J-M ; Parkin A ; Roessler MM ; Armstrong FA ; Fontecilla-Camps JC X-Ray Crystallographic and Computational Studies of the O2-Tolerant [NiFe]-Hydrogenase 1 from Escherichia Coli. Proc. Natl. Acad. Sci. U. S. A 2012, 109 , 5305–5310.22431599 (287) Beaton SE ; Evans RM ; Finney AJ ; Lamont CM ; Armstrong FA ; Sargent F ; Carr SB The Structure of Hydrogenase-2 from Escherichia Coli : Implications for H2-Driven Proton Pumping. Biochem. J 2018, 475 , 1353–1370.29555844 (288) Fritsch J ; Scheerer P ; Frielingsdorf S ; Kroschinsky S ; Friedrich B ; Lenz O ; Spahn CMT The Crystal Structure of an Oxygen-Tolerant Hydrogenase Uncovers a Novel Iron-Sulphur Centre. Nature 2011, 479 , 249–252.22002606 (289) Shomura Y ; Yoon KS ; Nishihara H ; Higuchi Y Structural Basis for a [4Fe-3S] Cluster in the Oxygen-Tolerant Membrane-Bound [NiFe]-Hydrogenase. Nature 2011, 479 , 253–256.22002607 (290) Frielingsdorf S ; Fritsch J ; Schmidt A ; Hammer M ; Löwenstein J ; Siebert E ; Pelmenschikov V ; Jaenicke T ; Kalms J ; Rippers Y ; Reversible [4Fe-3S] Cluster Morphing in an O2-Tolerant [NiFe] Hydrogenase. Nat. Chem. Biol 2014, 10 , 378–385.24705592 (291) Happe RP ; Roseboom W ; Pierik AJ ; Albracht SPJ ; Bagley KA Biological Activition of Hydrogen. Nature 1997, 385 , 126.8990114 (292) Pierik AJ ; Roseboom W ; Happe RP ; Bagley KA ; Albracht PJ Carbon Monoxide and Cyanide as Intrinsic Ligands to Iron in the Active Site of [NiFe]-Hydrogenases. J. Biol. Chem 1999, 274 , 3331–3337.9920874 (293) Armstrong FA ; Evans RM ; Hexter SV ; Murphy BJ ; Roessler MM ; Wulff P Guiding Principles of Hydrogenase Catalysis Instigated and Clarified by Protein Film Electrochemistry. Acc. Chem. Res 2016, 49 , 884–892.27104487 (294) Ash PA ; Hidalgo R ; Vincent KA Proton Transfer in the Catalytic Cycle of [NiFe] Hydrogenases: Insight from Vibrational Spectroscopy. ACS Catal 2017, 7 , 2471–2485.28413691 (295) Greene BL ; Vansuch GE ; Chica BC ; Adams MWW ; Dyer RB Applications of Photogating and Time Resolved Spectroscopy to Mechanistic Studies of Hydrogenases. Acc. Chem. Res 2017, 50 , 2718–2726.29083854 (296) Tai H ; Higuchi Y ; Hirota S Comprehensive Reaction Mechanisms at and near the Ni-Fe Active Sites of [NiFe] Hydrogenases. Dalt. Trans 2018, 47 , 4408–4423. (297) Tai H ; Hirota S ; Stripp ST Proton Transfer Mechanisms in Bimetallic Hydrogenases. Acc. Chem. Res 2021, 54 , 232–241.33326230 (298) Ogata H ; Lubitz W ; Higuchi Y Structure and Function of [NiFe] Hydrogenases. J. Biochem 2016, 160 , 251–258.27493211 (299) Baltazar CSA ; Marques MC ; Soares CM ; DeLacey AM ; Pereira IAC ; Matias PM Nickel-Iron-Selenium Hydrogenases - An Overview. Eur. J. Inorg. Chem 2011, 948–962. (300) Garcin E ; Vernede X ; Hatchikian EC ; Volbeda A ; Frey M ; Fontecilla-Camps JC The Crystal Structure of a Reduced [NiFeSe] Hydrogenase Provides an Image of the Activated Catalytic Center. Structure 1999, 7 , 557–566.10378275 (301) Marques MC ; Tapia C ; Gutiérrez-Sanz O ; Ramos AR ; Keller KL ; Wall JD ; De Lacey AL ; Matias PM ; Pereira IAC The Direct Role of Selenocysteine in [NiFeSe] Hydrogenase Maturation and Catalysis. Nat. Chem. Biol 2017, 13 , 544–550.28319099 (302) Ceccaldi P ; Marques MC ; Fourmond V ; Pereira IC ; Léger C Oxidative Inactivation of NiFeSe Hydrogenase. Chem. Commun 2015, 51 , 14223–14226. (303) Volbeda A ; Amara P ; Iannello M ; De Lacey AL ; Cavazza C ; Fontecilla-Camps JC Structural Foundations for the O2 Resistance of Desulfomicrobium Baculatum [NiFeSe]-Hydrogenase. Chem. Commun 2013, 49 , 7061–7063. (304) Yagi T ; Kimura K ; Daidoji H ; Sakai F ; Tamura S ; Properties of Purified Hydrogenase from the Particulate Fraction of Desulfovibrio Vulgaris, Miyazaki. J. Biochem 1976, 79 , 661–671.181372 (305) Yahata N ; Saitoh T ; Takayama Y ; Ozawa K ; Ogata H ; Higuchi Y ; Akutsu H Redox Interaction of Cytochrome C3 with [NiFe] Hydrogenase from Desulfovibrio Vulgaris Miyazaki F. Biochemistry 2006, 45 , 1653–1662.16460012 (306) Yagi T ; Higuchi Y Studies on Hydrogenase. Proc. Japan Acad. Ser. B-Physical Biol. Sci 2013, 89 , 16–33. (307) Volbeda A ; Darnault C ; Parkin A ; Sargent F ; Armstrong FA ; Fontecilla-Camps JC Crystal Structure of the O(2)-Tolerant Membrane-Bound Hydrogenase 1 from Escherichia Coli in Complex with Its Cognate Cytochrome B. Structure 2013, 21 , 184–190.23260654 (308) Tai H ; Hirota S Mechanism and Application of the Catalytic Reaction of [NiFe] Hydrogenase: Recent Developments. Chembiochem 2020, 21 , 1573–1581.32180334 (309) Leroux F ; Dementin S ; Burlat B ; Cournac L ; Volbeda A ; Champ S ; Martin L ; Guigliarelli B ; Bertrand P ; Fontecilla-Camps J ; Experimental Approaches to Kinetics of Gas Diffusion in Hydrogenase. Proc. Natl. Acad. Sci. U. S. A 2008, 105 , 11188–11193.18685111 (310) Liebgott P-P ; de Lacey AL ; Burlat B ; Cournac L ; Richaud P ; Brugna M ; Fernandez VM ; Guigliarelli B ; Rousset M ; Léger C ; Original Design of an Oxygen-Tolerant [NiFe] Hydrogenase: Major Effect of a Valine-to-Cysteine Mutation near the Active Site. J. Am. Chem. Soc 2011, 133 , 986–997.21175174 (311) Abou Hamdan A ; Dementin S ; Liebgott PP ; Gutierrez-Sanz O ; Richaud P ; De Lacey AL ; Roussett M ; Bertrand P ; Cournac L ; Léger C Understanding and Tuning the Catalytic Bias of Hydrogenase. J. Am. Chem. Soc 2012, 134 , 8368–8371.22540997 (312) van Gastel M ; Stein M ; Brecht M ; Schroder O ; Lendzian F ; Bittl R ; Ogata H ; Higuchi Y ; Lubitz W A Single-Crystal ENDOR and Density Functional Theory Study of the Oxidized States of the [NiFe] Hydrogenase from Desulfovibrio Vulgaris Miyazaki F. J. Biol. Inorg. Chem 2006, 11 , 41–51.16292669 (313) Pandelia M-E ; Ogata H ; Lubitz W Intermediates in the Catalytic Cycle of [NiFe] Hydrogenase: Functional Spectroscopy of the Active Site. ChemPhysChem 2010, 11 , 1127–1140.20301175 (314) Fernandez VM ; Hatchikian EC ; Patil DS ; Cammack R ESR-Detectable Nickel and Iron-Sulphur Centres in Relation to the Reversible Activation of Desulfovibrio Gigas Hydrogenase. Biochim. Biophys. Acta - Gen. Subj 1986, 883 , 145–154. (315) Barilone JL ; Ogata H ; Lubitz W ; Van Gastel M Structural Differences between the Active Sites of the Ni-A and Ni-B States of the [NiFe] Hydrogenase: An Approach by Quantum Chemistry and Single Crystal ENDOR Spectroscopy. Phys. Chem. Chem. Phys 2015, 17 , 16204–16212.26035632 (316) Maroney MJ Nailing down Hydrogenase. Nat. Chem. Biol 2013, 9 , 11–12.23238681 (317) Pandelia ME ; Ogata H ; Currell LJ ; Flores M ; Lubitz W Probing Intermediates in the Activation Cycle of [NiFe] Hydrogenase by Infrared Spectroscopy: The Ni-SIr State and Its Light Sensitivity. J. Biol. Inorg. Chem 2009, 14 , 1227–1241.19626348 (318) Fichtner C ; Laurich C ; Bothe E ; Lubitz W Spectroelectrochemical Characterization of the [NiFe] Hydrogenase of Desulfovibrio Vulgaris Miyazaki F. Biochemistry 2006, 45 , 9706–9716.16893172 (319) Bleijlevens B ; van Broekhuizen F. a ; De Lacey AL ; Roseboom W ; Fernandez VM ; Albracht SPJ The Activation of the [NiFe]-Hydrogenase from Allochromatium Vinosum. An Infrared Spectro-Electrochemical Study. J. Biol. Inorg. Chem 2004, 9 , 743–752.15243788 (320) De Lacey AL ; Hatchikian EC ; Volbeda A ; Frey M ; Fontecilla-Camps JC ; Fernandez VM Infrared-Spectroelectrochemical Characterization of the [NiFe] Hydrogenase of Desulfovibrio Gigas. J. Am. Chem. Soc 1997, 119 , 7181–7189. (321) Millo D ; Pandelia M-E ; Utesch T ; Wisitruangsakul N ; Mroginski MA ; Lubitz W ; Hildebrandt P ; Zebger I Spectroelectrochemical Study of the [NiFe] Hydrogenase from Desulfovibrio Vulgaris Miyazaki F in Solution and Immobilized on Biocompatible Gold Surfaces. J. Phys. Chem. B 2009, 113 , 15344–15351.19845323 (322) Breglia R ; Greco C ; Fantucci P ; De Gioia L ; Bruschi M Reactivation of the Ready and Unready Oxidized States of [NiFe]-Hydrogenases: Mechanistic Insights from DFT Calculations. Inorg. Chem 2019, 58 , 279–293.30576127 (323) De Lacey AL ; Pardo A ; Fernández VM ; Dementin S ; Adryanczyk-Perrier G ; Hatchikian CE ; Rousset M FTIR Spectroelectrochemical Study of the Activation and Inactivation Processes of [NiFe] Hydrogenases: Effects of Solvent Isotope Replacement and Site-Directed Mutagenesis. J. Biol. Inorg. Chem 2004, 9 , 636–642.15175937 (324) DeLacey AL ; Fernandez VM ; Rousset M ; Cavazza C ; Hatchikian EC Spectroscopic and Kinetic Characterization of Active Site Mutants of Desulfovibrio Fructosovorans Ni-Fe Hydrogenase. J. Biol. Inorg. Chem 2003, 8 , 129–134.12459907 (325) Evans RM ; Ash PA ; Beaton SE ; Brooke EJ ; Vincent KA ; Carr SB ; Armstrong FA Mechanistic Exploitation of a Self-Repairing, Blocked Proton Transfer Pathway in an O2-Tolerant [NiFe]-Hydrogenase. J. Am. Chem. Soc 2018, 140 , 10208–10220.30070475 (326) Ogata H ; Kramer T ; Wang H ; Schilter D ; Pelmenschikov V ; van Gastel M ; Neese F ; Rauchfuss TB ; Gee LB ; Scott AD ; Hydride Bridge in [NiFe]-Hydrogenase Observed by Nuclear Resonance Vibrational Spectroscopy. Nat. Commun 2015, 6 , 7890.26259066 (327) Wang HX ; Yoda Y ; Ogata H ; Tanaka Y ; Lubitz W A Strenuous Experimental Journey Searching for Spectroscopic Evidence of a Bridging Nickel-Iron-Hydride in [NiFe] Hydrogenase. J. Synchrotron Radiat 2015, 22 , 1334–1344.26524296 (328) Kamali S ; Wang HX ; Mitra D ; Ogata H ; Lubitz W ; Manor BC ; Rauchfuss TB ; Byrne D ; Bonnefoy V ; Jenney FE ; Observation of the Fe-CN and Fe-CO Vibrations in the Active Site of [NiFe] Hydrogenase by Nuclear Resonance Vibrational Spectroscopy. Angew. Chem. Int. Ed 2013, 52 , 724–728. (329) Ogo S ; Ichikawa K ; Kishima T ; Matsumoto T ; Nakai H ; Kusaka K ; Ohhara T A Functional [NiFe]-hydrogenase Mimic That Catalyzes Electron and Hydride Transfer from H2. Science 2013, 339 , 682–684.23393260 (330) Manor BC ; Rauchfuss TB Hydrogen Activation by Biomimetic [NiFe]-Hydrogenase Model Containing Protected Cyanide Cofactors. J. Am. Chem. Soc 2013, 135 , 11895–11900.23899049 (331) Chambers GM ; Huynh MT ; Li Y ; Hammes-Schiffer S ; Rauchfuss TB ; Reijerse E ; Lubitz W Models of the Ni-L and Ni-SIa States of the [NiFe]-Hydrogenase Active Site. Inorg. Chem 2016, 55 , 419–431.26421729 (332) Ulloa OA ; Huynh MT ; Richers CP ; Bertke JA ; Nilges MJ ; Hammes-Schiffer S ; Rauchfuss TB Mechanism of H2 Production by Models for the [NiFe]-Hydrogenases: Role of Reduced Hydrides. J. Am. Chem. Soc 2016, 138 , 9234–9245.27328053 (333) Dong G ; Phung QM ; Hallaert SD ; Pierloot K ; Ryde U H2 Binding to the Active Site of [NiFe] Hydrogenase Studied by Multiconfigurational and Coupled-Cluster Methods. Phys. Chem. Chem. Phys 2017, 19 , 10590–10601.28397891 (334) Dong G ; Ryde U ; Aa Jensen HJ ; Hedegard ED Exploration of H2 Binding to the [NiFe]-Hydrogenase Active Site with Multiconfigurational Density Functional Theory. Phys. Chem. Chem. Phys 2018, 20 , 794–801.29205241 (335) Bruschi M ; Tiberti M ; Guerra A ; De Gioia L Disclosure of Key Stereoelectronic Factors for Efficient H2 Binding and Cleavage in the Active Site of [NiFe]-Hydrogenases. J. Am. Chem. Soc 2014, 136 , 1803–1814.24392667 (336) Brazzolotto D ; Gennari M ; Queyriaux N ; Simmons TR ; Pécaut J ; Demeshko S ; Meyer F ; Orio M ; Artero V ; Duboc C Nickel-Centred Proton Reduction Catalysis in a Model of [NiFe] Hydrogenase. Nat. Chem 2016, 8 , 1054–1060.27768098 (337) Dong G ; Phung QM ; Pierloot K ; Ryde U Reaction Mechanism of [NiFe] Hydrogenase Studied by Computational Methods. Inorg. Chem 2018, 57 , 15289–15298.30500163 (338) Pandelia ME ; Ogata H ; Currell LJ ; Flores M ; Lubitz W Inhibition of the [NiFe] Hydrogenase from Desulfovibrio Vulgaris Miyazaki F by Carbon Monoxide: An FTIR and EPR Spectroscopic Study. Biochim. Biophys. Acta - Bioenerg 2010, 1797 , 304–313. (339) Ilina Y ; Lorent C ; Katz S ; Jeoung J ; Shima S ; Horch M ; Zebger I ; Dobbek H X-ray Crystallography and Vibrational Spectroscopy Reveal the Key Determinants of Biocatalytic Dihydrogen Cycling by [NiFe] Hydrogenases. Angew. Chem. Int. Ed 2019, 58 , 18710–18714. (340) Horch M ; Schoknecht J ; Mroginski MA ; Lenz O ; Hildebrandt P ; Zebger I Resonance Raman Spectroscopy on [NiFe] Hydrogenase Provides Structural Insights into Catalytic Intermediates and Reactions. J. Am. Chem. Soc 2014, 136 , 9870–9873.24956459 (341) Stein M ; van Lenthe E ; Baerends EJ ; Lubitz W Relativistic DFT Calculations of the Paramagnetic Intermediates of [NiFe] Hydrogenase. Implications for the Enzymatic Mechanism. J. Am. Chem. Soc 2001, 123 , 5839–5840.11403633 (342) Brecht M ; Van Gastel M ; Buhrke T ; Friedrich B ; Lubitz W Direct Detection of a Hydrogen Ligand in the [NiFe] Center of the Regulatory H2-Sensing Hydrogenase from Ralstonia Eutropha in Its Reduced State by HYSCORE and ENDOR Spectroscopy. J. Am. Chem. Soc 2003, 125 , 13075–13083.14570480 (343) Foerster S ; Stein M ; Brecht M ; Ogata H ; Higuchi Y ; Lubitz W Single Crystal EPR Studies of the Reduced Active Site of [NiFe] Hydrogenase from Desulfovibrio Vulgaris Miyazaki F. J. Am. Chem. Soc 2003, 125 , 83–93.12515509 (344) Hidalgo R ; Ash PA ; Healy AJ ; Vincent KA Infrared Spectroscopy during Electrocatalytic Turnover Reveals the Ni-L Active Site State during H2 Oxidation by a NiFe Hydrogenase. Angew. Chem. Int. Ed 2015, 54 , 7110–7113. (345) Murphy BJ ; Hidalgo R ; Roessler MM ; Evans RM ; Ash PA ; Myers WK ; Vincent KA ; Armstrong FA Discovery of Dark PH-Dependent H+ Migration in a [NiFe]-Hydrogenase and Its Mechanistic Relevance: Mobilizing the Hydrido Ligand of the Ni-C Intermediate. J. Am. Chem. Soc 2015, 137 , 8484–8489.26103582 (346) Tai H ; Nishikawa K ; Higuchi Y ; Mao Z-W ; Hirota S Cysteine SH and Glutamate COOH Contributions to [NiFe] Hydrogenase Proton Transfer Revealed by Highly Sensitive FTIR Spectroscopy. Angew. Chem. Int. Ed 2019, 58 , 13285–13290. (347) Tai H ; Nishikawa K ; Suzuki M ; Higuchi Y ; Hirota S Control of the Transition between Ni-C and Ni-SIastates by the Redox State of the Proximal Fe-S Cluster in the Catalytic Cycle of [NiFe] Hydrogenase. Angew. Chem. Int. Ed 2014, 53 , 13817–13820. (348) Tai H ; Nishikawa K ; Inoue S ; Higuchi Y ; Hirota S FT-IR Characterization of the Light-Induced Ni-L2 and Ni-L3 States of [NiFe] Hydrogenase from Desulfovibrio Vulgaris Miyazaki F. J. Phys. Chem. B 2015, 119 , 13688–13674. (349) Greene BL ; Wu CH ; McTernan PM ; Adams MWW ; Dyer RB Proton-Coupled Electron Transfer Dynamics in the Catalytic Mechanism of a [NiFe]-Hydrogenase. J. Am. Chem. Soc 2015, 137 , 4558–4566.25790178 (350) Greene BL ; Wu C ; Vansuch GE ; Adams MWW ; Dyer RB Proton Inventory and Dynamics in the Nia-S to Nia-C Transition of a [NiFe] Hydrogenase. Biochemistry 2016, 55 , 1813–1825.26956769 (351) Greene BL ; Vansuch GE ; Wu CH ; Adams MWW ; Dyer RB Glutamate Gated Proton-Coupled Electron Transfer Activity of a [NiFe]-Hydrogenase. J. Am. Chem. Soc 2016, 138 , 13013–13021.27617712 (352) Kampa M ; Pandelia ME ; Lubitz W ; Van Gastel M ; Neese F A Metal-Metal Bond in the Light-Induced State of [Nife] Hydrogenases with Relevance to Hydrogen Evolution. J. Am. Chem. Soc 2013, 135 , 3915–3925.23402569 (353) Siebert E ; Horch M ; Rippers Y ; Fritsch J ; Frielingsdorf S ; Lenz O ; Velazquez Escobar F ; Siebert F ; Paasche L ; Kuhlmann U ; Resonance Raman Spectroscopy as a Tool to Monitor the Active Site of Hydrogenases. Angew. Chem. Int. Ed 2013, 52 , 5162–5165. (354) Ogata H ; Lubitz W ; Higuchi Y [NiFe] Hydrogenases: Structural and Spectroscopic Studies of the Reaction Mechanism. Dalton Trans 2009, 9226 , 7577–7587. (355) Teixeira VH ; Soares CM ; Baptista AM Proton Pathways in a [NiFe]-Hydrogenase: A Theoretical Study. Proteins 2008, 70 , 1010–1022.17847093 (356) Galván IF ; Volbeda A ; Fontecilla-Camps JC ; Field MJ A QM/MM Study of Proton Transport Pathways in a [NiFe] Hydrogenase. Proteins Struct. Funct. Genet 2008, 73 , 195–203.18412257 (357) Dementin S ; Burlat B ; De Lacey AL ; Pardo A ; Adryanczyk-Perrier G ; Guigliarelli B ; Fernandez VM ; Rousset M A Glutamate Is the Essential Proton Transfer Gate during the Catalytic Cycle of the [NiFe] Hydrogenase. J. Biol. Chem 2004, 279 , 10508–10513.14688251 (358) Oteri F ; Baaden M ; Lojou E ; Sacquin-Mora S Multiscale Simulations Give Insight into the Hydrogen in and out Pathways of [NiFe]-Hydrogenases from Aquifex Aeolicus and Desulfovibrio Fructosovorans. J. Phys. Chem. B 2014, 118 , 13800–13811.25399809 (359) Dong G ; Ryde U Protonation States of Intermediates in the Reaction Mechanism of [NiFe] Hydrogenase Studied by Computational Methods. J. Biol. Inorg. Chem 2016, 21 , 383–394.26940957 (360) Evans RM ; Brooke EJ ; Wehlin SAM ; Nomerotskaia E ; Sargent F ; Carr SB ; Phillips SEV ; Armstrong FA Mechanism of Hydrogen Activation by [NiFe] Hydrogenases. Nat. Chem. Biol 2016, 12 , 46–50.26619250 (361) Brooke EJ ; Evans RM ; Islam STA ; Roberts GM ; Wehlin SAM ; Carr SB ; Phillips SEV ; Armstrong FA Importance of the Active Site “Canopy” Residues in an O2-Tolerant [NiFe]-Hydrogenase. Biochemistry 2017, 56 , 132–142.28001048 (362) Vansuch GE ; Wu C-H ; Haja DK ; Blair SA ; Chica B ; Johnson MK ; Adams MWW ; Dyer RB Metal-Ligand Cooperativity in the Soluble Hydrogenase-1 from Pyrococcus Furiosus. Chem. Sci 2020, 11 , 8572–8581.34123117 (363) Fichtner C ; van Gastel M ; Lubitz W Wavelength Dependence of the Photo-Induced Conversion of the Ni-C to the Ni-L Redox State in the [NiFe] Hydrogenase of Desulfovibrio Vulgaris Miyazaki F. Phys. Chem. Chem. Phys 2003, 5 , 5507–5513. (364) van der Zwaan JW ; Albracht SPJ ; Fontijn RD ; Slater EC Monovalent Nickel in Hydrogenase from Chromatium Vinosum. Light Sensitivity and Evidence for Direct Interaction with Hydrogen. FEBS Lett 1985, 179 , 271–277.2981705 (365) Medina M ; Williams R ; Cammack R ; Hatchikian EC Studies of Light-Induced Nickel EPR Signals in Desulfovibrio Gigas Hydrogenase. J. Chem. Soc. Faraday Trans 1994, 90 , 2921–2924. (366) Medina M ; Hatchikian EC ; Cammack R Studies of Light-Induced Nickel EPR Signals in Hydrogenase: Comparison of Enzymes with and without Selenium. Biochim. Biophys. Acta - Bioenerg 1996, 1275 , 227–236. (367) Bondar AN ; Elstner M ; Suhai S ; Smith JC ; Fischer S Mechanism of Primary Proton Transfer in Bacteriorhodopsin. Structure 2004, 12 , 1281–1288.15242604 (368) Abou-Hamdan A ; Ceccaldi P ; Lebrette H ; Gutierrez-Sanz O ; Richaud P ; Cournac L ; Guigliarelli B ; de Lacey AL ; Léger C ; Volbeda A ; A Threonine Stabilizes the NiC and NiR Catalytic Intermediates of [NiFe]-Hydrogenase. J. Biol. Chem 2015, 290 , 8550–8558.25666617 (369) Roberts LM ; Lindahl PA Analysis of Oxidative Titrations of Desulfovibrio Gigas Hydrogenase; Implications for the Catalytic Mechanism. Biochemistry 1994, 33 , 14339–14350.7947844 (370) Radu V ; Frielingsdorf S ; Lenz O ; Jeuken LJC Reactivation from the Ni-B State in [NiFe] Hydrogenase of Ralstonia Eutropha Is Controlled by Reduction of the Superoxidised Proximal Cluster. Chem. Commun 2016, 52 , 2632–2635. (371) Evans RM ; Parkin A ; Roessler MM ; Murphy BJ ; Adamson H ; Lukey MJ ; Sargent F ; Volbeda A ; Fontecilla-Camps JC ; Armstrong FA a. Principles of Sustained Enzymatic Hydrogen Oxidation in the Presence of Oxygen - The Crucial Influence of High Potential Fe-S Clusters in the Electron Relay of [NiFe]-Hydrogenases. J. Am. Chem. Soc 2013, 135 , 2694–2707.23398301 (372) Pandelia ME ; Fourmond V ; Tron-Infossi P ; Lojou E ; Bertrand P ; Léger C ; Giudici-Orticoni MT ; Lubitz W Membrane-Bound Hydrogenase I from the Hyperthermophilic Bacterium Aquifex Aeolicus: Enzyme Activation, Redox Intermediates and Oxygen Tolerance. J. Am. Chem. Soc 2010, 132 , 6991–7004.20441192 (373) Lubitz W ; Reijerse E ; van Gastel M [NiFe] and [FeFe] Hydrogenases Studied by Advanced Magnetic Resonance Techniques. Chem. Rev 2007, 107 , 4331–4365.17845059 (374) Pandelia ME ; Nitschke W ; Infossi P ; Giudici-Orticoni MT ; Bill E ; Lubitz W Characterization of a Unique [FeS] Cluster in the Electron Transfer Chain of the Oxygen Tolerant [NiFe] Hydrogenase from Aquifex Aeolicus. Proc. Natl. Acad. Sci. U. S. A 2011, 108 , 6097–6102.21444783 (375) Pandelia ME ; Bykov D ; Izsak R ; Infossi P ; Giudici-Orticoni MT ; Bill E ; Neese F ; Lubitz W Electronic Structure of the Unique [4Fe-3S] Cluster in O 2-Tolerant Hydrogenases Characterized by 57Fe Mössbauer and EPR Spectroscopy. Proc. Natl. Acad. Sci. U. S. A 2013, 110 , 483–488.23267108 (376) Mouesca J-M ; Amara P ; Fontecilla-Camps JC Electronic States of the O2-Tolerant [NiFe] Hydrogenase Proximal Cluster. Proc. Natl. Acad. Sci. U. S. A 2013, 110 , No. E2538.23716706 (377) Pandelia M-E ; Bykov D ; Izsak R ; Infossi P ; Giudici-Orticoni M-T ; Bill E ; Neese F ; Lubitz W Reply to Mouesca : Electronic Structure of the Proximal [4Fe-3S] Cluster of O2-Tolerant [NiFe] Hydrogenases. Proc. Natl. Acad. Sci. U. S. A 2013, 110 , No. E2539.24003456 (378) Mouesca JM ; Fontecilla-Camps JC ; Amara P The Structural Plasticity of the Proximal [4Fe3S] Cluster Is Responsible for the O2 Tolerance of Membrane-Bound [NiFe] Hydrogenases. Angew. Chem. Int. Ed 2013, 52 , 2002–2006. (379) Goris T ; Wait AF ; Saggu M ; Fritsch J ; Heidary N ; Stein M ; Zebger I ; Lendzian F ; Armstrong FA ; Friedrich B ; A Unique Iron-Sulfur Cluster Is Crucial for Oxygen Tolerance of a [NiFe]-Hydrogenase. Nat. Chem. Biol 2011, 7 , 310–318.21390036 (380) Lukey MJ ; Roessler MM ; Parkin A ; Evans RM ; Davies RA ; Lenz O ; Friedrich B ; Sargent F ; Armstrong FA Oxygen-Tolerant [NiFe]-Hydrogenases: The Individual and Collective Importance of Supernumerary Cysteines at the Proximal Fe-S Cluster. J. Am. Chem. Soc 2011, 133 , 16881–16892.21916508 (381) Lauterbach L ; Lenz O Catalytic Production of Hydrogen Peroxide and Water by Oxygen-Tolerant [NiFe]-Hydrogenase during H-2 Cycling in the Presence of O-2. J. Am. Chem. Soc 2013, 135 , 17897–17905.24180286 (382) Karstens K ; Wahlefeld S ; Horch M ; Grunzel M ; Lauterbach L ; Lendzian F ; Zebger I ; Lenz O Impact of the Iron-Sulfur Cluster Proximal to the Active Site on the Catalytic Function of an O-2-Tolerant NAD(+)-Reducing [NiFe]-Hydrogenase. Biochemistry 2015, 54 , 389–403.25517969 (383) Dance I What Is the Trigger Mechanism for the Reversal of Electron Flow in Oxygen-Tolerant [NiFe] Hydrogenases? Chem. Sci 2015, 6 , 1433–1443.29560232 (384) Roessler MM ; Evans RM ; Davies RA ; Harmer J ; Armstrong FA EPR Spectroscopic Studies of the Fe-S Clusters in the O-2-Tolerant [NiFe]-Hydrogenase Hyd-1 from Escherichia Coli and Characterization of the Unique [4Fe-3S] Cluster by HYSCORE. J. Am. Chem. Soc 2012, 134 , 15581–15594.22900997 (385) Tai H ; Xu L ; Inoue S ; Nishikawa K ; Higuchi Y ; Hirota S Photoactivation of the Ni-SIr State to the Ni-SIa State in [NiFe] Hydrogenase: FT-IR Study on the Light Reactivity of the Ready Ni-SIr State and as-Isolated Enzyme Revisited. Phys. Chem. Chem. Phys 2016, 18 , 22025–22030.27456760 (386) Tai H ; Xu L ; Nishikawa K ; Higuchi Y ; Hirota S Equilibrium between Inactive Ready Ni-SIr and Active Ni-SIa States of [NiFe] Hydrogenase Studied by Utilizing Ni-SIr-to-Ni-SIa Photoactivation. Chem. Commun 2017, 53 , 10444–10447. (387) Rousset M ; Montet Y ; Guigliarelli B ; Forget N ; Asso M ; Bertrand P ; Fontecilla-Camps JC ; Hatchikian EC [3Fe-4S] to [4Fe-4S] Cluster Conversion in Desulfovibrio Fructosovorans [NiFe] Hydrogenase by Site-Directed Mutagenesis. Proc. Natl. Acad. Sci. U. S. A 1998, 95 , 11625–11630.9751716 (388) Dementin S ; Belle V ; Bertrand P ; Guigliarelli B ; Adryanczyk-Perrier G ; De Lacey AL ; Fernandez VM ; Rousset M ; Léger C Changing the Ligation of the Distal [4Fe4S] Cluster in NiFe Hydrogenase Impairs Inter- and Intramolecular Electron Transfers. J. Am. Chem. Soc 2006, 128 , 5209–5218.16608357 (389) Dementin S ; Burlat B ; Fourmond V ; Leroux F ; Liebgott P-P ; Abou Hamdan A ; Léger C ; Rousset M ; Guigliarelli B ; Bertrand P Rates of Intra- and Intermolecular Electron Transfers in Hydrogenase Deduced from Steady-State Activity Measurements. J. Am. Chem. Soc 2011, 133 , 10211–10221.21615141 (390) Petrenko A ; Stein M Distal [FeS]-Cluster Coordination in [NiFe]-Hydrogenase Facilitates Intermolecular Electron Transfer. Int. J. Mol. Sci 2017, 18 , 100. (391) Yonemoto IT ; Smith HO ; Weyman PD Designed Surface Residue Substitutions in [NiFe] Hydrogenase That Improve Electron Transfer Characteristics. Int. J. Mol. Sci 2015, 16 , 2020–2033.25603181 (392) Yonemoto IT ; Clarkson BR ; Smith HO ; Weyman PD A Broad Survey Reveals Substitution Tolerance of Residues Ligating FeS Clusters in [NiFe] Hydrogenase. BMC Biochem 2014, 15 , 10.24934472 (393) Yonemoto IT ; Matteri CW ; Nguyen TA ; Smith HO ; Weyman PD Dual Organism Design Cycle Reveals Small Subunit Substitutions That Improve [NiFe] Hydrogenase Hydrogen Evolution. J. Biol. Eng 2013, 7 , 17.23819621 (394) Adamson H ; Robinson M ; Wright JJ ; Flanagan LA ; Walton J ; Elton D ; Gavaghan DJ ; Bond AM ; Roessler MM ; Parkin A Retuning the Catalytic Bias and Overpotential of a [NiFe]-Hydrogenase via a Single Amino Acid Exchange at the Electron Entry/Exit Site. J. Am. Chem. Soc 2017, 139 , 10677–10686.28697596 (395) Vidal-Limon AM ; Tafoya P ; Santini BL ; Contreras OE ; Aguila SA Electron Transfer Pathways Analysis of Oxygen Tolerant [NiFe]-Hydrogenases for Hydrogen Production: A Quantum Mechanics/Molecular Mechanics - Statistical Coupled Analysis. Int. J. Hydrog. Energy 2017, 42 , 20494–20502. (396) Montet Y ; Amara P ; Volbeda A ; Gas Access to the Active Site of Ni-Fe Hydrogenases Probed by X-Ray Crystallography and Molecular Dynamics. Nat. Struct. Biol 1997, 4 , 523–526.9228943 (397) Volbeda A High-Resolution Crystallographic Analysis of Desulfovibrio Fructosovorans [NiFe] Hydrogenase. Int. J. Hydrogen Energy 2002, 27 , 1449–1461. (398) Topin J ; Rousset M ; Antonczak S ; Golebiowski J Kinetics and Thermodynamics of Gas Diffusion in a NiFe Hydrogenase. Proteins-Structure Funct. Bioinforma 2012, 80 , 677–682. (399) Kalms J ; Schmidt A ; Frielingsdorf S ; van der Linden P ; von Stetten D ; Lenz O ; Carpentier P ; Scheerer P Krypton Derivatization of an O2 -Tolerant Membrane-Bound [NiFe] Hydrogenase Reveals a Hydrophobic Tunnel Network for Gas Transport. Angew. Chem. Int. Ed 2016, 55 , 5586–5590. (400) Kalms J ; Schmidt A ; Frielingsdorf S ; Utesch T ; Gotthard G ; von Stetten D ; van der Linden P ; Royant A ; Mroginski MA ; Carpentier P ; Tracking the Route of Molecular Oxygen in O2-Tolerant Membrane-Bound [NiFe] Hydrogenase. Proc. Natl. Acad. Sci. U.S.A 2018, 115 , E2229–E2237.29463722 (401) Wang PH ; Best RB ; Blumberger J Multiscale Simulation Reveals Multiple Pathways for H2 and O2 Transport in a [NiFe]-Hydrogenase. J. Am. Chem. Soc 2011, 133 , 3548–3556.21341658 (402) Buhrke T ; Lenz O ; Krauss N ; Friedrich B Oxygen Tolerance of the H2-Sensing [NiFe] Hydrogenase from Ralstonia Eutropha H16 Is Based on Limited Access of Oxygen to the Active Site. J. Biol. Chem 2005, 280 , 23791–23796.15849358 (403) Duché O ; Elsen S ; Cournac L ; Colbeau A Enlarging the Gas Access Channel to the Active Site Renders the Regulatory Hydrogenase HupUV of Rhodobacter Capsulatus O2 Sensitive without Affecting Its Transductory Activity. FEBS J 2005, 272 , 3899–3908.16045760 (404) Abou Hamdan A ; Liebgott P-P ; Fourmond V ; Gutierrez-Sanz O ; De Lacey AL ; Infossi P ; Rousset M ; Dementin S ; Léger C Relation between Anaerobic Inactivation and Oxygen Tolerance in a Large Series of NiFe Hydrogenase Mutants. Proc. Natl. Acad. Sci. U. S. A 2012, 109 , 19916–19921.23169623 (405) Liebgott P ; Leroux F ; Burlat B ; Dementin S ; Relating Diffusion along the Substrate Tunnel and Oxygen Sensitivity in Hydrogenase. Nat. Chem. Biol 2010, 6 , 63–70.19966788 (406) Dementin S ; Leroux F ; Cournac L ; de Lacey AL ; Volbeda A ; Léger C ; Burlat B ; Martinez N ; Champ S ; Martin L ; Introduction of Methionines in the Gas Channel Makes [NiFe] Hydrogenase Aero-Tolerant. J. Am. Chem. Soc 2009, 131 , 10156–10164.19580279 (407) Wang PH ; Blumberger J Mechanistic Insight into the Blocking of CO Diffusion in [NiFe]-Hydrogenase Mutants through Multiscale Simulation. Proc. Natl. Acad. Sci. U. S. A 2012, 109 , 6399–6404.22493222 (408) Liebgott PP ; Leroux F ; Burlat B ; Dementin S ; Baffert C ; Lautier T ; Fourmond V ; Ceccaldi P ; Cavazza C ; Meynial-Salles I ; Relating Diffusion along the Substrate Tunnel and Oxygen Sensitivity in Hydrogenase. Nat. Chem. Biol 2010, 6 , 63–70.19966788 (409) Fourmond V ; Infossi P ; Giudici-Orticoni M-T ; Bertrand P ; Léger C Two-Step” Chronoamperometric Method for Studying the Anaerobic Inactivation of an Oxygen Tolerant NiFe Hydrogenase. J. Am. Chem. Soc 2010, 132 , 4848–4857.20230028 (410) del Barrio M ; Guendon C ; Kpebe A ; Baffert C ; Fourmond V ; Brugna M ; Léger C Valine-to-Cysteine Mutation Further Increases the Oxygen Tolerance of Escherichia Coli NiFe Hydrogenase Hyd-1. ACS Catal 2019, 9 , 4084–4088. (411) Marques MC ; Coelho R ; Pereira IAC ; Matias PM Redox State-Dependent Changes in the Crystal Structure of [NiFeSe] Hydrogenase from Desulfovibrio Vulgaris Hildenborough. Int. J. Hydrog. Energy 2013, 38 , 8664–8682. (412) Zacarias S ; Temporao A ; del Barrio M ; Fourmond V ; Léger C ; Matias PM ; Pereira IAC A Hydrophilic Channel Is Involved in Oxidative Inactivation of a [NiFeSe] Hydrogenase. ACS Catal 2019, 9 , 8509–8519. (413) Peters JW ; Lanzilotta WN ; Lemon BJ ; Seefeldt LC X-Ray Crystal Structure of the Fe-Only Hydrogenase (CpI) from Clostridium Pasteurianum to 1.8 Angstrom Resolution. Science 1998, 282 , 1853–1858.9836629 (414) Lemon BJ ; Peters JW Binding of Exogenously Added Carbon Monoxide at the Active Site of the Iron-Only Hydrogenase (CpI) from Clostridium Pasteurianum. Biochemistry 1999, 38 , 12969–12973.10529166 (415) Lemon BJ ; Peters JW Photochemistry at the Active Site of the Carbon Monoxide Inhibited Form of the Iron-Only Hydrogenase (CpI). J. Am. Chem. Soc 2000, 122 , 3793–3794. (416) Gorwa MF ; Croux C ; Soucaille P Molecular Characterization and Transcriptional Analysis of the Putative Hydrogenase Gene of Clostridium Acetobutylicum ATCC 824. J. Bacteriol 1996, 178 , 2668–2675.8626337 (417) Gauquelin C ; Baffert C ; Richaud P ; Kamionka E ; Etienne E ; Guieysse D ; Girbal L ; Fourmond V ; André I ; Guigliarelli B ; Roles of the F-Domain in [FeFe] Hydrogenase. Biochim. Biophys. Acta - Bioenerg 2018, 1859 , 69–77.28842179 (418) Artz JH ; Mulder DW ; Ratzloff MW ; Lubner CE ; Zadvornyy OA ; LeVan AX ; Williams SG ; Adams MWW ; Jones AK ; King PW ; Reduction Potentials of [FeFe]-Hydrogenase Accessory Iron-Sulfur Clusters Provide Insights into the Energetics of Proton Reduction Catalysis. J. Am. Chem. Soc 2017, 139 , 9544–9550.28635269 (419) Pierik AJ ; Hagen WR ; Redeker JS ; Wolbert RBG ; Boersma M ; Verhagen MFJM ; Grande HJ ; Veeger C ; Mutsaers PH ; Sands RH ; Redox Properties of the Iron-Sulfur Clusters in Activated Fe-Hydrogenase from Desulfovibrio Vulgaris (Hildenborough). Eur. J. Biochem 1992, 209 , 63–72.1396719 (420) Nicolet Y ; Piras C ; Legrand P ; Hatchikian CE ; Fontecilla-Camps JC Desulfovibrio Desulfuricans Iron Hydrogenase: The Structure Shows Unusual Coordination to an Active Site Fe Binuclear Center. Structure 1999, 7 , 13–23.10368269 (421) Nicolet Y ; de Lacey AL ; Vernède X ; Fernandez VM ; Hatchikian EC ; Fontecilla-Camps JC Crystallographic and FTIR Spectroscopic Evidence of Changes in Fe Coordination Upon Reduction of the Active Site of the Fe-Only Hydrogenase from Desulfovibrio d Esulfuricans. J. Am. Chem. Soc 2001, 123 , 1596–1601.11456758 (422) Caserta G ; Adamska-Venkatesh A ; Pecqueur L ; Atta M ; Artero V ; Roy S ; Reijerse E ; Lubitz W ; Fontecave M Chemical Assembly of Multiple Metal Cofactors: The Heterologously Expressed Multidomain [FeFe]-Hydrogenase from Megasphaera Elsdenii. Biochim. Biophys. Acta - Bioenerg 2016, 1857 , 1734–1740. (423) Caserta G ; Papini C ; Adamska-Venkatesh A ; Pecqueur L ; Sommer C ; Reijerse E ; Lubitz W ; Gauquelin C ; Meynial-Salles I ; Pramanik D ; Engineering an [FeFe]-Hydrogenase: Do Accessory Clusters Influence O2 Resistance and Catalytic Bias? J. Am. Chem. Soc 2018, 140 , 5516–5526.29595965 (424) Winkler M ; Duan J ; Rutz A ; Felbek C ; Scholtysek L ; Lampret O ; Jaenecke J ; Apfel U-P ; Gilardi G ; Valetti F ; A Safety Cap Protects Hydrogenase from Oxygen Attack. Nat. Commun 2021, 12 , 756.33531463 (425) Happe T ; Naber JD Isolation, Characterization and N-terminal Amino Acid Sequence of Hydrogenase from the Green Alga Chlamydomonas Reinhardtii. Eur. J. Biochem 1993, 214 , 475–481.8513797 (426) Winkler M ; Hemschemeier A ; Gotor C ; Melis A ; Happe T [FeFe]-Hydrogenases in Green Algae: Photo-Fermentation and Hydrogen Evolution under Sulfur Deprivation. Int. J. Hydrogen Energy 2002, 27 , 1431–1439. (427) Kamp C ; Silakov A ; Winkler M ; Reijerse EJ ; Lubitz W ; Happe T Isolation and First EPR Characterization of the [FeFe]-Hydrogenases from Green Algae. Biochim. Biophys. Acta 2008, 1777 , 410–416.18355437 (428) Sybirna K ; Antoine T ; Lindberg P ; Fourmond V ; Rousset M ; Méjean V ; Bottin H Shewanella Oneidensis: A New and Efficient System for Expression and Maturation of Heterologous [Fe-Fe] Hydrogenase from Chlamydomonas Reinhardtii. BMC Biotechnol 2008, 8 , 73.18801156 (429) Stripp ST ; Sanganas O ; Happe T ; Haumann M The Structure of the Active Site H-Cluster of [FeFe] Hydrogenase from the Green Alga Chlamydomonas Reinhardtii Studied by X-Ray Absorption Spectroscopy. Biochemistry 2009, 48 , 5042–5049.19397274 (430) Mulder DW ; Boyd ES ; Sarma R ; Lange RK ; Endrizzi JA ; Broderick JB ; Peters JW Stepwise [FeFe]-Hydrogenase H-Cluster Assembly Revealed in the Structure of HydAΔEFG. Nature 2010, 465 , 248–251.20418861 (431) Schuchmann K ; Müller V Direct and Reversible Hydrogenation of CO2 to Formate by a Bacterial Carbon Dioxide Reductase. Science 2013, 342 , 1382–1386.24337298 (432) Ceccaldi P ; Schuchmann K ; Müller V ; Elliott SJ The Hydrogen Dependent CO2 Reductase: The First Completely Co Tolerant FeFe-Hydrogenase. Energy Environ. Sci 2017, 10 , 503–508. (433) Lipscomb GL ; Schut GJ ; Thorgersen MP ; Nixon WJ ; Kelly RM ; Adams MWW Engineering Hydrogen Gas Production from Formate in a Hyperthermophile by Heterologous Production of an 18-Subunit Membrane-Bound Complex. J. Biol. Chem 2014, 289 , 2873–2879.24318960 (434) Pierik AJ ; Hulstein M ; Hagen WR ; Albracht SPJ A Low-Spin Iron with CN and CO as Intrinsic Ligands Forms the Core of the Active Site in [Fe]-Hydrogenases. Eur. J. Biochem 1998, 258 , 572–578.9874225 (435) Pandey AS ; Harris TV ; Giles LJ ; Peters JW ; Szilagyi RK Dithiomethylether as a Ligand in the Hydrogenase H-Cluster. J. Am. Chem. Soc 2008, 130 , 4533–4540.18324814 (436) Silakov A ; Wenk B ; Reijerse E ; Lubitz W 14N HYSCORE Investigation of the H-Cluster of [FeFe] Hydrogenase: Evidence for a Nitrogen in the Dithiol Bridge. Phys. Chem. Chem. Phys 2009, 11 , 6592–6599.19639134 (437) Vignais PM ; Billoud B ; Meyer J Classification and Phylogeny of Hydrogenases. FEMS Microbiol. Rev 2001, 25 , 455–501.11524134 (438) Horner DS ; Heil B ; Happe T ; Embley TM Iron Hydrogenases - Ancient Enzymes in Modern Eukaryotes. Trends Biochem. Sci 2002, 27 , 148–153.11893512 (439) Calusinska M ; Happe T ; Joris B ; Wilmotte A The Surprising Diversity of Clostridial Hydrogenases: A Comparative Genomic Perspective. Microbiology 2010, 156 , 1575–1588.20395274 (440) Chongdar N ; Birrell JA ; Pawlak K ; Sommer C ; Reijerse EJ ; Rüdiger O ; Lubitz W ; Ogata H Unique Spectroscopic Properties of the H-Cluster in a Putative Sensory [FeFe] Hydrogenase. J. Am. Chem. Soc 2018, 140 , 1057–1068.29251926 (441) Chongdar N ; Pawlak K ; Rüdiger O ; Reijerse EJ ; Rodríguez-Maciá P ; Lubitz W ; Birrell JA ; Ogata H Spectroscopic and Biochemical Insight into an Electron-Bifurcating [FeFe] Hydrogenase. J. Biol. Inorg. Chem 2020, 25 , 135–149.31823008 (442) Land H ; Ceccaldi P ; Mészáros LS ; Lorenzi M ; Redman HJ ; Senger M ; Stripp ST ; Berggren G Discovery of Novel [FeFe]-Hydrogenases for Biocatalytic H2-Production. Chem. Sci 2019, 10 , 9941–9948.32055351 (443) Land H ; Sekretareva A ; Huang P ; Redman HJ ; Nemeth B ; Polidori N ; Meszaros LS ; Senger M ; Stripp ST ; Berggren G Characterization of a Putative Sensory [FeFe]-Hydrogenase Provides New Insight into the Role of the Active Site Architecture. Chem. Sci 2020, 11 , 12789–12801.34094474 (444) Furlan C ; Chongdar N ; Gupta P ; Lubitz W ; Ogata H ; James N ; Birrell JA Structural Insight on the Mechanism of an Electron-Bifurcating [FeFe] Hydrogenase. ChemRxiv 2021, 1–13. (445) Lampret O ; Esselborn J ; Haas R ; Rutz A ; Booth RL ; Kertess L ; Wittkamp F ; Megarity CF ; Armstrong FA ; Winkler M ; The Final Steps of [FeFe]-Hydrogenase Maturation. Proc. Natl. Acad. Sci. U. S. A 2019, 116 , 15802–15810.31337676 (446) Winkler M ; Senger M ; Duan J ; Esselborn J ; Wittkamp F ; Hofmann E ; Apfel U-P ; Stripp ST ; Happe T Accumulating the Hydride State in the Catalytic Cycle of [FeFe]-Hydrogenases. Nat. Commun 2017, 8 , 16115.28722011 (447) Lampret O ; Duan J ; Hofmann E ; Winkler M ; Armstrong FA ; Happe T The Roles of Long-Range Proton-Coupled Electron Transfer in the Directionality and Efficiency of [FeFe]-Hydrogenases. Proc. Natl. Acad. Sci. U. S. A 2020, 117 , 20520–20529.32796105 (448) Duan J ; Senger M ; Esselborn J ; Engelbrecht V ; Wittkamp F ; Apfel U-P ; Hofmann E ; Stripp ST ; Happe T ; Winkler M Crystallographic and Spectroscopic Assignment of the Proton Transfer Pathway in [FeFe]-Hydrogenases. Nat. Commun 2018, 9 , 4726.30413719 (449) Cornish AJ ; Ginovska B ; Thelen A ; Da Silva JCS ; Soares TA ; Raugei S ; Dupuis M ; Shaw WJ ; Hegg EL Single-Amino Acid Modifications Reveal Additional Controls on the Proton Pathway of [FeFe]-Hydrogenase. Biochemistry 2016, 55 , 3165–3173.27186945 (450) Knörzer P ; Silakov A ; Foster CE ; Armstrong FA ; Lubitz W ; Happe T Importance of the Protein Framework for Catalytic Activity of [FeFe]-Hydrogenases. J. Biol. Chem 2012, 287 , 1489–1499.22110126 (451) Bingham AS ; Smith PR ; Swartz JR Evolution of an [FeFe] Hydrogenase with Decreased Oxygen Sensitivity. Int. J. Hydrogen Energy 2012, 37 , 2965–2976. (452) Koo J ; Swartz JR System Analysis and Improved [FeFe] Hydrogenase O2 Tolerance Suggest Feasibility for Photosynthetic H2 Production. Metab. Eng 2018, 49 , 21–27.30057338 (453) Koo J Enhanced Aerobic H2 Production by Engineering an [FeFe] Hydrogenase from Clostridium Pasteurianum. Int. J. Hydrogen Energy 2020, 45 , 10673–10679. (454) Morra S ; Giraudo A ; Di Nardo G ; King PW ; Gilardi G ; Valetti F Site Saturation Mutagenesis Demonstrates a Central Role for Cysteine 298 as Proton Donor to the Catalytic Site in CaHydA [FeFe]-Hydrogenase. PLoS One 2012, 7 , No. e48400.23133586 (455) Morra S ; Maurelli S ; Chiesa M ; Mulder DW ; Ratzloff MW ; Giamello E ; King PW ; Gilardi G ; Valetti F The Effect of a C298D Mutation in CaHydA [FeFe]-Hydrogenase: Insights into the Protein-Metal Cluster Interaction by EPR and FTIR Spectroscopic Investigation. Biochim. Biophys. Acta - Bioenerg 2016, 1857 , 98–106. (456) Mulder DW ; Guo Y ; Ratzloff MW ; King PW Identification of a Catalytic Iron-Hydride at the H-Cluster of [FeFe]-Hydrogenase. J. Am. Chem. Soc 2017, 139 , 83–86.27973768 (457) Mulder DW ; Ratzloff MW ; Bruschi M ; Greco C ; Koonce E ; Peters JW ; King PW Investigations on the Role of Proton-Coupled Electron Transfer in Hydrogen Activation by [FeFe]-Hydrogenase. J. Am. Chem. Soc 2014, 136 , 15394–15402.25286239 (458) Pham CC ; Mulder DW ; Pelmenschikov V ; King PW ; Ratzloff MW ; Wang H ; Mishra N ; Alp EE ; Zhao J ; Hu MY ; Terminal Hydride Species in [FeFe]-Hydrogenases Are Vibrationally Coupled to the Active Site Environment. Angew. Chemie Int. Ed 2018, 57 , 10605–10609. (459) Rumpel S ; Sommer C ; Reijerse E ; Farès C ; Lubitz W Direct Detection of the Terminal Hydride Intermediate in [FeFe] Hydrogenase by NMR Spectroscopy. J. Am. Chem. Soc 2018, 140 , 3863–3866.29521088 (460) Rodríguez-Maciá P ; Kertess L ; Burnik J ; Birrell JA ; Hofmann E ; Lubitz W ; Happe T ; Rüdiger O His-Ligation to the [4Fe-4S] Subcluster Tunes the Catalytic Bias of [FeFe] Hydrogenase. J. Am. Chem. Soc 2019, 141 , 472–481.30545220 (461) Kertess L ; Adamska-Venkatesh A ; Rodríguez-Maciá P ; Rüdiger O ; Lubitz W ; Happe T Influence of the [4Fe-4S] Cluster Coordinating Cysteines on Active Site Maturation and Catalytic Properties of C. Reinhardtii [FeFe]-Hydrogenase. Chem. Sci 2017, 8 , 8127–8137.29568461 (462) Orain C ; Saujet L ; Gauquelin C ; Soucaille P ; Meynial-Salles I ; Baffert C ; Fourmond V ; Bottin H ; Léger C Electrochemical Measurements of the Kinetics of Inhibition of Two FeFe Hydrogenases by O 2 Demonstrate That the Reaction Is Partly Reversible. J. Am. Chem. Soc 2015, 137 , 12580–12587.26352172 (463) Sybirna K ; Ezanno P ; Baffert C ; Léger C ; Bottin H Arginine171 of Chlamydomonas Reinhardtii [Fe-Fe] Hydrogenase HydA1 Plays a Crucial Role in Electron Transfer to Its Catalytic Center. Int. J. Hydrogen Energy 2013, 38 , 2998–3002. (464) Lampret O ; Adamska-Venkatesh A ; Konegger H ; Wittkamp F ; Apfel U-P ; Reijerse EJ ; Lubitz W ; Rüdiger O ; Happe T ; Winkler M Interplay between CN - Ligands and the Secondary Coordination Sphere of the H-Cluster in [FeFe]-Hydrogenases. J. Am. Chem. Soc 2017, 139 , 18222–18230.29179539 (465) Kubas A ; Orain C ; De Sancho D ; Saujet L ; Sensi M ; Gauquelin C ; Meynial-Salles I ; Soucaille P ; Bottin H ; Baffert C ; Fourmond V ; Best RB ; Blumberger J ; Léger C Mechanism of O2 Diffusion and Reduction in FeFe Hydrogenases. Nat. Chem 2017, 9 , 88–95.27995927 (466) Winkler M ; Kuhlgert S ; Hippler M ; Happe T Characterization of the Key Step for Light-Driven Hydrogen Evolution in Green Algae. J. Biol. Chem 2009, 284 , 36620–36627.19846550 (467) Engelbrecht V ; Liedtke K ; Rutz A ; Yadav S ; Günzel A ; Happe T One Isoform for One Task? The Second Hydrogenase of Chlamydomonas Reinhardtii Prefers Hydrogen Uptake. Int. J. Hydrogen Energy 2021, 46 , 7165–7175. (468) Berggren G ; Adamska A ; Lambertz C ; Simmons TR ; Esselborn J ; Atta M ; Gambarelli S ; Mouesca J-M ; Reijerse E ; Lubitz W ; Biomimetic Assembly and Activation of [FeFe]-Hydrogenases. Nature 2013, 499 , 66–69.23803769 (469) Esselborn J ; Lambertz C ; Adamska-Venkatesh A ; Simmons T ; Berggren G ; Noth J ; Siebel J ; Hemschemeier A ; Artero V ; Reijerse E ; Spontaneous Activation of [FeFe]-Hydrogenases by an Inorganic [2Fe] Active Site Mimic. Nat. Chem. Biol 2013, 9 , 607–609.23934246 (470) Adamska-Venkatesh A ; Krawietz D ; Siebel JF ; Weber K ; Happe T ; Reijerse E ; Lubitz W New Redox States Observed in [FeFe] Hydrogenases Reveal Redox Coupling within the H-Cluster. J. Am. Chem. Soc 2014, 136 , 11339–11346.25025613 (471) Siebel JF ; Adamska-Venkatesh A ; Weber K ; Rumpel S ; Reijerse E ; Lubitz W Hybrid [FeFe]-Hydrogenases with Modified Active Sites Show Remarkable Residual Enzymatic Activity. Biochemistry 2015, 54 , 1474–1483.25633077 (472) Adamska-Venkatesh A ; Simmons TR ; Siebel JF ; Artero V ; Fontecave M ; Reijerse E ; Lubitz W Artificially Maturated [FeFe] Hydrogenase from Chlamydomonas Reinhardtii: A HYSCORE and ENDOR Study of a Non-Natural H-Cluster. Phys. Chem. Chem. Phys 2015, 17 , 5421–5430.25613229 (473) Adamska-Venkatesh A ; Roy S ; Siebel JF ; Simmons TR ; Fontecave M ; Artero V ; Reijerse E ; Lubitz W Spectroscopic Characterization of the Bridging Amine in the Active Site of [FeFe] Hydrogenase Using Isotopologues of the H-Cluster. J. Am. Chem. Soc 2015, 137 , 12744–12747.26393426 (474) Reijerse EJ ; Pham CC ; Pelmenschikov V ; Gilbert-Wilson R ; Adamska-Venkatesh A ; Siebel JF ; Gee LB ; Yoda Y ; Tamasaku K ; Lubitz W ; Direct Observation of an Iron-Bound Terminal Hydride in [FeFe]-Hydrogenase by Nuclear Resonance Vibrational Spectroscopy. J. Am. Chem. Soc 2017, 139 , 4306–4309.28291336 (475) Noth J ; Esselborn J ; Güldenhaupt J ; Brünje A ; Sawyer A ; Apfel U-P ; Gerwert K ; Hofmann E ; Winkler M ; Happe T [FeFe]-Hydrogenase with Chalcogenide Substitutions at the H-Cluster Maintains Full H2 Evolution Activity. Angew. Chemie Int. Ed 2016, 55 , 8396–8400. (476) Kertess L ; Wittkamp F ; Sommer C ; Esselborn J ; Rüdiger O ; Reijerse EJ ; Hofmann E ; Lubitz W ; Winkler M ; Happe T ; Chalcogenide Substitution in the [2Fe] Cluster of [FeFe]-Hydrogenases Conserves High Enzymatic Activity. Dalt. Trans 2017, 46 , 16947–16958. (477) Sommer C ; Rumpel S ; Roy S ; Farès C ; Artero V ; Fontecave M ; Reijerse E ; Lubitz W Spectroscopic Investigations of a Semi-Synthetic [FeFe] Hydrogenase with Propane Di-Selenol as Bridging Ligand in the Binuclear Subsite: Comparison to the Wild Type and Propane Di-Thiol Variants. JBIC, J. Biol. Inorg. Chem 2018, 23 , 481–491.29627860 (478) Sommer C ; Richers CP ; Lubitz W ; Rauchfuss TB ; Reijerse EJA [RuRu] Analogue of an [FeFe]-Hydrogenase Traps the Key Hydride Intermediate of the Catalytic Cycle. Angew. Chemie Int. Ed 2018, 57 , 5429–5432. (479) Duan J ; Mebs S ; Laun K ; Wittkamp F ; Heberle J ; Hofmann E ; Apfel U-P ; Winkler M ; Senger M ; Haumann M ; Geometry of the Catalytic Active Site in [FeFe]-Hydrogenase Is Determined by Hydrogen Bonding and Proton Transfer. ACS Catal 2019, 9 , 9140–9149. (480) Esselborn J ; Muraki N ; Klein K ; Engelbrecht V ; Metzler-Nolte N ; Apfel U-P ; Hofmann E ; Kurisu G ; Happe T A Structural View of Synthetic Cofactor Integration into [FeFe]-Hydrogenases. Chem. Sci 2016, 7 , 959–968.29896366 (481) Popescu C ; Münck E Electronic Structure of the H Cluster in [FeFe]-Hydrogenases. J. Am. Chem. Soc 1999, 121 (34 ), 7877–7884. (482) Pereira ASS ; Tavares P ; Moura I ; Moura JJGJ ; Huynh BHH Mössbauer Characterization of the Iron-Sulfur Clusters in Desulfovibrio v Ulgaris Hydrogenase. J. Am. Chem. Soc 2001, 123 , 2771–2782.11456963 (483) Silakov A ; Reijerse EJ ; Albracht SPJ ; Hatchikian EC ; Lubitz W The Electronic Structure of the H-Cluster in the [FeFe]-Hydrogenase from Desulfovibrio Desulfuricans: A Q-Band 57Fe-ENDOR and HYSCORE Study. J. Am. Chem. Soc 2007, 129 , 11447–11458.17722921 (484) Sidabras JW ; Duan J ; Winkler M ; Happe T ; Hussein R ; Zouni A ; Suter D ; Schnegg A ; Lubitz W ; Reijerse EJ Extending Electron Paramagnetic Resonance to Nanoliter Volume Protein Single Crystals Using a Self-Resonant Microhelix. Sci. Adv 2019, 5 , No. eaay1394.31620561 (485) De Lacey AL ; Stadler C ; Cavazza C ; Hatchikian EC ; Fernandez VM FTIR Characterization of the Active Site of the Fe-Hydrogenase from Desulfovibrio Desulfuricans. J. Am. Chem. Soc 2000, 122 , 11232–11233. (486) Chen Z ; Lemon BJ ; Huang S ; Swartz DJ ; Peters JW ; Bagley KA Infrared Studies of the CO-Inhibited Form of the Fe-Only Hydrogenase from Clostridium Pasteurianum I: Examination of Its Light Sensitivity at Cryogenic Temperatures. Biochemistry 2002, 41 , 2036–2043.11827551 (487) Roseboom W ; De Lacey AL ; Fernandez VM ; Hatchikian EC ; Albracht SPJ The Active Site of the [FeFe]-Hydrogenase from Desulfovibrio Desulfuricans. II. Redox Properties, Light Sensitivity and CO-Ligand Exchange as Observed by Infrared Spectroscopy. J. Biol. Inorg. Chem 2006, 11 , 102–118.16323019 (488) Senger M ; Mebs S ; Duan J ; Shulenina O ; Laun K ; Kertess L ; Wittkamp F ; Apfel U-P ; Happe T ; Winkler M ; Protonation/Reduction Dynamics at the [4Fe-4S] Cluster of the Hydrogen-Forming Cofactor in [FeFe]-Hydrogenases. Phys. Chem. Chem. Phys 2018, 20 , 3128–3140.28884175 (489) Goldet G ; Brandmayr C ; Stripp ST ; Happe T ; Cavazza C ; Fontecilla-Camps JC ; Armstrong FA Electrochemical Kinetic Investigations of the Reactions of [FeFe]-Hydrogenases with Carbon Monoxide and Oxygen: Comparing the Importance of Gas Tunnels and Active-Site Electronic/Redox Effects. J. Am. Chem. Soc 2009, 131 , 14979–14989.19824734 (490) Telser J ; Benecky MJ ; Adams MWW ; Mortenson LE ; Hoffman BM An EPR and Electron Nuclear Double Resonance Investigation of Carbon Monoxide Binding to Hydrogenase I (Bidirectional) from Clostridium Pasteurianum W5. J. Biol. Chem 1986, 261 , 13536–13541.3020036 (491) Telser J ; Benecky MJ ; Adams MWW ; Mortenson LE ; Hoffman BM EPR and Electron Nuclear Double Resonance Investigation of Oxidized Hydrogenase II (Uptake) from Clostridium Pasteurianum W5. Effects of Carbon Monoxide Binding. J. Biol. Chem 1987, 262 , 6589–6594.3032973 (492) Reijerse E ; Birrell JA ; Lubitz W Spin Polarization Reveals the Coordination Geometry of the [FeFe] Hydrogenase Active Site in Its CO-Inhibited State. J. Phys. Chem. Lett 2020, 11 , 4597–4602.32420744 (493) Senger M ; Mebs S ; Duan J ; Wittkamp F ; Apfel U-P ; Heberle J ; Haumann M ; Stripp ST Stepwise Isotope Editing of [FeFe]-Hydrogenases Exposes Cofactor Dynamics. Proc. Natl. Acad. Sci. U. S. A 2016, 113 , 8454–8459.27432985 (494) Laun K ; Mebs S ; Duan J ; Wittkamp F ; Apfel U-P ; Happe T ; Winkler M ; Haumann M ; Stripp ST Spectroscopical Investigations on the Redox Chemistry of [FeFe]-Hydrogenases in the Presence of Carbon Monoxide. Molecules 2018, 23 , 1669. (495) Zilberman S ; Stiefel EI ; Cohen MH ; Car R Resolving the CO/CN Ligand Arrangement in CO-Inactivated [FeFe] Hydrogenase by First Principles Density Functional Theory Calculations. Inorg. Chem 2006, 45 , 5715–5717.16841968 (496) Greco C ; Bruschi M ; Heimdal J ; Fantucci P ; De Gioia L ; Ryde U Structural Insights into the Active-Ready Form of [FeFe]-Hydrogenase and Mechanistic Details of Its Inhibition by Carbon Monoxide. Inorg. Chem 2007, 46 , 7256–7258.17676838 (497) Yu L ; Greco C ; Bruschi M ; Ryde U ; De Gioia L ; Reiher M Targeting Intermediates of [FeFe]-Hydrogenase by CO and CN Vibrational Signatures. Inorg. Chem 2011, 50 , 3888–3900.21443182 (498) Sommer C ; Adamska-Venkatesh A ; Pawlak K ; Birrell JA ; Rüdiger O ; Reijerse EJ ; Lubitz W Proton Coupled Electronic Rearrangement within the H-Cluster as an Essential Step in the Catalytic Cycle of [FeFe] Hydrogenases. J. Am. Chem. Soc 2017, 139 , 1440–1443.28075576 (499) Katz S ; Noth J ; Horch M ; Shafaat HS ; Happe T ; Hildebrandt P ; Zebger I Vibrational Spectroscopy Reveals the Initial Steps of Biological Hydrogen Evolution. Chem. Sci 2016, 7 , 6746–6752.28451119 (500) Singleton ML ; Bhuvanesh N ; Reibenspies JH ; Darensbourg MY Synthetic Support of De Novo Design: Sterically Bulky [FeFe]-Hydrogenase Models. Angew. Chemie Int. Ed 2008, 47 , 9492–9495. (501) Senger M ; Laun K ; Wittkamp F ; Duan J ; Haumann M ; Happe T ; Winkler M ; Apfel U-P ; Stripp ST Proton-Coupled Reduction of the Catalytic [4Fe-4S] Cluster in [FeFe]-Hydrogenases. Angew. Chemie Int. Ed 2017, 56 , 16503–16506. (502) Mebs S ; Senger M ; Duan J ; Wittkamp F ; Apfel U-P ; Happe T ; Winkler M ; Stripp ST ; Haumann M Bridging Hydride at Reduced H-Cluster Species in [FeFe]-Hydrogenases Revealed by Infrared Spectroscopy, Isotope Editing, and Quantum Chemistry. J. Am. Chem. Soc 2017, 139 , 12157–12160.28825810 (503) Birrell JA ; Pelmenschikov V ; Mishra N ; Wang H ; Yoda Y ; Rauchfuss TB ; Cramer SP ; Lubitz W ; Debeer S ; Spectroscopic and Computational Evidence That [FeFe] Hydrogenases Operate Exclusively with CO-Bridged Intermediates. J. Am. Chem. Soc 2020, 142 , 222–232.31820961 (504) Lorent C ; Katz S ; Duan J ; Kulka CJ ; Caserta G ; Teutloff C ; Yadav S ; Apfel U-P ; Winkler M ; Happe T ; Shedding Light on Proton and Electron Dynamics in [FeFe] Hydrogenases. J. Am. Chem. Soc 2020, 142 , 5493–5497.32125830 (505) Stripp ST ; Mebs S ; Haumann M Temperature Dependence of Structural Dynamics at the Catalytic Cofactor of [FeFe]-Hydrogenase. Inorg. Chem 2020, 59 , 16474–16488.33147959 (506) Haumann M ; Stripp ST The Molecular Proceedings of Biological Hydrogen Turnover. Acc. Chem. Res 2018, 51 , 1755–1763.30001117 (507) Siegbahn PEM ; Liao RZ Energetics for Proton Reduction in FeFe Hydrogenase. J. Phys. Chem. A 2020, 124 , 10540–10549.33275428 (508) Kleinhaus JT ; Wittkamp F ; Yadav S ; Siegmund D ; Apfel U-P [FeFe]-Hydrogenases: Maturation and Reactivity of Enzymatic Systems and Overview of Biomimetic Models. Chem. Soc. Rev 2021, 50 (50 ), 1668–1784.33305760 (509) Birrell JA ; Rodríguez-Maciá P ; Reijerse EJ ; Martini MA ; Lubitz W The Catalytic Cycle of [FeFe] Hydrogenase: A Tale of Two Sites. Coord. Chem. Rev 2021, 449 , 214191. (510) Albracht SPJ ; Roseboom W ; Hatchikian EC The Active Site of the [FeFe]-Hydrogenase from Desulfovibrio Desulfuricans. I. Light Sensitivity and Magnetic Hyperfine Interactions as Observed by Electron Paramagnetic Resonance. J. Biol. Inorg. Chem 2006, 11 , 88–101.16323020 (511) Adamska-Venkatesh A ; Silakov A ; Lambertz C ; Rüdiger O ; Happe T ; Reijerse E ; Lubitz W Identification and Characterization of the “Super-Reduced” State of the H-Cluster in [FeFe] Hydrogenase: A New Building Block for the Catalytic Cycle? Angew. Chemie Int. Ed 2012, 51 , 11458–11462. (512) Laun K ; Baranova I ; Duan J ; Kertess L ; Wittkamp F ; Apfel U-P ; Happe T ; Senger M ; Stripp ST Site-Selective Protonation of the One-Electron Reduced Cofactor in [FeFe]-Hydrogenase. Dalt. Trans 2021, 50 , 3641–3650. (513) Krasna AI ; Rittenberg D The Mechanism of Action of the Enzyme Hydrogenase. J. Am. Chem. Soc 1954, 76 , 3015–3020. (514) Pelmenschikov V ; Birrell JA ; Pham CC ; Mishra N ; Wang H ; Sommer C ; Reijerse E ; Richers CP ; Tamasaku K ; Yoda Y ; Reaction Coordinate Leading to H2 Production in [FeFe]-Hydrogenase Identified by Nuclear Resonance Vibrational Spectroscopy and Density Functional Theory. J. Am. Chem. Soc 2017, 139 , 16894–16902.29054130 (515) Hajj V ; Baffert C ; Sybirna K ; Meynial-Salles I ; Soucaille P ; Bottin H ; Fourmond V ; Léger C FeFe Hydrogenase Reductive Inactivation and Implication for Catalysis. Energy Environ. Sci 2014, 7 , 715–719. (516) Patil DS ; Moura JJ ; He SH ; Teixeira M ; Prickril BC ; DerVartanian DV ; Peck HD ; LeGall J ; Huynh BH EPR-Detectable Redox Centers of the Periplasmic Hydrogenase from Desulfovibrio Vulgaris. J. Biol. Chem 1988, 263 , 18732–18738.2848804 (517) van der Spek TM ; Arendsen AF ; Happe RP ; Yun S ; Bagley KA ; Stufkens DJ ; Hagen WR ; Albracht SPJ ; Spek TM ; Arendsen AF ; Similarities in the Architecture of the Active Sites of Ni-Hydrogenases and Fe-Hydrogenases Detected by Means of Infrared Spectroscopy. Eur. J. Biochem 1996, 237 , 629–634.8647106 (518) van der Westen HM ; Mayhew SG ; Veeger C Separation of Hydrogenase from Intact Cells of Desulfovibrio Vulgaris Purification and Properties. FEBS Lett 1978, 86 , 122–126.620819 (519) Rodríguez-Maciá P ; Reijerse EJ ; Van Gastel M ; Debeer S ; Lubitz W ; Rüdiger O ; Birrell JA Sulfide Protects [FeFe] Hydrogenases from O2. J. Am. Chem. Soc 2018, 140 , 9346–9350.30008217 (520) Rodríguez-Maciá P ; Galle LM ; Bjornsson R ; Lorent C ; Zebger I ; Yoda Y ; Cramer SP ; DeBeer S ; Span I ; Birrell JA Caught in the Hinact: Crystal Structure and Spectroscopy Reveal a Sulfur Bound to the Active Site of an O2-Stable State of [FeFe] Hydrogenase. Angew. Chem. Int. Ed 2020, 59 , 16786–16794. (521) Felbek C ; Arrigoni F ; de Sancho D ; Jacq-Bailly A ; Best RB ; Fourmond V ; Bertini L ; Léger C Mechanism of Hydrogen Sulfide-Dependent Inhibition of FeFe Hydrogenase. ACS Catal 2021, 11 , 15162–15176. (522) Parkin A ; Cavazza C ; Fontecilla-Camps JC ; Armstrong F. a. Electrochemical Investigations of the Interconversions between Catalytic and Inhibited States of the [FeFe]-Hydrogenase from Desulfovibrio Desulfuricans. J. Am. Chem. Soc 2006, 128 , 16808–16815.17177431 (523) del Barrio M ; Sensi M ; Fradale L ; Bruschi M ; Greco C ; de Gioia L ; Bertini L ; Fourmond V ; Léger C Interaction of the H-Cluster of FeFe Hydrogenase with Halides. J. Am. Chem. Soc 2018, 140 , 5485–5492.29590528 (524) Morra S ; Arizzi M ; Valetti F ; Gilardi G Oxygen Stability in the New [FeFe]-Hydrogenase from Clostridium Beijerinckii SM10 (CbA5H). Biochemistry 2016, 55 , 5897–5900.27749036 (525) Bruska MK ; Stiebritz MT ; Reiher M Regioselectivity of H-Cluster Oxidation. J. Am. Chem. Soc 2011, 133 , 20588–20603.22106822 (526) Stiebritz MT ; Reiher M Hydrogenases and Oxygen. Chem. Sci 2012, 3 , 1739–1751. (527) Finkelmann AR ; Stiebritz MT ; Reiher M Activation Barriers of Oxygen Transformation at the Active Site of [FeFe] Hydrogenases. Inorg. Chem 2014, 53 , 11890–11902.25345467 (528) Kubas A ; De Sancho D ; Best RB ; Blumberger J Aerobic Damage to [FeFe]-Hydrogenases: Activation Barriers for the Chemical Attachment of O2. Angew. Chem. Int. Ed 2014, 53 , 4081–4084. (529) Stripp ST ; Goldet G ; Brandmayr C ; Sanganas O ; Vincent KA ; Haumann M ; Armstrong FA ; Happe T How Oxygen Attacks [FeFe] Hydrogenases from Photosynthetic Organisms. Proc. Natl. Acad. Sci. U. S. A 2009, 106 , 17331–17336.19805068 (530) Lambertz C ; Leidel N ; Havelius KGV ; Noth J ; Chernev P ; Winkler M ; Happe T ; Haumann M O2 Reactions at the Six-Iron Active Site (H-Cluster) in [FeFe]-Hydrogenase. J. Biol. Chem 2011, 286 , 40614–40623.21930709 (531) Swanson KD ; Ratzloff MW ; Mulder DW ; Artz JH ; Ghose S ; Hoffman A ; White S ; Zadvornyy OA ; Broderick JB ; Bothner B ; [FeFe]-Hydrogenase Oxygen Inactivation Is Initiated at the H Cluster 2Fe Subcluster. J. Am. Chem. Soc 2015, 137 , 1809–1816.25579778 (532) Mebs S ; Kositzki R ; Duan J ; Kertess L ; Senger M ; Wittkamp F ; Apfel U-P ; Happe T ; Stripp ST ; Winkler M ; Hydrogen and Oxygen Trapping at the H-Cluster of [FeFe]-Hydrogenase Revealed by Site-Selective Spectroscopy and QM/MM Calculations. BBA - Bioenerg 2018, 1859 , 28–41. (533) Esselborn J ; Kertess L ; Apfel U-P ; Hofmann E ; Happe T Loss of Specific Active-Site Iron Atoms in Oxygen-Exposed [FeFe]-Hydrogenase Determined by Detailed X-Ray Structure Analyses. J. Am. Chem. Soc 2019, 141 , 17721–17728.31609603 (534) Kuchenreuther JM ; Britt RD ; Swartz JR New Insights into [FeFe] Hydrogenase Activation and Maturase Function. PLoS One 2012, 7 , No. e45850.23049878 (535) Mulder DW ; Ortillo DO ; Gardenghi DJ ; Naumov AV ; Ruebush SS ; Szilagyi RK ; Huynh B ; Broderick JB ; Peters JW Activation of HydA(DeltaEFG) Requires a Preformed [4Fe-4S] Cluster. Biochemistry 2009, 48 , 6240–6248.19435321 (536) Böck A ; King PW ; Blokesch M ; Posewitz MC Maturation of Hydrogenases. Advances In Microbial Physiology 2006, 51 , 1–71.17091562 (537) McGlynn SE ; Shepard EM ; Winslow MA ; Naumov AV ; Duschene KS ; Posewitz MC ; Broderick WE ; Broderick JB ; Peters JW HydF as a Scaffold Protein in [FeFe] Hydrogenase H-Cluster Biosynthesis. FEBS Lett 2008, 582 , 2183–2187.18501709 (538) Caserta G ; Pecqueur L ; Adamska-Venkatesh A ; Papini C ; Roy S ; Artero V ; Atta M ; Reijerse E ; Lubitz W ; Fontecave M Structural and Functional Characterization of the Hydrogenase-Maturation HydF Protein. Nat. Chem. Biol 2017, 13 , 779–784.28553946 (539) Scott AG ; Szilagyi RK ; Mulder DW ; Ratzloff MW ; Byer AS ; King PW ; Broderick WE ; Shepard EM ; Broderick JB Compositional and Structural Insights into the Nature of the H-Cluster Precursor on HydF. Dalt. Trans 2018, 47 , 9521–9535. (540) Shepard EM ; Duffus BR ; George SJ ; McGlynn SE ; Challand MR ; Swanson KD ; Roach PL ; Cramer SP ; Peters JW ; Broderick JB [FeFe]-Hydrogenase Maturation: HydG-Catalyzed Synthesis of Carbon Monoxide. J. Am. Chem. Soc 2010, 132 , 9247–9249.20565074 (541) Driesener RC ; Challand MR ; McGlynn SE ; Shepard EM ; Boyd ES ; Broderick JB ; Peters JW ; Roach PL [FeFe]-Hydrogenase Cyanide Ligands Derived from S-Adenosylmethionine-Dependent Cleavage of Tyrosine. Angew. Chemie Int. Ed 2010, 49 , 1687–1690. (542) Kuchenreuther JM ; Myers WK ; Suess DLM ; Stich TA ; Pelmenschikov V ; Shiigi S. a ; Cramer SP ; Swartz JR ; Britt RD ; George SJ The HydG Enzyme Generates an Fe(CO)2(CN) Synthon in Assembly of the FeFe Hydrogenase H-Cluster. Science 2014, 343 , 424–427.24458644 (543) Nicolet Y ; Pagnier A ; Zeppieri L ; Martin L ; Amara P ; Fontecilla-Camps JC Crystal Structure of HydG from Carboxydothermus Hydrogenoformans: A Trifunctional [FeFe]-Hydrogenase Maturase. ChemBioChem 2015, 16 , 397–402.25504963 (544) Rao G ; Tao L ; Suess DLM ; Britt RD A [4Fe-4S]-Fe(CO)(CN)-l-Cysteine Intermediate Is the First Organometallic Precursor in [FeFe] Hydrogenase H-Cluster Bioassembly. Nat. Chem 2018, 10 , 555–560.29632334 (545) Betz JN ; Boswell NW ; Fugate CJ ; Holliday GL ; Akiva E ; Scott AG ; Babbitt PC ; Peters JW ; Shepard EM ; Broderick JB Hydrogenase Maturation: Insights into the Role Hyde Plays in Dithiomethylamine Biosynthesis. Biochemistry 2015, 54 , 1807–1818.25654171 (546) Rohac R ; Amara P ; Benjdia A ; Martin L ; Ruffié P ; Favier A ; Berteau O ; Mouesca JM ; Fontecilla-Camps JC ; Nicolet Y Carbon-Sulfur Bond-Forming Reaction Catalysed by the Radical SAM Enzyme HydE. Nat. Chem 2016, 8 , 491–500.27102684 (547) Rao G ; Tao L ; Britt RD Serine Is the Molecular Source of the NH(CH2)2 Bridgehead Moiety of the in Vitro Assembled [FeFe] Hydrogenase H-Cluster Guodong. Chem. Sci 2020, 11 , 1241–1247. (548) Simmons TR ; Berggren G ; Bacchi M ; Fontecave M ; Artero V Mimicking Hydrogenases: From Biomimetics to Artificial Enzymes. Coord. Chem. Rev 2014, 270–271 , 127–150. (549) Artero V ; Berggren G ; Atta M ; Caserta G ; Roy S ; Pecqueur L ; Fontecave M From Enzyme Maturation to Synthetic Chemistry: The Case of Hydrogenases. Acc. Chem. Res 2015, 48 , 2380–2387.26165393 (550) Peters JW Structure and Mechanism of Iron-Only Hydrogenases. Curr. Opin. Struct. Biol 1999, 9 , 670–676.10607666 (551) Adamska-Venkatesh A ; Simmons TR ; Siebel JF ; Artero V ; Fontecave M ; Reijerse E ; Lubitz W Artificially Maturated [FeFe] Hydrogenase from Chlamydomonas Reinhardtii: A HYSCORE and ENDOR Study of a Non-Natural H-Cluster. Phys. Chem. Chem. Phys 2015, 17 , 5421–5430.25613229 (552) Bruschi M ; Greco C ; Kaukonen M ; Fantucci P ; Ryde U ; De Gioia L Influence of the [2Fe]H Subcluster Environment on the Properties of Key Intermediates in the Catalytic Cycle of [FeFe] Hydrogenases: Hints for the Rational Design of Synthetic Catalysts. Angew. Chemie Int. Ed 2009, 48 , 3503–3506. (553) Braga D ; Grepioni F Hydrogen-Bonding Interactions with the CO Ligand in the Solid State. Acc. Chem. Res 1997, 30 , 81–87. (554) Artz JH ; Zadvornyy OA ; Mulder DW ; Keable SM ; Cohen AE ; Ratzloff MW ; Williams SG ; Ginovska B ; Kumar N ; Song J ; Tuning Catalytic Bias of Hydrogen Gas Producing Hydrogenases. J. Am. Chem. Soc 2020, 142 , 1227–1235.31816235 (555) Ratzloff MW ; Artz JH ; Mulder DW ; Collins RT ; Furtak TE ; King PW CO-Bridged H-Cluster Intermediates in the Catalytic Mechanism of [FeFe]-Hydrogenase CaI. J. Am. Chem. Soc 2018, 140 , 7623–7628.29792026 (556) Ginovska-Pangovska B ; Dutta A ; Reback ML ; Linehan JC ; Shaw WJ Beyond the Active Site: The Impact of the Outer Coordination Sphere on Electrocatalysts for Hydrogen Production and Oxidation. Acc. Chem. Res 2014, 47 , 2621–2630.24945095 (557) Esmieu C ; Raleiras P ; Berggren G From Protein Engineering to Artificial Enzymes-Biological and Biomimetic Approaches towards Sustainable Hydrogen Production. Sustain. Energy Fuels 2018, 2 , 724–750.31497651 (558) Amanullah S ; Saha P ; Nayek A ; Ahmed ME ; Dey A Biochemical and Artificial Pathways for the Reduction of Carbon Dioxide, Nitrite and the Competing Proton Reduction: Effect of 2nd sphere Interactions in Catalysis. Chem. Soc. Rev 2021, 50 , 3755–3823.33514959 (559) Hong G ; Cornish AJ ; Hegg EL ; Pachter R On Understanding Proton Transfer to the Biocatalytic [Fe-Fe](H) Sub-Cluster in [Fe-Fe]H(2)Ases: QM/MM MD Simulations. Biochim. Biophys. Acta 2011, 1807 , 510–517.21296047 (560) Sode O ; Voth GA Electron Transfer Activation of a Second Water Channel for Proton Transport in [FeFe]-Hydrogenase. J. Chem. Phys 2014, 141 , 22D527. (561) Ginovska-Pangovska B ; Ho M-H ; Linehan JC ; Cheng Y ; Dupuis M ; Raugei S ; Shaw WJ Molecular Dynamics Study of the Proposed Proton Transport Pathways in [FeFe]-Hydrogenase. Biochim. Biophys. Acta 2014, 1837 , 131–138.23981729 (562) Long H ; King PW ; Chang CH Proton Transport in Clostridium Pasteurianum [FeFe] Hydrogenase I: A Computational Study. J. Phys. Chem. B 2014, 118 , 890–900.24405487 (563) Cornish AJ ; Gärtner K ; Yang H ; Peters JW ; Hegg EL Mechanism of Proton Transfer in [FeFe]-Hydrogenase from Clostridium Pasteurianum. J. Biol. Chem 2011, 19 , 38341–38347. (564) Senger M ; Eichmann V ; Laun K ; Duan J ; Wittkamp F ; Knör G ; Apfel U-P ; Happe T ; Winkler M ; Heberle J ; How [FeFe]-Hydrogenase Facilitates Bidirectional Proton Transfer. J. Am. Chem. Soc 2019, 141 , 17394–17403.31580662 (565) Mebs S ; Duan J ; Wittkamp F ; Stripp ST ; Happe T ; Apfel U-P ; Winkler M ; Haumann M Differential Protonation at the Catalytic Six-Iron Cofactor of [FeFe]-Hydrogenases Revealed by 57 Fe Nuclear Resonance X-Ray Scattering and Quantum Mechanics/Molecular Mechanics Analyses. Inorg. Chem 2019, 58 , 4000–4013.30802044 (566) Gray HB ; Winkler JR Long-Range Electron Transfer. Proc. Natl. Acad. Sci. U. S. A 2005, 102 , 3534–3539.15738403 (567) Kaila VRI Long-Range Proton-Coupled Electron Transfer in Biological Energy Conversion: Towards Mechanistic Understanding of Respiratory Complex I. J. R. Soc. Interface 2018, 15 , 20170916.29643224 (568) Rodríguez-Maciá P ; Breuer N ; DeBeer S ; Birrell JA Insight into the Redox Behavior of the [4Fe-4S] Subcluster in [FeFe] Hydrogenases. ACS Catal 2020, 10 , 13084–13095. (569) Cohen J ; Kim K ; King P ; Seibert M ; Schulten K Finding Gas Diffusion Pathways in Proteins: Application to O2 and H2 Transport in CpI [FeFe]-Hydrogenase and the Role of Packing Defects. Structure 2005, 13 , 1321–1329.16154089 (570) Mohammadi M ; Vashisth H Pathways and Thermodynamics of Oxygen Diffusion in [FeFe]-Hydrogenase. J. Phys. Chem. B 2017, 121 , 10007.28975788 (571) Liu Y ; Mohammadi M ; Vashisth H Diffusion Network of CO in FeFe-Hydrogenase. J. Chem. Phys 2018, 149 , 204108.30501239 (572) Stapleton JA ; Swartz JR A Cell-Free Microtiter Plate Screen for Improved [FeFe] Hydrogenases. PLoS One 2010, 5 , No. e10554.20479937 (573) Stapleton JA ; Swartz JR Development of an in Vitro Compartmentalization Screen for High-Throughput Directed Evolution of [FeFe] Hydrogenases. PLoS One 2010, 5 , No. e15275 21151915 (574) Appel AM ; Bercaw JE ; Bocarsly AB ; Dobbek H ; Dubois DL ; Dupuis M ; Ferry JG ; Fujita E ; Hille R ; Kenis PJA ; Frontiers, Opportunities, and Challenges in Biochemical and Chemical Catalysis of CO2 Fixation. Chem. Rev 2013, 113 , 6621–6658.23767781 (575) Hartmann T ; Schwanhold N ; Leimkühler S Assembly and Catalysis of Molybdenum or Tungsten-Containing Formate Dehydrogenases from Bacteria. BBA - Proteins Proteomics 2015, 1854 , 1090–1100.25514355 (576) Maia LB ; Moura I ; Moura JJG Carbon Dioxide UtilisationThe Formate Route. In Enzymes for Solving Humankind’s Problems; Moura JJG , Moura I , Maia LB , Eds.; Springer Nature Switzerland: Cham, Switzerland, 2021; pp 29–81. (577) Thauer RK ; Jungermann K ; Decker K Energy Conservation in Chemotrophic Anaerobic Bacteria. Bacteriol. Rev 1977, 41 , 100–180.860983 (578) Nielsen CF ; Lange L ; Meyer AS Classification and Enzyme Kinetics of Formate Dehydrogenases for Biomanufacturing via CO2 Utilization. Biotechnol. Adv 2019, 37 , 107408.31200015 (579) Popov VO ; Lamzin VS NAD+-Dependent Formate Dehydrogenase. Biochem. J 1994, 301 , 625–643.8053888 (580) Schirwitz K ; Schmidt A ; Lamzin VS High-Resolution Structures of Formate Dehydrogenase from Candida Boidinii. Protein Sci 2007, 16 , 1146–1156.17525463 (581) Maia LB ; Moura JJ ; Moura I Molybdenum and Tungsten-Dependent Formate Dehydrogenases. J. Biol. Inorg. Chem 2015, 20 , 287–309.25476858 (582) Niks D ; Hille R Molybdenum- and Tungsten-Containing Formate Dehydrogenases and Formylmethanofuran Dehydrogenases: Structure, Mechanism, and Cofactor Insertion. Protein Sci 2019, 28 , 111–122.30120799 (583) Hille R The Mononuclear Molybdenum Enzymes. Chem. Rev 1996, 96 , 2757–2816.11848841 (584) Hille R ; Hall J ; Basu P The Mononuclear Molybdenum Enzymes. Chem. Rev 2014, 114 , 3963–4038.24467397 (585) Axley MJ ; Böck A ; Stadtman TC Catalytic Properties of an Escherichia Coli Formate Dehydrogenase Mutant in Which Sulfur Replaces Selenium. Proc. Natl. Acad. Sci. U. S. A 1991, 88 , 8450–8454.1924303 (586) Kletzin A ; Adams MW Tungsten in Biological Systems. FEMS Microbiol. Rev 1996, 18 , 5–63.8672295 (587) Johnson MK ; Rees DC ; Adams MW Tungstoenzymes. Chem. Rev 1996, 96 , 2817–2840.11848842 (588) Brondino CD ; Passeggi MC ; Caldeira J ; Almendra MJ ; Feio MJ ; Moura JJ ; Moura I Incorporation of Either Molybdenum or Tungsten into Formate Dehydrogenase from Desulfovibrio Alaskensis NCIMB 13491; EPR Assignment of the Proximal Iron-Sulfur Cluster to the Pterin Cofactor in Formate Dehydrogenases from Sulfate-Reducing Bacteria. J. Biol. Inorg. Chem 2004, 9 , 145–151.14669076 (589) Moura JJ ; Brondino CD ; Trincao J ; Romao MJ Mo and W Bis-MGD Enzymes: Nitrate Reductases and Formate Dehydrogenases. J. Biol. Inorg. Chem 2004, 9 , 791–799.15311335 (590) Almendra MJ ; Brondino CD ; Gavel O ; Pereira AS ; Tavares P ; Bursakov S ; Duarte R ; Caldeira J ; Moura JJG ; Moura I Purification and Characterization of a Tungsten-Containing Formate Dehydrogenase from Desulfovibrio Gigas. Biochemistry 1999, 38 , 16366–16372.10587462 (591) Radon C ; Mittelstädt G ; Duffus BR ; Bürger J ; Hartmann T ; Mielke T ; Teutloff C ; Leimkühler S ; Wendler P Cryo-EM Structures Reveal Intricate Fe-S Cluster Arrangement and Charging in Rhodobacter Capsulatus Formate Dehydrogenase. Nat. Commun 2020, 11 , 1912.32313256 (592) Meneghello M ; Oliveira AR ; Jacq-Bailly A ; Pereira IAC ; Léger C ; Fourmond V Formate Dehydrogenases Reduce CO2 Rather than HCO3-: An Electrochemical Demonstration. Angew. Chemie Int. Ed 2021, 60 , 9964–9967. (593) Meneghello M ; Léger C ; Fourmond V Electrochemical Studies of CO 2 -Reducing Metalloenzymes. Chem. - A Eur. J 2021, 27 , 17542–17553. (594) Axley MJ ; Grahame DA Kinetics for Formate Dehydrogenase of Escherichia Coli Formate-Hydrogenlyase. J. Biol. Chem 1991, 266 , 13731–13736.1906883 (595) da Silva SM ; Pimentel C ; Valente FMA ; Rodrigues-Pousada C ; Pereira IAC Tungsten and Molybdenum Regulation of Formate Dehydrogenase Expression in Desulfovibrio Vulgaris Hildenborough. J. Bacteriol 2011, 193 , 2909–2916.21498650 (596) Hartmann T ; Leimkühler S The Oxygen-Tolerant and NAD +-Dependent Formate Dehydrogenase from Rhodobacter Capsulatus Is Able to Catalyze the Reduction of CO2 to Formate. FEBS J 2013, 280 , 6083–6096.24034888 (597) Schmitz RA ; Richter M ; Linder D ; Thauer RK A Tungsten-Containing Active Formylmethanofuran Dehydrogenase in the Thermophilic Archaeon Methanobacterium Wolfei. Eur. J. Biochem 1992, 207 , 559–565.1633810 (598) Hartmann T ; Schrapers P ; Utesch T ; Nimtz M ; Rippers Y ; Dau H ; Mroginski MA ; Haumann M ; Leimkühler S The Molybdenum Active Site of Formate Dehydrogenase Is Capable of Catalyzing C-H Bond Cleavage and Oxygen Atom Transfer Reactions. Biochemistry 2016, 55 , 2381–2389.27054466 (599) Oliveira AR ; Mota C ; Mourato C ; Domingos RM ; Santos MFA ; Gesto D ; Guigliarelli B ; Santos-Silva T ; Romao MJ ; Pereira IAC Toward the Mechanistic Understanding of Enzymatic CO2 Reduction. ACS Catal 2020, 10 , 3844–3856. (600) Plugge CM ; Zhang W ; Scholten JC ; Stams AJ Metabolic Flexibility of Sulfate-Reducing Bacteria. Front. Microbiol 2011, 2 , 81.21734907 (601) Stams AJ ; Plugge CM Electron Transfer in Syntrophic Communities of Anaerobic Bacteria and Archaea. Nat. Rev. Microbiol 2009, 7 , 568–577.19609258 (602) Rossmann R ; Sawers G ; Böck A Mechanism of Regulation of the Formate-Hydrogenlyase Pathway by Oxygen, Nitrate, and PH: Definition of the Formate Regulon. Mol. Microbiol 1991, 5 , 2807–2814.1779767 (603) Lamont CM ; Kelly CL ; Pinske C ; Buchanan G ; Palmer T ; Sargent F Expanding the Substrates for a Bacterial Hydrogenlyase Reaction. Microbiology 2017, 163 , 649–653.28488566 (604) Finney AJ ; Lowden R ; Fleszar M ; Albareda M ; Coulthurst SJ ; Sargent F The Plant Pathogen Pectobacterium Atrosepticum Contains a Functional Formate Hydrogenlyase-2 Complex. Mol. Microbiol 2019, 112 , 1440–1452.31420965 (605) Ramio-Pujol S ; Ganigue R ; Baneras L ; Colprim J Impact of Formate on the Growth and Productivity of Clostridium Ljungdahlii PETC and Clostridium Carboxidivorans P7 Grown on Syngas. Int. Microbiol 2014, 17 , 195–204.26421736 (606) Sawers G ; Suppmann B Anaerobic Induction of Pyruvate Formate-Lyase Gene Expression Is Mediated by the ArcA and FNR Proteins. J. Bacteriol 1992, 174 , 3474–3478.1592804 (607) Sawers G The Hydrogenases and Formate Dehydrogenases of Escherichia Coli. Antonie Van Leeuwenhoek 1994, 66 , 57–88.7747941 (608) Schwarz FM ; Schuchmann K ; Müller V Hydrogenation of CO2 at Ambient Pressure Catalyzed by a Highly Active Thermostable Biocatalyst. Biotechnol. Biofuels 2018, 11 , 237.30186365 (609) Vorholt JA ; Thauer RK Molybdenum and Tungsten Enzymes in C1 Metabolism. In Metals Ions in Biological System: Vol. 39: Molybdenum and Tungsten: Their Roles in Biological Processes; Sigel A , Sigel H , Eds.; CRC Press: Boca Raton, FL, 2002; Vol. 39 , pp 571–619. (610) Thauer RK Biochemistry of Methanogenesis: A Tribute to Marjory Stephenson:1998 Marjory Stephenson Prize Lecture. Microbiology 1998, 144 , 2377–2406.9782487 (611) Börner G ; Karrasch M ; Thauer RK Formylmethanofuran Dehydrogenase-Activity in Cell-Extracts of Methanobacterium-Thermoautotrophicum and of Methanosarcina-Barkeri. Febs Lett 1989, 244 , 21–25. (612) Wagner T ; Ermler U ; Shima S The Methanogenic CO2 Reducing-and-Fixing Enzyme Is Bifunctional and Contains 46 [4Fe-4S] Clusters. Science 2016, 354 , 114–117.27846502 (613) Welte C ; Kröninger L ; Deppenmeier U Experimental Evidence of an Acetate Transporter Protein and Characterization of Acetate Activation in Aceticlastic Methanogenesis of Methanosarcina Mazei. FEMS Microbiol. Lett 2014, 359 , 147–153.25088360 (614) Kurth JM ; Op den Camp HJM ; Welte CU Several Ways One Goal—Methanogenesis from Unconventional Substrates. Appl. Microbiol. Biotechnol 2020, 104 , 6839–6854.32542472 (615) Schmitz RA ; Albracht SP ; Thauer RK A Molybdenum and a Tungsten Isoenzyme of Formylmethanofuran Dehydrogenase in the Thermophilic Archaeon Methanobacterium Wolfei. Eur. J. Biochem 1992, 209 , 1013–1018.1330558 (616) Schmitz RA ; Albracht SPJ ; Thauer RK Properties of the Tungsten-Substituted Molybdenum Formylmethanofuran Dehydrogenase from Methanobacterium Wolfei. Febs Lett 1992, 309 , 78–81.1324851 (617) Bertram PA ; Karrasch M ; Schmitz RA ; Bocher R ; Albracht SPJ ; Thauer RK Formylmethanofuran Dehydrogenases from Methanogenic Archaea: Substrate Specificity, EPR Properties and Reversible Inactivation by Cyanide of the Molybdenum or Tungsten Iron-Sulfur Proteins. Eur. J. Biochem 1994, 220 , 477–484.8125106 (618) Bertram PA ; Schmitz RA ; Linder D ; Thauer RK Tungstate Can Substitute for Molybdate in Sustaining Growth of Methanobacterium Thermoautotrophicum. Identification and Characterization of a Tungsten Isoenzyme of Formylmethanofuran Dehydrogenase. Arch. Microbiol 1994, 161 , 220–228.8161283 (619) Schmitz RA ; Bertram PA ; Thauer RK Tungstate Does Not Support Synthesis of Active Formylmethanofuran Dehydrogenase in Methanosarcina Barkeri. Arch. Microbiol 1994, 161 , 528–530. (620) Hochheimer I ; Schmitz RA ; Thauer RK ; Hedderich R The Tungsten Formylmethanofuran Dehydrogenase from Methanobacterium Thermoautotrophicum Contains Sequence Motifs Characteristic for Enzymes Containing Molybdopterin Dinucleotide. Eur. J. Biochem 1995, 234 , 910–920.8575452 (621) Bertram PA ; Thauer RK Thermodynamics of the Formylmethanofuran Dehydrogenase Reaction in Methanobacterium Thermoautotrophicum. Eur. J. Biochem 1994, 226 , 811–818.7813470 (622) Hochheimer A ; Hedderich R ; Thauer RK The Formylmethanofuran Dehydrogenase Isoenzymes in Methanobacterium Wolfei and Methanobacterium Thermoautotrophicum: Induction of the Molybdenum Isoenzyme by Molybdate and Constitutive Synthesis of the Tungsten Isoenzyme. Arch. Microbiol 1998, 170 , 389–393.9818358 (623) Hochheimer A ; Linder D ; Thauer RK ; Hedderich R The Molybdenum Formylmethanofuran Dehydrogenase Operon and the Tungsten Formylmethanofuran Dehydrogenase Operon from Methanobacterium Thermoautotrophicum - Structures and Transcriptional Regulation. Eur. J. Biochem 1996, 242 , 156–162.8954165 (624) Vorholt JA ; Vaupel M ; Thauer RK A Selenium-Dependent and a Selenium-Independent Formylmethanofuran Dehydrogenase and Their Transcriptional Regulation in the Hyperthermophilic Methanopyrus Kandleri. Mol. Microbiol 1997, 23 , 1033–1042.9076739 (625) Börner G ; Karrasch M ; Thauer RK Molybdopterin Adenine Dinucleotide and Molybdopterin Hypoxanthine Dinucleotide in Formylmethanofuran Dehydrogenase from Methanobacterium Thermoautotrophicum (Marburg). FEBS Lett 1991, 290 , 31–34.1915887 (626) Axley MJ ; Grahame DA ; Stadtman TC Escherichia Coli Formate-Hydrogen Lyase. Purification and Properties of the Selenium-Dependent Formate Dehydrogenase Component. J. Biol. Chem 1990, 265 , 18213–18218.2211698 (627) Boyington JC ; Gladyshev VN ; Khangulov SV ; Stadtman TC ; Sun PD Crystal Structure of Formate Dehydrogenase H: Catalysis Involving Mo, Molybdopterin, Selenocysteine, and an Fe4S4 Cluster. Science 1997, 275 , 1305–1308.9036855 (628) Gladyshev VN ; Boyington JC ; Khangulov SV ; Grahame DA ; Stadtman TC ; Sun PD Characterization of Crystalline Formate Dehydrogenase H from Escherichia Coli. Stabilization, EPR Spectroscopy, and Preliminary Crystallographic Analysis. J. Biol. Chem 1996, 271 , 8095–8100.8626495 (629) Gladyshev VN ; Khangulov SV ; Axley MJ ; Stadtman TC Coordination of Selenium to Molybdenum in Formate Dehydrogenase H from Escherichia Coli. Proc. Natl. Acad. Sci. U. S. A 1994, 91 , 7708–7711.8052647 (630) Enoch HG ; Lester RL The Purification and Properties of Formate Dehydrogenase and Nitrate Reductase from Escherichia Coli. J. Biol. Chem 1975, 250 , 6693–6705.1099093 (631) Oh JI ; Bowien B Structural Analysis of the Fds Operon Encoding the NAD+-Linked Formate Dehydrogenase of Ralstonia Eutropha. J. Biol. Chem 1998, 273 , 26349–26360.9756865 (632) Niks D ; Duvvuru J ; Escalona M ; Hille R Spectroscopic and Kinetic Properties of the Molybdenum-Containing, NAD+-Dependent Formate Dehydrogenase from Ralstonia Eutropha. J. Biol. Chem 2016, 291 , 1162–1174.26553877 (633) Riederer-Henderson MA ; Peck HD In Vitro Requirements for Formate Dehydrogenase Activity from Desulfovibrio. Can. J. Microbiol 1986, 32 , 425–429. (634) Raaijmakers H ; Macieira S ; Dias JM ; Teixeira S ; Bursakov S ; Huber R ; Moura JJG ; Moura I ; Romao MJ Gene Sequence and the 1.8 A Crystal Structure of the Tungsten-Containing Formate Dehydrogenase from Desulfovibrio Gigas. Structure 2002, 10 , 1261–1272.12220497 (635) Raaijmakers H ; Teixeira S ; Dias JM ; Almendra MJ ; Brondino CD ; Moura I ; Moura JJ ; Romao MJ Tungsten-Containing Formate Dehydrogenase from Desulfovibrio Gigas: Metal Identification and Preliminary Structural Data by Multi-Wavelength Crystallography. J. Biol. Chem 2001, 6 , 398–404. (636) Riederer-Henderson MA ; Peck HD Properties of Formate Dehydrogenase from Desulfovibrio Gigas. Can. J. Microbiol 1986, 32 , 430–435. (637) Laukel M ; Chistoserdova L ; Lidstrom ME ; Vorholt JA The Tungsten-Containing Formate Dehydrogenase from Methylobacterium Extorquens AM1: Purification and Properties. Eur. J. Biochem 2003, 270 , 325–333.12605683 (638) Duffus BR ; Schrapers P ; Schuth N ; Mebs S ; Dau H ; Leimkühler S ; Haumann M Anion Binding and Oxidative Modification at the Molybdenum Cofactor of Formate Dehydrogenase from Rhodobacter Capsulatus Studied by X-Ray Absorption Spectroscopy. Inorg. Chem 2020, 59 , 214–225.31814403 (639) Schrapers P ; Hartmann T ; Kositzki R ; Dau H ; Reschke S ; Schulzke C ; Leimkühler S ; Haumann M Sulfido and Cysteine Ligation Changes at the Molybdenum Cofactor during Substrate Conversion by Formate Dehydrogenase (FDH) from Rhodobacter Capsulatus. Inorg. Chem 2015, 54 , 3260–3271.25803130 (640) Iobbi-Nivol C ; Leimkühler S Molybdenum Enzymes, Their Maturation and Molybdenum Cofactor Biosynthesis in Escherichia Coli. Biochim. Biophys. Acta 2013, 1827 , 1086–1101.23201473 (641) Abaibou H ; Pommier J ; Benoit S ; Giordano G ; Mandrand-Berthelot MA Expression and Characterization of the Escherichia Coli Fdo Locus and a Possible Physiological Role for Aerobic Formate Dehydrogenase. J. Bacteriol 1995, 177 , 7141–7149.8522521 (642) Thomé R ; Gust A ; Toci R ; Mendel R ; Bittner F ; Magalon A ; Walburger A A Sulfurtransferase Is Essential for Activity of Formate Dehydrogenases in Escherichia Coli. J. Biol. Chem 2012, 287 , 4671–4678.22194618 (643) Arnoux P ; Ruppelt C ; Oudouhou F ; Lavergne J ; Siponen MI ; Toci R ; Mendel RR ; Bittner F ; Pignol D ; Magalon A ; Sulphur Shuttling across a Chaperone during Molybdenum Cofactor Maturation. Nat. Commun 2015, 6 , 6148.25649206 (644) Böhmer N ; Hartmann T ; Leimkühler S The Chaperone FdsC for Rhodobacter Capsulatus Formate Dehydrogenase Binds the Bis-Molybdopterin Guanine Dinucleotide Cofactor. FEBS Lett 2014, 588 , 531–537.24444607 (645) Schwanhold N ; Iobbi-Nivol C ; Lehmann A ; Leimkühler S Same but Different: Comparison of Two System-Specific Molecular Chaperones for the Maturation of Formate Dehydrogenases. PLoS One 2018, 13 , No. e0201935.30444874 (646) Raaijmakers HC ; Romao MJ Formate-Reduced E. Coli Formate Dehydrogenase H: The Reinterpretation of the Crystal Structure Suggests a New Reaction Mechanism. J. Biol. Inorg. Chem 2006, 11 , 849–854.16830149 (647) Khangulov SV ; Gladyshev VN ; Dismukes GC ; Stadtman TC Selenium-Containing Formate Dehydrogenase H from Escherichia Coli: A Molybdopterin Enzyme That Catalyzes Formate Oxidation without Oxygen Transfer. Biochemistry 1998, 37 , 3518–3528.9521673 (648) Rivas MG ; Gonzalez PJ ; Brondino CD ; Moura JJ ; Moura I EPR Characterization of the Molybdenum(V) Forms of Formate Dehydrogenase from Desulfovibrio Desulfuricans ATCC 27774 upon Formate Reduction. J. Inorg. Biochem 2007, 101 , 1617–1622.17574676 (649) Barber MJ ; Siegel LM ; Schauer NL ; May HD ; Ferry JG Formate Dehydrogenase from Methanobacterium Formicicum: Electron Paramagnetic Resonance Spectroscopy of the Molybdenum and Iron-Sulfur Centers. J. Biol. Chem 1983, 258 , 839–845. (650) Barber MJ ; May HD ; Ferry JG Inactivation of Formate Dehydrogenase from Methanobacterium Formicicum by Cyanide. Biochemistry 1986, 25 , 8150–8155. (651) Malthouse JP ; Gutteridge S ; Bray RC Rapid Type 2 Molybdenum(V) Electron-Paramagnetic Resonance Signals from Xanthine Oxidase and the Structure of the Active Centre of the Enzyme. Biochem. J 1980, 185 , 767–770.6248034 (652) Thauer RK ; Fuchs G ; Käufer B Reduced Ferredoxin: CO2 Oxidoreductase from Clostridium Pasteurianum. Effect of Ligands to Transition Metals on the Activity and the Stability of the Enzyme. Hoppe Seylers Zeitschrift für Physiol. Chemie 1975, 356 , 653–662. (653) Mota CS ; Rivas MG ; Brondino CD ; Moura I ; Moura JJ ; Gonzalez PJ ; Cerqueira NM The Mechanism of Formate Oxidation by Metal-Dependent Formate Dehydrogenases. J. Biol. Inorg. Chem 2011, 16 , 1255–1268.21773834 (654) Pinsent J The Need for Selenite and Molybdate in the Formation of Formic Dehydrogenase by Members of the Coli-Aerogenes Group of Bacteria. Biochem. J 1954, 57 , 10–16.13159942 (655) Zinoni F ; Birkmann A ; Leinfelder W ; Böck A Cotranslational Insertion of Selenocysteine into Formate Dehydrogenase from Escherichia-Coli Directed by a Uga Codon. Proc. Natl. Acad. Sci. U. S. A 1987, 84 , 3156–3160.3033637 (656) Reich HJ ; Hondal RJ Why Nature Chose Selenium. ACS Chem. Biol 2016, 11 , 821–841.26949981 (657) George GN ; Colangelo CM ; Dong J ; Scott RA ; Khangulov SV ; Gladyshev VN ; Stadtman TC X-Ray Absorption Spectroscopy of the Molybdenum Site of Escherichia Coli Formate Dehydrogenase. J. Am. Chem. Soc 1998, 120 , 1267–1273. (658) George GN ; Costa C ; Moura JG ; Moura I Observation of Ligand-Based Redox Chemistry at the Active Site of a Molybdenum Enzyme. J. Am. Chem. Soc 1999, 121 , 2625–2626. (659) Heider J ; Bock A Selenium Metabolism in Micro-Organisms. Adv. Microb. Physiol 1993, 35 , 71–109.8310883 (660) Jormakka M ; Törnroth S ; Byrne B ; Iwata S Molecular Basis of Proton Motive Force Generation: Structure of Formate Dehydrogenase-N. Science 2002, 295 , 1863–1868.11884747 (661) Jollie DR ; Lipscomb JD Formate Dehydrogenase from Methylosinus Trichosporium Ob3b: Purification and Spectroscopic Characterization of the Cofactors. J. Biol. Chem 1991, 266 , 21853–21863.1657982 (662) Friedebold J ; Bowien B Physiological and Biochemical Characterization of the Soluble Formate Dehydrogenase, a Molybdoenzyme from Alcaligenes Eutrophus. J. Bacteriol 1993, 175 , 4719–4728.8335630 (663) Yamamoto I ; Saiki T ; Liu S-M ; Ljungdahl LG Purification and Properties of NADP-Dependent Formate Dehydrogenase from Clostridium Thermoaceticum, a Tungsten-Selenium-Iron Protein. J. Biol. Chem 1983, 258 , 1826–1832.6822536 (664) Schauer NL ; Ferry JG Properties of Formate Dehydrogenase in Methanobacterium Formicicum. J. Bacteriol 1982, 150 , 1–7.7061389 (665) Robinson WE ; Bassegoda A ; Reisner E ; Hirst J Oxidation-State-Dependent Binding Properties of the Active Site in a Mo-Containing Formate Dehydrogenase. J. Am. Chem. Soc 2017, 139 , 9927–9936.28635274 (666) Thauer RK ; Käufer B ; Fuchs G The Active Species of “CO2” Utilized by Reduced Ferredoxin: CO2 Oxidoreductase from Clostridium Pasteurianum. Eur. J. Biochem 1975, 55 , 111–117.240689 (667) Cooper TG ; Tchen TT ; Wood HG ; Benedict CR The Carboxylation of Phosphoenolpyruvate and Pyruvate. I. The Active Species of “CO2” Utilized by Phosphoenolpyruvate Carboxykinase, Carboxytransphosphorylase, and Pyruvate Carboxylase. J. Biol. Chem 1968, 243 , 3857–3863.5661712 (668) Chan MK ; Mukund S ; Kletzin A ; Adams MWW ; Rees DC Structure of a Hyperthermophilic Tungstopterin Enzyme, Aldehyde Ferredoxin Oxidoreductase. Science 1995, 267 , 1463–1469.7878465 (669) Tiberti M ; Papaleo E ; Russo N ; De Gioia L ; Zampella G Evidence for the Formation of a Mo - H Intermediate in the Catalytic Cycle of Formate Dehydrogenase. Inorg. Chem 2012, 51 , 8331–8339.22800191 (670) Dong G ; Ryde U Reaction Mechanism of Formate Dehydrogenase Studied by Computational Methods. J. Biol. Inorg. Chem 2018, 23 , 1243–1254.30173398 (671) Niks D ; Hille R Reductive Activation of CO2 by Formate Dehydrogenases. Methods Enzymol 2018, 613 , 277–295.30509470 (672) Najmudin S ; Gonzalez PJ ; Trincao J ; Coelho C ; Mukhopadhyay A ; Cerqueira NM ; Romao CC ; Moura I ; Moura JJ ; Brondino CD ; Periplasmic Nitrate Reductase Revisited: A Sulfur Atom Completes the Sixth Coordination of the Catalytic Molybdenum. J. Biol. Inorg. Chem 2008, 13 , 737–753.18327621 (673) Hille R ; Dingwall S ; Wilcoxen J The Aerobic CO Dehydrogenase from Oligotropha Carboxidovorans. J. Biol. Inorg. Chem 2015, 20 , 243–251.25156151 (674) Kaufmann P ; Duffus BR ; Teutloff C ; Leimkühler S Functional Studies on Oligotropha Carboxidovorans Molybdenum-Copper CO Dehydrogenase Produced in Escherichia Coli. Biochemistry 2018, 57 , 2889–2901.29676148 (675) Can M ; Armstrong FA ; Ragsdale SW Structure, Function, and Mechanism of the Nickel Metalloenzymes, CO Dehydrogenase, and Acetyl-CoA Synthase. Chem. Rev 2014, 114 , 4149–4174.24521136 (676) Wu M ; Ren Q ; Durkin AS ; Daugherty SC ; Brinkac LM ; Dodson RJ ; Madupu R ; Sullivan SA ; Kolonay JF ; Nelson WC ; Life in Hot Carbon Monoxide: The Complete Genome Sequence of Carboxydothermus Hydrogenoformans Z-2901. PLoS Genet 2005, 1 , No. e65.16311624 (677) Wittenborn EC ; Merrouch M ; Ueda C ; Fradale L ; Léger C ; Fourmond V ; Pandelia M-E ; Dementin S ; Drennan CL Redox-Dependent Rearrangements of the NiFeS Cluster of Carbon Monoxide Dehydrogenase. Elife 2018, 7 , No. e39451.30277213 (678) Ragsdale SW ; Pierce E Acetogenesis and the Wood-Ljungdahl Pathway of CO2 Fixation. Biochim. Biophys. Acta - Proteins Proteomics 2008, 1784 , 1873–1898. (679) Gong W ; Hao B ; Wei Z ; Ferguson DJ ; Tallant T ; Krzycki JA ; Chan MK Structure of the A2ε2 Ni-Dependent CO Dehydrogenase Component of the Methanosarcina Barkeri Acetyl-CoA Decarbonylase/Synthase Complex. Proc. Natl. Acad. Sci. U. S. A 2008, 105 , 9558–9563.18621675 (680) Doukov TI ; Iverson TM ; Seravalli J ; Ragsdale SW ; Drennan CL A Ni-Fe-Cu Center in a Bifunctional Carbon Monoxide Dehydrogenase/Acetyl-CoA Synthase. Science 2002, 298 , 567–572.12386327 (681) Domnik L ; Merrouch M ; Goetzl S ; Jeoung JH ; Léger C ; Dementin S ; Fourmond V ; Dobbek H CODH-IV: A High-Efficiency CO-Scavenging CO Dehydrogenase with Resistance to O2. Angew. Chem. Int. Ed 2017, 56 , 15466–15469. (682) Jeoung J-H ; Dobbek H Carbon Dioxide Activation at the Ni,Fe-Cluster of Anaerobic Carbon Monoxide Dehydrogenase. Science 2007, 318 , 1461–1464.18048691 (683) Inoue M ; Nakamoto I ; Omae K ; Oguro T ; Ogata H ; Yoshida T ; Sako Y Structural and Phylogenetic Diversity of Anaerobic Carbon-Monoxide Dehydrogenases. Front. Microbiol 2019, 9 , 3353.30705673 (684) Kim EJ ; Feng J ; Bramlett MR ; Lindahl PA Evidence for a Proton Transfer Network and a Required Persulfide-Bond-Forming Cysteine Residue in Ni-Containing Carbon Monoxide Dehydrogenases. Biochemistry 2004, 43 , 5728–5734.15134447 (685) Wittenborn EC ; Cohen SE ; Merrouch M ; Léger C ; Fourmond V ; Dementin S ; Drennan CL Structural Insight into Metallocofactor Maturation in Carbon Monoxide Dehydrogenase. J. Biol. Chem 2019, 294 , 13017–13026.31296570 (686) Merrouch M ; Hadj-Saïd J ; Domnik L ; Dobbek H ; Léger C ; Dementin S ; Fourmond V O2 Inhibition of Ni-Containing CO Dehydrogenase Is Partly Reversible. Chem. - A Eur. J 2015, 21 , 18934–18938. (687) Wittenborn EC ; Guendon C ; Merrouch M ; Benvenuti M ; Fourmond V ; Léger C ; Drennan CL ; Dementin S The Solvent-Exposed Fe-S D-Cluster Contributes to Oxygen-Resistance in Desulfovibrio Vulgaris Ni-Fe Carbon Monoxide Dehydrogenase. ACS Catal 2020, 10 , 7328–7335.32655979 (688) Inoue T ; Takao K ; Yoshida T ; Wada K ; Daifuku T ; Yoneda Y ; Fukuyama K ; Sako Y Cysteine 295 Indirectly Affects Ni Coordination of Carbon Monoxide Dehydrogenase-II C-Cluster. Biochem. Biophys. Res. Commun 2013, 441 , 13–17.24120497 (689) Jeon WB ; Singer SW ; Ludden PW ; Rubio LM New Insights into the Mechanism of Nickel Insertion into Carbon Monoxide Dehydrogenase: Analysis of Rhodospirillum Rubrum Carbon Monoxide Dehydrogenase Variants with Substituted Ligands to the [Fe3S4] Portion of the Active-Site C-Cluster. J. Biol. Inorg. Chem 2005, 10 , 903–912.16283394 (690) Fraser DM ; Lindahl PA Evidence for a Proposed Intermediate Redox State in the CO/CO2 Active Site of Acetyl-CoA Synthase (Carbon Monoxide Dehydrogenase) from Clostridium Thermoaceticum. Biochemistry 1999, 38 , 15706–15711.10625436 (691) Lindahl PA Implications of a Carboxylate-Bound C-Cluster Structure of Carbon Monoxide Dehydrogenase. Angew. Chem. Int. Ed 2008, 47 , 4054–4056. (692) Amara P ; Mouesca J-M ; Volbeda A ; Fontecilla-Camps JC Carbon Monoxide Dehydrogenase Reaction Mechanism: A Likely Case of Abnormal CO 2 Insertion to a Ni-H - Bond. Inorg. Chem 2011, 50 , 1868–1878.21247090 (693) Yang JY ; Kerr TA ; Wang XS ; Barlow JM Reducing CO2 to HCO2- at Mild Potentials: Lessons from Formate Dehydrogenase. J. Am. Chem. Soc 2020, 142 , 19438–19445.33141560 (694) Breglia R ; Arrigoni F ; Sensi M ; Greco C ; Fantucci P ; De Gioia L ; Bruschi M First-Principles Calculations on Ni,Fe-Containing Carbon Monoxide Dehydrogenases Reveal Key Stereoelectronic Features for Binding and Release of CO2to/from the C-Cluster. Inorg. Chem 2021, 60 , 387–402.33321036 (695) Maynard EL ; Lindahl PA Evidence of a Molecular Tunnel Connecting the Active Sites for CO2 Reduction and Acetyl-CoA Synthesis in Acetyl-CoA Synthase from Clostridium Thermoaceticum. J. Am. Chem. Soc 1999, 121 , 9221–9222. (696) Seravalli J ; Ragsdale SW Channeling of Carbon Monoxide during Anaerobic Carbon Dioxide Fixation. Biochemistry 2000, 39 , 1274–1277.10684606 (697) Darnault C ; Volbeda A ; Kim EJ ; Legrand P ; Vernède X ; Lindahl PA ; Fontecilla-Camps JC Ni-Zn-[Fe4-S4] and Ni-Ni-[Fe4-S4] Clusters in Closed and Open α Subunits of Acetyl-CoA Synthase/Carbon Monoxide Dehydrogenase. Nat. Struct. Biol 2003, 10 , 271–279.12627225 (698) Doukov TI ; Blasiak LC ; Seravalli J ; Ragsdale SW ; Drennan CL Xenon in and at the End of the Tunnel of Bifunctional Carbon Monoxide Dehydrogenase/Acetyl-CoA Synthase. Biochemistry 2008, 47 , 3474–3483.18293927 (699) Volbeda A ; Fontecilla-Camps JC Crystallographic Evidence for a CO/CO2 Tunnel Gating Mechanism in the Bifunctional Carbon Monoxide Dehydrogenase/Acetyl Coenzyme A Synthase from Moorella Thermoacetica. J. Biol. Inorg. Chem 2004, 9 , 525–532.15221479 (700) Tan X ; Loke HK ; Fitch S ; Lindahl PA The Tunnel of Acetyl-Coenzyme A Synthase/Carbon Monoxide Dehydrogenase Regulates Delivery of CO to the Active Site. J. Am. Chem. Soc 2005, 127 , 5833–5839.15839681 (701) Tan X ; Lindahl PA Tunnel Mutagenesis and Ni-Dependent Reduction and Methylation of the α Subunit of Acetyl Coenzyme A Synthase/Carbon Monoxide Dehydrogenase. J. Biol. Inorg. Chem 2008, 13 , 771–778.18365259 (702) Biester A ; Dementin S ; Drennan CL Visualizing the Gas Channel of a Monofunctional Carbon Monoxide Dehydrogenase. J. Inorg. Biochem 2022, 230 , 111774.35278753 (703) Tan X ; Volbeda A ; Fontecilla-Camps JC ; Lindahl PA Function of the Tunnel in Acetylcoenzyme A Synthase/Carbon Monoxide Dehydrogenase. J. Biol. Inorg. Chem 2006, 11 , 371–378.16502006 (704) Wang PH ; Bruschi M ; De Gioia L ; Blumberger J Uncovering a Dynamically Formed Substrate Access Tunnel in Carbon Monoxide Dehydrogenase/Acetyl-CoA Synthase. J. Am. Chem. Soc 2013, 135 , 9493–9502.23713976 (705) Dobbek H ; Svetlitchnyi V ; Gremer L ; Huber R ; Meyer O Crystal Structure of a Carbon Monoxide Dehydrogenase Reveals a [Ni-4Fe-5S] Cluster. Science 2001, 293 , 1281–1285.11509720 (706) Drennan CL ; Heo J ; Sintchak MD ; Schreiter E ; Ludden PW Life on Carbon Monoxide: X-Ray Structure of Rhodospirillum Rubrum Ni-Fe-S Carbon Monoxide Dehydrogenase. Proc. Natl. Acad. Sci. U. S. A 2001, 98 , 11973–11978.11593006 (707) Ensign SA ; Hyman MR ; Ludden PW Nickel-Specific, Slow-Binding Inhibition of Carbon Monoxide Dehydrogenase from Rhodospirillum Rubrum by Cyanide. Biochemistry 1989, 28 , 4973–4979.2504285 (708) Hadj-Saïd J ; Pandelia ME ; Léger C ; Fourmond V ; Dementin S The Carbon Monoxide Dehydrogenase from Desulfovibrio Vulgaris. Biochim. Biophys. Acta - Bioenerg 2015, 1847 , 1574–1583. (709) Benvenuti M ; Meneghello M ; Guendon C ; Jacq-Bailly A ; Jeoung J-H ; Dobbek H ; Léger C ; Fourmond V ; Dementin S The Two CO-Dehydrogenases of Thermococcus Sp. AM4. Biochim. Biophys. Acta - Bioenerg 2020, 1861 , 148188.32209322 (710) Parkin A ; Seravalli J ; Vincent KA ; Ragsdale SW ; Armstrong FA Rapid and Efficient Electrocatalytic CO2/CO Interconversions by Carboxydothermus Hydrogenoformans CO Dehydrogenase I on an Electrode. J. Am. Chem. Soc 2007, 129 , 10328–10329.17672466 (711) Ensign SA Reactivity of Carbon Monoxide Dehydrogenase from Rhodospirillum Rubrum with Carbon Dioxide, Carbonyl Sulfide, and Carbon Disulfide. Biochemistry 1995, 34 , 5372–5381.7727395 (712) Bonam D ; Murrell SA ; Ludden PW Carbon Monoxide Dehydrogenase from Rhodospirillum Rubrum. J. Bacteriol 1984, 159 , 693–699.6430875 (713) Snider MG ; Temple BS ; Wolfenden R The Path to the Transition State in Enzyme Reactions: A Survey of Catalytic Efficiencies. J. Phys. Org. Chem 2004, 17 , 586–591. (714) Prangé T ; Schiltz M ; Pernot L ; Colloc’h N ; Longhi S ; Bourguet W ; Fourme R Exploring Hydrophobic Sites in Proteins With Xenon or Krypton. Proteins Struct. Funct. Genet 1998, 30 , 61–73.9443341 (715) Richards FM Areas, Volumes, Packing and Protein Structure. Annu. Rev. Biophys. Bioeng 1977, 6 , 151–176 June.326146 (716) Plumeré N ; Rüdiger O ; Oughli AA ; Williams R ; Vivekananthan J ; Pöller S ; Schuhmann W ; Lubitz W A Redox Hydrogel Protects Hydrogenase from High-Potential Deactivation and Oxygen Damage. Nat. Chem 2014, 6 , 822–827.25143219 (717) Li H ; Buesen D ; Dementin S ; Léger C ; Fourmond V ; Plumeré N Complete Protection of O2-Sensitive Catalysts in Thin Films. J. Am. Chem. Soc 2019, 141 , 16734–16742.31525046 (718) Hardt S ; Stapf S ; Filmon DT ; Birrell JA ; Rüdiger O ; Fourmond V ; Léger C ; Plumeré N Reversible H2 Oxidation and Evolution by Hydrogenase Embedded in a Redox Polymer Film. Nat. Catal 2021, 4 , 251–258.33842839 (719) Wraight CA Chance and Design-Proton Transfer in Water, Channels and Bioenergetic Proteins. Biochim. Biophys. Acta - Bioenerg 2006, 1757 , 886–912. (720) Gray HB ; Winkler JR Electron Tunneling through Proteins. Q. Rev. Biophys 2003, 36 , 341–372.15029828 (721) Feng X ; Schut GJ ; Haja DK ; Adams MWW ; Li H Structure and Electron Transfer Pathways of an Electron-Bifurcating NiFe-Hydrogenase. Sci. Adv 2022, 8 , 1–14. (722) Oughli AA ; Ruff A ; Boralugodage NP ; Rodríguez-Maciá P ; Plumeré N ; Lubitz W ; Shaw WJ ; Schuhmann W ; Rüdiger O Dual Properties of a Hydrogen Oxidation Ni-Catalyst Entrapped within a Polymer Promote Self-Defense against Oxygen. Nat. Commun 2018, 9 , 864.29491416 (723) Lee CC ; Kang W ; Jasniewski AJ ; Stiebritz MT ; Tanifuji K ; Ribbe MW ; Hu Y Evidence of Substrate Binding and Product Release via Belt-Sulfur Mobilization of the Nitrogenase Cofactor. Nat. Catal 2022, 5 , 443–454. (724) Zuchan K ; Baymann F ; Baffert C ; Brugna M ; Nitschke W The dyad of the Y-junction- and a flavin module unites diverse redox enzymes. Biochimica et Biophysica Acta (BBA) - Bioenergetics 2021, 1862 (6 ), 148401.33684340

PMC009xxxxxx/PMC9555301.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0413066 2830 Cell Cell Cell 0092-8674 1097-4172 35835100 9555301 10.1016/j.cell.2022.06.035 NIHMS1822511 Article Cancer Vaccines: Building a Bridge over Troubled Waters Sellars MacLean C. 1 Wu Catherine J. 1234 Fritsch Edward F. 13 1. Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA. 2. Harvard Medical School, Boston, MA, USA. 3. Broad Institute of MIT and Harvard, Cambridge, MA, USA. 4. Department of Medicine, Brigham and Women’s Hospital, Boston, MA, USA. Contact Info: Catherine J. Wu. CWU@Partners.org; Edward_Fritsch@dfci.harvard.edu 15 7 2022 21 7 2022 13 7 2022 21 7 2023 185 15 27702788 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Cancer vaccines aim to direct the immune system to eradicate cancer cells. We review the essential immunologic concepts underpinning natural immunity, and highlight the multiple unique challenges faced by vaccines targeting cancer. Recent technological advances in mass spectrometry, neoantigen prediction, genetically and pharmacologically engineered mouse models, and single cell ‘omics have revealed new biology which can help to bridge this divide. We particularly focus on translationally relevant aspects, such as antigen selection and delivery and monitoring human post-vaccination responses, and encourage more aggressive exploration of novel approaches. ETOC Blurb For many decades, vaccines have provided effective protection against a variety of infectious diseases. Attention is now focussing on utilising the power of the immune system to target tumours through the development of cancer vaccines. This review provides a background in how the immune system responds to cancer, how these immune responses can be harnessed for cancer vaccines and highlights some of the remaining challenges. pmcIntroduction The goal for cancer vaccines is obvious – apply what has been done for a few infectious diseases and focus the power of human immunity on eliminating tumor cells. There are encouraging signs that this aim is achievable, but success remains elusive. Immunity has evolved over millennia to effectively respond to and prevent future pathogen attack, providing a well-tested template for vaccines for infectious diseases. However, tumors are not acute foreign events, but rather evolve from and within the host and are only apparent clinically after escape from immune pressure that controls incipient neoplasms (Dunn et al., 2002). Moreover, surviving tumor cells accumulate intrinsic defenses and co-opt other cells, including immune cells, to support escape. The resulting complex ecosystem of the immunosuppressive microenvironment impacts both innate and adaptive immune cells, can extend beyond the tumor border (van den Hout et al., 2017) and can even alter newly infiltrated cells (Li et al., 2022). Thus, therapeutic cancer vaccines, meant to treat that extant disease, face distinct challenges from prophylactic vaccines for infectious pathogens. Undeniably a disease of mutation, tumor possess enormous molecular diversity and exquisite individuality; cancer is thus not one or a hundred, but thousands of diseases. Those mutations also bring opportunity: novel non-self-protein targets for immune recognition (‘neoantigens’). However, the vastness of that mutational landscape makes target identification a far more difficult task than for foreign infectious pathogens with relatively low genetic complexity. Herein, we review the essential immunologic concepts underpinning productive immune responses and highlight the unique demands faced by vaccines targeting cancer. Numerous recent technological advances – from mass spectrometry and machine learning in neoantigen prediction, to genetically and pharmacologically engineered mouse models, to advances in single cell transcriptomics and genomics – have revealed new biology and point to how to bridge this divide. We particularly focus on those aspects that are translationally relevant, such as antigen selection and delivery, as well as unique challenges, like the immunosuppressive tumor immune microenvironment (TIME) and how to gather evidence for vaccine efficacy. Natural Immunity – The starting point Vaccines are pharmacologic manipulations of our immune system designed to achieve the physiological effect of defense and protection from a pathogen. Hence, a logical starting point is natural immunity, which has been ‘educated’ over eons by endless challenges from extant and emergent pathogens. Here, we review our current understanding of the contributing key molecular and cellular players and how these principles can be applied to cancer vaccines. It takes a village Perhaps the most remarkable feature of natural immunity is its ability parlay the initial events of pathogen infection into a multi-pronged, target-agnostic and target-specific attack (Figure 1). At the start, many cell types, but particularly innate immune myeloid cells, use pattern recognition receptors (PRR) to sense pathogen-derived foreign materials as well as byproducts from damaged tissue. These ‘danger signals’ (i.e. Pathogen- and Damage-Associated Molecular Patterns; PAMPs and DAMPs, respectively) trigger inflammatory signals which recruit and activate other innate immune cell types early on to broadly and non-specifically kill cells altered by infection. Importantly, PAMP/DAMP signals also trigger innate immune cells to initiate long-lasting adaptive immunity, the sine qua non for cancer vaccines. Adaptive immunity utilizes protein-protein interactions to provide its hallmark exquisite specificity in responding to epitopes from protein antigens encoded by the genome of the invading organism. PAMP/DAMP-activated myeloid lineage antigen-presenting cells (APCs, e.g. dendritic cells [DCs]) recognize and ingest dying cells and cell debris, proteolytically processing them into small peptides. A subset of those peptides will specifically bind one or more members of the large family of highly divergent major histocompatibility complexes (MHCI and MHCII; often referred to as HLAI and II, respectively, in humans) such that they are displayed on the cell surface. These MHC-epitope complexes are then recognized by the highly diverse family of T-cell receptors (TCR), each uniquely expressed on the surface of CD8+ and CD4+ T cells. Importantly, those same pathogen proteins are proteolytically processed within, and displayed on the surface of infected cells to facilitate T cell immune surveillance. Cell-cell communication, critically dependent on highly specific MHC:peptide∷TCR interactions, enables and strengthens such responses. Pathogen-specific CD4+ T cells provide critical immunologic ‘help’ to produce more effective CD8+ T cells by licensing APCs via costimulatory receptors (e.g. CD40-CD40L) and by local cytokine secretion (acting on APCs and CD8+ T cells). Simultaneously, rare target-specific B cells scan the extracellular milieu for proteins (released from stressed/dying cells) that bind to their unique but similarly highly diverse B-cell receptor (BCR). Antigen binding stimulates B cell clonal expansion and maturation to produce one or more forms of a secreted target-specific antibody. B cells may also function as APCs, presenting processed peptides from those target (and other co-internalized) proteins on MHCII to cognate CD4+ T cells, which in turn, stimulate the presenting B cell. Although initiated in infected tissue, these cellular and molecular interactions largely occur, and are spatially and temporally enabled, within highly specialized lymph nodes (LN) (Grant et al., 2020). Adaptive immunity thus depends on the recognition of epitope-containing protein antigens. A critical challenge is separating the wheat of desirable targets from the chaff of self. Which epitopes an individual can even ‘see’ is in part defined by the limited subset of MHC alleles encoded in their genome, amongst the hundreds that exist across the species (allele diversity across the species contributes to population-level control). Foreign- vs self-specificity arises at multiple levels beginning with central tolerance during progenitor T and B cell development, wherein TCRs and BCRs with strong reactivity to self are removed from repertoires, to peripheral tolerance, featuring target-specific Treg cells as well as varied immune dampening mechanisms that prevent response in the absence of clear danger. Additionally, immunologic history and environment (e.g. the highly diverse gut microbiome (Fluckiger et al., 2020)), and the possible reactivity of a given TCR to different MHC:peptides (Birnbaum et al., 2014) can affect the TCR profile, potentially leaving irreplaceable ‘holes’ in the repertoire. What then is unique about cancer and cancer vaccines with respect to natural immunity? First, cancer is a disease of self, without (except for a handful of virally-induced cancers) a ‘foreign invader.’ This provides cancer major advantages: natural immunity is limited by the loss of foreign PAMPs and completely foreign antigens as surveillance tools, and by physiologic dampening mechanisms that prevent damage to self. Fortunately, cancer cell intrinsic mutations central to transformation also create neoantigen targets for the immune system, and the dysregulated growth and DNA repair processes of malignant cells lead to further ‘passenger’ mutations, unique to the cancer cell and its descendants, and provide substantially more neoantigens. Mutations in cancer thus establish an evolutionary competition with the immune system, in which the immune system sometimes wins (‘Elimination’) or generates a persistent draw (‘Equilibrium’); or the cancer eventually wins (‘Escape’) (Dunn et al., 2002). But even this Achille’s Heel can have two sides, as genetically heterogeneous tumors enhance the opportunities for immune escape by target antigen loss (Anagnostou et al., 2017). Chronic infections illuminate another important aspect of the immunity-cancer interface. Continual antigen exposure, also at play in cancer, can phenotypically shift cytolytic T cells to an ‘exhausted’ state. This shift compromises the ability of T cells to eliminate infected target cells (Philip and Schietinger, 2021), resulting in an equilibrium between viral control and tissue damage. An immunosuppressive, tissue-specific microenvironment – The endpoint Tumors contain not only cancerous cells but also numerous non-malignant cell types, paradoxically including, immune cells, creating a suppressive niche that protects the tumor from immune attack. Long characterized as highly vascularized, glucose-depleted and hypoxic, the rich cellular immune features of the TIME were linked to clinical outcomes almost two decades ago and have been extensively explored since. The TIME in a primary tumor may differ from that in a metastasis, and the types of suppressive niches formed are associated with target tissue-specific-(Zagorulya et al., 2020), and tumor histology specific- patterns (Wellenstein and de Visser, 2018); this highlights both far-reaching tissue of origin consequences and the potential for diverse mechanisms in the same patient. The TIME is composed of multiple pro-tumor cell types and even pro-tumor acellular structures (Figure 2): Tumor-associated macrophages (TAMs): Long considered solely in tissue repair and clean-up, macrophage lineage cells are now appreciated as phenotypically heterogeneous and plastic. Among the most abundant pro-tumor cell types in many cancers (Mehta et al., 2021), TAMs are co-opted by tumors to suppress local immunity. This ‘renaissance’ lineage can promote tumor growth and metastasis via tissue remodeling by stimulating angiogenesis and producing TGFβ (to stimulate tissue growth and Treg differentiation), and via multiple secreted and cell surface factors. Myeloid-derived suppressor cells (MDSCs): MDSCs are increasingly associated with progressing cancer. Initially defined functionally (suppressive) and morphologically (‘immature’), high dimensional technologies have now more precisely characterized them, revealing nuanced functional/molecular ‘archetypes’ driven by myeloid cell plasticity, multiple suppressive mechanisms and tissue specificity (Hegde et al., 2021). Regulatory T cells (Tregs): Tregs have been long recognized as suppressive. Effector CD8+ to Treg cell ratio in tumors is a positive prognostic marker (Togashi et al., 2019). However, the importance of antigen-specificity - expected based on the presence of unique TCRs - awaits more definitive experiments (Moorman et al., 2021). Of note, melanoma neoantigen-specific Treg expansion has been found amongst tumor infiltrating lymphocytes (TIL) in MHCII+ melanomas (Oliveira et al., 2022), potentially representing a unique escape mechanism for MHCII-expressing tumors. Cancer associated fibroblasts (CAFs): CAFs are the predominant non-immune cell type within tumors. Like TAMs, they are a heterogeneous and plastic population associated with both pro- or anti-tumor properties (Desbois and Wang, 2021), and even direct antigen presentation to CD4+ T cells (Huang et al., 2022). CAFs contribute to dense, T cell excluding, extracellular stroma via secreted cytokines and chemokines, and recruit suppressive cells (Tregs and MDSCs) or directly inhibit T cells. mregDCs: While dendritic cells are typically thought to initiate immunity, this subset (discussed below) appears to have suppressive capability. Multiple chemokines, cytokines and other signaling molecules (e.g. VEGF) and exosomes produced by tumors are involved in the recruitment of a complex mix of cells, their conversion to a pro-tumor phenotype (Propper and Balkwill, 2022) and the structural features of the TIME. This environment creates a final challenge to immune cells, either induced physiologically or via vaccine, and must be considered in the design of any vaccine strategy. The logistics of building the vaccine bridge How do we train an immune system that has evolved to fight external pathogens to target a tumor that continually evolves to evade that same immune system? Using natural immunity as a blueprint, we discuss 4 inter-dependent ‘towers’ that support this bridge: cellular players, antigen, delivery to APCs, and co-therapies (Figure 3). LYMPHOCYTES Cancer vaccines have generally focused on eliciting tumor-specific CD8+ cytotoxic T cells (CTLs). Indeed, CTLs are undeniably central to tumor cell killing and foundational to most immune therapies, and across many tumor types survival correlates with CD8+ TIL numbers (Reiser and Banerjee, 2016). Because of this centrality, improving the number or phenotype, and ideally both, of the CTL repertoire is a critical unmet need for the field. A key to this objective is engaging the full range of adaptive immune cells to both complement and augment CTL responses. CD4+ T cells may have cytotoxic functions (Cachot et al., 2021), and transfer of neoantigen-specific CD4+ TIL can control metastatic cholangiocarcinoma (Tran et al., 2014), either by direct cytolytic activity or licensing other immune cells. CD4+ T cells can also directly recognize MHCI presented peptides without co-receptor involvement (Oliveira et al., 2022), although how to induce this phenotype or select epitopes with this feature is currently unknown. Perhaps more important is the role CD4+ T cells play in providing help to CD8+ CTLs. In a syngeneic tumor model, CTL-mediated tumor control and checkpoint inhibitor efficacy were dependent on CD4+ T cell recognition of an MHCII restricted neoantigen (Alspach et al., 2019). The presence of both an MHCI/CTL epitope and an MHCII/CD4+ T cell epitope in the same tumor cell (vs. in different tumor cells) was required to elicit maximal tumor-specific CTL responses, including after vaccination with irradiated tumor cells. These data support the long-standing dogma that optimal CTL stimulation by an APC requires interaction of that APC with an interferon-gamma producing CD4+ ‘helper’ Th1 T cell recognizing an MHCII restricted epitope on the same APC. This licensing by the APC requires CD40L (on the T cell)-CD40 (on the APC) interaction, and can be mimicked by an agonistic CD40 antibody (Morrison et al., 2020). Still unclear are whether epitope-intrinsic features (e.g. abundance or TCR avidity) are needed to induce effective help, or if it is determined by the local APC environment. Simply considering helper and cytotoxic T cell responses does not reflect the tremendous progress over the last decade, driven by bulk and single cell RNA sequencing, epigenetics, and animal models, in understanding the breadth, inter-relationships and trajectories of different T cell subsets. This complex T cell biology has been reviewed in depth elsewhere (Kumar et al., 2018), but we highlight two vaccine-related features. First, T cells are phenotypically dichotomized as effector, which are short lived (weeks) and have active functionality (e.g. cytotoxicity), or memory, which persist for months to years, ready to respond to later challenge. Interconversion from memory to effector states is routine, but the reverse, less common. In cancer and chronic infection, effectors often take on an ‘exhausted’ state with limited functionality, while a subset of ‘precursor exhausted’ cells maintain the proliferative, self-renewal and differentiation potential that are critical to successful immune checkpoint blockade (ICB) therapy (Liu et al., 2021; Sade-Feldman et al., 2018). Second, memory cells have three major forms (Kok et al., 2022; Kumar et al., 2018). Effector memory (TEM) cells are found in tissues and blood and are poised to rapidly expand and gain effector functionality upon antigen encounter; Resident memory (TRM) cells are found only in tissues, and are similarly poised to rapidly expand and gain effector function. Central memory (TCM) cells are found in tissues and blood, but respond to antigen by migrating to draining LNs to proliferate extensively and provide a second round of effectors. TCM cells also can self-renew. How vaccine approaches can proactively modulate T cell phenotypes is unclear, as are the optimal pattern and dynamics. We hypothesize that vaccine strategies that induce strong memory populations will be more efficacious for cancer than those that favor short-lived effectors. In this vein, stimulation of naïve cells to create de novo responses may bring more to the table than stimulating pre-existing, potentially exhausted effector cells. In contrast to T cells, B cells and their interactions with T cells have been far less well-studied (Downs-Canner et al., 2022) in cancer vaccinology. B cell tumor infiltrates, especially in the context of intra- or peri-tumoral tertiary lymphoid structures, are associated with longer survival and response to ICB in patients (Schumacher and Thommen, 2022). Antibodies targeting surface receptors have been used clinically for years, and engage multiple cells types to kill cancer cells and can even induce immunogenic cell death (Pozzi et al., 2016). Further, neoantigen specific IgA can control ovarian cancer by redirecting myeloid cells to attack malignant cells (Biswas et al., 2021). B cell:T cell interactions are important in two ways. First, antibody maturation and production are enhanced in the presence of activated CD4+ T cells that recognize epitopes associated with the target protein, ensuring that antibodies are made in the context of a broader immune response (classically defined IL-4-producing Th2 T cell help). Second, B cells can support T cell function. For example, in a murine model, interactions between tumor-antigen specific B cells and tumor-antigen specific CD4+ T follicular helper cells (TFH) triggered IL-21 dependent TFH cell support of CTL differentiation and promoted ICB responsiveness (Cui et al., 2021). For cancer vaccines to engage the full range of adaptive immune cells and maximize CTL responses, antigen structures and formats recognized and internalized by both APCs and B cells, and with the complexity to contain MHCI, MHCII and B cell epitopes, will be needed. Conversely, this argues against the reductionist approaches extensively employed in the past (i.e. inclusion of only minimal length MHCI/CD8+ epitopes or of only a few epitopes). ANTIGENS and MHC The complex and evolutionarily optimized cellular and molecular output of natural immunity begins with two central components, antigens and antigen-presenting cells; thus, optimizing both these parameters is warranted. Most current generations of cancer vaccines (and to reflect that current interest, our discussion here of antigens) focus on neoantigens, which are truly non-self antigens. Other antigen classes include tumor associated antigens, which are also expressed in normal tissues, and have been extensively tested with only limited success (reviewed over the years), and viral antigens, which are relevant to only a handful of virus-driven cancers; many of the principles we discuss are relevant to these antigens as well. Neoantigen-targeting vaccines broadly require identifying translated, cancer-specific mutations and predicting those most likely to be presented in MHC molecules (still with a primary focus on MHCI, due to the well-described cytolytic role of CTLs and also out of technological necessity [see below]). The results of neoantigen vaccination have been tantalizing (Ott et al., 2017; Sahin et al., 2017), but these trials and new and more facile approaches to examining tumor and peripheral blood T cell populations for tumor-reactive cells (Danilova et al., 2018; Lam et al., 2021) show that only a small fraction of neoantigens predicted to be immunologically useful are spontaneously and/or durably immunogenic. Thus, our understanding of what constitutes an immunogenic epitope remains rudimentary. Several pressing hurdles regarding neoantigen selection remain. First, the types and numbers of neoantigens observable in each tumor sample are currently limited by sequencing and/or bioinformatic technologies (Figure 4A). Single nucleotide variants (SNV) and small insertions/deletions (indels) in the annotated genome are now standardly detected. However, still beyond routine detection are SNVs in unannotated open reading frames (nuORFs), more complex novel ORFs (neoORFs) encoding potentially longer neoantigens from gene rearrangement/duplication transcripts, splicing variations, ‘exitrons’, and dysregulated splicing due to mutated or dysregulated splicing machinery, dysregulated translation, endogenous retroviral elements, and post-genetic changes such as A to I editing (Gupta et al., 2021; Ouspenskaia et al., 2021; Wang et al., 2021; Wu et al., 2018). Depending on the underlying defects in DNA repair and RNA splicing, conventional methods may thus miss more than half of a given tumor’s neoantigens (Wu et al., 2018). Still unanswered is if any of these classes are more effective targets than others. Second, determining which putative neoantigens are actually presented on MHC and if they will be immunogenic is yet unresolved. Three general approaches, ordered based on decreasing scalability, have been used to identify target neoantigens. (A) Prediction: From in vitro binding affinities measured for thousands of short peptides across a subset of MHC class I alleles, binding prediction algorithms have been developed (Nielsen et al., 2007) and further refined by training with substantially larger mass spectrometry datasets (Reynisson et al., 2020). MHCII binding predictors have lagged due to length variability and more limited binding affinity data, but breakthroughs have come via immunopeptidomic analyses (Abelin et al., 2019). However, despite improvement, there are still limitations and unexpected findings. Across multiple trials with fundamentally different formats of neoantigen-targeting vaccines (Ott et al., 2017; Sahin et al., 2017), only ~15–30% of predicted epitopes yielded CD8+ responses, a fraction consistent with an evaluation of in vitro immunogenicity (Stronen et al., 2016), and these responses were generally weak. Moreover, systematic screening of predicted peptides has revealed limited capacity to generate spontaneous T cell responses for the majority of predicted class I and II neoantigens (Lam et al., 2021) and observations of antigens with low binding score to MHCI having efficacy in vaccine-induced tumor models highlight shortcomings in current algorithms (Ebrahimi-Nik et al., 2021). Finally, the majority of epitopes predicted for MHCI binding, when given as part of longer peptides have tended to more effectively elicit CD4+ T cell responses; this observation about the efficiency of CD4+ stimulation is still not satisfactorily explained and the functional significance remains uncertain. (B) Observation: Identifying MHC-bound peptides on tumor cells by MS provides enormous confidence that the peptides have cleared numerous hurdles toward immunogenicity (expression, processing, transport, and MHC binding to an observable level – all of which may be modulated by the tumor to evade immune detection (Jhunjhunwala et al., 2021)). Direct immunopeptidome analysis of melanoma tissue from 5 patients identified (only) 11 bona fide MHCI neoantigens, of which 4 stimulated specific T cell responses (Bassani-Sternberg et al., 2016). Thus, this strategy is limited by low sensitivity and also by tissue availability (i.e. biopsy size), inaccessibility of MHCII presented peptides on APCs (APCs would be rare in biopsy samples) and the requirement for advanced MS/bioinformatic technology. (C) Screening: Multiple groups have explored empiric strategies to directly test the ability of individual epitopes to stimulate patient-derived peripheral blood T cells (Danilova et al., 2018; Lam et al., 2021; NCT03633110)). For example, Lam identified responses to fewer than 10% of predicted ‘strong’ neoantigen epitopes and ~30% of positive responses came from neoantigens that would not have been selected using conventional criteria (see (A)). This particular assay also detected epitopes that inhibited baseline response (‘Inhibigens’); it remains unclear if these epitopes should be discarded. An important caveat to these strategies is that they likely only reproducibly measure pre-existing patient responses. These models might be improved by adding other features. A better understanding of peptide processing dynamics due to (immuno)proteasome or cellular compartment in APCs (Grotzke et al., 2017) and cancer cells (Jhunjhunwala et al., 2021) may be especially important. Other avenues to consider include: peptide-MHC stability (Harndahl et al., 2011), similarity to self or to pathogens (Richman et al., 2019), and TCR∷peptide:MHC dynamics (Ebrahimi-Nik et al., 2021) (Figure 4B–C). How to weigh each feature is challenging. Third, future iterations of vaccines will likely benefit from inclusion of epitopes to stimulate not just CD4+ and CD8+ cells, but also B cells. As previously discussed, CD4+ help is critical to efficient CTL priming, and so it follows that including both MHCI and MHCII presented peptides will improve efficacy. Likewise, tumor-specific B cells/antibodies have been shown to play an important role in suppressing cancer growth and augmenting CTL responses (Biswas et al., 2021). It remains unclear whether universal class II (e.g. PADRE, Tetanus Toxin p30) and B cell epitopes (e.g. Tetanus toxoid) (Fletcher et al., 2018; Swartz et al., 2021) are equally or less effective than similar tumor-specific epitopes and how important physical linkage is. Incorporating tumor-specific B cell epitopes, at least in peptide vaccines, may be challenging as BCRs generally survey non-linear protein epitopes (Potocnakova et al., 2016). Fourth, in choosing antigens, we must take into account that vaccines enter the equation towards the end of a long evolutionary battle between immunity and cancer, a struggle that likely selects for clones with less spontaneously immunogenic mutations, profoundly affecting the mutational landscape (Turajlic et al., 2019). Further, the phenomenon of antigen immunodominance (Burger et al., 2021) may limit the number of neoantigens which are naturally recognized without intervention. Indeed, only a small fraction of known potential neoantigens exhibit detectable T cell responses in tumors (Kristensen et al., 2022). On the other side of this evolutionary battle, the T cell compartment no doubt bears scars as well, and tumor-immune co-existence may shape the breadth of future T cell responses. Neoantigen-specific TILs are exhausted (Caushi et al., 2021; Oliveira et al., 2021), and as has been elegantly shown in viral infections, the exhausted state leaves epigenetic scars which may be difficult or impossible to remove (Abdel-Hakeem et al., 2021; Yates et al., 2021). Similarly, ICB response in lung cancer is not associated with expansion or reinvigoration of terminally exhausted T cells, but rather with ‘clonal revival’ - the local and peripheral expansion of new and pre-existing T cell clonotypes which acquire precursor exhausted phenotype (Liu et al., 2021). Further, neoantigen-specific Tregs have been identified (Oliveira et al., 2022). These observations raise clinically relevant questions: Can vaccination reverse the predominance of exhausted cells? Are naive rather than antigen-experienced T cells a better target? Will vaccination stimulate immunosuppressive neoantigen-specific Tregs? DELIVERY TO APCs Just as a natural adaptive immune response grows from innate immune processing at the start of an infection, productive vaccine responses are driven by effective delivery to innate immune cells. Delivery is thus about getting the target antigens to the right innate immune cells, with the proper maturation cues to properly activate antigen-specific T cells. Two important considerations include antigen format (Box 1) and delivery. Vaccine formats have been extensively reviewed elsewhere (see (Irvine et al., 2020)), but here we highlight key cross-cutting principles: antigen presenting cell targets, adjuvants, and delivery. Antigen-presenting cells. Which APCs should be targeted by a vaccine? A key property of APCs for vaccine applications is antigen cross-presentation. Typically the provenance of specialized cells, such as some dendritic cells (Wculek et al., 2020), cross-presentation uniquely allows for MHCI presentation of external peptides by delivering endocytosed external antigens (e.g. from cancer cells or vaccines) to the cytoplasm for proteasomal degradation, with the peptide products transported into the endoplasmic reticulum for MHCI loading (Figure 4B). This capability complements the constitutive presentation of endogenously-produced peptides from most cells, including cancer cells, through the same cytoplasm/proteasome/MHCI pathway, enabling T cell surveillance. Four classical DC subtypes have been defined (Box 2): conventional DC 1 (cDC1), cDC2, and plasmacytoid DCs (pDCs), all derived from common DC precursors, and monocyte-derived DCs (moDCs) (Wculek et al., 2020). Recent single cell transcriptional analyses have provided higher resolution understanding of the DC compartment (Villani et al., 2017), with functional studies revealing additional subsets (e.g. mregDCs; (Maier et al., 2020)). These DC subsets differ by frequency, PRR expression, scavenging receptors, trafficking properties, and capacity for antigen (cross) presentation (Wculek et al., 2020). Human cDC1s, which are highly specialized in cross-presentation, are uniquely marked by CLEC9A and XCR1, both of which (Fossum et al., 2015; Zeng et al., 2018), in addition to DEC-205 (Bhardwaj et al., 2020), have been used as targets to efficiently deliver antigens to this rare DC subset in vivo. Other cell types can also serve as APCs. Just as for moDCs, neutrophils (Mysore et al., 2021), T cells (Booty et al., 2022; Veatch et al., 2021), B cells (Booty et al., 2022), unfractionated human PBMCs (Booty et al., 2022), or immunoproteasome-engineered mesenchymal stem cells (Abusarah et al., 2021) can be coaxed ex vivo to effectively cross-present antigen and generate strong anti-tumor immunity as part of cell-based vaccine therapies. Further, improvements in bead-based isolation and stimulation of various DC subsets have enabled ex vivo production of blood-derived DC subsets (Hanel et al., 2021). Ex vivo cell manipulation adds substantial cost and complexity to vaccine delivery, which may be offset by the benefit of direct antigen (or antigen-encoding nucleic acid) delivery to APCs. An open question is whether high-quality T cell responses are optimally generated through only a specific APC (e.g. cDC1s), or if combinatorial strategies to engage multiple APC subsets, potentially including B cells (Sagiv Barfi et al., 2021), would be more effective. Approaches such as the use of FLT-3 ligand to enhance natural dendritic cell numbers in the blood, and presumably therefore in key tissues, may also be required (Bhardwaj et al., 2020). Adjuvants. Effective antigen presentation depends upon DCs taking up antigens with activating signals to induce their maturation. For vaccination, this maturation can be driven by ‘danger signals’ (DAMPs or PAMPs; Figure 5) from adjuvants and/or co-stimulatory molecules (e.g. FLT3, CD40). PAMPs and DAMPs are recognized by similar PRRs, which have been extensively reviewed (Wicherska-Pawlowska et al., 2021). However, PAMP- and DAMP-triggered signals are not necessarily equivalent; these differences may allow unique responses to the simple presence of a pathogen vs. a pathogen that is killing host cells. For example, DCs treated with the gram-negative bacterial cell wall component lipopolysaccharide (LPS; a PAMP), or oxidized 1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphorylcholine (oxPAPC; a DAMP), both induce IL1β production via Caspase-11 mediated inflammasome activation. oxPAPC, however, protects DCs from LPS-induced pyroptosis, yielding a longer-lived “hyperactivated” DC with improved capacity to migrate to lymph nodes, induce CTL responses and control murine tumors (Zhivaki et al., 2020). Moreover, different DC subtypes may express different patterns of PRRs; for example, human cDC1s specifically express high levels of TLR3 (which recognizes the common adjuvant polyI:C). Thus, one avenue to improve cancer vaccine design will likely be based on better understanding how DAMP/PAMP combinations differentially generate unique APC phenotypes. Alternatively, DC interactions with other cells present at sites of inflammation, such as platelets, have generated effective DCs (Han et al., 2020). An important related consideration is consistent co-delivery of antigen and adjuvant to the same APC to overcome the potential for induction of tolerogenic responses by non-activated APCs (Toes et al., 1996). Nanoparticles containing TLR 7/8 agonists and specially modified peptide antigens elegantly solved this problem, ensuring the uptake by each APC of multiple copies of both peptide antigen and adjuvant molecules, which improved the magnitude and breadth of CD8+ responses (Lynn et al., 2020). Genetic or chemical linkage strategies have also been developed for protein antigens (Belnoue et al., 2019) and for DC-targeted RNA-based vaccines (using the intrinsic immunostimulatory properties of RNA (Kranz et al., 2016)). Solutions for in vivo delivered DNA-based vaccines are less apparent due to uncertainty and perhaps heterogeneity with which cells express the encoded antigens, although encoding of the DAMP HMBG1 has shown some promise (Fagone et al., 2011). A consistent theme is the superiority of linked antigen and adjuvant. Schedule, boosting and route. A relatively under-explored aspect of vaccination is how the schedule of antigen delivery impacts downstream outcomes. In natural acute infections, immunogenic antigens rise exponentially to peak levels over days to a week and are generally cleared within 2–3 weeks (Irvine et al., 2020), resulting in sterilizing immunity. These kinetics differ from chronic infection and cancer, where some level of antigen exposure continues for an extended period (months), resulting in an undesirable ‘exhausted’ T cell population which, without other intervention, is unable to mount effective responses. An exponential dosing scheme over 4–8 days, maximizes T and/or B cell responses and tumor control in murine models (Johansen et al., 2008), and has support also in non-human primates (Pauthner et al., 2017). To our knowledge, no clinical trials have yet tested this natural immunity-mimicking concept directly. However, trials using self-amplifying RNA vaccines based on alphaviruses, which replicate exponentially to increase antigen production (Melo et al., 2019) like in natural infection, are ongoing for rabies (NCT04062669), SARS-CoV-2 (NCT05132907) and cancer (as part of a heterologous prime-boost strategy (see below); NCT05141721). Similar unresolved questions are the timing of boosting immunizations and whether a ‘heterologous prime-boost’ (different immunogenic formats for the prime and boost) provides even stronger or broader immunity. Undoubtedly, ‘homologous’ boosting is effective, at least with respect to antibody responses, as recently observed with the approved SARS-CoV-2 vaccines. For effective anti-cancer vaccines, however, where the most relevant measure may be T cell immunity, optimal timing and dosing of booster immunization (during expansion, contraction or ‘resting’ of primed cells) in humans is non-existent. Heterologous prime-boost strategies were initially necessitated by strong anti-vector (typically a viral vector) responses on priming, which were thought to limit T cell responses following boosting. Over the years the concept has been extended beyond vector changes to include alterations which engage the immune system in different ways (adjuvants, APCs, dose kinetics, vaccine formats) and many preclinical studies (Hofer et al., 2021) have demonstrated improvements in magnitude, breadth of response and efficacy. While cancer trials are recruiting (e.g. NCT03794128), early definitive clinical outcomes are most likely to come from ongoing COVID heterologous prime-boost studies (e.g. NCT05074368, NCT04907331) although analysis of T cell responses has typically been lower priority in infectious disease studies. Finally, the anatomical site of delivery is emerging as an important variable. In a comparison of intravenous vs subcutaneous delivery of antigen/adjuvant nanoparticles, when the doses were adjusted to yield equivalent circulating T cell numbers, tumor control was significantly better when delivered intravenously, associating with a higher proportion of TCF-1+ stem-like cells (Baharom et al., 2021). Similar differences between intravenous vs. subcutaneous delivery have been observed with some long peptides (but not all, potentially due to peptide biochemical properties) admixed with adjuvant (Sultan et al., 2019), and in a peptide-multi-adjuvanted liposome format, improved responses were shown with intraperitoneal rather than subcutaneous immunization (Korsholm et al., 2014). Finally, targeting vaccines to appropriate anatomic sites may also improve efficacy by enhancing TRM induction, as observed with mucosal targeting via albumin (Rakhra et al., 2021), and with epicutaneous administration of a modified vaccinia ankara virus vector, mimicking a ‘pox’ lesion (Pan et al., 2021). CO-THERAPIES Given the extensive tumor-immune evolution of ‘escaped’ tumors, we posit that vaccines will require combination with other therapies to unlock the power of the immune system. These co-therapies fall into three partially overlapping buckets. Maximizing innate T cell priming. Taking a cue from natural immunity, one strategy is to mimic the inter-cellular cross-talk mechanisms between the innate and adaptive immune cells that drive priming. For example, local CD40 agonist antibodies induced robust CTL responses in mice immunized with epitope-length peptides that were otherwise tolerogenic (Diehl et al., 1999), presumably by mimicking the APC licensing normally provided by CD4+ T cell help. Indeed, agonist CD40 antibodies or soluble CD40L are under renewed clinical development after initial toxicity concerns (Vonderheide, 2020). Similarly, CD27 agonist antibodies (mimicking CD70 signals from DCs) augmented peptide vaccine responses and vaccine-induced syngeneic tumor control (Riccione et al., 2018). Thus, replicating the licensing signals that helper CD4+ T cell give DCs (via CD40) or that DCs give to T cells (via CD27) can amplify vaccine responses. Another route amplifying productive APC-T cell interactions may be simply increasing APC numbers. Pretreating patients with FLT3L increased circulating DC numbers by ~30x, and augmented antibody and T cell responses (Bhardwaj et al., 2020). Collectively, these studies indicate that vaccine design can continue to mine the cytokines and co-receptors critical in natural immune responses. Immune checkpoint blockade. ICB targeting CTLA-4, PD-1/L1 and LAG-3 can re-invigorate (some) ‘exhausted’ T cells resulting in dramatic and durable clinical responses that have revolutionized treatment for many cancers. Although these agents may be also useful in augmenting vaccine de novo induced T cell responses, questions remain regarding when and how to combine ICB with vaccine priming. PD-1/L1 has been most extensively investigated. In mice, PD1 is required in CD8+ T cells during viral infection for effective memory T cell formation (Pauken et al., 2020). In a therapeutic murine syngeneic model, concomitant vaccination and PD-1 blockade unveiled strong T cell responses that controlled tumor growth, while PD-1 blockade beginning just 3 days prior to vaccination resulted in dysfunctional PD-1+ CD38hi T cell responses (Verma et al., 2019). Finally, treatment with anti-PD-1 prior to tumor irradiation dramatically reduced the abscopal effect in a dual flank model and shortened survival compared to treatment with anti-PD-1 after irradiation (Wei et al., 2021). These data caution against vaccination closely preceded by anti-PD-1 therapy, raising a timing conundrum, as standard-of-care for multiple diseases often begins with ≥1 years of uninterrupted ICB. Possible approaches to this timing issue are discussed below but we emphasize that to be welcomed as a routine part of combination immunotherapy, personal neoantigen vaccines need to make significant manufacturing improvements (Fritsch et al., 2020). Short-circuiting the suppressive TIME. Vaccine strategies almost certainly need to include mechanisms that relieve immune suppression at the tumor site, especially if treating advanced disease is envisaged since the TIME generally evolves to impede immunity. Multiple strategies to facilitate T cell function in this environment are being developed and should be considered in vaccine trials (Box 3). Outcomes What can be expected of cancer vaccines and how should we look? We discuss four pertinent concepts for cancer vaccine trials. Clinical Impact. Despite an impressive increase in technical capability to monitor immune responses and more sophisticated understanding of cell phenotypes afforded by recent profiling technologies, clinical impact remains the most satisfying and trustworthy outcome. Thus, an important design variable is conducting trials in scenarios where a clinical benefit could be discernible. We posit that trials focused only on the less useful question of ‘are immune responses generated?’ should ideally be evaluated quickly (while analyzing immune correlatives thoroughly) and, if warranted, advanced to small trials that can generate a signal of clinical benefit. Simon two-phase designs can be useful to maximize the utility of recruitment. Design and deconvolution of co-therapy trials. In hindsight, a major flaw in the design of many past vaccine efforts was immunization alone in advanced disease with a highly evolved and suppressive TIME. Vaccines may effectively generate tumor-killing T cells (and maybe tumor-specific antibodies), but getting these warriors to the tumor, in sufficient numbers and quickly enough, and with the ability to overcome the tumor defenses is challenging as monotherapy in the setting of advanced disease. The use of co-therapies raises two important considerations: (1) how will the co-therapy affect the vaccine response – will it be complementary and synergistic or antagonistic? (2) de-convoluting effects of the combined therapy from those of the co-therapy alone. For the first question, unless the co-therapy is chosen based on a hypothesis of concomitant synergistic effects, temporally separating the vaccine and co-therapy (and recognizing that for de novo responses, priming and boosting are two critical and immunologically distinct phases) seems warranted. Since ICB is standard-of-care in many disease settings, weaving a vaccine-only phase into patient care can be challenging, arguing that less well-studied (or where immunotherapy has already failed) diseases need to be explored. Anti-PD-1 co-therapy, for example is a widely used standard-of-care, where sequencing of administration has biologic impact (discussed above). For the second question, historical controls are unreliable unless huge effect differences are observed; active control arms are preferred so that consideration of patients appropriate for the trial are made by the same investigators and contemporaneously. Disease settings. The other obvious approach to confronting TIME challenges is to look at early or minimal disease settings, where immunosuppression may be less well developed. One frequently pursued strategy is adjuvant therapy to prevent recurrence following surgical resection with curative intent. An important question regarding adjuvant vaccination is whether surgery sets up an immunologically compromised state and for what duration (Cheng et al., 2022); prolonged delays to adjuvant therapy are likely undesirable. Neoadjuvant vaccination potentially avoids surgery-induced immunosuppression during vaccination and provides correlative efficacy data in the form of clinical and pathological response, as observed with ICB (Menzies et al., 2021; Rozeman et al., 2021). The multiple weeks needed for personal neoantigen vaccine manufacture generally eliminates such an approach from consideration. Potentially even more desirable would be vaccinating during incipient disease, targeting, for example, ductal carcinoma in situ, pre-neoplastic colonic adenomas observed in colorectal cancer or pre-neoplastic vulvar intraepithelial neoplasia ((Finn, 2018); NCT04144023) or true prophylactic vaccination in high-risk groups (NCT04367675). In these settings, vaccine alone may be effective. Although such studies have focused to date on classic tumor-associated antigens, viral or common frameshift (Lynch Syndrome) antigens, personal neoantigens should be possible where precursor lesions portend increased risk of clonally related invasive disease even after removal (e.g. DCIS). Genes expressed in pluripotent stem cells, which overlap considerably with cancer stem cell gene programs, are another emerging source for targets (Ouyang et al., 2019). Mouse models – the most commonly studied tumor vaccine setting – are an obvious place from which to draw inspiration. However, the complexities of a TIME established by immune pressure over years in humans vs days in mice, shaped by immune systems separated by 80 million years of evolution and by very different pathogen exposures, and typically in anatomically irrelevant sites challenge the translatability of murine models (Lal et al., 2021). Nonetheless, such models can certainly provide a basis for hypothesis testing, especially when coupled with human correlative data, although they may only provide preclinical rationale for a subset of important questions about human biology. While the growing availability of autochthonous models may provide somewhat more translatable information, we posit that less expensive but safe approaches to more rapidly testing new concepts in humans need to be developed, fostering ‘reverse translation’ using the increasingly sophisticated suite of analytical technologies now available. Immune correlatives. The tools to investigate the ‘surrogate’ immunology of clinical samples have advanced dramatically over the past decade due to the generation of disruptive new technologies and advancements in existing tools. This includes prediction of peptide epitopes, epitope-specific MHC multimer availability, multiplex flow analysis (fluorophores or mass-labeled), single cell RNA sequencing, and bulk and single cell (including paired) TCR and BCR sequencing. Moreover, if tumor samples are available post-vaccination, multiplexed immunohistochemistry has advanced such that more discriminating spatial analysis can be performed (Tan et al., 2020) and molecular approaches are now being applied to FFPE sections (Dries et al., 2021). How is such correlative analysis best applied to assess vaccine activity? Analysis of blood for circulating antigen-specific T cells, readily accomplished in various ways (tetramer staining, flow-based intracellular staining, IFNγ ELISPOT) ex vivo or after in vitro stimulation with one or many specific epitopes or total tumor lysate can reveal if responsive T cells are circulating - and ex vivo detection of CD8+ responses should be a minimum goal - but such analysis does not necessarily indicate tumor killing capability. Paradoxically, T cells may be rapidly sequestered in inflamed tissue or tumor, as response to checkpoint therapy has been associated with the disappearance of pre-existing T cell responses from the blood (Bochem et al., 2021), and hence circulating and intratumoral responses may differ. Analysis of longitudinal blood specimens can be highly informative for addressing the dynamics of immune responses. Analysis of tumor-infiltrating T cells using single cell transcriptome and paired TCR sequencing can be even more informative, especially when T cell phenotypic properties can be linked to detailed features of antigen specificity (Caushi et al., 2021; Oliveira et al., 2021). An extensive analysis using single cell sequencing and other techniques on multiple tumor samples and bulk TCR sequencing on blood samples to follow T cell dynamics demonstrated ‘clonal revival’ (see above) of tumor-specific clones in tumor and blood in patients responding to ICB and chemotherapy (Liu et al., 2021). A demonstration that vaccine therapy increased the number, and changed the phenotype, of T cells in tumors would be highly encouraging, especially if coupled with clinical response, but surgical sample availability is often limiting. In the more typical setting of limited tumor fragments, small volume analytics like immunohistochemistry (IHC) can follow total changes in T cells and other populations. More advanced approaches are applying nucleic acid tools to interrogate cell genotype and spatial context at the single cell level in small biopsy specimens (Dries et al., 2021). Genetic analysis of tumor cells can also reveal signs of immunoediting as a marker of immune pressure (Anagnostou et al., 2017). Changes in blood or tumor TCR diversity and clonotype expansion may be tracked by bulk TCR sequencing and with more granularity by single cell paired TCR-sequencing tools, providing a snapshot of T cell dynamics if adequate samples are available. The current challenge with this approach is that confirming epitope-specificity is time/resource consuming and so is restricted in widespread use. One approach which may optimally balance practicality with high informative content is evaluating blood samples for epitope spreading – an increase in the circulating T cells specific for tumor antigens that were not included in the vaccine (Figure 6B). Epitope spread may be the best indicator of tumor cytolysis, which in turn may provide a surrogate molecular indicator of clinical impact (Ott et al., 2020), is not technically challenging to conduct (ELISPOT, tetramer, ICS), only requires blood samples and has been detected following vaccination of patients with no clinical evidence of disease, perhaps pointing to its sensitivity (Hu et al., 2021). Finally, cell-free DNA analysis, particularly for tumor-specific mutations, is steadily emerging as a tool to quantitatively follow tumor genomes in the blood (Zou et al., 2021) and could potentially reveal target-specific immune pressure by tracking both vaccine targeted neoantigens and non-targeted mutations. Where to Find Help Where can we learn lessons to accelerate the pace of tumor-immune evolution in favor of the immune system with cancer vaccines? Multiple lines of evidence support our premise of the therapeutic benefits of mimicking natural immunity. TIL therapy – expanding T cells already present in tumors, the product of natural immunity – has provided durable benefit to late-stage patients (Kumar et al., 2021). The oncolytic virus T-VEC, an approved therapy for accessible melanoma, mediates its effect not by direct killing, but via the immune response generated by dying, virally-infected cells – a raison d’etre of the immune system (Ferrucci et al., 2021). Certain chemotherapies can induce immunogenic cell death (ICD), which provides sufficient DAMPs to initiate immune responses, and may be an important component of efficacy (Zitvogel et al., 2011). A host of pre-clinical and clinical studies explored parameters to optimize this interaction (Hernandez et al., 2021), culminating in the approval of anti-PD-1 in combination with first line chemotherapy for lung cancer. The abscopal effect of radiation therapy inspired the concept of in situ vaccination, turning the tumor bed into the site of systemic immune cell priming. Directly delivering some combination of tumor cell death inducers and danger signals to the tumor, sometimes enhanced with immunotherapies, is under active preclinical and clinical investigation (Hammerich et al., 2016) and relies on generating systemic responses primed by an in situ response (Chen and Mellman, 2013). These approaches share the theme of dying whole tumor cells (WTC) as an antigen and adjuvant source and return to one of our initial dogmas – antigen presenting cells taking up dead/dying cells to initiate natural immunity to extant antigens. Simpler than using the latest sequencing and bioinformatic tools to find the ‘best’ antigens, and less biased from incomplete knowledge, WTC vaccines rely on the ‘undefined’ antigens within a tumor. Particularly note-worthy in their preclinical and clinical resurgence are the advancement to phase II studies of a dendritic cell-AML cell fusion vaccine (Rosenblatt et al., 2016) and a promising study of a vaccine consisting of monocyte-derived dendritic cells loaded with autologous lysate from oxidized ovarian WTC (Tanyi et al., 2018). In the latter work, coupled mutation analysis enabled tracking of responses to the mutation-induced neoantigens, and demonstrated de novo induction, expansion, and TCR avidity enhancement following vaccination and a positive correlation between immunogenicity and time to progression or overall survival. These trials are encouraging, but still limited by tumor sample size and the logistics of implementation. Remarkably, despite the overwhelming abundance of self-antigens in these WTC formulations, there was no indication that auto-immunity, even as strong immunity was induced against ‘foreign’ antigen from the same immune milieu. Why is this? Central tolerance is impressively effective, but maybe less so for tissue restricted antigens (Legoux et al., 2015). Treg’s provide effective post-thymic tolerance, but depending on antigen specificity requirements, the breadth of the required Treg repertoire could be energetically unfavorable. A possible intrinsic contributor may be a limited diversity of immunogenic self-epitopes; the same evolutionary pressure that operates on tumors within a fraction of a human lifetime could have operated over eons on the ‘immunogenomic’ proteome, eliminating protein sequences that engender self-immunogenic response. Irrespective of an explanation, which certainly requires further exploration, there is little evidence that self-reactivity is an issue either as a result of natural anti-tumor immunity or with whole-cell based antigens. Conclusions and Directions The Golden Gate Bridge is a now timeless example of simple beauty – two towers supporting suspension cables to a roadway arching the turbulent waters of the narrow Golden Gate strait, a functional but elegant mark on an already picturesque scene. Innovative thinking and over a year of negotiating overcame an initial plan for an awkward cantilevered structure with multiple supporting trusses. By analogy, cancer vaccines, after 50 years of middling success, can stand to use more out-of-the box thinking (See Box 4 for key questions and ideas). Neoantigens have provided one such novel direction, but our ability to see through all the complexity to the other side (the right epitopes) is still clouded by gaps in knowledge, requiring still developing technologies and bioinformatic trusses (tools). An area of renewed interest is WTC vaccines, since by mimicking the stressed and dying cells that are the starting point of ‘natural immunity’, they provide access to all the antigens of the cell, at levels reflective of what is in the cell, and without any observer selection or bias, potentially providing a clearer image of a personal tumor than any technology can yet deliver. The importance of tumor-specific T cells to immune therapy and to durable patient survival cannot be over-emphasized. Cancer vaccines have been proven to enhance both the number of such cells and their repertoire, but there are still difficult waters to cross to arrive at the induction of truly effective T cell populations to achieve a lifeline to cure for our patients. To borrow from Joseph Strauss, the Chief Engineer for the Golden Gate Bridge, ‘When you build a [vaccine], you might build something [for the patient] for all time.’ Declaration of Interests M.C.S. has no interests to declare. E.F.F. is an equity holder and consultant for BioNTech, an equity holder and scientific advisory board member of BioEntre, and a founder and equity holder of Dionis Therapeutics. C.J.W. is an equity holder of BioNTech. An immediate family member of C.J.W. is an advisor and equity holder for Related Sciences and receives research funding from Bristol-Myers Squibb. Patent applications have been filed that relate to the reviewed material, as follows: ‘Compositions and methods for personalized neoplasia vaccines’ (E.F.F. and C.J.W.), ‘Methods for identifying tumor specific neo-antigens’ (C.J.W.), ‘Formulations for neoplasia vaccines’ (E.F.F.), ‘Combination therapy for neoantigen vaccine’ (C.J.W. and E.F.F.), and ‘Multi-domain protein vaccine’ (E.F.F). Figure 1. Natural immunity to an intracellular pathogen. Infected and stressed host cells (top left) release PAMP/DAMPs, activating APCs, like DCs. Activated DCs mature, upregulating phagocytosis, antigen presentation and co-stimulatory molecules, and migrate to a draining lymph node (dLN); lymph drainage also transports local antigens to the dLN. The activated DCs present MHCII restricted peptides to CD4+ T cells, which can in turn license DCs (including via CD40L:CD40 signals; see Inset A) to cross-present antigen on MHCI and activate CD8+ T cells. Cytokines from DCs also shape CD4+ Helper T cell differentiation. DCs, via p:MHCI stimulation of TCR and co-stimulatory signals (Inset B) activate CD8 cells to respond, produce IL2 and IL-12 (for autocrine signaling) and differentiate into cytotoxic T lymphocytes (CTLs). CD4+ T cells potentiate CTL differentiation via IL2, maintain CTL effector function in viral infection via IL21 and induce CD8+ resident memory differentiation via IFNγ (Inset C). CD4+ T cells that recognize cognate p:MHCII on B cells, can provide co-stimulation (Inset D) to support affinity maturation, antibody class switching and plasma cell differentiation. Abbreviations: Pathogen- /Damage- associated molecular pattern (PAMP/DAMP), Dendritic cell (DC), T4 (CD4+ Helper T cell), T8 (CD8+ T cell), CTL (Cytotoxic T lymphocyte), B (B cell), p:MHC (peptide bound multi-histocompatibility complex), TCR (T Cell Receptor). Figure 2: Immunosuppression in the Tumor Immune Microenvironment (TIME). The TIME is made up of malignant cells, stomal cells and immune cells, all of which can suppress tumor directed immunity by acting directly or indirectly on effector T cells (Teff). Boxes depict select immunosuppressive mechanisms employed by these various cell types: cancer cells (Ca), regulatory T cells (Treg), dendritic cells (DC), macrophages (MΦ) and myeloid derived suppressor cells (MDSC; (Veglia et al., 2018)), and cancer associated fibroblasts (CAF). Figure 3: Vaccines, a bridge to cure? While there are many barriers to natural immune responses to cancer, we propose that focusing on four pillars (innate and adaptive immune cells, antigen targets, delivery strategies and co-therapies) will help vaccines become a bridge from cancer to immune-based cure. Figure 4: Neoantigens, our blind spots and their presentation. (A) Potential neoantigen sources in cancer, adapted with modifications from Gupta et. al. (Gupta et al., 2021). Conventional bioinformatic techniques efficiently detect neoantigens derived from single nucleotide variations (SNVs) and insertions/deletions (Indels) (circled in blue). Largely beyond current clinical prediction techniques (red text) are neoantigens from some somatic tandem duplication and fusion transcripts, endogenous retroviruses, transcriptional variations (alternative splicing, exitrons and A-to-I editing), and translational abnormalities (unannotated NuORFs, defective translation products and peptides from alternative start sites). (B) and (C) depict antigen presentation pathways in MHCI and MHCII respectively. (B) For MHCI, the proteasome process endogenous proteins into short peptides, which the TAP complex transports into the endoplasmic reticulum. Immune cells can diversify the peptidome by, processing proteins from phagosomes to allow exogenous peptide cross-presentation (i.e. cDC1s), and/or by immunoproteasome processing, which provides different protease specificities via replacement of three constitutive proteasome subunits with alternative subunits (Murata et al., 2018). In the ER, peptides are further processed and chaperones/editors, including Tapasin, facilitate MHCI loading with stable 8–10 aa peptides, before transport to the cell surface where T cells survey them via TCR:p:MHCI interactions that generally require CD8αβ:MHCI interactions. (C) For MHCII, proteins are processed into smaller peptides in phagosomes, auto-phagosomes (products of autophagy) and lysosomes (Roche and Furuta, 2015), which eventually fuse with the late endosome. There proteolytic cleavage of the invariant chain, CLIP (which facilitates MHCII folding during translation and resides partially in its antigen binding grove), allows loading of 13–25 amino-acid peptides on to MHCII; of note, MHCII is more promiscuous in peptide binding and peptides are considerably longer than for MHCI. p:MHCII complexes can then be edited by HLA-DM, which selects for peptides that are more stably bound to MHCII, before being transported to the cell surface for surveillance by where CD4+ T cell TCRs. In contrast to the importance of CD8 in TCR:p:MHCI interactions, TCR:p:MHCII interactions are not as dependent on CD4 recognition of MHC. Figure 5: Dendritic cells, PAMPs and DAMPs. The diverse dendritic cell lineage expresses a broad array of pattern recognition receptors (PRR). PRRs include Toll like receptors (TLRs), which are present as transmembrane proteins to directly detect PAMPs externally (i.e. cell surface; TLR 1,2,4,5,6) or within endosomes (TLR 3,7,8,9; which focus on nucleic acids released from internalized pathogens) (Fitzgerald and Kagan, 2020); cell surface C-Lectin type Receptors (CLRs; (Geijtenbeek and Gringhuis, 2009)), which directly detect external PAMPs/DAMPs, and in the case of CLEC9A/DNRG1 and DEC205/CLEC13B, can facilitate phagocytosis of dead and dying cells. Intracellular PRRs, including cGAS/STING and RIG-I/MDA5, recognize intracellular nucleic acids. Nod-like receptors (NLR), including NLRP3, can participate in PAMP/DAMP sensing via interactions with caspases and the inflammasome. (Inset A) DC subsets most commonly considered in vaccine production, with lineage defining cell surface markers and transcription factors, differential PRR expression and unique functions. (Inset B) Select PAMPs/DAMPs, the PRRs that detect them, and mimics that are either commonly used as clinical vaccine adjuvants or are under pre-clinical testing. (Inset C) LPS and oxPAPC signaling via caspase-11 highlights differential processing of DAMP and PAMP signals, even via the same PRR pathway. LPS and oxPAPC both induce IL-1β production via caspase 11 and downstream NLRP3/atypical inflammasome signaling. While LPS stimulation leads to short lived, pyroptotic IL-1β-expressing DCs, OxPACP blocks pyroptosis, inducing a hyperactivated state and efficient CTL priming. Figure 6: Measuring success. Ideally a cancer vaccine will demonstrate improved survival or progression free survival (A-i), although cancer vaccine trials are not usually powered to do so. Trials more typically focus on putative correlates of vaccine effectiveness via ex vivo analysis of T cells from the blood or tumor samples (A-ii) or analysis of tumor biopsies (A-iii). (A-ii) T cell assays include: ELISPOT, which enumerates T cell clones responsive to a defined peptide/epitope; flow cytometry based assays, which can identify antigen-specific T cells via p:MHC conjugates (e.g. tetramers), and define T cell function, activation and cytotoxic potential via cytokine, surface 41BB and CD107a exposure; bulk TCR sequencing can temporally track clonotype frequencies; single cell RNA and TCR sequencing (not broadly used in vaccine studies) may enable the linkage of clonotype changes to changes in cell state (e.g. effector, central memory, progenitor exhausted and exhausted). Linking TCR sequence to antigen specificity is not readily possible except through more extensive functional studies. (A-iii) Re-biopsy of tumor can provide evidence of vaccine effectiveness, for example, if genomic or expression analyses demonstrate loss of antigen presenting machinery or neoantigens, or decreased neoantigen expression. Newer spatial multiplexed IHC, in situ hybridization (Nanostring) and sequencing (Slide-seq), may show differences in T cell infiltration patterns (especially when paired with TCR sequencing methods). (B) Schema of epitope spreading. We propose that analysis of epitope spreading may best balance feasibility and sensitivity for vaccine efficacy. Epitope spreading is the concept that one antigen-specific immune response begets another: First, a vaccine primes a CD8+ T cell (purple T8 cell recognizing a purple neoantigen; B-i). That vaccine specific CTL migrates to the tumor (B-ii), where it recognizes and destroys a cancer cell, releasing DAMPs that stimulate a DC maturation and phagocytoses tumor cell remnants, including non-vaccine targeted neoantigens (red). The DC migrates to a draining LN (B-iii), where it presents neoantigens to T cells. A CD8+ T cell (red) with specificity for a new neoantigen is de novo primed and expanded, returning to the tumor (B-iv). B-bottom, model ELISPOT data reflecting epitope spreading is shown, with vaccine induced neoantigen responses appearing after vaccine priming, but non-vaccine neoantigen responses from epitope spreading, appearing only later. Box 1: Vaccine Formats. Protein: Initially recombinant proteins and epitope-length peptides, but now more commonly synthetic long peptides (SLPs, 15–30 amino acids). Advantages include modularity, the favoring of MHCI loading via cross-presentation by professional APCs (e.g. DCs), and the possibility of co-delivering peptide/protein domains that target DCs. Hurdles for SLPs include antigen prediction, technically difficult synthesis and solubility, the lack of intrinsic immunogenicity and difficulty uniting epitopes for T and B cells in one SLP. mRNA: Proven during the COVID-19 pandemic with the rapid production and approval of BNT and mRNA1273 vaccines, mRNAs can encode target antigen(s) and potentially DAMPs, cytokines and co-stimulatory molecules. Certain liposome formulations of mRNAs are preferentially internalized by DCs and mRNA can intrinsically function as adjuvant – in fact, this property must be dampened for optimal efficacy. Self-replicating mRNAs offer extended antigen expression kinetics, potentially mimicking natural infection. Drawbacks include inherent instability. DNA: The DNA platform, usually a plasmid, can encode antigen(s), DAMPs, cytokines and co-stimulatory molecules. DNA vectors tend to induce longer lasting expression than mRNA platforms, potentially augmenting responses. Hurdles include delivery, which requires electroporation or other delivery device, and inability to target to preferred cell types (professional APCs) via the typical intramuscular injection route. Microbial Vector: Weakened or attenuated viral and bacterial vectors can be modified to express neoantigens. They are inherently immunogenic and may be engineered to encode costimulatory molecules to augment responses. Vector directed immune responses can limit effective boosting. Cell Vectors: Most often DCs are differentiated in vitro from blood monocytes (moDCs), stimulated with synthetic PAMPs and loaded with recombinant proteins/peptides or whole tumor lysate. An advantage is that APCs can be directly loaded with antigen and readily manipulated ex vivo. The labor-intensive cell preparation, however, is limiting. Non-moDC cell types are actively being investigated as alternative cellular delivery vehicles. In Situ: Adjuvants, oncolytic viruses, mRNAs encoding immune stimulants and activated autologous or allogeneic dendritic cells can all be injected directly into sites of tumor. This antigen agnostic strategy bypasses the need for antigen prediction and directly modifies the TIME. However, only patients with accessible disease are candidates for treatment. Box 2. Dendritic Cells cDC1: cDC1s are especially effective antigen cross-presenters (Lee and Radford, 2019), express scavenging molecules like CLEC9A which facilitate dead-cell uptake and cross-priming, are central to tumor immunity in murine models (Ferris et al., 2020) and are associated via gene signatures with improved survival in multiple cancer types (Lee and Radford, 2019). Further, cDC1s are critical for priming of both CD4+ and CD8+ T cell responses in syngeneic tumor models (Ferris et al., 2020). cDC2: cDC2s are important for priming CD4 responses (Lee and Radford, 2019). They can also efficiently prime CD4+ helper T cell responses upon Treg depletion in mouse syngeneic models, leading to tumor eradication; in keeping with this, an increased cDC2/Treg ratio in head & neck cancer is associated with increased survival following ICB therapy (Binnewies et al., 2019). Interferon gamma-stimulated cDC2 can also prime effective tumor specific CTL responses in mice by directly acquiring peptide:MHCI complexes from tumor cells – fulfilling the role of cross-presentation (Duong et al., 2021). MoDCs: In vivo, monocyte derived DCs, driven by inflammation, may prime T cell responses after chemotherapy (Wculek et al., 2020) or viral infection. Ex vivo moDCs expansion is routinely used in cell-based vaccine delivery approaches (Nava et al., 2021), with new approaches still under investigation (Han et al., 2020). While adding complexity to vaccine production, ex vivo differentiated moDCs allow direct antigen delivery to the APC, potentially overcoming substantial hurdles compared to in vivo antigen delivery. mregDCs: Recent studies have identified a subset of regulatory DCs which negatively regulate T cell responses to tumors via AXL-dependent PD-L1 upregulation (Maier et al., 2020); interestingly, these mregDCs can also positively regulate T cell responses upon IL4 signaling blockade. Thus, care may be needed to avoid activating the ‘wrong’ DC during vaccination and it will be important to untangle their tolerogenic vs immunogenic programs. pDCs: Plasmacytoid DCs, although ontologically related to cDCs, are typically considered first responders which amplify type I interferon responses. However, they may present, and even cross-present, antigen, or cooperate with other cross-presenting dendritic cells to optimize responses (Fu et al., 2022). Box 3: TIME management strategies Low dose cyclophosphamide can deplete Treg’s, a strategy exploited in DC based vaccines for ovarian cancer (Tanyi et al., 2018), among other vaccine formats. In multiple mouse models, depletion of regulatory T cells is crucial to the activity of anti-CTLA-4 therapy, although in humans this may not be the case for activity (Sharma et al., 2019), and other Treg abrogating approaches are being developed (Chen et al., 2022). A myriad of TAM-targeted therapies are under evaluation (Allavena et al., 2021). Early (and ongoing) trials targeted the CSF-1 receptor or chemokines to block their activity and/or recruitment. More recent approaches attempt to capitalize on macrophage plasticity by re-programming them to anti-tumor phenotypes, targeting molecules like PI3K, STAT3 and IDO-1. Specific targets for MDSC inhibition have been difficult to define, although multiple studies are examining CCR2/CXCR2 axis inhibition; this axis is a generic marker targeting inflamed tissue (Bullock and Richmond, 2021). Targeting fibroblast activation protein (FAP), which marks immune suppressive CAFs, is most advanced clinically (Xin et al., 2021). Cytokines or adjuvants can be targeted to tumors to modify the TIME and recruit immune cells, including IL-2 (Ren et al., 2022), STING agonist (Perera et al., 2021), CCL4 (Williford et al., 2019), mRNA encoding multiple cytokines (Hotz et al., 2021) and photodynamic therapy (Huis In ‘t Veld et al., 2021). Box 4: A box of out-of-the-box thinking With a long history of middling cancer vaccine success, it is time to shake things up. Be inspired by natural immunity and look to whole tumor cell approaches to better capture the entire antigen repertoire. Reverse the epitope information flow by using undefined whole tumor vaccines with a bigger epitope repertoire and reverse engineering to learn what makes effective epitopes. Improve antigen target selection through better prediction of peptide processing (in APCs and cancer cells) or modelling of interactions between peptide epitopes, MHC and the TCR repertoire. What is the best way to get antigen (and activating signals) into APCs? Consider investment in rapid ex vivo preparation of antigen presenting/delivering cells, especially allogeneic cells, to better control cell phenotype and antigen delivery. Include mimics of innate-adaptive immune cell cross-talk in response to natural infections to maximize vaccine responses. CD40-CD40L or CD27–CD70, for example. CTL responses remain a central challenge. Engage CD4+ T cells and B cells with antigen structures to complement and enhance CTL activity. Clarify the origin of effective helper epitopes and antigen structure. Do CD4 helper epitopes need to be tumor-specific or are universal helper epitopes sufficient? Treg biology needs to be better understood – especially the importance of antigen specificity; Does/can vaccination inadvertently enhance/induce Tregs? Exercise patience and humility – not to undermine vaccine potential in late-stage disease, vaccines will undeniably be more effective in early disease, which will eventually reduce the costly and difficult burden of late-stage disease. Funders, regulatory authorities and researchers need to align with this reality and opportunity. Identify a setting - in humans - where delivery dynamics surrounding priming and boosting can be studied with several formats to determine if an optimal schedule exists. Respect the effectiveness and uniqueness of the evolved suppressive TIME. Follow (or precede) a prime/boost vaccination cycle quickly with a ‘bucket’ trial of TIME modulators. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. References Abdel-Hakeem MS , Manne S , Beltra JC , Stelekati E , Chen Z , Nzingha K , Ali MA , Johnson JL , Giles JR , Mathew D , Greenplate AR , Vahedi G , and Wherry EJ (2021). Epigenetic scarring of exhausted T cells hinders memory differentiation upon eliminating chronic antigenic stimulation. Nat Immunol 22 , 1008–1019. 10.1038/s41590-021-00975-5.34312545 Abelin JG , Harjanto D , Malloy M , Suri P , Colson T , Goulding SP , Creech AL , Serrano LR , Nasir G , Nasrullah Y , McGann CD , Velez D , Ting YS , Poran A , Rothenberg DA , Chhangawala S , Rubinsteyn A , Hammerbacher J , Gaynor RB , Fritsch EF , Greshock J , Oslund RC , Barthelme D , Addona TA , Arieta CM , and Rooney MS (2019). Defining HLA-II Ligand Processing and Binding Rules with Mass Spectrometry Enhances Cancer Epitope Prediction. Immunity 51 , 766–779 e717. 10.1016/j.immuni.2019.08.012.31495665 Abusarah J , Khodayarian F , El-Hachem N , Salame N , Olivier M , Balood M , Roversi K , Talbot S , Bikorimana JP , Chen J , Jolicoeur M , Trudeau LE , Kamyabiazar S , Annabi B , Robert F , Pelletier J , El-Kadiry AE , Shammaa R , and Rafei M (2021). Engineering immunoproteasome-expressing mesenchymal stromal cells: A potent cellular vaccine for lymphoma and melanoma in mice. Cell Rep Med 2 , 100455. 10.1016/j.xcrm.2021.100455.35028603 Allavena P , Anfray C , Ummarino A , and Andon FT (2021). Therapeutic Manipulation of Tumor-associated Macrophages: Facts and Hopes from a Clinical and Translational Perspective. Clin Cancer Res 27 , 3291–3297. 10.1158/1078-0432.CCR-20-1679.33531428 Alspach E , Lussier DM , Miceli AP , Kizhvatov I , DuPage M , Luoma AM , Meng W , Lichti CF , Esaulova E , Vomund AN , Runci D , Ward JP , Gubin MM , Medrano RFV , Arthur CD , White JM , Sheehan KCF , Chen A , Wucherpfennig KW , Jacks T , Unanue ER , Artyomov MN , and Schreiber RD (2019). MHC-II neoantigens shape tumour immunity and response to immunotherapy. Nature 574 , 696–701. 10.1038/s41586-019-1671-8.31645760 Anagnostou V , Smith KN , Forde PM , Niknafs N , Bhattacharya R , White J , Zhang T , Adleff V , Phallen J , Wali N , Hruban C , Guthrie VB , Rodgers K , Naidoo J , Kang H , Sharfman W , Georgiades C , Verde F , Illei P , Li QK , Gabrielson E , Brock MV , Zahnow CA , Baylin SB , Scharpf RB , Brahmer JR , Karchin R , Pardoll DM , and Velculescu VE (2017). Evolution of Neoantigen Landscape during Immune Checkpoint Blockade in Non-Small Cell Lung Cancer. Cancer Discov 7 , 264–276. 10.1158/2159-8290.CD-16-0828.28031159 Baharom F , Ramirez-Valdez RA , Tobin KKS , Yamane H , Dutertre CA , Khalilnezhad A , Reynoso GV , Coble VL , Lynn GM , Mule MP , Martins AJ , Finnigan JP , Zhang XM , Hamerman JA , Bhardwaj N , Tsang JS , Hickman HD , Ginhoux F , Ishizuka AS , and Seder RA (2021). Intravenous nanoparticle vaccination generates stem-like TCF1(+) neoantigen-specific CD8(+) T cells. Nat Immunol 22 , 41–52. 10.1038/s41590-020-00810-3.33139915 Bassani-Sternberg M , Braunlein E , Klar R , Engleitner T , Sinitcyn P , Audehm S , Straub M , Weber J , Slotta-Huspenina J , Specht K , Martignoni ME , Werner A , Hein R , D HB , Peschel C , Rad R , Cox J , Mann M , and Krackhardt AM (2016). Direct identification of clinically relevant neoepitopes presented on native human melanoma tissue by mass spectrometry. Nat Commun 7 , 13404. 10.1038/ncomms13404.27869121 Belnoue E , Mayol JF , Carboni S , Di Berardino Besson W , Dupuychaffray E , Nelde A , Stevanovic S , Santiago-Raber ML , Walker PR , and Derouazi M (2019). Targeting self and neoepitopes with a modular self-adjuvanting cancer vaccine. JCI Insight 5 . 10.1172/jci.insight.127305. Bhardwaj N , Friedlander PA , Pavlick AC , Ernstoff MS , Gastman BR , Hanks BA , Curti BD , Albertini MR , Luke JJ , Blazquez AB , Balan S , Bedognetti D , Beechem JM , Crocker AS , D’Amico L , Danaher P , Davis TA , Hawthorne T , Hess BW , Keler T , Lundgren L , Morishima C , Ramchurren N , Rinchai D , Salazar AM , Salim BA , Sharon E , Vitale LA , Wang E , Warren S , Yellin MJ , Disis ML , Cheever MA , and Fling SP (2020). Flt3 ligand augments immune responses to anti-DEC-205-NY-ESO-1 vaccine through expansion of dendritic cell subsets. Nature Cancer 1 , 1204–1217. 10.1038/s43018-020-00143-y.35121932 Binnewies M , Mujal AM , Pollack JL , Combes AJ , Hardison EA , Barry KC , Tsui J , Ruhland MK , Kersten K , Abushawish MA , Spasic M , Giurintano JP , Chan V , Daud AI , Ha P , Ye CJ , Roberts EW , and Krummel MF (2019). Unleashing Type-2 Dendritic Cells to Drive Protective Antitumor CD4(+) T Cell Immunity. Cell 177 , 556–571 e516. 10.1016/j.cell.2019.02.005.30955881 Birnbaum ME , Mendoza JL , Sethi DK , Dong S , Glanville J , Dobbins J , Ozkan E , Davis MM , Wucherpfennig KW , and Garcia KC (2014). Deconstructing the peptide-MHC specificity of T cell recognition. Cell 157 , 1073–1087. 10.1016/j.cell.2014.03.047.24855945 Biswas S , Mandal G , Payne KK , Anadon CM , Gatenbee CD , Chaurio RA , Costich TL , Moran C , Harro CM , Rigolizzo KE , Mine JA , Trillo-Tinoco J , Sasamoto N , Terry KL , Marchion D , Buras A , Wenham RM , Yu X , Townsend MK , Tworoger SS , Rodriguez PC , Anderson AR , and Conejo-Garcia JR (2021). IgA transcytosis and antigen recognition govern ovarian cancer immunity. Nature 591 , 464–470. 10.1038/s41586-020-03144-0.33536615 Bochem J , Zelba H , Spreuer J , Amaral T , Gaissler A , Pop OT , Thiel K , Yurttas C , Soffel D , Forchhammer S , Sinnberg T , Niessner H , Meier F , Terheyden P , Konigsrainer A , Garbe C , Flatz L , Pawelec G , Eigentler TK , Loffler MW , Weide B , and Wistuba-Hamprecht K (2021). Early disappearance of tumor antigen-reactive T cells from peripheral blood correlates with superior clinical outcomes in melanoma under anti-PD-1 therapy. J Immunother Cancer 9 . 10.1136/jitc-2021-003439. Booty MG , Hlavaty KA , Stockmann A , Ozay EI , Smith C , Tian L , How E , Subramanya D , Venkitaraman A , Yee C , Pryor O , Volk K , Blagovic K , Vicente-Suarez I , Yarar D , Myint M , Merino A , Chow J , Abdeljawad T , An H , Liu S , Mao S , Heimann M , Talarico L , Jacques MK , Chong E , Pomerance L , Gonzalez JT , von Andrian UH , Jensen KF , Langer R , Knoetgen H , Trumpfheller C , Umana P , Bernstein H , Sharei A , and Loughhead SM (2022). Microfluidic Squeezing Enables MHC Class I Antigen Presentation by Diverse Immune Cells to Elicit CD8(+) T Cell Responses with Antitumor Activity. J Immunol. 10.4049/jimmunol.2100656. Bullock K , and Richmond A (2021). Suppressing MDSC Recruitment to the Tumor Microenvironment by Antagonizing CXCR2 to Enhance the Efficacy of Immunotherapy. Cancers (Basel) 13 . 10.3390/cancers13246293. Burger ML , Cruz AM , Crossland GE , Gaglia G , Ritch CC , Blatt SE , Bhutkar A , Canner D , Kienka T , Tavana SZ , Barandiaran AL , Garmilla A , Schenkel JM , Hillman M , de Los Rios Kobara I , Li A , Jaeger AM , Hwang WL , Westcott PMK , Manos MP , Holovatska MM , Hodi FS , Regev A , Santagata S , and Jacks T (2021). Antigen dominance hierarchies shape TCF1(+) progenitor CD8 T cell phenotypes in tumors. Cell 184 , 4996–5014 e4926. 10.1016/j.cell.2021.08.020.34534464 Cachot A , Bilous M , Liu YC , Li X , Saillard M , Cenerenti M , Rockinger GA , Wyss T , Guillaume P , Schmidt J , Genolet R , Ercolano G , Protti MP , Reith W , Ioannidou K , de Leval L , Trapani JA , Coukos G , Harari A , Speiser DE , Mathis A , Gfeller D , Altug H , Romero P , and Jandus C (2021). Tumor-specific cytolytic CD4 T cells mediate immunity against human cancer. Sci Adv 7 . 10.1126/sciadv.abe3348. Caushi JX , Zhang J , Ji Z , Vaghasia A , Zhang B , Hsiue EH , Mog BJ , Hou W , Justesen S , Blosser R , Tam A , Anagnostou V , Cottrell TR , Guo H , Chan HY , Singh D , Thapa S , Dykema AG , Burman P , Choudhury B , Aparicio L , Cheung LS , Lanis M , Belcaid Z , El Asmar M , Illei PB , Wang R , Meyers J , Schuebel K , Gupta A , Skaist A , Wheelan S , Naidoo J , Marrone KA , Brock M , Ha J , Bush EL , Park BJ , Bott M , Jones DR , Reuss JE , Velculescu VE , Chaft JE , Kinzler KW , Zhou S , Vogelstein B , Taube JM , Hellmann MD , Brahmer JR , Merghoub T , Forde PM , Yegnasubramanian S , Ji H , Pardoll DM , and Smith KN (2021). Transcriptional programs of neoantigen-specific TIL in anti-PD-1-treated lung cancers. Nature 596 , 126–132. 10.1038/s41586-021-03752-4.34290408 Chen BJ , Zhao JW , Zhang DH , Zheng AH , and Wu GQ (2022). Immunotherapy of Cancer by Targeting Regulatory T cells. Int Immunopharmacol 104 , 108469. 10.1016/j.intimp.2021.108469.35008005 Chen DS , and Mellman I (2013). Oncology meets immunology: the cancer-immunity cycle. Immunity 39 , 1–10. 10.1016/j.immuni.2013.07.012.23890059 Cheng X , Zhang H , Hamad A , Huang H , and Tsung A (2022). Surgery-mediated tumor-promoting effects on the immune microenvironment. Semin Cancer Biol. 10.1016/j.semcancer.2022.01.006. Cui C , Wang J , Fagerberg E , Chen PM , Connolly KA , Damo M , Cheung JF , Mao T , Askari AS , Chen S , Fitzgerald B , Foster GG , Eisenbarth SC , Zhao H , Craft J , and Joshi NS (2021). Neoantigen-driven B cell and CD4 T follicular helper cell collaboration promotes anti-tumor CD8 T cell responses. Cell 184 , 6101–6118 e6113. 10.1016/j.cell.2021.11.007.34852236 Danilova L , Anagnostou V , Caushi JX , Sidhom JW , Guo H , Chan HY , Suri P , Tam A , Zhang J , Asmar ME , Marrone KA , Naidoo J , Brahmer JR , Forde PM , Baras AS , Cope L , Velculescu VE , Pardoll DM , Housseau F , and Smith KN (2018). The Mutation-Associated Neoantigen Functional Expansion of Specific T Cells (MANAFEST) Assay: A Sensitive Platform for Monitoring Antitumor Immunity. Cancer Immunol Res 6 , 888–899. 10.1158/2326-6066.CIR-18-0129.29895573 Desbois M , and Wang Y (2021). Cancer-associated fibroblasts: Key players in shaping the tumor immune microenvironment. Immunol Rev 302 , 241–258. 10.1111/imr.12982.34075584 Diehl L , den Boer AT , Schoenberger SP , van der Voort EI , Schumacher TN , Melief CJ , Offringa R , and Toes RE (1999). CD40 activation in vivo overcomes peptide-induced peripheral cytotoxic T-lymphocyte tolerance and augments anti-tumor vaccine efficacy. Nat Med 5 , 774–779. 10.1038/10495.10395322 Downs-Canner SM , Meier J , Vincent BG , and Serody JS (2022). B Cell Function in the Tumor Microenvironment. Annu Rev Immunol. 10.1146/annurev-immunol-101220-015603. Dries R , Chen J , Del Rossi N , Khan MM , Sistig A , and Yuan GC (2021). Advances in spatial transcriptomic data analysis. Genome Res 31 , 1706–1718. 10.1101/gr.275224.121.34599004 Dunn GP , Bruce AT , Ikeda H , Old LJ , and Schreiber RD (2002). Cancer immunoediting: from immunosurveillance to tumor escape. Nat Immunol 3 , 991–998. 10.1038/ni1102-991.12407406 Duong E , Fessenden TB , Lutz E , Dinter T , Yim L , Blatt S , Bhutkar A , Wittrup KD , and Spranger S (2021). Type I interferon activates MHC class I-dressed CD11b(+) conventional dendritic cells to promote protective anti-tumor CD8(+) T cell immunity. Immunity. 10.1016/j.immuni.2021.10.020. Ebrahimi-Nik H , Moussa M , Englander RP , Singhaviranon S , Michaux J , Pak H , Miyadera H , Corwin WL , Keller GLJ , Hagymasi AT , Shcheglova TV , Coukos G , Baker BM , Mandoiu II , Bassani-Sternberg M , and Srivastava PK (2021). Reversion analysis reveals the in vivo immunogenicity of a poorly MHC I-binding cancer neoepitope. Nat Commun 12 , 6423. 10.1038/s41467-021-26646-5.34741035 Fagone P , Shedlock DJ , Bao H , Kawalekar OU , Yan J , Gupta D , Morrow MP , Patel A , Kobinger GP , Muthumani K , and Weiner DB (2011). Molecular adjuvant HMGB1 enhances anti-influenza immunity during DNA vaccination. Gene Ther 18 , 1070–1077. 10.1038/gt.2011.59.21544096 Ferris ST , Durai V , Wu R , Theisen DJ , Ward JP , Bern MD , Davidson J.T.t. , Bagadia P , Liu T , Briseno CG , Li L , Gillanders WE , Wu GF , Yokoyama WM , Murphy TL , Schreiber RD , and Murphy KM (2020). cDC1 prime and are licensed by CD4(+) T cells to induce anti-tumour immunity. Nature 584 , 624–629. 10.1038/s41586-020-2611-3.32788723 Ferrucci PF , Pala L , Conforti F , and Cocorocchio E (2021). Talimogene Laherparepvec (T-VEC): An Intralesional Cancer Immunotherapy for Advanced Melanoma. Cancers (Basel) 13 . 10.3390/cancers13061383. Finn OJ (2018). The dawn of vaccines for cancer prevention. Nat Rev Immunol 18 , 183–194. 10.1038/nri.2017.140.29279613 Fitzgerald KA , and Kagan JC (2020). Toll-like Receptors and the Control of Immunity. Cell 180 , 1044–1066. 10.1016/j.cell.2020.02.041.32164908 Fletcher EAK , van Maren W , Cordfunke R , Dinkelaar J , Codee JDC , van der Marel G , Melief CJM , Ossendorp F , Drijfhout JW , and Mangsbo SM (2018). Formation of Immune Complexes with a Tetanus-Derived B Cell Epitope Boosts Human T Cell Responses to Covalently Linked Peptides in an Ex Vivo Blood Loop System. J Immunol 201 , 87–97. 10.4049/jimmunol.1700911.29752315 Fluckiger A , Daillere R , Sassi M , Sixt BS , Liu P , Loos F , Richard C , Rabu C , Alou MT , Goubet AG , Lemaitre F , Ferrere G , Derosa L , Duong CPM , Messaoudene M , Gagne A , Joubert P , De Sordi L , Debarbieux L , Simon S , Scarlata CM , Ayyoub M , Palermo B , Facciolo F , Boidot R , Wheeler R , Boneca IG , Sztupinszki Z , Papp K , Csabai I , Pasolli E , Segata N , Lopez-Otin C , Szallasi Z , Andre F , Iebba V , Quiniou V , Klatzmann D , Boukhalil J , Khelaifia S , Raoult D , Albiges L , Escudier B , Eggermont A , Mami-Chouaib F , Nistico P , Ghiringhelli F , Routy B , Labarriere N , Cattoir V , Kroemer G , and Zitvogel L (2020). Cross-reactivity between tumor MHC class I-restricted antigens and an enterococcal bacteriophage. Science 369 , 936–942. 10.1126/science.aax0701.32820119 Fossum E , Grodeland G , Terhorst D , Tveita AA , Vikse E , Mjaaland S , Henri S , Malissen B , and Bogen B (2015). Vaccine molecules targeting Xcr1 on cross-presenting DCs induce protective CD8+ T-cell responses against influenza virus. Eur J Immunol 45 , 624–635. 10.1002/eji.201445080.25410055 Fritsch EF , Burkhardt UE , Hacohen N , and Wu CJ (2020). Personal Neoantigen Cancer Vaccines: A Road Not Fully Paved. Cancer Immunol Res 8 , 1465–1469. 10.1158/2326-6066.CIR-20-0526.33262163 Fu C , Zhou L , Mi QS , and Jiang A (2022). Plasmacytoid Dendritic Cells and Cancer Immunotherapy. Cells 11 . 10.3390/cells11020222. Geijtenbeek TB , and Gringhuis SI (2009). Signalling through C-type lectin receptors: shaping immune responses. Nat Rev Immunol 9 , 465–479. 10.1038/nri2569.19521399 Grant SM , Lou M , Yao L , Germain RN , and Radtke AJ (2020). The lymph node at a glance - how spatial organization optimizes the immune response. J Cell Sci 133 . 10.1242/jcs.241828. Grotzke JE , Sengupta D , Lu Q , and Cresswell P (2017). The ongoing saga of the mechanism(s) of MHC class I-restricted cross-presentation. Curr Opin Immunol 46 , 89–96. 10.1016/j.coi.2017.03.015.28528219 Gupta RG , Li F , Roszik J , and Lizee G (2021). Exploiting Tumor Neoantigens to Target Cancer Evolution: Current Challenges and Promising Therapeutic Approaches. Cancer Discov 11 , 1024–1039. 10.1158/2159-8290.CD-20-1575.33722796 Hammerich L , Bhardwaj N , Kohrt HE , and Brody JD (2016). In situ vaccination for the treatment of cancer. Immunotherapy 8 , 315–330. 10.2217/imt.15.120.26860335 Han P , Hanlon D , Arshad N , Lee JS , Tatsuno K , Robinson E , Filler R , Sobolev O , Cote C , Rivera-Molina F , Toomre D , Fahmy T , and Edelson R (2020). Platelet P-selectin initiates cross-presentation and dendritic cell differentiation in blood monocytes. Sci Adv 6 , eaaz1580. 10.1126/sciadv.aaz1580. Hanel G , Angerer C , Petry K , Lichtenegger FS , and Subklewe M (2021). Blood DCs activated with R848 and poly(I:C) induce antigen-specific immune responses against viral and tumor-associated antigens. Cancer Immunol Immunother. 10.1007/s00262-021-03109-w. Harndahl M , Rasmussen M , Roder G , and Buus S (2011). Real-time, high-throughput measurements of peptide-MHC-I dissociation using a scintillation proximity assay. J Immunol Methods 374 , 5–12. 10.1016/j.jim.2010.10.012.21044632 Hegde S , Leader AM , and Merad M (2021). MDSC: Markers, development, states, and unaddressed complexity. Immunity 54 , 875–884. 10.1016/j.immuni.2021.04.004.33979585 Hernandez AP , Juanes-Velasco P , Landeira-Vinuela A , Bareke H , Montalvillo E , Gongora R , and Fuentes M (2021). Restoring the Immunity in the Tumor Microenvironment: Insights into Immunogenic Cell Death in Onco-Therapies. Cancers (Basel) 13 . 10.3390/cancers13112821. Hofer T , Rossi M , Carboni S , Di Berardino Besson W , von Laer D , Wollmann G , Derouazi M , and Santiago-Raber ML (2021). Heterologous Prime-Boost Vaccination with a Peptide-Based Vaccine and Viral Vector Reshapes Dendritic Cell, CD4+ and CD8+ T Cell Phenotypes to Improve the Antitumor Therapeutic Effect. Cancers (Basel) 13 . 10.3390/cancers13236107. Hotz C , Wagenaar TR , Gieseke F , Bangari DS , Callahan M , Cao H , Diekmann J , Diken M , Grunwitz C , Hebert A , Hsu K , Bernardo M , Kariko K , Kreiter S , Kuhn AN , Levit M , Malkova N , Masciari S , Pollard J , Qu H , Ryan S , Selmi A , Schlereth J , Singh K , Sun F , Tillmann B , Tolstykh T , Weber W , Wicke L , Witzel S , Yu Q , Zhang YA , Zheng G , Lager J , Nabel GJ , Sahin U , and Wiederschain D (2021). Local delivery of mRNA-encoded cytokines promotes antitumor immunity and tumor eradication across multiple preclinical tumor models. Sci Transl Med 13 , eabc7804. 10.1126/scitranslmed.abc7804. Hu Z , Leet DE , Allesoe RL , Oliveira G , Li S , Luoma AM , Liu J , Forman J , Huang T , Iorgulescu JB , Holden R , Sarkizova S , Gohil SH , Redd RA , Sun J , Elagina L , Giobbie-Hurder A , Zhang W , Peter L , Ciantra Z , Rodig S , Olive O , Shetty K , Pyrdol J , Uduman M , Lee PC , Bachireddy P , Buchbinder EI , Yoon CH , Neuberg D , Pentelute BL , Hacohen N , Livak KJ , Shukla SA , Olsen LR , Barouch DH , Wucherpfennig KW , Fritsch EF , Keskin DB , Wu CJ , and Ott PA (2021). Personal neoantigen vaccines induce persistent memory T cell responses and epitope spreading in patients with melanoma. Nat Med 27 , 515–525. 10.1038/s41591-020-01206-4.33479501 Huang H , Wang Z , Zhang Y , Pradhan RN , Ganguly D , Chandra R , Murimwa G , Wright S , Gu X , Maddipati R , Muller S , Turley SJ , and Brekken RA (2022). Mesothelial cell-derived antigen-presenting cancer-associated fibroblasts induce expansion of regulatory T cells in pancreatic cancer. Cancer Cell. 10.1016/j.ccell.2022.04.011. Huis In ‘t Veld RV , Da Silva CG , Jager MJ , Cruz LJ , and Ossendorp F (2021). Combining Photodynamic Therapy with Immunostimulatory Nanoparticles Elicits Effective Anti-Tumor Immune Responses in Preclinical Murine Models. Pharmaceutics 13 . 10.3390/pharmaceutics13091470. Irvine DJ , Aung A , and Silva M (2020). Controlling timing and location in vaccines. Adv Drug Deliv Rev 158 , 91–115. 10.1016/j.addr.2020.06.019.32598970 Jhunjhunwala S , Hammer C , and Delamarre L (2021). Antigen presentation in cancer: insights into tumour immunogenicity and immune evasion. Nat Rev Cancer 21 , 298–312. 10.1038/s41568-021-00339-z.33750922 Johansen P , Storni T , Rettig L , Qiu Z , Der-Sarkissian A , Smith KA , Manolova V , Lang KS , Senti G , Mullhaupt B , Gerlach T , Speck RF , Bot A , and Kundig TM (2008). Antigen kinetics determines immune reactivity. Proc Natl Acad Sci U S A 105 , 5189–5194. 10.1073/pnas.0706296105.18362362 Kok L , Masopust D , and Schumacher TN (2022). The precursors of CD8(+) tissue resident memory T cells: from lymphoid organs to infected tissues. Nat Rev Immunol 22 , 283–293. 10.1038/s41577-021-00590-3.34480118 Korsholm KS , Hansen J , Karlsen K , Filskov J , Mikkelsen M , Lindenstrom T , Schmidt ST , Andersen P , and Christensen D (2014). Induction of CD8+ T-cell responses against subunit antigens by the novel cationic liposomal CAF09 adjuvant. Vaccine 32 , 3927–3935. 10.1016/j.vaccine.2014.05.050.24877765 Kranz LM , Diken M , Haas H , Kreiter S , Loquai C , Reuter KC , Meng M , Fritz D , Vascotto F , Hefesha H , Grunwitz C , Vormehr M , Husemann Y , Selmi A , Kuhn AN , Buck J , Derhovanessian E , Rae R , Attig S , Diekmann J , Jabulowsky RA , Heesch S , Hassel J , Langguth P , Grabbe S , Huber C , Tureci O , and Sahin U (2016). Systemic RNA delivery to dendritic cells exploits antiviral defence for cancer immunotherapy. Nature 534 , 396–401. 10.1038/nature18300.27281205 Kristensen NP , Heeke C , Tvingsholm SA , Borch A , Draghi A , Crowther MD , Carri I , Munk KK , Holm JS , Bjerregaard AM , Bentzen AK , Marquard AM , Szallasi Z , McGranahan N , Andersen R , Nielsen M , Jonsson GB , Donia M , Svane IM , and Hadrup SR (2022). Neoantigen-reactive CD8+ T cells affect clinical outcome of adoptive cell therapy with tumor-infiltrating lymphocytes in melanoma. J Clin Invest 132 . 10.1172/JCI150535. Kumar A , Watkins R , and Vilgelm AE (2021). Cell Therapy With TILs: Training and Taming T Cells to Fight Cancer. Front Immunol 12 , 690499. 10.3389/fimmu.2021.690499.34140957 Kumar BV , Connors TJ , and Farber DL (2018). Human T Cell Development, Localization, and Function throughout Life. Immunity 48 , 202–213. 10.1016/j.immuni.2018.01.007.29466753 Lal JC , Townsend MG , Mehta AK , Oliwa M , Miller E , Sotayo A , Cheney E , Mittendorf EA , Letai A , and Guerriero JL (2021). Comparing syngeneic and autochthonous models of breast cancer to identify tumor immune components that correlate with response to immunotherapy in breast cancer. Breast Cancer Res 23 , 83. 10.1186/s13058-021-01448-1.34353349 Lam H , McNeil LK , Starobinets H , DeVault VL , Cohen RB , Twardowski P , Johnson ML , Gillison ML , Stein MN , Vaishampayan UN , DeCillis AP , Foti JJ , Vemulapalli V , Tjon E , Ferber K , DeOliveira DB , Broom W , Agnihotri P , Jaffee EM , Wong KK , Drake CG , Carroll PM , Davis TA , and Flechtner JB (2021). An Empirical Antigen Selection Method Identifies Neoantigens That Either Elicit Broad Antitumor T-cell Responses or Drive Tumor Growth. Cancer Discov 11 , 696–713. 10.1158/2159-8290.CD-20-0377.33504579 Lee YS , and Radford KJ (2019). The role of dendritic cells in cancer. Int Rev Cell Mol Biol 348 , 123–178. 10.1016/bs.ircmb.2019.07.006.31810552 Legoux FP , Lim JB , Cauley AW , Dikiy S , Ertelt J , Mariani TJ , Sparwasser T , Way SS , and Moon JJ (2015). CD4+ T Cell Tolerance to Tissue-Restricted Self Antigens Is Mediated by Antigen-Specific Regulatory T Cells Rather Than Deletion. Immunity 43 , 896–908. 10.1016/j.immuni.2015.10.011.26572061 Li Z , Tuong ZK , Dean I , Willis C , Gaspal F , Fiancette R , Idris S , Kennedy B , Ferdinand JR , Penalver A , Cabantous M , Murtuza Baker S , Fry JW , Carlesso G , Hammond SA , Dovedi SJ , Hepworth MR , Clatworthy MR , and Withers DR (2022). In vivo labeling reveals continuous trafficking of TCF-1+ T cells between tumor and lymphoid tissue. J Exp Med 219 . 10.1084/jem.20210749. Liu B , Hu X , Feng K , Gao R , Xue Z , Zhang S , Zhang Y , Corse E , Hu Y , Han W , and Zhang Z (2021). Temporal single-cell tracing reveals clonal revival and expansion of precursor exhausted T cells during anti-PD-1 therapy in lung cancer. Nature Cancer 3 , 108–121. 10.1038/s43018-021-00292-8.35121991 Lynn GM , Sedlik C , Baharom F , Zhu Y , Ramirez-Valdez RA , Coble VL , Tobin K , Nichols SR , Itzkowitz Y , Zaidi N , Gammon JM , Blobel NJ , Denizeau J , de la Rochere P , Francica BJ , Decker B , Maciejewski M , Cheung J , Yamane H , Smelkinson MG , Francica JR , Laga R , Bernstock JD , Seymour LW , Drake CG , Jewell CM , Lantz O , Piaggio E , Ishizuka AS , and Seder RA (2020). Peptide-TLR-7/8a conjugate vaccines chemically programmed for nanoparticle self-assembly enhance CD8 T-cell immunity to tumor antigens. Nat Biotechnol 38 , 320–332. 10.1038/s41587-019-0390-x.31932728 Maier B , Leader AM , Chen ST , Tung N , Chang C , LeBerichel J , Chudnovskiy A , Maskey S , Walker L , Finnigan JP , Kirkling ME , Reizis B , Ghosh S , D’Amore NR , Bhardwaj N , Rothlin CV , Wolf A , Flores R , Marron T , Rahman AH , Kenigsberg E , Brown BD , and Merad M (2020). A conserved dendritic-cell regulatory program limits antitumour immunity. Nature 580 , 257–262. 10.1038/s41586-020-2134-y.32269339 Mehta AK , Kadel S , Townsend MG , Oliwa M , and Guerriero JL (2021). Macrophage Biology and Mechanisms of Immune Suppression in Breast Cancer. Front Immunol 12 , 643771. 10.3389/fimmu.2021.643771.33968034 Melo M , Porter E , Zhang Y , Silva M , Li N , Dobosh B , Liguori A , Skog P , Landais E , Menis S , Sok D , Nemazee D , Schief WR , Weiss R , and Irvine DJ (2019). Immunogenicity of RNA Replicons Encoding HIV Env Immunogens Designed for Self-Assembly into Nanoparticles. Mol Ther 27 , 2080–2090. 10.1016/j.ymthe.2019.08.007.31515132 Menzies AM , Amaria RN , Rozeman EA , Huang AC , Tetzlaff MT , van de Wiel BA , Lo S , Tarhini AA , Burton EM , Pennington TE , Saw RPM , Xu X , Karakousis GC , Ascierto PA , Spillane AJ , van Akkooi ACJ , Davies MA , Mitchell TC , Tawbi HA , Scolyer RA , Wargo JA , Blank CU , and Long GV (2021). Pathological response and survival with neoadjuvant therapy in melanoma: a pooled analysis from the International Neoadjuvant Melanoma Consortium (INMC). Nat Med 27 , 301–309. 10.1038/s41591-020-01188-3.33558722 Moorman CD , Sohn SJ , and Phee H (2021). Emerging Therapeutics for Immune Tolerance: Tolerogenic Vaccines, T cell Therapy, and IL-2 Therapy. Front Immunol 12 , 657768. 10.3389/fimmu.2021.657768.33854514 Morrison AH , Diamond MS , Hay CA , Byrne KT , and Vonderheide RH (2020). Sufficiency of CD40 activation and immune checkpoint blockade for T cell priming and tumor immunity. Proc Natl Acad Sci U S A 117 , 8022–8031. 10.1073/pnas.1918971117.32213589 Murata S , Takahama Y , Kasahara M , and Tanaka K (2018). The immunoproteasome and thymoproteasome: functions, evolution and human disease. Nat Immunol 19 , 923–931. 10.1038/s41590-018-0186-z.30104634 Mysore V , Cullere X , Mears J , Rosetti F , Okubo K , Liew PX , Zhang F , Madera-Salcedo I , Rosenbauer F , Stone RM , Aster JC , von Andrian UH , Lichtman AH , Raychaudhuri S , and Mayadas TN (2021). FcgammaR engagement reprograms neutrophils into antigen cross-presenting cells that elicit acquired anti-tumor immunity. Nat Commun 12 , 4791. 10.1038/s41467-021-24591-x.34373452 Nava S , Lisini D , Frigerio S , and Bersano A (2021). Dendritic Cells and Cancer Immunotherapy: The Adjuvant Effect. Int J Mol Sci 22 . 10.3390/ijms222212339. Nielsen M , Lundegaard C , Blicher T , Lamberth K , Harndahl M , Justesen S , Roder G , Peters B , Sette A , Lund O , and Buus S (2007). NetMHCpan, a method for quantitative predictions of peptide binding to any HLA-A and -B locus protein of known sequence. PLoS One 2 , e796. 10.1371/journal.pone.0000796.17726526 Oliveira G , Stromhaug K , Cieri N , Iorgulescu JB , Klaeger S , Wolff JO , Rachimi S , Chea V , Krause K , Freeman SS , Zhang W , Li S , Braun DA , Neuberg D , Carr SA , Livak KJ , Frederick DT , Fritsch EF , Wind-Rotolo M , Hacohen N , Sade-Feldman M , Yoon CH , Keskin DB , Ott PA , Rodig SJ , Boland GM , and Wu CJ (2022). Landscape of helper and regulatory antitumour CD4(+) T cells in melanoma. Nature. 10.1038/s41586-022-04682-5. Oliveira G , Stromhaug K , Klaeger S , Kula T , Frederick DT , Le PM , Forman J , Huang T , Li S , Zhang W , Xu Q , Cieri N , Clauser KR , Shukla SA , Neuberg D , Justesen S , MacBeath G , Carr SA , Fritsch EF , Hacohen N , Sade-Feldman M , Livak KJ , Boland GM , Ott PA , Keskin DB , and Wu CJ (2021). Phenotype, specificity and avidity of antitumour CD8(+) T cells in melanoma. Nature 596 , 119–125. 10.1038/s41586-021-03704-y.34290406 Ott PA , Hu Z , Keskin DB , Shukla SA , Sun J , Bozym DJ , Zhang W , Luoma A , Giobbie-Hurder A , Peter L , Chen C , Olive O , Carter TA , Li S , Lieb DJ , Eisenhaure T , Gjini E , Stevens J , Lane WJ , Javeri I , Nellaiappan K , Salazar AM , Daley H , Seaman M , Buchbinder EI , Yoon CH , Harden M , Lennon N , Gabriel S , Rodig SJ , Barouch DH , Aster JC , Getz G , Wucherpfennig K , Neuberg D , Ritz J , Lander ES , Fritsch EF , Hacohen N , and Wu CJ (2017). An immunogenic personal neoantigen vaccine for patients with melanoma. Nature 547 , 217–221. 10.1038/nature22991.28678778 Ott PA , Hu-Lieskovan S , Chmielowski B , Govindan R , Naing A , Bhardwaj N , Margolin K , Awad MM , Hellmann MD , Lin JJ , Friedlander T , Bushway ME , Balogh KN , Sciuto TE , Kohler V , Turnbull SJ , Besada R , Curran RR , Trapp B , Scherer J , Poran A , Harjanto D , Barthelme D , Ting YS , Dong JZ , Ware Y , Huang Y , Huang Z , Wanamaker A , Cleary LD , Moles MA , Manson K , Greshock J , Khondker ZS , Fritsch E , Rooney MS , DeMario M , Gaynor RB , and Srinivasan L (2020). A Phase Ib Trial of Personalized Neoantigen Therapy Plus Anti-PD-1 in Patients with Advanced Melanoma, Non-small Cell Lung Cancer, or Bladder Cancer. Cell 183 , 347–362 e324. 10.1016/j.cell.2020.08.053.33064988 Ouspenskaia T , Law T , Clauser KR , Klaeger S , Sarkizova S , Aguet F , Li B , Christian E , Knisbacher BA , Le PM , Hartigan CR , Keshishian H , Apffel A , Oliveira G , Zhang W , Chen S , Chow YT , Ji Z , Jungreis I , Shukla SA , Justesen S , Bachireddy P , Kellis M , Getz G , Hacohen N , Keskin DB , Carr SA , Wu CJ , and Regev A (2021). Unannotated proteins expand the MHC-I-restricted immunopeptidome in cancer. Nat Biotechnol. 10.1038/s41587-021-01021-3. Ouyang X , Telli ML , and Wu JC (2019). Induced Pluripotent Stem Cell-Based Cancer Vaccines. Front Immunol 10 , 1510. 10.3389/fimmu.2019.01510.31338094 Pan Y , Liu L , Tian T , Zhao J , Park CO , Lofftus SY , Stingley CA , Yan Y , Mei S , Liu X , and Kupper TS (2021). Epicutaneous immunization with modified vaccinia Ankara viral vectors generates superior T cell immunity against a respiratory viral challenge. NPJ Vaccines 6 , 1. 10.1038/s41541-020-00265-5.33398010 Pauken KE , Godec J , Odorizzi PM , Brown KE , Yates KB , Ngiow SF , Burke KP , Maleri S , Grande SM , Francisco LM , Ali MA , Imam S , Freeman GJ , Haining WN , Wherry EJ , and Sharpe AH (2020). The PD-1 Pathway Regulates Development and Function of Memory CD8(+) T Cells following Respiratory Viral Infection. Cell Rep 31 , 107827. 10.1016/j.celrep.2020.107827.32610128 Pauthner M , Havenar-Daughton C , Sok D , Nkolola JP , Bastidas R , Boopathy AV , Carnathan DG , Chandrashekar A , Cirelli KM , Cottrell CA , Eroshkin AM , Guenaga J , Kaushik K , Kulp DW , Liu J , McCoy LE , Oom AL , Ozorowski G , Post KW , Sharma SK , Steichen JM , de Taeye SW , Tokatlian T , Torrents de la Pena A , Butera ST , LaBranche CC , Montefiori DC , Silvestri G , Wilson IA , Irvine DJ , Sanders RW , Schief WR , Ward AB , Wyatt RT , Barouch DH , Crotty S , and Burton DR (2017). Elicitation of Robust Tier 2 Neutralizing Antibody Responses in Nonhuman Primates by HIV Envelope Trimer Immunization Using Optimized Approaches. Immunity 46 , 1073–1088 e1076. 10.1016/j.immuni.2017.05.007.28636956 Perera SA , Kopinja JE , Ma Y , Muise ES , Laskey J , Chakravarthy K , Chen Y , Cui L , Presland J , Sathe M , Javaid S , Minnihan E , Ferguson H , Piesvaux J , Pan BS , Zhao S , Sharma SK , Woo HC , Pucci V , Knemeyer I , Cemerski S , Cumming J , Trotter BW , Tse A , Khilnani A , Ranganath S , Long BJ , Bennett DJ , and Addona GH (2021). STimulator of INterferon Genes Agonism Accelerates Anti-tumor Activity in Poorly Immunogenic Tumors. Mol Cancer Ther. 10.1158/1535-7163.MCT-21-0136. Philip M , and Schietinger A (2021). CD8(+) T cell differentiation and dysfunction in cancer. Nat Rev Immunol. 10.1038/s41577-021-00574-3. Potocnakova L , Bhide M , and Pulzova LB (2016). An Introduction to B-Cell Epitope Mapping and In Silico Epitope Prediction. J Immunol Res 2016 , 6760830. 10.1155/2016/6760830.28127568 Pozzi C , Cuomo A , Spadoni I , Magni E , Silvola A , Conte A , Sigismund S , Ravenda PS , Bonaldi T , Zampino MG , Cancelliere C , Di Fiore PP , Bardelli A , Penna G , and Rescigno M (2016). The EGFR-specific antibody cetuximab combined with chemotherapy triggers immunogenic cell death. Nat Med 22 , 624–631. 10.1038/nm.4078.27135741 Propper DJ , and Balkwill FR (2022). Harnessing cytokines and chemokines for cancer therapy. Nat Rev Clin Oncol. 10.1038/s41571-021-00588-9. Rakhra K , Abraham W , Wang C , Moynihan KD , Li N , Donahue N , Baldeon AD , and Irvine DJ (2021). Exploiting albumin as a mucosal vaccine chaperone for robust generation of lung-resident memory T cells. Sci Immunol 6 . 10.1126/sciimmunol.abd8003. Reiser J , and Banerjee A (2016). Effector, Memory, and Dysfunctional CD8(+) T Cell Fates in the Antitumor Immune Response. J Immunol Res 2016 , 8941260. 10.1155/2016/8941260.27314056 Ren Z , Zhang A , Sun Z , Liang Y , Ye J , Qiao J , Li B , and Fu YX (2022). Selective delivery of low-affinity IL-2 to PD-1+ T cells rejuvenates antitumor immunity with reduced toxicity. J Clin Invest 132 . 10.1172/JCI153604. Reynisson B , Alvarez B , Paul S , Peters B , and Nielsen M (2020). NetMHCpan-4.1 and NetMHCIIpan-4.0: improved predictions of MHC antigen presentation by concurrent motif deconvolution and integration of MS MHC eluted ligand data. Nucleic Acids Res 48 , W449–W454. 10.1093/nar/gkaa379.32406916 Riccione KA , He LZ , Fecci PE , Norberg PK , Suryadevara CM , Swartz A , Healy P , Reap E , Keler T , Li QJ , Congdon KL , Sanchez-Perez L , and Sampson JH (2018). CD27 stimulation unveils the efficacy of linked class I/II peptide vaccines in poorly immunogenic tumors by orchestrating a coordinated CD4/CD8 T cell response. Oncoimmunology 7 , e1502904. 10.1080/2162402X.2018.1502904.30524899 Richman LP , Vonderheide RH , and Rech AJ (2019). Neoantigen Dissimilarity to the Self-Proteome Predicts Immunogenicity and Response to Immune Checkpoint Blockade. Cell Syst 9 , 375–382 e374. 10.1016/j.cels.2019.08.009.31606370 Roche PA , and Furuta K (2015). The ins and outs of MHC class II-mediated antigen processing and presentation. Nat Rev Immunol 15 , 203–216. 10.1038/nri3818.25720354 Rosenblatt J , Stone RM , Uhl L , Neuberg D , Joyce R , Levine JD , Arnason J , McMasters M , Luptakova K , Jain S , Zwicker JI , Hamdan A , Boussiotis V , Steensma DP , DeAngelo DJ , Galinsky I , Dutt PS , Logan E , Bryant MP , Stroopinsky D , Werner L , Palmer K , Coll M , Washington A , Cole L , Kufe D , and Avigan D (2016). Individualized vaccination of AML patients in remission is associated with induction of antileukemia immunity and prolonged remissions. Sci Transl Med 8 , 368ra171. 10.1126/scitranslmed.aag1298. Rozeman EA , Hoefsmit EP , Reijers ILM , Saw RPM , Versluis JM , Krijgsman O , Dimitriadis P , Sikorska K , van de Wiel BA , Eriksson H , Gonzalez M , Torres Acosta A , Grijpink-Ongering LG , Shannon K , Haanen J , Stretch J , Ch’ng S , Nieweg OE , Mallo HA , Adriaansz S , Kerkhoven RM , Cornelissen S , Broeks A , Klop WMC , Zuur CL , van Houdt WJ , Peeper DS , Spillane AJ , van Akkooi ACJ , Scolyer RA , Schumacher TNM , Menzies AM , Long GV , and Blank CU (2021). Survival and biomarker analyses from the OpACIN-neo and OpACIN neoadjuvant immunotherapy trials in stage III melanoma. Nat Med 27 , 256–263. 10.1038/s41591-020-01211-7.33558721 Sade-Feldman M , Yizhak K , Bjorgaard SL , Ray JP , de Boer CG , Jenkins RW , Lieb DJ , Chen JH , Frederick DT , Barzily-Rokni M , Freeman SS , Reuben A , Hoover PJ , Villani AC , Ivanova E , Portell A , Lizotte PH , Aref AR , Eliane JP , Hammond MR , Vitzthum H , Blackmon SM , Li B , Gopalakrishnan V , Reddy SM , Cooper ZA , Paweletz CP , Barbie DA , Stemmer-Rachamimov A , Flaherty KT , Wargo JA , Boland GM , Sullivan RJ , Getz G , and Hacohen N (2018). Defining T Cell States Associated with Response to Checkpoint Immunotherapy in Melanoma. Cell 175 , 998–1013 e1020. 10.1016/j.cell.2018.10.038.30388456 Sagiv Barfi I , Czerwinski DK , and Levy R (2021). In Situ Vaccination with IL-12Fc and TLR Agonist - a Crucial Role for B Cells in Generating Anti-Tumor T Cell Immunity. Blood 138 , 3514–3514. 10.1182/blood-2021-146628. Sahin U , Derhovanessian E , Miller M , Kloke BP , Simon P , Lower M , Bukur V , Tadmor AD , Luxemburger U , Schrors B , Omokoko T , Vormehr M , Albrecht C , Paruzynski A , Kuhn AN , Buck J , Heesch S , Schreeb KH , Muller F , Ortseifer I , Vogler I , Godehardt E , Attig S , Rae R , Breitkreuz A , Tolliver C , Suchan M , Martic G , Hohberger A , Sorn P , Diekmann J , Ciesla J , Waksmann O , Bruck AK , Witt M , Zillgen M , Rothermel A , Kasemann B , Langer D , Bolte S , Diken M , Kreiter S , Nemecek R , Gebhardt C , Grabbe S , Holler C , Utikal J , Huber C , Loquai C , and Tureci O (2017). Personalized RNA mutanome vaccines mobilize poly-specific therapeutic immunity against cancer. Nature 547 , 222–226. 10.1038/nature23003.28678784 Schumacher TN , and Thommen DS (2022). Tertiary lymphoid structures in cancer. Science 375 , eabf9419. 10.1126/science.abf9419. Sharma A , Subudhi SK , Blando J , Scutti J , Vence L , Wargo J , Allison JP , Ribas A , and Sharma P (2019). Anti-CTLA-4 Immunotherapy Does Not Deplete FOXP3(+) Regulatory T Cells (Tregs) in Human Cancers. Clin Cancer Res 25 , 1233–1238. 10.1158/1078-0432.CCR-18-0762.30054281 Stronen E , Toebes M , Kelderman S , van Buuren MM , Yang W , van Rooij N , Donia M , Boschen ML , Lund-Johansen F , Olweus J , and Schumacher TN (2016). Targeting of cancer neoantigens with donor-derived T cell receptor repertoires. Science 352 , 1337–1341. 10.1126/science.aaf2288.27198675 Sultan H , Kumai T , Nagato T , Wu J , Salazar AM , and Celis E (2019). The route of administration dictates the immunogenicity of peptide-based cancer vaccines in mice. Cancer Immunol Immunother 68 , 455–466. 10.1007/s00262-018-02294-5.30604041 Swartz AM , Congdon KL , Nair SK , Li QJ , Herndon JE 2nd , Suryadevara CM , Riccione KA , Archer GE , Norberg PK , Sanchez-Perez LA , and Sampson JH (2021). A conjoined universal helper epitope can unveil antitumor effects of a neoantigen vaccine targeting an MHC class I-restricted neoepitope. NPJ Vaccines 6 , 12. 10.1038/s41541-020-00273-5.33462231 Tan WCC , Nerurkar SN , Cai HY , Ng HHM , Wu D , Wee YTF , Lim JCT , Yeong J , and Lim TKH (2020). Overview of multiplex immunohistochemistry/immunofluorescence techniques in the era of cancer immunotherapy. Cancer Commun (Lond) 40 , 135–153. 10.1002/cac2.12023.32301585 Tanyi JL , Bobisse S , Ophir E , Tuyaerts S , Roberti A , Genolet R , Baumgartner P , Stevenson BJ , Iseli C , Dangaj D , Czerniecki B , Semilietof A , Racle J , Michel A , Xenarios I , Chiang C , Monos DS , Torigian DA , Nisenbaum HL , Michielin O , June CH , Levine BL , Powell DJ Jr. , Gfeller D , Mick R , Dafni U , Zoete V , Harari A , Coukos G , and Kandalaft LE (2018). Personalized cancer vaccine effectively mobilizes antitumor T cell immunity in ovarian cancer. Sci Transl Med 10 . 10.1126/scitranslmed.aao5931. Toes RE , Offringa R , Blom RJ , Melief CJ , and Kast WM (1996). Peptide vaccination can lead to enhanced tumor growth through specific T-cell tolerance induction. Proc Natl Acad Sci U S A 93 , 7855–7860. 10.1073/pnas.93.15.7855.8755566 Togashi Y , Shitara K , and Nishikawa H (2019). Regulatory T cells in cancer immunosuppression - implications for anticancer therapy. Nat Rev Clin Oncol 16 , 356–371. 10.1038/s41571-019-0175-7.30705439 Tran E , Turcotte S , Gros A , Robbins PF , Lu YC , Dudley ME , Wunderlich JR , Somerville RP , Hogan K , Hinrichs CS , Parkhurst MR , Yang JC , and Rosenberg SA (2014). Cancer immunotherapy based on mutation-specific CD4+ T cells in a patient with epithelial cancer. Science 344 , 641–645. 10.1126/science.1251102.24812403 Turajlic S , Sottoriva A , Graham T , and Swanton C (2019). Resolving genetic heterogeneity in cancer. Nat Rev Genet 20 , 404–416. 10.1038/s41576-019-0114-6.30918367 van den Hout M , Koster BD , Sluijter BJR , Molenkamp BG , van de Ven R , van den Eertwegh AJM , Scheper RJ , van Leeuwen PAM , van den Tol MP , and de Gruijl TD (2017). Melanoma Sequentially Suppresses Different DC Subsets in the Sentinel Lymph Node, Affecting Disease Spread and Recurrence. Cancer Immunol Res 5 , 969–977. 10.1158/2326-6066.CIR-17-0110.28935649 Veatch JR , Singhi N , Srivastava S , Szeto JL , Jesernig B , Stull SM , Fitzgibbon M , Sarvothama M , Yechan-Gunja S , James SE , and Riddell SR (2021). A therapeutic cancer vaccine delivers antigens and adjuvants to lymphoid tissues using genetically modified T cells. J Clin Invest 131 . 10.1172/JCI144195. Veglia F , Perego M , and Gabrilovich D (2018). Myeloid-derived suppressor cells coming of age. Nat Immunol 19 , 108–119. 10.1038/s41590-017-0022-x.29348500 Verma V , Shrimali RK , Ahmad S , Dai W , Wang H , Lu S , Nandre R , Gaur P , Lopez J , Sade-Feldman M , Yizhak K , Bjorgaard SL , Flaherty KT , Wargo JA , Boland GM , Sullivan RJ , Getz G , Hammond SA , Tan M , Qi J , Wong P , Merghoub T , Wolchok J , Hacohen N , Janik JE , Mkrtichyan M , Gupta S , and Khleif SN (2019). PD-1 blockade in subprimed CD8 cells induces dysfunctional PD-1(+)CD38(hi) cells and anti-PD-1 resistance. Nat Immunol 20 , 1231–1243. 10.1038/s41590-019-0441-y.31358999 Villani AC , Satija R , Reynolds G , Sarkizova S , Shekhar K , Fletcher J , Griesbeck M , Butler A , Zheng S , Lazo S , Jardine L , Dixon D , Stephenson E , Nilsson E , Grundberg I , McDonald D , Filby A , Li W , De Jager PL , Rozenblatt-Rosen O , Lane AA , Haniffa M , Regev A , and Hacohen N (2017). Single-cell RNA-seq reveals new types of human blood dendritic cells, monocytes, and progenitors. Science 356 . 10.1126/science.aah4573. Vonderheide RH (2020). CD40 Agonist Antibodies in Cancer Immunotherapy. Annu Rev Med 71 , 47–58. 10.1146/annurev-med-062518-045435.31412220 Wang TY , Liu Q , Ren Y , Alam SK , Wang L , Zhu Z , Hoeppner LH , Dehm SM , Cao Q , and Yang R (2021). A pan-cancer transcriptome analysis of exitron splicing identifies novel cancer driver genes and neoepitopes. Mol Cell 81 , 2246–2260 e2212. 10.1016/j.molcel.2021.03.028.33861991 Wculek SK , Cueto FJ , Mujal AM , Melero I , Krummel MF , and Sancho D (2020). Dendritic cells in cancer immunology and immunotherapy. Nat Rev Immunol 20 , 7–24. 10.1038/s41577-019-0210-z.31467405 Wei J , Montalvo-Ortiz W , Yu L , Krasco A , Ebstein S , Cortez C , Lowy I , Murphy AJ , Sleeman MA , and Skokos D (2021). Sequence of alphaPD-1 relative to local tumor irradiation determines the induction of abscopal antitumor immune responses. Sci Immunol 6 . 10.1126/sciimmunol.abg0117. Wellenstein MD , and de Visser KE (2018). Cancer-Cell-Intrinsic Mechanisms Shaping the Tumor Immune Landscape. Immunity 48 , 399–416. 10.1016/j.immuni.2018.03.004.29562192 Wicherska-Pawlowska K , Wrobel T , and Rybka J (2021). Toll-Like Receptors (TLRs), NOD-Like Receptors (NLRs), and RIG-I-Like Receptors (RLRs) in Innate Immunity. TLRs, NLRs, and RLRs Ligands as Immunotherapeutic Agents for Hematopoietic Diseases. Int J Mol Sci 22 . 10.3390/ijms222413397. Williford JM , Ishihara J , Ishihara A , Mansurov A , Hosseinchi P , Marchell TM , Potin L , Swartz MA , and Hubbell JA (2019). Recruitment of CD103(+) dendritic cells via tumor-targeted chemokine delivery enhances efficacy of checkpoint inhibitor immunotherapy. Sci Adv 5 , eaay1357. 10.1126/sciadv.aay1357. Wu YM , Cieslik M , Lonigro RJ , Vats P , Reimers MA , Cao X , Ning Y , Wang L , Kunju LP , de Sarkar N , Heath EI , Chou J , Feng FY , Nelson PS , de Bono JS , Zou W , Montgomery B , Alva A , Team PSCIPCD , Robinson DR , and Chinnaiyan AM (2018). Inactivation of CDK12 Delineates a Distinct Immunogenic Class of Advanced Prostate Cancer. Cell 173 , 1770–1782 e1714. 10.1016/j.cell.2018.04.034.29906450 Xin L , Gao J , Zheng Z , Chen Y , Lv S , Zhao Z , Yu C , Yang X , and Zhang R (2021). Fibroblast Activation Protein-alpha as a Target in the Bench-to-Bedside Diagnosis and Treatment of Tumors: A Narrative Review. Front Oncol 11 , 648187. 10.3389/fonc.2021.648187.34490078 Yates KB , Tonnerre P , Martin GE , Gerdemann U , Al Abosy R , Comstock DE , Weiss SA , Wolski D , Tully DC , Chung RT , Allen TM , Kim AY , Fidler S , Fox J , Frater J , Lauer GM , Haining WN , and Sen DR (2021). Epigenetic scars of CD8(+) T cell exhaustion persist after cure of chronic infection in humans. Nat Immunol 22 , 1020–1029. 10.1038/s41590-021-00979-1.34312547 Zagorulya M , Duong E , and Spranger S (2020). Impact of anatomic site on antigen-presenting cells in cancer. J Immunother Cancer 8 . 10.1136/jitc-2020-001204. Zeng B , Middelberg AP , Gemiarto A , MacDonald K , Baxter AG , Talekar M , Moi D , Tullett KM , Caminschi I , Lahoud MH , Mazzieri R , Dolcetti R , and Thomas R (2018). Self-adjuvanting nanoemulsion targeting dendritic cell receptor Clec9A enables antigen-specific immunotherapy. J Clin Invest 128 , 1971–1984. 10.1172/JCI96791.29485973 Zhivaki D , Borriello F , Chow OA , Doran B , Fleming I , Theisen DJ , Pallis P , Shalek AK , Sokol CL , Zanoni I , and Kagan JC (2020). Inflammasomes within Hyperactive Murine Dendritic Cells Stimulate Long-Lived T Cell-Mediated Anti-tumor Immunity. Cell Rep 33 , 108381. 10.1016/j.celrep.2020.108381.33207188 Zitvogel L , Kepp O , and Kroemer G (2011). Immune parameters affecting the efficacy of chemotherapeutic regimens. Nat Rev Clin Oncol 8 , 151–160. 10.1038/nrclinonc.2010.223.21364688 Zou W , Yaung SJ , Fuhlbruck F , Ballinger M , Peters E , Palma JF , Shames DS , Gandara D , Jiang Y , and Patil NS (2021). ctDNA Predicts Overall Survival in Patients With NSCLC Treated With PD-L1 Blockade or With Chemotherapy. JCO Precis Oncol 5 , 827–838. 10.1200/PO.21.00057.34994614

PMC009xxxxxx/PMC9579187.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8110298 7170 Semin Nephrol Semin Nephrol Seminars in nephrology 0270-9295 1558-4488 36435682 9579187 10.1016/j.semnephrol.2022.10.005 NIHMS1849315 Article COVID-19 and the Kidney: Recent Advances and Controversies Menez Steven MD, MHS 1 Parikh Chirag R. MD, PhD 1 1. Division of Nephrology, Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland Corresponding author: Chirag R. Parikh, MD, PhD, Director, Division of Nephrology, Johns Hopkins University School of Medicine, 1830 E Monument St, Ste 416, Baltimore, Maryland 21287, chirag.parikh@jhmi.edu 10 11 2022 19 10 2022 08 1 2023 42 3 151279151279 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Kidney involvement is common in COVID-19, and our understanding of the effects of COVID-19 on short- and long-term kidney outcomes has evolved over the course of the pandemic. Initial key questions centered on the spectrum and degree of acute kidney injury (AKI) in patients hospitalized with severe COVID-19. Investigators worldwide explored the association between COVID-19–associated AKI and short-term outcomes, including inpatient mortality and disease severity. Even as treatments evolved, vaccinations were developed, and newer viral variants arose, subsets of patients were identified as at continued high risk for major adverse kidney outcomes. In this review, we explore key topics of continued relevance including: 1) a comparison of COVID-19–associated AKI with AKI developing in other clinical settings; 2) the ongoing controversy over kidney tropism in the setting of COVID-19 and the potential for competitive binding of the SARS-CoV-2 virus with ACE2 to prevent viral cell entry; and 3) the identification of high-risk patients for adverse outcomes in order to inform long-term outpatient management. Patients at particularly high risk for adverse kidney outcomes include those with APOL1 high-risk genotype status, and biomarkers of injury, inflammation, and tubular health and repair measured in both the blood and urine may hold prognostic significance. acute kidney injury APOL1 chronic kidney disease COVID-19 pmcIntroduction The SARS-CoV-2 virus has continued to affect people across the world for over 2 years,1,2 with the novel coronavirus disease of 2019 (COVID-19) pandemic persisting as new variants have arisen.3 Acute kidney injury (AKI) has been well described in patients with COVID-19, especially in patients hospitalized with severe COVID-19 early in the pandemic.4–6 We and others have explored risk factors for both short- and intermediate-term adverse kidney outcomes in patients hospitalized with COVID-19,7–9 and more recent work has explored longer-term effects of COVID-19 on kidney function.10,11 However, with SARS-CoV-2 variants arising over time and continued concerns about future strains evading vaccine-mediated immunity, it becomes increasingly important to identify the patients at greatest risk for severe COVID-19 and adverse post–COVID-19 outcomes. From a kidney standpoint, such major adverse kidney events (MAKE) include the risk of COVID-19–associated AKI and long-term accelerated decline in kidney function. Central to the question of MAKE in the setting of COVID-19 is the kidney tropism associated with SARS-CoV-2 infection, given the high expression of angiotensin converting enzyme-2 (ACE2) receptors in the kidney, a continued area of controversy. In this review, we aim to synthesize recent advances in the study of COVID-19–associated kidney injury while highlighting a several areas of ongoing and future research: the comparison of COVID-19–associated AKI with AKI in other clinical settings; kidney tropism in the setting of COVID-19; and active research to identify key subgroups of patients at the highest risk for major adverse kidney events, including AKI and long-term CKD. COVID-19–associated AKI in comparison to AKI in other clinical settings In an editorial published early in the course of the global pandemic, Kellum et al.12 identified several similarities and distinctions between AKI associated with COVID-19 compared with other causes of sepsis. They also raised the question of whether COVID-19–associated AKI should be conceptualized or managed differently from general sepsis-associated AKI, which set the framework for future studies. Early studies evaluating histopathological evidence of kidney tissue at autopsy suggested acute tubular injury as the predominant cause of AKI in the vast majority of patients dying from COVID-19, with the clear caveat being that these studies were conducted using postmortem samples.13,14 Alexander and colleagues15 compared COVID-19–associated AKI with sepsis-associated AKI using a multi-omics approach. These investigators interrogated kidney tissue from 17 patients who died from COVID-19 compared with samples from 14 patients without COVID-19, which served as controls: 7 with sepsis-associated AKI (s-AKI) and 7 with non–sepsis-associated AKI (ns-AKI). All 17 patients who died from COVID-19 had evidence of at least mild acute tubular injury on histopathological examination. Using spatial transcriptomics and proteomic analysis, these authors found similar patterns of decreased oxidative phosphorylation, with upregulation in the ceramide signaling pathway and microvascular dysfunction/inflammation similar to that seen in s-AKI, findings not observed in ns-AKI. Outside of infectious etiologies of AKI, we sought to compare the degree of kidney injury and inflammation in patients with COVID-19 with AKI in other clinical settings (Figure 1). Comparison groups included patients after cardiac surgery (from the TRIBE-AKI [Translational Research investigating Biomarker Endpoints in AKI Study] cohort), patients after brain death before kidney donation (Deceased Donor Study cohort), and patients in the setting of exercise stress (marathon-associated AKI). Using pre-cardiac surgery biomarker levels as a reference, we noted that among patients with COVID-19–associated AKI, 40%, 50%, and 60% of patients with stage 1, 2, and 3 AKI, respectively, had KIM-1 levels above the 90th percentile of reference. Similarly, among kidney donors after brain death, 70%, 80%, and 90% of patients with stage 1, 2, and 3 AKI, respectively, had MCP-1 levels above the 90th percentile of reference. In general, the degree of kidney inflammation in the setting of COVID-19–associated AKI was comparable to other clinical settings. Despite a number of similarities, however, several studies have highlighted how COVID-19–associated AKI is distinct in several respects from AKI in other settings, including its transmissibility, rapidity of infection, and disease severity. The incidence of AKI incidence has been observed to be increased in patients with COVID-19 compared with AKI from bacterial sepsis, severe influenza infection, or in the general hospitalized population.12,16,17 Moledina et al.16 demonstrated that COVID-19 was associated with AKI even after adjusting for key demographic factors, inflammatory markers, use of vasopressors, and other potential confounders indicating illness severity, compared with patients admitted within the same timeframe who were COVID-19–negative (adjusted hazard ratio [aHR]: 1.40; 95% CI: 1.29–1.53). These results further suggested the existence of mechanisms leading to AKI in the setting of COVID-19 that extend beyond general hospitalized AKI. Strohbehn and colleagues17 used a historical cohort of patients with severe influenza infection as a more suitable comparison group to investigate adverse kidney outcomes in patients with severe COVID-19. These investigators demonstrated that compared with patients hospitalized with influenza, patients hospitalized with COVID-19 had a greater risk of AKI incidence (HR: 1.58; 95% CI: 1.29–1.94), higher overall mortality (aHR: 7.17; 95% CI: 4.78–10.76), and less AKI recovery at the time of hospital discharge. One notable limitation for many of these translational and clinical studies surrounding COVID-19 associated AKI is the temporal trend in AKI incidence as new strains have emerged. Following the initial wave of the pandemic, later waves have been associated with less disease severity including mortality in general,18 with decreasing AKI incidence.19 Similar to Alexander et al,15 Volbeda and colleagues20 sought to compare COVID-19– and sepsis-associated AKI beyond clinical outcomes and histopathology, measuring differential mRNA gene expression in a similarly sized cohort of patients (6 with COVID-19–associated AKI, 27 with bacterial sepsis–associated AKI, and 12 reference samples from total nephrectomy). Volbeda et al.18 found differential gene expression by clinical setting, with mRNA expression of the genes for NGAL and KIM-1 significantly lower in patients with COVID-19, compared with patients with bacterial sepsis, which is somewhat counterintuitive given the higher degree of tubular injury seen on histopathology in patients with COVID-19. Furthermore, these investigators found less kidney inflammation and endothelial cell activation in patients with COVID-19 compared with patients with bacterial sepsis, based on the relative lack of upregulation of E-selectin, VCAM-1, and ICAM-1. Kidney tropism in COVID-19: evidence for and against direct kidney involvement Based on previous studies from the early 2000s with the SARS-CoV-1 virus,21 it was quickly established that SARS-CoV-2 viral entry into cells occurs through the ACE2 receptor, which is highly expressed in the proximal tubule and, to a lesser extent, in the distal tubule and collecting duct.22 Indeed, early post-mortem findings in patients with severe COVID-19 demonstrated acute tubular injury being nearly universally present on histology.14,23 Later studies of kidney biopsy findings in critically ill patients with COVID-19 similarly demonstrated acute tubular injury as a prominent histological finding (Table 1).24–27 However, it remained unclear to what extent such viral entry might have a direct impact on kidney injury and whether inhibition of viral entry at the level of the ACE2 receptor would make a clinically meaningful difference.28 Hassler and colleagues28 argued that the detection of virus or viral particles within kidney tissue has been inconsistent, with a number of studies showing no presence of viral protein present on immunohistochemistry in biopsy studies, with the argument that viral like proteins seen in post-mortem studies were likely artifact.26,27,29 These investigators summarized the differences noted across several biopsy-based or postmortem tissue–based studies on the presence of either the spike protein or RNA using a variety of techniques, including immunohistochemistry, immunofluorescence, reverse transcription–PCR, or in situ hybridization. In March of 2020, Batlle and colleagues30 postulated that a soluble form of ACE2 could competitively bind to SARS-CoV-2, thereby preventing binding of the virus to the ACE2 receptor to limit viral entry and replication. However, studies would be required to provide evidence for these hypotheses in vitro and later in organoids and in vivo animal studies. Studies of kidney tropism in COVID-19—in vitro and in kidney organoids Monteil et al.31 tested this experimentally by adding human recombinant soluble ACE2 (ACE2 1–740) to Vero-E6 cells inoculated with SARS-CoV-2 in vitro. With varying degrees of ACE2 1–740 administered over time, these investigators demonstrated significant reductions in SARS-CoV-2 viral load. These investigators then used human embryonic stem cells to generate kidney organoids with proximal tubular–like epithelial cells present. As expected, addition of ACE2 1–740 reduced SARS-CoV-2 entry into these human kidney organoids in a dose-dependent fashion. Further research explored the use of a novel, bioengineered, soluble human ACE2 with an extended duration of action as a means of reducing SARS-CoV-2 infectivity of kidney organoids.32 Wysocki and colleagues specifically fused a human short ACE2 variant with an albumin-binding domain (ABD) to increase the duration of action.32 They found similar reductions in SARS-CoV-2 viral loads using the ACE2 1–740 tested by Monteil et al.,31 with similar efficacy using a novel, shorter human soluble ACE2-ABD with the added benefit of a longer duration of action. Studies of kidney tropism in COVID-19 – murine models In a follow-up study, Batlle et al.33 performed in vivo studies of their soluble ACE2-ABD protein linked via a dimerization motif hinge-like 4-cysteine dodecapeptide (DDC), which, in addition to a longer duration of activity, showed a greater binding affinity for SARS-CoV-2. Using a lethal murine model of COVID-19 in k18-hACE2 mice expressing human ACE2 and therefore susceptible to SARS-CoV-2 infection,34,35 the bioengineered ACE2–1–618-DDC-ABD was administered both intranasally and intraperitoneally. Compared with human ACE2 (ACE2 1–740) and their original bioengineered ACE2 (ACE2 1–618-ABD), ACE2–1–618-DDC-ABD had the greatest binding affinity to SARS-CoV-2. Mice receiving ACE2–1–618-DDC-ABD experienced less weight loss, with marked improvements in clinical scores and reduced mortality compared with untreated animals, with only 1 in 10 treated mice requiring euthanization. Histological analysis showed that the treated animals had less severe tubular injury, based on NGAL tissue staining, compared with untreated mice. Clinically, the presence of SARS-CoV-2 in the urine as a marker of disease severity or adverse outcomes remains controversial, with conflicting evidence. Frithiof et al.36 evaluated for the presence of SARS-CoV-2 RNA in 81 critically-ill patients with PCR-proven COVID-19. They were able to detect viral RNA in only 6 (7%) patients, and they did not find any association between either the presence of the virus or viral load (range: 300–2,800 copies/mL) with disease severity or mortality. A later study by Caceres et al.37 found that the presence of SARS-CoV-2 virus in the urine was associated with AKI and worse kidney outcomes in 52 patients hospitalized with COVID-19. Specifically, the presence of SARS-CoV-2 virus in the urine was associated with AKI incidence and that viral load correlated with subsequent mortality. Identifying patients at highest risk for MAKE and evaluation of long-term kidney function after COVID-19 APOL1 high-risk genotype status and MAKE after COVID-19 Early case reports and case series,38,39 later supported by larger studies conducted as the pandemic progressed,29 demonstrated an association between APOL1 high-risk genotype status and collapsing glomerulopathy in patients with COVID-19. COVID-19–associated nephropathy or COVAN was the term used to describe this phenomenon, similar to HIV-associated nephropathy or HIVAN and with a potentially shared pathophysiology.40 Later studies explored clinical outcomes in patients with APOL1 high-risk genotype status diagnosed with COVID-19 (Table 2). Larsen et al.39 showed, in a combined inpatient/outpatient cohort of 126 self-reported Black adult patients, that the presence of 2 APOL1 risk alleles (either homozygous G1/G1 or G2/G2 or heterozygous G1/G2) was associated with an increased risk of AKI incidence, AKI persistence, and need for kidney replacement therapy. More recently, Hung and colleagues41 investigated how APOL1 risk variants associated with the incidence of AKI and death in patients hospitalized with COVID-19. Of 990 patients in the Veterans Affairs (VA) health care system of African ancestry who were hospitalized with COVID-19 between March 2020 and January 2021, 125 (12.6%) patients had 2 APOL1 risk alleles. Among these patients, over 50% developed AKI and nearly 20% died. Compared with patients without high-risk genotype status, those in the high-risk group had significantly higher odds of AKI incidence (odds ratio [OR]: 1.95; 95% CI: 1.27–3.02) in fully adjusted analysis. Furthermore, these patients had significantly higher odds of developing KDIGO stages 2 or 3 AKI (OR: 2.03; 95% CI: 1.37–2.99) and of mortality (OR: 2.15; 95% CI: 1.22–3.72). Notably, these investigators did not find any significant association between APOL1 high-risk genotype status and hematuria or proteinuria. Blood biomarker–enriched risk prognostication in COVID-19 Clinically-available biomarkers of disease severity, especially markers of inflammation such as D-dimer and C-reactive protein, have been associated with COVID-19 disease severity and short-term outcomes, including in-hospital mortality.42,43 The use of novel biomarkers in both blood and urine has been studied extensively across multiple clinical settings to improve the precision and timeliness of AKI diagnosis, as well as to prognosticate longer-term outcomes after acute kidney injury, both clinical and subclinical,44–50 which is being increasingly investigated in the setting of COVID-19 (Table 3). Soluble tumor necrosis factor receptor 1 (sTNFR1) has been associated with adverse outcomes in patients with COVID-19, including COVID-19 severity51 and intensive care unit (ICU) mortality.52 With a focus on kidney-related outcomes, Ferrando and colleagues9 investigated the prospective association between AKI in patients with COVID-19 and the biomarkers sTNFR1 and sTNFR1. A total of 122 patients with COVID-19 were consented at the time of ICU admission and followed longitudinally over the course of their admission. Levels of sTNFR1 and sTNFR2 were higher in patients with severe COVID-19 compared with healthy blood donors as controls. Furthermore, sTNFR1 and sTNFR2 levels trended higher by AKI stage (P < 0.001 and P = 0.02, respectively). Finally, sTNFR1 showed moderate to strong discrimination for the prediction of 30-day mortality, after adjustment for age and respiratory failure, as a marker of COVID-19 severity (AUC: 0.73; 95% CI: 0.62–0.84). We have similarly shown that higher levels of plasma sTNFR1 and sTNFR2 are both strongly associated with increased risk of MAKE, defined in our study as development of AKI stage 3, dialysis, or death within 60 days of admission in patients hospitalized with COVID-19.53 Furthermore, sTNFR1 showed strong discrimination for the prediction of MAKE (AUC: 0.88). A cutpoint value for sTNFR1 of 3,005 pg/mL had a negative predictive value of 92%, suggesting that sTNFR1 may be used as a potential rule-out test for MAKE at the time of COVID-19 hospitalization. Urine biomarker–enriched risk prognostication in COVID-19 Bezerra and colleagues54 investigated the association of urinary biomarkers of kidney injury with death in patients admitted to the ICU with COVID-19. They measured urinary levels of neutrophil gelatinase-associated lipocalin (NGAL), kidney injury molecule-1 (KIM-1), monocyte chemoattractant protein-1 (MCP-1), and nephrin in patients admitted to the ICU with COVID-19. In fully adjusted Cox proportional hazards regression modeling, higher levels of NGAL (above a cutpoint of 118.8 ng/mg Cr) were significantly associated with increased risk of death at 2 months. Beyond mortality, our group has investigated the association between urinary biomarkers of injury, inflammation, and repair in patients hospitalized with COVID-19 with MAKE (stage 3 AKI, new dialysis, or death within 60 days of hospital admission).7 We measured biomarkers using urine samples obtained throughout the course of admission, demonstrating that higher levels of urinary NGAL, KIM-1, and MCP-1 were associated with MAKE (Figure 2). Conversely, higher levels of urinary epidermal growth factor (EGF), a marker of intact distal tubular repair and a surrogate for healthy repair mechanisms in the kidney, were associated with decreased risk of MAKE. Similarly, Xu et al.55 demonstrated, in a cohort of 444 patients in New York City, that urinary NGAL measured on admission for COVID-19 was strongly associated with AKI diagnosis and severity. Furthermore, a urinary NGAL level >150 ng/mL had 80% specificity and 75% sensitivity to diagnosed AKIN stages 2 or 3 AKI. Furthermore, NGAL levels on admission were associated with increased odds of sustained AKI, dialysis, death, and hospital length of stay. Long-term kidney function after COVID-19 Within months of the start of the pandemic, various groups investigated intermediate to long-term health consequences after recovery from acute COVID-19, variously termed long-COVID syndrome, post-COVID-19 syndrome, and post-acute sequelae of COVID-19 among others (Table 4).56 Huang and colleagues8 were among the first to report long-term outcomes up to 6 months after acute COVID-19 in patients surviving to hospital discharge. These investigators showed that patients commonly experienced long-term sequelae after recovery from the acute phase of COVID-19, including persistent dyspnea, fatigue, and weakness. In the subset of patients with available follow-up laboratory data, 35% of individuals who had experienced COVID-19–associated AKI had decreased eGFR at follow-up, compared with only 13% of individuals without COVID-19–associated AKI. Nugent et al.57 later investigated longer-term kidney function in patients with hospitalized AKI in the setting of COVID-19 compared with patients who experienced hospitalized AKI without COVID-19. After adjusting for demographic factors, baseline comorbidities, peak creatinine value in the hospital, and need for acute dialysis, patients with COVID-19–associated AKI had a greater decreased in eGFR over the course of follow-up compared with noninfected controls with hospitalized AKI (−14.0 mL/min/1.72 m2; 95% CI: −25.1 to −2.9). Using data from the VA health care system, Bowe and colleagues58 evaluated MAKE up to 365 days after COVID-19 diagnosis in both the ambulatory and hospital settings. Compared with over 1.6 million patients without a diagnosis of COVID-19, nearly 90,000 patients who survived at least 30 days after COVID-19 diagnosis had a higher incidence of AKI and greater declines in eGFR. Furthermore, patients with COVID-19 were more likely to develop end-stage kidney disease (HR 2.96; 95% CI: 2.49–3.51). Not unexpectedly, among patients diagnosed with COVID-19, excess eGFR decline was greatest in hospitalized patients who developed AKI, compared to hospitalized patients without AKI and the non-hospitalized population. Gu and colleagues11 later investigated kidney function trends in a cohort of 1,734 hospitalized patients with COVID-19 out of China. Patients who experienced AKI had significantly greater declines in eGFR compared with patients without AKI. In particular, patients with KDIGO stage 3 AKI had a 17.8% (95% CI: 9.1–26.4) greater decline in eGFR compared with patients without AKI. Summary The long-term impact of COVID-19 after initial recovery from the disease has become more evident over time, with an increasing focus on the identification of patients at the highest risk for long-term adverse outcomes. Our conceptualization of COVID-19–associated kidney injury as a distinct entity has been informed by studies worldwide, though similarities exist between this and other infectious and inflammatory clinical scenarios. With proof-of-concept studies in vitro leading to newer studies in kidney organoids and mouse models, there is compelling evidence that a soluble form of ACE2 as a competitive binder to SARS-CoV-2 may decrease infectivity and improve clinical outcomes. However, further studies in human subjects are essential before firm clinical implications can be determined. Research in therapeutic discovery for patients with COVID-19 and to identify the patients most at risk for MAKE remains essential. Such research will be increasingly important as new SARS-CoV-2 variants continue to emerge, the lessons from which may be applicable beyond COVID-19 in the future. Financial Support: SM receives grant support from the NIDDK (K23DK128538) and additional research funding from RenalytixAI. Conflicts of Interest: CRP serves as a member of the advisory board of and own equity in Renalytix AI. CRP is named as a co-inventor on a pending patent, “Methods and Systems for Diagnosis of Acute Interstitial Nephritis,”. SM has received consulting fees from the Dedham Group and research funding from Renalytix AI. Figure 1. Heat map showing the percentage of patients with AKI in the setting of COVID-19, after cardiac surgery (TRIBE-AKI cohort), after brain death (DDS cohort), and in the setting of exercise stress (marathon-associated AKI). For exercise stress–associated AKI, blood and urine biomarkers were measured from biosamples obtained within 30 minutes of completing a marathon. The colors denote the percentage of patients by AKI stage with biomarker levels above the 90th percentile of the reference value, based on cardiac surgery preoperative values. Among patients with COVID-19–associated AKI, 40%, 50%, and 60% of patients with stage 1, 2, and 3 AKI, respectively, had KIM-1 levels above the 90th percentile of reference. Similarly, among kidney donors after brain death, 70%, 80%, and 90% of patients with stage 1, 2, and 3 AKI, respectively, had MCP-1 levels above the 90th percentile of the reference value. Note: Sample sizes by AKI stage: COVID-19: stage 1 (n = 56), stage 2 (n = 31), stage 3 (n = 20); cardiac surgery: stage 1 (n = 456), stage 2 (n = 34), stage 3 (n = 31); brain death: stage 1 (n = 275), stage 2 (n = 93), stage 3 (n = 76); exercise stress: stage 1 (n = 9), stage 2 (n = 2). Abbreviations: AKI, acute kidney injury; COVID-19, coronavirus disease 2019; DDS, Deceased Donor Study; EGF, epidermal growth factor; IL, interleukin; KIM-1, kidney injury molecule 1; MCP1, monocyte chemoattractant protein 1; NGAL, neutrophil gelatinase-associated lipocalin; TRIBE-AKI, Translational Research Investigating Biomarker Endpoints in Acute Kidney Injury; UMOD, uromodulin; YKL-40, chitinase-3-like protein 1. Figure reproduced with permission from American Journal of Kidney Diseases, Volume 79, Issue 2, 2022, Pages 257–267.e1, Copywright Elsevier Figure 2. Risk of stage 3 AKI, new dialysis initiation, or death within 60 days of hospital admission by urinary biomarker level, indexed to urine creatinine and adjusted for World Health Organization disease severity scale. Abbreviations: AKI, acute kidney injury; EGF, epidermal growth factor; IL, interleukin; KIM-1, kidney injury molecule 1; MCP-1, monocyte chemoattractant protein 1; NGAL, neutrophil gelatinase-associated lipocalin; OPN, osteopontin; UMOD, uromodulin; WHO, World Health Organization; YKL-40, chitinase-3-like protein 1 Figure reproduced with permission from American Journal of Kidney Diseases, Volume 79, Issue 2, 2022, Pages 257–267.e1, Copywright Elsevier Table 1. Overview of histological findings on kidney biopsies from patients with COVID-19 Study Acute tubular injury Glomerular findings Other findings Kudose et al.24 (n=14)* present in 11/14 Collapsing FSGS in 5/14 MCD in 1/14 Membranous GN in 2/14 Anti-GBM in 1/14 LN class IV/V in 1/14 Pigmented casts in 1/14 Nasr et al.25 (n=13) present in 13/13 Collapsing FSGS in 8/13 Diabetic Neph. in 4/13 Membranous GN in 2/13 IgAN in 1/13 Crescentic GN in 1/13 TRI’s in 4/13 Sharma et al.26 (n=10) present in 10/10 Crescentic GN in 1/10 “Healed” collapse in 1/10 TMA in 2/10 Myoglobin casts in 1/10 Akilesh et al.27 (n=14)* present in 12/14 Collapsing FSGS in 7/14 FSGS in 3/14 Diabetic Neph. in 1/14 MCD in 1/14 TMA in 6/14 AIN in 2/14 * among native kidney biopsies only FSGS = focal segmental glomerulosclerosis; GN = glomerulonephritis; IgAN = IgA nephropathy; LN = lupus nephritis; MCD = minimal change disease; TRI = endothelial tubuloreticular inclusion Table 2. Summary of cohort studies investigating APOL1 high-risk genotype status and kidney outcomes in COVID-19 Study Study Details Patient Population Comparison Groups Outcomes May et al.29 Retrospective cohort study between March 2020 and March 2021 240 patients with PCR-positive COVID-19 who provided kidney biopsies, of whom 107 underwent APOL1 genetic testing1 2 APOL1 high-risk alleles (n = 65) vs 0 or 1 APOL1 high-risk alleles (n = 42) Kidney pathology by high-risk allele status (2 vs. 0/1):Increased FSGS lesions (P = 0.03) Increased podocyte foot process effacement (P < 0.001) 44/48 (91.7%) patients with collapsing GN had 2 high-risk alleles Larsen et al.39 Retrospective cohort study between March 2020 and October 2020, New Orleans, LA 126 adult self-identified Black patients with PCR-positive COVID-19 2 APOL1 high-risk alleles (n = 16) vs 0 or 1 APOL1 high-risk alleles (n = 110) aOR of AKI = 4.4 (95% CI: 1.1–17.5) aOR of persistent AKI = 3.5 (95% CI: 1.1–11.6) aOR of kidney replacement therapy = 5.0 (95% CI: 1.0–24.4) Hung et al.41 Retrospective cohort study between March 2020 and January 2021 990 adult participants of African ancestry in the VA Health System with PCR-positive COVID-19 2 APOL1 high-risk alleles (n = 125) vs 0 or 1 APOL1 high-risk alleles (n = 865) Overall aOR of AKI = 1.95 (95% CI: 1.27–3.02) aOR of death = 2.15 (95% CI: 1.22–3.72) Patients with baseline eGFR > 60 mL/min/1.73 m2 aOR of AKI = 1.93 (95% CI: 1.15–3.26) aOR of death = 2.51 (95% CI: 1.21–5.05) AKI, acute kidney injury; aOR = adjusted odds ratio; GN = glomerulopathy; PCR = polymerase chain reaction Table 3. Summary of studies showing biomarker-enriched prognostication of adverse kidney events Study Study Details Patient Population Control/Comparison Group Biomarkers Tested Kidney outcomes Blood Ferrando et al.9 Prospective cohort study between March 2020 and September 2020 from the Uppsala PRONMED-study cohort 122 patients admitted to the hospital with COVID-19, with blood samples drawn within 6 days of ICU admission 25 healthy blood donors plasma sTFNR1, sTNFR2 Significantly increased sTNFR1 and sTNFR2 in COVID-19 vs. control (P < 0.001) sTNFR1 levels increase with increasing AKI stage (P < 0.001) sTNFR2 levels increase with increasing AKI stage (P < 0.02) Menez et al.53 Prospective cohort study between April 2020 and June 2020 in Baltimore, MD, New Haven, CT, and New York, NY 576 patients admitted to the hospital with COVID-19, taking first available plasma sample for biomarker measurement N/A Plasma sTNFR1, sTNFR2, NGAL, YKL-40, KIM-1, IL-2, IL-10, IL-18, sFTL1, TNF-α, Ang2 Higher aHR of MAKE by increase in biomarker: sTNFR1 = 2.30 (95% CI: 1.86–2.85) sTNFR2 = 2.26 (95% CI: 1.73–2.95) Model discrimination for MAKE: AUC of sTNFR1 = 0.88 (95% CI: 0.85–0.91) AUC of sTNFR2 = 0.83 (95% CI: 0.80–0.87) Urine Menez et al.7 Prospective cohort study between April 2020 and June 2020 in Baltimore, MD and New Haven, CT 178 patients admitted to the hospital with COVID-19, with urinary biomarkers measured on all available urine samples (n = 218 samples) N/A Urinary EGF, UMOD, interluekin-18, YKL-40, albumin, NGAL, OPN, MCP-1, KIM-1 Lower aHR of MAKE by increase in biomarker: EGF = 0.61 (95% CI: 0.47–0.79) Higher aHR of MAKE by increase in biomarker: YKL-40 = 1.18 (95% CI: 1.04–1.34) NGAL = 1.34 (95% CI: 1.14–1.57) MCP-1 = 1.42 (95% CI: 1.09–1.84) KIM-1 = 2.03 (95% CI: 1.38–2.99) Xu et al.55 Prospective cohort study between March 2020 and April 2020 in New York, NY 440 patients presenting to the Columbia University emergency department with COVID-19 Historical cohort of 426 patients admitted to Columbia University between 2017–2019 Urinary NGAL NGAL at admission associated with AKI diagnosis (p < 0.001) NGAL >150 ng/mL to diagnose stage 2/3 AKI:Sensitivity 75% Specificity 80% NGAL <150 ng/mL to rule out stage 2/3 AKI: negative predictive value = 0.95 (95% CI: 0.92–0.97) * Combined model containing NGAL, KIM-1, and proteinuria Abbreviations: aHR = adjusted hazard ratio; Ang2 = angiopoietin 2, EGF, epidermal growth factor; KIM-1, kidney injury molecule 1; MAKE = major adverse kidney events (stage 3 AKI, dialysis, death within 60 days of hospital admission); MCP-1, monocyte chemoattractant protein 1; NGAL, neutrophil gelatinase-associated lipocalin; OPN, osteopontin; sFLT1 = soluble fms-like tyrosine kinase 1; sTNFR1/2 = soluble tumor necrosis factor receptor 1/2; UMOD, uromodulin; YKL-40, chitinase-3-like protein 1 Table 4. Long-term outcomes in COVID-19 Authors Study Details Patient Population Control/Comparison Group Outcomes Huang et al.8 Ambidirectional cohort study of patients hospitalized with COVID-19, surviving to discharge from January 2020 to May 2020 in Wuhan, China 1,733 patients admitted to the hospital with COVID-19 who survived to hospital discharge. N/A Persistent fatigue or weakness in 63% of survivors Anxiety/depression in 23% of survivors Decreased eGFR in 35% of survivors at follow-up Decreased eGFR* in 13% of survivors without AKI at time of acute COVID-19 Nugent et al.57 Retrospective cohort study of patients hospitalized with COVID-19 surviving to discharge from March 2020 to August 2020 in 5 hospitals across CT and RI 178 patients who had AKI in the setting of COVID-19 during hospitalization 1,430 patients who had hospitalized AKI but did not have COVID-19 Patients with AKI and COVID-19, compared with patients with AKI and no COVID-19: eGFR decline 14.0 mL/min/1.73m2 greater aHR of kidney recovery during follow-up, in those without AKI recovery at discharge = 0.57 (95% CI: 0.35–0.92) Bowe et al.58 Retrospective cohort study of patients in the VA health care system with COVID-19, in both the ambulatory and hospital setting from March 2020 to March 2021 89,216 patients in the VA system with COVID-19 who survived at least 30 days from date of diagnosis 1.6 million ambulatory patients in the VA system over the same time period without COVID-19 COVID-19 survivors compared to noninfected controls: aHR of AKI = 1.94 (95% CI: 1.86–2.04) aHR of eGFR decline ≥ 30% = 1.25 (95% CI: 1.14–1.37) aHR of ESKD = 2.96 (95% CI: 2.49–3.51) aHR of MAKE = 1.66 (95% CI: 1.58–1.74) Gu et al.11 ambidirectional cohort study of patients hospitalized with COVID-19, surviving to discharge from January 2020 to May 2020 in Wuhan, China 1,734 patients admitted to the hospital with COVID-19 who survived to hospital discharge N/A Patients with AKI compared to patients without AKI during acute COVID-19: eGFR decline 8.3% greater (overall) eGFR decline 17.8% greater (stage 3 AKI) OR of reduced kidney function at follow-up = 4.60 (95% CI: 2.10–10.1) * in patients without AKI and with eGFR above 90 at time of acute COVID-19, 13% had eGFR < 90 mL/min/1.73m2 at the time of follow-up AKI = acute kidney injury; eGFR = estimated glomerular filtration rate; ESKD = end-stage kidney disease; MAKE = major adverse kidney events (eGFR decline ≥50%, ESKD, or all-cause mortality); VA = Veterans Affairs References 1. Bhatraju PK , Ghassemieh BJ , Nichols M , Covid-19 in Critically Ill Patients in the Seattle Region - Case Series. N Engl J Med. 2020. 2. Argenziano MG , Bruce SL , Slater CL , Characterization and clinical course of 1000 patients with coronavirus disease 2019 in New York: retrospective case series. BMJ. 2020;369 :m1996.32471884 3. Otto SP , Day T , Arino J , The origins and potential future of SARS-CoV-2 variants of concern in the evolving COVID-19 pandemic. Curr Biol. 2021;31 (14 ):R918–r929.34314723 4. Chan L , Chaudhary K , Saha A , AKI in Hospitalized Patients with COVID-19. J Am Soc Nephrol. 2020;32 (1 ):151–160.32883700 5. Gupta S , Hayek SS , Wang W , Factors Associated With Death in Critically Ill Patients With Coronavirus Disease 2019 in the US. JAMA Internal Medicine. 2020;180 (11 ):1436–1447.32667668 6. Ronco C , Reis T , Husain-Syed F . Management of acute kidney injury in patients with COVID-19. Lancet Respir Med. 2020. 7. Menez S , Moledina DG , Thiessen-Philbrook H , Prognostic Significance of Urinary Biomarkers in Patients Hospitalized With COVID-19. American Journal of Kidney Diseases. 2022;79 (2 ):257–267.e251.34710516 8. Huang C , Huang L , Wang Y , 6-month consequences of COVID-19 in patients discharged from hospital: a cohort study. Lancet. 2021;397 (10270 ):220–232.33428867 9. Sancho Ferrando E , Hanslin K , Hultström M , Larsson A , Frithiof R , Lipcsey M . Soluble TNF receptors predict acute kidney injury and mortality in critically ill COVID-19 patients: A prospective observational study. Cytokine. 2022;149 :155727.34628127 10. Bowe B , Cai M , Xie Y , Gibson AK , Maddukuri G , Al-Aly Z . Acute Kidney Injury in a National Cohort of Hospitalized US Veterans with COVID-19. Clinical Journal of the American Society of Nephrology. 2021;16 (1 ):14. 11. Gu X , Huang L , Cui D , Association of acute kidney injury with 1-year outcome of kidney function in hospital survivors with COVID-19: A cohort study. eBioMedicine. 2022;76. 12. Kellum JA , Nadim MK , Forni LG . Sepsis-associated acute kidney injury: is COVID-19 different? Kidney International. 2020;98 (6 ):1370–1372.32920023 13. Golmai P , Larsen CP , DeVita MV , Histopathologic and Ultrastructural Findings in Postmortem Kidney Biopsy Material in 12 Patients with AKI and COVID-19. Journal of the American Society of Nephrology. 2020;31 (9 ):1944.32675304 14. Su H , Yang M , Wan C , Renal histopathological analysis of 26 postmortem findings of patients with COVID-19 in China. Kidney Int. 2020;98 (1 ):219–227.32327202 15. Alexander MP , Mangalaparthi KK , Madugundu AK , Acute Kidney Injury in Severe COVID-19 Has Similarities to Sepsis-Associated Kidney Injury: A Multi-Omics Study. Mayo Clinic Proceedings. 2021;96 (10 ):2561–2575.34425963 16. Moledina DG , Simonov M , Yamamoto Y , The Association of COVID-19 With Acute Kidney Injury Independent of Severity of Illness: A Multicenter Cohort Study. American Journal of Kidney Diseases. 2021;77 (4 ):490–499.33422598 17. Strohbehn IA , Zhao S , Seethapathy H , Acute Kidney Injury Incidence, Recovery, and Long-term Kidney Outcomes Among Hospitalized Patients With COVID-19 and Influenza. Kidney International Reports. 2021;6 (10 ):2565–2574.34307971 18. Bechman K , Yates M , Mann K , Inpatient COVID-19 mortality has reduced over time: Results from an observational cohort. PLOS ONE. 2022;17 (1 ):e0261142.35025917 19. Charytan DM , Parnia S , Khatri M , Decreasing Incidence of Acute Kidney Injury in Patients with COVID-19 Critical Illness in New York City. Kidney Int Rep. 2021;6 (4 ):916–927.33558853 20. Volbeda M , Jou-Valencia D , van den Heuvel MC , Comparison of renal histopathology and gene expression profiles between severe COVID-19 and bacterial sepsis in critically ill patients. Crit Care. 2021;25 (1 ):202.34112226 21. Li W , Moore MJ , Vasilieva N , Angiotensin-converting enzyme 2 is a functional receptor for the SARS coronavirus. Nature. 2003;426 (6965 ):450–454.14647384 22. Hamming I , Timens W , Bulthuis ML , Lely AT , Navis G , van Goor H . Tissue distribution of ACE2 protein, the functional receptor for SARS coronavirus. A first step in understanding SARS pathogenesis. J Pathol. 2004;203 (2 ):631–637.15141377 23. Santoriello D , Khairallah P , Bomback AS , Postmortem Kidney Pathology Findings in Patients with COVID-19. J Am Soc Nephrol. 2020;31 (9 ):2158–2167.32727719 24. Kudose S , Batal I , Santoriello D , Kidney Biopsy Findings in Patients with COVID-19. J Am Soc Nephrol. 2020;31 (9 ):1959–1968.32680910 25. Nasr SH , Alexander MP , Cornell LD , Kidney Biopsy Findings in Patients With COVID-19, Kidney Injury, and Proteinuria. American Journal of Kidney Diseases. 2021;77 (3 ):465–468.33217501 26. Sharma P , Uppal NN , Wanchoo R , COVID-19-Associated Kidney Injury: A Case Series of Kidney Biopsy Findings. J Am Soc Nephrol. 2020;31 (9 ):1948–1958.32660970 27. Akilesh S , Nast CC , Yamashita M , Multicenter Clinicopathologic Correlation of Kidney Biopsies Performed in COVID-19 Patients Presenting With Acute Kidney Injury or Proteinuria. Am J Kidney Dis. 2021;77 (1 ):82–93.e81.33045255 28. Hassler L , Reyes F , Sparks MA , Welling P , Batlle D . Evidence For and Against Direct Kidney Infection by SARS-CoV-2 in Patients with COVID-19. Clin J Am Soc Nephrol. 2021;16 (11 ):1755–1765.34127485 29. May RM , Cassol C , Hannoudi A , A multi-center retrospective cohort study defines the spectrum of kidney pathology in Coronavirus 2019 Disease (COVID-19). Kidney International. 2021;100 (6 ):1303–1315.34352311 30. Batlle D , Wysocki J , Satchell K . Soluble angiotensin-converting enzyme 2: a potential approach for coronavirus infection therapy? Clinical Science. 2020;134 (5 ):543–545.32167153 31. Monteil V , Kwon H , Prado P , Inhibition of SARS-CoV-2 Infections in Engineered Human Tissues Using Clinical-Grade Soluble Human ACE2. Cell. 2020;181 (4 ):905–913.e907.32333836 32. Wysocki J , Ye M , Hassler L , A Novel Soluble ACE2 Variant with Prolonged Duration of Action Neutralizes SARS-CoV-2 Infection in Human Kidney Organoids. Journal of the American Society of Nephrology. 2021;32 (4 ):795.33526471 33. Hassler L , Wysocki J , Gelarden I , A Novel Soluble ACE2 Protein Provides Lung and Kidney Protection in Mice Susceptible to Lethal SARS-CoV-2 Infection. J Am Soc Nephrol. 2022;33 (7 ):1293–1307.35236774 34. Winkler ES , Bailey AL , Kafai NM , SARS-CoV-2 infection of human ACE2-transgenic mice causes severe lung inflammation and impaired function. Nat Immunol. 2020;21 (11 ):1327–1335.32839612 35. Golden JW , Cline CR , Zeng X , Human angiotensin-converting enzyme 2 transgenic mice infected with SARS-CoV-2 develop severe and fatal respiratory disease. JCI Insight. 2020;5 (19 ). 36. Frithiof R , Bergqvist A , Järhult JD , Lipcsey M , Hultström M . Presence of SARS-CoV-2 in urine is rare and not associated with acute kidney injury in critically ill COVID-19 patients. Critical Care. 2020;24 (1 ):587.32993742 37. Caceres P , Savickas G , Murray S , High SARS-CoV-2 Viral Load in Urine Sediment Correlates with Acute Kidney Injury and Poor COVID-19 Outcome. Journal of the American Society of Nephrology. 2021:ASN.2021010059. 38. Wu H , Larsen CP , Hernandez-Arroyo CF , AKI and Collapsing Glomerulopathy Associated with COVID-19 and APOL1 High-Risk Genotype. Journal of the American Society of Nephrology. 2020;31 (8 ):1688.32561682 39. Larsen CP , Wickman TJ , Braga JR , APOL1 Risk Variants and Acute Kidney Injury in Black Americans with COVID-19. Clinical Journal of the American Society of Nephrology. 2021;16 (12 ):1790. 40. Velez JCQ , Caza T , Larsen CP . COVAN is the new HIVAN: the re-emergence of collapsing glomerulopathy with COVID-19. Nat Rev Nephrol. 2020;16 (10 ):565–567.32753739 41. Hung AM , Shah SC , Bick AG , APOL1 Risk Variants, Acute Kidney Injury, and Death in Participants With African Ancestry Hospitalized With COVID-19 From the Million Veteran Program. JAMA Internal Medicine. 2022. 42. Malik P , Patel U , Mehta D , Biomarkers and outcomes of COVID-19 hospitalisations: systematic review and meta-analysis. BMJ Evidence-Based Medicine. 2021;26 (3 ):107. 43. Garibaldi BT , Fiksel J , Muschelli J , Patient Trajectories Among Persons Hospitalized for COVID-19. Annals of Internal Medicine. 2020;174 (1 ):33–41.32960645 44. Parikh CR , Coca SG , Thiessen-Philbrook H , Postoperative biomarkers predict acute kidney injury and poor outcomes after adult cardiac surgery. Journal of the American Society of Nephrology : JASN. 2011;22 (9 ):1748–1757.21836143 45. Belcher JM , Garcia-Tsao G , Sanyal AJ , Urinary biomarkers and progression of AKI in patients with cirrhosis. Clin J Am Soc Nephrol. 2014;9 (11 ):1857–1867.25183658 46. Coca SG , Garg AX , Thiessen-Philbrook H , Urinary biomarkers of AKI and mortality 3 years after cardiac surgery. Journal of the American Society of Nephrology : JASN. 2014;25 (5 ):1063–1071.24357673 47. Moledina DG , Isguven S , McArthur E , Plasma Monocyte Chemotactic Protein-1 Is Associated With Acute Kidney Injury and Death After Cardiac Operations. Ann Thorac Surg. 2017;104 (2 ):613–620.28223055 48. Menez S , Moledina DG , Garg AX , Results from the TRIBE-AKI Study found associations between post-operative blood biomarkers and risk of chronic kidney disease after cardiac surgery. Kidney International. 2021;99 (3 ):716–724.32721447 49. Vijayan A , Faubel S , Askenazi DJ , Clinical Use of the Urine Biomarker [TIMP-2] x [IGFBP7] for Acute Kidney Injury Risk Assessment. American Journal of Kidney Diseases : The Official Journal of the National Kidney Foundation. 2016;68 (1 ):19–28.26948834 50. Tidbury N , Browning N , Shaw M , Morgan M , Kemp I , Matata B . Neutrophil gelatinase-associated lipocalin as a marker of postoperative acute kidney injury following cardiac surgery in patients with pre-operative kidney impairment. Cardiovasc Hematol Disord Drug Targets. 2019. 51. Palacios Y , Ruiz A , Ramón-Luing LA , Severe COVID-19 Patients Show an Increase in Soluble TNFR1 and ADAM17, with a Relationship to Mortality. Int J Mol Sci. 2021;22 (16 ). 52. Mortaz E , Tabarsi P , Jamaati H , Increased Serum Levels of Soluble TNF-α Receptor Is Associated With ICU Mortality in COVID-19 Patients. Frontiers in Immunology. 2021;12 . 53. Menez S , Coca SG , Moledina DG , Plasma TNFR1 Predicts Major Adverse Kidney Events in Hospitalized Patients with COVID-19 [abstract]. Am Soc Nephrol Kidney Week. 2021. In:2021. 54. Bezerra GF , Meneses GC , Albuquerque PL , Urinary tubular biomarkers as predictors of death in critically ill patients with COVID-19. Biomark Med. 2022;16 (9 ):681–692.35531623 55. Xu K , Shang N , Levitman A , Elevated Neutrophil Gelatinase-Associated Lipocalin Is Associated With the Severity of Kidney Injury and Poor Prognosis of Patients With COVID-19. Kidney International Reports. 2021;6 (12 ):2979–2992.34642645 56. Lopez-Leon S , Wegman-Ostrosky T , Perelman C , More than 50 long-term effects of COVID-19: a systematic review and meta-analysis. Scientific Reports. 2021;11 (1 ):16144.34373540 57. Nugent J , Aklilu A , Yamamoto Y , Assessment of Acute Kidney Injury and Longitudinal Kidney Function After Hospital Discharge Among Patients With and Without COVID-19. JAMA Network Open. 2021;4 (3 ):e211095–e211095.33688965 58. Bowe B , Xie Y , Xu E , Al-Aly Z . Kidney Outcomes in Long COVID. Journal of the American Society of Nephrology. 2021;32 (11 ):2851.34470828

PMC009xxxxxx/PMC9631267.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101313252 34584 J Vis Exp J Vis Exp Journal of visualized experiments : JoVE 1940-087X 35969101 9631267 10.3791/64028 NIHMS1843032 Article Intestinal Epithelial Regeneration in Response to Ionizing Irradiation Orzechowska-Licari Emilia J. 1 LaComb Joseph F. 1 Giarrizzo Michael 1 Yang Vincent W. 12 Bialkowska Agnieszka B. 1 1 Department of Medicine, Renaissance School of Medicine at Stony Brook University 2 Department of Physiology and Biophysics, Renaissance School of Medicine at Stony Brook University Corresponding Author: Agnieszka B. Bialkowska, Agnieszka.Bialkowska@stonybrookmedicine.edu 24 10 2022 27 7 2022 27 7 2022 27 7 2023 185 10.3791/64028This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. The intestinal epithelium consists of a single layer of cells yet contains multiple types of terminally differentiated cells, which are generated by the active proliferation of intestinal stem cells located at the bottom of intestinal crypts. However, during events of acute intestinal injury, these active intestinal stem cells undergo cell death. Gamma irradiation is a widely used colorectal cancer treatment, which, while therapeutically efficacious, has the side effect of depleting the active stem cell pool. Indeed, patients frequently experience gastrointestinal radiation syndrome while undergoing radiotherapy, in part due to active stem cell depletion. The loss of active intestinal stem cells in intestinal crypts activates a pool of typically quiescent reserve intestinal stem cells and induces dedifferentiation of secretory and enterocyte precursor cells. If not for these cells, the intestinal epithelium would lack the ability to recover from radiotherapy and other such major tissue insults. New advances in lineage-tracing technologies allow tracking of the activation, differentiation, and migration of cells during regeneration and have been successfully employed for studying this in the gut. This study aims to depict a method for the analysis of cells within the mouse intestinal epithelium following radiation injury. pmcIntroduction The human intestinal epithelium would cover approximately the surface of half a badminton court if placed completely flat1. Instead, this single cell layer separating humans from the contents of their guts is compacted into a series of finger-like projections, villi, and indentations, crypts that maximize the surface area of the intestines. The cells of the epithelium differentiate along a crypt-villus axis. The villus primarily consists of nutrient-absorbing enterocytes, mucus-secreting goblet cells, and the hormone-producing enteroendocrine cells, while the crypts primarily consist of defensin-producing Paneth cells, active and reserve stem cells, and progenitor cells2,3,4,5. Furthermore, the bi-directional communication these cells have with the stromal and immune cells of the underlying mesenchymal compartment and the microbiota of the lumen generate a complex network of interactions that maintains gut homeostasis and is critical to recovery after injury6,7,8. The intestinal epithelium is the most rapidly self-renewing tissue in the human body, with a turnover rate of 2–6 days9,10,11. During homeostasis, active stem cells at the base of intestinal crypts (crypt base columnar cells), marked by the expression of leucine-rich repeat-containing G-protein coupled receptor 5 (LGR5), rapidly divide and provide progenitor cells that differentiate into all other intestinal epithelial lineages. However, owing to their high mitotic rate, active stem cells and their immediate progenitors are particularly sensitive to gamma-radiation injury and undergo apoptosis following irradiation5,12,13,14. Upon their loss, reserve stem cells and non-stem cells (subpopulation of progenitors and some terminally differentiated cells) within intestinal crypts undergo activation and replenish the basal crypt compartment, which can then reconstitute cell populations of the villi and, thus, regenerate the intestinal epithelium15. Using lineage tracing techniques, multiple research groups have demonstrated that reserve (quiescent) stem cells are capable of supporting regeneration upon the loss of active stem cells13,16,17,18,19,20,21,22. These cells are characterized by the presence of polycomb complex protein 1 oncogene (Bmi1), mouse telomerase reverse transcriptase gene (mTert), Hop homeobox (Hopx), and leucine-rich repeat protein 1 gene (Lrig1). In addition, it has been shown that non-stem cells are capable of replenishing intestinal crypts upon injury23,24,25,26,27,28,29,30,31. In particular, it has been shown that progenitors of secretory cells and enterocytes undergo dedifferentiation upon injury, revert to stem-like cells, and support the regeneration of the intestinal epithelium. Recent studies have identified cells expressing multiple markers that possess the capacity of acquiring stem-like characteristics upon injury (such as DLL+, ATOH1+, PROX1+, MIST1+, DCLK1+)32,33,34,35,36. Surprisingly, Yu et al. showed that even mature Paneth cells (LYZ+) can contribute to intestinal regeneration37. Furthermore, in addition to causing apoptosis of intestinal epithelial cells and disrupting epithelial barrier function, irradiation results in dysbiosis of the gut flora, immune cell activation and the initiation of a pro-inflammatory response, and the activation of mesenchymal and stromal cells38,39. Gamma radiation is a valuable therapeutic tool in cancer treatment, especially so for colorectal tumors40. However, irradiation significantly affects intestinal homeostasis by inducing damage to the cells, which leads to apoptosis. Radiation exposure causes multiple perturbations that slow down a patient’s recovery and is marked by mucosal injury and inflammation in the acute phase and diarrhea, incontinence, bleeding, and abdominal pain long term. This panoply of manifestations is referred to as gastrointestinal radiation toxicity. Additionally, radiation-induced progression of transmural fibrosis and/or vascular sclerosis may only manifest years after the treatment38,41. Simultaneous to the injury itself, radiation induces a repair response in intestinal cells that activates signaling pathways responsible for initiating and orchestrating regeneration42. Radiation-induced small bowel disease can originate from pelvic or abdominal radiotherapy provided to other organs (such as cervix, prostate, pancreas, rectum)41,43,44,45,46. Intestinal irradiation injury is, thus, a significant clinical issue, and a better understanding of the resulting pathophysiology is likely to advance the development of interventions to alleviate the gastrointestinal complications associated with radiotherapy. There are other techniques that allow for investigating the regenerative purpose of the intestinal epithelium apart from radiation. Transgenic and chemical murine models to study inflammation and the regeneration thereafter have been developed47. Dextran sodium sulfate (DSS) induces inflammation in the intestine and leads to the development of characteristics similar to those of inflammatory bowel disease48. A combination of DSS treatment with the pro-carcinogenic compound azoxymethane (AOM) can result in the development of colitis-associated cancer48,49. Ischemia reperfusion-induced injury is another method employed to study the regenerative potential of the intestinal epithelium. This technique requires experience and surgical knowledge50. Furthermore, the aforementioned techniques cause different types of injury than radiation and may lead to the involvement of different mechanisms of regeneration. In addition, these models are time-consuming, while the radiation technique is fairly brief. Recently, in vitro methods utilizing enteroids and colonoids generated from the intestine and colon have been used in combination with radiation injury to study the mechanisms of intestinal regeneration51,52. However, these techniques do not fully recapitulate the organ they model53,54. The protocol presented includes the description of a murine model of gamma-radiation injury in combination with a genetic model that, following tamoxifen treatment, permits tracing of lineages originating from the reserve stem cell population (Bmi1-CreER;Rosa26eYFP). This model utilizes a 12 Gy total-body irradiation, which induces significant enough intestinal injury to activate reserve stem cells while still allowing for the subsequent investigation of intestinal regenerative capability within 7 days of injury55. Protocol All mice were housed in the Division of Laboratory Animal Resources (DLAR) at Stony Brook University. The Stony Brook University Institutional Animal Care and Use Committee (IACUC) approved all studies and procedures involving animal subjects. Experiments involving animal subjects were conducted strictly in accordance with the approved animal handling protocol (IACUC #245094). NOTE: Mouse strains Bmi1-CreER and Rosa26eYFP were commercially obtained (see Table of Materials) and crossed to obtain Bmi1-CreER;Rosa26eYFP (Bmi1ctrl) mice, as described previously56,57,58. 1. Housing of Bmi1-Cre ER;Rosa26 eYFP mice Keep the mice under pathogen-free conditions at constant temperature and humidity, in a 12 h/12 h light/dark cycle, with water and normal chow ad libitum. Prior to the experiments, confirm mice genotypes using a standard PCR genotyping technique57,58. 2. Preparation of animals and materials Transfer mice to the conventional housing room at least 7 days before any experimentation to let the mice acclimatize. Match experimental and control animals according to gender and age. Subject the control animals to tamoxifen-induced Cre-mediated recombination but not to irradiation (sham treatment, 0 Gy) while ensuring that the experimental animals receive the tamoxifen injection and are exposed to gamma irradiation. Prepare the tamoxifen solution. Resuspend tamoxifen powder in sterile corn oil at a concentration of 30 mg/mL. Sonicate for 3 min in cycles of 30 s ON and 30 s OFF with 60% amplitude and then rotate in the dark for 1 h at room temperature (RT). Tamoxifen solution can be frozen at −20 °C but do not freeze/thaw it more than once. Preferably, prepare a fresh solution each time and do not store it. NOTE: Tamoxifen is light sensitive. Wrap it with tin foil during room temperature rotation. CAUTION: Tamoxifen is a potentially hazardous substance: Health Hazard (GHS08) and Environment Hazard (GHS09). Prepare 5-ethynyl-2′-deoxyuridine (EdU) stock solution. Resuspend EdU powder in 1/5 of the total volume of sterile dimethyl sulfoxide (DMSO) and slowly add the remaining 4/5 of the total volume of sterile ultrapure water. When completely dissolved, aliquot and store at −20 °C. CAUTION: 5-ethynyl-2′-deoxyuridine (EdU) and dimethyl sulfoxide (DMSO) are potentially hazardous substances: Health Hazard (H340 and H631, and H227, H315, and H319, respectively). EdU may cause genetic defects and is suspected of damaging fertility or unborn children, and DMSO can cause skin and eye irritation. Prepare reagents for tissue collection and fixation: dilute ethanol to prepare a 70% water solution, cool down DPBS to 4 °C, and prepare modified Bouin’s fixative buffer (50% ethanol and 5% acetic acid in distilled H2O) and 10% buffered formalin. CAUTION: Ethyl alcohol is a potentially hazardous substance: Health Hazard (H225-H319), flammable (GHS02), and causes acute toxicity (GHS07). Acetic acid is a potentially hazardous substance: Health Hazard (H226-H314), flammable (GHS02), and corrosive (GHS05). Formalin is a potentially hazardous substance: Health Hazard (H350, H315, H317, H318, H370), and corrosive. Formalin may cause cancer, skin and eye irritation, damage to organs, and allergic skin reactions. Prepare the equipment necessary for euthanasia according to an approved method (CO2 chamber, for instance), a mice dissection kit (scissors, forceps), Petri dishes, a 16 G gavage needle attached to a 10 mL syringe to flush the intestines, and histological cassettes. 3. Total-body gamma irradiation (TBI) and tissue collection Two days prior to gamma-irradiation exposure, inject the experimental animals with a single dose of tamoxifen to induce Cre-mediated recombination and BMI1/EYFP+ cell linage tracing. Weigh each animal and calculate a dose of 40 mg/kg of body weight of tamoxifen resuspended in corn oil. Disinfect the abdominal area with 70% ethanol and administer tamoxifen intraperitoneally using a 27G needle attached to a 1 mL syringe. Observe animals for the following 48 h to exclude potential tamoxifen toxicity. Forty-eight hours post injection transfer the animals to the irradiation room. Calculate the irradiation exposure time released by the 137Cs source according to the current exactor dose rate. For example, if the current dose rate equals 75.9 rad/min (0.759 Gy min−1 = 75.9 cGy m−1), the time of exposure in minutes is calculated as the desired dose/0.759. To irradiate the animals to a dose of 12 Gy TBI, the exposure time is ~15.81 min (15 min and 48 s). Disinfect the sample chamber in the gamma irradiator with 70% ethanol solution; place the absorbent mat and animals inside the sample chamber. NOTE: Sham-treated animals should be placed in the irradiation room but not exposed to gamma irradiation. Place the lid and close the chamber. Program the gamma irradiator to expose animals to 12 Gy TBI. Turn on the machine by turning the key into the START position. Enter the operator number using a numeric keyboard and confirm by pressing ENTER. Enter the PIN number using a numeric keyboard and confirm by pressing ENTER. Press 1 for options. Press 1 for timer settings. Press 1 for irradiation time. Enter the timer setting for the next cycle by using a numeric keyboard (hh:mm:ss format) and confirm by pressing ENTER. Confirm the settings once more by pressing ENTER. Go back to the home menu by pressing CLEAR 2x. Press START to initiate the exposure. Leave the room for the time of active exposure. The machine will stop automatically after the preset time elapses and will start beeping. CAUTION: Cesium-137 (137Cs) is a deadly hazardous substance (Health Hazard: H314). External exposure to large amounts of 137Cs can cause burns, acute radiation sickness, and even death. Turn off the machine by turning the key into the STOP position. Open the sample chamber, take off the lid, transfer the animals back to the cage, and disinfect the sample chamber with 70% ethanol solution. Transfer the animals back to the conventional housing room and observe their condition post treatment. Monitor the animals’ weight every day and euthanize them immediately (despite the planned time point) if any symptoms of altered well-being, distress, or weight loss exceeding 15% of the starting body weight are observed. Three hours prior to the planned euthanasia disinfect the abdominal area and, using a 28G insulin syringe, inject the mice with 100 μL of EdU stock solution (administered intraperitoneally). Collect proximal intestines at 0 h, 3 h, 6 h, 24 h, 48 h, 72 h, 96 h, and 168 h post irradiation. First, sacrifice the mice by asphyxiation. Place the animals in a chamber connected to a CO2 gas source. Observe the animals while gas is flowing into the chamber and close the valve when the mice are unconscious, and all movement stops. Leave the animals for another few minutes in the chamber and perform cervical dislocation to ensure euthanasia. CAUTION: Carbon dioxide is a hazardous substance (H280). Contains gas under pressure; may explode if heated. May displace oxygen and cause rapid suffocation. Then, dissect the proximal part of the small intestine, remove attached tissues, flush with cold DPBS using a 16 G straight feeding needle attached to a 10 mL syringe, fix with modified Bouin’s fixative buffer using a 16 G straight feeding needle attached to a 10 mL syringe, cut open longitudinally, and roll the proximal intestines using the swiss-roll technique as described previously59. Place the tissues in a histological cassette and leave for 24–48 h at room temperature in a container with 10% buffered formalin. The volume of the formalin should be sufficient to fully cover the histological cassettes. When the incubation time elapses, using forceps, transfer the histological cassettes to the container filled with 70% ethanol. Then, proceed with tissue paraffin embedding. NOTE: Tissue paraffin embedding was performed by the Research Histology Core Laboratory at Stony Brook University. The procedure can be stopped here, and paraffin blocks can be stored at room temperature. CAUTION: Formalin is a potentially hazardous substance: Health Hazard (H350, H315, H317, H318, H370), and corrosive. Formalin may cause cancer, skin and eye irritation, damage to organs, and allergic skin reactions. 4. Histological analysis Cool down the histological cassettes containing the paraffin-embedded tissue specimens by placing them on ice for at least 1 h. Using a microtome, prepare 5 μm thick sections for staining. Every tissue should be sliced horizontally to cover the entire rolled specimen. After cutting the paraffin block, transfer the tissue sections to a water bath warmed up to 45 °C and place the sections on charged slides. Leave the slides on a rack overnight at room temperature to dry. NOTE: The procedure can be stopped here, and the slides can be stored at room temperature. Place the slides in a slide holder and bake in a 65 °C oven overnight. The slides can be baked for a shorter time, 1–2 h. However, this is not advisable in the case of freshly (less than 1 month old) cut slides. NOTE: Place the manual slide staining set under the chemical hood and perform the incubations with xylene, ethanol, and hematoxylin under the chemical hood. The next day, cool down the slides by placing them in a slide holder under the chemical hood for 10 min. Deparaffinize the tissues by placing the slide holder in a container filled with 100% xylene for 3 min. Repeat incubation using fresh 100% xylene. NOTE: From that moment, ensure the specimens are kept wet at all times. CAUTION: Xylene is a potentially hazardous substance: flammable (GHS02), causing acute toxicity (GHS07), and a Health hazard (H226 - H304 - H312 + H332 - H315 - H319 - H335 - H373 - H412). The potential hazardous effects are that it is fatal if swallowed or enters airways and can cause skin, eye, and respiratory irritation. Additionally, it may affect motor functions by causing drowsiness or dizziness and cause organ damage in case of prolonged or repeated exposure. Rehydrate sections in an ethanol gradient by placing the slide holder sequentially in containers filled with the following solutions: 100% ethanol, 2 min; 95% ethanol, 2 min; 70% ethanol, 2 min. Transfer the slide holder to the container filled with distilled water and rinse in running distilled water for 2 min. Place the slide holder in a container filled with hematoxylin solution, and stain for 5 min (under the chemical hood). Transfer the slide holder to the container filled with tap water and rinse in running tap water until the hematoxylin staining turns bluish, typically after 2 min. Transfer the slide holder to the container filled with 5% (w/v) lithium carbonate solution. Dip 10x. CAUTION: Lithium carbonate is a potentially hazardous substance: causing acute toxicity (GHS07), and a Health Hazard (H302 - H319). It is harmful if swallowed, harmful in contact with skin, and causes serious eye irritation. Transfer the slide holder to the container filled with distilled water and rinse in running distilled water for 2 min. Place the slide holder in a container filled with eosin and stain for 5 min. Transfer the slide holder to the container filled with 70% ethanol and rinse briefly (10 s). Dehydrate the sections by placing the slide holder sequentially in the containers filled with ethanol gradient solutions: 95% ethanol, 5 dips; 100% ethanol, 5 dips. Clear the slides by placing the slide holder in the container filled with 100% xylene for 3 min. Repeat incubation using fresh 100% xylene. Take the slide out of the rack, dry the area around the tissues using a paper towel, and depending on the size of the specimen, add one drop or two of xylene-based mounting medium on the surface of the specimen. Mount by gently placing a coverslip on top of the glass slide (be careful not to leave any air bubbles). Press gently if needed. The whole surface of the glass should be covered and sealed. Dry the slides by leaving them under the chemical hood overnight. Typically, the mounting medium solidifies in less than 24 h. To ensure the slides are dry, a sample slide can be made. Using an empty slide, place a drop of the mounting medium and leave under the chemical hood overnight. The next day, touch the drop and check if it is solidified. When the slides dry out, analyze the histology of the tissues using a light microscope or a more advanced microscope with a bright field channel. Place a microscope slide on the stage, move until the sample is in the center of the field of view, adjust the focus using the focus knob, and use a 10x and/or 20x magnification objective lens to analyze the histology in comparison to control tissues (sham irradiated). If the microscope is equipped with a camera, take images as well. NOTE: The procedure can be stopped here; the slides can be stored at room temperature and analyzed later. Do not keep the slides longer than 30 days as the staining slowly fades away. 5. Immunofluorescence staining Cool down the histological cassettes containing paraffin-embedded tissue specimens by placing them on ice for at least 1 h. Using a microtome, prepare 5 μm thick sections for staining. Every tissue should be sliced horizontally to cover the entire rolled specimen. After cutting the paraffin block, transfer the tissue sections to a water bath warmed to 45 °C and place the sections on charged slides. Leave the slides on a rack overnight at room temperature to dry. NOTE: The procedure can be stopped here, and the slides can be stored at room temperature. Place the slides in a slide holder and bake in a 65 °C oven overnight. The slides can be baked for a shorter time, 1–2 h. However, this is not advisable in the case of freshly (less than 1 month ago) cut slides. NOTE: Place the manual slide staining set under the chemical hood and perform the incubations with xylene, ethanol, and H2O2/methanol under the chemical hood. The next day, cool down the slides by placing them in a slide holder under the chemical hood for 10 min. Deparaffinize the tissues by placing the slide holder in a container filled with 100% xylene for 3 min. Repeat incubation using fresh 100% xylene. NOTE: From that moment, ensure the specimens are kept wet at all times. Quench endogenous peroxidase by incubating the slides for 30 min in a container filled with 2% hydrogen peroxide solution in methanol under the chemical hood. CAUTION: Methyl alcohol is a potentially hazardous substance: Health Hazard (H225-H301, H311, H331-H370), flammable (GHS02), and causes acute toxicity (oral, dermal, inhalation) (GHS08). Hydrogen peroxide is a potentially hazardous substance: corrosive (GHS05) and causes acute toxicity (GHS07). Rehydrate the sections in an ethanol gradient by placing the slide holder sequentially in containers filled with the following solutions: 100% ethanol, 3 min; 95% ethanol, 3 min; 70% ethanol, 3 min. Transfer the slide holder to a container filled with distilled water and rinse in running distilled water for 2 min. To retrieve the antigens, transfer the slide holder to a container with 250 mL of citrate buffer solution (10 mM sodium citrate, 0.05% Tween-20, pH 6.0) and cook at 110 °C for 10 min using a decloaking chamber. Transfer the whole container to the cold room and allow gradual cooling down over the course of 30 min. After 30 min, replace the citrate buffer solution with distilled water and rinse in running distilled water until all floating paraffin pieces are removed. Take the slides out of the holder (one at a time), tap on a paper towel, and carefully dry the area around the specimen with a paper towel. Draw a closed shape around the specimen using a pen that provides a hydrophobic barrier (PAP pen). Place in a humidified chamber. Block with 5% bovine serum albumin (BSA) in TBS-Tween by adding 200 μL of the solution onto the surface of a specimen. Ensure there is no leakage. Incubate at 37 °C in a humidified chamber for 30 min. Remove the blocking solution by tapping the glass slide on a paper towel. Place back in a humidified chamber. Add 100 μL of the appropriate concentration of the primary antibodies resuspended in a blocking buffer and incubate at 4 °C with gentle rocking overnight. For BMI1/EYFP, use chicken anti-GFP (dilution 1:500); for Ki-67, use rabbit anti-Ki-67 (dilution 1:200). Remove the antibody solution by tapping the glass slide on a paper towel; place the slides in a slide holder and transfer to the container filled with TBS-Tween. Wash the slides with shaking for 5 min. Repeat 3x. Each time, use a fresh portion of TBS-Tween buffer. Take the slides out of the holder (one at a time), remove excess washing solution by tapping the glass slide on a paper towel, and place in a humidified chamber. Add 100 μL of the appropriate concentration of secondary antibodies resuspended in a blocking buffer, and incubate at 37 °C for 30 min. Secondary antibodies should be conjugated with a fluorophore. For BMI1/EYFP, use donkey anti-chicken Alexa Fluor 647 (dilution 1:500); for Ki-67, use goat anti-rabbit Alexa Fluor 488 (dilution 1:500). Remove the antibody solution by tapping the glass slide on a paper towel; place the slides in a slide holder and transfer to the container filled with TBS-Tween. Wash the slides with shaking for 5 min. Repeat 3x. Each time, use a fresh portion of TBS-Tween buffer. Take the slides out of the holder (one at a time), remove excess washing solution by tapping the glass slide on a paper towel, and place in a humidified chamber. Add 100 μL of the EdU staining solution prepared according to the manufacturer’s instructions and using Alexa Fluor 555 fluorophore. Remove the EdU solution by tapping the glass slide on a paper towel; place the slides in a slide holder and transfer to the container filled with TBS-Tween. Wash the slides with shaking for 5 min. Repeat 2x. Each time, use a fresh portion of TBS-Tween buffer. Take the slides out of the holder (one at a time); remove excess washing solution by tapping the glass slide on a paper towel and place in a humidified chamber. Perform Hoechst 33258 counterstaining by adding 100 μL of the Hoechst 33258 solution (dilution 1:1000 in TBS-Tween) onto the surface of the specimen. Incubate in the dark at room temperature for 5 min. Remove the Hoechst 33258 solution by tapping the glass slide on a paper towel; place the slides in a slide holder and transfer to a container filled with TBS-Tween. Wash the slides with shaking for 5 min. Take the slide out of the slide holder, dry the area around the tissues using a paper towel, and depending on the size of the specimen, add one drop or two of aqua-based mounting medium onto the surface of the specimen. Mount by gently placing a coverslip on top of the glass slide (be careful not to leave any air bubbles). Press gently if needed. The whole surface of the glass should be covered and sealed. Dry the slides by leaving them in the dark at room temperature overnight. Typically, the mounting media solidifies in less than 24 h. To ensure the slides are dry, a sample slide can be made. Use an empty slide; place a drop of the mounting medium and leave under the chemical hood overnight. The next day, touch the drop and check if it is solidified. When the slides dry out, stained tissues can be analyzed using a fluorescence microscope equipped with emission wavelength filters allowing the visualization of Ki-67, EdU, and BMI1/EYFP staining (535 nm, 646 nm, and 700 nm, respectively). Place a microscope slide on the stage, move until the sample is in the center of the field of view, adjust the focus using the focus knob, and use the 10x and/or 20x magnification objective lens to analyze the staining in comparison to control tissues (sham irradiated). If the microscope is equipped with a camera, images can be taken as well. Take images for each channel separately and merge later. NOTE: The procedure can be stopped here; the slides can be stored at 4 °C and analyzed later. Do not keep the slides longer than 14 days as the staining slowly fades away. 6. TUNEL staining Cool down the histological cassettes containing paraffin-embedded tissue specimens by placing them on ice for at least 1 h. Using a microtome, prepare 5 μm thick sections for staining. Every tissue should be sliced horizontally to cover the entire rolled specimen. After cutting the paraffin block, transfer the tissue sections to a water bath warmed up to 45 °C and place the sections on charged slides. Leave the slides on a rack overnight at room temperature to dry. NOTE: The procedure can be stopped here, and the slides can be stored at room temperature. Place the slides in a slide holder and bake in a 65 °C oven overnight. NOTE: The slides can be baked for a shorter time, 1–2 hr. However, it is not advisable in the case of freshly (less than 1 month ago) cut slides. Place the manual slide-staining set under the chemical hood and perform the incubations with xylene and ethanol under the chemical hood. The next day, cool down the slides by placing them in a slide holder under the chemical hood for 10 min. Deparaffinize the tissues by placing the slides holder in a container filled with 100% xylene for 3 min. Repeat incubation using fresh 100% xylene. NOTE: From this point, ensure the specimens are kept wet at all times. Rehydrate the sections in an ethanol gradient by placing the slides holder sequentially in containers filled with the following solutions: 100% ethanol, 3 min; 95% ethanol, 3 min; 90% ethanol, 3 min; 80% ethanol, 3 min; 70% ethanol, 3 min. Transfer the slide holder to a container filled with distilled water and rinse in running distilled water for 1 min. Take the slides out of the holder (one at a time), tap on a paper towel, and carefully dry the area around the specimen with a paper towel. Draw a closed shape around the specimen using a pen that provides a hydrophobic barrier (PAP pen). Place in a humidified chamber. Incubate with DNase-free proteinase K. Add 100 μL of the DNase-free proteinase K solution (dilution 1:25 in DPBS) onto the surface of the specimen within the hydrophobic barrier. Incubate at room temperature in a humidified chamber for 15 min. CAUTION: Proteinase K is a potentially hazardous substance: Health Hazard (H315 - H319 - H334). Remove the proteinase K solution by tapping the glass slide on a paper towel; place the slides in a slide holder and transfer to the container filled with DPBS. Wash the slides with shaking for 5 min. Repeat 2x. Each time, use a fresh portion of DPBS. Prepare the TUNEL reaction mixture: Mix label solution (1x) with enzyme solution (10x). Pipette well. Take the slides out of the holder (one at a time); remove excess washing solution by tapping the glass slide on a paper towel and place in a humidified chamber. Add 50 μL of TUNEL reaction mixture onto the surface of the specimen and incubate in the dark at 37 °C in a humidified chamber for 60 min. For a negative control, use label solution only (without TdT). Remove the TUNEL solution by tapping the glass slide on a paper towel; place the slides in a slide holder and transfer to a container filled with DPBS. Wash the slides with shaking for 5 min. Repeat 2x. Each time, use a fresh portion of DPBS. Take the slides out of the holder (one at a time); remove excess washing solution by tapping the glass slide on a paper towel and place in a humidified chamber. Perform Hoechst 33258 counterstaining by adding 100 μL of the Hoechst 33258 solution (dilution 1:1000 in TBS-Tween) onto the surface of the specimen. Incubate in the dark at room temperature for 5 min. Remove Hoechst 33258 solution by tapping the glass slide on a paper towel; place the slides in a slide holder and transfer to the container filled with DPBS. Wash the slides with shaking for 5 min. Take the slide out of the slide holder; dry the area around the tissues using a paper towel, and depending on the size of the specimen, add one drop or two of aqua-based mounting medium onto the surface of the specimen. Mount by gently placing a coverslip on top of the glass slide (be careful not to leave any air bubbles). Press gently if needed. The whole surface of the glass should be covered and sealed. Dry the slides by leaving them in the dark at room temperature overnight. Typically, the mounting medium solidifies in less than 24 h. To ensure the slides are dry, a sample slide can be made. Place a drop of the mounting medium and leave in the dark at room temperature overnight. The next day, touch the drop and check if it is solidified. When the slides dry out, stained tissues can be analyzed using a fluorescence microscope equipped with emission wavelength filters allowing the visualization of TUNEL staining (535 nm). Place a microscope slide on the stage, move until the sample is in the center of the field of view, adjust the focus using the focus knob, and use a 10x and/or 20x magnification objective lens to analyze the staining in comparison to control tissues (sham irradiated). If the microscope is equipped with a camera, images can be taken as well. Take images for each channel separately and merge later. NOTE: The procedure can be stopped here; slides can be stored at 4 °C and analyzed later. Do not keep the slides longer than 14 days as the staining slowly fades away. Representative Results The use of 12 Gy total-body irradiation (TBI) in combination with murine genetic lineage tracing allows for a thorough analysis of the consequences of radiation injury in the gut. To start, Bmi1-CreER;Rosa26eYFP mice received a single tamoxifen injection, which induces enhanced yellow fluorescent protein (EYFP) expression within a Bmi1+ reserve stem cell population. Two days subsequent to the tamoxifen injection, the mice underwent irradiation or sham irradiation. Three hours before euthanasia, the mice were injected with EdU. Following euthanasia, small intestine specimens were collected for analysis at 0 h, 3 h, 6 h, 24 h, 48 h, 72 h, 96 h, and 7 days post irradiation. Representative hematoxylin and eosin (H&E) images (Figure 1) from the aforementioned time course illustrate the intestinal epithelium response to injury. The 0 h time point is illustrative of homeostatic intestinal epithelium (Figure 1); this is followed by the apoptotic phase-characterized by a loss of cells within crypt compartments between 3 h and 48 h (Figure 1). Then follows the regenerative phase, where highly proliferative cells can be found populating the crypts between 72 h and 96 h (Figure 1). Finally, this leads to the normalization phase, here represented by the 7 day time point (Figure 1). Figure 2 illustrates changes in the proliferative status of intestinal crypt cells assessed by immunofluorescent staining of EdU (marker of cells in S-phase) and Ki-67 (marker of cells in G1, S, and G2/M phases), while lineage tracing (EYFP+ cells) marks cells originating from Bmi1+ reserve stem cells. During homeostasis (Figure 2, 0 h sham), the Bmi1+-positive cells marked by the expression of EYFP are restricted to the +4-+6 position within crypts. During the apoptotic phase (24 h and 48 h), EYFP-positive cells decrease in number (Figure 2). The regenerative phase (72 h and 96 h) is characterized by rapidly proliferating Bmi1-positive cells and their progenitors (Figure 2); by 7 days post-irradiation, further proliferation and migration have replenished the intestinal crypt cells and restored intestinal epithelium integrity (Figure 2). In sham-irradiated mice (control), EdU and Ki-67 stain EYFP negative proliferative cells of the intestinal crypts, typical of normal intestinal homeostasis, extending from active stem cells through the transit-amplifying zone (Figure 2). Radiation damage causes a loss of highly proliferative cells during the apoptotic phase, as shown by the reduction of EdU and Ki-67 staining (Figure 2). The activation of reserve stem cells (in this example, EYFP marked Bmi1+ positive cells) and their proliferation result in an increase in EdU and Ki-67 positive cells clearly visible at 72 h and 96 h post injury, where crypts are comprised almost exclusively of cells co-stained with EYFP, EdU, and Ki-67. The observed levels of proliferation normalize 7 days post injury (Figure 2). In order to illustrate apoptosis caused by radiation damage, TUNEL staining was performed, and representative images are included in Figure 3. During homeostasis (Figure 3), few cells are undergoing apoptosis, typical of the normally rapid intestinal epithelial turnover. A clear increase in TUNEL staining can be observed late in the apoptotic phase (24 h and 48 h) following irradiation (Figure 3). During the regenerative phase, TUNEL staining steadily decreases within the regenerating crypts and is again nearly absent when the crypts have normalized (Figure 3). The presented protocol describes a simple and approachable immunofluorescent staining method utilizing a straightforward and widely employed combination of labeling techniques (TUNEL, EdU, Ki-67, EYFP) that is able to provide insights into the behavior of intestinal cells following injury (here, irradiation). This approach can be employed to study the mechanisms regulating regenerative processes in the intestinal epithelium in specific circumstances. Employing this and similar approaches will illuminate alternate regenerative processes following different intestinal insults and, further, will allow the investigation of therapeutic strategies for the prevention of loss of barrier function. Discussion This protocol describes a robust and reproducible radiation injury model. It allows for the precise analysis of the changes in the intestinal epithelium over the course of 7 days post injury. Importantly, the selected time points reflect crucial stages of injury and are characterized by distinct alterations to the intestine (injury, apoptosis, regeneration, and normalization phases)60. This model of irradiation has been established and carefully assessed, demonstrating a suitable manifestation of injury to mimic that experienced by patients undergoing radiotherapy61. More than half of patients with colorectal cancer and gynecological cancers undergo abdominal radiotherapy40,62,63. In spite of advanced methods for targeting radiotherapy, patients experience irradiation of healthy intestinal tissue, which produces pathophysiology very similar to that observed with the total-body irradiation model presented here. Therefore, the radiation model described herein can potentially address the repercussions of accidental radiation exposure of patients and/or medical personnel and, thus, is vital for human health. The Bmi1-CreER;Rosa26eYFP mice model carries tamoxifeninducible Cre under the transcriptional control of the mouse Bmi1 (Bmi1 polycomb ring finger oncogene) promoter. Additionally, these mice carry an R26-stop-EYFP sequence, where the STOP codon is flanked by loxP sides followed by the Enhanced Yellow Fluorescent Protein gene (EYFP) inserted into the Gt(ROSA)26Sor locus. The expression of EYFP is blocked by an upstream loxP-flanked STOP sequence but can be activated by tamoxifen-induced, Cre-mediated targeted deletions, allowing lineage tracing of Bmi1-expressing cells, a reserve stem cell-specific subpopulation of crypt cells. The radiation protocol in combination with the Bmi1-CreER;Rosa26eYFP mice model has been employed previously to study the role of transcription factors in regulating intestinal regeneration18,56. The representative results presented in Figure 1 to Figure 3 show results akin to those presented in recent publications18,56. Of note, Bmi1-CreER;Rosa26eYFP mice are only one of multiple transgenic animal models employed to study intestinal regeneration. Other murine lineage tracing models have been combined with radiation to demonstrate the roles of specific cells and pathways in the regenerative capacity of the intestinal epithelium18,26,32,33,34,35,36,37,56,64,65,66,67. These models are designed to investigate the role of different populations of reserved stem cells, explore the precursors of secretory and enterocyte lineages, and study the function of specific genes within each subpopulation of the intestinal epithelial cells. Combined results from these investigations will help to expand the understanding of the regenerative capability of the intestinal epithelium. Importantly, with minor modifications, this technique should allow for the investigation of the relationships between microbiota and different epithelial, stromal, and immune cell populations upon radiation injury. Introducing distinct lineage-tracing animal models will allow the study of real-time behavior and the interaction of various subpopulations of cells post injury. Alternatively, the staining of tissues could be replaced by RNA-sequencing, single-cell RNA-sequencing, fluorescence-activated cell sorting (FACS), or proteomic analysis to glean more in-depth knowledge about the role of specific cell lineages during intestinal regeneration following injury68,69. This injury model could be utilized to test new treatment modalities intended to minimize the side effects of irradiation, shorten recovery time, and improve the quality of patient care70. Although the radiation model described here can serve as a useful tool to study the regeneration of the intestinal epithelium, it also has several limitations. The total-body radiation dose invariably results in subsequent hematopoietic syndrome, which hinders the investigation of intestinal regeneration. This can be averted by bone marrow transplantation. Further, the modification to an abdominal-only injury protocol may potentially ameliorate some of the side effects of the irradiation71. However, this model requires additional steps that may affect gut response to the injury (e.g., anesthesia). Incidentally, abdominal radiation produces radiation-induced gastrointestinal syndrome (RIGS) without hematopoietic syndrome, differing from whole-body irradiation (WBI), in that WBI produces similar GI symptomology but includes partial loss of hematopoietic function following injury71,72. Of note, genetically engineered animal models such as those described in this protocol are sophisticated and allow for the lineage tracing of specific cells. However, these models, though very beneficial, require additional effort to maintain. Transgenic animal models that allow for conditional gene deletions are difficult to generate, and their timely deletion upon injury may not be easily accomplished. Moreover, most transgenic models allow only for a conditional deletion or inactivation of a gene of interest but are not inducible and thus, are incapable of restoring gene function and may impede the studies. Furthermore, transgenic animal models usually require treatment with additional chemicals (e.g., tamoxifen, doxycycline) to achieve inactivation or overexpression of the gene of interest. Treatments with these substances may, to some degree, alter the response of the intestinal epithelium or add to the severity of the injury73,74,75,76. In addition, the induction of lineage tracing by tamoxifen can be introduced closer to the time of radiation injury to improve the labeling of the cells involved in regeneration (e.g., 6–24 h before injury). Thus, it is possible to substitute a transgenic model with wild-type mice and a combination of IF staining (e.g., Ki-67, PCNA, EdU, TUNEL, Caspase 3). This model allows for the analysis of the regenerative capacity of the intestine, although it does not permit the study of a specific subpopulation of the cells. The protocol described in this manuscript has several crucial steps. It is of utmost importance to ensure that healthy mice are used for experimentation, as they will be exposed to radiation and treatment with chemicals. If experiments with different mice groups are performed days, weeks, or months apart, it is good practice to adhere to the same timeline, e.g., tamoxifen injections at the same time of the day. The irradiation process should be done promptly, and mice should be returned to their original cages straight away. In addition, providing softened food and plenty of water can help to ameliorate some of the adverse effects of radiation. The swiss-roll technique is suggested to allow for comprehensive analysis of the intestinal tissue59. This technique is not trivial, and prior tryouts should be done to ensure that impeccable quality of the tissue for staining is achieved. All solutions for mice experiments and staining should be prepared freshly and stored under appropriate conditions. When working with antibodies, it is important to test slides using positive and negative controls and, if possible, use antibodies with the same catalog number and lot number. Immunofluorescence slides should be stored at 4 °C and imaged within 1 week of preparation. Prolonged storage, even at −20 °C, may result in the loss of signal. H&E slides can be stored at room temperature as there is no concern of loss of fluorophore stability. In summary, the lineage-tracing radiation injury model presented here is a reproducible and relevant model to track intestinal cell fate following insult and can be combined with diverse murine models and molecular techniques to improve the understanding of the pathophysiology of the intestinal epithelium. Insight into the molecular and cellular mechanisms governing intestinal epithelium regeneration could lead to the development of new treatments and beneficial therapeutic interventions. Supplementary Material Supplementary Material Acknowledgments The authors wish to acknowledge the Stony Brook Cancer Center Histology Research Core for expert assistance with tissue specimen preparation and the Division of Laboratory Animal Resources at Stony Brook University for assistance with animal care and handling. This work was supported by grants from the National Institutes of Health DK124342 awarded to Agnieszka B. Bialkowska and DK052230 to Dr. Vincent W. Yang. Figure 1: Representative images of hematoxylin and eosin (H&E)-stained sections of small intestinal tissues following a time course after total body irradiation. Bmi1-CreER;Rosa26eYFP mice received one dose of tamoxifen 2 days before 12 Gy total-body irradiation or sham irradiation. Mice were sacrificed, and small intestine tissues were collected at time points indicated in the figure. All images were taken at 10x magnification. Figure 2: Representative images of immunofluorescent staining of small intestine sections following irradiation. Bmi1-CreER;Rosa26eYFP mice received one dose of tamoxifen 2 days before 12 Gy total-body irradiation or sham irradiation. Mice were sacrificed, and small intestine tissues were collected at time points indicated in the figure. Tissue sections were stained with EdU to mark cells in S-phase (pink) and with Ki-67 to mark cells in G0, S, and G2/M phases (yellow). EYFP (green) marks Bmi1+ cells and their progenitors and demonstrates lineage tracing. The sections were further counterstained with DAPI (blue) to visualize nuclei. The data are shown as merged images at 10x magnification and insets of individual stains at 20x magnification (these originated from 10x images). Figure 3: Representative images of immunofluorescent TUNEL staining of small intestine sections upon irradiation. Bmi1-CreER;Rosa26eYFP mice received one dose of tamoxifen 2 days before 12 Gy total-body irradiation or sham irradiation. Mice were sacrificed, and small intestine tissues were collected at time points indicated in the figure. TUNEL (red) stains apoptotic cells. These images were counterstained with DAPI (blue) to visualize nuclei. The images shown were taken at 20x magnification. A complete version of this article that includes the video component is available at http://dx.doi.org/10.3791/64028. Disclosures The authors have no conflicts of interest. References 1. Helander HF , Fandriks L Surface area of the digestive tract - Revisited. Scandinavian Journal of Gastroenterology. 49 (6 ), 681–689 (2014).24694282 2. van der Flier LG , Clevers H Stem cells, self-renewal, and differentiation in the intestinal epithelium. Annual Review of Physiology. 71 , 241–260 (2009). 3. Clevers H The intestinal crypt, a prototype stem cell compartment. Cell. 154 (2 ), 274–284 (2013).23870119 4. Barker N Identification of stem cells in small intestine and colon by marker gene Lgr5. Nature. 449 (7165 ), 1003–1007 (2007).17934449 5. Yan KS The intestinal stem cell markers Bmi1 and Lgr5 identify two functionally distinct populations. Proceedings of the National Academy of Sciences of the United States of America. 109 (2 ), 466–471 (2012).22190486 6. Liao Z , Hu C , Gao Y Mechanisms modulating the activities of intestinal stem cells upon radiation or chemical agent exposure. Journal of Radiation Research. 63 (2 ), 149–157 (2022).35021216 7. Meyer AR , Brown ME , McGrath PS , Dempsey PJ Injury-Induced Cellular Plasticity Drives Intestinal Regeneration. Cellular and Molecular Gastroenterology and Hepatology. 13 (3 ), 843–856 (2022).34915204 8. Owens BM , Simmons A Intestinal stromal cells in mucosal immunity and homeostasis. Mucosal Immunology. 6 (2 ), 224–234 (2013).23235744 9. Barker N Adult intestinal stem cells: Critical drivers of epithelial homeostasis and regeneration. Nature Reviews Molecular Cell Biology. 15 (1 ), 19–33 (2014).24326621 10. Cheng H , Leblond CP Origin, differentiation and renewal of the four main epithelial cell types in the mouse small intestine. V. Unitarian Theory of the origin of the four epithelial cell types. The American Journal of Anatomy. 141 (4 ), 537–561 (1974).4440635 11. Sender R , Milo R The distribution of cellular turnover in the human body. Nature Medicine. 27 (1 ), 45–48 (2021). 12. Metcalfe C , Kljavin NM , Ybarra R , de Sauvage FJ Lgr5+ stem cells are indispensable for radiation-induced intestinal regeneration. Cell Stem Cell. 14 (2 ), 149–159 (2014).24332836 13. Tian H A reserve stem cell population in small intestine renders Lgr5-positive cells dispensable. Nature. 478 (7368 ), 255–259 (2011).21927002 14. Tirado FR Radiation-induced toxicity in rectal epithelial stem cell contributes to acute radiation injury in rectum. Stem Cell Research & Therapy. 12 (1 ), 63 (2021).33451351 15. Tetteh PW , Farin HF , Clevers H Plasticity within stem cell hierarchies in mammalian epithelia. Trends in Cell Biology. 25 (2 ), 100–108 (2015).25308311 16. Breault DT Generation of mTert-GFP mice as a model to identify and study tissue progenitor cells. Proceedings of the National Academy of Sciences of the United States of America. 105 (30 ), 10420–10425 (2008).18650388 17. Montgomery RK Mouse telomerase reverse transcriptase (mTert) expression marks slowly cycling intestinal stem cells. Proceedings of the National Academy of Sciences of the United States of America. 108 (1 ), 179–184 (2011).21173232 18. Orzechowska EJ , Katano T , Bialkowska AB , Yang VW Interplay among p21(Waf1/Cip1), MUSASHI-1 and Kruppel-like factor 4 in activation of Bmi1-Cre(ER) reserve intestinal stem cells after gamma radiation-induced injury. Scientific Reports. 10 (1 ), 18300 (2020).33110120 19. Takeda N Interconversion between intestinal stem cell populations in distinct niches. Science. 334 (6061 ), 1420–1424 (2011).22075725 20. Wong VW Lrig1 controls intestinal stem-cell homeostasis by negative regulation of ErbB signalling. Nature Cell Biology. 14 (4 ), 401–408 (2012).22388892 21. Powell AE The pan-ErbB negative regulator Lrig1 is an intestinal stem cell marker that functions as a tumor suppressor. Cell. 149 (1 ), 146–158 (2012).22464327 22. Ayyaz A Single-cell transcriptomes of the regenerating intestine reveal a revival stem cell. Nature. 569 (7754 ), 121–125 (2019).31019301 23. Tomic G Phospho-regulation of ATOH1 is required for plasticity of secretory progenitors and tissue regeneration. Cell Stem Cell. 23 (3 ), 436–443.e7 (2018).30100168 24. Castillo-Azofeifa D Atoh1(+) secretory progenitors possess renewal capacity independent of Lgr5(+) cells during colonic regeneration. The EMBO Journal. 38 (4 ), e99984 (2019).30635334 25. Van Landeghem L Activation of two distinct Sox9-EGFP-expressing intestinal stem cell populations during crypt regeneration after irradiation. American Journal of Physiology-Gastrointestinal and Liver Physiology. 302 (10 ), G1111–1132 (2012).22361729 26. Roche KC SOX9 maintains reserve stem cells and preserves radioresistance in mouse small intestine. Gastroenterology. 149 (6 ), 1553–1563.e10 (2015).26170137 27. Barriga FM Mex3a marks a slowly dividing subpopulation of Lgr5+ intestinal stem cells. Cell Stem Cell. 20 (6 ), 801–816.e7 (2017).28285904 28. May R Brief report: Dclk1 deletion in tuft cells results in impaired epithelial repair after radiation injury. Stem Cells. 32 (3 ), 822–827 (2014).24123696 29. Tetteh PW Replacement of lost Lgr5-positive stem cells through plasticity of their enterocyte-lineage daughters. Cell Stem Cell. 18 (2 ), 203–213 (2016).26831517 30. Bohin N Rapid crypt cell remodeling regenerates the intestinal stem cell niche after Notch inhibition. Stem Cell Reports. 15 (1 ), 156–170 (2020).32531190 31. Li N Single-cell analysis of proxy reporter allele-marked epithelial cells establishes intestinal stem cell hierarchy. Stem Cell Reports. 3 (5 ), 876–891 (2014).25418730 32. van Es JH Dll1+ secretory progenitor cells revert to stem cells upon crypt damage. Nature Cell Biology. 14 (10 ), 1099–1104 (2012).23000963 33. Durand A Functional intestinal stem cells after Paneth cell ablation induced by the loss of transcription factor Math1 (Atoh1). Proceedings of the National Academy of Sciences of the United States of America. 109 (23 ), 8965–8970 (2012).22586121 34. Hayakawa Y BHLHA15-positive secretory precursor cells can give rise to tumors in intestine and colon in mice. Gastroenterology. 156 (4 ), 1066–1081.e16 (2019).30448068 35. Yan KS Intestinal enteroendocrine lineage cells possess homeostatic and injury-inducible stem cell activity. Cell Stem Cell. 21 (1 ), 78–90.e6 (2017).28686870 36. Chandrakesan P Intestinal tuft cells regulate the ATM mediated DNA damage response via Dclk1 dependent mechanism for crypt restitution following radiation injury. Scientific Reports. 6 , 37667 (2016).27876863 37. Yu S Paneth cell multipotency induced by Notch activation following Injury. Cell Stem Cell. 23 (1 ), 46–59.e5 (2018).29887318 38. Moussa L Bowel radiation injury: Complexity of the pathophysiology and promises of cell and tissue engineering. Cell Transplantation. 25 (10 ), 1723–1746 (2016).27197023 39. Gong W Mesenchymal stem cells stimulate intestinal stem cells to repair radiation-induced intestinal injury. Cell Death & Disease. 7 (9 ), e2387 (2016).27685631 40. Tam SY , Wu VWC A review on the special radiotherapy techniques of colorectal cancer. Frontiers in Oncology. 9 , 208 (2019).31001474 41. Shadad AK , Sullivan FJ , Martin JD , Egan LJ Gastrointestinal radiation injury: Symptoms, risk factors and mechanisms. World Journal of Gastroenterology. 19 (2 ), 185–198 (2013).23345941 42. Serrano Martinez P , Giuranno L , Vooijs M , Coppes RP The radiation-induced regenerative response of adult tissue-specific stem cells: Models and signaling pathways. Cancers. 13 (4 ), 855 (2021).33670536 43. Stacey R , Green JT Radiation-induced small bowel disease: Latest developments and clinical guidance. Therapeutic Advances in Chronic Disease. 5 (1 ), 15–29 (2014).24381725 44. Pan YB , Maeda Y , Wilson A , Glynne-Jones R , Vaizey CJ Late gastrointestinal toxicity after radiotherapy for anal cancer: A systematic literature review. Acta Oncologica. 57 (11 ), 1427–1437 (2018).30264638 45. Elhammali A Late gastrointestinal tissue effects after hypofractionated radiation therapy of the pancreas. Radiation Oncology. 10 , 186 (2015).26337917 46. You SH , Cho MY , Sohn JH , Lee CG Pancreatic radiation effect in apoptosis-related rectal radiation toxicity. Journal of Radiation Research. 59 (5 ), 529–540 (2018).29901726 47. Jiminez JA , Uwiera TC , Douglas Inglis G , Uwiera RR Animal models to study acute and chronic intestinal inflammation in mammals. Gut Pathogens. 7 , 29 (2015).26561503 48. Snider AJ Murine model for colitis-associated cancer of the colon. Methods in Molecular Biology. 1438 , 245–254 (2016).27150094 49. Clapper ML , Cooper HS , Chang WC Dextran sulfate sodium-induced colitis-associated neoplasia: A promising model for the development of chemopreventive interventions. Acta Pharmacologica Sinica. 28 (9 ), 1450–1459 (2007).17723178 50. Gonzalez LM , Moeser AJ , Blikslager AT Animal models of ischemia-reperfusion-induced intestinal injury: Progress and promise for translational research. American Journal of Physiology-Gastrointestinal and Liver Physiology. 308 (2 ), G63–75 (2015).25414098 51. Fujimichi Y , Otsuka K , Tomita M , Iwasaki T Ionizing radiation alters organoid forming potential and replenishment rate in a dose/dose-rate dependent manner. Journal of Radiation Research. 63 (2 ), 166–173 (2022).34977948 52. Montenegro-Miranda PS A novel organoid model of damage and repair identifies HNF4alpha as a critical regulator of intestinal epithelial regeneration. Cellular and Molecular Gastroenterology and Hepatology. 10 (2 ), 209–223 (2020).32145468 53. Nagle PW , Coppes RP Current and future perspectives of the use of organoids in radiobiology. Cells. 9 (12 ), 2649 (2020). 54. Taelman J , Diaz M , Guiu J Human Intestinal Organoids: Promise and Challenge. Frontiers in Cell and Developmental Biology. 10 , 854740 (2022).35359445 55. Kim CK , Yang VW , Bialkowska AB The role of intestinal stem cells in epithelial regeneration following radiation-induced gut injury. Current Stem Cell Reports. 3 (4 ), 320–332 (2017).29497599 56. Kuruvilla JG Kruppel-like factor 4 modulates development of BMI1(+) intestinal stem cell-derived lineage following gamma-radiation-induced gut injury in mice. Stem Cell Reports. 6 (6 ), 815–824, (2016).27237377 57. Sangiorgi E , Capecchi MR Bmi1 is expressed in vivo in intestinal stem cells. Nature Genetics. 40 (7 ), 915–920 (2008).18536716 58. Srinivas S Cre reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus. BMC Developmental Biology. 1 , 4 (2001).11299042 59. Bialkowska AB , Ghaleb AM , Nandan MO , Yang VW Improved swiss-rolling technique for intestinal tissue preparation for immunohistochemical and immunofluorescent analyses. Journal of Visualized Experiments. (113 ), 54161 (2016). 60. Booth C , Tudor G , Tudor J , Katz BP , MacVittie TJ Acute gastrointestinal syndrome in high-dose irradiated mice. Health Physics. 103 (4 ), 383–399 (2012).23091876 61. Lu L , Jiang M , Zhu C , He J , Fan S Amelioration of whole abdominal irradiation-induced intestinal injury in mice with 3,3’-Diindolylmethane (DIM). Free Radical Biology & Medicine. 130 , 244–255 (2019).30352304 62. Karlsson JA , Andersen BL Radiation therapy and psychological distress in gynecologic oncology patients: Outcomes and recommendations for enhancing adjustment. Journal of Psychosomatic Obstetrics & Gynecology. 5 (4 ), 283–294 (1986).19844609 63. Yang J , Cai H , Xiao ZX , Wang H , Yang P Effect of radiotherapy on the survival of cervical cancer patients: An analysis based on SEER database. Medicine. 98 (30 ), e16421 (2019).31348242 64. Giroux V Mouse intestinal Krt15+ crypt cells are radio-resistant and tumor initiating. Stem Cell Reports. 10 (6 ), 1947–1958 (2018).29805107 65. Kim CK Kruppel-like factor 5 regulates stemness, lineage specification, and regeneration of intestinal epithelial stem cells. Cellular and Molecular Gastroenterology and Hepatology. 9 (4 ), 587–609 (2020).31778829 66. Sheng X Cycling stem cells are radioresistant and regenerate the intestine. Cell Reports. 32 (4 ), 107952 (2020).32726617 67. Gross S Nkx2.2 is expressed in a subset of enteroendocrine cells with expanded lineage potential. American Journal of Physiology-Gastrointestinal and Liver Physiology. 309 (12 ), G975–987 (2015).26492922 68. Sato T Characterization of radioresistant epithelial stem cell heterogeneity in the damaged mouse intestine. Scientific Reports. 10 (1 ), 8308 (2020).32444673 69. Roth S Paneth cells in intestinal homeostasis and tissue injury. PLoS One. 7 (6 ), e38965 (2012).22745693 70. Bohin N Insulin-like growth factor-1 and mTORC1 signaling promote the intestinal regenerative response after irradiation injury. Cellular and Molecular Gastroenterology and Hepatology. 10 (4 ), 797–810 (2020).32502530 71. Romesser PB Preclinical murine platform to evaluate therapeutic countermeasures against radiation-induced gastrointestinal syndrome. Proceedings of the National Academy of Sciences of the United States of America. 116 (41 ), 20672–20678 (2019).31551264 72. Gu J At what dose can total body and whole abdominal irradiation cause lethal intestinal injury among C57BL/6J mice? Dose Response. 18 (3 ), 1559325820956783 (2020). 73. Huh WJ Tamoxifen induces rapid, reversible atrophy, and metaplasia in mouse stomach. Gastroenterology. 142 (1 ), 21–24.e7 (2012).22001866 74. Keeley TM , Horita N , Samuelson LC Tamoxifen-induced gastric injury: Effects of dose and method of administration. Cellular and Molecular Gastroenterology and Hepatology. 8 (3 ), 365–367 (2019).31233898 75. Bohin N , Carlson EA , Samuelson LC Genome toxicity and impaired stem cell function after conditional activation of CreER(T2) in the intestine. Stem Cell Reports. 11 (6 ), 1337–1346 (2018).30449703 76. Boynton FDD , Ericsson AC , Uchihashi M , Dunbar ML , Wilkinson JE Doxycycline induces dysbiosis in female C57BL/6NCrl mice. BMC Research Notes. 10 (1 ), 644 (2017).29187243

PMC009xxxxxx/PMC9638987.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8301365 4429 J Am Coll Cardiol J Am Coll Cardiol Journal of the American College of Cardiology 0735-1097 1558-3597 35772913 9638987 10.1016/j.jacc.2022.04.042 NIHMS1842309 Article Cardiovascular Disease Risk Among Cancer Survivors: The Atherosclerosis Risk in Communities (ARIC) Study Florido Roberta MD, MHS ab Daya Natalie R. MPH c Ndumele Chiadi E. MD, PhD, MHS ab Koton Silvia PhD, MOccH, RN cd Russell Stuart D. MD e Prizment Anna PhD f Blumenthal Roger S. MD ab Matsushita Kunihiro MD, PhD ac Mok Yejin PhD, MPH c Felix Ashley S. PhD, MPH g Coresh Josef MD, PhD c Joshu Corinne E PhD ch Platz Elizabeth A. ScD, MPH ch Selvin Elizabeth PhD, MPH c a Division of Cardiology, Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, MD b Johns Hopkins Ciccarone Center for the Prevention of Cardiovascular Disease, Division of Cardiology, Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, MD c Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD d Stanley Steyer School of Health Professions, Sackler Faculty of Medicine, Tel-Aviv University, Tel Aviv, Israel. e Division of Cardiology, Department of Medicine, Duke University, Durham, NC f Division of Hematology, Oncology and Transplantation, Department of Medicine, University of Minnesota, Minneapolis, MN g Division of Epidemiology, The Ohio State University College of Public Health, Columbus, OH h Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD Address for Correspondence: Roberta Florido, MD, MHS, Johns Hopkins University, 600 N. Wolfe St. Carnegie 591A, Baltimore, MD 21287, Telephone: 410-955-3996, Fax: 410-614-9990, Twitter: @FloridoRoberta 22 10 2022 05 7 2022 07 11 2022 80 1 2232 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Background: Over 80% of adult patients diagnosed with cancer survive long-term. Long-term complications of cancer and its therapies may increase the risk of cardiovascular disease (CVD), but prospective studies utilizing adjudicated cancer and CVD events are lacking. Objectives: Assess the risk of CVD in cancer survivors in a prospective community-based study. Methods: We included 12,414 ARIC participants. Cancer diagnoses were ascertained via linkage with state registries supplemented with medical records. Incident CVD outcomes were coronary heart disease (CHD), heart failure (HF), stroke, and a composite of these endpoints. We used multivariable Poisson and Cox regression to estimate the association of cancer with incident CVD. Results: Mean age was 54, 55% were female, and 25% Black. 3,250 (25%) participants had incident cancer over a median 13.6 years of follow-up. Age-adjusted IR of CVD (per 1,000 person-years) were 27.0 (24.7, 29.1) for cancer survivors and 12.0 (11.5, 12.4) for non-cancer controls. After adjustment for cardiovascular risk factors, cancer survivors had significantly higher risks of CVD (HR 1.37, 95% CI 1.26, 1.50), HF (HR 1.52, 95% CI 1.38, 1.68), and stroke (HR 1.22, 95% CI 1.03, 1.44), but not CHD (HR 1.11, 95% CI 0.97, 1.28). Breast, lung, colorectal, and hematologic/lymphatic cancers, but not prostate cancer, were significantly associated with CVD risk. Conclusions: Compared to persons without cancer, adult cancer survivors have significantly higher risk of CVD, especially HF, independent of traditional cardiovascular risk factors. There is an unmet need to define strategies for CVD prevention in this high-risk population. Condensed abstract: There is a growing number of cancer survivors in the general population. We used data from the prospective ARIC cohort to estimate the excess burden of CVD and individual CVD subtypes in cancer survivors compared to persons without prior cancer, adjusting for baseline and time-varying shared risk factors. We found that cancer survivors had 37% greater risk of incident CVD. The excess CVD risk was independent of shared risk factors, varied by primary cancer, and was predominantly driven by HF, with less robust associations between cancer and stroke and CHD. Cancer survivors will benefit from enhanced CVD prevention strategies. Cancer cardiovascular disease cardio-oncology heart failure epidemiology prevention pmcIntroduction Cancer survivors are a rapidly growing population with specific health needs (1). It is estimated that 17 million adults living in the US are cancer survivors, representing 5% of the adult population, and this number is projected to increase (2). The growing number of cancer survivors results in part from an increase in cancer incidence due to aging of the population from advances in cancer screening, improvements in early detection, and therapeutics leading to significant improvements in cancer prognosis and long-term survival (3). However, patients diagnosed with cancer often have a high burden of chronic health conditions related to late effects of cancer and its treatments and are now living long enough such that non-cancer deaths are surpassing the risk of cancer-related deaths (4, 5). Cardiovascular disease (CVD) is highly prevalent and the leading cause of death among some cancer survivors (1). CVD and cancer share numerous risk factors and pathophysiological mechanisms that may predispose patients to both conditions (6). Additionally, several cancer treatments may cause cardiac toxicity contributing to a higher risk of CVD in cancer survivors (7). Despite increasing recognition of a close link between cancer and CVD, few prospective studies have rigorously assessed the excess risk of CVD in cancer survivors, particularly for cancers with onset in adulthood. Until recently, the published literature was limited to randomized clinical trials of highly selected patient populations with limited follow-up and lack of generalizability (8, 9). More recently, large observational studies have demonstrated an excess risk of CVD in cancer survivors compared to controls (5, 10–12). However, these studies have important limitations including retrospective designs, use of prescription claims and/or billing codes for classification of CVD, and variable quality of information on risk factors that may confound the associations of cancer and CVD (5, 10, 11, 13). Understanding the true excess burden of CVD and its subtypes in cancer survivors, as well as the degree to which CVD risk in this population is explained by shared risk factors, can inform clinical and public health strategies for CVD prevention in this unique patient population. We undertook a prospective cohort analysis of data from the community-based, Atherosclerosis Risk in Communities (ARIC) Study to estimate the associations of adult cancer survivorship with incident CVD and CVD subtypes (coronary heart disease (CHD), stroke, and heart failure (HF)). We evaluated whether the burden of CVD related to cancer was independent of traditional CVD risk factors and whether associations differed by type of primary cancer. Methods Study Population The ARIC Study is a prospective community-based cohort initiated in 1987 with the intent of studying the risk factors and natural history of CVD. A total of 15,792 participants from one of 4 US communities (Jackson, Mississippi; Washington County, Maryland; suburbs of Minneapolis, Minnesota; and Forsyth County, North Carolina) were enrolled at the initial study visit (1987–1989). Participants were aged 45–64 years at enrollment and predominantly Black or White adults. Participants were followed prospectively with continuous surveillance for incident CVD and serial study examinations that occurred every 3 years following baseline (1987–1989; 1990–1992; 1993–1995; and 1996–1998). ARIC visits 5–8 occurred in 2011–2013, 2016–2017, 2018–2019, and 2020. Additional study details have been previously published (14). All participants provided informed consent and the institutional review boards associated with all ARIC Study centers approved the study protocol. Within ARIC, 15,641 participants consented to cancer research and were linked to cancer registries. From these, we excluded 139 participants with missing data on cancer; 910 participants with self-reported prevalent cancer at baseline (visit 1, 1987–1989), as information about these cases was limited; 1,430 participants with prevalent CVD at baseline; 88 participants who did not self-identify as Black or White race, as well as Black participants from the Minneapolis suburbs and Washington county centers (due to small numbers); and 804 participants missing information on the covariates of interest, leaving a total of 12,421 participants included in the main analyses. Of those, 3,250 (25%) participants developed cancer during ARIC follow up. Ascertainment of Cancer Cases Data on cancer cases occurring between 1987 and December 31st, 2015 were obtained via linkage of the ARIC study with state cancer registries from Minnesota, North Carolina, Maryland, and Mississippi. These data were supplemented with information obtained directly from ARIC participants or their family members, hospital discharge summary codes, and review of medical records (15). Cancer survivors included in this analysis were those participants who had a diagnosis of a first primary invasive cancer (excluding those with non-melanoma skin cancer) during ARIC follow-up and who were free of CVD at the time of cancer diagnosis. An expert panel adjudicated all cases of bladder, breast, colorectal, liver, lung, pancreatic, and prostate cancer (15). For adjudicated cases, when possible, stage at diagnosis was determined from the cancer registry or medical records using the pathologic TNM stage (tumor extent, lymph node involvement, presence of metastasis). When this was not available, staging was determined from the cancer registry or clinical TNM stage from cancer registry or medical records according to Surveillance, Epidemiology, and End Results (SEER) summary stage. Ascertainment of CVD Events Incident CVD was the main outcome of interest and was defined as a composite of incident CHD, stroke, or HF from baseline through December 31st, 2015, as information on cancer cases is not yet available beyond this date. As secondary outcomes, we also considered incident CHD, stroke, and HF as individual endpoints. Participants were followed continuously for any possible CVD event via annual telephone calls, community-wide hospital surveillance, and linkage to state and national death indexes. Additional details of CVD surveillance in ARIC have been previously published (16, 17). An expert panel adjudicated all cases of CHD and stroke. Incident CHD was defined as a definite or probable non-fatal myocardial infarction, or definite fatal CHD. Stroke was defined as definite or probable ischemic or hemorrhagic stroke. HF was defined as the first hospitalization or death related to HF, using ICD-9 code 428 or ICD-10 code I-50 in the main analyses. We performed sensitivity analyses considering HF events based on ICD codes between baseline and December, 2004 and HF events adjudicated by an expert panel from January 2005 (start of ARIC HF adjudication) onwards (18). Covariates of Interest Information about participants’ demographics and alcohol drinking status (former, current, never) was obtained via questionnaire at visit 1. Information on all other covariates of interest was obtained at visit 1 and at each subsequent ARIC visit and included as time-varying covariates in regression models. Smoking status was categorized as former, current, or never, and additional information was obtained on number of cigarette-pack years consumed. All medications used in the prior 2 weeks were brought in by participants and recorded at each study visit. Body mass index (BMI) was calculated based on measured height and weight in kg/m2. High-density lipoprotein (HDL) cholesterol and triglycerides were measured in serum using enzymatic assays, and low-density lipoprotein (LDL) cholesterol was calculated using the Friedewald equation (19). Hypertension was defined as a measured blood pressure ≥140/90 or use of antihypertensive medication. Diabetes mellitus was defined as a fasting blood glucose ≥ 126 mg/dl, non-fasting blood glucose ≥ 200 mg/dl, history of physician diagnosis, or current use of hypoglycemic medication. Estimated glomerular filtration rate (eGFR) was calculated based on measured creatinine using the CKD-EPI equation (20). Statistical Analyses We compared the Visit 1 (pre-cancer) characteristics of those individuals who did and did not develop incident cancer during follow-up, using the t-test for continuous variables and the chi square test for categorical variables. We used survival analysis methods to estimate the association of the cancer with subsequent risk of CVD. In our main analyses, incident cancer was modeled as a time varying exposure. Participants without prior cancer contributed person time to the non-cancer group from Visit 1 until the development of CVD, incident cancer, censoring or death. Those with incident cancer contributed person time to the cancer group starting at the date of cancer diagnosis until development of CVD, censoring or death. We used Poisson regression to calculate adjusted incidence rates of CVD and CVD subtypes for cancer survivors and for persons who did not develop cancer during follow up. We used Cox regression models to assess the association of cancer with incident CVD, CHD, stroke, and HF, with two levels of adjustment. Model 1 included demographics: age, sex, race-center and educational level. We used a second adjustment model with robust adjustment for covariates thought to be potential confounders of the associations of cancer with CVD. Model 2 included baseline age, sex, race-center, educational level, drinking status and non-steroidal anti-inflammatory drug use (including aspirin), as well as baseline and time-varying smoking status, smoking pack-years, BMI, LDL-cholesterol, HDL-cholesterol, triglycerides, use of lipid lowering medications, prevalent hypertension, prevalent diabetes mellitus, and eGFR. By including time-varying covariates in this model, we ensured CVD risk factors assessed at multiple ARIC visits were included in the adjustment model. These analyses were performed overall and stratified by race and sex, with tests for interaction with cancer survivorship status. For analyses stratified by sex, we created a category of non-sex-related cancers including all cancers that could have occurred in both men and women, and excluding breast, cervical, endometrial, ovarian, and prostate cancers. We performed similar exploratory analyses considering the associations of the most common cancers with incident CVD and CVD subtypes. These included post-menopausal female breast cancer, prostate, lung, colorectal, and hematopoietic and lymphatic cancers. We conducted multiple sensitivity analyses to confirm our findings. First, we repeated the main analyses excluding participants who died within one year of cancer diagnosis since these deaths most likely occurred as a result of the cancer itself or its treatments. Second, we created a nested matched cohort sample, where each participant who developed cancer was matched on sex, race, and age at the time of cancer diagnosis (±5 years) to two participants without a history of cancer. Participants were followed from time of matching until development of CVD, censoring, or death. If a control developed cancer after matching, follow up in the non-cancer group would be censored and from that point on they would become a cancer case. In these matched analyses, we used adjusted Cox regression to assess the association of cancer survivorship with incident CVD, CHD, stroke, or HF. This study design also allowed us to construct analyses stratified by date of matching (prior to 1995, 1995–2000, 2000–2005, 2005 onwards). All covariates were selected a priori based on previously published literature. Analyses were performed using Stata SE 15. Results At visit 1, the mean age of the study population was 54 years, 55% were female, and 25% were Black adults. Participants who developed cancer during ARIC follow-up were older, more likely to be male, and less likely to be Black compared to participants who did not develop cancer (Table 1). They were also more likely to be current or former smokers and to have higher pack-years of smoking and lower HDL-cholesterol, but less likely to have diabetes, compared to persons who did not develop cancer. There were no significant differences in BMI, LDL-cholesterol, triglycerides, use of cholesterol lowering medication, prevalence of hypertension, or NSAID use between the two groups. A total of 3,250 (25%) participants free of CVD were diagnosed with a first primary cancer after visit 1, with a median time to cancer diagnosis of 13.6 years. Post-menopausal breast cancer was the most common cancer among women (35%), whereas prostate cancer was the most common cancer among men (40%). Lung, colorectal, and hematopoietic and lymphatic were the most common primary non-sex-related cancers (Table 2). The median follow-up time to CVD was 14 years (from visit 1 date) among those who never developed cancer and 5.2 years (from date of cancer diagnosis) among those who developed cancer. During the follow-up period there were 3,723 incident CVD events; 1,824 CHD, 1,162 strokes, and 2,665 HF events. Median times from cancer diagnosis to any CVD event were: 6.2 years for breast, 6.3 years for prostate, 1.3 years for lung, 5.1 years for colorectal, and 3.1 years for hematopoietic and lymphatic cancers. Age-adjusted incidence rates (IR) of CVD per 1,000 person-years were 12.0 (95% CI 11.5, 12.4) for participants who did not develop cancer and 23.1 (95% CI 21.4, 25.1) person-years for all cancer survivors. After robust adjustment for potential confounders (Model 2), IR remained higher among those who developed cancer (17.4 (95% CI 14.8, 20.5) person-years) than among those without cancer (11.0 (95% CI 9.6, 12.7) person-years), resulting in an IR difference of 6.4 (95% CI 5.2, 7.8) person-years. In analyses adjusted for demographics (Model 1), cancer survivors had 42% higher risk of developing incident CVD compared to those who did not develop cancer (HR 1.42, 95% CI 1.30, 1.56; Supplemental Table 1). Results were minimally attenuated and remained significant after robust adjustment for shared risk factors between cancer and CVD (Model 2: HR 1.37, 95% CI 1.26, 1.50; Figure 1 and Supplemental Table 1). When considering specific subtypes of CVD, overall cancer survivorship was significantly associated with incident HF (HR 1.52, 95% CI 1.38, 1.68) and stroke (HR 1.22, 95% CI 1.03, 1.44), but not with CHD (HR 1.11, 95% CI 0.97, 1.28) (Model 2; Table 3). Results were unchanged in sensitivity analyses considering adjudicated cases of HF (HR for the association of cancer survivorship with incident HF 1.58, 95% CI 1.43, 1.75). We did not find significant differences in the association of cancer and incident CVD by race (p-for-interaction = 0.76) (Table 4). There was a stronger association of survivorship from non-sex-related cancers (i.e., excluding breast, cervical, endometrial, ovarian, and prostate cancers) with incident CVD among women (HR 1.96, 95% CI 1.66, 2.31, Model 2) versus men (HR 1.57, 95% CI 1.35, 1.83, Model 2; p-for-interaction <0.01; Table 4). Age-adjusted incidence rates of CVD per 1,000 person years for survivors of specific cancers were: 16.6 (95% CI 13.7, 20.0) person-years for breast, 21.0 (95% CI 18.0, 24.6) person-years for prostate, 50.0 (95% CI 39.0, 64.2) person-years for lung, 25.4 (95% CI 20.1, 32.0) person-years for colorectal, and 41.0 (95% CI 32.1, 52.4) person-years for survivors of hematopoietic and lymphatic cancers. In fully adjusted analyses, we found that survivorship from breast, lung, colorectal, and hematopoietic and lymphatic cancers were each independently associated with incident CVD (Figure 1 and Supplemental Table 1). There was no significant association between prostate cancer and incident CVD. We observed some differences in the associations of specific cancers with CVD subtype. In the fully adjusted analyses, increased HF risk was seen among survivors of breast (HR 1.58, 95% CI 1.28, 1.95), lung (HR 2.73, 95% CI 2.10, 3.55), colorectal (HR 1.32, 95% CI 1.00, 1.75), and hematopoietic and lymphatic (HR 3.22, 95% CI 2.49, 4.15), but not of prostate cancer (HR 1.08, 95% CI 0.89, 1.31), compared to persons without prior cancer (Table 3). Survivors of lung cancer had a more than two-fold increased risk of stroke than persons without cancer (HR 2.40, 95% CI 1.53–3.78; Table 3), but other cancers were not significantly associated with incident stroke. Only survivors of hematopoietic and lymphatic cancers were at significantly higher risk for CHD compared to persons without cancer (HR 1.76, 95%CI 1.15–2.69; Table 3). Our results were not appreciably different in sensitivity analyses excluding participants who died within 1 year of cancer diagnosis, with a HR for incident CVD comparing cancer survivors to persons without cancer of 1.21 (95% CI 1.11, 1.33). We also found similar results for the associations of cancer survivorship with incident CVD, CHD, stroke, and HF when using a nested study population with matching of cancer survivors and individuals without cancer (Supplemental Table 2). However, in these analyses, we did not observe significant associations between breast or colorectal cancer survivorship and incident CVD (Supplemental Table 3). In analyses stratified by date of cancer diagnosis at the time of matching, we found that the association of cancer with incident CVD was similar for different time intervals of cancer diagnosis except for those with remote cancer diagnosis, prior to 1995, who had null associations (Table 5). Discussion In this prospective analysis of the community-based ARIC Study, we found strong independent associations of adult cancer survivorship with incident CVD and, in particular, with HF. After accounting for shared risk factors between cancer and CVD, cancer diagnosis was associated with an excess of 6.4 CVD cases per 1,000 person-years. Compared to persons without prior cancer, cancer survivors had a 37% higher risk of incident CVD and 52% higher risk of HF, and less strong associations with stroke or CHD. The risk of incident CVD and CVD subtypes varied by primary cancer type, with significant risk associations with for breast, lung, colorectal, and hematopoietic and lymphatic cancers, but no significant associations for prostate cancer. Importantly, the associations of cancer survivorship with CVD were largely unchanged between minimally adjusted analyses accounting for demographics and those robustly adjusted for traditional CVD risk factors, suggesting additional cancer-specific mechanisms likely contribute to the excess burden and risk of CVD in this population. Few population-based studies have examined the risk of CVD in cancer survivors compared to controls, with important limitations. In a prior retrospective matched cohort study, Schoormans et al. found that only survivors of prostate and lung and trachea cancers had an increased risk of CVD compared to non-cancer controls (11). Despite inclusion of a large sample size, the study was limited by the use of drug dispensing information for the definition of CVD and confounding comorbidities. Two other large retrospective cohort studies have demonstrated variable associations between survivorship from different cancers and incident CVD (10, 13). Inferences from these studies are limited by retrospective design, variable information on shared risk factors between cancer and CVD, lack of CVD or cancer event adjudication, and variable follow up times. Additionally, most studies to date have had limited information on smoking, an important confounder of the associations of cancer with CVD. Our study adds to the literature by demonstrating strong associations between adult cancer survivorship and incident CVD in a diverse, prospective cohort, with continuously adjudicated cancer and CVD events, as well as detailed information on shared risk factors at multiple time points, for the first time. Our results are in line with a growing body of literature indicating that cancer survivors are a population at increased risk for CVD who may benefit from enhanced preventive measures. While the mechanisms underlying the associations of cancer CVD are uncertain, several groups have speculated that shared cancer and CVD risk factors may be important contributors (6). Prior studies have found variable differences in the burden of CVD risk factors between cancer survivors and non-cancer patients. Using data from electronic health records, Armenian et al. found that cancer survivors were more likely to have hypertension, diabetes, dyslipidemia, excess weight, and a history of smoking than persons without cancer (10). Conversely, another large study using multiple linked electronic health records from the United Kingdom found only smoking, hypertension, and CKD were marginally more prevalent in cancer survivors (13). In the present study, aside from smoking, we did not find that persons who developed cancer had a significantly higher burden of pre-existing CVD risk factors. Differences in our findings may be explained by the timing of assessment of comorbidities which was done prior to as opposed to following cancer diagnosis as it is well established that the burden of CVD risk factors may increase following cancer diagnosis and treatment (21, 22). More importantly, our study benefited from direct assessment of these comorbidities rather than relying on data from electronic health records or drug dispensing information. In our study, the associations of cancer with incident CVD were largely independent of traditional CVD risk factors. While CVD risk factors may mediate some of the observed associations observed, our findings suggest alternative mechanisms may contribute to the link between cancer and CVD (Central Illustration). Potential pathways involved in the cancer and CVD association include shared disease mechanisms such as systemic inflammation and oxidative stress (23, 24), a pro-inflammatory and prothrombotic state promoted by cancer itself (25), as well as cancer therapies. Variation in CVD risk across primary cancers suggest that the malignancy itself or cardiotoxicity from specific cancer treatments are likely central to CVD risk in this population (26). This is further supported by our findings of variable associations of cancer with CVD subtypes, with stronger associations with incident HF. For example, breast and hematopoietic and lymphatic cancers are typically managed with a combination of chemotherapy, often anthracycline-based, as well as chest radiation, both with well-established cardiotoxic potential (27). Similarly, chest radiation may be at least partly responsible for the increased risk of CHD in patients with hematopoietic and lymphatic cancers, that was not observed among other cancer survivors (28). Conversely, prostate cancer may be managed with active surveillance or local therapies without the risk of cardiotoxicity, which may explain why we did not observe an excess CVD risk in this subgroup (29). Prior epidemiologic studies evaluating the contributions of cancer therapies to CVD risk have been limited by factors such as retrospective design and limitations in assessments of cancer therapies and confounders. Additional studies are needed to elucidate the contribution of cancer therapies to the development of CVD in cancer survivors. Our findings have important clinical and public health implications. CVD screening and prevention practices among cancer survivors are highly variable and often neglected due to limited evidence guiding its practice as well as misconceptions regarding competing risks of cancer mortality (30). In the present study, close to half of cancer survivors developed CVD following cancer diagnosis, indicating that this population would likely benefit from aggressive screening and preventive interventions. However, we also demonstrate that the links between cancer and CVD go above and beyond traditional risk factors. Therefore, while attention to shared risk factors between cancer and CVD is needed, our data suggests traditional risk assessment tools are likely to underestimate the risks in this population and risk factor modification alone is likely insufficient to fully address CVD risk in this population. Furthermore, it is important to consider the variable associations of specific cancer subtypes with CVD and CVD subtypes, with some subsets of adult cancer survivors having particularly high risk. Further studies are needed to inform screening and preventive strategies specific to this unique patient population. Our study has important limitations to consider. First, its observational nature means that we cannot eliminate the possibility of residual confounding. Second, while CHD and stroke outcomes were adjudicated by an expert panel, HF events were defined based on ICD codes with the possibility of misclassification. However, a prior study has showed high specificity of this definition (18). Furthermore, our sensitivity analyses considering adjudicated cases from 2005 onwards demonstrated similar results. Third, even with over 12,000 middle-aged adults at baseline, our power to detect small to moderate associations, especially for specific cancer subtypes and demographic groups, was likely limited. Lack of power may also explain the lack of association between breast and colorectal cancer and CVD in our matched analyses. While we had limited information on cancer staging, which may influence cancer treatments, we did not have sufficient power for stratified analyses. Similarly, we did not have information on cancer treatments, which could be directly contributing to the observed variability in CVD risk across cancer subtypes. Lastly, we cannot exclude the possibility of increased medical care and CVD surveillance among cancer survivors increasing the likelihood of CVD detection. Despite such limitations, our study is unique as a large, diverse, community-based cohort study of men and women, with long-term follow-up. Our study also benefited from comprehensive cardiovascular surveillance, adjudication of both cancer cases and cardiovascular events, the ability to distinguish CVD subtypes, and rigorous measurement of cardiovascular risk factors at multiple time points, allowing us to conduct time-varying adjustment. Furthermore, we were able to conduct time to event analyses due to validation of cancer diagnosis with exact timing. In summary, we found cancer survivors have an increased risk of CVD, particularly HF, compared to persons without cancer, and that this excess risk is not explained by traditional CVD risk factors. Elucidating the mechanisms underlying the excess risk of CVD among adult cancer survivors, from treatment toxicities to shared biological pathways, is needed in order to define novel strategies for predicting and preventing CVD in this population. Clinical perspectives: Clinical competencies: Overall cancer survivors have an increased risk of CVD compared to persons without prior cancer, independently of shared cancer and CVD risk factors. This excess risk is predominantly driven by HF, with less robust associations between cancer and stroke, and in particular, CHD. Some cancer survivors are at greater CVD risk than others. Cancer survivors should be considered a high-risk population and may benefit from enhanced CVD preventative interventions. While CVD risk factor modification should be emphasized, it may not be sufficient to address this excess risk. Translational outlook: Cancer survivors have an increased risk of CVD, particularly HF, compared to persons without cancer, independently of shared risk factors. Future studies are needed to elucidate the mechanisms underlying the excess risk of CVD among adult cancer survivors, from treatment toxicities to shared biological pathways. Supplementary Material Supplemental Tables Acknowledgements: The authors thank the staff and participants of the Atherosclerosis Risk in Communities Study for their important contributions. Cancer incidence data have been provided by the Maryland Cancer Registry, Center for Cancer Surveillance and Control, Maryland Department of Health. We acknowledge the State of Maryland, the Maryland Cigarette Restitution Fund, and the National Program of Cancer Registries of the Centers for Disease Control and Prevention for the funds that helped support the availability of the cancer registry data. Sources of Funding: The ARIC study has been funded in whole or in part with federal funds from the National Heart, Lung, and Blood Institute (NHLBI); the National Institutes of Health (NIH); and the Department of Health and Human Services, under contract numbers 75N92022D00001, 75N92022D00002, 75N92022D00003, 75N92022D00004, and 75N92022D00005. Studies on cancer in the Atherosclerosis Risk in Communities Study are also supported by the National Cancer Institute (grant U01CA164975). This research was additionally supported by a Cancer Center Support Grant from the National Cancer Institute (grant P30 CA006973). Dr Selvin was supported by NIH/National Institute of Diabetes and Digestive and Kidney Diseases grants K24 HL152440 and R01 DK089174. Dr. Ndumele was supported by NIH/NHLBI grant R01 HL146907 and by AHA grant 20SFRN35120152. Abbreviation list: ARIC Atherosclerosis Risk in Communities BMI Body mass index CHD coronary heart disease CVD cardiovascular disease eGFR estimated glomerular filtration rate HDL high-density lipoprotein HF heart failure LDL Low-density lipoprotein Figure 1. Association of Cancer Survivorship with Cardiovascular Disease, by Cancer Type Caption: Forest plot demonstrating the association of cancer and cancer subtypes with incident CVD, after adjustment for demographic variables and shared risk factors between cancer and CVD (Model 2). a) Model 2: Adjusted for baseline age, sex, race-center, educational level, drinking status, and time-varying smoking status and cigarette pack-years, NSAID use (including aspirin), and time-varying, body mass index, LDL-cholesterol, HDL-cholesterol, triglycerides, use of lipid lowering medications, hypertension, diabetes mellitus and eGFR. b) N = total number of participants included in the analyses, including cancer survivors and persons without cancer. Central Illustration. Cardiovascular Disease Risk in Cancer Survivors and Proposed Pathways Linking Cancer and Cardiovascular Disease. Cancer related pathways underlying the associations of cancer with CVD include but are not limited to common biologic predisposition; inflammation, oxidative stress, a pro-thrombotic state promoted by the cancer itself; and cardiotoxic effects of various cancer therapies. Non-cancer related pathways include lifestyle factors such as diet and physical activity, and other shared risk factors between cancer and CVD such as smoking, obesity and diabetes. Table1. Baseline (Visit 1) Characteristics of the Study Population by Incident Cancer During ARIC Follow-upa Variable No Cancer (n=9,171) Cancer (n=3,250) P value Age (yrs) 53.6 (5.7) 54.5 (5.7) <0.001 Sex (%) Female (%) 58.3 47.1 <0.001 Male (%) 41.7 52.9 Race (%) 0.005   Black 26.1 23.4   White 73.9 76.6 Education Level (%) 0.66   Less than high school 22.2 21.6   High school or vocational school 41.1 41.9   Some college education of higher 36.7 36.4 Cigarette smoking status (%) <0.001   Current 24.0 28.8   Former 30.7 32.2   Never 45.3 39.0 Pack-years of smoking 276 (400) 354 (443) <0.001 Drinking status (%) <0.001   Current 56.0 60.5   Former 18.1 17.2   Never 26.0 22.3 Body mass index (kg/m2) 27.5 (5.3) 27.4 (5.0) 0.39 Lipids (mg/dl)   LDL-cholesterol 137.3 (39.7) 136.6 (36.9) 0.42   HDL-cholesterol 52.7 (17.0) 51.7 (16.8) 0.003   Triglycerides 122.2 (63.7 123.9 (62.4) 0.20 Cholesterol lowering medication (%) 2.2 2.5 0.48 Hypertension (%) 31.4 30.1 0.18 Diabetes mellitus (%) 10.6 7.8 <0.001 eGFR, mL/min/1.73 m2 103.2 (15.4) 102.2 (14.3) 0.002 NSAID use (%) 54.8 53.8 0.34 a) Values are means (SD) or percentages. Table 2. Distribution of First Primary Cancer in Women and Men Cancer Women (n=1,532) Men (n=1,718) Breast 529 (34.5) -   Local 193 (12.6) -   Regional 130 (8.5) -   Distant 16 (1.0) -   Stage unknown 190 (12.4) - Cervical and endometrial 121(7.9) - Ovarian 52 (3.4) - Prostate - 681 (39.6)   Pathological or clinical TNM 1 - 34 (2.0)   Pathological or clinical TNM in (2A, 2B, 2C0) - 414 (24.1)   Pathological or clinical TNM 3 - 59 (3.4)   Pathological TNM 4 - 31 (1.8) Stage unknown - 143 (8.3) Lung 184 (12.0) 236 (13.7)   Local 33 (2.2) 39 (2.3)   Regional 43 (2.8) 58 (3.4)   Distant 54 (3.5) 67 (3.9)   Stage unknown 54 (3.5) 72 (4.2) Colorectal 156 (10.2) 158 (9.2)   Local 43 (2.8) 42 (2.4)   Regional 44 (2.9) 47 (2.7)   Distant 21 (1.4) 15 (0.9)   Stage unknown 48 (3.1) 54 (3.1) Hematopoietic and lymphatic 123 (8.0) 133 (7.7) Renal 48 (3.1) 58 (3.4) Bladder 38 (2.5) 116 (6.8) Melanoma 38 (2.5) 58 (3.4) Pancreatic 35 (2.3) 49 (2.9) Stomach 18 (1.2) 26 (1.5) Other digestive 28 (1.8) 41 (2.4) Central nervous system 38 (2.5) 30 (1.8) Thyroid 16 (1.0) 7 (0.4) Other 91 (5.9) 268 (15.6) Table 3. Associations (HR, 95% CI) of Cancer with Cardiovascular Disease Subtypes Coronary Heart Disease Stroke Heart Failure Any cancer Model 1 1.15 (1.01, 1.32) 1.24 (1.05, 1.46) 1.59 (1.44, 1.75) Model 2 1.11 (0.97, 1.28) 1.22 (1.03, 1.44) 1.52 (1.38, 1.68) Breast cancer Model 1 1.23 (0.90, 1.68) 0.82 (0.53, 1.25) 1.64 (1.33, 2.03) Model 2 1.21 (0.88, 1.65) 0.80 (0.52, 1.23) 1.58 (1.28, 1.95) Prostate cancer Model 1 0.96 (0.76, 1.22) 1.10 (0.82, 1.48) 1.05 (0.87, 1.28) Model 2 0.99 (0.78, 1.26) 1.12 (0.84, 1.51) 1.08 (0.89, 1.31) Lung cancer Model 1 2.08 (1.35, 3.20) 3.11 (1.99, 4.86) 3.78 (2.92, 4.90) Model 2 1.41 (0.91, 2.18) 2.40 (1.53, 3.78) 2.73 (2.10, 3.55) Colorectal cancer Model 1 1.17 (0.82, 1.69) 1.29 (0.83, 1.99) 1.32 (1.00, 1.75) Model 2 1.19 (0.83, 1.72) 1.28 (0.83, 1.98) 1.32 (1.00, 1.75) Hematopoietic and lymphatic cancer Model 1 1.74 (1.14, 2.65) 1.58 (0.93, 2.68) 3.05 (2.37, 3.93) Model 2 1.76 (1.15, 2.69) 1.60 (0.94, 2.71) 3.22 (2.51, 4.18) a) Model 1: Adjusted for baseline age, sex, race-center and educational level. b) Model 2: Adjusted for baseline age, sex, race-center, educational level, drinking status and NSAID use (including aspirin), and baseline and time-varying smoking status, cigarette pack-years, body mass index, LDL-cholesterol, HDL-cholesterol, triglycerides, use of lipid lowering medications, hypertension, diabetes mellitus and eGFR Note: Analyses of breast cancer also adjust for hormone use in Model 2. Table 4. Associations (HR, 95% CI) of Cancer with Cardiovascular Disease by Race and Sex Cancer Black White p Interaction Any cancer Model 1 1.44 (1.22, 1.71) 1.43 (1.29, 1.59) 0.26 Model 2 1.40 (1.18, 1.66) 1.36 (1.23, 1.51) 0.76 Cancer Women Men p Interaction Non-sex-related cancer Model 1 2.12 (1.80, 2.51) 1.69 (1.45, 1.97) <0.01 Model 2 1.96 (1.66, 2.31) 1.57 (1.35, 1.83) <0.01 a) Model 1: Adjusted for baseline age, sex, race-center and educational level. b) Model 2: Adjusted for baseline age, sex, race-center, educational level, drinking status and NSAID use (including aspirin), and baseline and time-varying smoking status, cigarette pack-years, body mass index, LDL-cholesterol, HDL-cholesterol, triglycerides, use of lipid lowering medications, hypertension, diabetes mellitus and eGFR Table 5. Associations of Cancer with Cardiovascular Disease in Nested Case-Control Sample, by Date of Cancer Diagnosis. Date of Matching HR (95% CI) Model 1 Model 2 <1995 (n=1,525) 1.13 (0.92, 1.39) 1.09 (0.89, 1.35) 1995-<2000 (n=1,565) 1.40 (1.14, 1.72) 1.33 (1.08, 1.63) 2000-<2005 (n=1,869) 1.39 (1.13, 1.69) 1.28 (1.04, 1.56) ≥2005 (n=2,726) 1.53 (1.23, 1.92) 1.44 (1.14, 1.80) a) Model 1: Adjusted for baseline educational level. b) Model 2: Adjusted for baseline educational level, drinking status and NSAID use (including aspirin), and baseline and time-varying smoking status, cigarette pack-years, body mass index, LDL-cholesterol, HDL-cholesterol, triglycerides, use of lipid lowering medications, hypertension, diabetes mellitus and eGFR Relationship with Industry: None. Tweet: Cancer survivors have a higher risk of CVD, particularly #heartfailure, compared to persons without cancer. Risk varies by cancer subtype and is not fully explained by shared risk factors. Cancer survivors may need more aggressive CVD prevention. #cardioonc #cardiooncology References: 1. Miller KD , Nogueira L , Mariotto AB , Rowland JH , Yabroff KR , Alfano CM , Cancer treatment and survivorship statistics, 2019. CA Cancer J Clin. 2019;69 (5 ):363–85. Epub 2019/06/12. doi: 10.3322/caac.21565.31184787 2. Bluethmann SM , Mariotto AB , Rowland JH . Anticipating the “Silver Tsunami”: Prevalence Trajectories and Comorbidity Burden among Older Cancer Survivors in the United States. Cancer Epidemiol Biomarkers Prev. 2016;25 (7 ):1029–36. Epub 2016/07/03. doi: 10.1158/1055-9965.EPI-16-0133.27371756 3. Byers T , Wender RC , Jemal A , Baskies AM , Ward EE , Brawley OW . The American Cancer Society challenge goal to reduce US cancer mortality by 50% between 1990 and 2015: Results and reflections. CA Cancer J Clin. 2016;66 (5 ):359–69. Epub 2016/05/14. doi: 10.3322/caac.21348.27175568 4. Zaorsky NG , Churilla TM , Egleston BL , Fisher SG , Ridge JA , Horwitz EM , Causes of death among cancer patients. Ann Oncol. 2017;28 (2 ):400–7. Epub 2016/11/11. doi: 10.1093/annonc/mdw604.27831506 5. Sturgeon KM , Deng L , Bluethmann SM , Zhou S , Trifiletti DM , Jiang C , A population-based study of cardiovascular disease mortality risk in US cancer patients. Eur Heart J. 2019;40 (48 ):3889–97. Epub 2019/11/26. doi: 10.1093/eurheartj/ehz766.31761945 6. Koene RJ , Prizment AE , Blaes A , Konety SH . Shared Risk Factors in Cardiovascular Disease and Cancer. Circulation. 2016;133 (11 ):1104–14. Epub 2016/03/16. doi: 10.1161/CIRCULATIONAHA.115.020406.26976915 7. Moslehi JJ . Cardiovascular Toxic Effects of Targeted Cancer Therapies. N Engl J Med. 2016;375 (15 ):1457–67. Epub 2016/10/13. doi: 10.1056/NEJMra1100265.27732808 8. Swain SM , Whaley FS , Ewer MS . Congestive heart failure in patients treated with doxorubicin: a retrospective analysis of three trials. Cancer. 2003;97 (11 ):2869–79. Epub 2003/05/27. doi: 10.1002/cncr.11407.12767102 9. Seidman A , Hudis C , Pierri MK , Shak S , Paton V , Ashby M , Cardiac dysfunction in the trastuzumab clinical trials experience. J Clin Oncol. 2002;20 (5 ):1215–21. Epub 2002/03/01. doi: 10.1200/JCO.2002.20.5.1215.11870163 10. Armenian SH , Xu L , Ky B , Sun C , Farol LT , Pal SK , Cardiovascular Disease Among Survivors of Adult-Onset Cancer: A Community-Based Retrospective Cohort Study. J Clin Oncol. 2016;34 (10 ):1122–30. Epub 2016/02/03. doi: 10.1200/JCO.2015.64.0409.26834065 11. Schoormans D , Vissers PAJ , van Herk-Sukel MPP , Denollet J , Pedersen SS , Dalton SO , Incidence of cardiovascular disease up to 13 year after cancer diagnosis: A matched cohort study among 32 757 cancer survivors. Cancer Med. 2018;7 (10 ):4952–63. Epub 2018/09/17. doi: 10.1002/cam4.1754.30220107 12. Kenzik KM , Balentine C , Richman J , Kilgore M , Bhatia S , Williams GR . New-Onset Cardiovascular Morbidity in Older Adults With Stage I to III Colorectal Cancer. J Clin Oncol. 2018;36 (6 ):609–16. Epub 2018/01/18. doi: 10.1200/JCO.2017.74.9739.29337636 13. Strongman H , Gadd S , Matthews A , Mansfield KE , Stanway S , Lyon AR , Medium and long-term risks of specific cardiovascular diseases in survivors of 20 adult cancers: a population-based cohort study using multiple linked UK electronic health records databases. Lancet. 2019;394 (10203 ):1041–54. Epub 2019/08/25. doi: 10.1016/S0140-6736(19)31674-5.31443926 14. The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. The ARIC investigators. Am J Epidemiol. 1989;129 (4 ):687–702.2646917 15. Joshu CE , Barber JR , Coresh J , Couper DJ , Mosley TH , Vitolins MZ , Enhancing the Infrastructure of the Atherosclerosis Risk in Communities (ARIC) Study for Cancer Epidemiology Research: ARIC Cancer. Cancer Epidemiol Biomarkers Prev. 2018;27 (3 ):295–305. Epub 2017/12/22. doi: 10.1158/1055-9965.EPI-17-0696.29263187 16. White AD , Folsom AR , Chambless LE , Sharret AR , Yang K , Conwill D , Community surveillance of coronary heart disease in the Atherosclerosis Risk in Communities (ARIC) Study: methods and initial two years’ experience. J Clin Epidemiol. 1996;49 (2 ):223–33. Epub 1996/02/01. doi: 10.1016/0895-4356(95)00041-0.8606324 17. Rosamond WD , Folsom AR , Chambless LE , Wang CH , McGovern PG , Howard G , Stroke incidence and survival among middle-aged adults: 9-year follow-up of the Atherosclerosis Risk in Communities (ARIC) cohort. Stroke. 1999;30 (4 ):736–43. Epub 1999/04/03. doi: 10.1161/01.str.30.4.736.10187871 18. Rosamond WD , Chang PP , Baggett C , Johnson A , Bertoni AG , Shahar E , Classification of heart failure in the atherosclerosis risk in communities (ARIC) study: a comparison of diagnostic criteria. Circ Heart Fail. 2012;5 (2 ):152–9. Epub 2012/01/25. doi: 10.1161/CIRCHEARTFAILURE.111.963199.22271752 19. Friedewald WT , Levy RI , Fredrickson DS . Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18 (6 ):499–502.4337382 20. Liu X , Wang Y , Wang C , Shi C , Cheng C , Chen J , A new equation to estimate glomerular filtration rate in Chinese elderly population. PLoS One. 2013;8 (11 ):e79675. Epub 2013/11/19. doi: 10.1371/journal.pone.0079675.24244543 21. van den Berg MM , Winkels RM , de Kruif JT , van Laarhoven HW , Visser M , de Vries JH , Weight change during chemotherapy in breast cancer patients: a meta-analysis. BMC Cancer. 2017;17 (1 ):259. Epub 2017/04/14. doi: 10.1186/s12885-017-3242-4.28403873 22. Jo A , Scarton L , O’Neal LJ , Larson S , Schafer N , George TJ , New onset of type 2 diabetes as a complication after cancer diagnosis: A systematic review. Cancer Med. 2021;10 (2 ):439–46. Epub 2020/12/29. doi: 10.1002/cam4.3666.33355998 23. Coussens LM , Werb Z . Inflammation and cancer. Nature. 2002;420 (6917 ):860–7. Epub 2002/12/20. doi: 10.1038/nature01322.12490959 24. Adamo L , Rocha-Resende C , Prabhu SD , Mann DL . Reappraising the role of inflammation in heart failure. Nat Rev Cardiol. 2020;17 (5 ):269–85. Epub 2020/01/24. doi: 10.1038/s41569-019-0315-x.31969688 25. Panova-Noeva M , Schulz A , Arnold N , Hermanns MI , Prochaska JH , Laubert-Reh D , Coagulation and inflammation in long-term cancer survivors: results from the adult population. J Thromb Haemost. 2018;16 (4 ):699–708. Epub 2018/02/13. doi: 10.1111/jth.13975.29431889 26. Lenneman CG , Sawyer DB . Cardio-Oncology: An Update on Cardiotoxicity of Cancer-Related Treatment. Circ Res. 2016;118 (6 ):1008–20. Epub 2016/03/19. doi: 10.1161/CIRCRESAHA.115.303633.26987914 27. Henriksen PA . Anthracycline cardiotoxicity: an update on mechanisms, monitoring and prevention. Heart. 2018;104 (12 ):971–7. Epub 2017/12/09. doi: 10.1136/heartjnl-2017-312103.29217634 28. van Nimwegen FA , Schaapveld M , Janus CP , Krol AD , Petersen EJ , Raemaekers JM , Cardiovascular disease after Hodgkin lymphoma treatment: 40-year disease risk. JAMA Intern Med. 2015;175 (6 ):1007–17. Epub 2015/04/29. doi: 10.1001/jamainternmed.2015.1180.25915855 29. Nguyen PL , Je Y , Schutz FA , Hoffman KE , Hu JC , Parekh A , Association of androgen deprivation therapy with cardiovascular death in patients with prostate cancer: a meta-analysis of randomized trials. JAMA. 2011;306 (21 ):2359–66. Epub 2011/12/08. doi: 10.1001/jama.2011.1745.22147380 30. Weaver KE , Foraker RE , Alfano CM , Rowland JH , Arora NK , Bellizzi KM , Cardiovascular risk factors among long-term survivors of breast, prostate, colorectal, and gynecologic cancers: a gap in survivorship care? J Cancer Surviv. 2013;7 (2 ):253–61. Epub 2013/02/19. doi: 10.1007/s11764-013-0267-9.23417882

PMC009xxxxxx/PMC9772400.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 2985117R J Immunol J Immunol Journal of immunology (Baltimore, Md. : 1950) 0022-1767 1550-6606 36427001 9772400 10.4049/jimmunol.2200609 EMS156475 Article Diversity In Cortical Thymic Epithelial Cells Occurs Through Loss Of A Foxn1-Dependent Gene Signature Driven By Stage-Specific Thymocyte Crosstalk1 White Andrea J. * Parnell Sonia M. * Handel Adam ‡¶ Maio Stefano ‡ Bacon Andrea * Cosway Emilie J. * Lucas Beth * James Kieran D. * Cowan Jennifer E. † Jenkinson William E. * Hollander Georg A. ‡§# Anderson Graham * * Institute of Immunology and Immunotherapy, University of Birmingham, Birmingham, UK ‡ Department of Paediatrics and Institute of Developmental and Regenerative Medicine, University of Oxford, Oxford, UK ¶ Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK § Paediatric Immunology, Department of Biomedicine, University of Basel and University Children’s Hospital Basel, Basel, Switzerland # Department of Biosystems Science and Engineering, ETH Zurich, Basel, Switzerland † Division of Infection and Immunity, University College London, London, UK Correspondence to: Professor Graham Anderson, Institute for Immunology and Immunotherapy, Floor 4 Institute for Biomedical Research, Medical School, University of Birmingham, B15 2TT, United Kingdom. Tel: (44)1214146817. g.anderson@bham.ac.uk 01 11 2022 14 11 2022 01 1 2023 ji2200609This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. In the thymus, cortical and medullary thymic epithelial cells (TEC) support αβT-cell development from lymphoid progenitors. For cortical TEC (cTEC), expression of a specialised gene signature that includes Cxcl12, Delta-like 4 (Dll4) and Psmb11 enables the cortex to support T-lineage commitment and the generation and selection of CD4+CD8+ thymocytes. While the importance of cTEC in T-cell development is well defined, mechanisms that shape the cTEC compartment and regulate its functional specialisation are unclear. Using a Cxcl12DsRed reporter mouse model, we show that changes in Cxcl12 expression reveal a developmentally regulated programme of cTEC heterogeneity. While cTEC are uniformly Cxcl12DsRed+ during neonatal stages, progression through postnatal life triggers the appearance of Cxcl12DsRed- cTEC that continue to reside in the cortex alongside their Cxcl12DsRed+ counterparts. This appearance of Cxcl12DsRed- cTEC is controlled by maturation of CD4-CD8-, but not CD4+CD8+ thymocytes, demonstrating stage-specific thymocyte crosstalk controls cTEC heterogeneity. Importantly, while fate mapping experiments show both Cxcl12DsRed+ and Cxcl12DsRed- cTEC share a common Foxn1+ cell origin, RNA sequencing analysis shows Cxcl12DsRed- cTEC no longer express Foxn1, which results in loss of the FOXN1-dependent cTEC gene signature and may explain the reduced capacity of Cxcl12DsRed- cTEC for thymocyte interactions. In sum, our study shows that shaping of the cTEC compartment during the life-course occurs via stage-specific thymocyte crosstalk, which drives loss of Foxn1 expression and its key target genes which may then determine the functional competence of the thymic cortex. pmcIntroduction Self-tolerant MHC restricted CD4+ and CD8+ αβT-cells are produced exclusively in the thymus, a primary lymphoid organ that guides lymphoid progenitors through multiple developmental events. Importantly, many studies have shown the key roles that thymic stromal cells play in controlling thymocyte development (1–3). In particular, thymic epithelial cells (TEC) are functionally important during multiple developmental events that occur within anatomically distinct thymic areas(4). For example, EpCAM1+UEA1+Ly51- medullary thymic epithelial cells (mTEC) are key in controlling T-cell tolerance induction through the induction of both negative selection and Foxp3+ T-cell development (5, 6). In contrast, cortex-resident cortical TEC (cTEC), typically defined as EpCAM1+ UEA1- Ly51+ cells, are critical regulators of early T-cell development. For example, upon entry to the thymus, lymphoid progenitors undergo interactions with Delta-like 4 (DLL4) expressing cTEC, which induces Notch signalling and directs progenitors towards a T-cell fate (7–9). Immature thymocytes then transit through a series of CD4-CD8- double negative (DN) stages, including CD44+CD25- DN1, CD44+CD25+ DN2 and CD44-CD25+ DN3, where they rearrange the Tcrb gene and express TCRβ protein as part of the cell surface pre-TCR complex. Importantly, selection of TCRβ-expressing DN3 cells is also controlled by cTEC products, with CXCL12 and DLL4 acting in concert with the pre-TCR to generate large cohorts of pre-selection CD4+CD8+αβTCRlow thymocytes (10, 11). cTEC expression of MHC/self-peptide complexes then enables the cortex to support positive selection of CD4+CD8+ thymocytes, that results in the generation of single positive (SP) CD4+ and CD8+ thymocytes. Here, the unique ability of cTEC to support positive selection is at least in part attributed to their specialised antigen processing capabilities (12). For example, unique expression of Psmb11, the gene encoding the thymoproteosomal subunit β5t, enables cTEC to produce MHC class I bound self-peptides that result in the effective positive selection of CD8+ thymocytes (13, 14). Similarly, cTEC expression of Cathepsin-L (15) and Prss16 (16) enables the generation MHC class II/self-peptide complexes that drive efficient CD4+ thymocyte selection. Autophagic properties of cTEC may also aid in their control of positive selection(17). Significantly, many of the genes expressed by cTEC that underpin their functional specialisation, including Cxcl12, Dll4, Psmb11 and Ctsl, are known targets of FOXN1 (18, 19), a transcription factor that plays an essential role in TEC development and function (20–22). Thus, cTEC expression of FOXN1 plays an important role in controlling a key gene expression signature that enables the cortex to support multiple stages of T-cell development. Despite this importance of cTEC for thymus function, our understanding of the mechanisms that control their development remains incomplete. To address cTEC development and heterogeneity, we examined Ly51+UEA1- cTEC for evidence of heterogeneity using mice in which the fluorescent protein DsRed reports expression of the functionally important cTEC gene Cxcl12 (23). We found that cTEC in adult mice can be readily subdivided into Cxcl12DsRed+ and Cxcl12DsRed- subsets that both reside within the thymic cortex, with qPCR analysis confirming their differential Cxcl12 gene expression. Interestingly, examination of cTEC heterogeneity across the life-course revealed a developmentally regulated programme where cTEC were uniformly Cxcl12DsRed+ at neonatal stages, with Cxcl12DsRed- cTEC appearing 1 week after birth and persisting into adulthood. Importantly, while fate mapping experiments show Cxcl12DsRed+ and Cxcl12DsRed- cTEC both derive from FOXN1+ cells, RNA sequencing analysis showed these populations to be transcriptionally distinct. Unlike Cxcl12DsRed+ cTEC, Cxcl12DsRed- cTEC lacked Foxn1 expression, and this was accompanied by a change in the gene expression profiles of FOXN1 targets, including Cxcl12 itself, as well as Psmbl11, and the Notch ligand Dll4. Furthermore, the emergence of Cxcl12DsRed- cTEC was impaired in Rag2-/- but not Tcra-/- mice, and Cxcl12DsRed- cTEC were impaired in their ability to form successful cellular interactions with thymocytes when compared to their Cxcl12DsRed+ counterparts. Taken together, our study identifies a developmentally regulated programme of cTEC heterogeneity, where signals arising from the maturation of immature DN3 thymocytes cause transcriptional changes in the cTEC population that result in loss of Foxn1 expression and transcripts of its downstream targets. This then creates epithelial heterogeneity in the thymic cortex that may influence functionality within the cTEC compartment. Materials and Methods Mice The following mice on a C57BL/6 background were purchased from The Jackson Laboratory and used at 10 weeks of age unless otherwise stated: Cxcl12DsRed knockin (stock no. 022458) (23), which were used in isolation or crossed with the following: Tcrα-/- (stock no. 002116, (24), Rag2-/-(stock no. 008449) (25), Foxn1Cre (stock no. 018448)(26) and Rosa26-stop-EYFP (stock no. 006148) (27). Control mice for experiments involving and Tcra-/- and Rag2-/- mice were heterozygous littermate controls. RANKVenus BAC transgenic mice were generated as described previously (28). Husbandry, housing, and experimental methods involving mice were performed at the Biomedical Services Unit at the University of Birmingham in accordance with the local Ethical Review Panel and UK Home Office Regulations (Animal project Licence number P3ACFED06). Flow Cytometry, Cell Sorting and Antibodies For thymic epithelial cell analysis, single cell suspensions were generated by digesting thymic lobes with collagenase dispase (2.5mg/ml, Roche) and DNase 1 (40mg/ml Roche). CD45- cells were enriched by the depletion of CD45+ cells using anti-CD45 beads and LS columns (Miltenyi Biotec). The following antibodies were used for TEC analysis: anti-CD45 clone 30-F11 (eBioscience), anti-EpCAM1 clone G8.8 (eBioscience), anti-Ly51 clone 6C3 (Biolegend), anti-MHCII clone M5/114.15.2 (eBioscience), anti-CD80 clone 16-10A1 (Biolegend), CD104 clone 346-11A (Biolegend), and anti-MHCI 28-14-8. Biotinylated UEA-1 (Vector laboratories) was detected using streptavidin PECy7 (eBioscience). Cells were analysed using a LSR Fortessa (Becton Dickinson) with data analysis carried out using Flowjo v10 (Becton Dickinson). For cell sorting, TEC subsets were identified using the antibodies above, and isolated using a FACS Aria Fusion 1 cell sorter (Becton Dickinson). The sorting strategy for the different TEC subsets were as follows, Cxcl12DsRed+ cTEC: CD45-EpCAM1+UEA1-Ly51+CXCL12DsRed+; CXCL12DsRed- cTEC: CD45-EpCAM1+UEA1-Ly51+CXCL12DsRed-; mTEClo: CD45-EpCAM1+UEA1+Ly51-CD80-MHCII-; mTEChi: CD45-EpCAM1+UEA-1-Ly51+CD80+MHCII+; CD104+ mTEClo, CD45-EpCAM1+UEA1+Ly51-CD80-MHCII-CD104+; CD104- mTEClo, CD45-EpCAM1+UEA1+Ly51-CD80-MHCII-CD104-. Immunohistochemistry and Confocal Microscropy Thymus tissue from Foxn1Cre/Rosa26YFP/Cxcl12 mice was isolated and fixed in 2% PFA (Sigma) for 2 hours, then overnight in 15% Sucrose (Sigma). Thymic lobes were frozen on dry ice, and sectioned at 7μm within 24 hours of freezing. eYFP protein in sections from Foxn1Cre/Rosa26YFP/Cxcl12DsRed was amplified using rabbit anti-GFP (ThermoFisher) and donkey anti-rabbit 488 (ThermoFisher). Sections were counter stained with DAPI (4’,6-diamidino-2-phenylindole)(Sigma) and mounted using Prolong Diamond (ThermoFisher). Sections were imaged using Zeiss Zen 880 microscope (Zeiss) and analysis using Zeiss Zen Black (Zeiss). qPCR Analysis Real-time PCR was performed as described previously(29) on a Corbett Rotor Gene-3000 PCR machine (Qiagen) using a SensiMix SYBR No ROX Kit (Meridian Bioscience-Bioline) and primers specific for Actb (β-actin) (Qiagen), and indicated genes of interest (Sigma-Merck). Data shown are typical of at least two independently sorted sample sets, histograms represent the mean (±SEM) of replicate reactions. Primer sequences used were: Foxn1: forward 5′-CAAATTCTGCAGGGGTCAGA-3′ and reverse 5′-TGGGGTGCAATCCTCTGATA-3′; Cxcl12: forward 5′-GCTCTGCATCAGTGACGGTA-3′ and reverse 5′-TGTCTGTTGTTGTTCTTCAGC-3′; Psmb11: forward 5′-ATCGCTGCGGCTGATACTC-3′ and reverse 5′-GCAGGACATCATAGCTGCCAA-3′; Prss16: forward 5′-GTATTTCTGCACATAGGAGGCG-3′ and reverse 5′-TGTTCTAGGCTTATCACCAGGG-3′; Cd83: forward 5′-AGGGCCTATTCCCTGACGAT-3′ and reverse 5′-CTTCCTTGGGGCATCCTGTC-3′; Dll4: forward 5′-GAAGCGCGATGACCACTTCG-3′ and reverse 5′-TGGACGGCAGATGCACTCAT-3′; Ly75: forward 5′-GCTCAGGTAATGATCCATTCACC-3′ and reverse 5′-TTAGTTCCGCTACAGTCCTGG-3′; Ctsl: forward 5′-ATCAAACCTTTAGTGCAGAGTGG-3′ and reverse 5 ′-CTGTATTCCCCGTTGTGTAGC-3′; Epcam1: forward 5′-TTGCTCCAAACTGGCGTCTAA-3′ and reverse 5′-GCAGTCGGGGTCGTACA-3′; Aire: forward 5′-TGCATAGCATCCTGGACGGCTTCC-3′ and reverse 5′-CCTGGGCTGGAGACGCTCTTTGAG-3′; Trpm5: forward 5′-CCAGCATAAGCGACAACATCT-3′ and reverse 5′-GAGCATACAGTAGTTGGCCTG-3′; Ccl21a: forward 5′-ATCCCGGCAATCCTGTTCTC-3′ and reverse 5′-GGGGCTTTGTTTCCCTGGG-3′; Actb (β-actin): QuantiTect Mm Actb 1SG Primer Assay (Qiagen, QT00095242). Bulk RNA Sequencing RNA samples were extracted using the QIagen RNAeasy kit. Libraries were prepared using the SMARTer Ultra Low Input RNA Kit for Sequencing as per manufacturer instructions and sequenced on an Illumina NovaSeq platform. Reads were trimmed for adapter contamination using Trimmomatic (version 0.36) and aligned to the mm10 mouse genome using STAR (version 2.7.3a) (30, 31). Reads were assigned to genes using HTSeq (version 0.12.4) with the option “intersection-nonempty” (32). Differentially expressed genes were identified using edgeR (FDR < 0.05) (33). Enrichment of Foxn1 high confidence genes (18) was assessed by comparing the log2 fold expression for Foxn1 high confidence genes to a control set of genes matched by expression decile using a Wilcoxon rank sum test. Sequencing data is available at the Gene Expression Omnibus GEO (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE205940. Accession number GSE205940). Gene ontology analysis was performed using clusterProfiler (34). Cell Conjugate Analysis Thymocyte-TEC cell conjugate experiments were carried out using protocol adapted from Hare et al (35). In short, CD45-EpCAM1+ TEC were FACS-sorted from 10 week Cxcl12DsRed mice and neonatal day 2 wildtype mice, and labelled with CFSE according to manufacturer instructions (ThermoFisher). A single cell suspension of WT adult thymocytes were labelled with Cell Trace Violet according to manufacturer’s instructions (ThermoFisher), and the two cell types were mixed at a 5:1 ratio (Thymocytes:TEC). The mixed suspension was then centrifuged, the supernatant removed, and the cell pellet vortexed and incubated at 37°C for 20 minutes, a timepoint that enables successful conjugate formation between WT TEC and thymocytes (35). Samples were resuspended in a volume of 200μl of PBS (Sigma), and analysed using a BD LSR Fortessa. Results Progressive Loss of Cxcl12 Expression Identifies A Developmentally Regulated Programme of cTEC Heterogeneity In the thymus, cTEC are classically defined as an Ly51+UEA1- subset of EpCAM1+ TEC. While the functional properties of cTEC are well described, relatively little is known about the cellular and molecular interactions that control their development and potential functional heterogeneity. To investigate this, we made use of Cxcl12DsRed reporter mice(23) in which DsRed expression identifies cells expressing Cxcl12, a cTEC-expressed chemokine that is an important regulator of thymocyte migration and development. Surprisingly, flow cytometric analysis of Ly51+UEA1- cTEC from 10 week old adult Cxcl12DsRed mice revealed striking heterogeneity with regard to DsRed expression, with the presence of distinct subsets of Cxcl12DsRed+ and Cxcl12DsRed- cTEC (Fig. 1A). Importantly, when FACS-purified DsRed+ and DsRed- cTEC cells were analysed for Cxcl12 mRNA expression by qPCR, we saw that the abundant expression of Cxcl12 mRNA in DsRed+ cells was lacking in DsRed- cells (Fig 1B). Thus, heterogeneity in adult cTEC described here reflects true heterogeneity in their Cxcl12 expression, and is not merely a feature of DsRed reporter expression. To examine cTEC heterogeneity further, we performed time-course analysis from birth up to 20 weeks of adulthood. Interestingly, we saw that cTEC from neonatal (postnatal day 1, P1) were uniformly Cxcl12DsRed+ (Fig. 1C). While the vast majority of cTEC were also Cxcl12DsRed+ at the 1 week stage, we detected a distinct Cxcl12DsRed- cTEC subset at 6 weeks of life (Fig. 1C), with the proportions of Cxcl12DsRed+ and Cxcl12DsRed- cTEC remaining constant for the remainder of the observation period (Fig. 1C, D). Collectively, these findings identify Cxcl12+ and Cxcl12- subsets within the bulk cTEC compartment that are ordered in their appearance during development, suggesting the cTEC compartment undergoes developmentally regulated changes that can be measured by differences in Cxcl12 expression. Cxcl12DsRed- cTEC Are Transcriptionally Distinct from Their Cxcl12DsRed+ Counterparts, And Lack Foxn1 Expression and a FOXN1-Dependent Gene Signature To understand the events underlying this cTEC heterogeneity, we used RNA sequencing to compare the transcriptomes of Cxcl12DsRed+ and Cxcl12DsRed- cTEC. Here, Cxcl12DsRed+ and Cxcl12DsRed- subsets of total CD45-EpCAM1+UEA1-Ly51+ cTEC were FACS-sorted from 10 week old adult Cxcl12DsRed reporter mice, with experiments performed in triplicate to produce 3 independent biological replicates for each subset. This approach identified 946 genes differentially expressed between DsRed+ and DsRed- cTEC (Fig 2A). Much of this transcriptomic difference was driven by the lower expression of genes known to be direct targets of FOXN1 in Cxcl12DsRed- cTEC relative to Cxcl12DsRed+ cTEC, and this correlated with the lack of expression of Foxn1 in the former (p < 0.0001, Wilcoxon rank sum test; Fig 2B-D) (18). For example, the heatmap analysis in Figure 2D shows clear differences in expression of Foxn1 and several of its direct targets including Cxcl12, Dll4, Cd83, Ccl25, Ly75, Psmb11, and Prss16. Further quantitative PCR (qPCR) analyses confirmed data obtained from RNAseq experiments, including the absence of Foxn1 transcripts in Cxcl12DsRed- cTEC (Fig 3A), as well as the absence of transcripts encoding FOXN1 target genes that play key roles in specific stages of thymocyte development including thymocyte migration (Cxcl12), Notch signalling (Dll4), and antigen processing/presentation (Prss16, Psmb11, Ctsl, Ly75). By contrast, Cxcl12DsRed+ and Cxcl12DsRed- cTEC subsets showed no reduction in levels of Epcam1 mRNA (Fig 3A). Importantly, Cxcl12DsRed+ and Cxcl12DsRed- cTEC showed comparable levels of Enpep expression, the gene encoding the cTEC marker Ly51 (Figure 3B). qPCR analysis showed both Cxcl12DsRed+ and Cxcl12DsRed- cTEC subsets lacked expression of mTEC markers, including the tuft cell marker Trpm5, as well as Aire and Ccl21a that were readily detectable within mTEC subsets (Figure 3C). Moreover, by crossing Cxcl12DsRed with RANKVenus reporter mice, we saw both cTEC subsets lacked expression of RANK, a key marker and regulator of mTEC (Figure 3D). These findings support the idea that Cxcl12DsRed- Ly51+UEA1- cells belong to the cTEC lineage, and do not contain mTEC lineage cells. Finally, although both Cxcl12DsRed+ and Cxcl12DsRed- cTEC expressed MHC class I and MHC class II, their cell surface expression levels were significantly lower on Cxcl12DsRed- cTEC (Fig. 3E). To examine further the nature of Cxcl12DsRed- cTEC in relation to their Cxcl12DsRed+ counterparts, we searched for genes that were differentially expressed between the two subsets (Supp. Table 1). When we analysed the expression of cTEC marker genes (36), removing those known to be FOXN1-dependent (18), we saw the expression of cTEC marker genes in Cxcl12DsRed+ and Cxcl12DsRed- cTEC subsets were similar (Figure 4A). Interestingly however, gene ontology analysis pointed towards some potential differences. For example, in Cxcl12DsRed+ cTEC we saw enrichment of pathways associated with regulation of endothelial cell proliferation, angiogenesis and vascular development, while Cxcl12DsRed- cTEC showed enrichment of other distinct pathways including and serine-type endopeptidase activity regulation of granulocyte migration (Figure 4B, C). Collectively, these data suggest that while the major difference between Cxcl12DsRed+ and Cxcl12DsRed- cTEC relates to expression of Foxn1 and a FOXN1-dependent cTEC signature, they may also harbour gene expression patterns that point towards functional differences between the two subsets. The presence of Foxn1- cTEC in the adult thymus could occur as a result of the downregulation of FOXN1 in cells that had previously expressed FOXN1, or via the progressive emergence of a cTEC subset with no prior history of FOXN1 expression. To distinguish between these possibilities, we used a fate mapping approach to examine the history of FOXN1 expression in Cxcl12DsRed+ and Cxcl12DsRed- cTEC. In adult Foxn1Cre/Rosa26YFP/Cxcl12DsRed mice, the vast majority of both Cxcl12DsRed+ and Cxcl12DsRed- cTEC were Foxn1Cre fate mapped (Fig. 5A, B), indicating both cTEC subpopulations were generated from FOXN1 expressing cells. Confocal analysis of thymus sections from these mice demonstrated that both Cxcl12DsRed+ and Cxcl12DsRed- Foxn1Cre-fate mapped cells were present within thymic cortex areas (Fig 5C). Use of confocal microscopy to further examine the phenotypic properties of cortex-resident Cxcl12DsRed- cells was unfortunately hampered by the impact of PFA fixation, required to preserve DsRed protein, on successful antibody staining. Collectively, these findings show FOXN1 is not uniformly expressed within the adult cTEC compartment, with the presence of FOXN1- cTEC providing an explanation for the presence of those cells that lack expression of the target gene Cxcl12. Importantly, our findings also show that heterogeneity in FOXN1 expression by cTEC extends beyond differences in Cxcl12 expression and includes the differential expression of FOXN11-controlled loci (e.g. Dll4, Ccl25, Psmbl1, Prss16) that are important in the regulation of cortical T-cell development. Despite this change in the cTEC-specific mRNA signature, Cxcl12DsRed- cTEC continue to reside within cortical areas alongside their Cxcl12DsRed+ counterparts, where they contribute to the reticular epithelial network of the adult thymic cortex. Stage-Specific Thymocyte Crosstalk Regulates cTEC Heterogeneity Signals from developing thymocytes are known to regulate the development and formation thymic microenvironments, a process termed thymic crosstalk (37, 38). Much of our understanding of this process comes from studies examining the cellular interactions that govern events in the thymus medulla. For example, crosstalk with mTEC regulates development of Aire+ mTEC (39, 40) and post-Aire stages (29, 41). In contrast, how thymic crosstalk signals influence the thymic cortex, and in particular how they might control the Cxcl12/Foxn1 cTEC heterogeneity described here, is unclear. To examine this specific aspect, we crossed Cxcl12DsRed mice with Rag2-/- and Tcra-/- mice, where T-cell development is blocked at the CD4-CD8- or CD4+CD8+ stages, respectively. Interestingly, in Tcra-/- Cxcl12DsRed mice, cTEC heterogeneity was comparable to littermate controls (Fig 6A,B), with no alterations in the proportions of Cxcl12DsRed+ and Cxcl12DsRed- cTEC (Fig. 6C) or the ratio of DsRed+:DsRed- cTEC (Fig. 6D). MFI levels of DsRed in Cxcl12DsRed+ cTEC were also comparable (Fig. 6E). Thus, the appearance of Cxcl12DsRed- cTEC occurs normally in the absence of CD4+ and CD8+ single positive thymocytes, suggesting that positive selection of CD4+CD8+ thymocytes is not essential for the generation of Cxcl12DsRed cTEC heterogeneity. In contrast, when we performed similar analysis of Rag2-/-Cxcl12DsRed mice (Fig. 6F), we saw that the proportion of Cxcl12DsRed-cTEC was decreased, with a concomitant increase in Cxcl12DsRed+ cTEC (Fig. 6G, H). This finding was accompanied by a skewing of the DsRed+:DsRed- cTEC ratio in favour of DsRed+ cells (Fig 6I), with Cxcl12DsRed+ cTEC in Rag2-/- mice also showing higher levels of DsRed compared to littermate controls (Fig 6J). These findings show that in the absence of CD4+CD8+ thymocytes, the appearance of Cxcl12DsRed- cTEC is impaired, suggesting that maturation of CD4-CD8- thymocytes is an important regulator of cTEC heterogeneity in the adult thymus. The functional ability of cTEC is regulated by their expression of several key genes now known to be Foxn1 targets (18). Interestingly, a recent study (42) has shown that the formation of successful cellular interactions with thymocytes requires CXCL12 and DLL4, both of which are Foxn1 targets that are absent from Cxcl12DsRed- cTEC. Given these differences between Cxcl12DsRed+ and Cxcl12DsRed- cTEC, we wondered whether this may have functional consequences for their abilities to influence T-cell development. To investigate this, we performed a flow cytometry based cell conjugate assay where TEC-thymocyte interactions occur in a TCR-MHC independent manner (35) to compare the ability of Cxcl12DsRed+ and Cxcl12DsRed- cTEC to form successful TEC-thymocyte conjugates. Here, purified EpCAM1+ TEC were FACS-sorted from adult Cxcl12DsRed mice, labelled with the fluorescent dye CFSE, and mixed with Cell Trace Violet labelled thymocytes at a ratio of 5:1 Thymocytes:TEC (Fig 7A). Following centrifugation and 20 minutes incubation, pellets were gently disrupted and conjugate formation was assessed by flow cytometry following gating on Cxcl12DsRed+ and Cxcl12DsRed- cTEC within the total cTEC population (Fig 7B). While both Cxcl12DsRed+ and Cxcl12DsRed- cTEC were capable of conjugate formation, we saw a significant decrease in conjugates formed from Cxcl12DsRed- cTEC (Fig 7B), suggesting Cxcl12DsRed- cTEC may be less effective than their Cxcl12DsRed+ counterparts to influence T-cell development. Interestingly, when we compared the efficiency of TEC-conjugate formation using adult Cxcl12DsRed+ cTEC and neonatal cTEC, the latter being uniformly Cxcl12DsRed+ (Fig. 1C), we found them to be equally effective in mediating thymocyte interactions (Fig. 7B). Thus, the ability of Cxcl12DsRed+ cTEC to influence cortex-dependent thymocyte development may be consistent throughout the life-course, and any changes in this process may occur as a result of the progressive emergence of Cxcl12DsRed- cTEC. Discussion Interactions between thymocytes and cTEC/mTEC populations support the intrathymic development and selection of αβT-cells. Through examination of the cTEC compartment, we identified a developmentally regulated programme of heterogeneity which occurs over the life course and is defined by loss of expression of Foxn1 and its downstream targets. While our finding that all TEC arise from Foxn1-expressing cells is consistent with previous reports (20), what causes some cTEC to downregulate Foxn1, and Foxn1 dependent genes, is not known. Importantly, while Foxn1- TEC have been described previously (43–45), multiple features including their intrathymic positioning, transcriptomic profile, and intrathymic generation have remained poorly understood. Here, by identifying the gene profile of these cells, including their loss of a functionally important cTEC gene signature, we provide evidence they are transcriptionally distinct from their Foxn1-expressing counterparts. Moreover, the intrathymic positioning within the cortex of the cTEC subsets defined here, together with their regulation by CD4-CD8- but not CD4+CD8+ thymocytes, extends our understanding of complexity of the cTEC compartment and the mechanisms that control this. Indeed, as the appearance of cTEC that lack Foxn1 and its key target genes is regulated by thymocyte crosstalk, in particular events specific to CD4-CD8- thymocytes, it may be that early stages of T-cell development generate signals that cause loss of Foxn1, which then results in cTEC heterogeneity. Interestingly, analysis from birth up to 20 weeks of age showed that the frequency of Cxcl12DsRed- cTEC had plateaued by around 10 weeks, which may indicate that turnover of Cxcl12DsRed- cells takes place, rather than a process that results in their progressive accumulation during the life course. The presence within the adult thymic cortex of cTEC that no longer express key genes regulating specific stages of thymocyte development raises multiple interesting scenarios. For example, it may be relevant to understanding progressive changes in thymus function under homeostatic conditions. Here, as both Cxcl12 and Dll4 are important regulators of the β-selection checkpoint. (46), absence of these genes in Foxn1- cTEC may impact the ability of the thymus to support transition to the CD4+CD8+ stage. Also significant is that while Psmb11, the cTEC-specific gene encoding the thymoproteosome component β5t, is unique to cTEC (12), our data suggest that not all adult cTEC express transcripts of Psmb11. Thus, it may be the case that in the adult thymus, both Psmb11+ and Psmb11- cTEC contribute to CD8+ SP positive selection, but they generate distinct αβTCR repertoires as a result of differences in the MHC class I bound self-peptides they can produce (thymoproteosome/β5t dependent peptides for Cxcl12DsRed+ cTEC versus non-thymoproteosome/β5t independent peptides for Cxcl12DsRed- cTEC). Here, it is important to note that β5t-deficient mice are still able to positively select some SP8+ thymocytes (47), a finding that may be consistent with the scenario that cTEC lacking Psmb11 can to contribute to SP8 generation in normal mice. Alternatively, adult Foxn1- cTEC that lack Psmb11 may be incapable of positive selection due to other functional defects, such as a failure to interact with CD4+CD8+ thymocytes. While it is interesting to note that Cxcl12DsRed- cTEC express significantly lower levels of MHC class I relative to their Cxcl12DsRed+ counterparts, and form fewer cell-cell conjugates with thymocytes, further studies are required to examine the functional properties of the cTEC subsets described here. Relevant to this, our attempts to compare the functional abilities of FACS-sorted Cxcl12DsRed+ and Cxcl12DsRed- cTEC from adult mice in reaggregate thymus organ cultures were unsuccessful. Here, intact 3-dimensional structures consistently failed to form when using TEC isolated from adult mice, which is in contrast to the efficient generation of intact RTOC from embryonic TEC (48, 49). The reasons for the inability of adult TEC to effectively form RTOC under conditions that support embryonic TEC reaggregation are not clear. However, it is interesting to note that early studies on the capacity of embryonic tissues to undergo effective reaggregation attributed this to their ability to undergo what was termed ‘inductive interactions’(50), which may be missing from adult TEC. Whatever the case, further studies are required to compare the functional capacity of cTEC subsets described here, which would also benefit from the creation of improved experimental systems to study adult TEC functions in vitro. Beyond directly influencing specific stages of thymocyte development, Cxcl12DsRed- cTEC may also play role in physically supporting the epithelial scaffold within the thymus cortex, a possibility raised recently in the context of the presence of FOXN1- TEC in thymus (44). Such a possibility may be compatible with our finding that Cxcl12DsRed- cTEC are interspersed in the cortex alongside Cxcl12DsRed+ cells. A final possibility is that alongside loss of Foxn1-mediated functional properties, Cxcl12DsRed- cTEC acquire new functional features that are important in adult thymus cortex organisation and/or function. Again, further examination requires approaches to directly assess the functional properties of defined cTEC subsets. In sum, we show that the Ly51+UEA1- cTEC compartment undergoes developmentally regulated changes in its cellular makeup that are driven by interactions with the maturation of immature CD4-CD8- thymocytes. We identify the emergence of a cTEC subset that retains its Ly51+UEA1- phenotype and positioning within the cortex, but has ceased to express Foxn1, resulting in the lack of expression of key Foxn1 target genes that define the functional properties of cTEC. These findings demonstrate the emerging complexity of the thymic cortex, and will aid in future studies that examine the role of this intrathymic site in thymocyte development. Supplementary Material Supplementary Table 1 Acknowledgements We thank BMSU staff at The University of Birmingham for expert animal husbandry, and all members of the Wellcome Trust SynThy Collaborative Award for discussion. Key Points cTEC undergo progressive changes during postnatal life caused by loss of Foxn1. Foxn1- cTEC remain within the cortex, but lack cTEC signature genes. Foxn1- cTEC show an impaired ability to influence developing thymocytes. Figure 1 Cxcl12 Expression Defines Developmentally Controlled Heterogeneity in cTEC. (A) Flow cytometric analysis of EpCAM1+CD45- thymic epithelial cells (TEC) from adult 10 week Cxcl12DsRed mice, separated into UEA1+Ly51- mTEC and UEA1-Ly51+ cTEC. Levels of Cxcl12DsRed expression in cTEC (gray line) and mTEC (black line) are shown. (B) shows qPCR expression of Cxcl12 mRNA in FACS-sorted Cxcl12DsRed+ and Cxcl12DsRed- cTEC subsets, with mTEC shown for comparison. (C) shows time-course analysis of Cxcl12DsRed expression in cTEC, identified using the gating shown, as UEA1-Ly51+ cells, from mice at indicated ages. Gates are set using mTEC as in (A). (D) shows quantitation of Cxcl12DsRed+ and Cxcl12DsRed- cTEC subsets. Each time point is from a minimum of n=4 mice and at least 3 separate experiments: P1 n=5, 1W n=4, 6W n=5, 10W n=9, and 20W n=6. P values are as follows and indicate the significance relative to P1, using a Mann-Whitney non-parametric test: **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001. Error bars represent mean ± SEM. Figure 2 Cxcl12DsRed+ and Cxcl12DsRed- cTEC Subsets Are Transcriptionally Distinct. RNA-seq analysis of FACS-sorted Cxcl12DsRed+ and Cxcl12DsRed- cTEC from 10 week Cxcl12DsRed mice. (A) shows a volcano plot of differentially expressed genes between cTEC subsets, red dots represent FDR<0.05 and black dots represent no significance. (B) shows a volcano plot of differentially expressed genes, emphasising significant FOXN1 high confidence target genes shown by red dots, with all other genes represented by grey dots. (C) shows an MA plot for all genes highlighting high confidence FOXN1 target genes in red, other genes shown in grey. Graphs show a technical triplicate of a single experiment that is representative of 3 individally sorted biological replicates. (D) shows a heatmap of significantly differentially expressed FOXN1 target genes as identified in Zuklys et al (18), and scaled mean expression of all FOXN1 target genes. Only FOXN1 targets with mean expression >1 counts per million (CPM) were included. Genes associated with cTEC phenotype/function are highlighted in red. Figure 3 Cxcl12DsRed- cTEC Lack Expression of Foxn1 and A FOXN1 Target Gene Signature. (A) shows analysis of gene expression by qPCR in Cxcl12DsRed+ (gray bars) and Cxcl12DsRed- (white bars) cTEC that were FACS-sorted from 10 week Cxcl12DsRed mice. (B) shows levels of expression of Enpep obtained from bulk RNA sequencing data in Cxcl12DsRed+ and Cxcl12DsRed- cTEC. (C) shows qPCR analysis of mTEC-expressed genes Trpm5, Aire and Ccl21a in Cxcl12DsRed+ and Cxcl12DsRed- cTEC compared to relevant mTEC subsets. For all qPCR, graphs represent of data obtained from at least 2 independentally sorted biological samples, with dots showing technical repeats. Error bars represent mean ± SEM (D) Flow cytometric analysis of RANKVenus expression by total UEA1+ mTEC, Cxcl12DsRed+ and Cxcl12DsRed- cTEC from Cxc12DsRedRANKVenus reporter mice, n=5 from 5 separate experiments. (E) Flow cytometric analysis of indicated cell surface markers in Cxcl12DsRed+ (gray line) and Cxcl12DsRed- (black line) cTEC from 10 week old Cxcl12DsRed mice. Control staining levels obtained via omission of primary antibodies are shown as a grey line. Panel (E) also shows MFI analysis of indicated markers in Cxcl12DsRed cTEC subsets. Data is from least 3 experiments, for MHC II n=8, and MHC I n=4. P values are as follows and indicate the significance relative to P1, using a Mann-Whitney non-parametric test: *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001. Error bars represent mean ± SEM. Figure 4 Comparative Analysis of Gene Expression In Cxcl12DsRed+ and Cxcl12DsRed- cTEC. (A) shows a boxplot of Foxn1-independent cTEC gene expression and genes matched by decile expression. cTEC marker genes were defined as those expressed more highly in perinatal or mature cTEC than other cell types in a reference dataset (36). cTEC markers that were Foxn1-enhanced (significantly upregulated or ≥0.25 log2 fold higher with increased Foxn1 (18) were removed to leave only FOXN1-independent cTEC markers. Expression of FOXN1-independent cTEC markers were similar between Cxcl12DsRed+ and Cxcl12DsRed- cTEC. Below are shown dotplots of gene ontology analysis for biological processes (B) and molecular functions (C). Figure 5 Both Cxcl12DsRed+ and Cxcl12DsRed- cTEC Are Derived From Foxn1-Expressing Cells. Panel (A) shows gating for the identification of Cxcl12DsRed+ and Cxcl12DsRed- cTEC in 10 week old Foxn1Cre/Rosa26-YFP/Cxcl12DsRed mice, and levels of YFP expression in these cells, where YFP indicates a history of Foxn1 expression. Gates are set following gating on YFP levels in CD45+ cells, where Foxn1Cre-mediated fate mapping is absent. Quantitation is shown in (B). Data is from 5 mice across 3 experiments. (C) shows confocal analysis of PFA-treated thymus sections from Foxn1Cre/Rosa26RYFP/Cxcl12DsRed mice, analyzed for expression of YFP (shown in green), DsRed (red), with co-expression appearing yellow. Upper panels are x10 magnification and show cortex (C) and medulla (M) areas defined by DAPI, dotted line is the CMJ. Scale bar denotes 50μm. The boxed area highlighted in upper panels represents an area of the cortex that is shown at x40 magnification in the image row below. Scale bar in the lower images denotes 20μm. Arrows identify Foxn1Cre-fate mapped YFP+Cxcl12DsRed+ cTEC, while arrowheads identify Foxn1Cre-fate mapped YFP+Cxcl12DsRed- cTEC. Images are examples of 4 sections randomly chosen from 4 separate mice across 2 separate experiments. Figure 6 Stage-Specific Thymocyte Crosstalk Controls cTEC Heterogeneity. (A) shows identification of cTEC and mTEC in 10 week old Cxcl12DsRed/Tcra-/- mice and Cxcl12DsRed/Tcra+/- littermate controls, with panel (B) showing levels of Cxcl12DsRed expression after gating on cTEC. Percentages (C) and ratios (D) of Cxcl12DsRed+ and Cxcl12DsRed- cTEC in Tcra-/- mice and Tcra+/- mice are shown alongside MFI of DsRed in cTEC subsets (E). Panels (F-J) shows similar analysis of Cxcl12DsRed/Rag2-/- mice and Cxcl12DsRed/Rag2+/- littermate controls. All data are representative of at least 3 independent experiments, using the following numbers of mice: Cxcl12DsRed/Tcra-/- n=12; Cxcl12DsRed/Tcra+/- n=10; Cxcl12DsRed/Rag2-/- n=6; Cxcl12DsRed/Rag+/- n=6. P values indicate significance using a Mann-Whitney non-parametric test: **, P < 0.01. Error bars represent mean ± SEM. Figure 7 Cxcl12DsRed- cTEC Demonstrate An Impaired Capacity For Thymocyte Interactions. Panel (A) shows the experimental approach used to study cTEC:thymocyte conjugate interactions using flow cytometry. Panel (B) shows the gating approach used to compare the ability of Cxcl12DsRed+ and Cxcl12DsRed- cTEC to form conjugates with thymocytes. Successful thymocyte/cTEC conjugates appear as CFSE+TEC:CTV+ thymocyte events within Cxcl12DsRed+ and Cxcl12DsRed- cTEC subsets. Quantitation of cTEC:thymocyte conjugate formation is also shown in (B), with comparison of conjugate formation with Cxcl12DsRed+ cTEC (gray bar), Cxcl12DsRed- (white bar) cTEC, and neonatal cTEC (black bar). All data is representative of 4 individual experiments and 4 samples. P values indicate significance using a Mann-Whitney non-parametric test: *, P < 0.05. Error bars represent mean ± SEM. 1 This work was supported by an MRC Programme Grant to GA (MR/T029765/1) and a Wellcome Trust funded Collaborative Award (SynThy, 211944/Z/18/Z) where GA and GAH are partners. GAH also received funding from the Swiss National Science Foundation (IZLJZ3_171050; 310030_184672) and the Wellcome Trust (105045/Z/14/Z). JEC is a Sir Henry Dale Fellow funded by The WellcomeTrust. 1 Takahama Y Ohigashi I Baik S Anderson G Generation of diversity in thymic epithelial cells Nat Rev Immunol 2017 17 295 305 28317923 2 Han J Zuniga-Pflucker JC A 2020 View of Thymus Stromal Cells in T Cell Development J Immunol 2021 206 249 256 33397738 3 Kadouri N Nevo S Goldfarb Y Abramson J Thymic epithelial cell heterogeneity: TEC by TEC Nat Rev Immunol 2020 20 239 253 31804611 4 Petrie HT Zuniga-Pflucker JC Zoned out: functional mapping of stromal signaling microenvironments in the thymus Annu Rev Immunol 2007 25 649 679 17291187 5 Cowan JE Parnell SM Nakamura K Caamano JH Lane PJ Jenkinson EJ Jenkinson WE Anderson G The thymic medulla is required for Foxp3+ regulatory but not conventional CD4+ thymocyte development J Exp Med 2013 210 675 681 23530124 6 Perry JSA Lio CJ Kau AL Nutsch K Yang Z Gordon JI Murphy KM Hsieh CS Distinct contributions of Aire and antigen-presenting-cell subsets to the generation of self-tolerance in the thymus Immunity 2014 41 414 426 25220213 7 Fiorini E Ferrero I Merck E Favre S Pierres M Luther SA MacDonald HR Cutting edge: thymic crosstalk regulates delta-like 4 expression on cortical epithelial cells J Immunol 2008 181 8199 8203 19050235 8 Koch U Fiorini E Benedito R Besseyrias V Schuster-Gossler K Pierres M Manley NR Duarte A Macdonald HR Radtke F Delta-like 4 is the essential, nonredundant ligand for Notch1 during thymic T cell lineage commitment J Exp Med 2008 205 2515 2523 18824585 9 Chen ELY Thompson PK Zuniga-Pflucker JC RBPJ-dependent Notch signaling initiates the T cell program in a subset of thymus-seeding progenitors Nat Immunol 2019 20 1456 1468 31636466 10 Janas ML Varano G Gudmundsson K Noda M Nagasawa T Turner M Thymic development beyond beta-selection requires phosphatidylinositol 3-kinase activation by CXCR4 J Exp Med 2010 207 247 261 20038597 11 Trampont PC Tosello-Trampont AC Shen Y Duley AK Sutherland AE Bender TP Littman DR Ravichandran KS CXCR4 acts as a costimulator during thymic beta-selection Nat Immunol 2010 11 162 170 20010845 12 Ohigashi I Matsuda-Lennikov M Takahama Y Peptides for T cell selection in the thymus Peptides 2021 146 170671 34624431 13 Sasaki K Takada K Ohte Y Kondo H Sorimachi H Tanaka K Takahama Y Murata S Thymoproteasomes produce unique peptide motifs for positive selection of CD8(+) T cells Nat Commun 2015 6 7484 26099460 14 Takada K Van Laethem F Xing Y Akane K Suzuki H Murata S Tanaka K Jameson SC Singer A Takahama Y TCR affinity for thymoproteasome-dependent positively selecting peptides conditions antigen responsiveness in CD8(+) T cells Nat Immunol 2015 16 1069 1076 26301566 15 Honey K Nakagawa T Peters C Rudensky A Cathepsin L regulates CD4+ T cell selection independently of its effect on invariant chain: a role in the generation of positively selecting peptide ligands J Exp Med 2002 195 1349 1358 12021314 16 Fornari TA Marques MM Nguyen C Carrier A Passos GA Transcription profiling of Prss16 (Tssp) can be used to find additional peptidase genes that are candidates for self-peptide generation in the thymus Mol Biol Rep 2012 39 4051 4058 21773946 17 Rodrigues PM Sousa LG Perrod C Maceiras AR Ferreirinha P Pombinho R Romera-Cardenas G Gomez-Lazaro M Senkara M Pistolic J Cabanes D LAMP2 regulates autophagy in the thymic epithelium and thymic stroma-dependent CD4 T cell development Autophagy 2022 1 14 18 Zuklys S Handel A Zhanybekova S Govani F Keller M Maio S Mayer CE Teh HY Hafen K Gallone G Barthlott T Foxn1 regulates key target genes essential for T cell development in postnatal thymic epithelial cells Nat Immunol 2016 17 1206 1215 27548434 19 Uddin MM Ohigashi I Motosugi R Nakayama T Sakata M Hamazaki J Nishito Y Rode I Tanaka K Takemoto T Murata S Foxn1-beta5t transcriptional axis controls CD8(+) T-cell production in the thymus Nat Commun 2017 8 14419 28176764 20 Corbeaux T Hess I Swann JB Kanzler B Haas-Assenbaum A Boehm T Thymopoiesis in mice depends on a Foxn1-positive thymic epithelial cell lineage Proc Natl Acad Sci U S A 2010 107 16613 16618 20823228 21 Nehls M Pfeifer D Schorpp M Hedrich H Boehm T New member of the winged-helix protein family disrupted in mouse and rat nude mutations Nature 1994 372 103 107 7969402 22 Blackburn CC Augustine CL Li R Harvey RP Malin MA Boyd RL Miller JF Morahan G The nu gene acts cell-autonomously and is required for differentiation of thymic epithelial progenitors Proc Natl Acad Sci U S A 1996 93 5742 5746 8650163 23 Ding L Morrison SJ Haematopoietic stem cells and early lymphoid progenitors occupy distinct bone marrow niches Nature 2013 495 231 235 23434755 24 Mombaerts P Clarke AR Rudnicki MA Iacomini J Itohara S Lafaille JJ Wang L Ichikawa Y Jaenisch R Hooper ML Mutations in T-cell antigen receptor genes alpha and beta block thymocyte development at different stages Nature 1992 360 225 231 1359428 25 Hao Z Rajewsky K Homeostasis of peripheral B cells in the absence of B cell influx from the bone marrow J Exp Med 2001 194 1151 1164 11602643 26 Gordon J Xiao S Hughes B 3rd Su DM Navarre SP Condie BG Manley NR Specific expression of lacZ and cre recombinase in fetal thymic epithelial cells by multiplex gene targeting at the Foxn1 locus BMC Dev Biol 2007 7 69 17577402 27 Srinivas S Watanabe T Lin CS William CM Tanabe Y Jessell TM Costantini F Cre reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus BMC Dev Biol 2001 1 4 11299042 28 McCarthy NI Cowan JE Nakamura K Bacon A Baik S White AJ Parnell SM Jenkinson EJ Jenkinson WE Anderson G Osteoprotegerin-Mediated Homeostasis of Rank+ Thymic Epithelial Cells Does Not Limit Foxp3+ Regulatory T Cell Development J Immunol 2015 195 2675 2682 26254339 29 Lucas B White AJ Cosway EJ Parnell SM James KD Jones ND Ohigashi I Takahama Y Jenkinson WE Anderson G Diversity in medullary thymic epithelial cells controls the activity and availability of iNKT cells Nat Commun 2020 11 2198 32366944 30 Bolger AM Lohse M Usadel B Trimmomatic: a flexible trimmer for Illumina sequence data Bioinformatics 2014 30 2114 2120 24695404 31 Dobin A Davis CA Schlesinger F Drenkow J Zaleski C Jha S Batut P Chaisson M Gingeras TR STAR: ultrafast universal RNA-seq aligner Bioinformatics 2013 29 15 21 23104886 32 Anders S Pyl PT Huber W HTSeq--a Python framework to work with high-throughput sequencing data Bioinformatics 2015 31 166 169 25260700 33 Robinson MD McCarthy DJ Smyth GK edgeR: a Bioconductor package for differential expression analysis of digital gene expression data Bioinformatics 2010 26 139 140 19910308 34 Yu G Wang LG Han Y He QY clusterProfiler: an R package for comparing biological themes among gene clusters OMICS 2012 16 284 287 22455463 35 Hare KJ Pongracz J Jenkinson EJ Anderson G Modeling TCR signaling complex formation in positive selection J Immunol 2003 171 2825 2831 12960303 36 Baran-Gale J Morgan MD Maio S Dhalla F Calvo-Asensio I Deadman ME Handel AE Maynard A Chen S Green F Sit RV Ageing compromises mouse thymus function and remodels epithelial cell differentiation Elife 2020 9 37 Shores EW Van Ewijk W Singer A Disorganization and restoration of thymic medullary epithelial cells in T cell receptor-negative scid mice: evidence that receptor-bearing lymphocytes influence maturation of the thymic microenvironment Eur J Immunol 1991 21 1657 1661 2060577 38 van Ewijk W Shores EW Singer A Crosstalk in the mouse thymus Immunol Today 1994 15 214 217 8024681 39 Rossi SW Kim MY Leibbrandt A Parnell SM Jenkinson WE Glanville SH McConnell FM Scott HS Penninger JM Jenkinson EJ Lane PJ RANK signals from CD4(+)3(-) inducer cells regulate development of Aire-expressing epithelial cells in the thymic medulla J Exp Med 2007 204 1267 1272 17502664 40 Hikosaka Y Nitta T Ohigashi I Yano K Ishimaru N Hayashi Y Matsumoto M Matsuo K Penninger JM Takayanagi H Yokota Y The cytokine RANKL produced by positively selected thymocytes fosters medullary thymic epithelial cells that express autoimmune regulator Immunity 2008 29 438 450 18799150 41 White AJ Nakamura K Jenkinson WE Saini M Sinclair C Seddon B Narendran P Pfeffer K Nitta T Takahama Y Caamano JH Lymphotoxin signals from positively selected thymocytes regulate the terminal differentiation of medullary thymic epithelial cells J Immunol 2010 185 4769 4776 20861360 42 Allam AH Charnley M Pham K Russell SM Developing T cells form an immunological synapse for passage through the beta-selection checkpoint J Cell Biol 2021 220 43 O’Neill KE Bredenkamp N Tischner C Vaidya HJ Stenhouse FH Peddie CD Nowell CS Gaskell T Blackburn CC Foxn1 Is Dynamically Regulated in Thymic Epithelial Cells during Embryogenesis and at the Onset of Thymic Involution PLoS One 2016 11 e0151666 26983083 44 Rode I Martins VC Kublbeck G Maltry N Tessmer C Rodewald HR Foxn1 Protein Expression in the Developing, Aging, and Regenerating Thymus J Immunol 2015 195 5678 5687 26538393 45 Barsanti M Lim JM Hun ML Lister N Wong K Hammett MV Lepletier A Boyd RL Giudice A Chidgey AP A novel Foxn1(eGFP/+) mouse model identifies Bmp4-induced maintenance of Foxn1 expression and thymic epithelial progenitor populations Eur J Immunol 2017 47 291 304 27861793 46 Janas ML Turner M Stromal cell-derived factor 1alpha and CXCR4: newly defined requirements for efficient thymic beta-selection Trends Immunol 2010 31 370 376 20829112 47 Murata S Sasaki K Kishimoto T Niwa S Hayashi H Takahama Y Tanaka K Regulation of CD8+ T cell development by thymus-specific proteasomes Science 2007 316 1349 1353 17540904 48 Jenkinson EJ Anderson G Owen JJ Studies on T cell maturation on defined thymic stromal cell populations in vitro J Exp Med 1992 176 845 853 1512547 49 Anderson G Jenkinson EJ Moore NC Owen JJ MHC class II-positive epithelium and mesenchyme cells are both required for T-cell development in the thymus Nature 1993 362 70 73 8446171 50 Auerbach R Grobstein C Inductive interaction of embryonic tissues after dissociation and reaggregation Exp Cell Res 1958 15 384 397 13597896

PMC009xxxxxx/PMC9797435.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101500303 35779 Biodemography Soc Biol Biodemography Soc Biol Biodemography and social biology 1948-5565 1948-5573 35892204 9797435 10.1080/19485565.2022.2104691 NIHMS1837588 Article The Intergenerational Transmission of Sexual Frequency Yabiku Scott T. * Newmyer Lauren Department of Sociology & Criminology, The Pennsylvania State University * Address correspondence to Scott T. Yabiku, Population Research Institute, The Pennsylvania State University, University Park, PA, 16802; phone: 814-863-0145; fax: 814-863-7216; sty105@psu.edu 24 9 2022 Jul-Dec 2022 27 7 2022 27 7 2023 67 3-4 175186 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Intergenerational relationships are one of the most frequently studied topics in the social sciences. Within the area of family, researchers find intergenerational similarity in family behaviors such as marriage, divorce, and fertility. Yet less research has examined the intergenerational aspects of a key proximate determinant of fertility: sexual frequency. We use the National Survey of Families and Households to examine the relationship between sexual frequency of parents and the sexual frequency of children when adults. We link parental sexual frequency in 1987/1988, when children were ages 5–18, to the sexual frequency of the children in 2001–2003 when these grown children were ages 18–34. We find a modest, yet significant association, between parental and adult children sexual frequency. A mechanism behind this association appears to be the higher likelihood of being in a union among children of parents with high sexual frequency. pmcIntroduction Intergenerational relationships are one of the most frequently studied topics in the social sciences. These links have been examined in domains such as occupational attainment (Biblarz et al., 1997; Korupp et al., 2002), health (Bowers & Yehuda, 2016; Goodman, 2020), and family behaviors. Within family behaviors, fertility has also found to be similar across generations (Kolk, 2014a; Lappegård & Thomson, 2018). Proposed mechanisms include the desire of children to reproduce their own family to that of their family of origin size (Kolk, 2014a), the transmission of family preferences and lifestyles (Barber, 2000; Lappegård & Thomson, 2018; Pouta et al., 2005), or the transmission of socioeconomic factors (such as education or wealth) that constrain or promote fertility similarly in both generations (Kolk, 2014b). Sexual intercourse is a necessary precursor to fertility, but sexual frequency remains less examined. Although Bongaarts (1978) argues that sexual frequency is one of the least important determinants of fertility, sexual frequency can play a role in childbearing, particularly in contexts below replacement-level fertility (Guzzo & Hayford, 2022; Rindfuss & Morgan, 1983). Studying sexual frequency is also important to better understand sexual and reproductive health (Blanc & Rutenberg, 1991; Gibbs et al., 2019) and sexually transmitted infections (Parish, Laumann, & Mojola, 2007). Additionally, sexual frequency is an important aspect of relationship quality (Dewitte & Mayer, 2018; Gager & Yabiku, 2010), sexual satisfaction (Higgins et al., 2011), and stability (Yabiku & Gager, 2009). Furthermore, research on between-person variance in sexual frequency remain incomplete, and research has not sufficiently accounted for parental factors. With this motivation our study asks the following questions: does the sexual frequency of a child’s parents predict the sexual frequency of the child? If so, what mechanisms may explain this intergenerational association? There is substantial theory and evidence to suggest that sexual frequency could have an intergenerational component. Contemporary research conceptualizes sexuality as something learned and modeled (Hogben & Byrne, 1998). Even by the time children are 3 or 4 years old, they have learned important issues relating to sexuality, such as touching, holding, and exposing of the body (Fausto-Sterling, 2019; Darling & Hicks, 1982). Most of this learning is not direct, but indirect and via non-verbal communication (Darling & Hicks, 1982; Yarber & Greer, 1986). Upchurch and colleagues (1999) argue that adolescent sexual activity is greatly influenced by the family because it is a primary location of social learning and role modeling. Although sexual behavior is a private activity, behaviors might still transmit intergenerationally through the transference of romantic relationship structures, sexual attitudes and views, and subtle cues. Past research has found that parental views on sexual activity may be more important to their children than their peers’ own sexual behaviors (Longmore et al., 2009). Parents may also guide, via indirect and direct pathways, their children’s relationship trajectories (Hiekel & Vidal, 2020; Manlove et al., 2012). Parent’s relationships with each other, their relationship to their children, their attitudes and views on sex, and the environment they offer their children may all shape their children’s sexual behaviors (Browning et al., 2004; Longmore et al., 2009; Manlove et al., 2012). Parents may also influence the sexual activity of their children through their own attitudes and views on sex. Research finds that the relationship between parents’ attitudes towards sex and their children’s sexual activity is positive (Davis & Friel, 2001). We propose three mechanisms through which sexual frequency may be transmitted from parents to adult children: union formation, fertility, and family attitudes. Although we focus on these three mechanisms, there are multiple other factors that might shape sexual frequency such as relationship quality, health status, age, state of residence, gender, and other important sociodemographic factors (Dewitte & Mayer, 2018; Tillman et al., 2019), but given our focus on parental impacts, we do not focus on these. Regarding union formation, if parental sexual frequency is high and adult children have been socialized to similar levels, then adult children may be more likely to enter marriages and cohabitations. We hypothesize this behavior because the highest sexual frequency is observed for married and cohabiting individuals because these people have ready access to a sexual partner (Yabiku & Gager, 2009). In addition, although pre-marital sex, defined as sex before marriage and withing an informal union, in the United States is common, there were still widespread norms against it when our data were collected (Horne, 2004). However, these norms may have weakened in modern times where premarital sex is even more common and discussed (Finer 2007, Graf 2019). Individuals may have a higher likelihood of entering a formal union due to this stigma around sexual activity. Parents’ own relationship might also extend onto their children’s sexual behavior (David & Friel, 2001). Another component is that children might replicate the union of their parents, which in turn might shape their sexual frequency. The second mechanism we theorize leads to the intergenerational transfer of sexual frequency is fertility. Although reproduction is not the only reason individuals engage in sexual frequency, it is often a motive (Gaskins et al., 2018). Adult children might engage in high sexual frequency to have a large family size, possibly to replicate that of their parents (Kolk, 2014a) In other words, people who have more children have more sexual intercourse or vice versa as higher sexual frequency can increase fecundity (Lorenz et al., 2015). Therefore, the adult children’s fertility may be the link between their sexual frequency and their parents’ sexual frequency. The third mechanism we hypothesize leads to this intergenerational transfer is family and sexual attitudes. Parents might transfer values that reinforce family formation, such as marital expectations. Because people in unions have higher sexual frequency, marital expectations that lead to marriage might explain higher sexual frequency in young adulthood. Parents may also shape their children’s views on sex which in turn shapes their sexual frequency. Sexual frequency has been linked with sexual self-schemas and ideologies ( Andersen et al., 1999). Parents influence the development of these attitudes in their children through both verbal and nonverbal direct and indirect communication (Darling & Hicks, 1982). Additionally, parents might shape their children’s views of sexual attitudes through the transmission of other related sociodemographic factors such as religion (Copen & Silverstein, 2007). Although our research does not investigate sexual schemas specifically, this may be a mechanism. Data and Methods Data The National Survey of Families and Households (NSFH) is a nationally representative study of households in the United States fielded in 1987/1988 (wave 1), 1992–1994 (wave 2), and 2001–2003 (wave 3). Although these data may not reflect contemporary sexual behaviors, the NSFH is the only nationally representative data available to examine these intergenerational associations. Contemporary intergenerational datasets (Add Health, NLSY) or older datasets with freshened samples (e.g., PSID) do not have these unique measures. The NSFH is the only nationally representative study that has measurement of parental sexual frequency early in the child’s life as well as the child’s own sexual frequency later as a young adult. In wave 1, respondents with children had one of their children randomly selected as a “focal child,” and additional information was collected about them. In wave 2, the focal children themselves were interviewed. In wave 3, when the focal children had become adults, they completed an extensive individual interview. Sexual frequency was measured in both the parent wave 1 interview and the adult child wave 3 interview. The research design and the temporal ordering of measurement is ideal for the study of the intergenerational transmission of sexual frequency. Although these data may not reflect contemporary sexual behaviors, the NSFH is the only nationally representative data available to examine these intergenerational associations Dependent Variable. Our dependent variable is a continuous measure of the adult child’s sexual frequency. The adult children were asked, “About how many times have you had sexual intercourse over the last 30 days?” This number varies from 0 to a maximum of 30 (the highest category was capped at 30 or more). Respondents who were not sexually active were assigned 0 times in the last 30 days. Independent Variables. The primary independent variable of interest is the parent’s sexual frequency, asked in the first wave of the NSFH in 1987/1988; question wording is similar to the adult child’s question. Other independent variables of interest are the proposed mechanisms linking parental and adult child sexual frequency: union formation, fertility, and marital expectations. Union formation is the adult child’s status at the most recent survey (2001–2003): married, single, or cohabiting. Fertility is the number of children the adult child has had, as of the 2001–2003 survey. Marital expectations, which is measured at NSFH 2 in 1992–1994, asked “How do you feel about getting married someday? Would you say that you definitely don’t want to, probably don’t want to, probably want to, or definitely want to get married?” This is recoded on a 1 to 4 scale, with higher values representing greater expectations for marriage. Although the question was asked slightly differently depending on if the child was a younger (age 10–17 at interview) or older (age 18–24 at interview) focal child, the measures are comparable. Control Variables. Parent controls include sex, age, education, family income, marital status, age at first marriage, race and ethnicity, and religion. All these variables were measured at the first wave of the NSFH in 1987/1988. In coding sex, females are assigned 1 and males 0. Age is continuous years, and education is four categories: less than high school, high school, some college, and college degree. Family income, which is measured in thousands, is the earnings of all the members of the household; it is logged to reduce skew. Marital status is coded into three categories: married, cohabiting, and not married (single, divorced, widowed, or separated). In the analysis, married is the reference group. Parental age at first marriage is coded in categories: never married, married before age 20, married age 20 up to 22, married age 22 up to 25, married age 25 up to 30, and marriage at age 30 and up. Parental age at first marriage is an important control because one of the proposed mechanisms – adult children’s union formation – is likely to be correlated with their parent’s union formation history, e.g., early marriage. Parental marital status, as of the time of NSFH wave 1, is measured in three categories: married, cohabiting, and not married. The not married category has a mix of single, divorced/separated, and widowed. Thus, the marital status category of “not married” is not collinear with the “never married” category in the parental age at first marriage variable. Note that even if a parent is not married, it is still possible for children to receive norms related to sexual behavior because children are often aware of their parents’ dating behaviors (Ferguson & Dickson, 1995). Race and ethnicity are coded as non-Hispanic White, Black, Hispanic, and other. Religion coding follows the categories outlined by Lehrer & Chiswick (1993) and includes mainline Protestant, conservative Protestant, Catholic, Jewish, Mormon, other religion, or no religion. At the child level, controls include sex, age in 2001–2003, whether the child pursued any education after high school (1 if higher education was pursued, 0 otherwise), and the child’s number of siblings. Because the adult child sample is as young as 18, the pursuit of higher education is a more appropriate measure of schooling. Race, ethnicity, and religion are not controlled at the child level due to their strong correlation with parents. Because race, ethnicity, and religion are controlled at the parent level, it is likely that this variation is accounted for in these parent variables. For both parents and children, we control for seasonality of interview. A dummy variable indicates if the interview was in a warm month (April through September) or a cool month. This approach to seasonality was based on prior research that indicated sexual frequency varied by temperature (Demir, Uslu & Arslan, 2016). Methods We use linear regression to estimate the relationships between the predictor variables and adult children’s sexual frequency (as a robustness check, we also categorized children’s sexual frequency as ordinal – low, medium, and high – and observed a similar pattern of ordinal logistic regression results to what is presented here). To address missing data, we use multiple imputation (Allison, 2002). We created 20 imputations, and the analysis of each imputation is combined into the final results. In our modeling strategy, we first estimate a model with control variables and our primary predictor of interest: parental sexual frequency. Parental sexual frequency is measured at NSFH wave 1, and thus is causally prior to the adult child’s sexual activity. We then add, one at a time, potential intervening mechanisms that might explain this relationship: adult children’s union formation, family attitudes, and fertility. These adult child activities happen after NSFH wave 1, but before NSFH wave 3, and thus they intervene between parental and adult children’s sexual frequency. If coefficients for parental sexual frequency decrease when these adult child variables are added, it is consistent with the explanation that these variables explain or mediate the relationship. Results Table 1 shows the descriptive statistics of the sample. We note a few key patterns. The mean sexual frequency in the parental and adult child generation is similar: 7.5 times in the past 30 days for the adult children, and 7.1 times for the parents. Only about 1/3 of the adult children had entered marriage, yet almost ¾ of the parents were currently married at NSFH wave 1. Age at parental first marriage was early – slightly over half of the parents married by age 22. This seems young, but these parents, on average, were born in 1949. In the early 1970s, median age at marriage was 23 for men, and 21 for women (United States Census Bureau, 2019). By the time of NSFH wave 3, the average fertility of the child generation was less than 1. Marital expectations when the children were younger, at NSFH wave 2, averaged 3.4 on the 1 to 4 scale, with higher values being more positive towards marriage. When younger, these adult children overall expected to be married someday. Table 2 presents the multivariate results. The dependent variable for the models in Table 2 is the adult child’s sexual frequency, which was measured in 2001–2003 (NSFH wave 3). Model 1 tests if there is an overall association between parental and adult child sexual frequency, after including typical sociodemographic and relevant controls. The results suggest that there is a significant intergenerational relationship. For each increase in the parent’s frequency of sexual intercourse in the past 30 days, we find an associated increase of the adult child’s frequency of sexual intercourse by .08 times (p<0.01). Clearly, the magnitude of this coefficient is not large, but we stress the very long-time span separating these two measures: as much as 15 years. The fact that any significant relationship, even if small, is found across such a large time span is notable. In Model 2, union formation has an association with adult child sexual frequency. The coefficient of 5.8 (p<0.001) means that adult children who are married report almost 6 additional events of sexual intercourse in the past 30 days compared to single adult children. Cohabitation has an even bigger effect: cohabitors report 7.3 (p<0.001) more events of sexual intercourse compared to singles. Furthermore, it appears that union formation mediates a substantial portion of the intergenerational influence of parental sexual frequency. The coefficient for parental sexual frequency is no longer significant, and compared to Model 1, the magnitude of the coefficient is reduced by 38% (.048 is 38% less than .077). Formal tests of the average causal mediation effects (ACME) showed that cohabitation was a significant mediator (p<.05) and marriage approached significance as a mediator (p = .08). Model 3 adds the adult child’s fertility. Fertility has a significant relationship with sexual frequency: for each child ever born, there is an associated increase in the respondent’s sexual frequency in the past 30 days of about .9 events (p<0.001). This finding suggests that, as hypothesized, adults with higher fertility also have higher sexual frequency. Fertility, however, is not a mediator of parental sexual frequency. The coefficient for parent’s sexual frequency is significant and remains mostly unchanged in Model 3 compared to Model 1. In Model 4, we examine the child’s childhood marital expectations. Marital expectations, which was reported in childhood at NSFH wave 2, are significantly and positively associated (coefficient 0.75; p<0.05) with sexual frequency at NSFH wave 3. Children who reported greater expectations for marriage in 1992–1994 had higher sexual frequency in 2001–2003. Marital expectations, however, is not a mediator of parental sexual frequency, which remains significant in Model 4 and of nearly the same magnitude as it was in Model 1. Finally, Model 5 examines all three mechanisms in the same model. The coefficients for being in a union (marriage and cohabitation) are essentially unchanged in Model 5 as compared to Model 2. The coefficient for fertility is reduced in Model 5 and no longer significant. This result is likely due to fertility being correlated with being in a union (marriage or cohabitation). Thus, Model 5 shows that there is no independent association of the adult child’s fertility with sexual frequency, but rather it is union status that is the primary mediator. The coefficient for marital expectations is also insignificant in Model 5 and reduced in magnitude compared to Model 4, likely due to its correlation with the family formation variables (the bivariate correlation between marital expectations and being married at wave 3 was r =.14). In sum, the results suggest that, of the tested mechanisms, adult children’s union formation is the most important intergenerational mechanism linking the sexual frequency of parents and adult children. Discussion Although intergenerational relationships are frequently a topic of study, the intergenerational aspect of sexual frequency has received less attention. Our work examines this gap. We show that despite a time span of up to 15 years, there are significant associations between parental sexual frequency and adult child sexual frequency many years later. These associations exist even after controlling for relevant sociodemographic characteristics of parents and adult children. Although we cannot rule out spuriousness, a rich set of controls reduces this possibility. Genetic influence may be important, but this pathway is beyond the scope and data available in this analysis. We examined three mechanisms through which parents may transmit behaviors in sexual frequency: union formation, fertility, and family attitudes. Although all had significant associations with the adult children’s sexual frequency, only the children’s union formation is an intervening mechanism independent from the others. The evidence is consistent with an interpretation that parents with higher sexual frequency have adult children who enter unions at higher rates. Because individuals in unions have more ready access to sexual partners, these adult children then have higher sexual frequency than similar young adults not in a union. Our findings complement other intergenerational research that has noted strong correlations between parents and children in fertility outcomes (Lappegård & Thomson, 2018; Axinn et al., 1994). Additionally, as we find that the intergenerational transmission of sexual frequency is mediated by union formation, future research should further consider the causes and consequences of the intergenerational transmission of union formation. It is important to note that the relationship between parental and adult child sexual frequency exists even when controlling for parental age at marriage; thus, it does not appear that this relationship is simply due to an intergenerational transmission of parental marriage timing. Furthermore, as marriage continues to decline, it is possible that although younger generations might be less likely to form a formal union, they might still replicate their parents’ sexual frequency through a high likelihood of being in informal unions like cohabitation. Our findings should not be interpreted to represent current trends in sexual frequency which may differ due to recent declines in young adult sexual frequency (Lei & South, 2020). A limitation to our research is the age of our data source. However, these data allow us to assess this specific intergenerational phenomenon with uniquely timed self-reports of sexual frequency from both the parent and, later, his/her adult child. Another limitation is that sexuality and sexual behaviors are diverse, and a single measure such as sexual frequency provides an incomplete view. It is likely there are many dimensions of sexuality that have an intergenerational component, but our data cannot speak to these. Additionally, our analysis is unable to differentiate between whether it is parental tendencies towards sexual frequency that are transmitted, or rather if it is another parental behavior or background factor, such as union formation (although we control for parental marriage timing, which should capture some of the union formation mechanism). Future research should seek to disentangle these factors to better understand the intergenerational transmission of sexual frequency. As our work relates to broader research in sexuality, investigating the potential intergenerational transmission of other sexual behaviors and processes, such as sexual schemas, may lend more insight to sexual function and well-being. Our work suggests that there is merit in investigating the long-term influence of the intergenerational transmission of sexual behaviors. To better explore pathways and mechanisms, creative research designs are likely needed. These include high frequency measurement studies or other empirical designs that could provide more direct causal inference than traditional statistical methods, such as the use of instrumental variables, fixed effects, or exogenous shocks. Sibling data may also be especially insightful for isolating parental effects. Acknowledgements: This research was supported by the Population Research Institute at Penn State University, which is supported by an infrastructure grant by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2CHD041025), and by a Eunice Kennedy Shriver National Institute of Child Health and Human Development institutional predoctoral traineeship grant (T-32HD007514). Table 1: Descriptive Statistics Mean SD Adult Child’s Sexual Frequency (2001–2003) 7.50 8.16 Parent’s Sexual Frequency (1987/1988) 7.07 7.55 Adult Child Intervening Mechanisms  Currently Single .51 .50  Currently Married .34 .47  Currently Cohabiting .15 .36  Fertility (Children ever born) .68 1.02  Marital Expectations (1992–1994) 3.44 .66 Adult Child Controls  Female .54 .50  Age in 2001–2003 25.21 4.45  Education After Highschool (1=Yes) .68 .47  Number of siblings 2.66 2.14  Interview was during warm months .46 .50 Parent Controls  Female .65 .48  Age (1987/1988) 37.71 7.71  Education   Less than High School .11 .31   High School .42 .49   Some College .25 .43   College and above .22 .41  Family Income (1987/1988), thousands 38.15 4.92  Marital Status (1987/1988)   Married .72 .45   Cohabiting .04 .19   Not Currently Married .25 .43  Age at First Marriage   Never married .04 .20   Before 20 .35 .48   20 up to 22 .22 .42   22 up to 25 .22 .42   25 up to 30 .13 .34   30 and above .04 .19  Race & Ethnicity   Non-Hispanic White .80 .40   Black .14 .34   Hispanic .05 .22   Other .01 .10  Religion   Catholic .24 .42   Jewish .02 .14   Other .01 .11   Mainline Protestant .32 .47   Conservative Protestant .30 .46   Mormon .04 .19   None .08 .27  Interview was during warm months .73 .44 N=1952 Table 2: Relationships between Parental Sexual Frequency and Adult Child Sexual Frequency Model 1 Model 2 Model 3 Model 4 Model 5 Parent’s Sexual Frequency (1987/1988) 0.077 * 0.048 0.077 * 0.076 * 0.047 (0.036) (0.033) (0.036) (0.036) (0.033) Adult Child Intervening Mechanisms  Currently Married (ref=single) 5.814 *** 5.661 *** (0.470) (0.498)  Currently Cohabiting (ref=single) 7.253 *** 7.224 *** (0.578) (0.578)  Fertility (Children ever born) 0.857 *** 0.112 (0.217) (0.217)  Marital Expectations (1992–1994) 0.746 * 0.517 (0.342) (0.319) Adult Child Controls  Female (ref=male) −0.106 −0.860 * −0.330 −0.177 −0.924 * (0.396) (0.378) (0.400) (0.401) (0.384)  Age in 2001–2003 0.147 * −0.139 * 0.056 0.131 * −0.154 * (0.060) (0.061) (0.064) (0.060) (0.063)  Education After Highschool (1=Yes) −0.278 −0.074 −0.041 −0.316 −0.069 (0.451) (0.425) (0.457) (0.453) (0.433)  Number of siblings 0.069 0.016 0.052 0.064 0.010 (0.098) (0.092) (0.097) (0.098) (0.092)  Interview was during warm months 0.053 0.178 0.054 0.062 0.181 (0.489) (0.471) (0.489) (0.491) (0.472) Parent Controls  Female (ref=male) 0.135 0.282 0.295 0.123 0.283 (0.441) (0.417) (0.443) (0.439) (0.419)  Age (1987/1988) −0.075 * −0.052 −0.066 −0.073 * −0.050 (0.037) (0.034) (0.036) (0.037) (0.034)  Education   Less than High School 0.931 0.462 0.634 0.944 0.439 (0.833) (0.783) (0.829) (0.833) (0.784)   High School 0.459 0.206 0.353 0.498 0.224 (0.572) (0.539) (0.571) (0.570) (0.539)   Some College 0.090 −0.395 0.000 0.131 −0.371 (0.589) (0.556) (0.587) (0.588) (0.556)  Family Income (1987/1988), logged −0.386 −0.253 −0.335 −0.410 −0.264 (0.217) (0.205) (0.215) (0.217) (0.206)  Marital Status (1987/1988) (ref=married)   Cohabiting 0.474 0.396 0.393 0.570 0.453 (1.160) (1.099) (1.157) (1.159) (1.099)   Not Currently Married 0.339 0.249 0.371 0.303 0.225 (0.564) (0.532) (0.561) (0.569) (0.535)  Age at First Marriage (ref=never married)   Before 20 1.874 1.148 1.900 1.890 1.174 (1.095) (1.037) (1.092) (1.094) (1.038)   20 up to 22 1.446 1.047 1.548 1.450 1.071 (1.125) (1.064) (1.123) (1.123) (1.065)   22 up to 25 1.552 1.020 1.715 1.550 1.046 (1.164) (1.106) (1.161) (1.162) (1.105)   25 up to 30 1.047 0.832 1.226 0.999 0.822 (1.234) (1.170) (1.232) (1.231) (1.170)   30 and above 1.458 1.228 1.422 1.569 1.305 (1.683) (1.595) (1.677) (1.684) (1.599)  Race & Ethnicity (ref=Non-Hispanic White)   Black −1.530 * 0.163 −1.511 * −1.485 * 0.163 (0.672) (0.640) (0.669) (0.671) (0.641)   Hispanic 0.130 0.376 0.070 0.222 0.429 (0.938) (0.886) (0.933) (0.939) (0.888)   Other −1.406 −0.925 −1.344 −1.583 −1.043 (2.161) (2.050) (2.154) (2.160) (2.047)  Religion (ref=Catholic)   Jewish 2.786 2.501 2.904 * 2.580 2.367 (1.436) (1.348) (1.430) (1.430) (1.345)   Other −0.112 −0.163 −0.412 −0.110 −0.190 (1.835) (1.723) (1.828) (1.830) (1.721)   Mainline Protestant 1.774 ** 1.550 ** 1.710 ** 1.729 ** 1.516 ** (0.549) (0.515) (0.547) (0.552) (0.517)   Conservative Protestant 1.293 * 1.203 * 1.141 1.244 * 1.159 * (0.597) (0.562) (0.594) (0.596) (0.561)   Mormon −0.333 −0.664 −0.677 −0.500 −0.797 (1.049) (0.993) (1.046) (1.052) (0.994)   None 2.222 ** 2.239 ** 2.160 ** 2.298 ** 2.291 ** (0.823) (0.771) (0.819) (0.824) (0.771)  Interview was during warm months −0.275 −0.308 −0.327 −0.228 −0.281 (0.429) (0.405) (0.428) (0.429) (0.406) Intercept 4.580 * 8.607 *** 5.823 ** 2.461 7.171 *** (1.878) (1.795) (1.886) (2.078) (2.026) N 1952 1952 1952 1952 1952 References Allison PD (2002). Missing Data. SAGE Publications. Andersen BL , Cyranowski JM , & Espindle D (1999). Men’s Sexual Self-Schema. Journal of Personality and Social Psychology, 76 (4 ), 645–661.10234850 Axinn WG , Clarkberg ME , & Thornton A (1994). Family influences on family size preferences. Demography, 31 (1 ), 65–79.8005343 Barber JS (2000). Intergenerational Influences on the Entry into Parenthood: Mothers’ Preferences for Family and Nonfamily Behavior. Social Forces, 79 (1 ), 319–348. Biblarz TJ , Raftery AE , & Bucur A (1997). Family Structure and Social Mobility. Social Forces, 75 (4 ), 1319–1339. Blanc AK , & Rutenberg N (1991). Coitus and Contraception: The Utility of Data on Sexual Intercourse for Family Planning Programs. Studies in Family Planning, 22 (3 ), 162–176.1949099 Bongaarts J (1978). A Framework for Analyzing the Proximate Determinants of Fertility. Population and Development Review, 4 (1 ), 105–132. Bowers ME , & Yehuda R (2016). Intergenerational Transmission of Stress in Humans. Neuropsychopharmacology, 41 , 232–244.26279078 Browning CR , Leventhal T , & Brooks-Gunn J (2004). Neighborhood Context and Racial Differences in Early Adolescent Sexual Activity. Demography, 41 (4 ), 697–720.15622950 Copen CE , & Silverstein M (2008). The transmission of religious beliefs across generations: Do grandparents matter?. Journal of Comparative Family Studies, 39 (1 ), 59–71. Darling CA , & Hicks MW (1982). Parental Influence on Adolescent Sexuality: Implications for Parents as Educators. Journal of Youth and Adolescence, 11 (3 ), 231–245.12338446 Davis EC , & Friel LV (2001). Adolescent Sexuality: Disentangling the Effects of Family Structure and Family Context. Journal of Marriage and Family, 63 (3 ), 669–681. Dewitte M , & Mayer A (2018). Exploring the Link Between Daily Relationship Quality, Sexual Desire, and Sexual Activity in Couples. Archives of Sexual Behavior, 47 , 1675–1686.29497914 Fausto-Sterling A (2019). Gender/sex, sexual orientation, and identity are in the body: How did they get there?. The Journal of Sex Research, 56 (4–5 ), 529–555.30875248 Ferguson SM , & Dickson FC (1995). Children’s expectations of their single parents’ dating behaviors: A preliminary investigation of emergent themes relevant to single parent dating. Journal of Applied Communication Research 23 : 308–324. Finer LB (2007). Trends in premarital sex in the United States, 1954–2003. Public Health Reports, 122 (1 ), 73–78.17236611 Gager CT , & Yabiku ST (2010). Who has the time? The relationship between household labor time and sexual frequency. Journal of Family Issues, 31 (2 ), 135–163. Gaskins AJ , Sundaram R , Buck Louis GM , & Chavarro JE (2018). Predictors of Sexual Intercourse Frequency among Couples Trying to Conceive. Journal of Sexual Medicine, 15 (4 ), 519–528.29523477 Gibbs SE , Kusunoki Y , & Moreau C (2019). Sexual Activity and Weekly Contraceptive Discontinuation and Selection Among Young Adult Women in Michigan. The Journal of Sex Research, 56 (8 ), 977–984.30632833 Goodman SH (2020). Intergenerational Transmission of Depression. Annual Review of Clinical Psychology, 16 , 213–238. Graf N (2019). “Key findings on marriage and cohabitation in the U.S” Pew Research Center, https://www.pewresearch.org/fact-tank/2019/11/06/key-findings-on-marriage-and-cohabitation-in-the-u-s/ Guzzo KB , & Hayford SR (2020). Pathways to parenthood in social and family contexts: Decade in review, 2020. Journal of Marriage and Family, 82 (1 ), 117–144.34012172 Hiekel N , & Vidal S (2020). Childhood Family Structure and Complexity in Partnership Life Courses. Social Science Research, 87 , 102400. Higgins JA , Mullinax M , Trussell J , Davidson JK Sr , & Moore NB (2011). Sexual satisfaction and sexual health among university students in the United States. American Journal of Public Health, 101 (9 ), 1643–1654.21778509 Hogben M , & Byrne B (1998). Using Social Learning Theory to Explain Individual Differences in Human Sexuality. The Journal of Sex Research, 35 (1 ), 58–71. 10.1080/00224499809551917 Horne C (2004). Values and Evolutionary Psychology. Sociological Theory, 22 (3 ), 477–493. Kolk M (2014a). Multigenerational transmission of family size in contemporary Sweden. Population Studies, 68 (1 ), 111–129.23957693 Kolk M (2014b). Understanding Transmission of Fertility across Multiple Generations – Socialization or Socioeconomics? Research in Social Stratification and Mobility, 35 , 89–103. Korupp SE , Sanders K , & Ganzeboom HBG (2002). The intergenerational transmission of occupational status and sex-typing at children’s labour market entry. European Journal of Women’s Studies, 9 (1 ), 7–29. Lappegård T , & Thomson E (2018). Intergenerational Transmission of Multipartner Fertility. Demography, 55 , 2205–2228.30378006 Lehrer EL , & Chiswick CU (1993). Religion as a Determinant of Marital Stability. Demography, 30 (3 ), 385–404.8405605 Lei L , & South SJ (2020). Explaining the Decline in Young Adult Sexual Activity in the United States. Journal of Marriage and Family. 10.1111/jomf.12723 Longmore MA , Eng AL , Giordano PC , & Manning WD (2009). Parenting and Adolescents’ Sexual Initiation. Journal of Marriage and Family, 71 (November ), 969–982.20160871 Lorenz TK , Demas GE , & Heiman JR (2015). Interaction of menstrual cycle phase and sexual activity predicts mucosal and systemic humoral immunity in healthy women. Physiology & behavior, 152 , 92–98.26394125 Manlove J , Wildsmith E , Ikramullah E , Terry-humen E , & Schelar E (2012). Family environments and the relationship context of first adolescent sex: Correlates of first sex in a casual versus steady relationship. Social Science Research, 41 (4 ), 861–875.23017856 Parish WL , Laumann EO , & Mojola SA (2007). Sexual Behavior in China: Trends and Comparisons. Population and Development Review, 33 (4 ), 729–756. Pouta A , Järvelin M-R , Hemminki E , Sovio U , & Hartikainen A-L (2005). Mothers and Daughters: Intergenerational Patterns of Reproduction. European Journal of Public Health, 15 (2 ), 195–199.15941762 Rindfuss RR , & Morgan SP (1983). Marriage, Sex, and the First Birth Interval: The Quiet Revolution in Asia. Population and Development Review, 9 (2 ), 259–278. Tillman KH , Brewster KL , & Holway GV (2019). Sexual and Romantic Relationships in Young Adulthood. Annual Review of Sociology, 45 , 133–153. Upchurch Dawn M. , Aneshensel Carol S. , Sucoff Clea A. , and Levy-Storms Lené . 1999. “Neighborhood and Family Contexts of Adolescent Sexual Activity.” Journal of Marriage and Family 61 (4 ):920–33. Yabiku ST , & Gager CT (2009). Sexual Frequency and the Stability of Marital and Cohabiting Unions. Journal of Marriage and Family, 71 (November ), 983–1000. Yarber WL , & Greer JM (1986). The Relationship Between the Sexual Attitudes of Parents and their College Daughters’ or Sons’ Sexual Attitudes and Sexual Behavior. Journal of School Health, 56 (2 ), 68–72.3633366

PMC009xxxxxx/PMC9817215.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7503056 4435 J Am Chem Soc J Am Chem Soc Journal of the American Chemical Society 0002-7863 1520-5126 35833781 9817215 10.1021/jacs.2c05044 NIHMS1862043 Article A Base-Promoted Reductive Coupling Platform for the Divergent Defluorofunctionalization of Trifluoromethylarenes Wright Shawn E. Bandar Jeffrey S. * Department of Chemistry, Colorado State University, Fort Collins, Colorado 80523, United States * Corresponding Author: jeff.bandar@colostate.edu 2 1 2023 27 7 2022 14 7 2022 27 7 2023 144 29 1303213038 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. We report a trifluoromethylarene reductive coupling method that dramatically expands the scope of difluorobenzylic substructures accessible via C–F bond functionalization. Catalytic quantities of a Lewis base, in conjunction with a disilane reagent in formamide solvent, leads to the replacement of a single trifluoromethyl fluorine atom with a silylated hemiaminal functional group. The reaction proceeds through a difluorobenzyl silane intermediate that can also be isolated. Together, these defluorinated products are shown to provide rapid access to over 20 unique difluoroalkylarene scaffolds. Graphical Abstract pmcThe α,α-difluorobenzylic substructure (ArCF2R) is often studied in pharmaceutical and agrochemical research as a means to modulate bioavailability and metabolic stability, amongst other potential benefits of fluorine incorporation.1 A key feature of an aromatic difluoroalkyl substituent is the structural modularity possible via the R group, allowing further optimization of a compound’s desired properties. The challenge of exploring this chemical space lies in the lack of general methods to access derivatives from a single precursor, typically requiring the synthesis of a unique reagent for each target of interest.2 For example, carbonyl deoxyfluorination3,4 and cross-coupling5 reactions are commonly used to access such motifs, but these routes first require access to the carbonyl or RCF2X coupling partner, respectively.6 Therefore, a method that can access valuable α,α-difluorobenzylic frameworks in a diversifiable fashion could greatly accelerate investigation of this substructure.7 The C–F functionalization of trifluoromethylarenes is an ideal route to α,α-difluorobenzylic compounds due to the wide availability of trifluoromethylarenes and their prevalence in late-stage settings.8 The impact of such methodology hinges on the ability to access α,α-difluorobenzylic derivatives that reflect the structural diversity found in bioactive compounds (Figure 1).9 A major challenge for the single C–F substitution of a trifluoromethyl group is the fact that the C–F bonds become weaker as defluorination proceeds10, typically resulting in overfunctionalization.11 Early efforts to address this challenge using electrochemical12 and metal13 reducing conditions are either limited to simple trifluoromethylbenzenes or are unselective. Recent reports by König, Jui, and Gouverneur use photoredox catalysis to achieve monoselective C–F reduction and hydroalkylation on a wide range of trifluoromethylarenes.14 An alternative strategy reported by Young employs frustrated Lewis pairs to form C–F substituted pyridinium and phosphonium salts, primarily used as electrophilic difluorobenzylic reagents.15,16 Despite these impressive advancements, there is still the need for a unified method that accesses a greater breadth of α,α-difluorobenzylic substructures from trifluoromethylarenes. Our group recently reported a fluoride-initiated protocol for the selective defluoroallylation of trifluoromethylarenes using allyltrimethylsilane coupling partners (Figure 2a).17,18 While practical, this reaction can only access difluoroalkyl substituents that map onto the allyl coupling fragment. To address this limitation, we herein report the development of a base-initiated, silane-mediated, reductive coupling platform of trifluoromethylarenes (Figure 2b). This method expands the C–F transformations accessible from trifluoromethylarenes by providing a versatile silylated hemiaminal synthon that possesses the reactivity of both an aldehyde and an iminium ion. The identification of a difluorobenzyl silane as the key intermediate for the reductive coupling reaction also allowed for its isolation and use as a general difluorobenzylic pronucleophile. This work began with the goal of discovering a single silane reagent that couples with trifluoromethylarenes to generate synthetically versatile difluorobenzylic products. While investigating disilane19 coupling partners, we observed that catalytic activation of commercially available tris(trimethylsilyl)silane (TTMSS) with Lewis basic salts in DMF generates silylated hemiaminal adduct 2 (Scheme 1). Given that a silylated hemiaminal could potentially serve as a branching point for a wealth of derivatization reactions, we sought to further optimize this reaction. Notably, similar silylated hemiacetal and hemiaminal adducts were proposed as intermediates in Lalic’s dual Pd/Cu-catalyzed selective trifluoromethylarene reduction protocol that is conducted with triphenylsilane in DMF.20 Strong Lewis bases, such as fluoride and alkoxide salts, provide low yields of 2, while carboxylate salts lead to significantly higher yields (entries 1–3). Ultimately, we found 18-crown-6-ligated cesium formate to be the optimal catalyst system (76% yield, entry 4). The reaction proceeds in slightly lower yield without 18-crown-6 (63%, entry 5) or using tetrabutylammonium acetate (69%, entry 6). Other commercial disilanes are less effective at promoting this reaction (entries 7 and 8). Unfortunately, all attempts to isolate product 2 via chromatography resulted in decomposition. Evaluation of other formamides led to the identification of 4-formylmorpholine (4FM) in NMP (1:1 mixture) as a satisfactory substitute for DMF, providing chromatographically stable 3 in 66% isolated yield (Scheme 1b, entry 10). Benzotrifluoride can also be used as a cosolvent (entry 11) and, for some substrates, results in improved yields (vide infra). Table 1 shows representative trifluoromethylarenes that are amenable to the base-promoted coupling reaction. 1,3-Bis(trifluoromethyl)arenes are effective substrates, including when scaled to 10 mmol (3) or with free phenolic O–H (4) and terminal alkene (5) functional groups. Sulfonamide (6, characterized by X-ray crystallography) and phosphonate ester (7) aryl substituents also sufficiently activate the trifluoromethylarene towards functionalization. Heteroaryl and drug-like trifluoromethylarenes are also effective, including 2-, 3- and 4-trifluoromethylpyridines (8–11), a benzylated aprepitant precursor (12) and a fluoxetine-trifluoromethylpyrimidine derivative (13) that selectively couples at the electron-deficient heteroarene. Under the current reaction conditions, trifluoromethylbenzenes that lack an electron-withdrawing group do not react, while substrates with electrophilic functional groups (e.g. ketone) undergo competitive side reactions with the silane.21 We expected these silylated hemiaminal products to be versatile synthetic intermediates based on their resemblance to reported trifluoromethyl formamide adducts.22 The diversity of difluorobenzylic frameworks accessible from the silylated hemiaminal unit is demonstrated in Scheme 2, with sixteen transformations shown starting from product 3 (prepared on multigram scale). These reactions employ common reagents and require one purification step, with detailed procedures described in the Supporting Information (Section VII, pages S14–25). Numerous reactions can be conducted directly with the silylated hemiaminal (Scheme 2a), including cleavage to a deuterated difluoromethylarene (14), condensation to an oxime (15), condensation-dehydration to a nitrile (16), a Petasis-type styrenylation process (17), oxidation to an amide (18), reduction to a tertiary amine (19), and a Mannich-type addition reaction (20). Conversion of 3 into a hemiacetal is facile with catalytic acid in ethanol without the need for isolation.23,24 From this intermediate, many transformations are possible (Scheme 2b), including reduction (21), reductive amination (22), Wittig (23, 24), heterocycle condensation (25), Henry (26), oxidation (27), cyanide addition (28) and silylation (29) reactions. An additional application of this method is its use for late-stage C–F functionalization via sequenced reductive coupling and derivatization. A series of pharmaceutical derivatives and drug-like structures are shown in Scheme 3 that underwent defluorofunctionalization using one total purification step. This includes trifluoromethylaryl derivatives of apriprazole (30 and 31), fluoxetine (34) and bepotastine (35), as well as an aprepitant precursor (32 and 33) and a trifluoromethylquinoline substrate (36). These examples demonstrate the ability to modify trifluoromethyl substituents of bioactive compounds, as well as the ability to carry the typically inert trifluoromethyl group through multistep syntheses before derivatization. Insight into the mechanism of this reductive coupling process first came while varying the reaction conditions. We observed the product identity to be dependent on the solvent used; in NMP, the major product is difluorobenzylsilane 37, and in MeCN, the major product is difluoromethylarene 38 (Figure 3a).25 We reasoned that formation of the benzylsilane (37) and subsequent in situ base-promoted desilylation could explain the solvent dependence.26 Subjection of benzylsilane 37 to cesium formate in MeCN or DMF provides difluoromethylarene 38 or silylated hemiaminal 2, respectively (Figure 3b).27 A profile of the model reaction in DMF shows the concurrent formation of benzylsilane 37 and the silylated hemiaminal 2, and once the trifluoromethylarene has been consumed, the remaining benzylsilane is converted to the silylated hemiaminal (Figure 3c). These observations support defluorosilylation as the key process en route to the formamide addition product 2. Each reaction of this sequence generates an anion (fluoride or oxyanion) that could regenerate the formate anion or propagate silane activation via an anionic chain process, explaining why only catalytic quantities of formate salt are required (Figure 3d).28 Defluorosilylation likely occurs via initial formation of a silicate29 or silyl anion from TTMSS30,31, and we have obtained evidence that both TMS and HSi(TMS)2 anions may be generated under the reaction conditions.32,33 As silyl anions are known to be potent reductants34, bases35 and halophilic nucleophiles36, numerous mechanistic pathways for defluorosilylation seem plausible. Interestingly, when TTMSS is replaced with other disilane reagents (e.g. Si2Me6 or Si(TMS)4) for the model reaction, consumption of the trifluoromethylarene is observed but with little formation of the hemiaminal product.37 These comparisons indicate TTMSS strikes the right balance of Lewis acidity and capability of silyl anion generation to mediate the selective reductive coupling reaction. Details of these control studies and a discussion of possible pathways for the initiation of this reaction are provided in the Supporting Information. Investigations are underway to gain more insight into the defluorosilylation process and to identify disilanes that can activate a wider scope of trifluoromethylarenes.38,39 The discovery of a defluorosilylation pathway provides an opportunity to expand the scope of accessible C–F coupling products. Use of the α,α-difluorobenzylsilane as a masked carbanion can access derivatives that are challenging to prepare from the hemiaminal intermediates.26 The model defluorosilylation product 37 was first isolated from a reaction conducted in NMP on a 5 mmol scale in 40% yield. From 37, our recently reported fluoride-promoted protocol for benzylsilane cross-coupling to aryl nitriles can be used to generate defluoroarylation products (Scheme 4a).40 This route provides an alternative to Zhang’s recently developed light-promoted Pd-catalyzed trifluoromethylarene C–F arylation method.41 We also sought to show how defluorosilylation could serve as an entry to assembling difluoroalkylarene libraries with minor structural differences (Scheme 4b). Fluoride-activation of 37 promotes facile substitution with alkyl iodides, providing the defluoromethylation product (41), its isotopologues (42 and 43), and the ethyl derivative (44). Substitution using Togni reagent II42 provides pentafluoroethyl product 45, thus accomplishing a net extension of a trifluoromethyl substituent into a pentafluoroethyl group. In summary, this reductive coupling platform expands the scope of α,α-difluorobenzylic substructures accessible from trifluoromethylarenes to better reflect the structural diversity found in bioactive compounds (Figure 1). The reaction leverages the continuous generation of anionic intermediates to propogate a disilane-mediated defluorosilylation and formamide addition sequence. This ensemble allows a trifluoromethyl C–F bond to formally serve as a masked nucleophile, thus delivering new difluoroalkylarene synthetic linchpins. Supplementary Material Supporting Info ACKNOWLEDGMENT Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under award number R35GM138350. The content is solely the responsibility of the authors and does not necessarily reflect the official views of the National Institutes of Health. We are grateful for support from Colorado State University (CSU) and the Research Corporation for Science Advancement (Cottrell Scholar Award for J.S.B). We thank the Analytical Resources Core (ARC, RRID: SCR_021758) at CSU for instrument access, training and assistance with sample analysis. We also thank Dr. Brian S. Newell of the CSU ARC for performing X-ray diffraction analyses on compounds 6 and 37. Figure 1. Goals for defluorofunctionalization methodology. Figure 2. Overview of base-promoted ArCF3 functionalization. Figure 3. Studies and observations into reaction mechanism. Yields determined by 1H or 19F NMR spectroscopy. a Under alternative conditions, the yield of 38 is 86% if the reaction is conducted at 80 °C and 80% if CsF is used at rt in place of HCO2Cs. Scheme 1. Development of ArCF3 reductive coupling reaction. a Yields determined by 1H NMR spectroscopy. b Yields in parentheses are isolated yields by column chromatography. Scheme 2. Synthetic utility of hemiaminal adduct.a a Isolated product yields; see Supporting Information for full synthetic details for each derivatization reaction. Scheme 3. Divergent late-stage ArCF3 C–F functionalization.a a Isolated product yields starting from trifluoromethylarene; see Supporting Information for full synthetic details for each entry. Scheme 4. Isolation and utility of α,α-difluorobenzylsilane.a a Isolated product yields; see Supporting Information for full synthetic details for each entry. b 53% 19F NMR yield; isolated yield reduced due to coelution with protodesilylated compound 38. c 76% 1H NMR yield; isolated yield reduced due to coelution with protodesilylation side product. Table 1. Substrate Scope of Trifluoromethylarenes.a a Yields are of purified product on a 0.25–1 mmol scale; see Supporting Information for full details. b Isolated as adduct with NEt3. c PhCF3 used in place of NMP. d Reaction conducted at 80 °C. e Additional base (20 mol%) and TTMSS (1.2 equiv) added after 16 h. ASSOCIATED CONTENT Supporting Information The Supporting Information is available free of charge on the ACS Publications website. Experimental procedures and characterization data for all compounds (PDF). REFERENCES (1) For reviews on the general properties of fluorine incorporation in medicinal and agrochemical compounds, see: (a) Gillis EP ; Eastman KJ ; Hill MD ; Donnelly DJ ; Meanwell NA Applications of Fluorine in Medicinal Chemistry. J. Med. Chem 2015, 58 , 8315–8359.26200936 (b) Meanwell N , Fluorine A and Fluorinated Motifs in the Design and Application of Bioisosteres for Drug Design. J. Med. Chem 2018, 61 , 5822–5880.29400967 (c) Johnson BM ; Shu Y-Z ; Zhuo X ; Meanwell NA Metabolic and Pharmaceutical Aspects of Fluorinated Compounds. J. Med. Chem 2020, 63 , 6315–6386.32182061 (d) Jeschke P . Latest generation of halogen-containing pesticides. Pest Manage. Sci 2017, 73 , 1053–1066. (2) For selected reviews on preparing difluorobenzylic compounds, see: (a) Belhomme D-C ; Besset T ; Poisson T ; Pannecoucke X Recent Progress toward the Introduction of Functionalized Difluoromethylated Building Blocks onto C(sp2) and C(sp) Center. Chem. Eur. J 2015, 21 , 12836–12865.26178870 (b) Carvalho DR ; Christian AH Modern approaches towards the synthesis of geminal difluoroalkyl groups. Org. Biomol. Chem 2021, 19 , 947–964.33406177 (3) Ni C ; Hu M ; Hu J Good Partnership between Sulfur and Fluorine: Sulfur-Based Fluorination and Fluoroalkylation Reagents for Organic Synthesis. Chem. Rev 2015, 115 , 765–825.25144886 (4) Benzylic C–H fluorination is emerging as an alternative to carbonyl deoxyfluorination; see: (a) Koperniku A ; Liu H ; Hurley PB Mono- and Difluorination of Benzylic Carbon Atoms. Eur. J. Org. Chem 2016, 871–886. (b) Meanwell M ; Britton R Synthesis of Heterobenzylic Fluorides. Synthesis 2018, 50 , 1228–1236. (c) Szpera R ; Moseley DFJ ; Smith LB ; Sterling AJ ; Gouverneur V The Fluorination of C–H Bonds: Developments and Perspectives. Angew. Chem. Int. Ed 2019, 58 , 14824–14848. (5) For selected reviews and examples on cross-coupling reactions using functionalized difluoroalkyl reagents, see: (a) Feng Z ; Xiao Y-L ; Zhang X Transition-Metal (Cu, Pd, Ni)-Catalyzed Difluoroalkylation via Cross-Coupling with Difluoroalkyl Halides. Acc. Chem. Res 2018, 51 , 2264–2278.30132322 (b) Dong D-Q ; Yang H ; Shi J-L ; Si W-J ; Wang Z-L ; Xu X-M Promising reagents for difluoroalkylation. Org. Chem. Front 2020, 7 , 2538–2575. (c) Xiao Y-L ; Min Q-Q ; Xu C ; Wang R-W ; Zhang X Nickel-Catalyzed Difluoroalkylation of (Hetero)Arylborons with Unactivated 1-Bromo-1,1-difluoroalkanes. Angew. Chem., Int. Ed 2016, 55 , 5837–5841. (d) Ge S ; Arlow SI ; Mormino MG ; Hartwig JF Pd-Catalyzed α-Arylation of Trimethylsilyl Enolates of α,α-Difluoroacetamides. J. Am. Chem. Soc 2014, 136 , 14401–14404.25254966 (6) For reviews and examples of radical-based routes to difluorobenzylic products, see ref 2b and (a) Lemos A ; Lemaire C ; Luxen A Progress in Difluoroalkylation of Organic Substrates by Visible Light Photoredox Catalysis. Adv. Synth. Catal 2019, 361 , 1500–1537.31680791 (b) Zhou Q ; Ruffoni A ; Gianatassio R ; Fujiwara Y ; Sella E ; Shabat D ; Baran PS Direct Synthesis of Fluorinated Heteroarylether Bioisosteres. Angew. Chem. Int. Ed 2013, 52 , 3949–3952. (c) Douglas JJ ; Albright H ; Sevrin MJ ; Cole KP ; Stephenson CRJ A Visible-Light-Mediated Radical Smiles Rearrangement and its Application to the Synthesis of a Difluoro-Substituted Spirocyclic ORL-1 Antagonist. Angew. Chem. Int. Ed 2015, 54 , 14898–14902. (7) For recent methods that functionalize aryldifluoromethyl reagents, see: (a) Geri JB ; Wade Wolfe MM ; Szymczak NK The Difluoromethyl Group as a Masked Nucleophile: A Lewis Acid/Base Approach. J. Am. Chem. Soc 2018, 140 , 9404–9408.30040403 (b) Wade Wolfe MM ; Shanahan JP ; Kampf JW ; Szymczak NK Defluorinative Functionalization of Pd(II) Fluoroalkyl Complexes. J. Am. Chem. Soc 2020, 142 , 18698–18705.33073563 (c) Santos L ; Panossian A ; Donnard M ; Vors J-P ; Pazenok S ; Bernier D ; Leroux FR Deprotonative Functionalization of the Difluoromethyl Group. Org. Lett 2020, 22 , 8741–8745.33089999 (8) (a) Jaroschik F Picking One out of Three: Selective Single C–F Activation in Trifluoromethyl Groups. Chem. Eur. J 2018, 24 , 14572–14582.29923644 (b) Yan G ; Qiu K ; Guo M Recent advance in the C–F bond functionalization of trifluoromethyl-containing compounds. Org. Chem. Front 2021, 8 , 3915–3942. (c) Zhao F ; Zhou W ; Zuo Z Recent Advances in the Synthesis of Difluorinated Architectures from Trifluoromethyl Groups. Adv. Synth. Catal 2022, 364 , 234–267. (9) For the full structures and references for the compounds shown in Figure 1, see page S56 of the Supporting Information file. (10) For discussions and values of C–F bond strengths, see: (a) Luo YR Comprehensive Handbook of Chemical Bond Energies; CRC Press: Boca Raton, FL, 2007. (b) O’Hagan D Understanding organofluorine chemistry. An introduction to the C–F bond. Chem. Soc. Rev 2008, 37 , 308–319.18197347 (c) Wiberg KB ; Rablen PR Origin of the stability of carbon tetrafluoride: negative hyperconjugation reexamined. J. Am. Chem. Soc 1993, 115 , 614–625. (11) For selected examples and reviews of multiple C–F functionalization reactions, see: (a) Saito K ; Umi T ; Yamada T ; Suga T ; Akiyama T Niobium(V)-catalyzed defluorinative triallylation of α,α,α-trifluorotoluene derivatives by triple C−F bond activation. Org. Biomol. Chem 2017, 15 , 1767–1770.28139811 (b) Gu W ; Haneline MR ; Douvris C ; Ozerov OV Carbon−Carbon Coupling of C(sp3)−F Bonds Using Alumenium Catalysis. J. Am. Chem. Soc 2009, 131 , 11203–11212.19722677 (c) Zhu J ; Pérez M ; Caputo CB ; Stephan DW Use of Trifluoromethyl Groups for Catalytic Benzylation and Alkylation with Subsequent Hydrodefluorination. Angew. Chem. Int. Ed 2016, 55 , 1417–1421. (d) Stahl T ; Klare HFT ; Oestreich M Main-Group Lewis Acids for C–F Bond Activation. ACS Catal 2013, 3 , 1578–1587. (e) Shen Q ; Huang Y-G ; Liu C ; Xiao J-C ; Chen Q-Y ; Guo Y Review of recent advances in C−F bond activation of aliphatic fluorides. J. Fluor. Chem 2015, 179 , 14–22. (12) (a) Saboureau C ; Troupel M ; Sibille S ; Périchon J Electroreductive Coupling of Trifluoromethylarenes with Electrophiles: Synthetic Applications. J. Chem. Soc., Chem. Commun 1989, 1138–1139. (b) Clavel P ; Léger-Lambert M-P ; Biran C ; Serein-Spirau F ; Bordeau M ; Roques N ; Marzouk H Selective Electrosynthesis of (Trimethylsilyldifluoro)methylbenzene, a PhCF2− Precursor; Conditions for a Molar Scale Preparation without HMPA. Synthesis 1999, 5 , 829–834. (c) Clavel P ; Lessene G ; Biran C ; Bordeau M ; Roques N ; Trévin S ; de Montauzon D Selective electrochemical synthesis and reactivity of functional benzylic fluorosilylsynthons. J. Fluor. Chem 2001, 107 , 301–310. (13) (a) Utsumi S ; Katagiri T ; Uneyama K Cu-deposits on Mg metal surfaces promote electron transfer reactions. Tetrahedron 2012, 68 , 1085–1091. (b) Utsumi S ; Katagiri T ; Uneyama K Mg–Cu bimetal system for selective C–F bond activation. J. Fluor. Chem 2013, 152 , 84–89. (c) Munoz SB ; Ni C ; Zhang Z ; Wang F ; Shao N ; Mathew T ; Olah GA ; Prakash GKS Selective Late-Stage Hydrodefluorination of Trifluoromethylarenes: A Facile Access to Difluoromethylarenes. Eur. J. Org. Chem 2017, 2017 , 2322–2326. (14) For photoredox-catalyzed trifluoromethylarene C–F functionalization methods, see: (a) Chen K ; Berg N ; Gschwind R ; König B Selective Single C(sp3)−F Bond Cleavage in Trifluoromethylarenes: Merging VisibleLight Catalysis with Lewis Acid Activation. J. Am. Chem. Soc 2017, 139 , 18444–18447.29182283 (b) Wang H ; Jui NT Catalytic Defluoroalkylation of Trifluoromethylaromatics with Unactivated Alkenes. J. Am. Chem. Soc 2018, 140 , 163–166.29256607 (c) Vogt DB ; Seath CP ; Wang H ; Jui NT Selective C−F Functionalization of Unactivated Trifluoromethylarenes. J. Am. Chem. Soc 2019, 141 , 13203–13211.31369264 (d) Sap JBI ; Straathof NJW ; Knauber T ; Meyer CF ; Médebielle M ; Buglioni L ; Genicot C ; Trabanco AA ; Noël T ; am Ende CW ; Gouverneur V Organophotoredox Hydrodefluorination of Trifluoromethylarenes with Translational Applicability to Drug Discovery. J. Am. Chem. Soc 2020, 142 , 9181–9187.32379965 For additional recent photoredox-based methods, see: (e) Yuan X ; Zhuang K-Q ; Cui Y-S ; Qin L-Z ; Sun Q ; Duan X ; Chen L ; Zhu N ; Li G ; Qui J-K ; Guo K Photocatalytic radical defluoroalkylation of unactivated alkenes via distal heteroaryl ipso-migration. Commun. Chem 2020, 3 , 98. (f) Liu C ; Shen N ; Shang R Photocatalytic defluoroalkylation and hydrodefluorination of trifluoromethyls using o-phosphinophenolate. Nat. Commun 2022, 13 , 354.35039496 (g) Liu C ; Li K ; Shang R Arenethiolate as a Dual Function Catalyst for Photocatalytic Defluoroalkylation and Hydrodefluorination of Trifluoromethyls. ACS Catal 2022, 12 , 4103–4109. (15) (a) Mandal D ; Gupta R ; Jaiswal AK ; Young RD Frustrated Lewis-Pair-Mediated Selective Single Fluoride Substitution in Trifluoromethyl Groups. J. Am. Chem. Soc 2020, 142 , 2572–2578.31935080 (b) Gupta R ; Mandal D ; Jaiswal AK ; Young RD FLP-Catalyzed Monoselective C–F Functionalization in Polyfluorocarbons at Geminal or Distal Sites. Org. Lett 2021, 23 , 1915–1920.33624500 (16) For monoselective C–F functionalization reactions that are facilitated by an ortho-silylium ion, see: (a) Yoshida S ; Shinomori K ; Kim Y ; Hosoya T Single C–F Bond Cleavage of Trifluoromethylarenes with an ortho-Silyl Group. Angew. Chem. Int. Ed 2016, 55 , 10406–10409. (b) Idogawa R ; Kim Y ; Shinomori K ; Hosoya T ; Yoshida S Single C–F Transformations of o-Hydrosilyl Benzotrifluorides with Trityl Compounds as All-in-One Reagents. Org. Lett 2020, 22 , 9292–9297.33180503 (c) Kim Y ; Kanemoto K ; Shinomori K ; Hosoya T ; Yoshida S Functionalization of a Single C–F Bond of Trifluoromethylarenes Assisted by an ortho-Silyl Group Using a Trityl-Based All-in-One Reagent with Ytterbium Triflate Catalyst. Chem. Eur. J 2020, 26 , 6136–6140.32185819 (17) Luo C ; Bandar JS Selective Defluoroallylation of Trifluoromethylarenes. J. Am. Chem. Soc 2019, 141 , 14120–14125.31437395 (18) For recently reported alternative methods for defluoroallylation of perfluoroalkylarenes, see: (a) Sugihara N ; Suzuki K ; Nishimoto Y ; Yasuda M Photoredox-Catalyzed C–F Bond Allylation of Perfluoroalkylarenes at the Benzylic Position. J. Am. Chem. Soc 2021, 143 , 9308–9313.34075740 (b) Zhou F-Y ; Jiao L Asymmetric Defluoroallylation of 4-Trifluoromethylpyridines Enabled by Umpolung C−F Bond Activation. Angew. Chem. Int. Ed 2022, 61 , e202201102. (19) We herein refer to any compound with a silicon-silicon bond as a disilane. (20) Dang H ; Whittaker AM ; Lalic G Catalytic activation of a single C–F bond in trifluoromethylarenes. Chem. Sci 2016, 7 , 505–509.29861997 (21) For trifluoromethylbenzenes, electron-withdrawing groups in the meta positions lead to the highest yields, although electron-withdrawing groups at any position are sufficient for activation. For example, 1,2- and 1,4-bis(trifluoromethyl)benzenes undergo reductive coupling at room temperature or 50 °C in moderate yields. Additionally, 2-, 3-, and 4-trifluoromethyl substituted pyridines perform similarly as shown in Table 1. Benzotrifluoride does not react under the standard reaction conditions. A comparison of these electronic effects is documented in the Supporting Information on page S7. (22) For selected studies on formamide adducts of fluoroform and their synthetic utility that inspired the derivatization reactions in this work, see: (a) Billard T ; Bruns S ; Langlois BR New Stable Reagents for the Nucleophilic Trifluoromethylation. 1. Trifluoromethylation of Carbonyl Compounds with N-Formylmorpholine Derivatives. Org. Lett 2000, 2 , 2101–2103.10891240 (b) Blond G ; Billard T ; Langlois BR Reactivity of Stable Trifluoroacetaldehyde Hemiaminals. 1. An Unexpected Reaction with Enolizable Carbonyl Compounds. J. Org. Chem 2001, 66 , 4826–4830.11442412 (c) Billard T ; Langlois BR Reactivity of Stable Trifluoroacetaldehyde Hemiaminals. 2. Generation and Synthetic Potentialities of Fluorinated Iminiums. J. Org. Chem 2002, 67 , 997–1000.11856050 (23) For representative references using fluorinated hemiacetals as synthetic intermediates, see (a) Tsukamoto T ; Kitazume T Difluorinated Maloaldehyde Derivatives as Useful Difluoromethylene-containing Building Blocks. J. Chem. Soc., Chem. Commun 1992, 540–541. (b) Tsukamoto T ; Kitazume T Preparation and Reaction of Difluorinated Malonaldehydic Acid Derivatives: a New Route to Functionalized ⍺,⍺-Difluorinated Esters and Amides. J. Chem. Soc., Perkin Trans 1 , 1993 , 1177–1181. (c) Kubota T ; Iijima M ; Tanaka T Facile Synthesis of ⍺-Trifluoromethylated Alcohols from Trifluoroacetaldehyde Ethyl Hemiacetal. Tetrahedron Lett 1992, 33 , 1351–1354. (24) Attempts to prepare and isolate the corresponding aldehyde were unsuccessful due to its instability. This instability mirrors that of related compounds, such as trifluoroacetaldehyde. For a discussion of trifluoroacetaldehyde instability, see: Lin P Trifluoroacetaldehyde. In Encyclopedia of Reagents for Organic Synthesis; Wiley, 2003. DOI: 10.1002/047084289X.rn00213. (25) Defluorosilylation has been reported under electrochemical and metal reducing conditions for several trifluoromethylbenzenes; see refs 12b-c and 13a-b. (26) For alternative routes to difluorobenzylic silanes and their use as nucleophilic reagents: (a) see ref 23. (b) Aikawa K ; Maruyama K ; Nitta J ; Hashimoto R ; Mikami K Siladifluoromethylation and Difluoromethylation onto C(sp3), C(sp2), and C(sp) Centers Using Ruppert–Prakash Reagent and Fluoroform. Org. Lett 2016, 18 , 3354–3357.27340753 (c) Guidotti J ; Metz F ; Tordeux M ; Wakselman C Preparation of (Phenyldifluoromethyl)- and (Phenoxydifluoromethyl)-silanes by Magnesium-Promoted Carbon-Chlorine Bond Activation. Synlett 2004, 10 , 1759–1762. (27) Because the yield of 38 is low under the “standard conditions” for the experiment shown in Figure 3b, we conducted additional control experiments that could mimic the conditions typical of the defluorofunctionalization reactions in Figure 3a. Thus, use of catalytic CsF as a base, which is formed as an intermediate under the defluorosilylation reaction conditions, promotes the desilylation reactions shown in Figure 3b in high yield. These results are shown as footnote a in Figure 3 and described in the Supporting Information. (28) Mixing fluoride salts with trimethylsilylated carboxylates (RCO2TMS) rapidly releases the carboxylate anion and forms fluorotrimethylsilane. This observation, and the fact that use of fluoride as an initiator for the model reaction (Scheme 1, entry 1) leads to low yields, leads us to hypothesize that cesium formate is continuously generated under the reaction conditions and is acting as a true catalyst. (29) For reviews on base activation of organosilanes, see: (a) Chuit C ; Corriu RJP ; Reye C ; Young JC Reactivity of penta- and hexacoordinate silicon compounds and their role as reaction intermediates. Chem. Rev 1993, 93 , 1371–1448. (b) Reich HJ Mechanism of C–Si Bond Cleavage Using Lewis Bases (n → σ*). In Lewis Base Catalysis in Organic Synthesis, Vol. 1 ; Wiley-VCH, 2016; pp 233–280. (c) García-Domínguez A ; Leach AG ; Llyod-Jones GC In Situ Studies of Arylboronic Acids/Esters and R3SiCF3 Reagents: Kinetics, Speciation, and Dysfunction at the Carbanion–Ate Interface. Acc. Chem. Res 2022, 55 , 1324–1336.35435655 (30) For reviews on silyl anion generation and reactivity, see: (a) Lickiss P D; Smith, C. M. Silicone derivatives of the metals of groups 1 and 2. Coord. Chem. Rev 1995, 145 , 75–124. (b) Tamao K ; Kawachi A Silyl Anions. In Advances in Organometallic Chemistry, Vol. 38 ; Academic Press, 1995, 38, pp 1–58. (c) Marschner C Chapter 7 – Silicon-Centered Anions. In Organosilicon Compounds, Vol. 1 ; Academic Press, 2017, pp 295–360. (31) For reviews on the use of TTMSS, see: (a) Giese B ; Dickhaut J ; Chatgilialoglu C ; Wu X ; Zhang Z Tris(trimethylsilyl)silane. In Encyclopedia of Reagents for Organic Synthesis; Wiley, 2022. Updated January 28, 2022. DOI: 10.1002/047084289X.rt420.pub3. (b) Chatgilialoglu C ; Ferreri C ; Landais Y ; Timokhin VI Thirty Years of (TMS)3SiH: A Milestone in Radical-Based Synthetic Chemistry. Chem. Rev 2018, 118 , 6516–6572.29938502 (32) It is well known that Lewis base activation of tetrakis(trimethylsilyl)silane (Si(TMS)4) can generate the Si(TMS)3 anion; see ref 30c. Similarly, it has been observed that Lewis base activation of TTMSS can generate the HSi(TMS)2 anion; see: (a) Marschner C A New and Easy Route to Polysilanylpotassium Compounds. Eur. J. Inorg. Chem 1998, 221–226. (33) As control experiments, we subjected TTMSS and cesium formate in DMF at rt to various electrophiles in place of trifluoromethylarenes. We observed substitution products that most likely arise from generation of both the TMS and HSi(TMS)2 anions; see Supporting Information for full details. Thus, we cannot rule out the generation of either silyl anion under the standard reaction conditions. (34) (a) Sakurai H ; Akane O ; Umino H ; Kira M Silyl anions. IV. New convenient method of producing radical anions involving one-electron transfer from trimethylsilylsodium. J. Am. Chem. Soc 1973, 95 , 955–956. (b) Hideki S ; Mitsuo K ; Hiroshi U Silyl Anions VII. Electron Transfer From Trimethylsilylpotassium To Benzophenone And Naphthalene. Generation Of Anion Radicals In Nonpolar Solvents Such As n-Hexane. Chem. Lett 1977, 6 , 1265–1268. (c) Smith AJ ; Young A ; Rohrbach S ; O’Connor EF ; Allison M ; Wang H-S ; Poole DL ; Tuttle T ; Murphy JA Electron-Transfer and Hydride-Transfer Pathways in the Stoltz–Grubbs Reducing System (KOtBu/Et3SiH). Angew. Chem. Int. Ed 2017, 56 , 13747–13751. (35) For pKa values of organosilanes, see: Fu Y ; Liu L ; Li R-Q ; Liu L ; Guo Q-X First-Principal Predictions of Absolute pKa’s of Organic Acids in Dimethyl Sulfoxide Solution. J. Am. Chem. Soc 2003, 126 , 814–822. (36) For representative halophilic substitution processes involving silyl anions, see: (a) Shippey MA ; Dervan PB Trimethylsilyl Anions. Direct Synthesis of Trimethylsilylbenzenes. J. Org. Chem. 1977, 42 , 2654–2655. (b) Uematsu R ; Yamamoto E ; Maeda S ; Ito H ; Taketsugo T Reaction Mechansim of the Anomalous Formal Nucleophilic Borylation of Organic Halides with Silylborane: Combined Theoretical and Experimental Studies. J. Am. Chem. Soc 2015, 137 , 4090–4099.25773395 (37) Use of other disilanes in place of TTMSS results in low yields of 2 and/or poor mass balance of 1; these studies are described in the Supporting Information. The disilane must be capable of being activated by the Lewis base to promote defluorosilylation. The inability of hexamethyldisilane, which can only generate the TMS anion, to promote the reaction in high yield suggests that the role of TTMSS constitutes more than just the generation of a TMS anion. (38) TTMSS is known to readily generate the Si(TMS)3 radical and engage in radical-based reactions (see ref 31). However, replacement of H in TTMSS with other substituents (e.g. alkyl or amino groups) still results in some yield of 2 for the model reaction, suggesting the Si–H bond is not critical for the reaction mechanism. Furthermore, replacement of cesium formate with radical initiators (e.g. AIBN and benzoyl peroxide) results in no reductive coupling reaction. As another control, a reaction conducted with catalytic TTMSS and stoichiometric Si(TMS)4 provided only low yields of product. Additionally, our optimization studies show that acetate bases perform similarly to formate bases, indicating that formate does not act as a reductant or radical anion precursor in this reaction; see: Hendy CM ; Smith GC ; Xu Z ; Lian T ; Jui NT Radical Chain Reduction via Carbon Dioxide Radical Anion (CO2•–). J. Am. Chem. Soc 2021, 143 , 8987–8992.34102836 (39) For base-promoted defluorosilylation of aryl and alkyl fluorides, see: Liu X-W ; Zarate C ; Martin R Base-Mediated Defluorosilylation of C(sp2)–F and C(sp3)–F Bonds. Angew. Chem. Int. Ed 2019, 58 , 2064–2068. (40) (a) Reidl TW ; Bandar JS Lewis Basic Salt-Promoted Organosilane Coupling Reactions with Aromatic Electrophiles. J. Am. Chem. Soc 2021, 143 , 11939–11945.34314159 (b) see also ref 26b. (41) (a) Luo Y-C ; Tong F-F ; Zhang Y ; He C-Y ; Zhang X Visible-Light-Induced Palladium-Catalyzed Selective Defluoroarylation of Trifluoromethylarenes with Arylboronic Acids. J. Am. Chem. Soc 2021, 143 , 139741–13979. (b) For several examples of formal defluoroarylation using aryl boronic acid coupling partners via Young’s frustrated Lewis pair approach, see ref 15b. (42) Charpentier J ; Früh N ; Togni A Electrophilic Trifluoromethylation by Use of Hypervalent Iodine Reagents. Chem. Rev 2015, 115 , 650–682 25152082

PMC009xxxxxx/PMC9822535.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101573691 39703 Cell Rep Cell Rep Cell reports 2211-1247 35977518 9822535 10.1016/j.celrep.2022.111218 NIHMS1831962 Article Genetic impairment of succinate metabolism disrupts bioenergetic sensing in adrenal neuroendocrine cancer Gupta Priyanka 1 Strange Keehn 1 Telange Rahul 2 Guo Ailan 3 Hatch Heather 4 Sobh Amin 5 Elie Jonathan 6 Carter Angela M. 1 Totenhagen John 7 Tan Chunfeng 8 Sonawane Yogesh A. 9 Neuzil Jiri 1011 Natarajan Amarnath 9 Ovens Ashley J. 1213 Oakhill Jonathan S. 1213 Wiederhold Thorsten 3 Pacak Karel 14 Ghayee Hans K. 15 Meijer Laurent 6 Reddy Sushanth 1 Bibb James A. 11617* 1 Department of Surgery, University of Alabama at Birmingham Heersink School of Medicine, Birmingham, AL 35233, USA 2 Department of Hematology, St Jude Children’s Research Hospital, Memphis, TN 38105, USA 3 Cell Signaling Technology, Danvers, MA 01923, USA 4 Department of Physiological Sciences, College of Veterinary Medicine, University of Florida, Gainesville, FL 32610, USA 5 Department of Medicine, Division of Hematology and Oncology, University of Florida, Gainesville, FL 32608, USA 6 Perha Pharmaceuticals, Hôtel de Recherche, Perharidy Peninsula, 29680 Roscoff, France 7 Department of Radiology, University of Alabama at Birmingham Heersink School of Medicine, Birmingham, AL 35233, USA 8 UT Health Science Center at Houston, Department of Neurology, University of Texas McGovern Medical School, Houston, TX 77030, USA 9 Eppley Institute for Research in Cancer and Allied Diseases, University of Nebraska Medical Center, Omaha, NE 68198, USA 10 Institute of Biotechnology, Czech Academy of Sciences, Prague-West 252 50, Czech Republic 11 School of Pharmacy Medical Science, Griffith University, Southport, QLD 4222, Australia 12 Metabolic Signalling Laboratory, St Vincent’s Institute of Medical Research, Fitzroy, VIC, Australia 13 Mary MacKillop Institute for Health Research, Australian Catholic University, Melbourne, VIC, Australia 14 Section on Medical Neuroendocrinology, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD 20892, USA 15 Department of Internal Medicine, Division of Endocrinology, University of Florida College of Medicine and Malcom Randall VA Medical Center, Gainesville, FL 32608, USA 16 O’Neal Comprehensive Cancer Center and the Department of Neurobiology, University of Alabama at Birmingham Heersink School of Medicine, Birmingham, AL 35233, USA 17 Lead contact AUTHOR CONTRIBUTIONS J.A.B. and P.G. conceptualized the project. P.G. designed and performed the experiments, analyzed the data, and wrote the original draft. L.M., A.N., J.E., and Y.A.S. provided Cdk5 inhibitors. K.S., R.T., C.T., H.K.G., H.H., A.S., J.N., J.S.O., A.J.O., A.M.C., J.T., K.P., A.G., and T.W. provided reagents or assisted with the experiments. J.A.B., S.R., K.P., and H.K.G. provided conceptual input. J.A.B. oversaw all study design, data analysis, and review and editing of the manuscript. J.A.B. and S.R. acquired funding and supervised the study. All authors discussed the results and commented on the manuscript. * Correspondence: jbibb@uab.edu 8 9 2022 16 8 2022 06 1 2023 40 7 111218111218 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY Metabolic dysfunction mutations can impair energy sensing and cause cancer. Loss of function of the mitochondrial tricarboxylic acid (TCA) cycle enzyme subunit succinate dehydrogenase B (SDHB) results in various forms of cancer typified by pheochromocytoma (PC). Here we delineate a signaling cascade where the loss of SDHB induces the Warburg effect, triggers dysregulation of [Ca2+]i, and aberrantly activates calpain and protein kinase Cdk5, through conversion of its cofactor from p35 to p25. Consequently, aberrant Cdk5 initiates a phospho-signaling cascade where GSK3 inhibition inactivates energy sensing by AMP kinase through dephosphorylation of the AMP kinase γ subunit, PRKAG2. Overexpression of p25-GFP in mouse adrenal chromaffin cells also elicits this phosphorylation signaling and causes PC. A potent Cdk5 inhibitor, MRT3-007, reverses this phospho-cascade, invoking a senescence-like phenotype. This therapeutic approach halted tumor progression in vivo. Thus, we reveal an important mechanistic feature of metabolic sensing and demonstrate that its dysregulation underlies tumor progression in PC and likely other cancers. In brief Gupta et al. describe a signaling cascade by which TCA cycle deficiency inactivates bioenergetic sensing to cause adrenal gland cancer. Animal models and drugs targeting the cascade support the discovery, which suggests an additional explanation for Warburg’s description of cancers as uncontrolled cell proliferation despite metabolic impairment. Graphical Abstract pmcINTRODUCTION The tricarboxylic acid (TCA) cycle is the critical energy-yielding component of cellular respiration and the centerpiece of cell metabolism. Impairments in TCA cycle components cause a spectrum of metabolic disorders (Krebs and Johnson, 1980). These alterations may still allow embryonic and postnatal development but result in organ-system dysfunction or context-specific pathology (Jeanmonod et al., 2020; Stockholm et al., 2005). Multiple enzymes of the TCA cycle, such as fumarate hydratase (FH), isocitrate dehydrogenase (IDH), and succinate dehydrogenase (SDH) are altered in numerous sporadic or hereditary forms of cancer, causing the production of onco-metabolites. Pheochromocytoma (PC) and paraganglioma (PG) are archetypical metabolic tumors in which TCA cycle malfunction results in tumors arising from chromaffin cells of the adrenal medulla or sympathetic and parasympathetic ganglia (Dahia, 2014; Erez et al., 2011). High rates of SDHx-related abnormalities correlate with an increased risk of metastatic PCPG in patients who often developed primary tumors in childhood or adolescence (King et al., 2011). The SDHx complex is a hetero-tetrameric protein (composed of four subunits: SDHA, SDHB, SDHC, and SDHD) that functions in both the TCA cycle and electron transport chain (ETC), catalyzing the oxidation of succinate to fumarate and converting ubiquinone to ubiquinol, respectively (Bardella et al., 2011). Patients harboring SDHB mutations have a lifetime cancer risk of 76%, with 50% penetrance by the age of 35 years (Neumann et al., 2004). A wide array of intragenic mutations occur in SDHx complex genes, including frameshifts, splicing defects, and single/multiple exon deletions. However, genetic analysis has clearly shown that heterozygous mutations of SDHx genes predominate in PCPG patients, and very few patients harbor tumors completely lacking SDH function. Approximately 60% of SDHB-related adrenal or extra-adrenal PCs present allelic imbalances (Gimenez-Roqueplo et al., 2002, 2003). A significant focal deletion in the SDHB locus of chromosome 1 in 91 out of 93 patients was identified in a human PC/PG dataset (NCIC3326) (https://progenetix.org/progenetix-cohorts/TCGA/). These exhibit complicated phenotypes that affect gene expression and enzymatic function to varying degrees (Huang et al., 2018; Solis et al., 2009). SDHB mutant PC typically exhibited intact mRNA expression but significantly reduced protein compared with non-SDHB PC/PG (Yang et al., 2012). Retention of a haploinsufficient wild-type allele results in reduced SDHB expression and function. For example, impaired SDH maturation and compartmentalization arising from germline mutation p.Tyr147Cys in SDHB manifested as reduced enzymatic function (Maignan et al., 2017). Germline SDHB mutations such as c.343C>T and c.541-2A>G still exhibit some level of protein expression in patients (Huang et al., 2018). Thus, there is a strong rationale to model heterozygous SDHB genotype that elicits reduced but not complete loss of SDH function. Malfunction due to loss of SDHx genes causes abnormal accumulation of succinate, thereby activating angiogenic/hypoxia-responsive genes, which further promote metabolic reprogramming and tumor progression (Anderson et al., 2018). Other malignancies affected by SDHB loss of function include renal carcinoma, gastrointestinal stromal tumors, pancreatic neuroendocrine tumors, pituitary adenoma, and pulmonary chondroma (Eijkelenkamp et al., 2020). For all these, attenuating SDHB impairs mitochondrial respiration and causes a metabolic shift in favor of aerobic glycolysis to meet the high energetic and biosynthetic demands of tumor cells, a process known as the Warburg effect (Favier et al., 2009; Kim and Dang, 2006). Impaired TCA cycle function due to SDHB loss not only perturbs cellular bioenergetics but also elicits tumor-promoting signaling mediated by Ca2+ or reactive oxygen species (ROS) released by mitochondria (Hadrava Vanova et al., 2020; Jana et al., 2019). Moreover, TCA cycle dysfunction can also override cell-cycle checkpoint regulators whereby cells may become poorly differentiated, and malignant cell-cycle progression can ensue. While SDHB loss characterizes highly aggressive forms of PC and other diseases (Jochmanova and Pacak, 2016), the underlying mechanisms by which metabolism is reprogrammed and connected to the loss of cell cycle control is not yet fully understood. As a result, it has been challenging to model these diseases or identify therapeutic targets. Consequently, outcome improvement for patients with these metabolic errors has been limited. Here we studied the impact of metabolic dysfunction caused by loss of SDHB on bioenergetics, Ca2+ dynamics, intracellular signaling, and tumor progression. Our findings provide additional understanding of the mechanisms that control proliferation and senescence-like features and reveal drug targets for the treatment of PC and other SDHB mutation-derived disorders. RESULTS SDHB loss alters cellular metabolism and dysregulates Cdk5 signaling Heterozygous SDHB mutations often associate with reduced mRNA expression, or truncated protein (Cascon et al., 2006; Huang et al., 2018; Yang et al., 2012), which may result in impaired but not necessarily complete loss of SDH activity. Analysis of a PCPG dataset revealed genetic alterations within SDHx complex in 54% of patients (n = 161). The highest portion of 39% occurs in SDHB/SDHD where a majority of samples exhibit reduced SDHB expression (Figures 1A and 1B). Also, ~60% of patients with shallow SDHB deletions have corresponding reductions in SDHB mRNA relative to individuals with an unaltered copy of genes (Figure 1C). A significant decrease in SDHB transcripts in PC compared with normal adrenals indicates SDHB haploinsufficiency and thus sensitivity to copy number alterations (Figure 1D). Haploinsufficiency in SDHB develops clinically aggressive metastatic tumors manifesting stem-like properties (Baysal and Maher, 2015; Buffet et al., 2012). This also indicates that SDHB null is relevant to only a small subset of patients and not the absolute presentation of SDHx-associated disease to study downstream signaling mechanisms. Therefore, we conducted CRISPR-Cas9 targeted gene deletion of SHDB from progenitor human PC tumor-derived cell line, hPheo1 (Ghayee et al., 2013) and selected a partial knockout (KO) clone expressing 20% of basal SDHB levels as an accurate cellular model of the human disease (Figure 1E). Altered metabolism in cancer cells is a common feature where TCA cycle impairment in tumors harboring mutations in SDHB or other key components of aerobic respiration contributes to the Warburg effect (Vander Heiden et al., 2009). In agreement, SDHB KO dramatically altered the cellular bioenergetics profile. Sequential metabolic perturbation with glucose, oligomycin, and 2-deoxy-D-glucose (2-DG) enhanced the extracellular acidification rate (ECAR), which corresponded to metabolic shifts toward increased glycolysis, glycolytic capacity, and glycolytic reserve in SDHB KO cells (Figure 1F). These effects in the KO hPheo1 recapitulate the metabolic phenotype that characterizes SDHB mutant human PC (Favier et al., 2009; Jochmanova and Pacak, 2016). Concomitant comparison of mitochondrial function indicated an elevated oxygen consumption rate (OCR) and higher basal and ATP-linked respiration in wild type (WT) hPheo1, suggesting decreased efficiency of mitochondrial function due to loss of SDHB (Figure 1G). Nevertheless, when challenged with high energy demand via disruption of mitochondrial membrane potential with carbonyl cyanide-4 (trifluoromethoxy) phenylhydrazone (FCCP), a significant spike in maximum respiration was induced in both cell types, suggesting glycolytic preference but not complete shut-off of mitochondrial function in SDHB KO PC cells. In accordance, it has been shown that chromaffin cells retain mitochondrial fitness despite SDHx loss (Kluckova et al., 2020). Loss of SDHx function not only alters metabolism but also may dysregulate Ca2+ homeostasis (Nasr et al., 2003; Ranganayaki et al., 2021). To evaluate the effect of SDHB loss on Ca2+, single-cell confocal imaging was performed to compare intracellular [Ca2+]i in WT versus SDHB KO hPheo1 cells. In WT cells, a rapid time-dependent recovery to the ionomycin-induced spike in [Ca2+]i occurred, where [Ca2+]i surge returned to basal levels within 5 min. This response was absent in SDHB KO cells where increased [Ca2+]i levels were sustained without significant reduction over the same period of analysis (Figure 2A). In addition to causing an imbalance in [Ca2+]i, SDHB loss can trigger buildup in the TCA metabolite, succinate. Indeed, succinate concentrations were significantly higher in SDHB KO cells compared with controls (Figure 2B). We hypothesized that this increase in basal intracellular succinate could contribute to the loss in [Ca2+]i dynamics observed in KO cells. To test this, WT PC cells were treated with cell-permeable dimethyl succinate (DMS). Similar to the effect of SDHB KO, the addition of exogenous succinate disrupted ionomycin-induced [Ca2+]i homeostasis recovery in parent cells (Figure 2C). Succinate may alter [Ca2+]i homeostasis through autocrine signaling. Specifically, excessive succinate freely shuttles between the mitochondria, cytosol, and across the cell membrane to stimulate the SUCNR1 receptor in SDHx-related PC (Matlac et al., 2021). Interestingly, hPheo1 expressed appreciable levels of SUCNR1 and SDHB KO causing a 1.75-fold increase in SUCNR1 expression (Figure S1A). To test if SDHB loss and consequent succinate accumulation triggers SUCNR1-mediated disruption of [Ca2+]i homeostasis, KO cells were treated with SUCNR1 antagonist (NF-56-EJ40). [Ca2+]i homeostasis was impaired, as was observed earlier in KO cells. However, pretreatment with NF-56-EJ40, rescued the WT phenotype for [Ca2+]i recovery in SDHB KO cells (Figures 2D and S1B). These data indicate that SDHB loss caused an increase in succinate, which then destabilized [Ca2+]i through constitutive activation of SUCNR1 receptors. The Ca2+-dependent protease calpain is a key downstream effector activated by loss of Ca2+ homeostasis (Crespo-Biel et al., 2007; Nasr et al., 2003; Pang et al., 2003). To determine if SDHB KO activated calpain, a fluorescence resonance energy transfer (FRET) probe harboring a calpain-specific substrate was used (Stockholm et al., 2005). SDHB KO caused an intracellular calpain activity increase, as indicated by a higher FRET index in parent hPheo1 versus SDHB KO cells expressing the calpain reporter (Figures 2E and S2A). Accordingly, an increase in the 145-kDa breakdown product of spectrin confirmed elevated ubiquitous calpain activity in SDHB-deficient cells (Figure S3A) (Rajgopal and Vemuri, 2002). The Cdk5 activating cofactor p35 is an important substrate of calpain that is cleaved to p25 in response to the loss of [Ca2+]i homeostasis. Increased expression of coactivators p35/25 is pivotal to stimulating aberrant or pathological Cdk5 hyper-activation (Patrick et al., 1999; Pozo et al., 2013), which can cause neuronal death and may drive neuroendocrine cell proliferation (Barros-Minones et al., 2013; Carter et al., 2020; Crespo-Biel et al., 2009). Interestingly, SDHB KO triggered a cumulative increase of both p35 and p25 levels compared with WT cells (Figure 2F). Concomitant immunofluorescent quantitation showed a 22.1 ± 2.4 pixel intensity increase in p35/p25 signals in KO versus parent cell lines (p = 0.002) (Figure 2G). The expression patterns of SDHB, p35/p25, and Cdk5 were recapitulated in a second independent partial knockout clone (KO-8) (Figure S3B), confirming that aberrant Cdk5 signaling is not selective to a particular clone but rather a potential common target linked to SDHB loss. Furthermore, exogenous succinate induced a dose-dependent increase in p35/p25 levels, with a simultaneous increase in spectrin cleavage in parent cells (Figures 2H, 2I, and S3C). The increased expression of p35 and p25 generation in response to SDHB KO was also observed in vivo when WT or SDHB KO hPheo1 cells were used to create xenografts in SCID mice. SDHB KO cells were more likely to produce viable tumors with rapid onset compared with WT cells (Figure S3D). Tumors derived from SDHB KO cells exhibited increased calpain activity (Figure S3E) and exhibited higher levels of p35 and p25, while Cdk5 levels were comparable in both WT and KO tumors (Figure 2J). Tumor sections from both WT and SDHB KO xenografts immunostained for tyrosine hydroxylase (TH) and chromogranin A (CgA), as hallmarks of human neuroendocrine PC (Figure S3F) (Fliedner et al., 2010). There was also a trend toward higher TH and ChrA levels in KO mice (Figure S3G). Together these data show that a shallow deletion of SDHB, similar to what occurs in humans, alters the metabolic profile, induces succinate buildup, and perturbs Ca2+/calpain/Cdk5 signaling in PC cells (Figures 2K and S3H). SDHB and Cdk5 correlation in human PC Aberrant Cdk5 has been previously implicated in human neuroendocrine tumors (NETs) (Pozo and Bibb, 2016). However, the link between Cdk5 and SDHx-related NETs has heretofore not been investigated. Given that SDHB loss elicits elevated p35/25 levels in human PC cells, we assessed p35 transcripts and protein levels in human PC tissues. Interestingly, SDHB and Cdk5R1 (i.e., encodes p35) were differentially expressed between PCs and adjacent normal adrenal medulla in The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) datasets. A significant reduction in SDHB mRNA (see Figure 1D) corresponds to increased Cdk5R1 expression in PC compared with normal adrenals (Figure 3A). In addition, co-expression analysis of 161 PCPGs highlighted a significant negative correlation between SDHB and Cdk5R1 (R = −0.41, p = 6.78 × 10−8; Figure 3B). No such significant correlation was observed between Cdk5R1 and other SDH subunit genes (such as SDHA, SDHC, and SDHD) or with other tumor-suppressive genes of the TCA cycle (e.g., FH, IDH1, CS) (Figures S4A–S4F). Also, a PCPG GEO dataset of 84 patients showed an inverse correlation between the expression of Cdk5 activating components (Cdk5, Cdk5R1, Cdk5R2) and SDHB (p < 0.001), while no such relationships were observed with SDHA, SDHC, and SDHD subunits (Figures S5A–S5C). Moreover, treatment of WT hPheo1 cells with cell-permeable succinate increased Cdk5R1 mRNA, suggesting that the inverse relationship between SDHB and Cdk5R1 may be mediated via succinate signaling (Figure S5D). Germline mutations in SDHx genes are responsible for 38%–80% of metastatic familial PCPGs, while ~10% of sporadic PC also harbored SDHx mutations (Gottlieb and Tomlinson, 2005; Korpershoek et al., 2011). To determine if the inverse correlation between SDHB and p35 gene expression characterizes PC tumors at the post-translation level, immunoblots were performed with sporadic PC tumors versus human normal adrenal medulla. Consistent with the mRNA levels, SDHB protein levels decreased, whereas p35/p25 cumulative protein band intensities were increased (Figure 3C). This effect was mirrored in PC in which SDHB mutations were observed (Figure 3D). Both sporadic and SDHx mutated PCs immunostained for Cdk5 and p35/p25 (Figure 3E). To determine cell type selectivity, differential expression of p35/25 was determined between PC tissues and a different form of adrenal tumor derived from cortical cells known as adrenocortical adenoma (ACA). Analysis of a human adrenal tissue microarray (30 cases of PC and 40 cases of ACA) showed significantly higher levels of p35/25 in PC compared with ACA (1.5-fold, p = 0.0002) with no measurable change in Cdk5 (Figure 3F), implicating selective functional significance for aberrant Cdk5 in chromaffin cell-derived tumors. These results illustrate that human PCs manifest a negative correlation between SDHB and p35/p25, supporting the concept that aberrant Cdk5 can function as a driver of tumor cell proliferation in PC. Aberrant activation of Cdk5 in mouse chromaffin cells causes human-like PC To determine if aberrant Cdk5 activation recapitulates human PC, a PiggyBac transgenic mouse carrying single-copy PNMT (gene encodes the enzyme catalyzing the final step of catecholamine biosynthesis) transgene promoter was designed to drive chromaffin cell-specific tetracycline transactivator (tTA) expression (Goldstein et al., 1972; Ross et al., 1990) (Figures 4A and S6A). This animal was then crossed with the Tet-Op-p25GFP line (Bujard, 1999) to generate bitransgenic mice in which p25 overexpression (p25OE) could be induced in adrenal medulla chromaffin cells by removal of dietary doxycycline (Figure S6B) (Cruz et al., 2003). Here, P25GFP expression status is denoted as p25-ON (Dox-OFF) or p25-OFF (Dox-ON), and PNMT-tTA littermate controls are referred to as WT. Aberrant Cdk5 induced via activation of p25OE caused bilateral 12- to 50-mm3 PCs to develop in 20–21 weeks with no effect on adrenal glands of WT mice (Figures 4B and 4C). Higher levels of p25GFP, TH, and ChrA occurred in p25-ON PC tissue compared with WT adrenal glands (Figure 4D). Moreover, p25-ON PC expressed the highest levels of these proteins compared with other organs, including liver, brain, spleen, and lung, demonstrating tissue type specificity. P25-ON tumor tissue showed neuroendocrine pseudo-rosettes surrounding blood vessels immunostained for TH/ChrA that typify human PC. While both WT adrenal medulla and p25-ON PC stained for Cdk5, only p25ON tumors were p25-GFP positive (Figure 4E). Patients exhibit hypertensive crises due to PC-derived overproduction of catecholamines (Eisenhofer et al., 1999). Similarly, p25-ON mice had elevated plasma levels of nor-metanephrine, metanephrine, and higher systolic blood pressure (Figures 4F–4H). These data show that aberrant Cdk5 activity, such as that resulting from SDHB mutations, can drive the formation and progression of PC and the clinical symptoms with which it is associated in a mouse model. Screening functional targets of aberrant Cdk5 Loss- or gain-of-function mutations in kinases dysregulate signaling pathways associated with cancer (Hanahan and Weinberg, 2000). To understand how the aberrant Cdk5 invokes PC tumorigenesis, we analyzed our published library of phosphorylation sites derived from p25-ON/OFF NETs (Carter et al., 2020). Of 2,000 proline-directed phosphorylation sites detected, 200 were upregulated, while 122 were downregulated, suggesting that p25OE both positively and negatively regulates protein phosphorylation in growing versus arrested tumors (Figure 5A). To gain a wider mechanistic perspective we focused on phosphorylation sites suppressed in p25-ON tumors. Forty-four phosphoproteins were downregulated by >60% in growing tumors (Figure 5B). Functional analysis of these phosphoproteins revealed a network of enriched pathways involved in metabolic processes, cell-cycle regulation, cell size, kinase activity, protein modification, and RNA processing (Figures 5C, S7, S8A, and S8B). To analyze phosphorylation sites of particular interest in PC, five targets were chosen based on the algorithm of having (1) unknown phosphorylation sites; (2) conserved phosphosite in humans; (3) implicated in cancer metabolism, protein trafficking, cell cycle, autophagy, or protein translation. These phospho-targets included phospho-Ser672 AAK-1, phospho-Ser244 DEPTOR, phosphoSer65 PRKAG2, phospho-Thr517 UVRAG, and phospho-Ser288 SDPR (Figure 5D). To assess functional relevance to PC tumorigenesis, each phosphorylation site was mutated to encode phospho-mimetic (D/E) or phospho-null (A) amino acids and transiently overexpressed in hPheo1 cells. Of the five phospho-targets, S65D/E mutants of PRKAG2 (5’-AMP-activated protein kinase subunit gamma-2) significantly suppressed cell proliferation compared with that of WT or the S65A form (Figure 5E). PRKAG2 is the non-catalytic regulatory γ2 subunit of AMP-activated protein kinase (AMPK), a heterotrimeric metabolic sensor composed of catalytic α subunit and two regulatory (β and γ) subunits (Figure 5F). AMPK regulates cellular energy homeostasis, glucose sensing, and immune response processes that affect cell growth and proliferation (Zadra et al., 2015). PRKAG2 has been chiefly studied in cardiac and skeletal muscles (Pinter et al., 2013; Zhan et al., 2018). However, the isoform-specific function of PRKAG2 is largely unexplored in cancer. Nucleotide binding to PRKAG2 confers phosphorylation/dephosphorylation upon residue Thr172 in the activation loop of the AMPK catalytic α subunit, as a prerequisite to kinase activation (Gowans et al., 2013; Oakhill et al., 2011; Shaw et al., 2004). Interestingly, D/E mutations of S65 PRKAG2 increased phospho-Thr172 AMPKα (Figures 5G and S8C). This suggests that phosphorylation at Ser65 PRKAG2 controls AMPK activity. These data also implicate the regulation of AMPK via phosphoSer65 PRKAG2 as an important downstream effector of aberrant Cdk5 in NE cells. Aberrant Cdk5/GSK3 deregulates AMPK pathway AMPK activity is regulated by either allosteric or non-canonical mechanisms (Hardie, 2014; Hawley et al., 2005). To better understand how Ser65 PRKAG2 phosphorylation could regulate AMPK activity, a phosphorylation state-specific antibody was generated to this site (Figures S9A–S9C). Analysis of human patient tissue showed that PC tumors had significant decreases in both phospho-Ser65 PRKAG2 and -Thr172 AMPKα compared with normal control adrenals, with a linear correlation between the two sites (Figures 5H and 5I). In contrast, no alterations in PRKAG2/AMPK gene expression were observed between normal and tumor tissues (TCGA-Gene Expression Profiling Interactive Analysis [GEPIA] dataset; n = 182; Figure S9D). Thus decreased Ser65 phosphorylation and AMPK inactivation characterize both human and mouse NE tumors. A decrease in phospho-Ser65 PRKAG2 and AMPK activity in response to aberrant activation of Cdk5 suggests an intermediary signaling step is involved. In fact, both Cdk5 and GSK3 are predicted to phosphorylate Ser65 PRKAG2 (Figure S9E). Since Cdk5 activation indirectly results in GSK3 inactivation (Morfini et al., 2004), we asked if Ser65 PRKAG2 phosphorylation by GSK3 is downstream of aberrant Cdk5/p25. Indeed, GSK3 efficiently phosphorylated purified recombinant PRKAG2 (Figure 6A). The site at which GSK3 phosphorylates PRKAG2 was confirmed as Ser65 by immunoblotting, while Cdk5 did not phosphorylate this site in vitro (Figures 6B and 6C). Also, active GSK3 immunoprecipitated from hPheo1 cell lysates efficiently phosphorylated WT PRKAG2 in vitro compared with the S65A mutant (Figures S9F and S9G). Aberrant Cdk5 activation leads to inhibitory phosphorylation of Ser9 GSK3b (Plattner et al., 2006; Wen et al., 2008). In agreement, SDHB KO PC cells exhibited higher phospho-Ser9 GSK3β levels compared with parent cells (Figure S9H). Also, the expression of kinase-dead (KD) versus WT Cdk5 reduced phospho-Ser9 GSK3 by 50% and caused a concomitant increase in phospho-Ser65 PRKAG2 and Thr172 AMPKα (Figure 6D). As evidence that this pathway is linked to upstream dysregulation of succinate signaling, treatment of hPheo1 KO cells with the SUCNR1 antagonist NF-56-EJ40 also disrupted aberrant Cdk5-GSK3-AMPK signaling (Figure S9I). Inhibitors or activators of GSK3 may induce cellular proliferation or suppression dependent on cell type (Li et al., 2014; Pap and Cooper, 1998; Tang et al., 2012). Here, we found that hPheo1 cells were unaffected by treatment with a GSK3 inhibitor (SB216763), while a Cdk5 inhibitor (CP681301) dose-dependently abrogated cell viability (Figure 6E). Moreover, metabolic modulators that activate AMPK, including AICAR, metformin (Vial et al., 2019), and 2-DG (Wang et al., 2011), induced growth-inhibitory effects, whereas an AMPK inhibitor, compound C (CC; dorsomorphin) (Zhou et al., 2001) had a minimal effect on cell growth (Figures 6F and 6G). These results suggest a pivotal signaling mechanism where aberrant Cdk5 activation inhibits GSK3, thereby disrupting phospho-PRKAG2-dependent AMPK activation (Figure 6H). Characterization of a potent Cdk5 inhibitor With the emergence of Cdk5 as a promising target in cancers, there is renewed demand for effective drugs that target this kinase. Currently available Cdk5 inhibitors such as roscovitine (CYC202, seliciclib) act as purine analogs that interfere with ATP binding (Bettayeb et al., 2010). However, lack of specificity, short half-life, rapid degradation, and weak potency limit their potential for clinical use (McClue and Stuart, 2008). Therefore, we screened a small library of selective Cdk5 inhibitors including 25–16, MRT3–007, MRT3–124, and CR8 using roscovitine as a positive control. These compounds share the same general chemical structures (Figure 6I) and relative kinase selectivity (Figure S10A). Each of these compounds exhibited dose-dependent effects on hPheo1 cell viability (Figures 6J and S10B). MRT3–007 showed the highest potency, with an half maximal inhibitory concentration (IC50) value approximately 1,000-fold lower than that of roscovitine (25 ± 10 nM for MRT3–007 versus 26 ± 10 μM for roscovitine). Effective cancer treatments using kinase inhibitors depend on the precise genetic constitution of individual patients so that differences in molecular signatures between tumor and normal cells can be defined (Broekman et al., 2011; McDermott et al., 2007). Furthermore, it is possible for inhibitors to be effective by targeting the kinase driving neoplasia while exhibiting broader activity in vitro. As a typical example, MRT3–007 shares overlapping selectivity for Cdk1, 2 and 9 in vitro. Therefore, we queried whether its growth-inhibitory effects in PC were Cdk5 dependent. Notably, PC/PG patients showed significantly higher gene expression of Cdk5 over Cdk1/2 (Figure S10C), while Cdk9 expression was absent in the experimental models used here. Correspondingly, MRT3–007 was more effective in suppressing in vitro growth of PC compared with a selective Cdk2 inhibitor (CVT-313) (Figure S10D). Also, MRT3–007 elicited limited efficacy on cancer cell lines expressing lower p25 levels, such as those derived from breast cancer (MDAMB231 cells), liver carcinoma (HepG2 cells), and small cell lung cancer (H1184 cells). In contrast, MRT3–007 showed higher potency for higher p25-expressing cervical cancer cells (HeLa cells), comparable with PC-derived cells (Figure S10E). Thus this class of compounds could be effective in treating tumors that are dependent upon hyperactive Cdk5. Additional support for ex vivo selectivity of MRT3–007 for Cdk5 over Cdk1/2 was derived by immunoprecipitating Cdk5 or Cdk1/2 from the lysates of hPheo1 cells treated with MRT3–007 using histone H1 as a reporter substrate. Cdk5 but not Cdk1/2-dependent phosphorylation of histone H1 was significantly attenuated, suggesting MRT3–007 is more selective for Cdk5 within the context of the intracellular milieu than in vitro (Figure S10F). MRT3–007 also selectively inhibited the proliferation of cells that overexpress p25 but had no effect on cells lacking p25 expression (Dox-ON; Figure S10G), implying aberrant Cdk5-dependent sensitivity. Together these results suggest that MRT3–007 can be exploited as a potent Cdk5 inhibitor (Cdk5in) in cells, which depend upon aberrant Cdk5 for their growth. Cdk5-GSK3-AMPK cascade controls bioenergetics and induces senescence Cdk5 plays a critical role in tumor-associated cell-cycle progression, DNA damage response, and mitochondrial dysfunction (Mao and Hinds, 2010; Sun et al., 2008; Zhang et al., 2015). Here, we showed that SDHB loss caused a metabolic shift toward glycolysis and triggered aberrant Cdk5-AMPK signaling. These findings prompted us to explore the impact of Cdk5 inhibitor on AMPK signaling, downstream metabolism, and cell-cycle progression. Treatment of SDHB KO hPheo1 cells with Cdk5in (MRT3–007) decreased basal glycolysis, glycolytic capacity and glycolytic reserve (Figure 6K). In addition, Cdk5in shifted the cell energy profile (ratio of OCR:ECAR) from a glycolytic to quiescence/low-energy state (Figure 6L). Concomitantly, Cdk5 inhibition caused time-dependent increases in phospho-S65 PRKAG2 and -T172 AMPKα (Figure 6M). These effects corresponded to increased phosphorylation of S79 acetyl-CoA carboxylase (ACC), a defined AMPK activity reporter, confirming that Cdk5in activates the AMPK pathway. To better understand the relationship of Cdk5-GSK3-AMPK signaling pathway, a selective GSK3 inhibitor, SB216763 was used. SB216763 pretreatment suppressed activation of S65-PRKAG2, T172-AMPK, and S79-ACC phosphorylation induced by Cdk5in (Figure 6N). Also, direct inhibition of AMPK via CC reversed the effects of Cdk5in on phospho-T172 AMPK (Figure 6O). These data further substantiate the Cdk5-GSK3-AMPK cascade as a critical signaling mechanism downstream of SDHB deficiency. On sensing bioenergetics stress, AMPK-dependent phosphorylation of Ser15 p53 stabilizes the protein causing cell-cycle arrest (Jones et al., 2005; Garcia and Shaw, 2017). In agreement, Cdk5in induced a time-dependent increase in phospho-Ser15 p53 (Figure S11A), implicating p53 as a critical downstream effector of AMPK in PC. P53 functions as a transcription factor controlling cell-cycle regulatory gene expression (Chen, 2016; Mijit et al., 2020), while AMPK-p53 activation can also induce cellular senescence in response to bioenergetics stress or chemotherapy (Jones et al., 2005; Lee et al., 2015; Xue et al., 2007). Here, in addition to increased phospho-Ser15 p53, Cdk5in induced time-dependent increases in the expression of senescence markers, p16INK4a and p27Kip, which associate with the primary G1-S checkpoint (Figures S11B and S11C). Cdk5in also caused a sharp rise in phospho-Ser139 histone H2AX, a potential indicator of a transition from senescent to apoptotic cell death (Figure S11D). These effects were negated in the presence of the p53-specific inhibitor, Pifthrin (Pftα) (Figure S11E). Cdk5in also induced senescence-like cell morphology, marked by enlarged flattened cells (Figure 7A), disorganized cytoskeleton, and elevated β-galactosidase, concomitant with overexpression of p16INK4a, p27, and phospho-S139 H2Ax (Figure 7B). Of note, inactivation of AMPK via CC reversed the cell spreading induced by Cdk5in, indicating that AMPK activation mediated the Cdk5in-induced morphological changes (Figure S11F). These phenotypic effects induced by Cdk5 inhibition corresponded to a significant shift toward cell-cycle arrest in G1 phase with a drastic expulsion of cells from G2 phase (Figure S11G). At the same time, a decrease in proliferation marker Ki67 was evident (Figure 7C). The G1/S cell-cycle arrest was also observed in cells that overexpressed S65D PRKAG2 compared with those expressing either WT or S65A forms of the AMPK regulatory subunit (Figure S11H). Thus, Cdk5 inhibition not only alters the bioenergetics of cancer cells but also provokes senescence-like characteristics following activation of PRKAG2/AMPK/p53 signaling cascade. Cdk5 inhibition as a preclinical treatment for SDHB-mediated disease The above results implicate Cdk5in as a promising targeted therapy for PC and other SDHB-related disorders. To evaluate its anti-tumor potential in vivo, the maximal tolerated dose (MTD) of Cdk5in was first determined in mice. MRT3–007 was well tolerated up to a dose of 1 mg/kg, intraperitoneal (i.p.) (Figure S11I). Subsequently, mice carrying SDHB KO hPheo1 xenografts were treated with 0.5 mg/kg Cdk5in, which induced a significant reduction in tumor volume and mitotic index (Figures 7D and 7E) with a minimal adverse effect on body weight (Figure S11J). Additionally, in vivo therapeutic efficacy of Cdk5in was confirmed using a metastatic allograft model of PC where luciferase-expressing mouse PC cells (MTT) were injected intravenously and metastases were imaged in vivo (Martiniova et al., 2009). Cdk5in dramatically reduced tumor signal compared with both vehicle and its parent compound, roscovitine (Figures 7F and 7G). Consistent with our ex vivo findings, Cdk5 inhibition reduced phospho-Ser21/9 GSK3 and increased phospho-Ser65 PRKAG2, -Thr172 AMPK, and -Ser79 ACC in the lysates of PC xenografts (Figures S12A–S12D). Additionally, Cdk5in treatment increased phospho-Ser15 p53, p16, and p27 levels, consistent with that of senescence-like markers observed in vitro (Figure S12E). These findings indicate that Cdk5in possesses therapeutic potential in the treatment of PC and other SDHB-related disorders. Finally, to validate that PC tumor progression is dependent upon aberrant Cdk5 activity, we assessed the effects of halting p25OE in the bitransgenic mouse PC model. Under Dox-OFF conditions, tumor size progressed by 2.5-fold over 18 weeks. However, the replacement of dietary Dox (Dox-ON) significantly limited tumor volume with a corresponding decrease in p25-GFP expression (Figures 7H–7J). Tumor arrest due to halt in p25OE (i.e., p25-OFF) resulted in decreased phospho-Ser21/9 GSK3 and increased phospho-Ser65 PRKAG2, -Thr172 AMPK, and -Ser79 ACC (Figures S12F–S12I). Once again, these tumors achieve a senescent-like state as indicated by increased phospho-Ser15 p53 and p16 levels (Figures S12J and S12K). Together, these data support a signaling cascade (Figure S12L) where SDHB loss causes accumulation of succinate, which, in the context of metabolic impairment, leads to loss of Ca2+ control and calpain activation. As a result, aberrant Cdk5/p25 accumulates and causes GSK3 inactivation. Consequently, AMPK is inactivated through the reduction in Ser65 PRKAG2 phosphorylation. Attenuated AMPK activity leads to reduced phosphorylation of Ser15 p53 promoting cell proliferation. This signaling cascade appears to be an important additional feature of the Warburg effect. DISCUSSION TCA-linked mitochondrial malfunction and elevated glucose utilization is considered the root cause of several human diseases, including cancer, diabetes, and neurodegenerative disorders (Blank et al., 2010; Hsu and Sabatini, 2008). Here, we delineated a signaling cascade that highlights critical phosphorylation hotspots on metabolic checkpoints disrupted by loss of the TCA cycle component, SDHB. Several downstream effects of succinate accumulation have been linked to altered metabolism, pseudohypoxia, and SUCNR1 activation (Dahia et al., 2005; Matlac et al., 2021; Pollard et al., 2006). Although these mechanisms have been demonstrated to serve as components in discrete cellular- or disease-specific contexts, their suggested interactions as part of the multistep process of carcinogenesis have not been fully delineated. Hence, the complete picture of how TCA perturbations can lead to cancer has not yet emerged. Previously, features of neurodegeneration or ischemic injury have been modeled by inhibiting SDH activity using 3-nitropropionic acid, and malonate, which activates calpain/Cdk5 signaling (Barros-Minones et al., 2013; Pang et al., 2003; Ranganayaki et al., 2021). These findings support the importance of Cdk5 in SDHx deficiency diseases. Here we deciphered several distinct aspects of tumorigenic signaling including metabolic shift coupled with dysregulation of [Ca2+]i/calpain/aberrant Cdk5 in response to SDHB loss in PC tumors. Recent studies have identified ancillary driver mutations in ATRX (Fishbein et al., 2015), KIF1B, and NF1 (Evenepoel et al., 2017) in patients harboring SDHB mutations. This is congruent with recently discovered mutations in KIF1B and NRAS genes in hPheo1 cells (Rossitti et al., 2020). In addition to the germline driver SDHB mutations, ancillary mutations in hPheo1 create the opportunity to study malignant PC behavior. Moreover, the majority of the SDHx mutations are not complete deletions as was previously thought, but human tumors harboring SDHx mutations still express some of the protein (Gimenez-Roqueplo et al., 2003; Neumann et al., 2004; Yang et al., 2012). Thus, the partial SDHB KO hPheo1 cells provided a valuable clinically relevant tool to decipher distinct tumor phenotypes, and signaling mechanisms. Extracellular succinate activates SUCNR1, which invokes multiple signaling outcomes, dependent on the cell type, while excess intracellular succinate accumulation causes inhibition of 2-oxoglutarate-dependent dioxygenases, histone, and DNA demethylases. Succinate accumulation could cause [Ca2+]i. dysregulation either through activation of SUCNR1 receptors or mitochondrial ROS (Andrienko et al., 2017). SUCNR1 are metabolic stress sensors that modulate intracellular Ca2+ through an inositol phosphate-dependent mechanism via PLCβ activation or pathways downstream of G protein invocation (Bhuniya et al., 2011; Gilissen et al., 2016; Sundstrom et al., 2013; Wu et al., 2020). We showed that succinate accumulation impaired intracellular Ca2+ dynamics, and caused aberrant Cdk5 activity, while inhibition of SUCNR1 rescued these effects. Thus, we demonstrate succinate-SUCNR1 pathway as a primary source of [Ca2+]i. dysregulation in cancer cells. Cdk5-GSK3β interactions have been reported in neurodegenerative disorders, although how Cdk5 induces GSK3 inhibition remains to be explained. This may involve either activation of ErbB/Akt (Wen et al., 2008) or inhibition of phosphatases PP1/PP2A (Morfini et al., 2004; Plattner et al., 2006). The data presented here strongly support Cdk5 as the arbitrator regulating the phospho-dynamics of a GSK3/PRKAG2/AMPK cascade, which determines the proliferative state of PC cells. It has been suggested that GSK3 interacts with the AMPKβ regulatory subunit and can inhibit AMPKα through phosphorylation at Thr479 (Suzuki et al., 2013). However, this mechanism is somewhat confounded by report of simultaneous activation of AMPKα and reduction of inhibitory Ser9 GSK3 phosphorylation in neuronal models (King et al., 2011). Thus, the relationship between GSK3 and AMPK may be context dependent. We identified a functional role for phospho-Ser65 PRKAG2 as a prerequisite to Thr172 AMPKα phosphorylation. In agreement, several mutations in PRAKG2, such as K475E and R302Q, lead to increased AMPK activity associated with cardiomyocyte hypertrophy (Xu et al., 2017; Zhan et al., 2018). AMPK activationcoupled PRKAG2 nuclear translocation also promotes cardioprotection against ischemic injury (Cao et al., 2017) and is likely followed by interactions between PRKAG2-AMPK-LKB1 (Xie et al., 2008). Considering these findings, a cardinal observation of our study is reporting an isoform-specific function for PRKAG2 in cancer. AMPK-mediated p53 activation is known to cause premature senescence in response to energetic stress. Moreover, AMPK-induced metabolic activation of p53 was not affected by the inhibition of ATM kinase, while loss of p53 negated AMPK-induced effects (Jones et al., 2005). Here, Cdk5in-induced p53 activation appears to be regulated by the metabolic state of the cell where a coordinated increase in P-Ser15 p53 and P-Thr172 AMPK was marked by senescence-like cell morphology. Mitochondrial ROS, DNA damage, and histone methylation are the main drivers of cellular senescence. However, modulation of stress-inducible kinases such as p38MAPK (Freund et al., 2011), AMPK, and mTOR/PTEN (Jung et al., 2019) can also mediate cellular senescence signals independent of DNA damage. Furthermore, AMPK-p53 activation incites anti-glycolytic effects (Thoreen and Sabatini, 2005), consistent with the ability of AMPK to negatively regulate glycolysis as a feature of its anti-tumor effects (Faubert et al., 2013). Interestingly, inhibiting Cdk5 attenuates glycolysis, suggesting an active role for this kinase in aerobic glycolysis while maintaining low levels of AMPK-p53 signaling. In summary, this study reveals a phospho-dynamic mechanism where a Cdk5/GSK3/PRKAG2-AMPK/p53 signaling cascade acts as a critical downstream effector of SDHB loss. We demonstrated key components of this cascade across cell-based and in vivo models as well as in human tumors. We also derived a clinically accurate model of PC by transgenically invoking aberrant Cdk5 activity. These findings serve as a mechanistic rationale to utilize combinations of Cdk5 inhibitors and AMPK agonists in tumors driven by Cdk5/AMPK-dependent metabolic checkpoints. Limitations of the study Our study uncovers an important signaling cascade, presenting a molecular hub of potential anti-cancer targets. However, there are a few missing links that will require further exploration. We show activation of aberrant Cdk5 downstream of SDHB-succinate-Ca2+ signaling via SUCNR1 as one crucial tumorigenic signaling mechanism in PC. However, it remains unclear precisely how SUCNR1 activation mediates these effects. Further delineation of the signaling steps between SUCNR1 and Ca2+/calpain/p25 is warranted. Second, TCGA analysis indicates a significant negative correlation between SDHB and Cdk5R1 in contrast to other subunits. This raises the question of why loss of SDH through different subunit genes causes distinct functional outcomes on the tumorigenic potential of PC patients. The causes of this selective negative correlation between SDHB and Cdk5R1 need further study. There are also likely additional functional aspects of PRKAG regulation to understand. The precise molecular mechanism by which GSK3-dependent Ser65 PRKAG2 phosphorylation mediates Thr172 AMPK phosphorylation remains to be understood, as do any additional features of PRKAG2-AMPK regulation. Finally, it remains to be fully explained how Cdk5 inhibition induces premature cellular senescence. Mechanistically, senescence is a complex orchestration and temporal coordination of numerous cell-cycle regulators. The molecular markers for tracking senescence in the context of SDHx tumors may be context, cell type, and species dependent and should be further studied. STAR★METHODS RESOURCE AVAILABILITY Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, James A. Bibb (jbibb@uabmc.edu). Materials availability This study did not generate new unique reagents. Data and code availability All data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. EXPERIMENTAL MODELS AND SUBJECT DETAILS Human tumor samples Human Pheochromocytoma (PC) were acquired from tumor bank of University of Alabama Birmingham, and Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) following institutional review board (IRB) regulations, office for human research protections, NIH guidelines for research involving human subjects, and the health insurance portability and accountability act. All samples were de-identified, coded with no patient data, stored at −80°C until required for assay. Normal adrenal medullae were obtained from cadaveric kidney transplant donors from University of Alabama Birmingham (used as controls for which patients have given consent). Cell lines and cell culture The human progenitor PC cell line, hPheo1 (Ghayee et al., 2013) and mouse MTT PC cells were used (Korpershoek et al., 2012). SDHB was deleted from parent hPheo1 using CRISPR/Cas9 technology The goal was to achieve 80–90% KO of SDHB where single KO clones were selected and, subcloned multiple times. The resultant cell line confirmed >80% KO efficiency with a strong glycolytic phenotype in comparison to the uncloned cells. Cells were cultured in RPMI 1640 with 10% FBS (Gibco). HEK293, HeLa, MDAMB-231, and H1184 were maintained in the ATCC-recommended culture. Pheochromocytoma animal model All animal research was approved by the University of Alabama Birmingham (UAB) Institutional Animal Care and Use Committee (IACUC). Mice were genotyped and maintained within UAB Animal Resource Program Facility (ARP) and the animal research facilities of the Department of Surgery. Bitransgenic mice- Both male and female wild-type C57BL/6 mice, ages eight to twelve weeks were utilized for all the experiments. PiggyBac technology was used to generate single-copy PNMT–tTA transgenic mice (Cyagen Biosciences). Transgene positive offsprings were confirmed by genotyping four weeks old pups (primer sequence given in Key Resource Table; annealing temp– 60°C; Product– 373 bp). All positive offsprings were confirmed by PCR to not contain any integration of the helper plasmid. Primers used in the PCR to test for helper plasmid integration, Forward– CTGGACGAGCAGAACGTGATCG; Reverse- CGAAGAAGGCGTAGATCTCGTCCTC. Bi-transgenic mice were generated by crossing eight weeks old Tet-Op-p25GFP (Meyer et al., 2008) with that of PNMT-tTA. Transgenes were confirmed by genotyping for PNMT and p25-GFP alleles while control littermates were positive only for PNMT-tTA (Figure S6). Mice were treated with water containing 0.1 g/L doxycycline as required and all experimental mice were group-housed 12 h light/dark cycle with access to food and water ad libitum. METHOD DETAILS Database mining SDHB copy number alteration, mRNA expression and co-expression analysis were analyzed in cBioportal Cancer Genomics Database (http://www.cbioportal.org) (Gao et al., 2013). According to cBioportal definitions, genomic alterations are defined as: −2 or deep deletion equivalent to homozygous deletion; −1 or shallow deletion indicative of heterozygous deletion; and 0 indicative of diploid with no alteration. UALCAN online TCGA transcriptomic database was used to compare expression levels of SDHB/Cdk5R1 between normal human medulla and PCPG patients (http://ualcan.path.uab.edu/) (Chandrashekar et al., 2017). Publicly available Gene expression Omnibus (GEO) dataset GSE19422 was used for data mining in Figure S5. ShinyGO v0.61 gene-set enrichment tool was used to derive Gene Ontology categories, enriched pathways and graphical pathway network trees (Ge et al., 2020). Gene expression analysis of PRKAG2, AMPKα, (Figure S9D), Cdk5, Cdk2 and Cdk1 (Figure S10A) were performed in GEPIA (The Gene Expression Profiling Interactive Analysis) database (Tang et al., 2017). Seahorse XF96 metabolic flux analysis Real time extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) in cells were determined using the Seahorse Extracellular Flux (XFe-96) analyzer (Seahorse Bioscience, MA, USA). 1 × 104 cells were seeded per well of XF96 cell culture plates and incubated for 24 h to allow adherence. Bioenergetic profile was determined by mitochondrial and glycolytic stress tests following the manufacturer’s protocol. The following day cells were washed with pre-warmed XF assay base media (note: for OCR measurement, assay media was supplemented with 10 mM glucose, 1 mM Pyruvate, 2 mM L-glutamine, 5 mM HEPES and adjusted at 7.4 pH; for the glycolysis stress test, cells were washed with glucose-free XF base media). No signs of cytotoxicity, cell detachment or variations in the cell density noticed between the wells. Cells were maintained in final volume of 180 μL/well of media at 37°C, in a non-CO2 incubator for 1 h. Meanwhile, we loaded the cartridge ports with effectors. [For glycolysis stress test–12.5 mM Glucose, 1.5 μM Oligomycin, 50 μM 2-DG. For Mito stress test– 1.5 μM Oligomycin, 1 μM FCCP, 0.5 μM Antimycin and 0.5 μM Rotenone]. Linear correlation between protein concentration and viable cell number was confirmed. Values normalized by CyQUANT DNA quantification (#C35014, Thermo Fisher Scientific) and protein content (BCA assay) was comparable. Data were analyzed using XF96 Wave software and GraphPad Prism. Ca2+ measurement and live cell imaging Cells were loaded with cell permeant intracellular Ca2+ flux indicator + in HBSS (Gibco) for 30 min following manufacturer’s instruction. Time-lapse live cell imaging was performed using Nikon A1R HD25 inverted confocal microscope equipped with perfect focus system. Cell imaging chambers were maintained in 5% CO2 and 37°C. Images were capture with Plan Apo λ 20× NA 0.8 wd 1000 objective, frame size– 1024 × 1024, scan speed– 2, time loop– 31, imaging zoom– 1.192, resolution– 3.7 pixels per micron, frame interval – 10 sec, bits per pixel–16, ex/em– 494/525. Pinhole and laser power settings were adjusted based on pre-stimulation background levels. Fiji software (Schindelin et al., 2012) was used to select and quantify ROIs and individual object integrated density (IntDen) values were calculated for each cell using the formula (IntDen- (area of selected cell × mean field background intensity). Individual kinetic profile of [Ca2+]i for WT and SDHB KO cells were generated by plotting fluorescence intensity as a function of time in seconds as described previously (Koh et al., 2016). Fluorescence resonance energy transfer (FRET) CMV pCalpain-sensor (Addgene #36182) composed of eCFP (donor) and eYFP (acceptor) linked to calpain cleavage site (GSG-QQEVY GAMPRDGSG) where inactive calpains = High FRET and vice versa). pCalpain-sensor transfected in PC cells using Fugene HD according to the manufacturer’s instruction (Promega). Donor and acceptor bleed through were corrected using eCFP and eYFP only fluorophores. FRET measurements were performed using Nikon C2 confocal microscope (Plan Apo 60x Oil λS DIC N2) attached to a stage- top live-cell incubator to maintain cells at 37°C with 5% CO2 (Tokai Hit environmental chamber). FRET imaging based on sensitized emission were performed to acquire images with following filter combinations: donor (eCFP, ex-430±20, em-485±20); acceptor (eYFP, ex- 480, em-535±25); and FRET channel with a 435 nm (ex) and 535 nm (em). Visualization of FRET index and quantitation were performed as described in the FRET analyzer image J plug-in (Hachet-Haas et al., 2006). Immunostaining and tissue microarray Each human PC tumor case was thoroughly reviewed, formalin-fixed, and paraffin-embedded blocks were acquired within Department of Surgery, University of Alabama at Birmingham, and Dept. of Pathology, NICHD followed by the protocol described previously using DAKO immunohistochemistry kit (Pozo et al., 2013). Human adrenal tumor tissue microarray (US Biomax Inc.) was de-waxed at 60°C for 2 h followed by standard IHC protocol. Primary antibodies used included those for Cdk5 (Rockland), -p35/25 (Cell Signaling), ChrA (Abcam), anti-tyrosine hydroxylase (Abcam), and GFP (Cell Signaling). Secondary antibody alone was used as negative control. Quantitative analysis of DAB stained images were performed by using optical density (color deconvolution algorithm) within IHC profiler plug-in compatible with ImageJ digital image analysis software (Varghese et al., 2014). Blood pressure Mice blood pressure measurements were evaluated using CODA noninvasive BP system (a tail-cuff Method, Kent Scientific Corporation) as described previously (Wang et al., 2017). Assessment of Cdk5 inhibitors in vivo Xenografts- 6–7 week-old C.B-Igh-1b/IcrTac-Prkdcscid mice (Taconic Biosciences) were used for xenografts as previously described (Rai et al., 2020). 2 × 106 hPheo1 tumor cells (WT and SDHB KO) were injected subcutaneously in the right flank of the mice. When average tumor volume reached 150 mm3, mice were divided into two groups and injected intraperitoneally with vehicle or MRT3–007 (0.5 mg/kg every alternate day for three weeks). Body weights and tumor diameters were measured 3 times/week and tumor volumes were calculated using the formula V = ab2 × 0.52, where a and b are major and minor axes of the tumor foci, respectively. The experiment was terminated on day 25, and the tumors were harvested for biochemistry and histological assessment. Allograft and in vivo imaging- 6–7 week old nude mice (nu/nu) (Jackson laboratory) were used for allograft assay described previously (Korpershoek et al., 2012). 1 × 106 MTT luciferase expressing cells were injected via tail vein and imaged one-week post-injection via bioluminescence imaging (Xenogen IVIS). Cohorts bearing comparable size of allografts received IP injection of substrate d-Luciferin (250 μL; 3.75 mg, Caliper Life Science, Hopkinton, MA, USA) 12 min before the whole-body imaging. Data was acquired and analyzed using the Live Imaging software version 3.0 (Caliper Life Science). Fourteen days after injecting tumor cells, all mice were divided into three groups, treated with Vehicle, MRT3–007 (0.75 mg/kg) or Roscovitine (50 mg/kg) every other day for 2 weeks. Animals were re-imaged to measure metastatic lesions via bioluminescence imaging. Plasmids, site directed mutagenesis and transfections Lentivirus gene expression vector (3rd generation; pLV-EGFP:T2A:Puro-EF1>ORF/FLAG) was used for subcloning ORF’s of PRKAG2 (NM_016203.3), DEPTOR(NM_022783.3), SDPR (NM_004657.5), AAK1(NM_014911.3) and UVRAG (NM_003369.3) driven by EF1A promoter (Vector Builder, Cyagen Biosciences). Cdk5 wild-type and kinase dead plasmids were described previously (Pozo et al., 2013). DNA modifications were performed in lentiviral pLV[Exp]-EGFP:T2A expression plasmids cloned by VectorBuilder, Cyagen Biosciences to generate phospho- and dephospho-mimetics for PRKAG2 (S-65D/E/A), DEPTOR (S-244D/E/A), SDPR(S-288D/E/A), AAK1(S-676D/E/A) and UVRAG (T-518D/E/A) using Q5-site directed mutagenesis Kit (NEB). Manufacturer’s instructions for mutagenic primer design were followed, and mutations were confirmed by DNA sequencing. Transfections were carried out using FuGene-HD (Promega). Transfection solutions were prepared in Opti-MEM using 1:3 ratio of plasmid DNA to Fugene transfection reagent. Cell growth assay Cells were seeded in 6-well plates at the density of 1 × 105, transfected with phospho- or dephospho-mimetics and control vectors. Cell growth was determined 48 h post-transfection by dual fluorescence acridine orange/propidium iodide (AO/PI) viability staining. Total cell number was counted using Cellometer Auto 2000 cell viability counter (Nexcelom) and normalized with total number of GFP-expressing cells. Dose-response curves were generated by cell viability tests using Cell Counting Kit-8 (CCK-8) (Dojindo). Assays were performed in five replicates and repeated at least 3 times. Phosphorylation state specific antibody generation Production and affinity purification of phosphorylation state specific polyclonal antibodies were performed as described previously (Hemmings, 1997). Phospho-Ser65 PRKAG2 were raised against a synthetic oligopeptide encompassing the amino acid sequence (RKVDS*PFGC). Cysteine containing phosphopeptide was conjugated to carrier protein Limulus hemocyanin using hetero-bifunctional crosslinker m-maleimidobenzoyl-N-hydroxysulfosuccinimide ester (Sigma# 803227). This conjugate was used to immunize New Zealand white rabbits (Charles River Laboratories). Preimmune sera were obtained and booster injections of 150 μg phosphopeptide conjugate were given at 2, 4, 6 and 8 weeks. Blood was collected at weeks– 5, 7, 9,11,13 and 14. The specificity of the antibodies in anti-serum was characterized by dot blot analysis using dephospho- and phospho-PRKAG2 standards. Phosphorylation-specific antibodies were purified using affinity purification method (Hemmings, 1997). Immunoblotting, cell cycle analysis, and in vitro phosphorylation Antibodies to the following phosphorylation sites and proteins were used: phospho-Thr172 AMPK, phospho-Ser79 ACC, phosphoSer21/9 GSK3, phospho-Ser16 p53, p27Kip1, phospho-Ser139 H2Ax from Cell Signaling Technology; and anti-p16INK4A, antispectrin and, anti-GAPDH, anti-actin from Thermo Fisher Scientific. SDS-PAGE and immunoblotting were conducted as previously described (Bibb et al., 1999). Following membrane blocking, immunoblots were incubated with primary and fluorescent secondary antibodies IRDye® 800CW, and IRDye® 680 (goat anti-rabbit or goat anti-mouse) from LI-COR and visualized with Odyssey CLx Imaging System (LI-COR Biosciences, NE). Thereafter, immunoblots’ signal intensity was computationally quantified and analyzed using ImageStudio software (LI-COR Biosciences–GmbH, www.licor.com). For cell cycle analysis, MRT3–007 treatment or phosphomimetics transfected cells were stained with 50 μg/mL of propidium iodide and analyzed for cell-cycle distribution as described previously (Erba et al., 1989). In vitro phosphorylation and immunoprecipitation-kinase assays were performed using optimized protocols described previously (Bibb et al., 1999; Pozo et al., 2013). Magnetic resonance imaging MRI experiments were performed using a Bruker Biospec 9.4 Tesla scanner with Paravision 5.1 software (Bruker Biospin, Billerica, MA). A Bruker 72 mm volume coil was used for signal excitation, with a 24 mm surface coil for reception (Doty Scientific Inc., Columbia, SC). Mice were anesthetized with isoflurane gas and respiration observed with an MRI-compatible physiological monitoring system (SA Instruments Inc., Stony Brook, NY). Animals were imaged in supine position on an animal bed with integrated circulating heated water to maintain temperature during the experiment. Scout images were collected in the axial, sagittal, and coronal planes to confirm animal positioning and coil placement. A 2D T2-weighted fast spin echo sequence was used for imaging of kidney and adrenal gland areas. Prospective respiratory gating was enabled to minimize motion artifacts. The following imaging parameters were used: TR/TE = 2000/25 ms, echo spacing = 12.5 ms, ETL = 4, 4 averages, 23 contiguous coronal slices with 0.5 mm thickness, FOV = 30 × 30 mm, and matrix = 300 × 300 for an in-plane resolution of 100 μ. All MRI images were obtained in the DICOM (Digital Imaging and Communications in Medicine) format and were imported into the image processing ITK software to obtain tumor volumes and perform 3D reconstructions. Mean tumor volumes were measured by drawing regions of interest (ROI), to circumscribe the entire tumor. QUANTIFICATION AND STATISTICAL ANALYSIS Data were analyzed by Student’s t-test for comparison of two groups or one-way ANOVA combined with Tukey’s post hoc test for multiple comparisons using Prism 8 version 8.4.2 (Graph Pad Software). The number of experimental replications or number of animals are represented as n, and the definitions of center and precision measures i.e, mean ± SD or mean ± SEM are indicated in the figure legends. p values <0.05 were considered statistically significant, reported as *< 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 in the figure legends. p values >0.05 were considered not significant (n.s.). Supplementary Material 1 ACKNOWLEDGMENTS We thank S. Thomas for assistance with MRI, E. Daniel for technical expertise, S. Barnes for LC-MS succinate measurements, D. Pollack for assistance in measuring mouse blood pressure, L. Bibb and SouthernBiotech for phosphorylation state-specific antibodies, and Pfizer for CP681301. This research was supported by the SDHB Para/Pheo Coalition and the Neuroendocrine Tumor Research Foundation (NETRF, J.A.B.). Portions of this work were also facilitated by the NIH (MH116896, MH126948, J.A.B.), the Robert E Reed Gastrointestinal Oncology Research Foundation, an American Cancer Society Institutional Research Grant Junior Faculty Development Award (S.R.), and Postdoctoral Fellowship (A.M.C.). This research was supported, in part, by Intramural Research Program of the Eunice Kennedy Shriver NICHD, NIH (K.P.); the Gatorade Trust through funds by the University of Florida, Department of Medicine (H.K.G.); and by a Eurostars grant (CYST-ARREST, L.M.). Studies were also supported by the National Cancer Institute Cancer Center Support Grant P30 CA013148 (UAB O’Neal Comprehensive Cancer Center). It was also supported by S10 OD028498-01 (UAB Preclinical Imaging Shared Facility). Some chemical synthesis was also supported by NCI SPORE P50 CA127297 (A.N., Eppley Institute for Research in Cancer and Allied Disease). This work was also supported in part by grant 1145265 from the National Health and Medical Research Council (NHMRC, to J.S.O.), St Vincent’s Institute of Medical Research (Australia), and the Victorian Government’s Operational Infrastructure Support Program. We are grateful to the UAB Diabetes Research Center (NIH P30 DK-079626) for providing core services. Figure 1. Alterations of SDHx genes in pheochromocytoma (A) Oncoprint depicts alterations in SDHx complex genes in patients. Genetic alterations are indicated by color codes. (B) Expression heatmap of SDHx complex genes. (C) Bar plots presenting copy number alteration and mRNA expression of SDHB in patients with PC tumors (del., deletions; alter., alterations). (D) Boxplots of SDHB gene expression in PCPG tumors (n = 179) versus normal adrenal (n = 3). (E) Exemplary immunoblot showing SDHB protein expression in WT versus SDHB KO hPheo1. (F) Glycolytic profile of WT versus SDHB KO hPheo1 cells indicated as extracellular acidification rate (ECAR). Quantitative ECAR analysis indicated by glycolytic capacity, glycolysis, and glycolytic reserve. (G) Oxygen consumption rate (OCR) measured in WT and SDHB KO cells. Bar plots show OCR analysis indicated as basal and ATP-linked respiration, n = 3, 10 replicates per group. Values are means ± SEM, ***p < 0.001, ****p < 0.0001 compared by Student’s t test. Figure 2. SDHB loss activates a succinate-Ca2+-calpain-Cdk5 cascade (A) [Ca2+]i activity reported by time-lapse live-cell imaging of cells loaded with Fluo-4 AM, images acquired pre or post stimulation with ionomycin (10 μM). Representative pseudo-colored images are shown with time course of fluorescence intensity quantification as %[Ca2+]i, scale bar, 80 μm. (B) Liquid chromatography-mass spectrometry (LC-MS) quantitation of succinate in WT and KO cell extracts. (C) Quantitation of fluorescence intensity and photomicrographs showing effects of ionomycin on %[Ca2+]i in WT hPheo1 cells treated with dimethyl succinate (DMS, 2 mM) or controls. (D) Time-lapse measurement of [Ca2+]i in SDHB KO cells pretreated with SUCNR1 antagonist, NF-56-EJ40, with representative images, scale bar, 50 μm (E) Schematic workflow of calpain sensor, and representative pseudocolor FRET map and normalized FRET index of WT and SDHB KO hPheo1. (F) Immunoblots of Cdk5, p25/p35 in WT and SDHB KO cells with quantitation. (G) 3D surface plot of immunostained cells comparing relative levels of p35/25 between WT and KO hPheo1, scale bar, 80 μm. (H and I) Representative confocal 3D photomicrographs and quantitation showing p35/25 expression in WT hPheo1 treated with increasing concentrations of succinate as indicated. (J) Quantitative immunoblotting of Cdk5/p35/p25 levels in tumor lysates derived from WT (n = 4) and KO (n = 6) xenografts. Values are means ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001; n.s., non-significant compared by t test and/or one-way ANOVA. (K) Schematic illustrating the mechanism of action mediated via succinate accumulation in chromaffin cells. Figure 3. SDHB and Cdk5 coactivator inversely correlate in human PC (A) Boxplots of Cdk5R1 in PCPG tumors (n = 179) versus normal adrenal (n = 3). (B) Correlation between SDHB and Cdk5R1, data quarried from cBioportal PCPG dataset; (R is Spearman coefficient; p = 6.78e). (C) Quantitation and representative immunoblots showing protein levels of SDHB, Cdk5, and p35/25 in human sporadic PC (n = 11) compared with normal medulla (n = 5). (D) Expression analysis of SDHB, Cdk5, p35/25 in human SDHB mutant tumors (n = 7) versus normal medulla (n = 7). (E) Hematoxylin eosin (HE) and immunostains of Cdk5 and p35/p25 in human PC, scale bar, 50 μm. (F) Histological assessment of Cdk5 and p35/25 in tissue microarray sections of adrenocortical adenoma (ACA, n = 40) and PC (n = 30); quantification presented as optical density; *p < 0.05, **p < 0.01, ***p < 0.001 compared using t test with Welch’s correction; n.s., non-significant. Figure 4. Aberrant Cdk5 develops PC in bitransgenic mice (A) Schematic showing design of PiggyBac PNMT expression system to generate Dox-ON/OFF bitransgenic model. (B) T2-weighted representative MRI coronal images indicating temporal changes in adrenal gland size in WT versus p25-ON mice. (C) MRI quantitation of adrenal size in p25-ON mice (n = 11) compared with WT (n = 6). (D) Immunoblots showing expression of p25-GFP, Cdk5, TH, and ChrA in tissue lysates derived from p25-ON or WT adrenal glands. (E) Histological assessment of NE markers in WT adrenals or PC tissue sections, scale bar 100 μM. (F–H) Measurements of plasma nor-metanephrines (F), metanephrines (G), and tail-cuff blood pressure (H); (n = 6–11), *p < 0.05, **p < 0.01 compared by using t test with Welch’s correction. Figure 5. Characterization of functional Cdk5 targets in human PC (A) Schematic showing NE tumors from p25-ON/OFF model analyzed for potential Cdk5 phosphorylation sites; function of phosphosite ratio (log2 fold change) between p25-ON and -OFF tumors is plotted against p values. (B) Table listing the Cdk5 phospho-targets downregulated in growing p25-ON tumors. (C) Graphical presentation of network tree highlighting enriched pathways associated with downregulated Cdk5 targets; each node represents enriched pathway where two nodes or pathways are connected if they share 20% or more genes; bigger nodes indicate larger gene sets. (D) Table summarizing phosphosites selected to evaluate their effects on PC cellular proliferation. (E) Effects of phospho- and dephospho-mimetics (D/E/A) on cell growth determined by dual fluorescence acridine orange/propidium iodide (AO/PI) viability staining. EV, empty vector. (F) Schematic of AMP-activated protein kinase (AMPK) heterotrimeric complex composed of α, β, and γ subunits. Phosphorylation site, S65, is located in PRKAG2 subunit of the trimeric complex. (G) Immunoblot detection of P-T172 AMPK in response to PRKAG2 S65D/E/A phosphomutant overexpression, n = 4. Values are means ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001, n.s. non-significant, one-way ANOVA multiple comparisons with Tukey’s method. (H) Quantitative immunoblotting of P-S65 PRKAG2 and -T172 AMPKα normalized to total PRKAG2 and -AMPKα in human PC tissues (n = 9) versus normal adrenal medulla (n = 5). Values are means ± SEM, *p < 0.05, ***p < 0.001, Student’s t test. (I) Correlation plot between P-S65 PRKAG2 and -T172 AMPK in patients with PC; r is Spearman’s correlation coefficient. Figure 6. Targeting Cdk5 to regulate the P-PRKAG2/P-AMPK cascade (A) In vitro phosphorylation of recombinant PRKAG2 by GSK3, with time-dependent 32P incorporation and Coomassie-Brilliant Blue (CBB) stained protein shown with stoichiometry. (B) Immunoblots showing GSK3 but not Cdk5 phosphorylates Ser65 PRKAG2 in vitro. (C) Immunoblots of recombinant AMPK holoenzyme trimeric complex (α1β1γ2) phosphorylated by GSK3β. (D) Effects of ectopic expression of WT versus kinase-dead (KD) Cdk5 on the levels of phosphorylation sites as shown. (E) Dose-dependent effects of GSK3 (SB216763, 24 h) versus CDK5 inhibition (CP681301, 24 h) on SDHB KO hPheo1 cell viability. (F) Plot showing dose-dependent effect of AMPK activator AICAR on cell viability (24 h). (G) Time-dependent effects of AMPK activators, metformin (20 mM), 2-deoxy-D-glucose (2-DG, 20 mM), and AMPK inhibitor, compound C (CC; 10 μM) on KO hPheo1 cell viability. n = 3, values are mean ± SEM, *p < 0.05, **p < 0.01 compared by one-way ANOVA. (H) Schematic of signaling mechanism showing aberrant CDK5-GSK3β crosstalk and deregulation of downstream phospho-dynamics of AMPK pathway. (I) Chemical structures of Cdk5 inhibitors, as indicated. (J) Dose-response effects of five different Cdk5 inhibitors on cell viability of hPheo1. n = 4. (K) Glycolytic profile of SDHB KO cells treated with or without MRT3–007 [Cdk5in], 25 nM for 12 h (left). Plots comparing basal glycolysis rate, glycolytic capacity, and glycolytic reserve between control versus Cdk5in (right). (L) Bioenergetic phenotype of KO cells in response to Cdk5in. Values are means ± SD, *p < 0.05, Student t test, n = 2 (10 replicates per group). (M) Immunoblot quantification showing time-dependent effect of Cdk5in on phosphorylation states as indicated, n = 3, values presented as fold change normalized with time = 0. (N) Immunoblot analysis of the impact of GSK3 inhibitor, SB216763 (5 μM, pretreatment 10 h) on Cdk5in-induced phosphorylations. (O) hPheo1 pre-incubated with or without CC (10 μM), immunoblot quantitation comparing effects of Cdk5in alone or in combination with CC. Figure 7. Cdk5in induces senescence-like phenotypic characteristics (A) Scanning electron microscopic images of hPheo1 cell morphology in control versus Cdk5in (Indolinone A). (B) Imaging of proliferating cells and those treated with Cdk5in (20 nM, 48 h) for common senescence markers. Representative confocal photomicrographs and quantitation show phalloidin stain of F-actin, p16INK4a (inset is senescence-associated β-gal), p27Kip, and P-H2AX, respectively; scale bar, 136 μm. (C) Confocal images of Ki67 staining and bar graph show percentage of Ki67-positive cells. (D and E) SDHB KO xenografts treated with vehicle (Con, n = 8) or Cdk5in (n = 10) (0.5 mg/kg) were analyzed for tumor volume (D), followed by histological (HE) and Ki67 expression analysis in xenograft tissues (E). (F and G) Efficacy of Cdk5 inhibition tested on a metastatic allograft PC model, examined via in vivo bioluminescent imaging. Representative images of vivisected liver and tumor spread indicated by histological analysis and photon flux quantitation; n = 4. (H and I) Representative coronal MRI images and quantitation showing changes in adrenal gland size of mice over time under Dox-ON/OFF conditions. (J) Quantitation of p25GFP expression determined by immunoblotting. KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Rabbit anti-Cdk5 Rockland Cat# 200-301-163; RRID: AB_11182476 Rabbit anti-SDHB Abcam Cat#154974; RRID: N/A Rabbit anti-ChrA Abcam Cat#15160; RRID: AB_368477 Rabbit anti-Tyrosine Hydroxylase Abcam Cat#6211; RRID: AB_2240393 Rabbit anti-GFP Cell Signaling Technology Cat#2956; RRID: AB_1196615 Rabbit anti-p35/25 Cell Signaling Technology Cat#2680; RRID: AB_1078214 Rabbit anti-phospho-Thr172 AMPKα Cell Signaling Technology Cat#2535; RRID: AB_331250 Rabbit anti-AMPKα Cell Signaling Technology Cat#2532; RRID: AB_330331 Rabbit anti-phospho-Ser79 ACC Cell Signaling Technology Cat#11818; RRID: AB_2687505 Rabbit anti-ACC Cell Signaling Technology Cat# #3662; RRID: N/A Rabbit anti-phospho-Ser21/9 GSK3 Cell Signaling Technology Cat# 9331; RRID: AB_329830 Rabbit anti-GSK3 Cell Signaling Technology Cat#9315; RRID: N/A Rabbit anti-phospho-Ser15 p53 Cell Signaling Technology Cat#9284; RRID: AB_331464 Mouse anti-p53 Thermo Fisher Scientific Cat# MA5-12557; RRID: AB_10989883 Rabbit anti-p27Kip1 Cell Signaling Technology Cat#3686; RRID: AB_2077850 Rabbit anti-phospho-Ser139 H2Ax Cell Signaling Technology Cat# 9718; RRID:AB_2118009 Rabbit anti-p16INK4A Droteintech Cat#10883-1-AP; RRID: AB_2078303 Mouse anti-Spectrin Millipore Sigma Cat#MAB1622; RRID: AB_94295 Mouse anti-GAPDH Thermo Fisher Scientific Cat# 39-8600, RRID: AB_2533438 Mouse anti-β actin Thermo Fisher Scientific Cat#PA5-78715; RRID: AB_2745831 IRDye® 800CW Goat anti-Rabbit or anti-Mouse IgG LI-COR Biosciences Cat#926-32211; RRID: AB_621843, Cat# 926-32210; RRID: AB_621842 IRDye® 680RD Goat anti-Rabbit or anti-Mouse IgG LI-COR Biosciences Cat# 926-68071; RRID:AB_10956166, Cat# 926-68070; RRID: AB_10956588 Biological samples Human adrenal tumor tissue microarray US Biomax Inc. Cat#AG801 Human PC tissues Tumor bank (University of Alabama Birmingham); Dr.Karel Pacak [Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)] N/A Chemicals, peptides, and recombinant proteins Roscovitine Dr. Laurent Meijer ManRos Therapeutics MRT3-007 Dr. Laurent Meijer ManRos Therapeutics MRT3-124 Dr. Laurent Meijer ManRos Therapeutics CR8 Dr. Laurent Meijer ManRos Therapeutics 25-16 Dr. Amarnath Natarajan (University of Nebraska) N/A CVT-313 Selleck Chemicals Cat#S6537 Metformin Sigma CAS#1115-70-4 AICAR Sigma CAS#2627-69-2 NF-56-EJ40 Med Chem Express Cat#HY-130246 2-DG Sigma Aldrich Cat# D6134 Sulfo-MBS Sigma Aldrich Cat# 803227 P-S65 PRKAG2 peptide - RKVDS*PFGC University of Texas Southwestern N/A HBSS Gibco Cat#14175 Gibco™ Fetal Bovine Serum Fisher Scientific Cat# 10-082-147 Doxycycline Sigma Aldrich Cat# D9891 Fluo-4AM Thermo Fischer Scientific Cat# F14201 FuGENE® HD Transfection Reagent Promega Cat# E2311 Critical commercial assays 2-MET Plasma ELISA Fast Track Rocky Mountain Diagnostics Cat#BA E-8300 Cell Counting Kit-8 Dojindo Molecular Technologies Cat#CK04-05 Q5-site directed mutagenesis Kit New England Biolabs Cat# E0554S DAKO Immunohistochemistry Kit Agilent Technologies, Inc. Cat# K064030-2 Deposited data Library of Cdk5 Phosphosites Carter et al. (2020) https://www.phosphosite.org/Supplemental_Files.action Experimental models: Cell lines hPheo1 (Wild-type, SDHB knockout) Dr. Hans Ghayee (University of Florida) Developed by Ghayee laboratory MTT cells Courtesy of Dr. Karel Pacak (NICHD) Martiniova et al. (2009) Experimental models: Organisms/Strains Mouse: PiggyBac PNMT-tTA Cyagen Biosciences N/A Mouse: TetOp - P25GFP Meyer et al. (2008) N/A Mouse: PNMT-tTA x TetOp-P25GFP Dr. James Bibb (University of Alabama at Birmingham) N/A Mouse: nu/nu Jackson laboratory JAX:002019 Mouse: C.B-Igh-1b/IcrTac-Prkdcscid Taconic Biosciences TAC:cb17sc Oligonucleotides Primer: PNMT-tTA Forwardl: CAGTAGTAGATAAAGGGATGGGGAG This paper N/A Primer: PNMT-tTA Reversel: GGGGCAGAAGTGGGTATGATG This paper N/A Primer: PNMT-tTA Forward2: CAGGAGCATCAAGTAGCAAAAGAG This paper N/A Primer: PNMT-tTA Reverse2: CACACCAGCCACCACCTTCT This paper N/A gRNA sequence: SDHB: ATGGCAAATTTCTTGATACG Fisher Scientific N/A Recombinant DNA pLV[Exp]-EGFP:T2A:Puro VectorBuilder, Cyagen Biosciences Cat#VB160420-1011mqh CMV pCalpain-sensor Addgene Cat#36182 Software and algorithms FIJI Schindelin et al., 2012 http://imagej.net/software/fiji/ cBioportal Gao et al., 2013 http://www.cbioportal.org ShinyGO v0.61 Ge et al., 2020 http://ge-lab.org/go/ GEPIA Tang et al., 2017 http://gepia.cancer-pku.cn/index.html UALCAN Chandrashekar et al., 2017 http://ualcan.path.uab.edu/ XF96 Seahorse Wave Agilent Technologies https://www.agilent.com/zh-cn/product/cell-analysis/real-time-cellmetabolic-analysis/xf-software/seahorse-wave-desktop-software-740897 ImageStudio LI-COR Biosciences https://www.licor.com/bio/image-studio/ GraphPad Prism GraphPad Software, Inc http://www.graphpad.com/scientific-software/prism/ Other CODA noninvasive BP system Courtesy of Dr. David Pollock, University of Alabama at Birmingham N/A Production of P-S56 PRKAG2 antiserum SouthernBiotech, Birmingham, AL N/A Magnetic Resonance Imaging Institutional Research Core, University of Alabama at Birmingham N/A Agilent Seahorse XFe96 Analyzers Bio-Analytical Redox Biology (BARB) Core, UAB Department of Nutrition Sciences https://www.uab.edu/shp/drc/cores/barb-core Bioluminescence imaging (IVIS Lumina III) Small Animal Imaging Shared Facility, Department of Radiology & Comprehensive Cancer Center, University of Alabama at Birmingham N/A Highlights Dysfunctional SDHB subunit causes aberrant activation of Cdk5 in pheochromocytoma (PC) Aberrantly activated Cdk5 dysregulates a GSK3/PRKAG2/AMPKα signaling cascade p25 overexpression in chromaffin cells and consequent aberrant Cdk5 activity causes PC Cdk5 inhibition activates AMPK/p53 axis to rescue senescence and block PC progression DECLARATION OF INTERESTS The authors declare no competing interests. SUPPLEMENTAL INFORMATION Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2022.111218. REFERENCES Anderson NM , Mucka P , Kern JG , and Feng H (2018). The emerging role and targetability of the TCA cycle in cancer metabolism. Protein Cell 9 , 216–237. 10.1007/s13238-017-0451-1.28748451 Andrienko TN , Pasdois P , Pereira GC , Ovens MJ , and Halestrap AP (2017). The role of succinate and ROS in reperfusion injury - a critical appraisal. J. Mol. Cell. Cardiol. 110 , 1–14. 10.1016/j.yjmcc.2017.06.016.28689004 Bardella C , Pollard PJ , and Tomlinson I (2011). SDH mutations in cancer. Biochim. Biophys. Acta 1807 , 1432–1443. 10.1016/j.bbabio.2011.07.003.21771581 Barros-Minones L , Martín-de-Saavedra D , Perez-Alvarez S , Orejana L , Suquía V , Goñi-Allo B , Hervias I , López MG , Jordan J , Aguirre N , and Puerta E . (2013). Inhibition of calpain-regulated p35/cdk5 plays a central role in sildenafil-induced protection against chemical hypoxia produced by malonate. Biochim. Biophys. Acta 1832 , 705–717. 10.1016/j.bbadis.2013.02.002.23415811 Baysal BE , and Maher ER (2015). 15 YEARS OF PARAGANGLIOMA: Genetics and mechanism of pheochromocytoma-paraganglioma syndromes characterized by germline SDHB and SDHD mutations. Endocr Relat Cancer 22 , T71–T82.26113606 Bettayeb K , Baunbæk D , Delehouze C , Loaëc N , Hole AJ , Baumli S , Endicott JA , Douc-Rasy S , Bénard J , Oumata N , (2010). CDK inhibitors roscovitine and CR8 trigger Mcl-1 down-regulation and apoptotic cell death in neuroblastoma cells. Genes Cancer 1 , 369–380. 10.1177/1947601910369817.21779453 Bhuniya D , Umrani D , Dave B , Salunke D , Kukreja G , Gundu J , Naykodi M , Shaikh NS , Shitole P , Kurhade S , (2011). Discovery of a potent and selective small molecule hGPR91 antagonist. Bioorg. Med. Chem. Lett. 21 , 3596–3602. 10.1016/j.bmcl.2011.04.091.21571530 Bibb JA , Snyder GL , Nishi A , Yan Z , Meijer L , Fienberg AA , Tsai LH , Kwon YT , Girault JA , Czernik AJ , (1999). Phosphorylation of DARPP-32 by Cdk5 modulates dopamine signalling in neurons. Nature 402 , 669–671. 10.1038/45251.10604473 Blank A , Schmitt AM , Korpershoek E , van Nederveen F , Rudolph T , Weber N , Strebel RT , de Krijger R , Komminoth P , and Perren A (2010). SDHB loss predicts malignancy in pheochromocytomas/sympathethic paragangliomas, but not through hypoxia signalling. Endocr. Relat. Cancer 17 , 919–928. 10.1677/ERC-09-0316.20702724 Broekman F , Giovannetti E , and Peters GJ (2011). Tyrosine kinase inhibitors: Multi-targeted or single-targeted? World J. Clin. Oncol. 2 , 80–93. 10.5306/wjco.v2.i2.80. Buffet A , Venisse A , Nau V , Roncellin I , Boccio V , Le Pottier N , Boussion M , Travers C , Simian C , Burnichon N , (2012). A decade (2001–2010) of genetic testing for pheochromocytoma and paraganglioma. Horm Metab Res 44 , 359–366.22517557 Bujard H (1999). Controlling genes with tetracyclines. J. Gene Med. 1 , 372–374. 10.1002/(SICI)1521-2254(199909/10)1:5<372::AID-JGM61>3.0.CO;2-T.10738554 Cao Y , Bojjireddy N , Kim M , Li T , Zhai P , Nagarajan N , Sadoshima J , Palmiter RD , and Tian R (2017). Activation of gamma2-AMPK suppresses ribosome biogenesis and protects against myocardial ischemia/reperfusion injury. Circ. Res. 121 , 1182–1191. 10.1161/CIRCRESAHA.117.311159.28835357 Carter AM , Tan C , Pozo K , Telange R , Molinaro R , Guo A , De Rosa E , Martinez JO , Zhang S , Kumar N , (2020). Phosphoprotein-based biomarkers as predictors for cancer therapy. Proc. Natl. Acad. Sci. USA 117 , 18401–18411. 10.1073/pnas.2010103117.32690709 Cascon A , Montero-Conde C , Ruiz-Llorente S , Mercadillo F , Letón R , Rodríguez-Antona C , Martínez-Delgado B , Delgado M , Díez A , Rovira A , (2006). Gross SDHB deletions in patients with paraganglioma detected by multiplex PCR: a possible hot spot? Genes Chromosomes Cancer 45 , 213–219. 10.1002/gcc.20283.16258955 Chandrashekar DS , Bashel B , Balasubramanya SAH , Creighton CJ , Ponce-Rodriguez I , Chakravarthi BVSK , and Varambally S (2017). UALCAN: a portal for facilitating tumor subgroup gene expression and survival analyses. Neoplasia 19 , 649–658. 10.1016/j.neo.2017.05.002.28732212 Chen J (2016). The cell-cycle arrest and apoptotic functions of p53 in tumor initiation and progression. Cold Spring Harb. Perspect. Med. 6 , a026104. 10.1101/cshperspect.a026104.26931810 Crespo-Biel N , Camins A , Pallàs M , and Canudas AM (2009). Evidence of calpain/cdk5 pathway inhibition by lithium in 3-nitropropionic acid toxicity in vivo and in vitro. Neuropharmacology 56 , 422–428. 10.1016/j.neuropharm.2008.09.012.18948125 Crespo-Biel N , Camins A , Pelegrí C , Vilaplana J , Pallàs M , and Canudas AM . (2007). 3-Nitropropionic acid activates calpain/cdk5 pathway in rat striatum. Neurosci. Lett. 421 , 77–81. 10.1016/j.neulet.2007.05.038.17566644 Cruz JC , Tseng HC , Goldman JA , Shih H , and Tsai LH (2003). Aberrant Cdk5 activation by p25 triggers pathological events leading to neurodegeneration and neurofibrillary tangles. Neuron 40 , 471–483. 10.1016/s0896-6273(03)00627-5.14642273 Dahia PL (2014). Pheochromocytoma and paraganglioma pathogenesis: learning from genetic heterogeneity. Nat. Rev. Cancer 14 , 108–119. 10.1038/nrc3648.24442145 Dahia PL , Ross KN , Wright ME , Hayashida CY , Santagata S , Barontini M , Kung AL , Sanso G , Powers JF , Tischler AS , (2005). A HIF1alpha regulatory loop links hypoxia and mitochondrial signals in pheochromocytomas. PLoS Genet. 1 , 72–80. 10.1371/journal.pgen.0010008.16103922 Eijkelenkamp K , Osinga TE , Links TP , and van der Horst-Schrivers ANA (2020). Clinical implications of the oncometabolite succinate in SDHx-mutation carriers. Clin. Genet. 97 , 39–53. 10.1111/cge.13553.30977114 Eisenhofer G , Lenders JW , Linehan WM , Walther MM , Goldstein DS , and Keiser HR (1999). Plasma normetanephrine and metanephrine for detecting pheochromocytoma in von Hippel-Lindau disease and multiple endocrine neoplasia type 2. N. Engl. J. Med. 340 , 1872–1879. 10.1056/NEJM199906173402404.10369850 Erba E , Ubezio P , Pepe S , Vaghi M , Marsoni S , Torri W , Mangioni C , Landoni F , and D’Incalci M (1989). Flow cytometric analysis of DNA content in human ovarian cancers. Br. J. Cancer 60 , 45–50. 10.1038/bjc.1989.217.2803914 Erez A , Shchelochkov OA , Plon SE , Scaglia F , and Lee B (2011). Insights into the pathogenesis and treatment of cancer from inborn errors of metabolism. Am. J. Hum. Genet. 88 , 402–421. 10.1016/j.ajhg.2011.03.005.21473982 Evenepoel L , Helaers R , Vroonen L , Aydin S , Hamoir M , Maiter D , Vikkula M , and Persu A (2017). KIF1B and NF1 are the most frequently mutated genes in paraganglioma and pheochromocytoma tumors. Endocr. Relat. Cancer 24 , L57–L61. 10.1530/ERC-17-0061.28515046 Faubert B , Boily G , Izreig S , Griss T , Samborska B , Dong Z , Dupuy F , Chambers C , Fuerth BJ , Viollet B , (2013). AMPK is a negative regulator of the Warburg effect and suppresses tumor growth in vivo. Cell Metab. 17 , 113–124. 10.1016/j.cmet.2012.12.001.23274086 Favier J , Brière JJ , Burnichon N , Rivière J , Vescovo L , Benit P , GiscosDouriez I , De Reyniès A , Bertherat J , Badoual C , (2009). The Warburg effect is genetically determined in inherited pheochromocytomas. PLoS One 4 , e7094. 10.1371/journal.pone.0007094.19763184 Fishbein L , Khare S , Wubbenhorst B , DeSloover D , D’Andrea K , Merrill S , Cho NW , Greenberg RA , Else T , Montone K , (2015). Whole-exome sequencing identifies somatic ATRX mutations in pheochromocytomas and paragangliomas. Nat. Commun. 6 , 6140. 10.1038/ncomms7140.25608029 Fliedner SM , Breza J , Kvetnansky R , Powers JF , Tischler AS , Wesley R , Merino M , Lehnert H , and Pacak K (2010). Tyrosine hydroxylase, chromogranin A, and steroidogenic acute regulator as markers for successful separation of human adrenal medulla. Cell Tissue Res. 340 , 607–612. 10.1007/s00441-010-0965-9.20440513 Freund A , Patil CK , and Campisi J (2011). p38MAPK is a novel DNA damage response-independent regulator of the senescence-associated secretory phenotype. EMBO J. 30 , 1536–1548. 10.1038/emboj.2011.69.21399611 Gao J , Aksoy BA , Dogrusoz U , Dresdner G , Gross B , Sumer SO , Sun Y , Jacobsen A , Sinha R , Larsson E , (2013). Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal. 6 , pl1. 10.1126/scisignal.2004088. Garcia D , and Shaw RJ (2017). AMPK: mechanisms of cellular energy sensing and restoration of metabolic balance. Mol. Cell 66 , 789–800. 10.1016/j.molcel.2017.05.032.28622524 Ge SX , Jung D , and Yao R (2020). ShinyGO: a graphical gene-set enrichment tool for animals and plants. Bioinformatics 36 , 2628–2629. 10.1093/bioinformatics/btz931.31882993 Ghayee HK , Bhagwandin VJ , Stastny V , Click A , Ding LH , Mizrachi D , Zou YS , Chari R , Lam WL , Bachoo RM , (2013). Progenitor cell line (hPheo1) derived from a human pheochromocytoma tumor. PLoS One 8 , e65624. 10.1371/journal.pone.0065624.23785438 Gilissen J , Jouret F , Pirotte B , and Hanson J (2016). Insight into SUCNR1 (GPR91) structure and function. Pharmacol. Ther. 159 , 56–65. 10.1016/j.pharmthera.2016.01.008.26808164 Gimenez-Roqueplo AP , Favier J , Rustin P , Rieubland C , Crespin M , Nau V , Khau Van Kien P , Corvol P , Plouin PF , and Jeunemaitre X ; COMETE Network (2003). Mutations in the SDHB gene are associated with extra-adrenal and/or malignant phaeochromocytomas. Cancer Res. 63 , 5615–5621.14500403 Gimenez-Roqueplo AP , Favier J , Rustin P , Rieubland C , Kerlan V , Plouin PF , Rötig A , and Jeunemaitre X (2002). Functional consequences of a SDHB gene mutation in an apparently sporadic pheochromocytoma. J. Clin. Endocrinol. Metab. 87 , 4771–4774. 10.1210/jc.2002-020525.12364472 Goldstein M , Fuxe K , and Hökfelt T (1972). Characterization and tissue localization of catecholamine synthesizing enzymes. Pharmacol. Rev. 24 , 293–309.4564603 Gottlieb E , and Tomlinson IPM (2005). Mitochondrial tumour suppressors: a genetic and biochemical update. Nat. Rev. Cancer 5 , 857–866. 10.1038/nrc1737.16327764 Gowans GJ , Hawley SA , Ross FA , and Hardie DG (2013). AMP is a true physiological regulator of AMP-activated protein kinase by both allosteric activation and enhancing net phosphorylation. Cell Metab. 18 , 556–566. 10.1016/j.cmet.2013.08.019.24093679 Hachet-Haas M , Converset N , Marchal O , Matthes H , Gioria S , Galzi JL , and Lecat S (2006). FRET and colocalization analyzer–a method to validate measurements of sensitized emission FRET acquired by confocal microscopy and available as an ImageJ Plug-in. Microsc. Res. Tech. 69 , 941–956. 10.1002/jemt.20376.17080432 Hadrava Vanova K , Kraus M , Neuzil J , and Rohlena J (2020). Mitochondrial complex II and reactive oxygen species in disease and therapy. Redox Rep. 25 , 26–32. 10.1080/13510002.2020.1752002.32290794 Hanahan D , and Weinberg RA (2000). The hallmarks of cancer. Cell 100 , 57–70. 10.1016/s0092-8674(00)81683-9.10647931 Hardie DG (2014). AMPK–sensing energy while talking to other signaling pathways. Cell Metab. 20 , 939–952. 10.1016/j.cmet.2014.09.013.25448702 Hawley SA , Pan DA , Mustard KJ , Ross L , Bain J , Edelman AM , Frenguelli BG , and Hardie DG (2005). Calmodulin-dependent protein kinase kinase-beta is an alternative upstream kinase for AMP-activated protein kinase. Cell Metab. 2 , 9–19. 10.1016/j.cmet.2005.05.009.16054095 Hemmings HC (1997). Regulatory Protein Modification: Techniques and Protocols (Springer). Hsu PP , and Sabatini DM (2008). Cancer cell metabolism: Warburg and beyond. Cell 134 , 703–707. 10.1016/j.cell.2008.08.021.18775299 Huang Y , Wang LA , Xie Q , Pang J , Wang L , Yi Y , Zhang J , Zhang Y , Chen R , Lan W , (2018). Germline SDHB and SDHD mutations in pheochromocytoma and paraganglioma patients. Endocr. Connect. 7 , 1217–1225. 10.1530/EC-18-0325. Jana F , Bustos G , Rivas J , Cruz P , Urra F , Basualto-Alarcón C , Sagredo E , Ríos M , Lovy A , Dong Z , (2019). Complex I and II are required for normal mitochondrial Ca(2+) homeostasis. Mitochondrion 49 , 73–82. 10.1016/j.mito.2019.07.004.31310854 Jeanmonod R , Asuka E , and Jeanmonod D (2020). Inborn Errors of Metabolism (StatPearls). Jochmanova I , and Pacak K (2016). Pheochromocytoma: the first metabolic endocrine cancer. Clin. Cancer Res. 22 , 5001–5011. 10.1158/1078-0432.CCR-16-0606.27742786 Jones RG , Plas DR , Kubek S , Buzzai M , Mu J , Xu Y , Birnbaum MJ , and Thompson CB (2005). AMP-activated protein kinase induces a p53-dependent metabolic checkpoint. Mol. Cell 18 , 283–293. 10.1016/j.molcel.2005.03.027.15866171 Jung SH , Hwang HJ , Kang D , Park HA , Lee HC , Jeong D , Lee K , Park HJ , Ko YG , and Lee JS (2019). mTOR kinase leads to PTEN-loss-induced cellular senescence by phosphorylating p53. Oncogene 38 , 1639–1650. 10.1038/s41388-018-0521-8.30337688 Kim JW , and Dang CV (2006). Cancer’s molecular sweet tooth and the Warburg effect. Cancer Res. 66 , 8927–8930. 10.1158/0008-5472.CAN-06-1501.16982728 King KS , Prodanov T , Kantorovich V , Fojo T , Hewitt JK , Zacharin M , Wesley R , Lodish M , Raygada M , Gimenez-Roqueplo AP , (2011). Metastatic pheochromocytoma/paraganglioma related to primary tumor development in childhood or adolescence: significant link to SDHB mutations. J. Clin. Oncol. 29 , 4137–4142. 10.1200/JCO.2011.34.6353.21969497 Kluckova K , Thakker A , Vettore L , Escribano-Gonzalez C , Hindshaw RL , Tearle JLE , Goncalves J , Kaul B , Lavery GG , Favier J , and Tennant DA (2020). Succinate dehydrogenase deficiency in a chromaffin cell model retains metabolic fitness through the maintenance of mitochondrial NADH oxidoreductase function. FASEB J 34 , 303–315. 10.1096/fj.201901456R.31914648 Koh J , Hogue JA , Wang Y , DiSalvo M , Allbritton NL , Shi Y , Olson JA Jr. , and Sosa JA (2016). Single-cell functional analysis of parathyroid adenomas reveals distinct classes of calcium sensing behaviour in primary hyperparathyroidism. J. Cell Mol. Med. 20 , 351–359. 10.1111/jcmm.12732.26638194 Korpershoek E , Favier J , Gaal J , Burnichon N , van Gessel B , Oudijk L , Badoual C , Gadessaud N , Venisse A , Bayley JP , (2011). SDHA immunohistochemistry detects germline SDHA gene mutations in apparently sporadic paragangliomas and pheochromocytomas. J. Clin. Endocrinol. Metab. 96 , E1472–E1476. 10.1210/jc.2011-1043.21752896 Korpershoek E , Pacak K , and Martiniova L (2012). Murine models and cell lines for the investigation of pheochromocytoma: applications for future therapies? Endocr. Pathol. 23 , 43–54. 10.1007/s12022-012-9194-y.22323007 Krebs HA , and Johnson WA (1980). The role of citric acid in intermediate metabolism in animal tissues. FEBS Lett. 117 , K1–K10. Lee JH , Jang H , Lee SM , Lee JE , Choi J , Kim TW , Cho EJ , and Youn HD (2015). ATP-citrate lyase regulates cellular senescence via an AMPK- and p53-dependent pathway. FEBS J. 282 , 361–371. 10.1111/febs.13139.25367309 Li H , Huang K , Liu X , Liu J , Lu X , Tao K , Wang G , and Wang J (2014). Lithium chloride suppresses colorectal cancer cell survival and proliferation through ROS/GSK-3beta/NF-kappaB signaling pathway. Oxid. Med. Cell. Longev. 2014 , 241864. 10.1155/2014/241864.25002914 Maignan A , Guerin C , Julliard V , Paladino NC , Kim E , Roche P , Castinetti F , Essamet W , Mancini J , Imperiale A , (2017). Implications of SDHB genetic testing in patients with sporadic pheochromocytoma. Langenbeck’s Arch. Surg. 402 , 787–798. 10.1007/s00423-017-1564-y.28229225 Mao D , and Hinds PW (2010). p35 is required for CDK5 activation in cellular senescence. J. Biol. Chem. 285 , 14671–14680. 10.1074/jbc.M109.066118.20181942 Martiniova L , Lai EW , Elkahloun AG , Abu-Asab M , Wickremasinghe A , Solis DC , Perera SM , Huynh TT , Lubensky IA , Tischler AS , (2009). Characterization of an animal model of aggressive metastatic pheochromocytoma linked to a specific gene signature. Clin. Exp. Metastasis 26 , 239–250. 10.1007/s10585-009-9236-0.19169894 Matlac DM , Hadrava Vanova K , Bechmann N , Richter S , Folberth J , Ghayee HK , Ge GB , Abunimer L , Wesley R , Aherrahrou R , (2021). Succinate mediates tumorigenic effects via succinate receptor 1: potential for new targeted treatment strategies in succinate dehydrogenase deficient paragangliomas. Front. Endocrinol. 12 , 589451. 10.3389/fendo.2021.589451. McClue SJ , and Stuart I (2008). Metabolism of the trisubstituted purine cyclin-dependent kinase inhibitor seliciclib (R-roscovitine) in vitro and in vivo. Drug Metab. Dispos. 36 , 561–570. 10.1124/dmd.107.019232.18048486 McDermott U , Sharma SV , Dowell L , Greninger P , Montagut C , Lamb J , Archibald H , Raudales R , Tam A , Lee D , (2007). Identification of genotype-correlated sensitivity to selective kinase inhibitors by using high-throughput tumor cell line profiling. Proc. Natl. Acad. Sci. USA 104 , 19936–19941. 10.1073/pnas.0707498104.18077425 Meyer DA , Richer E , Benkovic SA , Hayashi K , Kansy JW , Hale CF , Moy LY , Kim Y , O’Callaghan JP , Tsai LH , (2008). Striatal dysregulation of Cdk5 alters locomotor responses to cocaine, motor learning, and dendritic morphology. Proc. Natl. Acad. Sci. USA 105 , 18561–18566. 10.1073/pnas.0806078105.19017804 Mijit M , Caracciolo V , Melillo A , Amicarelli F , and Giordano A (2020). Role of p53 in the regulation of cellular senescence. Biomolecules 10 , E420. 10.3390/biom10030420. Morfini G , Szebenyi G , Brown H , Pant HC , Pigino G , DeBoer S , Beffert U , and Brady ST (2004). A novel CDK5-dependent pathway for regulating GSK3 activity and kinesin-driven motility in neurons. EMBO J. 23 , 2235–2245. 10.1038/sj.emboj.7600237.15152189 Nasr P , Gursahani HI , Pang Z , Bondada V , Lee J , Hadley RW , and Geddes JW (2003). Influence of cytosolic and mitochondrial Ca2+, ATP, mitochondrial membrane potential, and calpain activity on the mechanism of neuron death induced by 3-nitropropionic acid. Neurochem. Int. 43 , 89–99. 10.1016/s0197-0186(02)00229-2.12620277 Neumann HP , Pawlu C , Peczkowska M , Bausch B , McWhinney SR , Muresan M , Buchta M , Franke G , Klisch J , Bley TA , (2004). Distinct clinical features of paraganglioma syndromes associated with SDHB and SDHD gene mutations. JAMA 292 , 943–951. 10.1001/jama.292.8.943.15328326 Oakhill JS , Steel R , Chen ZP , Scott JW , Ling N , Tam S , and Kemp BE (2011). AMPK is a direct adenylate charge-regulated protein kinase. Science 332 , 1433–1435. 10.1126/science.1200094.21680840 Pang Z , Bondada V , Sengoku T , Siman R , and Geddes JW (2003). Calpain facilitates the neuron death induced by 3-nitropropionic acid and contributes to the necrotic morphology. J. Neuropathol. Exp. Neurol. 62 , 633–643. 10.1093/jnen/62.6.633.12834108 Pap M , and Cooper GM (1998). Role of glycogen synthase kinase-3 in the phosphatidylinositol 3-Kinase/Akt cell survival pathway. J. Biol. Chem. 273 , 19929–19932. 10.1074/jbc.273.32.19929.9685326 Patrick GN , Zukerberg L , Nikolic M , de la Monte S , Dikkes P , and Tsai LH (1999). Conversion of p35 to p25 deregulates Cdk5 activity and promotes neurodegeneration. Nature 402 , 615–622. 10.1038/45159.10604467 Pinter K , Grignani RT , Watkins H , and Redwood C (2013). Localisation of AMPK gamma subunits in cardiac and skeletal muscles. J. Muscle Res. Cell Motil. 34 , 369–378. 10.1007/s10974-013-9359-4.24037260 Plattner F , Angelo M , and Giese KP (2006). The roles of cyclin-dependent kinase 5 and glycogen synthase kinase 3 in tau hyperphosphorylation. J. Biol. Chem. 281 , 25457–25465. 10.1074/jbc.M603469200.16803897 Pollard PJ , El-Bahrawy M , Poulsom R , Elia G , Killick P , Kelly G , Hunt T , Jeffery R , Seedhar P , Barwell J , (2006). Expression of HIF-1alpha, HIF-2alpha (EPAS1), and their target genes in paraganglioma and pheochromocytoma with VHL and SDH mutations. J. Clin. Endocrinol. Metab. 91 , 4593–4598. 10.1210/jc.2006-0920.16954163 Pozo K , and Bibb JA (2016). The emerging role of Cdk5 in cancer. Trends Cancer 2 , 606–618. 10.1016/j.trecan.2016.09.001.27917404 Pozo K , Castro-Rivera E , Tan C , Plattner F , Schwach G , Siegl V , Meyer D , Guo A , Gundara J , Mettlach G , (2013). The role of Cdk5 in neuroendocrine thyroid cancer. Cancer Cell 24 , 499–511. 10.1016/j.ccr.2013.08.027.24135281 Rai SK , Bril F , Hatch HM , Xu Y , Shelton L , Kalavalapalli S , Click A , Lee D , Beecher C , Kirby A , (2020). Targeting pheochromocytoma/paraganglioma with polyamine inhibitors. Metabolism 110 , 154297.32562798 Rajgopal Y , and Vemuri MC (2002). Calpain activation and alpha-spectrin cleavage in rat brain by ethanol. Neurosci. Lett. 321 , 187–191. 10.1016/s0304-3940(02)00063-0.11880203 Ranganayaki S , Jamshidi N , Aiyaz M , Rashmi SK , Gayathri N , Harsha PK , Padmanabhan B , and Srinivas Bharath MM (2021). Inhibition of mitochondrial complex II in neuronal cells triggers unique pathways culminating in autophagy with implications for neurodegeneration. Sci. Rep. 11 , 1483. 10.1038/s41598-020-79339-2.33452321 Ross ME , Evinger MJ , Hyman SE , Carroll JM , Mucke L , Comb M , Reis DJ , Joh TH , and Goodman HM (1990). Identification of a functional glucocorticoid response element in the phenylethanolamine N-methyltransferase promoter using fusion genes introduced into chromaffin cells in primary culture. J. Neurosci. 10 , 520–530.2303857 Rossitti HM , Dutta RK , Larsson C , Ghayee HK , Söderkvist P , and Gimm O . (2020). Activation of RAS signalling is associated with altered cell adhesion in phaeochromocytoma. Int. J. Mol. Sci. 21 , E8072. 10.3390/ijms21218072. Schindelin J , Arganda-Carreras I , Frise E , Kaynig V , Longair M , Pietzsch T , Preibisch S , Rueden C , Saalfeld S , Schmid B , (2012). Fiji: an open-source platform for biological-image analysis. Nat. Methods 9 , 676–682. 10.1038/nmeth.2019.22743772 Shaw RJ , Kosmatka M , Bardeesy N , Hurley RL , Witters LA , DePinho RA , and Cantley LC (2004). The tumor suppressor LKB1 kinase directly activates AMP-activated kinase and regulates apoptosis in response to energy stress. Proc. Natl. Acad. Sci. USA 101 , 3329–3335. 10.1073/pnas.0308061100.14985505 Solis DC , Burnichon N , Timmers HJLM , Raygada MJ , Kozupa A , Merino MJ , Makey D , Adams KT , Venisse A , Gimenez-Roqueplo AP , and Pacak K (2009). Penetrance and clinical consequences of a gross SDHB deletion in a large family. Clin. Genet. 75 , 354–363. 10.1111/j.1399-0004.2009.01157.x.19389109 Stockholm D , Bartoli M , Sillon G , Bourg N , Davoust J , and Richard I (2005). Imaging calpain protease activity by multiphoton FRET in living mice. J. Mol. Biol. 346 , 215–222. 10.1016/j.jmb.2004.11.039.15663939 Sun KH , de Pablo Y , Vincent F , and Shah K (2008). Deregulated Cdk5 promotes oxidative stress and mitochondrial dysfunction. J. Neurochem. 107 , 265–278. 10.1111/j.1471-4159.2008.05616.x.18691386 Sundstrom L , Greasley PJ , Engberg S , Wallander M , and Ryberg E (2013). Succinate receptor GPR91, a Galpha(i) coupled receptor that increases intracellular calcium concentrations through PLCbeta. FEBS Lett. 587 , 2399–2404. 10.1016/j.febslet.2013.05.067.23770096 Suzuki T , Bridges D , Nakada D , Skiniotis G , Morrison SJ , Lin JD , Saltiel AR , and Inoki K (2013). Inhibition of AMPK catabolic action by GSK3. Mol. Cell 50 , 407–419. 10.1016/j.molcel.2013.03.022.23623684 Tang QL , Xie XB , Wang J , Chen Q , Han AJ , Zou CY , Yin JQ , Liu DW , Liang Y , Zhao ZQ , (2012). Glycogen synthase kinase-3beta, NF-kappaB signaling, and tumorigenesis of human osteosarcoma. J. Natl. Cancer Inst. 104 , 749–763. 10.1093/jnci/djs210.22534782 Tang Z , Li C , Kang B , Gao G , Li C , and Zhang Z (2017). GEPIA: a web server for cancer and normal gene expression profiling and interactive analyses. Nucleic Acids Res. 45 , W98–W102. 10.1093/nar/gkx247.28407145 Thoreen CC , and Sabatini DM (2005). AMPK and p53 help cells through lean times. Cell Metab. 1 , 287–288. 10.1016/j.cmet.2005.04.009.16054073 Vander Heiden MG , Cantley LC , and Thompson CB (2009). Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324 , 1029–1033. 10.1126/science.1160809.19460998 Varghese F , Bukhari AB , Malhotra R , and De A (2014). IHC Profiler: an open source plugin for the quantitative evaluation and automated scoring of immunohistochemistry images of human tissue samples. PLoS One 9 , e96801. 10.1371/journal.pone.0096801.24802416 Vial G , Detaille D , and Guigas B (2019). Role of mitochondria in the mechanism(s) of action of metformin. Front. Endocrinol. 10 , 294. 10.3389/fendo.2019.00294. Wang Q , Liang B , Shirwany NA , and Zou MH (2011). 2-Deoxy-D-glucose treatment of endothelial cells induces autophagy by reactive oxygen species-mediated activation of the AMP-activated protein kinase. PLoS One 6 , e17234. 10.1371/journal.pone.0017234.21386904 Wang Y , Thatcher SE , and Cassis LA (2017). Measuring blood pressure using a noninvasive tail cuff method in mice. Methods Mol. Biol. 1614 , 69–73. 10.1007/978-1-4939-7030-8_6. Wen Y , Planel E , Herman M , Figueroa HY , Wang L , Liu L , Lau LF , Yu WH , and Duff KE (2008). Interplay between cyclin-dependent kinase 5 and glycogen synthase kinase 3 beta mediated by neuregulin signaling leads to differential effects on tau phosphorylation and amyloid precursor protein processing. J. Neurosci. 28 , 2624–2632. 10.1523/JNEUROSCI.5245-07.2008.18322105 Wu JY , Huang TW , Hsieh YT , Wang YF , Yen CC , Lee GL , Yeh CC , Peng YJ , Kuo YY , Wen HT , (2020). Cancer-derived succinate promotes macrophage polarization and cancer metastasis via succinate receptor. Mol. Cell 77 , 213–227.e5. 10.1016/j.molcel.2019.10.023.31735641 Xie Z , Dong Y , Scholz R , Neumann D , and Zou MH (2008). Phosphorylation of LKB1 at serine 428 by protein kinase C-zeta is required for metforminenhanced activation of the AMP-activated protein kinase in endothelial cells. Circulation 117 , 952–962. 10.1161/CIRCULATIONAHA.107.744490.18250273 Xu Y , Gray A , Hardie DG , Uzun A , Shaw S , Padbury J , Phornphutkul C , and Tseng YT (2017). A novel, de novo mutation in the PRKAG2 gene: infantile-onset phenotype and the signaling pathway involved. Am. J. Physiol. Heart Circ. Physiol. 313 , H283–H292. 10.1152/ajpheart.00813.2016.28550180 Xue W , Zender L , Miething C , Dickins RA , Hernando E , Krizhanovsky V , Cordon-Cardo C , and Lowe SW (2007). Senescence and tumour clearance is triggered by p53 restoration in murine liver carcinomas. Nature 445 , 656–660. 10.1038/nature05529.17251933 Yang C , Matro JC , Huntoon KM , Ye DY , Huynh TT , Fliedner SMJ , Breza J , Zhuang Z , and Pacak K (2012). Missense mutations in the human SDHB gene increase protein degradation without altering intrinsic enzymatic function. FASEB J 26 , 4506–4516. 10.1096/fj.12-210146.22835832 Zadra G , Batista JL , and Loda M (2015). Dissecting the dual role of AMPK in cancer: from experimental to human studies. Mol. Cancer Res. 13 , 1059–1072. 10.1158/1541-7786.MCR-15-0068.25956158 Zhan Y , Sun X , Li B , Cai H , Xu C , Liang Q , Lu C , Qian R , Chen S , Yin L , (2018). Establishment of a PRKAG2 cardiac syndrome disease model and mechanism study using human induced pluripotent stem cells. J. Mol. Cell. Cardiol. 117 , 49–61. 10.1016/j.yjmcc.2018.02.007.29452156 Zhang S , Lu Z , Mao W , Ahmed AA , Yang H , Zhou J , Jennings N , Rodriguez-Aguayo C , Lopez-Berestein G , Miranda R , (2015). CDK5 regulates paclitaxel sensitivity in ovarian cancer cells by modulating AKT activation, p21Cip1- and p27Kip1-mediated G1 cell cycle arrest and apoptosis. PLoS One 10 , e0131833. 10.1371/journal.pone.0131833.26146988 Zhou G , Myers R , Li Y , Chen Y , Shen X , Fenyk-Melody J , Wu M , Ventre J , Doebber T , Fujii N , (2001). Role of AMP-activated protein kinase in mechanism of metformin action. J. Clin. Invest. 108 , 1167–1174. 10.1172/JCI13505.11602624

PMC009xxxxxx/PMC9825050.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0372354 646 Ann Surg Ann Surg Annals of surgery 0003-4932 1528-1140 32941272 9825050 10.1097/SLA.0000000000004493 NIHMS1858739 Article Disparity Between United States Adolescent Class II and III Obesity Trends and Bariatric Surgery Utilization, 2015–2018 Messiah Sarah E. PhD *†✉ Xie Luyu PharmD *† Atem Folefac PhD *† Mathew M. Sunil MS *† Qureshi Faisal G. MD ‡ Schneider Benjamin E. MD ‡ de la Cruz-Muñoz Nestor MD § * University of Texas Health Science Center, School of Public Health, Dallas, TX † Center for Pediatric Population Health, UTHealth School of Public Health and Children’s Health System of Texas ‡ Department of Surgery, University of Texas Southwestern Medical Center, Dallas, TX § Department of Surgery, University of Miami Miller School of Medicine, Miami, FL ✉ Sarah.E.Messiah@uth.tmc.edu, Sarah.Messiah@utsouthwestern.edu. 29 12 2022 01 8 2022 15 9 2020 07 1 2023 276 2 324333 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Objectives: Class II (120% ≥ body mass index [BMI] < 140% of the 95th percentile for age and sex) and Class III (BMI ≥140% of the 95th percentile for age and sex) obesity are the fastest growing subcategories of obesity in the United States pediatric population. Metabolic and bariatric surgery (MBS) is a safe and effective treatment option for class II/III obesity. The primary objectives of this analysis were to determine the (1) current US MBS utilization rates in those with class II/III obesity and (2) utilization rates and 30-day postoperative outcomes. Background: The 2015 to 2018 National Health and Nutrition Examination Survey cross-sectional data (N = 19,225) generated US class II/III obesity prevalence estimates. The 2015 to 2018 Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP) longitudinal (30 days) cohort data were used to compare adolescent and adult (N = 748,622) postoperative outcomes and to calculate utilization rates. Methods: The 2015 to 2018 youth and adult MBS utilization rates were calculated using MBSAQIP data (numerator) and National Health and Nutrition Examination Survey data (denominator). Two-sample tests of proportions were performed to compare the MBS utilization rates by age, ethnicity, and sex and expressed per 1000. Results: Mean age of the analytical MBSAQIP sample was 17.9 (1.15) years in youth (n = 3846) and 45.1 (11.5) in adults (N = 744,776), majority female (77.4%, 80.7%, respectively) and non-Hispanic White (68.5%, 59.4%, respectively). The overall 2015 to 2018 MBS utilization rate for youth was 1.81 per 1000 and 5.56 per 1000 for adults (P < 0.001). Adult patients had slightly higher percentage (4.2%) of hospital readmissions compared to youth (3.4%, P = 0.01) but there were no other post-MBS complication differences. From 2015 to 2018 the US prevalence of youth with class II/III obesity increased in Hispanics and non-Hispanic Blacks (P trend <0001), but among youth who did complete MBS non-Hispanic Whites had higher rates of utilization (45.8%) compared to Hispanics (22.7%) and non-Hispanic blacks 14.2% (P = 0.006). Conclusions: MBS is an underutilized obesity treatment tool for both youth and adults, and among ethnic minority groups in particular. adolescents adults metabolic and bariatric surgery population prevalence readmission class II obesity, class III obesity pmcClass II obesity (defined as 120% ≥ body mass index or BMI < 140% of the 95th percentile for age and sex) and class III obesity (defined as BMI ≥ 140% of the 95th percentile for age and sex) are the fastest growing subcategories of obesity in the US pediatric population.1 As of 2016, approximately 9% of 12- to 19-year-olds had class II/III obesity, triple the prevalence from 1988 to 1994.1 Even more concerning, almost 12% of non-Hispanic Black and 9% of Hispanic adolescents ages 12 to 19 have class II/III obesity compared to 7% of their non-Hispanic White counterparts.2 Class II/III obesity during the pediatric years has become such a prevalent issue that it has been labeled “an epidemic within an epidemic.”3 Obesity in youth is associated with many cardiometabolic comorbidities, liver and kidney disease, lower sleep quality, and mental health comorbidities resulting in lower quality of life scores.2–5 Moreover, class II/III obesity during adolescence tracks strongly into adulthood and is associated with adult asthma, arthritis, and poorer cardiometabolic and psychological risk profiles.6–10 Metabolic and bariatric surgery (MBS) is safe and efficacious in treating adolescents with class II/III obesity.11–13 As such, in late 2019, the American Academy of Pediatrics called for better access to MBS for adolescents and teenagers with class II/III obesity when medically indicated.14 In addition, weight loss behavioral, lifestyle, and pharmacotherapy treatment programs in inpatient and ambulatory settings have reported mixed findings, especially in terms of similar, and sustained weight loss trajectories.15–21 MBS continues to remain a popular adult elective procedure in the United States with about a quarter of a million patients undergoing the procedure annually.22 Over the past several years the overall number of surgeries has increased from 158,000 in 2011 to 228,000 in 2017.22 We report here the latest available (through 2018) national prevalence estimates for class II/III obesity in US youth and calculate MBS utilization rates during the same time period. We also compared youth and adult MBS utilization rates and 30-day complication rates. It was hypothesized that MBS utilization would be lower among US youth compared to adults, despite similar post-MBS complication rates. METHODS Study Design A cross-sectional design was used for the national obesity prevalence estimates. A prospective (30-day) cohort design was used for the Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP) data collection. A retrospective analysis of both datasets was conducted. Data Sources National Obesity Prevalence Estimates The National Health and Nutrition Examination Survey (NHANES) 2015 to 2016 and 2017 to 2018 cycles were used for this analysis. NHANES is a cross-sectional survey conducted in a 2-year cycle by the National Center for Health Statistics to monitor health and nutrition status of the US population.23 For each survey cycle, a complex 3-step weighting method is created to represent the US noninstitutional, civilian population. Participants A total of 19,225 participants were included in the 2015 to 2018 data cycles (15–19 years old = 2417). Metabolic and Bariatric Surgery Prevalence Estimates In 2012, the American College of Surgeons and the American Society for Metabolic and Bariatric Surgery merged their MBS accredited programs into the MBSAQIP.24 The merged 2015–2018 MBSAQIP participant use files (PUF) was used for this analysis (N = 760,076). The MBSAQIP PUF is a clinical data set of MBS patients who received their clinical care at an accredited center. The database includes Health Insurance Portability and Accountability Act–compliant data such as patient demographic characteristics, laboratory data, comorbidities, and complications within 30 days post-MBS. Data are collected at each accredited center by certified MBS clinical reviewers who undergo audits for reliability and consistency.25 The PUF does not identify healthcare providers or hospitals, region or area of surgery, and no personal health information is reported. Patients who were greater than 70 years old (n = 11,221) or with age data missing (n = 233) were excluded from our analysis. Participants The final analytical sample of the 2018 PUF file contains 173 Health Insurance Portability and Accountability Act compliant variables on 201,180 cases submitted from 854 centers, whereas the 2017 file contains 197,175 cases submitted by 832 centers, the 2016 file contains 184,004 cases submitted by 791 centers, and the 2015 file contained data on 166,216 patients from 742 centers (N=47 missing year information). The UT Health institutional review board determined that as a retrospective analyses of public, anonymized data sets the MBSAQIP Data Registry is exempt from review. Statistical Analysis National Health and Nutrition Examination Survey Analysis 2015 to 2018 NHANES data were used to calculate weighted prevalence estimates with 95% confidence intervals (CIs) of class II/III obesity in the US pediatric and adult population. Time trend analyses for overall and different race/ethnicities were performed by using adjusted linear regression models [including survey year, sex, age, race/ethnicities (non-Hispanic White, non-Hispanic Black, Hispanic, or other)]. Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program Analysis Descriptive analysis was performed for baseline characteristics and youth and adult samples were compared. Chi-square tests were used for comparison of categorical predictors including sex, race/ethnicity, categorical pre-op BMI, categorical highest pre-op BMI, procedure type, hypertension, hyperlipidemia, type 2 diabetes, gastroesophageal reflux disease (GERD), sleep apnea, and chronic steroid use. A 2 sample (independent groups) t test was performed to determine any statistical significance for all continuous variables such as age. A further chi-square test was performed to compare the following outcomes between youth and adults; mortality, reoperation, acute renal failure, bleeding event, at least 1 intervention within 30 days, unplanned intubation, sepsis ICU admission with 30 days, ventilator 48 hours, approach conversion, pulmonary embolism, Clostridium difficile colitis, ED visit with 30 days, urinary tract infection, pneumonia, and incisional surgical site infection. Youth and adult differences in continuous outcomes including hospital length of stay and operation length were examined with a 2-sample t test. Linear regression analysis was performed to compare the percentage weight loss between youth and adults. Total body (percentage) weight loss (TBWL) was computed by subtracting weight at 30 days from the presurgery weight then multiplying this by 100 and divide by the pre-surgery weight. Logistic regression analysis was performed to compare hospital readmission rates between youth and adults. MBS was the exposure variable of interest and all available postoperative complications were considered outcomes. In these models, age was analyzed as a categorical predictor with 2 levels, adult and youth with adult as the reference group. The following were also analyzed as binary (Y/N) predictors: hypertension, hyperlipidemia, type 2 diabetes, GERD, sleep apnea, and chronic steroid use with the presence of these conditions as the reference. The variable sex was also entered as a binary predictor with female as the reference group. Race/ethnicity was analyzed as a 4-level categorical predictor that included non-Hispanic White, Hispanic, non-Hispanic Black, and other (Asian, American Indian, and others) with non-Hispanic Whites as the reference category. A multivariable logistic regression model calculated the odds of hospital readmission by demographics and comorbidities. US Metabolic and Bariatric Surgery Utilization Rates The 2015 to 2018 youth and adult MBS utilization rates were calculated by the following formula using MBSAQIP data only for the numerator and NHANES data only for denominator: MBS utilization rate= Number of MBS completedNumber of NHANES individuals eligible for MBS×1000 The number of NHANES eligible individuals for MBS included patients with (1) class II obesity (defined as BMI ≥ 120% to < 140% of 95th percentile for age and sex in youths; and 35kg/m2≤BMI < 40kg/m2 in adults) with at least 1 comorbidity (elevated blood pressure, cholesterol, or type 2 diabetes or (2) class III obesity, defined as BMI ≥140% of the 95th percentile for age and sex in youths or BMI ≥ 40kg/m2 in adults.26 Comorbidity status was determined by objective measurements from NHANES laboratory and examination data. Specifically, following American Society for Metabolic and Bariatric Surgery guidelines, blood pressure 130/80 or higher,27 total cholesterol of 200 mg/dL or more (5.2 mmol/L),28 and hemoglobin A1c level of 6.5% or more29 were the cut-off values used to determine if a participant had a comorbidity in addition to class II obesity for this analysis. The MBS utilization rate was also calculated by age, sex, and ethnic group and expressed per 1000. Overall (aggregated) utilization rates for youth and adults were computed as a weighted average across all years and all groups. Two-sample test of proportions were performed to compare the MBS utilization rates within each subgroup. All statistical analyses were performed by a statistician using SAS v9.4 (SAS Institute, Cary, NC) and STATA v15.1 (College Station, TX). The type 1 error was maintained at 5%. RESULTS National Health and Nutrition Examination Survey The overall class II/III obesity rates increased from 2015 (5.6%, 95% CI 3.9%–7.4%) to 2018 (6.5%, 95% CI 4.6%–8.4%; P trend <0.001) (Fig. 1 A). NHANES 2017 to 2018 data show that overall 6.5%, or about 4.8 million US youth ages 2 to 19 years have class II/III obesity, whereas the MBSAQIP data during the same time period included a total of 1862 surgeries among those are 19 years or younger. Despite Hispanic and non-Hispanic Black youth having higher class II/III obesity rates compared to non-Hispanic White youth, rates of MBS utilization are lower among these 2 groups (Fig. 1B). In 2018, nearly half (45.8%) of youth who completed MBS were non-Hispanic White versus 22.7% Hispanic and 14.2% non-Hispanic Black (P = 0.006). Likewise, in adults compared to non-Hispanic Whites, significantly less Hispanic and non-Hispanic Black adults completed MBS (57.8% vs 9.4% vs 17.3%, P < 0.001) (data not shown in tables). Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program Table 1 shows no significant differences between youth and adult 2015 to 2018 samples on all baseline variables. There were slightly more women (80.7%) in the adult versus the youth (77.4%) sample (P < 0.001). The proportion of non-Hispanic Whites was higher in the youth sample (68.5%) compared to the adult sample (59.4%), but there were more Hispanics in the youth sample (18.4% vs 9.1%, respectively). There were significant differences in the highest pre-MBS BMI (48.4 in youth vs 46.5 in adults, P < 0.001). Almost a third (32.97%) of youth had a pre-op BMI of 50 kg/m2 or more compared to adults (26.0%). The majority of both adults and youth underwent laparoscopic sleeve gastrectomy (73.3% for youth, 61.5% for adults). There were no significant differences between youth and adult post-MBS outcomes with the exception of hospital readmission rate, which was slightly higher for the adult (4.2%) compared to youth sample (3.4%, P = 0.01) and the mean length of the operation time which was longer in adult (88.4 min) compared to youth (80.4 min, P < 0.001) (Table 2). TBWL in youth was 10.1 (SE = 0.7) and 15.0 (SE = 0.1) for adults at 30 days post-MBS (P < 0.001). The mean weight loss among youth was 6.5 kg (SE = 1.0) and 12.2 kg in adults (SE = 0.1) at 30 days post-MBS (P < 0.001) (data not shown in tables). Logistic regression results showed that youth and adult men were less likely than women to be re-admitted post-MBS [odds ratio (OR) 0.7, 95% CI 0.4–1.1 for youth; OR 0.9, 95% CI 0.8–0.9 for adults] (Table 3). Among adults only, Hispanic (OR 1.1, 95% CI 1.0–1.1) and non-Hispanic Black (OR 1.4, 95% CI 1.4–1.5) were more likely than non-Hispanic White patients to be readmitted post-MBS. Both youth and adults who completed the laparoscopic sleeve gastrectomy procedure were less likely than patients who completed the Roux-en-Y gastric bypass procedure to be readmitted post-MBS (OR 0.4, 95% CI 0.3–0.7 for youth; OR 0.5, 95% CI 0.4–0.5 for adults). Adults with hypertension, hyperlipidemia, type 2 diabetes, GERD, and sleep apnea were more likely to be readmitted versus those without comorbidities. For youth, those with hypertension and sleep apnea were more likely to be readmitted. Both youth and adults with chronic steroid use were more likely to be readmitted compared with those who did not use this medication (OR 3.5, 95% CI, 1.2–10.4 for youth; OR 1.5, 95% CI 1.4–1.6 for adults). The overall 2015 to 2018 MBS utilization rate was 1.81 surgeries per 1000 for youth and 5.56 per 1000 for adults (P < 0.001). The rate was 0.09 surgeries per 1000 among 13 to 16 year olds. Among youth, the rate was highest in non-Hispanic Whites (1.75 and 1.33 in 2015–2016 and 2017–2018, respectively) compared to non-Hispanic Blacks (0.10 and 0.06 per 1000 in 2015–2016 and 2017–2018, respectively) and Hispanics (0.12 and 0.23 per 1000 in 2015–2016 and 2017–2018, respectively (P < 0.001 in all groups except other race/ethnicities in 2015–2016) (Table 4). Among adults, the rate was highest in non-Hispanic Whites (6.29 and 5.31 in 2015–2016 and 2017–2018, respectively) compared to non-Hispanic Blacks (1.14 and 1.13 per 1000 in 2015–2016 and 2017–2018, respectively) and Hispanics (0.74 and 0.87 per 1000 in 2015–2016 and 2017–2018, respectively (P < 0.001 in all groups) (Table 5). DISCUSSION Analysis here took advantage of 2 of the most current population-level datasets to calculate national MBS completion rates (1.81 in youth and 5.56 in adults per 1000) while also comparing youth and adult MBS 30-day complication and TBWL outcomes. Results showed that while youth class II/III obesity rates continue to rise, the overall MBS utilization rate decreased in MBSAQIP centers during the same time period. Moreover, a higher proportion of youth enter MBS with a BMI of 50 kg/m2 compared with adults. During the 30-day postoperative period, youth and adult complication(s) and readmission rates (3.4%, 4.2%, respectively) were similar. Ethnic group disparities among youth were also noted; more than 10% of Hispanic and non-Hispanics now have class II/III obesity in the United States yet their rates of completing MBS are well below that of non-Hispanic White youth. These findings have implications for pediatricians, primary care, and other specialists who may have young patients with class II/III obesity in their care and thus have an option to refer these patients to MBS. Recent guidance from the American Academy of Pediatrics supports greater access to MBS for youth.14 This endorsement was guided by evidence-based reports to assist pediatric medical providers with patient selection for MBS, how to support families and patients through steps of the decision to pursue MBS process, and how to advocate for reimbursement.3,30 Although MBS is an invasive treatment option, and perhaps not an agreeable weight loss solution for many youth and their families, it is an evidence-based, safe (low mortality, readmission, and reoperation rates), and effective tool that healthcare providers can introduce to their patients. Yet Woolford et al31 found that 48% of physicians would never refer an adolescent with class II/III obesity for MBS and 46% would not make a referral until the patient was 18 years old. In addition, the American Academy of Pediatrics policy statement3 highlights “watchful waiting,” defined as long-term lifestyle management as a barrier to MBS since youth with class II/III obesity are unlikely to benefit from this treatment approach. This is particularly relevant for providers who take care of ethnic minority youth with class II/III obesity who are disproportionately impacted versus their non-Hispanic White peers. Yet, the disparity between those who qualify for MBS and who complete it remains wide. Results here suggest healthcare providers who have young patients with class II/III obesity wait longer to refer to MBS versus adult providers; almost 11% more youth enter MBS with a BMI of 40 kg/m2 or more compared with adults. In addition, results showed the odds of readmissions were higher for both adolescent and adult patients with more comorbidities compared with those with no comorbidities. Thus, it may be important to not only advocate for access to MBS for youth, but at younger ages, and ideally before these comorbidities advance, and thus may add to complications or readmissions postoperatively. As such the AAP policy statement3 recommends that pediatricians (1) recognize that patients with class II and III obesity are at risk for many chronic physical and mental health conditions; (2) identify patients who qualify for MBS, and provide referrals to experienced multidisciplinary centers and MBS programs/practices when appropriate; (3) acquire basic working knowledge of MBS including its risks, benefits, and short- and long-term health benefits to assist families with the decision to pursue surgery for their child; (4) coordinate pre- and post-MBS care with the patient, their family, and the surgical team; (5) provide post-MBS monitoring of micronutrient deficiencies and if necessary recommend iron, folate, and vitamin B12 supplementation as needed; and (6) provide post-MBS monitoring of risk-taking behavior and mental health problems. Given the ethnic group disparities in MBS utilization rates found here, the above policy recommendations will be particularly relevant among those pediatricians and pediatric practices that serve these patients and their families. Other socioecological (personal, social, community) factors beyond the policy level have been also cited as barriers and facilitators to MBS completion among ethnically diverse adult patients.32–35 Specifically, patients identified motivation for change in body image, less discrimination due to weight, and improvement in self-esteem as facilitators to completing MBS and lack of social support before and after surgery as barriers.32,33 Among adolescents, authors have called attention to other barriers to MBS utilization including missed school/university for youth, and time from work (parents) for pre- and postoperative appointments in addition to the surgery and recuperation period.34 Results here showed that between 2015 and 2018 the number of surgeries among youth through age 19 decreased by 269 surgeries while during the same time period increased by more than 35,000 surgeries in adults in MBSAQIP centers. Although some of this increase among adults may be due to the addition of 112 centers over the same time period the question of how, when the national class II/III obesity epidemic is deepening, the same trend is not seen in youth. It should be noted that this decrease was largely driven by youth with class II obesity, while in fact there was an increase in the MBS completion rate in youth with class III obesity, whereas adult rates increased in both obesity classes. This discrepancy again points to a trend in delay of MBS referral among youth, especially with class II obesity. Yet studies have consistently shown that MBS is the most effective and durable treatment for class II/III obesity resulting in significant weight loss and comorbidity resolution in both the short and long term.12,35–38 In another national sample of adolescent patients, our team has shown that MBS can safely and substantially reduce weight and related comorbidities in morbidly obese adolescents for at least 1 year,12 whereas other United States multisite longitudinal cohort studies have documented similar TBWL outcomes among youth and adults at 3,36 5,37 and 5 to 12 years38 post-MBS. Specifically, 3-year follow-up results have shown that in addition to weight loss, significant improvement in joint pain, impaired physical function, and impaired health-related quality of life significantly improve after MBS.36 A 5-year post-MBS comparison between adolescents and adults who underwent gastric bypass showed similar weight loss trajectories, yet adolescents showed greater remission of diabetes and hypertension than adults.37 Finally, 1 study followed up patients who had competed Roux-en-Y gastric bypass surgery at ages 13 to 21. Five to 12 years later (mean 8 years follow-up patients had maintained durable weight loss and cardio- metabolic benefits).38 Updated results here in a different dataset showed a mortality rate of 0.1 for both youth and adults and a reoperation rate of less than 2% for both age groups, and hospital readmission rate was 3.4% for youth and 4.2% for adults. This is similar to other studies that show the risk of death is 0.1%, and the overall likelihood of major complications is about 4%.39–44 Results here showed ~7% of both age groups made an emergency department visit within 30 days post-MBS. This rate is slightly lower than other recent international group reports (8%)45 and national reports ranging from 10% to 12%.46,47 A simulation study published 3 years ago assessed the risk for adult obesity based on childhood obesity status.48 The authors reported that if a child had class II/III obesity at age 19, their chances of being normal weight at age 35 was 3.5% among boys, and 8.2% among girls. Although this study correctly pointed out that obesity prevention must start very early in life, their findings also draw attention to only a fraction of teenagers with class II/III obesity who will transition to healthy weight through early adulthood. NHANES results here support these simulation models and may even be a bit chronologically ahead in terms of class II/III obesity rates increasing in US youth. Results here show MBS is a safe, effective, and underutilized option for weight loss in this group. Limitations and Strengths Only some limitations should be noted concerning this analysis. MBSAQIP center data was used for this analysis, and thus not representative of all MBS practices in the nation. Similarly, we calculated utilization rates using only MBSAQIP affiliated centers, so all unaffiliated centers were not included in the numerator thus potentially leading to an underestimation of the true positive/utilization rates, and thus lower sensitivity. However, a recent analysis of both MBSAQIP and National Inpatient Survey data that included both adolescents and adults reported that in 2018 approximately 80% of surgeries are performed at MBSAQIP-accredited centers. Moreover, National Inpatient Survey data showed only 8% of MBS centers were nonaccredited in 2015 to 2016 and this decreased to 4.9% in 2018.49 Perioperative data were only available for 30 days post-MBS which limits longer-term capture of weight loss measures and complication rates. However, the primary purpose of this study was not to examine the outcome of MBS, but rather the focus was on capturing the most recent national MBS utilization rates, especially given the rise in class II/III obesity. Although the MBSA-QIP database maintains a rigorous auditing process for data accuracy, some cases may contain errors or omissions that could produce reporting bias. We were not able to examine the impact of insurance coverage on the MBS utilization rate due to data unavailability; however, insurance is not the only influential factors for MBS utilization.34 As mentioned previously, others have called attention to a intrapersonal, interpersonal, environmental, and community barriers to MBS utilization among adolescents.32–34 Indeed, our group has shown that there are a number of socioecological (intra- and interpersonal, community, policy) factors that drive the decision to complete MBS among patients from a variety of ethnic back-grounds.32,33 These also include, other important factors such as age and comorbidities have been rigorously controlled in this study. Finally, the database does not document surgical variation in intra-operative technique, which may affect patient outcomes. Although previous studies have also observed racial disparities of MBS utilization,50–53 no studies have reported the actual national utilization rate among US youths aged 10 to 19 years, especially using the most current data available (up to 2018). Strengths of the study include large youth and adult sample sizes among diverse patients in several hundred centers and the first time a US MBS prevalence rate (as opposed to number or percentage) for both adults and youth has been calculated in the literature. CONCLUSIONS US 2017 to 2018 population-based estimates show that approximately 4.8 million US youth have class II/III obesity, whereas the MBSAQIP data during the same time period included a total of 1862 surgeries among those 19 years or younger. More than 80% of youth had a pre-op BMI of 40 kg/m2 or more compared to 73.05% in adults. In context to the increases in youth class II/III obesity trends, MBS is as safe and effective as in adults and may provide benefits such as the delay or elimination of chronic disease onset at an earlier age but is currently underutilized as a treatment option, especially among ethnic minority group. ACKNOWLEDGMENTS The authors would like to thank the American Society for Metabolic and Bariatric Surgery for the use of the 2015–2018 MBSAQIP PUF file for this analysis. This work was funded by the National Institutes of Health, National Institute on Minority Health, and Health Disparities (grant R01MD011686). FIGURE 1. A, class II/III obesity Rate of children aged 2 to 19 years with class II/III obesity, NHANES 1999 to 2018. TABLE 1. Preoperative Characteristics of Youth 19 Years or Younger and Adult Patients Who Completed Metabolic and Bariatric Surgery, Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program, 2015 to 2018 Variable Youth (n = 3846) Adults (n = 744,776) P Age, mean (SD) 17.9 (1.15) 45.1 (11.5) <0.001 Sex, female, n (%) 2978 (77.4%) 600,838 (80.7%) <0.001 Race/ethnicity, n (%) <0.001  Hispanic 709 (18.4%) 67,430 (9.1%)  Non-Hispanic White 1925 (68.5%) 44,2271 (59.4%)  Non-Hispanic Black 597 (15.5%) 131,732 (17.7%)  Asian 19 (0.5%) 3661 (0.5%)  American Indian 15 (0.4%) 3003 (0.4%)  Other/unknown 581 (15.1%) 96,679 (12.9%) Pre-op BMI closest to bariatric surgery, mean (SD) 46.7 (8.48) 44.5 (8.40) <0.001  BMI <35, n (%) 227 (5.9%) 68,238 (9.2%) <0.001  35 ≤ BMI < 40, n (%) 574 (14.9%) 167,555 (22.5%)  40 ≤ BM < 50, n (%) 1968 (51.2%) 353,886 (47.5%)  BMI ≥50, n (%) 1077 (28.0%) 155,097 (20.8%) Highest pre-op BMI, mean (SD) 48.4 (9.0) 46.5 (8.9) <0.001  BMI <35, n (%) 385 (10.1%) 88,126 (11.8%) <0.001  35 ≤ BMI < 40, n (%) 316 (18.2%) 112,595 (15.1%)  40 ≤ BMI < 50, n (%) 1877 (48.8%) 350,273 (47.0%)  BMI ≥50, n (%) 1268 (32.9%) 193,782 (26.0%) Procedure type, n (%) <0.001  LSG 2818 (73.3%) 457,698 (61.5%)  LRYGB 654 (17.0%) 177,344 (23.8%)  Other 374 (9.7%) 109,734 (14.7%) Comorbidities, n (%)  Hypertension 1532 (39.8%) 429,505 (57.7%) <0.001  Hyperlipidemia 178 (4.6%) 168,089 (22.6%) <0.001  Type 2 diabetes 521 (13.6%) 180,415 (24.2%) <0.001  GERD 554 (14.4%) 238,005 (31.9%) <0.001  Sleep apnea 655 (17.0%) 261,378 (35.1%) <0.001 Chronic steroid use, n (%) 39 (1.01%) 13,089 (1.8%) <0.001 N surgeries by year*, n (%) <0.001  2015 1199 (31.2%) 165,017 (22.2%)  2016 785 (20.4%) 183,219 (24.6%)  2017 932 (24.2%) 196,243 (26.4%)  2018 930 (24.2%) 200,250 (26.8%) LRYGB indicates laparoscopic Roux-en-Y gastric bypass; LSG, laparoscopic sleeve gastrectomy. * Total MBSAQIP centers by year; 2015 (n = 742); 2016 (n = 791); 2017(n = 832); 2018 (n = 854); n=47 for "N surgeries by year" in adults. TABLE 2. Outcome Data for All Pediatric and Adult Patients Who Completed Metabolic and Bariatric Surgery, Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program 2015–2018 N = 748,622 Variable Youth n = 3846 Adult n = 744,776 P  Mortality   6 (0.1%)   800 (0.1%)   0.359  Reoperation    48 (1.3%) 11,594 (1.6%)   0.123  Acute renal failure   2 (0.1%)   575 (0.1%)   0.574 Bleeding event    19 (0.5%)    5638 (0.8%)   0.060 ≥Intervention w/in 30 days    46 (1.2%) 11,498 (1.5%)   0.081  Unplanned intubation   3 (0.1%)   891 (0.1%)   0.456  Sepsis   2 (0.1%)   278 (0.1)   0.639  ICU admission    20 (0.5%)   595 (0.8%)   0.052  Hospital readmission  129 (3.4%) 30,984 (4.2%)   0.013  Ventilator, 48 h   0 (0.0%)     69 (0.0%)   0.550  Approach conversion   0 (0.0%)   321 (9.6%)   0.087 Pulmonary embolism   3 (0.1%)   886 (0.1%)   0.462  Clostridium difficile colitis   2 (0.1%)   759 (0.1%)   0.432  ED visits  177 (6.7%) 39,529 (6.8%)   0.789  UTI   2 (0.0%)   238 (0.0%)   0.489 Pneumonia   1 (0.0%)     94 (0.0%)   0.463 Incisional SSI   0 (0.0%)   124 (0.0%)   0.424 Length of stay, day  1.7 (1.8)    1.7 (2.1)   0.761 Operation length, min 80.4 (45.4)  88.4 (53.8) <0.001 SSI indicates surgical site infection; UTI, urinary tract infection. TABLE 3. Odds of Hospital Readmission for Youth and Adults by Patient Characteristics and Comorbidity Status, Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program, 2015–2018. Youth Adults Variable OR (95% CI) P OR (95% CI) P Sex  Female (ref) 1.0 — 1.0 —  Male 0.7 (0.4,1.1) 0.116 0.9 (0.8,0.9) <0.001 Race/ethnicity  Non-Hispanic White (ref) 1.0 — 1.0 —  Hispanic 1.0 (0.6,1.6) 0.873 1.1 (1.0, 1.1) <0.001  Non-Hispanic Black 1.0 (0.6,1.6) 0.975 1.4 (1.4,1.5) <0.001  Other 0.9 (0.5,1.5) 0.734 0.9 (0.9, 1.0) <0.001 Procedure type  LRYGB (ref) 1.0 — 1.0 —  LSG 0.4 (0.3,0.7) 0.001 0.5 (0.4, 0.5) <0.001  Other 0.7 (0.4,1.3) 0.786 0.9 (0.9, 1.0) <0.001 Hypertension  No (ref) 1.0 — 1.0 —  Yes 1.1 (0.8,1.6) 0.680 1.1 (1.0, 1.1) <0.001 Hyperlipidemia  No (ref) 1.0 — 1.0 —  Yes 0.9 (0.4,2.1) 0.724 1.1 (1.0, 1.1) <0.001 Type 2 diabetes  No (ref) 1.0 — 1.0 —  Yes 1.0 (0.6,1.6) 0.873 1.1 (1.0, 1.1) <0.001 GERD  No (ref) 1.0 — 1.0 —  Yes 1.0 (0.6,1.6) 0.996 1.4 (1.3, 1.4) <0.001 Sleep apnea  No (ref) 1.0 — 1.0 —  Yes 1.3 (0.8,2.1) 0.263 1.1 (1.0, 1.1) <0.001 Chronic steroid use  No (ref) 1.0 — 1.0 —  Yes 3.5 (1.2,10.4) 0.020 1.5 (1.4, 1.6) <0.001 LRYGB indicates laparoscopic Roux-en-Y gastric bypass; LSG, laparoscopic sleeve gastrectomy. TABLE 4. Metabolic and Bariatric Surgery Utilization Rates Among Youths by Demographic Characteristics, Calculated Using 2015 to 2018 Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program and National Health and Nutrition Examination Survey Data MBSAQIP NHANES* MBS Utilization Rate† (per 1000) 2015–2016 2017–2018 2015–2016 2017–2018 2015–2016 2017–2018 Age, N (%)‡  Class II obesity with ≥1 comorbidity§   10–12 y  53 (2.8)  70 (4.3)    210,989 (0.5)   42,838 (0.1)    0.25‖    1.63‖   13–16 y   113 (6.0)  27 (1.7)    126,593 (0.3)    385,546 (0.9)    0.89‖    0.07‖   17–19 y 1131 (60.3)   257 (15.8)    253,186 (0.6)    342,707 (0.8)    4.47‖    0.75‖  Class III obesity¶   10–12 y    8 (0.4)  26 (1.6)    506,372 (1.2)    771,092 (1.8)    0.01    0.03‖   13–16 y  32 (1.7) ‰86 (5.3) 1,139,338 (2.7)    985,283 (2.3)    0.03    0.09‖   17–19 y   453 (24.0) 1034 (63.3) 1,392,524 (3.3) 1,327,991 (3.1)    0.33‖    0.77‖  Total#   10–12 y  99 (5.3)   173 (10.6)    717,361 (1.7)    813,930 (1.9)    0.14    0.21‖   13–16 y   149 (7.9)   119 (7.2) 1,265,931 (3.0) 1,370,829 (3.2)    0.12    0.09‖   17–19 y 1636 (86.8) 1341 (82.1) 1,645,710 (3.9) 1,670,698 (3.9)    0.99‖    0.80‖ Race/ethnicity, N (%)**  Class II obesity with ≥1 comorbidity§   Non-Hispanic White   666 (35.4)   164 (10.1)   84,395 (0.2)      0 (0)    7.89‖     —   Non-Hispanic Black   181 (9.6)   168 (10.3)    210,988 (0.5)    813,930 (1.9)    0.86‖    0.21   Hispanic   224 (11.9)  71 (4.4)    506,372 (1.2)    428,384 (1.0)    0.44‖    0.17   Other   226 (12.0)   197 (12.1)      0 (0)    514,061 (1.2)    –    0.38  Class III obesity¶   Non-Hispanic White   234 (12.4)   541 (33.1)    464,175 (1.1)    599,738 (1.4)    0.50‖    0.90‖   Non-Hispanic Black  85 (4.5)  62 (3.8) 2,447,466 (5.8) 3,212,881 (7.5)    0.03‖    0.02‖   Hispanic  82 (4.4)   240 (14.7) 2,025,490 (4.8)    985,284 (2.3)    0.04‖    0.24‖   Other  92 (4.9)  57 (3.5)    168,791 (0.4)    899,607 (2.1)    0.54‖    0.06‖  Total#   Non-Hispanic White   962 (51.1)   797 (48.8)    548,570 (1.3)    599,738 (1.4)    1.75‖    1.33‖   Non-Hispanic Black   273 (14.5)   235 (14.4) 2,679,544 (6.3) 4,026,811 (9.4)    0.10‖    0.06‖   Hispanic   317 (16.8)   325 (19.9) 2,531,862 (6.0) 1,413,668 (3.3)    0.12‖    0.23‖   Other   332 (17.6)   276 (16.9)    168,791 (0.4) 1,413,668 (3.3)    1.96    0.20‖ Sex, N (%)††  Class II obesity with ≥1 comorbidity‡‡   Males   279 (14.8)   106 (6.5)    377,582 (0.8)    428,384 (1.0)    0.83‖    0.25‖   Females 1018 (54.0)   248 (15.2)   84,395 (0.2)    128,515 (0.3)     12.06‖    1.93‖  Class III obesity¶   Males   111 (5.9)   284 (17.4) 1,181,536 (2.8) 1,113,799 (2.6)    0.09‖    0.25‖   Females   382 (20.3)   862 (52.8)    886,152 (2.1) 1,028,122 (2.4)    0.43‖    0.84‖  Total#   Males   405 (21.5)   414 (25.4) 1,519,117 (3.6) 1,542,183 (3.6)    0.27‖    0.27‖   Females 1479 (78.5) 1219 (74.6)    970,547 (2.3) 1,156,637 (2.7)    1.52‖    1.05‖ Overall utilization rate across all 4 years, all groups‡‡ = 1.81 per 1000‖ * Population estimates obtained by using weighted obesity rates from NHANES multiplied by the number of US population ages 10 to 19 years from the US census bureau (https://www.census.gov/). † Numerator = total number of surgeries during time period; denominator = total number of class II obese patients with at least 1 comorbidity and class III patient with obesity during time period. ‡ Two-sample test of proportions compared MBS utilization rates by age groups; 10 to 12 years is the reference group. § Class II obesity defined as 120% ≤ BMI < 140% of the 95th percentile for age and sex. ‖ P < 0.001. ¶ Class III obesity defined as BMI ≥ 140% of the 95th percentile for age and sex. # The total number of MBSAQIP included all patients who had surgery regardless of obesity status. ** Two-sample test of proportions compared MBS utilization rates by race and ethnicity; non-Hispanic White is the reference group. †† Two-sample test of proportions compared MBS utilization rates by sex; males is the reference group. ‡‡ Two-sample test of proportions compared overall MBS utilization rates between youth and adults; adults is the reference group. TABLE 5. The Comparison of Metabolic Bariatric Surgery Utilization Rates Among Adults, Calculated Using 2015 to 2018 Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program and National Health and Nutrition Examination Survey Data MBSAQIP NHANES* MBS Utilization Rate† (per 1000) 2015–2016 2017–2018 2015–2016 2017–2018 2015–2016 2017–2018 Age, N (%)‡  Class II obesity with ≥1 comorbidity§   20–39 y   71,830 (21.8)   43,093 (12.1)   8,663,334 (3.6) 11,754,087 (4.8)    8.29‖    3.67‖   ≥40 y 167,353 (50.7) 148,939 (41.9) 16,845,371 (7.0) 18,120,884 (7.4)    9.93‖    8.22‖  Class III obesity¶   20 – 39 y   33,568 (10.2)   73,788 (20.8) 19,492,500 (8.1) 23,753,050 (9.7)    1.72‖    3.11‖   ≥40 y   24,772 (7.5)   53,661 (15.1) 18,529,908 (7.7) 22,773,543 (9.3)    1.34‖    2.36‖  Total#   20–39 y 112,561 (34.1) 124,328 (35.0) 28,155,834 (11.7) 35,507,137 (14.5)    4.0‖    3.50‖   ≥40 y 217,279 (65.9) 231,072 (65.0) 35,375,278 (14.7) 40,894,427 (16.7)    6.14‖    5.65‖ Race/ethnicity, N (%)**  Class II obesity with ≥1 comorbidity§   Non-Hispanic White 143,811 (43.6) 112,895 (31.8) 13,235,648 (5.5) 16,406,746 (6.7)     10.87‖    6.88‖   Non-Hispanic Black   39,840 (12.1)   37,013 (10.4) 18,289,260 (7.6) 19,345,268 (7.9)    2.18‖    1.91‖   Hispanic   20,318 (6.1)   15,481 (4.4) 15,882,778 (6.6) 16,406,746 (6.7)    1.28‖    0.94‖   Other   35,214 (10.7)   26,643 (7.5) 10,829,167 (4.5)   7,346,304 (3.0)    3.25‖    3.63‖  Class III obesity¶   Non-Hispanic White   33,365 (10.1)   71,523 (20.1) 18,289,260 (7.6) 22,773,543 (9.3)    1.82‖    3.14‖   Non-Hispanic Black  9,223 (2.8)   21,397 (6.0) 28,396,482 (11.8) 35,752,014 (14.6)    0.32‖    0.60‖   Hispanic  5,745 (1.7)   13,135 (3.7) 22,861,575 (9.5) 20,324,775 (8.3)    0.25‖    0.65‖   Other   10,007 (3.0)   21,394 (6.0)   6,738,148 (2.8) 15,916,993 (6.5)    1.49‖    1.34‖  Total#   Non-Hispanic White 198,356 (60.1) 207,869 (58.5) 31,524,908 (13.1) 39,180,289 (16.0)    6.29‖    5.31‖   Non-Hispanic Black   53,302 (15.9)   62,142 (17.5) 46,685,742 (19.4) 55,097,282 (22.5)    1.14‖    1.13‖   Hispanic   28,806 (8.7)   31,778 (8.9) 38,744,353 (16.1) 36,731,521 (15.0)    0.74‖    0.87‖   Other   52,376 (15.3)   53,611 (15.1) 17,567,315 (7.3) 23,263,297 (9.5)    2.98‖    2.30‖ Sex, N (%)††  Class II obesity with ≥1 comorbidity§   Males   52,962 (16.1)   47,303 (13.3) 11,551,111 (4.8) 15,427,239 (6.3)    4.59‖    3.07‖   Females 186,221 (56.5) 144,729 (40.7) 16,123,426 (6.7) 16,406,746 (6.7)     11.55‖    8.82‖  Class III obesity¶   Males  8,899 (2.7)   18,689 (5.3) 13,716,945 (5.7) 17,386,253 (7.1)    0.65‖    1.07‖   Females   49,441 (15.0) 108,760 (30.6) 23,824,167 (9.9) 28,650,587 (11.7)    2.08‖    3.80‖  Total#   Males   66,345 (20.1)   71,042 (20.0) 25,268,056 (10.5) 32,813,492 (13.4)    2.63‖    2.17‖   Females 263,495 (79.9) 284,358 (80.0) 39,947,593 (16.6) 45,057,333 (18.4)    6.60‖    6.31‖ Overall utilization rate across all 4 years, all groups‡‡ = 5.56 per 1000‖ * Population estimates obtained by using weighted obesity rates from NHANES multiplied by the number of US population ages 20 years and above from the US census bureau (https://www.census.gov/). † Numerator = total number of surgeries during time period; denominator = total number of class II obese patients with at least 1 comorbidity and class III obese patient during time period. ‡ Two-sample test of proportions compared MBS utilization rates by age groups. § Class II obesity defined as 35 ≤BMI < 40 in adults. ‖ P < 0.001. ¶ Class III obesity defined as BMI ≥40 in adults. # The total number of MBSAQIP included all patients who had surgery regardless of obesity status. ** Two-sample test of proportions compared MBS utilization rates by race and ethnicity; non-Hispanic White is the reference group. †† Two-sample test of proportions compared MBS utilization rates by sex. ‡‡ Two-sample test of proportions compared overall MBS utilization rates between youth and adults. The authors declare no conflict of interests. REFERENCES 1. Ogden CL , Fryar CD , Hales CM , Differences in obesity prevalence by demographics and urbanization in US children and adolescents, 2013–2016. JAMA. 2018;319 :2410–2418.29922826 2. Hales CM , Fryar CD , Carroll MD , Trends in obesity and severe obesity prevalence in US youth and adults by sex and age, 2007–2008 to 2015–2016. JAMA. 2018;319 :1723–1725.29570750 3. Armstrong SC , Bolling CF , Michalsky MP , , SECTION ON OBESITY, SECTION ON SURGERY. Pediatric metabolic and bariatric surgery: evidence, barriers, and best practices. Pediatrics. 2019;144 :e20193223.31656225 4. Kelly SA , Barlow SE , Rao HG , , American Heart Association Atherosclerosis, Hypertension, and Obesity in the Young Committee of the Council on Cardiovascular Disease in the Young, Council on Nutrition, Physical Activity and Metabolism, and Council on Clinical Cardiology.. Severe obesity in children and adolescents: Identification, associated health risks, and treatment approaches: a scientific statement from the American Heart Association. Circulation. 2013;128 :1689–1712.24016455 5. Rank M , Siegrist M Wilks D , .The cardio- metabolic risk of moderate and severe obesity in children and adolescents. J Pediatr. 2013;163 :137–142.23434121 6. Christou NV , Sampalis JS , Liberman M , Surgery decreases long-term mortality, morbidity, and health care use in morbidly obese patients. Ann Surg. 2004;240 :416–424.15319713 7. Golden SH , Brown A , Cauley JA , Health disparities in endocrine disorders: biological, clinical, and nonclinical factors–an Endocrine Society scientific statement. J Clin Endocrinol Metab. 2012;97 :E1579–E1639.22730516 8. Pratt LA , Brody DJ . Depression and obesity in the US adult household population, 2005–2010. NCHS Data Brief. 2014;1–8. 9. Robert Wood Johnson Foundation. The State of Obesity 2019: Better Policies for a Healthier America. Available at: https://www.stateofobesity.org/ Accessed January 20, 2020. 10. Xu J , Murphy S , Kochanek K , Mortality in the United States, 2015. NCHS Data Brief, no 267. Hyattsville, MD: US Department of Health and Human Services, CDC, National Center for Health Statistics; 2016. 11. Pratt JSA , Browne A , Browne NT , ASMBS pediatric metabolic and bariatric surgery guidelines, 2018. Surg Obes Relat Dis. 2018;14 :882–901.30077361 12. Messiah SE , Lopez-Mitnik G , Winegar D , Changes in weight and co-morbidities among adolescents undergoing bariatric surgery: 1-year results from the Bariatric Outcomes Longitudinal Database. Surg Obes Relat Dis. 2013;9 :503–513.22542199 13. Inge TH , Zeller MH , Jenkins TM , , Teen-LABS Consortium. Perioperative outcomes of adolescents undergoing bariatric surgery: the Teen-Longitudinal Assessment of Bariatric Surgery (Teen-LABS) study. JAMA Pediatr. 2014;168 :47–53.24189578 14. American Academy of Pediatrics Available at: https://www.aappublications.org/news/aapnewsmag/2019/10/27/bariatricsurgery102719.full.pdf). Accessed April 3, 2020. 15. Kelly AS , Auerbach P , Barrientos-Perez M , , NN8022-4180 Trial Investigators. A randomized, controlled trial of liraglutide for adolescents with obesity. N Engl J Med. 2020;382 :2117–2128.32233338 16. Clarson CL , Brown HK , De Jesus S , Effects of a comprehensive, intensive lifestyle intervention combined with metformin extended release in obese adolescents. Int Sch Res Notices. 2014;2014 :659410.27433488 17. Van der Baan-Slootweg O , Benninga MA , Beelen A , Inpatient treatment of children and adolescents with severe obesity in the Netherlands: a randomized clinical trial. JAMA Pediatr. 2014;168 :807–814.25022831 18. Danne T , Biester T , Kapitzke K , Liraglutide in an adolescent population with obesity: a randomized, double-blind, placebo-controlled 5-week trial to assess safety, tolerability, and pharmacokinetics of liraglutide in adolescents aged 12–17 years. J Pediatr. 2017;181 :146.e3–153.e3.27979579 19. Wilson DM , Abrams SH , Aye T , , Glaser Pediatric Research Network Obesity Study Group. Metformin extended release treatment of adolescent obesity: a 48-week randomized, double-blind, placebo-controlled trial with 48-week follow-up. Arch Pediatr Adolesc Med. 2010;164 :116–123.20124139 20. Makkes S , Renders CM , Bosmans JE , One-year effects of two intensive inpatient treatments for severely obese children and adolescents. BMC Pediatr. 2016;16 :120.27484299 21. Brufani C , Crinò A , Fintini D , Systematic review of metformin use in obese nondiabetic children and adolescents. Horm Res Paediatr. 2013;80 :78–85.23899569 22. American Society for Metabolic and Bariatric Surgery. Estimate of Bariatric Surgery Numbers, 2011–2018. Available at: https://asmbs.org/resources/estimate-of-bariatricsurgery-numbers, Accessed April 9, 2020. 23. CDC. National Health and Nutrition Examination Survey. 2018. Available at: https://wwwn.cdc.gov/nchs/nhanes/tutorials/module3.aspx. Accessed February 14, 2020. 24. Telem DA , Dimick JB . Practical guide to surgical data sets: Metabolic and Bariatric Surgery Accreditation and Quality Program (MBSAQIP). JAMA Surg. 2018;153 :766–767.29617538 25. User guide for the MBSAQIP 2017 participant use file [homepage on the Internet]. Chicago: American College of Surgeons; c1996-2019 [updated October 2018; cited June 2, 2019]. Available at: facs.org/-/media/files/quality-rograms/bariatric/mbsaqip_2017_puf_user_guide.ashx. 26. Pratt JSA , Browne A , Browne NT , ASMBS pediatric metabolic and bariatric surgery guidelines. Surg Obes Relat Dis. 2018;14 :882–901.30077361 27. Whelton PK , Carey RM , Aronow WS , 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines [published correction appears in Hypertension. 2018 Jun;71(6):e136-e139] [published correction appears in Hypertension. 2018 Sep;72(3):e33]. Hypertension. 2018;71 :1269–1324.29133354 28. Grundy SM , Stone NJ , Bailey AL , 2018 AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA guideline on the management of blood cholesterol: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines [published correction appears in J Am Coll Cardiol. 2019 Jun 25;73(24):3237-3241]. J Am Coll Cardiol. 2019;73 :e285–e350.30423393 29. American Diabetes, Association. 2. Classification and diagnosis of diabetes: standards of medical care in diabetes-2020. Diabetes Care. 2020;43 (suppl 1 ):S14–S31.31862745 30. Bolling CF , Armstrong SC , Reichard KW , , SECTION ON OBESITY, SECTION ON SURGERY. Metabolic and bariatric surgery for pediatric patients with severe obesity. Pediatrics. 2019;144 :e20193224 31656226 31. Keeton H , Ofori A , Booker Q , Socioecological factors that inform the decision to have metabolic and bariatric surgery utilization in ethnically diverse patients. Obes Surg. 2020 [Epub ahead of print]. 32. Ofori A , Keeton H , Booker Q , Socioecological factors associated with metabolic and bariatric surgery utilization: a qualitative study. Surg Obes Relat Dis. 2020;piiS1550-7289:30073–30083. 33. Ogle SB , Inge TH , Campbell EG . Comment on: Beyond insurance: race-based disparities in the use of metabolic and bariatric surgery for the management of severe pediatric obesity. Surg Obes Relat Dis. 2020;16 :419–421.32007434 34. Woolford SJ , Clark SJ , Gebremariam A , To cut or not to cut: physicians’ perspectives on referring adolescents for bariatric surgery. Obes Surg. 2010;20 :937–942.20401742 35. O’Brien PE , Hindle A , Brennan L , Long-term outcomes after bariatric surgery: a systematic review and meta-analysis of weight loss at 10 or more years for all bariatric procedures and a single-centre review of 20-year outcomes after adjustable gastric banding. Obes Surg. 2019;29 :3–14.30293134 36. Bout-Tabaku S , Gupta R , Jenkins TM , Musculoskeletal pain, physical function, and quality of life after bariatric surgery. Pediatrics. 2019;144 :e20191399.31744891 37. Inge TH , Jenkins TM , Xanthakos SA , Long-term outcomes of bariatric surgery in adolescents with severe obesity (FABS-5+): a prospective follow-up analysis. Lancet Diabetes Endocrinol. 2017;5 :165–173.28065736 38. Inge TH , Courcoulas AP , Jenkins TM , , Teen-LABS Consortium. Five-year outcomes of gastric bypass in adolescents as compared with adults. N Engl J Med. 2019;380 :2136–2145.31116917 39. Qi L , Guo Y , Liu CQ , Effects of bariatric surgery on glycemic and lipid metabolism, surgical complication and quality of life in adolescents with obesity: a systematic review and meta-analysis. Surg Obes Relat Dis. 2017;13 :2037–2055.29079384 40. Spirou D , Raman J , Smith E . Psychological outcomes following surgical and endoscopic bariatric procedures: a systematic review. Obes Rev. 2020;21 :e12998.31994311 41. Sjöström L , Narbro K , Sjöström CD , , Swedish Obese Subjects Study. Effects of bariatric surgery on mortality in Swedish obese subjects. N Engl J Med. 2007;357 :741–752.17715408 42. Agency for Healthcare Research and Quality (AHRQ). (2007). Statistical Brief 23. Bariatric Surgery Utilization and Outcomes in 1998 and 2004. Available at: http://www.hcup-us.ahrq.gov/reports/statbriefs/sb23.jsp. Accessed March 21, 2020. 43. Flum DR , Belle SH , King WC , , Longitudinal Assessment of Bariatric Surgery (LABS) Consortium. Perioperative safety in the longitudinal assessment of bariatric surgery. N Engl J Med. 2009;361 :445–454.19641201 44. Encinosa WE , Bernard DM , Du D , Recent improvements in bariatric surgery outcomes. Med Care. 2009;47 :531–535.19318997 45. Ahmed A , AlBuraikan D , ALMuqbil B , Readmissions and emergency department visits after bariatric surgery at Saudi Arabian Hospital: the rates, reasons, and risk factors. Obes Facts. 2017;10 :432–443.28988235 46. Alvarez R , Matusko N , Varban O . Characterizing the preventable emergency department visit after bariatric surgery. Surg Obes Relat Dis. 2020;16 :48–55.31744733 47. Telem DA , Yang J , Altieri M , Rates and risk factors for unplanned emergency department utilization and hospital readmission following bariatric surgery. Ann Surg. 2016;263 :956–960.26727087 48. Ward ZJ , Long MW , Resch SC , Simulation of growth trajectories of childhood obesity into adulthood. N Engl J Med. 2017;377 :2145–2153.29171811 49. English WJ , DeMaria EJ , Hutter MM , American Society for Metabolic and Bariatric Surgery 2018 estimate of metabolic and bariatric procedures performed in the United States. Surg Obes Relat Dis. 2020;16 :457–463.32029370 50. Perez NP , Westfal ML , Stapleton SM , Beyond insurance: race-based disparities in the use of metabolic and bariatric surgery for the management of severe pediatric obesity. Surg Obes Relat Dis. 2020;16 :414–419.31917198 51. Campos GM , Khoraki J , Browning MG , Changes in utilization of bariatric surgery in the United States from 1993 to 2016. Ann Surg. 2020;271 :201–209.31425292 52. Campoverde Reyes KJ , Misra M , Lee H , Weight loss surgery utilization in patients aged 14-25 with severe obesity among several healthcare institutions in the United States. Front Pediatr. 2018;6 :251.30283764 53. Kindel TL , Lomelin D , McBride C , Plateaued national utilization of adolescent bariatric surgery despite increasing prevalence of obesity-associated co-morbidities. Surg Obes Relat Dis. 2016;12 :868–873.26775046

PMC009xxxxxx/PMC9825120.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101573691 39703 Cell Rep Cell Rep Cell reports 2211-1247 36516757 9825120 10.1016/j.celrep.2022.111803 NIHMS1859705 Article The hepatic integrated stress response suppresses the somatotroph axis to control liver damage in nonalcoholic fatty liver disease Ohkubo Rika 1289 Mu Wei-Chieh 239 Wang Chih-Ling 29 Song Zehan 12 Barthez Marine 2 Wang Yifei 12 Mitchener Nathaniel 2 Abdullayev Rasul 2 Lee Yeong Rim 23 Ma Yuze 2 Curtin Megan 2 Srinivasan Suraj 2 Zhang Xingjia 2 Yang Fanghan 23 Sudmant Peter H. 45 Pisco Angela Oliveira 6 Neff Norma 6 Haynes Cole M. 7 Chen Danica 12310* 1 Metabolic Biology Graduate Program, University of California, Berkeley, Berkeley, CA 94720, USA 2 Department of Nutritional Sciences and Toxicology, University of California, Berkeley, Berkeley, CA 94720, USA 3 Endocrinology Graduate Program, University of California, Berkeley, Berkeley, CA 94720, USA 4 Department of Integrative Biology, University of California, Berkeley, Berkeley, CA 94720, USA 5 Center for Computational Biology, University of California, Berkeley, Berkeley, CA 94720, USA 6 Chan Zuckerberg Biohub, San Francisco, CA, USA 7 Department of Molecular, Cell and Cancer Biology, UMass-Chan Medical School, Worcester, MA 01605, USA 8 Present address: MSD K.K., Kitanomaru Square, 1-13-12 Kudan-kita, Chiyoda-Ku, Tokyo 102-8667, Japan 9 These authors contributed equally 10 Lead contact AUTHOR CONTRIBUTIONS D.C. conceived the study. C.-L.W. prepared the samples. W.-C.M. analyzed and A.O.P., N.N., and P.H.S. supervised single-cell RNA sequencing data. R.O. and C.-L.W. characterized SIRT7 KO mice, Myc KD mice, and SIRT7 KO mice with ATF3 KD. W.-C.M., C.-L.W., Y.M., and X.Z. characterized CD-HFD mice with ATF3 KD. C.-L.W., Y.W., N.M., and R.A. characterized IGF1-treated mice. C.-L.W., W.-C.M., M.C., and S.S. characterized 78c-treated mice. Z.S., W.-C.M., M.B., Y.R.L., and R.O. performed studies in cell culture. C.-L.W. and F.Y. studied the link between SIRT7 and NAFLD. C.M.H. advised on ISR. D.C. wrote the manuscript with contributions from all authors. * Correspondence: danicac@berkeley.edu 3 1 2023 13 12 2022 07 1 2023 41 11 111803111803 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. SUMMARY Nonalcoholic fatty liver disease (NAFLD) can be ameliorated by calorie restriction, which leads to the suppressed somatotroph axis. Paradoxically, the suppressed somatotroph axis is associated with patients with NAFLD and is correlated with the severity of fibrosis. How the somatotroph axis becomes dysregulated and whether the repressed somatotroph axis impacts liver damage during the progression of NAFLD are unclear. Here, we identify a regulatory branch of the hepatic integrated stress response (ISR), which represses the somatotroph axis in hepatocytes through ATF3, resulting in enhanced cell survival and reduced cell proliferation. In mouse models of NAFLD, the ISR represses the somatotroph axis, leading to reduced apoptosis and inflammation but decreased hepatocyte proliferation and exacerbated fibrosis in the liver. NAD+ repletion reduces the ISR, rescues the dysregulated somatotroph axis, and alleviates NAFLD. These results establish that the hepatic ISR suppresses the somatotroph axis to control cell fate decisions and liver damage in NAFLD. In brief Ohkubo et al. show that hepatic ER stress results in the suppression of the somatotroph axis, which controls liver damage in nonalcoholic fatty liver disease by preventing hepatocyte death and proliferation, resulting in reduced inflammation but increased fibrosis. This study has important implications in treating this prevalent metabolic disease. Graphical Abstract pmcINTRODUCTION The current challenges in developing therapeutics against nonalcoholic fatty liver disease (NAFLD) reflect its complex nature, raising the question of whether the solution requires a combination of drugs. NAFLD can be ameliorated by calorie restriction, which leads to the suppressed growth hormone/insulin-like growth factor-1 (IGF-1) somatotroph axis, a conserved regulator of lifespan that triggers the activation of the cellular protective program and the re-allocation of resources from growth to somatic preservation.1–9 Paradoxically, suppression of the somatotroph axis is associated with patients with NAFLD and, in particular, is correlated with the severity of fibrosis.10–19 Whether the somatotroph axis controls liver damage during the progression of NAFLD is unknown. NAFLD begins with hepatosteatosis and can progress to nonalcoholic steatohepatitis (NASH) in response to endoplasmic reticulum (ER) stress.20–22 The integrated stress response (ISR) is a critical regulator of protein homeostasis at the cellular and organismal level to control the pathogenesis of complex diseases.23 Little is known about the connectivity of the ISR to other intracellular signaling networks to determine cell fate decisions and physiological output. The growth hormone/IGF-1 somatotroph axis includes the secretion of growth hormone from the somatotropes of the pituitary gland into the circulation and the subsequent stimulation of IGF-1 production, which is synthesized and secreted by the liver.24 While evidence is emerging that systemic ER stress induction leads to the suppressed somatotroph axis,25 whether hepatic ER stress regulates the somatotroph axis autonomously and the molecular mechanism underlying such regulation remain unexplored. In this study, we show that hepatic ER stress suppresses the somatotroph axis autonomously through the transcription factor ATF3. We provide evidence that suppression of the somatotroph axis results in reduced apoptosis and inflammation but decreased hepatocyte proliferation and exacerbated fibrosis in the livers, offering explanations for the paradoxical observations that the suppressed somatotroph axis is associated with patients with NAFLD while calorie restriction suppresses the somatotroph axis and prevents the development of NAFLD at the early stage. Finally, we demonstrate the therapeutic implication of this regulatory pathway for NAFLD. RESULTS A mouse model of NAFLD with the suppressed somatotroph axis To investigate how ER stress and the ISR drive the progression of liver damage in NASH and avoid the confounding factors derived from the diets that are commonly used to induce NASH, we employed a mouse NASH model deficient in the histone deacetylase SIRT7 that develops spontaneous NASH resembling human fatty liver disease when fed a chow diet due to elevated ER stress.26–28 Single-cell RNA sequencing of the livers of wild-type and SIRT7−/− mice using the 10x Genomics Chromium platform and the pathway analysis of differentially expressed genes showed that NAFLD genes were highly enriched in several cell populations (hepatocytes, macrophages, and plasma B cells) of SIRT7−/− livers (Figures 1A–1C and S1A–S1E; Data S1; Table S1), validating the NAFLD mouse model. Microarray analysis of the livers of wild-type and SIRT7−/− mice showed that a number of genes in the somatotroph growth axis and other mitogenic signals were differentially expressed between these two genotypes. The expression of several pro-growth factors, such as growth hormone receptor (GHR), fibroblast growth factor 1 (FGF1), epidermal growth factor receptor (EGFR), FGF receptor 4 (FGFR4), was suppressed in the livers of SIRT7−/− mice (Figure S2; Data S2). IGF-binding proteins that positively correlate with the level of IGF-1, such as IGF-binding protein 3 (IGFBP3) and IGF-binding protein acid labile (IGFALS), were also suppressed in the livers of SIRT7−/− mice, while IGF-binding proteins that generally inhibit the activity of IGF-1, such as IGFBP1, were upregulated. This pattern of gene expression changes in the livers of SIRT7−/− mice and wild-type littermates was confirmed by quantitative real-time PCR (Figures 1D–1I). The analysis of the single-cell RNA sequencing data for the livers of wild-type and SIRT7−/− mice revealed that the expression of the somatotroph gene IGF-1 was reduced in the hepatocytes of SIRT7−/− liver (Figures 1J and 1K). Circulating IGF-1 levels in SIRT7−/− mice were significantly lower than their wild-type counterparts (Figure 1L). Consistent with reduced levels of blood IGF-1, the IGF-1 signaling was decreased in the livers of SIRT7−/− mice, as evidenced by reduced phosphorylation of Akt (Figure 1M and 1N). The downregulation of the growth hormone/IGF-1 somatotroph axis in the livers of SIRT7−/− mice is consistent with their post-natal growth retardation.27,28 Together, these data indicate suppressed somatotroph axis in SIRT7−/− mice. This mouse model was therefore used to investigate how the somatotroph axis becomes dysregulated in NAFLD and to dissect the role of the somatotroph axis in the progression of NASH. Hepatic ER stress suppresses the somatotroph axis autonomously SIRT7 deficiency results in constitutive hepatic ER stress.27 We asked whether suppression of the somatotroph axis in SIRT7−/− mice could result from hepatic ER stress and the induction of the ISR autonomously. SIRT7 suppresses ER stress by repressing the activity of the transcription factor Myc and reducing the expression of translation machinery.27 Consistently, the analysis of the single-cell RNA sequencing data for the livers of wild-type and SIRT7−/− mice showed that ribosome genes were among the most significant changes in various cell types of the liver associated with SIRT7 expression (Figure 1B, 1C, S1D, and S1E). We knocked down the expression of Myc in the livers of SIRT7−/− mice via adeno-associated virus 8 (AAV8)-mediated gene transfer. Myc inactivation repressed the ISR in the livers of SIRT7−/− mice as evidenced by the levels of phosphorylation of eIF2α (Figures 2A and 2B). Myc inactivation also rescued the expression of genes in the somatotroph axis that were dysregulated in the livers of SIRT7−/− mice (Figures 2C–2G), increased the plasma levels of IGF-1 (Figure 2H), and enhanced the hepatic IGF-1 signaling (Figures 2I and 2J), consistent with the suppression of the somatotroph axis by the hepatic ISR autonomously. Furthermore, treatment of hepatocytes with ER stress inducers thapsigargin or tunicamycin resulted in reduced expression of genes in the somatotroph axis (Figures S3A–S3E). Together, these data suggest that hepatic ER stress and the ISR induction are sufficient to trigger the response in the somatotroph axis autonomously. Hepatic ER stress and the ISR suppress the somatotroph axis by inducing ATF3 We next investigated how the hepatic ISR leads to the suppression of the somatotroph axis. ER stress elicits signaling transduction and stress response that allow the cells to restore protein homeostasis.29 Central to the ISR is the actions of the transcription factors ATF4 and ATF6. ATF3 is also induced by ER stress by a mechanism requiring eIF2 kinases and ATF4, although its role in stress response is obscure (Figures S4A and S4B and Jiang et al.30). We used the Harmonizome web portal, which is a collection of processed datasets to mine information related to genes and proteins,31 to determine whether the ER-stress-related transcription factors could regulate genes in the somatotroph axis. Chromatin immunoprecipitation (ChIP) sequencing data analyses revealed that ATF3 bound to the promotors or enhancers of a number of IGF-related genes (Figure S4C) and that ATF4 or ATF6 did not. The binding of ATF3 to the promoters of IGF-related genes was further confirmed by ChIP with an ATF3 antibody, followed by quantitative real-time PCR in parental hepatocytes (Figures 3A–3C) and mouse livers (Figures S4D–S4F), and was abrogated in ATF3 knockdown (KD) cells generated using two independent short hairpin RNAs (Figures 3D–3F). While treatment with the ER-stress-inducer tunicamycin reduced the expression of genes in the somatotroph axis, ATF3 inactivation blunted the effect (Figure 3G), suggesting that ER stress and the ISR induction repress the somatotroph axis in hepatocytes by inducing ATF3. Suppression of the somatotroph axis leads to metabolic changes that shift energy usage from growth and proliferation to cellular protection in order to enhance stress resistance, a phenomenon termed hormesis.1,3–8 ATF3-mediated suppression of the somatotroph axis in response to ER stress and the ISR induction suggests that this branch of the ISR might prevent cell growth and proliferation while activating cellular protective programs and preventing cell death. ATF3 KD hepatocytes proliferated faster than control cells (Figure 3H) and exhibited increased apoptosis upon treatment with tunicamycin compared with control cells (Figure 3I). Together, these data suggest that ER stress and the ISR induce ATF3 to repress the somatotroph axis, resulting in reduced proliferation and improved survival of hepatocytes. ATF3 is a member of the CREB family of basic leucine zipper transcription factors and functions both as a transcriptional activator or repressor.32 ATF3 is induced in the livers of a rat model of severe steatosis and human patients with NAFLD, correlative with the ER stress status.33 ATF3 was also induced in the livers of SIRT7−/− mice (Figure 3J; Data S2). Myc inactivation in the livers of SIRT7−/− mice via AAV8-mediated gene transfer suppressed the ISR (Figures 2A and 2B) and rescued the increased ATF3 expression (Figure 3J), consistent with the induction of ATF3 expression upon the hepatic ISR. To determine whether hepatic ISR results in suppression of the somatotroph axis due to the induction of ATF3, we knocked down the expression of ATF3 in the livers of SIRT7−/− mice via AAV8-mediated gene transfer (Figure 3K). ATF3 inactivation in the livers of SIRT7−/− mice rescued the dysregulated gene expression of the somatotroph axis (Figures 3L and 3M), in keeping with the binding of ATF3 to the promoters of IGF-related genes (Figures 3A–3C and S4C–S4F). ATF3 inactivation in the livers of SIRT7−/− mice also increased the plasma levels of IGF-1 (Figure 3N) and the IGF-1 signaling (Figures 3O and 3P). Together, these data suggest that ATF3 mediates the hepatic ISR-induced repression of the somatotroph axis in vivo. Suppression of the somatotroph axis controls liver damage in NAFLD The progression from hepatosteatosis to NASH is associated with increased hepatocyte apoptosis and liver damage, which initiate inflammation to clear out dead cells and damaged tissue and to facilitate tissue repair.34,35 Increased hepatocyte proliferation is one such attempt to repair liver damage and restore loss of mass,36–38 while hepatic stellate cells are also activated and transdifferentiate into myofibroblasts, which produce an excessive amount of extracellular matrix proteins that form fibrous connective tissues to replace normal parenchymal tissues.34 Hepatic fibrosis, the wound-healing process mediated by hepatic stellate cells, is a key feature used to determine the severity of NASH. Suppression of the somatotroph axis in response to ER stress associated with NASH suggests that this branch of the ISR might activate the cellular protective program and prevent cell death, resulting in reduced inflammation but compromised parenchymal repair due to repressed hepatocyte proliferation and compensatory fibrosis. To test this possibility, we examined the physiological effects of suppressing the somatotroph axis on liver damage in NASH. The livers of SIRT7−/− mice exhibited increased inflammation (Figures 4A and 4B), apoptosis (Figures 4A and 4C), proliferation (Figures 4A and 4D), and fibrosis (Figures 4A and 4E), characteristic of the cellular and pathophysiological features of NASH.21,34–37 The analysis of the single-cell RNA sequencing data for the livers of wild-type and SIRT7−/− mice revealed increased expression of cell-cycle genes in hepatocytes of SIRT7−/− mice, consistent with increased proliferation of hepatocytes as a way to repair damage and restore loss of mass (Figure S5). ATF3 inactivation in the livers of SIRT7−/− mice via AAV8-mediated gene transfer rescued the suppression of the somatotroph axis (Figure 3K–3P). Liver terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining demonstrated increased frequency of apoptotic cells (Figures 4A and 4C), while liver Ki67 staining showed increased frequency of proliferating cells (Figures 4A and 4D) in SIRT7−/− mice with ATF3 inactivation compared with SIRT7−/− control mice. Compared with SIRT7−/− control mice, SIRT7−/− mice with ATF3 inactivation showed increased inflammation in the livers as evidenced by staining of CD68, a marker for macrophages (Figures 4A and 4B). Hepatic fibrosis as measured with Sirius red staining was reduced in SIRT7−/− mice with ATF3 inactivation (Figures 4A and 4E). Consistent with these observations, ATF3 KO mice showed increased hepatic apoptosis, liver damage, and inflammation upon liver ischemia/reperfusion injury.39 These data suggest that suppression of the somatotroph axis prevents hepatocyte apoptosis, liver damage, and inflammation while suppressing hepatocyte proliferation and parenchymal repair and promoting compensatory fibrosis (Figure 4F). Diet-induced NASH mouse models show reduced plasma IGF-1 levels.40–42 We therefore next tested whether hepatic ER stress and the ISR suppress the somatotroph axis to control liver damage in commonly used preclinical NASH models. Wild-type mice with ATF3 inactivation in the livers via AAV8-mediated gene transfer and mice treated with control virus were fed a choline-deficient high-fat diet (CD-HFD) to induce hepatic steatosis, liver damage, and fibrosis35 (Figure 5A, 5B, S6A, and S6B). ATF3 was induced in the livers of mice fed a CD-HFD compared with mice fed a chow diet (Figures 5A and 5B). Compared with chow-fed mice, CD-HFD mice had reduced expression of the somatotroph genes in the livers (Figures S6C and S6D) and reduced plasma IGF-1 levels (Figure 5C). ATF3 inactivation in the livers of CD-HFD-fed mice increased the plasma IGF-1 levels (Figure 5C). Staining of liver samples showed increased frequency of Ki67 (Figures 5D and 5E), TUNEL (Figures 5D and 5F), and CD68-positive cells (Figures 5D and 5G) and decreased staining of Sirius red (Figures 5D and 5H) in CD-HFD mice with ATF3 inactivation compared with CD-HFD control mice. ATF3 inactivation also increased the expression of inflammatory marker genes in the livers of CD-HFD mice (Figures S6E and S6F). To test directly the effects of IGF-1 on liver damage in NASH, we treated either SIRT7−/− mice or CD-HFD mice with IGF-1 for 4 weeks. Staining of liver samples showed increased frequency of CD68-positive cells and decreased staining of Sirius red in SIRT7−/− mice (Figures 6A–6C) or CD-HFD mice (Figures 6D–6F) treated with IGF-1 compared with their respective controls. These results are consistent with the effects of upregulating the somatotroph axis via ATF3 KD on liver damage in NASH (Figures 4 and 5). Together, these data are consistent with the model that ATF3 activation represses the somatotroph axis, leading to reduced hepatic apoptosis and inflammation but decreased hepatic proliferation and increased fibrosis (Figure 4F). Therefore, an effective approach to ameliorate both inflammation and fibrosis, two major indications for effective NAFLD therapeutics, would be targeting an event upstream of the suppression of the somatotroph axis, such as ER stress. NAD+ repletion reduces hepatic ER stress and ameliorates liver damage in NAFLD We took a pharmacological approach to activate SIRT7 and suppress ER stress. NAD+ boosters are emerging to be attractive means to activate sirtuins.43–45 We treated CD-HFD mice with 78c, an NAD+ booster, for 4 weeks46 (Figures S7A and S7B). 78c treatment reduced ER stress and the ISR induction in the liver (Figures 7A–7C), rescued dysregulated somatotroph gene expression (Figures 7D and 7E), increased the plasma IGF-1 levels (Figure 7F), reduced hepatic triglyceride content (Figure 7G), and reduced hepatic inflammation (Figures 7H–7K) and fibrosis (Figures 7J and 7L). DISCUSSION Our studies establish suppression of the somatotroph axis as a physiological response to hepatic ER stress that controls liver damage during the progression of NASH. Suppression of the somatotroph axis results in improved hepatocyte survival and reduced inflammation, but repressed hepatocyte proliferation and parenchymal repair, and compensatory fibrosis (Figures 4, 5, and 6). These findings provide mechanistic insights into the epidemiological observations that suppression of the somatotroph axis is associated with patients with NAFLD, in particular the severity of fibrosis.10–19 These findings also offer an explanation that NAFLD can be ameliorated by calorie restriction at the early stage, which elicits the suppressed somatotroph axis and prevents hepatocyte cell death and further liver damage.1–9 Our studies identify a regulatory branch of the hepatic ISR and uncover ATF3 as a stress-induced transcription factor that orchestrates the gene expression of the somatotroph axis. Although ATF3 is known to be induced by ER stress,30 its role in stress response is obscure. We show that ATF3 binds to the promoters or enhances of the somatotroph genes to control their expression (Figures S4C–S4F and 3A–3F). The suppressed somatotroph axis leads to reduced cell proliferation but increased stress resistance to improve cell survival (Figures 3G–3I). Thus, this regulatory branch of ISR constitutes a stress response to prevent cell death. Overnutrition and obesity are strongly associated with NAFLD, while calorie restriction is an effective intervention that prevents NAFLD in humans.20,21,47–50 Sirtuins are nutrient sensors that mediate the responses to calorie restriction and overnutrition.26,27,51–55 Indeed, evidence is emerging showing dysregulated sirtuin expression in the livers of patients with NAFLD56 and linking sirtuins to nutritional regulation of PNPLA3, which is strongly linked to NAFLD.57 SIRT7 alleviates diet-induced NAFLD.27 Therefore, sirtuins are thought to be relevant to the pathogenesis and prevention of NAFLD associated with nutrition and obesity. Furthermore, dysregulated NAD+ metabolism has been linked to human NAFLD. For example, the levels of NAMPT, a rate-limiting enzyme for NAD+ biosynthesis, is reduced in the livers and plasma of patients with NAFLD.58 NAMPT functions to prevent hepatocyte apoptosis.58 The NAD+ level is reduced in the livers of patients with NASH.59 Sirtuins are the major NAD+-consuming enzymes that mediate the signaling effects of NAD+ and are thought to be the mediators of NAD+ metabolism in NAFLD. Indeed, overexpression of SIRT7 rescues diet-induced NAFLD in mice.27 Given the association of sirtuins to known risk factors of NAFLD, such as diet, obesity, and NAD+, the prominent NAFLD phenotype in the SIRT7−/− mouse model,26–28 and the observation that SIRT7 prevents the development of NAFLD by suppressing ER stress,27 a major driver of the progression from NAFLD to NASH,22 the SIRT7−/− mouse model is relevant to human NASH, although human genome-wide association study (GWAS) data linking SIRT7 to NAFLD have not emerged yet. Indeed, our single-cell RNA sequencing analysis provided further support that the SIRT7−/− mouse model develops NAFLD (Figures 1B, 1C, S1D, and S1E). Using the SIRT7−/− mouse model, we showed that suppression of the somatotroph axis reduces hepatic inflammation but promotes fibrosis (Figures 4 and 6A–6C). This finding was further validated using the CD-HFD mouse model (Figure 5, 6D–6F, 7, S6, and S7). The consistent findings in both mouse models of NAFLD further support the relevance of the SIRT7−/− mouse model to NAFLD. NAD+ boosting has demonstrated therapeutic potential for a number of diseases.43–45 Our studies show that NAD+ boosting via 78c can ameliorate NASH, a prevalent metabolic disease that needs a cure, at least in part by modulating the hepatic ISR and the somatotroph axis in mouse models (Figure 7), demonstrating the therapeutic potential of modulating this pathway. Suppression of the somatotroph axis in response to ER stress uncouples inflammation and fibrosis (Figures 4, 5, and 6), providing a basis for combination therapies or targeting an initiating event, such as ER stress, for this metabolic disease (Figure 7). Limitations of the study The role of the hepatic ISR and the somatotroph axis in controlling liver damage during NAFLD has been tested using two NAFLD mouse models, SIRT7−/− mice and CD-HFD mice. How this pathway operates in other NAFLD models has not been tested. The effects of IGF1 treatment were examined in mice treated for 4 weeks, but the effects after longer or shorter treatments have not been tested. STAR★METHODS RESOURCE AVAILABILITY Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Danica Chen (danicac@berkeley.edu). Materials availability Unique reagents generated in this study are available from the lead contact, Danica Chen (danicac@berkeley.edu). Data and code availability The sequencing data reported in this paper has been deposited in NCBI′ s Gene Expression Omnibus and are accessible through GEO: GSE216996. This paper does not report custom code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request (danicac@berkeley.edu). EXPERIMENTAL MODEL AND SUBJECT DETAILS Mice SIRT7−/− mice have been described previously.27,53 For a diet-induced NAFLD mouse model, 8-week-old C57BL/6 male mice were fed with choline-deficient high-fat diet (Research Diet, A06071302) consisting of 60 kcal% fat with 0.1% methionine and no added choline for 3 weeks before either 78c treatment or IGF-1 treatment. 78c was administered to mice by intraperitoneal injection (10 mg/kg/dose) twice daily for 4 weeks. Control mice received vehicle (5% DMSO, 15% PEG400, 80% of 15% hydroxypropyl-g-cyclodextrin (in citrate buffer pH 6.0)) injections. IGF-1 (Pepro Tech) dissolved in 0.1% BSA/PBS was administered to mice by subcutaneous injection (20 μg/kg/day) for 4 weeks. For IGF-1-treated SIRT7−/− mice, 7- to 11-month-old mice were used for the experiment. All mice were housed on a 12:12 h light:dark cycle at 25°C and were given free access to food and water. All animal procedures were in accordance with the animal care committee at the University of California, Berkeley. Cell culture Hepa 1–6 cells were acquired from cell culture facility at the University of California, Berkeley. Cells were cultured in advanced Dulbecco’s modified Eagle’s medium (Gibco) supplemented with 10% FBS (Gibco). For ER stress induction, cells were treated with tunicamycin (Sigma, 2 μg/mL) or thapsigargin (Sigma, 0.1μM) for 24 h before biochemical analysis. For ATF3 knockdown, Hepa 1–6 cells were transfected with AllStars Negative Control siRNA (Qiagen, 1027281) or ATF3 siRNA (Qiagen, GS11910) using RNAiMAX (Invitrogen, 13778100) according to manufacture’s instruction. To generate Hepa 1–6 cells with stable ATF3 knockdown, cells were infected with lentivirus. For lentiviral packaging, 293T cells were co-transfected with packaging vectors (pCMV-dR8.2 dvpr and pCMV-VSV-G) and the pLKO.1-ATF3 shRNA (Sigma, TRCN0000082129, TRCN0000082132) or control construct. Viral supernatant was harvested after 48 h and 72 h after transfection, as described previously.64 For transduction, cells were incubated with virus-containing supernatant in the presence of 10 μg/mL polybrene. After 48 h, infected cells were selected with puromycin (4 μg/mL). For cell proliferation, 0.3 × 106 cells were seeded in a 6-well plate. Two days later, 20% cells were passaged to a new well and were counted 24 h later. Primary hepatocytes were suspended in plating medium (DMEM low glucose, 5% FBS and 1%Pen/Strep) and plated on collagen-coated cell culture plates (Sigma-Aldrich C3867-1VL). After 3 h, it was changed to maintenance media (Williams E media, 1% Glutamine and 1% Pen/Strep). The next day cells were treated with tunicamycin for 24 h (Sigma, 4 μg/mL) before analysis. METHOD DETAILS Apoptosis assay Apoptotic cells were assayed using propidium iodide (BioLegend) and FITC Annexin V staining (BioLegend) according to the manufacturer’s instruction (BioLegend). All data were collected on an LSR Fortessa (BD Bioscience), and data analysis was performed with FlowJo (TreeStar). Chromatin immunoprecipitation Cells were prepared for ChIP as previously described,65 with the exception that DNA was washed and eluted using a QIAprep Spin Miniprep kit (Qiagen) rather than by phenol-chloroform extraction. For ChIP with mouse livers, 150 mg mouse liver were minced and dounce homogenized with 10 strokes in hypotonic lysis buffer (10 mM HEPES, pH7.5, 10 mM KCl, 1.5 mM MgCl2, 250 mM Sucrose, 0.5% NP40, and protease inhibitor cocktail). Lysates were filtered through a 100um cell strainer and spin at 1500 g for 5 min. Lipid and cytoplasmic fractions were removed and the nuclear pellet was resuspended in lysis buffer, cross-linked with fresh formaldehyde (1%) for 5 min at room temperature, quenched with glycine (125 mM), and washed twice with PBS. Affymetrix microarray Total RNA was isolated from the livers of wild type and SIRT7−/− mice using an RNA isolation kit (Qiagen). Microarray hybridizations were performed at the University of California, Berkeley Functional Genomics Laboratory using Affymetrix GeneChip mouse 430As according to the instructions of the manufacturer (Affymetrix). RMA normalization was applied and the limma package was used to identify the differentially expressed genes. Differentially expressed genes were selected using the Benjamini-Hochberg method to control the FDR at 15%. Single-cell RNA-sequencing of livers using 10x Genomics Chromium 5- to 6-month-old SIRT7−/− mice were used for single-cell RNA-sequencing of livers. Hepatocytes and non-parenchymal cells (NPCs) were isolated by a two-step collagenase perfusion method.66 Briefly, after the inferior vena cava was cannulated with a 25 gauge catheter and the portal vein was cut, the liver was perfused at 10 mL/min with Liver Perfusion Medium (Gibco 17701-038) at 37°C for 5 min, followed by perfusion with collagenase type IV (Worthington LS004188) in HBSS (GIBCO) at 37°C for 5 min. The liver was dissected out and transferred to Petri dish with William E medium (Gibco 12551-032) containing 200 mM L-glutamine, 1% pen/strep and 1% non-essential amino acid. Then gently shake out the cells from liver capsule. The released liver cells were passed through a 100 μm filter. Hepatocytes were separated from NPCs by low-speed centrifugation (50 × g, 4 min, 3x, brake = 2) and further purified by Percoll gradient centrifugation (50% v/v) to remove dead cells.67 NPCs were pelleted from supernatant by centrifugation (300 xg, 10 min) then purified by Percoll gradient centrifugation (33% v/v) to remove dead cells.68 Cell viability was confirmed by trypan blue exclusion. 3,000 hepatocytes and 3,000 NPCs were mixed and used directly for scRNA-seq analysis using 10X Genomics Chromium Single-Cell 3′ according to the manufacturer’s instructions. 10x genomics single-cell RNA-sequencing data pre-processing, UMAP analysis, and identification of cell clusters RNA reads from sequencing were demultiplexed and aligned to mouse transcriptome (mm10) using the Cell Ranger software (10x Genomics, v.6.0.0). The Scanpy Python package (v.1.6.0) was used for the pre-processing of the single-cell RNA seq data.60 Cells with less than 500 unique genes or more than 5% mitochondrial genes were removed. Genes detected in less than 3 cells were excluded. We included 11,610 cells with 3,270 cells from wild type and 8,340 cells from SIRT7−/−, and 16,623 genes for further analysis. The data was normalized such that every cell has 10,000 counts and then log transformed with an offset of 1. The batch correction was done by the bbknn batch-alignment algorithm.61 We computed the highly variable genes with the top 1,000 genes and the flavor set to ‘cell_ranger’. The highly variable genes were used for principal components analysis. The data was visualized by UMAP (Uniform Manifold Approximation and Projection) projection using Scanpy. Unsupervised clustering was done by the Leiden algorithm62 with a resolution of 0.35. Marker genes for each cluster were calculated by Wilcoxon rank-sum test. The cell identity of each cluster was determined by comparing the marker genes of each cluster with the marker genes identified in the literature. Differential gene expression analysis, bar plots, violin plots, and dot plots for gene expression in single cells, and pathway enrichment analysis Adaptive thresholding of the single-cell gene expression data was performed with the MAST R package (v1.12.0), and differential gene expression analysis of wild type and SIRT7−/− cells from each cluster using a hurdle model with the wild type cells as the ref. 63. To visualize the expression of genes, log-normalized expressions of genes were extracted from the data after adaptive thresholding and plotted for every cell with a violin plot and an overlying strip plot by the Seaborn Python package (v.0.9.0). The bar plots were generated by Seaborn. The UMAP plots, dot plots, and track plots were generated by Scanpy. The GSEAPY Python package (v.0.10.3) was used for pathway enrichment analysis. Quantitative real-time PCR RNA was isolated from cells or tissues using Trizol reagent (Invitrogen) following the manufacturer’s instructions. cDNA was generated using the qScript cDNA SuperMix (Quanta Biosciences). Gene expression was determined by quantitative real-time PCR using Eva qPCR SuperMix kit (BioChain Institute) on an ABI StepOnePlus system. All data were normalized to GAPDH expression. AAV8-mediated gene transfer For AAV8-mediated gene transfer to the mouse liver, Myc knockdown target sequence was cloned into dsAAV-RSVeGFP-U6 vector. AAV8 for knocking down Myc was produced by Vigene biosciences. AAV8 for knocking down ATF3 was acquired from Vector biolabs. Myc knockdown target sequence: 5′-CCCAAGGTAGTGATCCTCAAA-3′. ATF3 knockdown target sequence: 5′-TGCTGCCA AGTGTCGAAACAA-3′. Each mouse was injected with 3 × 1011 genome copies of virus via tail vein. Mice were characterized four weeks after viral infection (5- to 6-month-old wild-type and SIRT7−/− mice) or eight weeks after viral infection (8-week-old C57BL/6 mice on CD-HFD). Plasma IGF-1 levels To detect IGF-1 in the plasma, the plasma was pretreated with acid-ethanol extraction solution to release IGF-1 from binding proteins. Briefly, 120 μL of acid-ethanol extraction buffer (hydrochloric acid:water:ethanol = 1:4:35, v/v/v) was added to 30 μL of plasma. The extract was incubated for 30 min at room temperature with shaking. The extract was centrifuged at 10,000 rpm for 5 min and 100 μL of supernatant was collected. 200 uL of Tris buffer (pH = 7.6) was added to the supernatant. IGF-1 was detected using IGF-1 Mouse ELISA Kit (Invitrogen). Immunohistochemistry Tissue sections (5 μm) were mounted on glass slides. Slides were fixed with 10% formalin. Tissue processing and immunohistochemistry was performed on sections. Primary antibodies were: mouse anti-CD68 (Biolegend, 137001); Ki67 (Biolegend, 652409). After overnight incubation, primary antibody staining was revealed using fluorescence conjugated secondary antibodies. Nuclei were counter stained using DAPI. Images were taken with Zeiss AxioImager microscope. The positive cells were manually counted or counted using ImageJ. Fibrosis staining Liver sections were fixed with 10% formalin and then stained with Sirius Red (Sigma)/Fast Green (Sigma). Images were taken with Zeiss AxioImager microscope. The positive area was quantified using ImageJ. TUNEL staining Apoptosis was detected with Apo-Brdu in situ DNA fragmentation assay kit according to the manufacturer’s instruction (Biovision). Nuclei were counter stained using DAPI. TUNEL-positive cells were imaged using Zeiss AxioImager microscope. Western blot Tissues or cells were homogenized in a lysis buffer that contained protease inhibitor, and total protein was extracted with gentle rotation for 30 min at 4°C. The extract was centrifuged at 15,000 g for 15 min at 4°C. Supernatants were collected and total protein was quantified with BCA assay (Thermo Scientific, 23225). Proteins were resolved by SDS-PAGE and transferred to nitrocellulose membranes (Bio-Rad), which was incubated with specific primary antibodies and horseradish peroxidase-conjugated secondary antibodies, and enhanced chemiluminescence substrate (PerkinElmer, NEL103001EA), and visualized using ImageQuant™ LAS 4000 (GE Healthcare). Triglyceride quantification Triglycerides were extracted from liver tissues as described.69 Briefly, liver tissues were homogenized in the methanol/chloroform buffer (1:2, v/v) and lipids were extracted with gentle rotation for 2 h at room temperature. The homogenate was centrifuged at 15,000 g for 5 min. Supernatants were concentrated via nitrogen gas and reconstituted with the reconstitution buffer (1% Triton X-100 in 100% ethanol). Extracted triglyceride was quantified in accordance with the manufacturer’s instruction (Wako Diagnostics). QUANTIFICATION AND STATISTICAL ANALYSIS Mice were randomized to groups and analysis of mice and tissue samples was performed by investigators blinded to the treatment or the genetic background of the animals during experiments. Statistical analysis was performed with Student’s t test (Excel) unless specified. Wilcoxon rank-sum test for single-cell RNA sequencing analysis was performed using the SciPy Python package (v.1.4.1). Data are presented as means and error bars represent standard errors. In all corresponding figures, * represents p < 0.05. ** represents p < 0.01. *** represents p < 0.001. ns represents p > 0.05. Replicate information is indicated in the figures. Supplementary Material 1 2 3 4 5 ACKNOWLEDGMENTS We thank Y. Choi and the Berkeley Functional Genomics Lab for genome-wide gene expression. We thank the CNR Biological Imaging Facility. This work was supported by NIH R01DK 117481 (D.C.), R01AG063404 (D.C.), and R01AG 063389 (D.C.); the National Institute of Food and Agriculture (D.C.); NIH R35GM142916 (P.H.S.) and R01AG047182 (C.M.H.); the ITO Scholarship (R.O.); the Honjo International Scholarship (R.O.); the Dr. and Mrs. James C.Y. Soong Fellowship (W.-C.M.); and the Taiwan Government for Study Abroad Scholarship (W.-C.M.) and QB3 Frontiers in Medical Research Fellowship (W.-C.M.). Figure 1. A mouse model of NAFLD with the suppressed somatotroph axis (A) Single-cell RNA sequencing of the livers of WT and SIRT7−/− mice using the 10x Genomics Chromium platform. Uniform manifold approximation and projection (UMAP) clustering of single-cell transcriptomes (3,270 cells from WT and 8,340 cells from SIRT7−/− mice) colored by cell type. n = 3 mice. (B and C) Pathway analysis for the biological function of differentially expressed genes in hepatocyte 1 (pericentral) and hepatocyte 2 (periportal) of the livers of WT and SIRT7−/− mice. n = 3 mice. (D–I) Quantitative real-time PCR analyses for the mRNA levels of the indicated genes in the livers of SIRT7−/− mice and wild-type controls. GAPDH was used as an internal control. n = 9–13 mice. (J and K) Violin plots comparing log-normalized expression values of IGF-1 in hepatocyte 1 (pericentral) and hepatocyte 2 (periportal) in the livers of WT and SIRT7−/− mice. Each dot represents the gene expression levels in one cell. Wilcoxon rank-sum test. n = 3 mice. (L) ELISA quantification of plasma levels of IGF-1 in SIRT7−/− mice and wild-type controls. n = 8 mice. (M and N) Western analyses (M) and quantification (N) of phosphorylated Akt in the livers of SIRT7−/− mice and wild-type controls. n = 3 mice. Error bars represent standard errors. *p < 0.05; **p < 0.01; ***p < 0.001. See also Figure S1 and S2. Figure 2. Hepatic ER stress suppresses the somatotroph axis autonomously Comparison of wild-type and SIRT7−/− mice with or without Myc knockdown mediated by AAV8-mediated gene delivery. Mice were analyzed 4 weeks after viral infection. (A and B) Western analyses (A) and quantification (B) for phosphorylated eIF2α in the livers. n = 3 mice. (C–G) Quantitative real-time PCR analyses for the mRNA levels of the indicated genes in the livers. GAPDH was used as an internal control. n = 4–5 mice. (H) ELISA analyses of plasma levels of IGF-1. n = 4 mice. (I and J) Western analyses (I) and quantification (J) for phosphorylated Akt in the livers. n = 3 mice. Error bars represent standard errors. *p < 0.05; **p < 0.01; ***p < 0.001. See also Figure S3. Figure 3. Hepatic ER stress and the ISR suppress the somatotroph axis by inducing ATF3 (A–C) ChIP with ATF3 antibody followed by quantitative real-time PCR showing ATF3 occupancy at the gene promoters of IGFBP3 and IGF1R in Hepa 1–6 cells. Tubulin was used as a negative control. n = 2. (D) Western blots showing ATF3 expression in stable ATF3 knockdown Hepa 1–6 cells using shRNA. (E and F) ChIP with ATF3 antibody followed by quantitative real-time PCR showing reduced ATF3 occupancy at the gene promoters of IGFBP3 and IGF1R in ATF3 knockdown Hepa 1–6 cells. n = 2. (G) Western analyses of GHR and ATF3 in control and ATF3 knockdown Hepa 1–6 cells with or without tunicamycin induction. (H) Proliferation of stable ATF3 knockdown Hepa 1–6 cells and control cells. n = 3. (I) Annexin V staining of ATF3 knockdown and control Hepa 1–6 cells with or without tunicamycin induction was analyzed with flow cytometry. n = 3. (J) Quantitative real-time PCR analyses of mRNA levels of ATF3 in the livers of SIRT7−/− mice and wild-type mice with or without Myc knockdown mediated by AAV8-mediated gene delivery. Mice were analyzed 4 weeks after viral infection. n = 4 mice. (K–P) Comparison of SIRT7−/− mice and wild-type mice with or without ATF3 knockdown mediated by AAV8-mediated gene delivery. Mice were analyzed 4 weeks after viral infection. (K–M) Quantitative real-time PCR analyses of mRNA levels of indicated genes in the livers. GAPDH was used as an internal control. n = 4–5 mice. (N) Elisa analyses of plasma levels of IGF-1. n=4–5 mice. (O and P) Western analyses (O) and quantification (P) for phosphorylated Akt in the livers. n = 3 mice. Error bars represent standard errors. *p < 0.05; **p < 0.01; ***p < 0.001; ns p > 0.05. See also Figure S4. Figure 4. Suppression of the somatotroph axis controls liver damage in NAFLD (A–E) Liver sections stained for Ki67, TUNEL, CD68, and Sirius red (A) and their quantifications (B–E) for SIRT7−/− mice and wild-type mice with or without ATF3 knockdown mediated by AAV8-mediated gene delivery. Mice were analyzed 4 weeks after viral infection. n = 4–5 mice. Scale bar: 100 μm. (F) A proposed model. Hepatic ER stress and the ISR induce ATF3 expression and the suppression of the somatotroph axis, leading to reduced hepatocyte death, liver damage, and inflammation, while reducing hepatocyte proliferation and parenchymal repair, resulting in compensatory fibrosis. Error bars represent standard errors. *p < 0.05; **p < 0.01; ***p < 0.001; ns p > 0.05. See also Figure S5. Figure 5. Suppression of the somatotroph axis controls liver damage in mice fed a CD-HFD Comparison of wild-type mice with or without ATF3 knockdown in the livers fed a chow diet or a CD-HFD for 8 weeks. (A and B) Western analyses (A) and quantification (B) of ATF3 in the livers. n = 3 mice. (C) ELISA analyses of plasma levels of IGF-1. n = 8 mice. (D–H) Liver sections stained for Ki67, TUNEL, CD68, and Sirius red (D) and their quantifications (E–H). n = 5–6 mice. Scale bars: 200 (Ki67), 100 (TUNEL, Sirius red), and 50 μm (CD68). Error bars represent standard errors. *p < 0.05; **p < 0.01; ***p < 0.001. See also Figure S6. Figure 6. IGF-1 controls liver damage in NAFLD (A–C) Comparison of wild-type and SIRT7−/− mice treated with or without IGF-1 for 4 weeks. Data shown are liver sections stained for CD68 and Sirius red (A) and their quantifications (B and C). n = 7 mice. Scale bar: 100 μm. (D–F) Comparison of wild-type mice fed a CD-HFD for 3 weeks followed by treatment with or without IGF-1 for 4 weeks. Data shown are liver sections stained for CD68 and Sirius red (D) and their quantifications (E and F). n = 6–8 mice (E) and 7 mice (F). Scale bar: 100 μm. Error bars represent standard errors. *p < 0.05; **p < 0.01; ***p < 0.001. Figure 7. NAD+ repletion ameliorates hepatic ER stress, dysregulated somatotroph axis, and liver damage in NAFLD Comparison of mice fed a chow diet or a CD-HFD for 3 weeks followed by treatment with or without 78c for 4 weeks. (A–C) Western analyses (A) and quantification (B and C) for phosphorylated eIF2α and ATF3 in the livers. n = 3–4 mice. (D and E) Quantitative real-time PCR analyses for the mRNA levels of indicated genes in the livers. GAPDH was used as an internal control. n = 5–8 mice. (F) ELISA analyses of plasma levels of IGF-1. n = 5–8 mice. (G) Liver triglyceride quantification. n = 5–8 mice. (H and I) Quantitative real-time PCR analyses for the mRNA levels of the indicated genes in the livers. GAPDH was used as an internal control. n = 5–8 mice. (J–L) Liver sections stained for CD68 and Sirius red (J) and their quantifications (K and L). n = 5–8 mice. Scale bars: 100 (CD68) and 200 μm (Sirius red). Error bars represent standard errors. *p < 0.05; **p < 0.01; ***p < 0.001. See also Figure S7. KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies p-eIF2α (Ser52) polyclonal antibody Invitrogen Cat# 44728G; RRID:AB_1500038 eIF2α antibody CST Cat# 9722; RRID: AB_2230924 Phospho-Akt (Ser473) antibody CST Cat# 9271; RRID:AB_329825 Akt antibody CST Cat# 9272; RRID:AB_329827 Actin antibody Sigma Cat# A2066; RRID:AB_476693 GAPDH antibody CST Cat# 5174; RRID: AB_10622025 Mouse growth hormone R/GHR antibody R&D Cat# AF1360; RRID:AB_2111403 Mouse FGF acidic/FGF1 antibody R&D Cat# AF4686; RRID: AB_2924726 ATF-3 (D2Y5W) Rabbit antibody CST Cat# 33593S; RRID: AB_2799039 Normal Rabbit IgG CST Cat# 2729S; RRID: AB_1031062 Purified anti-mouse CD68 antibody BioLegend Cat#137001; RRID: AB_2044003 Goat anti-rat IgG (H + L) cross-absorbed secondary antibody, DyLight 488 ThermoFisher Scientific Cat# SA5-10018; RRID: AB_2556598 FITC anti-mouse Ki-67 antibody BioLegend Cat# 652409; RRID: AB_2562140 Chemicals, peptides, and recombinant proteins 78c (CD38 inhibitor) MedChemExpress Cat# HY-123999; CAS#1700637-55-3 Dimethyl Sulfoxide (DMSO) Sigma Cat# D8418 Polyethylene glycol 400 (PEG400) Sigma Cat# PX1286B Hydroxypropyl-g-cyclodextrin Santa Cruz biotechnology Cat# sc-238090A Recombinant human IGF1 PeproTech Cat# 100-11 BSA Sigma Cat# A7906 Dulbecco’s Modification of Eagle’s medium Gibco Cat# 11995065 Dulbecco’s Modification of Eagle’s medium (low glucose) Gibco Cat# 11885-084 Williams E media Gibco Cat# 12551-032 Liver perfusion medium Gibco Cat# 17701-038 Collagenase type IV Worthington Cat# LS004188 L-Glutamine Gibco Cat# 25030081 Non-essential amino acid (100X) Gibco Cat# 11140-050 Percoll™ PLSU Cytiva Cat# 17544702 Fetal Bovine Serum Invitrogen Cat#10437-028 Tunicamycin Sigma Cat# T7765 Thapsigargin Sigma Cat# T9033 RNAiMAX Invitrogen Cat# 13778100 Sirius red (direct red 80) Sigma Cat# 365548 Fast green Fisher Chemical Cat# F99-10 qScript™ cDNA SuperMix Quanta biosciences Cat# 95048 qPCR SuperMix kit BioChain Institute Cat# K5052400 Penicillin Streptomycin solution (100x) Invitrogen Cat# 15140122 Collagen, type I solution from rat tail Sigma Cat# C3867-1VL Trypsin-EDTA (0.25%) Gibco Cat# 25200056 TRIzol reagent Invitrogen Cat# 15596026 Lipofectamine 2000 Invitrogen Cat# 11668019 HEPES Gibco Cat# 15630080 HBSS, calcium, magnesium, no phenol red Gibco Cat# 14025092 HBSS, no calcium, no magnesium, no phenol red Gibco Cat# 14175095 Western (blotting) Lightning Plus-ECL substrate Perkin Elmer Cat# NEL103E001EA DAPI (4′,6-diamidino-2-phenylindole, dihydrochloride) Thermo Fisher Scientific Cat#62247 Propidium iodide solution Biolegend Cat#421301 FITC Annexin V BioLegend Cat# 640906 Formaldehyde Thermo Fisher Scientific Cat# F79-500 Critical commercial assays QIAprep spin Miniprep kit Qiagen Cat# 27106X4 10× Genomics single Cell 3′ reagent kits v3 10× Genomics Cat# PN-1000075 IGF-1 mouse ELISA kit Invitrogen Cat# EMIGF1 Apo-Brdu in situ DNA fragmentation assay kit Biovision Cat# K401 Pierce™ BCA protein assay kit Thermo Scientific Cat# 23225 L-type Triglyceride M enzyme color A Fujifilm Wako Diagnostics Cat# 996-02895 L-type Triglyceride M enzyme color B Fujifilm Wako Diagnostics Cat# 992-02995 Deposited data SIRT7 liver GEO: GSE216996 Experimental models: Cell lines Hepa 1–6 UC Berkeley Cell culture facility N/A HEK293T ATCC CRL-3216 Experimental models: Organisms/strains Mouse: SIRT7 KO Shin et al.27 N/A Mouse: C57BL/6J National Institute on Aging N/A Oligonucleotides qPCR primer sequences IDT (integrated DNA technologies) Table S2 IGFBP3 ChIP Forward primer: GTTCTCGCTGGGAAATGCCT IDT (integrated DNA technologies) N/A IGFBP3 ChIP Reverse primer: TCAGCGCCTGTGTACTTTGT IDT (integrated DNA technologies) N/A IGF-1R ChIP Forward primer: GGGAATTTCGTCCCAAATAAAAGGA IDT (integrated DNA technologies) N/A IGF-1R ChIP Reverse primer: GAGAGAAACACGAGCCCCC IDT (integrated DNA technologies) N/A Tubulin ChIP Forward primer: AGACGGAAGAGAACACTGCG IDT (integrated DNA technologies) N/A Tubulin ChIP Reverse primer: CTTCATCGGGCTTCAGTCGT IDT (integrated DNA technologies) N/A ATF3 siRNA TGCTGCCAAGTGTCGAAACAA Qiagen Cat# GS11910 Control siRNA Qiagen Cat# 1027281 Myc siRNA CCCAAGGTAGTGATCCTCAAA Shin et al.27 N/A Recombinant DNA pCMV-dR8.2 dvpr Addgene Plasmid: #8455 pCMV-VSV-G Addgene Plasmid: #8454 pLKO.1-ATF3 Sigma TRCN0000082129 TRCN0000082132 dsAAV-RSVeGFP-U6 Shin et al.27 N/A dsAAV-RSVShMyc Shin et al.27 N/A Ad-m-ATF3-shRNA Vector biolabs Cat# shADV-253206 Software and algorithms Cell ranger (v.6.0.0) 10X Genomics N/A Scanpy Python package (v.1.6.0) Luo et al.60 https://github.com/scverse/scanpy Bbknn batch-alignment algorithm Polański et al.61 https://github.com/Teichlab/bbknn Leiden algorithm Traag et al.62 https://github.com/vtraag/leidenalg MAST R package (v.1.12.0) Finak et al.63 https://github.com/RGLab/MAST Seaborn Python package (v.0.9.0) https://seaborn.pydata.org/citing.html GSEAPY Python package (v.0.10.3) https://github.com/zqfang/GSEApy/releases ImageJ https://imagej.nih.gov/ij/ iVision (v.4.5.6 r4) BioVision Technologies https://www.biovis.com GraphPad Prism GraphPad https://www.graphpad.com/ Other Choline-deficient high fat diet Research Diet Cat# A06071302 Highlights Hepatic ER stress suppresses the somatotroph axis by inducing ATF3 Suppression of the somatotroph axis controls liver damage in NAFLD NAD+ repletion ameliorates dysregulated somatotroph axis and liver damage in NAFLD DECLARATION OF INTERESTS The authors declare no competing interests. SUPPLEMENTAL INFORMATION Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2022.111803. REFERENCES 1. Bartke A (2005). Minireview: role of the growth hormone/insulin-like growth factor system in mammalian aging. Endocrinology 146 , 3718–3723. 10.1210/en.2005-0411.15919742 2. Fontana L , Partridge L , and Longo VD (2010). Extending healthy life span–from yeast to humans. Science 328 , 321–326. 10.1126/science.1172539.20395504 3. Garinis GA , Uittenboogaard LM , Stachelscheid H , Fousteri M , van Ijcken W , Breit TM , van Steeg H , Mullenders LHF , van der Horst GTJ , Brüning JC , (2009). Persistent transcription-blocking DNA lesions trigger somatic growth attenuation associated with longevity. Nat. Cell Biol. 11 , 604–615. 10.1038/ncb1866.19363488 4. Kanfi Y , Naiman S , Amir G , Peshti V , Zinman G , Nahum L , Bar-Joseph Z , and Cohen HY (2012). The sirtuin SIRT6 regulates lifespan in male mice. Nature 483 , 218–221. 10.1038/nature10815.22367546 5. Kenyon C (2005). The plasticity of aging: insights from long-lived mutants. Cell 120 , 449–460. 10.1016/j.cell.2005.02.002.15734678 6. Murphy CT , McCarroll SA , Bargmann CI , Fraser A , Kamath RS , Ahringer J , Li H , and Kenyon C (2003). Genes that act downstream of DAF-16 to influence the lifespan of Caenorhabditis elegans. Nature 424 , 277–283. 10.1038/nature01789.12845331 7. Niedernhofer LJ , Garinis GA , Raams A , Lalai AS , Robinson AR , Appeldoorn E , Odijk H , Oostendorp R , Ahmad A , van Leeuwen W , (2006). A new progeroid syndrome reveals that genotoxic stress suppresses the somatotroph axis. Nature 444 , 1038–1043. 10.1038/nature05456.17183314 8. Vermeij WP , Dollé MET , Reiling E , Jaarsma D , Payan-Gomez C , Bombardieri CR , Wu H , Roks AJM , Botter SM , van der Eerden BC , (2016). Restricted diet delays accelerated ageing and genomic stress in DNA-repair-deficient mice. Nature 537 , 427–431. 10.1038/nature19329.27556946 9. Weindruch R , and Walford RL (1988). The Retardation of Aging and Disease by Dietary Restriction (Charles C. Thomas). 10. Chishima S , Kogiso T , Matsushita N , Hashimoto E , and Tokushige K (2017). The relationship between the growth hormone/insulin-like growth factor system and the histological features of nonalcoholic fatty liver disease. Intern. Med. 56 , 473–480. 10.2169/internalmedicine.56.7626.28250290 11. Colak Y , Senates E , Ozturk O , Yilmaz Y , Zemheri E , Yilmaz Enc F , Ulasoglu C , Aksaray S , Bozbeyoglu SG , Kiziltas S , (2012). Serum concentrations of human insulin-like growth factor-1 and levels of insulin-like growth factor-binding protein-5 in patients with nonalcoholic fatty liver disease: association with liver histology. Eur. J. Gastroenterol. Hepatol. 24 , 255–261. 10.1097/MEG.0b013e32834e8041.22157234 12. Dal K , Bulur O , Ata N , Yeniova AO , Baser S , Karakaya S , Unsal O , Dagdeviren M , Karadag I , Beyan E , and Ertugrul DT (2017). The role of insulin - like growth factor - 1 on steatohepatitis. Acta Gastroenterol. Belg. 80 , 21–24.29364093 13. Dichtel LE , Corey KE , Misdraji J , Bredella MA , Schorr M , Osganian SA , Young BJ , Sung JC , and Miller KK (2017). The association between IGF-1 levels and the histologic severity of nonalcoholic fatty liver disease. Clin. Transl. Gastroenterol. 8 , e217. 10.1038/ctg.2016.72.28125073 14. Fusco A , Miele L , D’Uonnolo A , Forgione A , Riccardi L , Cefalo C , Barini A , Bianchi A , Giampietro A , Cimino V , (2012). Nonalcoholic fatty liver disease is associated with increased GHBP and reduced GH/IGF-I levels. Clin. Endocrinol. 77 , 531–536. 10.1111/j.1365-2265.2011.04291.x. 15. García-Galiano D , Sánchez-Garrido MA , Espejo I , Montero JL , Costán G , Marchal T , Membrives A , Gallardo-Valverde JM , Muñoz-Castañeda JR , Arévalo E , (2007). IL-6 and IGF-1 are independent prognostic factors of liver steatosis and non-alcoholic steatohepatitis in morbidly obese patients. Obes. Surg. 17 , 493–503. 10.1007/s11695-007-9087-1.17608262 16. Hribal ML , Procopio T , Petta S , Sciacqua A , Grimaudo S , Pipitone RM , Perticone F , and Sesti G (2013). Insulin-like growth factor-I, inflammatory proteins, and fibrosis in subjects with nonalcoholic fatty liver disease. J. Clin. Endocrinol. Metab. 98 , E304–E308. 10.1210/jc.2012-3290.23316084 17. Liang S , Cheng X , Hu Y , Song R , and Li G (2017). Insulin-like growth factor 1 and metabolic parameters are associated with nonalcoholic fatty liver disease in obese children and adolescents. Acta Paediatr. 106 , 298–303. 10.1111/apa.13685.27889912 18. Liang S , Yu Z , Song X , Wang Y , Li M , and Xue J (2018). Reduced growth hormone secretion is associated with nonalcoholic fatty liver disease in obese children. Horm. Metab. Res. 50 , 250–256. 10.1055/s-0043-124970.29341046 19. Mallea-Gil MS , Ballarino MC , Spiraquis A , Iriarte M , Kura M , Gimenez S , Oneto A , Guitelman M , Machado R , and Miguel CM (2012). IGF-1 levels in different stages of liver steatosis and its association with metabolic syndrome. Acta Gastroenterol. Latinoam. 42 , 20–26.22616493 20. Cohen JC , Horton JD , and Hobbs HH (2011). Human fatty liver disease: old questions and new insights. Science 332 , 1519–1523. 10.1126/science.1204265.21700865 21. Kim JY , Garcia-Carbonell R , Yamachika S , Zhao P , Dhar D , Loomba R , Kaufman RJ , Saltiel AR , and Karin M (2018). ER stress drives lipogenesis and steatohepatitis via caspase-2 activation of S1P. Cell 175 , 133–145.e15. 10.1016/j.cell.2018.08.020.30220454 22. Wang M , and Kaufman RJ (2016). Protein misfolding in the endoplasmic reticulum as a conduit to human disease. Nature 529 , 326–335. 10.1038/nature17041.26791723 23. Costa-Mattioli M , and Walter P (2020). The integrated stress response: from mechanism to disease. Science 368 , eaat5314. 10.1126/science.aat5314.32327570 24. Jones JI , and Clemmons DR (1995). Insulin-like growth factors and their binding proteins: biological actions. Endocr. Rev. 16 , 3–34. 10.1210/edrv-16-1-3.7758431 25. Xia W , Wang Y , Zhang Y , Ge X , Lv P , Cheng J , and Wei J (2020). Endoplasmic reticulum stress induces growth retardation by inhibiting growth hormone IGF-I axis. Growth Horm. IGF Res. 55 , 101341. 10.1016/j.ghir.2020.101341.32890915 26. Ryu D , Jo YS , Lo Sasso G , Stein S , Zhang H , Perino A , Lee JU , Zeviani M , Romand R , Hottiger MO , (2014). A SIRT7-dependent acetylation switch of GABPbeta1 controls mitochondrial function. Cell Metab. 20 , 856–869. 10.1016/j.cmet.2014.08.001.25200183 27. Shin J , He M , Liu Y , Paredes S , Villanova L , Brown K , Qiu X , Nabavi N , Mohrin M , Wojnoonski K , (2013). SIRT7 represses Myc activity to suppress ER stress and prevent fatty liver disease. Cell Rep. 5 , 654–665. 10.1016/j.celrep.2013.10.007.24210820 28. Vazquez BN , Thackray JK , Simonet NG , Kane-Goldsmith N , Martinez-Redondo P , Nguyen T , Bunting S , Vaquero A , Tischfield JA , and Serrano L (2016). SIRT7 promotes genome integrity and modulates non-homologous end joining DNA repair. EMBO J. 35 , 1488–1503. 10.15252/embj.201593499.27225932 29. Walter P , and Ron D (2011). The unfolded protein response: from stress pathway to homeostatic regulation. Science 334 , 1081–1086. 10.1126/science.1209038.22116877 30. Jiang HY , Wek SA , McGrath BC , Lu D , Hai T , Harding HP , Wang X , Ron D , Cavener DR , and Wek RC (2004). Activating transcription factor 3 is integral to the eukaryotic initiation factor 2 kinase stress response. Mol. Cell Biol. 24 , 1365–1377. 10.1128/mcb.24.3.1365-1377.2004.14729979 31. Rouillard AD , Gundersen GW , Fernandez NF , Wang Z , Monteiro CD , McDermott MG , and Ma’ayan A (2016). The harmonizome: a collection of processed datasets gathered to serve and mine knowledge about genes and proteins. Database 2016, baw100. 10.1093/database/baw100. 32. Mayr B , and Montminy M (2001). Transcriptional regulation by the phosphorylation-dependent factor CREB. Nat. Rev. Mol. Cell Biol. 2 , 599–609. 10.1038/35085068.11483993 33. Kim JY , Park KJ , Hwang JY , Kim GH , Lee D , Lee YJ , Song EH , Yoo MG , Kim BJ , Suh YH , (2017). Activating transcription factor 3 is a target molecule linking hepatic steatosis to impaired glucose homeostasis. J. Hepatol. 67 , 349–359. 10.1016/j.jhep.2017.03.023.28365312 34. Zhao P , and Saltiel AR (2020). From overnutrition to liver injury: AMP-activated protein kinase in nonalcoholic fatty liver diseases. J. Biol. Chem. 295 , 12279–12289. 10.1074/jbc.REV120.011356.32651233 35. Zhao P , Sun X , Chaggan C , Liao Z , In Wong K , He F , Singh S , Loomba R , Karin M , Witztum JL , and Saltiel AR (2020). An AMPK-caspase-6 axis controls liver damage in nonalcoholic steatohepatitis. Science 367 , 652–660. 10.1126/science.aay0542.32029622 36. Dorn C , Engelmann JC , Saugspier M , Koch A , Hartmann A , Müller M , Spang R , Bosserhoff A , and Hellerbrand C (2014). Increased expression of c-Jun in nonalcoholic fatty liver disease. Lab. Invest. 94 , 394–408. 10.1038/labinvest.2014.3.24492282 37. Vansaun MN , Mendonsa AM , and Lee Gorden D (2013). Hepatocellular proliferation correlates with inflammatory cell and cytokine changes in a murine model of nonalchoholic fatty liver disease. PLoS One 8 , e73054. 10.1371/journal.pone.0073054.24039859 38. Zhu C , Tabas I , Schwabe RF , and Pajvani UB (2021). Maladaptive regeneration - the reawakening of developmental pathways in NASH and fibrosis. Nat. Rev. Gastroenterol. Hepatol. 18 , 131–142. 10.1038/s41575-020-00365-6.33051603 39. Zhu Q , Wang H , Jiang B , Ni X , Jiang L , Li C , Wang X , Zhang F , Ke B , and Lu L (2018). Loss of ATF3 exacerbates liver damage through the activation of mTOR/p70S6K/HIF-1alpha signaling pathway in liver inflammatory injury. Cell Death Dis. 9 , 910. 10.1038/s41419-018-0894-1.30185770 40. Cabrera D , Ruiz A , Cabello-Verrugio C , Brandan E , Estrada L , Pizarro M , Solis N , Torres J , Barrera F , and Arrese M (2016). Diet-induced nonalcoholic fatty liver disease is associated with sarcopenia and decreased serum insulin-like growth factor-1. Dig. Dis. Sci. 61 , 3190–3198. 10.1007/s10620-016-4285-0.27572941 41. Farrell GC , Larter CZ , Hou JY , Zhang RH , Yeh MM , Williams J , dela Pena A , Francisco R , Osvath SR , Brooling J , (2009). Apoptosis in experimental NASH is associated with p53 activation and TRAIL receptor expression. J. Gastroenterol. Hepatol. 24 , 443–452. 10.1111/j.1440-1746.2009.05785.x.19226377 42. Guarino M , Kumar P , Felser A , Terracciano LM , Guixé-Muntet S , Humar B , Foti M , Nuoffer JM , St-Pierre MV , and Dufour JF (2020). Exercise attenuates the transition from fatty liver to steatohepatitis and reduces tumor formation in mice. Cancers 12 , E1407. 10.3390/cancers12061407. 43. Katsyuba E , Romani M , Hofer D , and Auwerx J (2020). NAD(+) homeostasis in health and disease. Nat. Metab. 2 , 9–31. 10.1038/s42255-019-0161-5.32694684 44. Montllor-Albalate C , Song Z , and Chen D (2021). The therapeutic promises of NAD(+) boosters. Cell Metab. 33 , 1274–1275. 10.1016/j.cmet.2021.06.008.34233170 45. Rajman L , Chwalek K , and Sinclair DA (2018). Therapeutic potential of NAD-boosting molecules: the in vivo evidence. Cell Metab. 27 , 529–547. 10.1016/j.cmet.2018.02.011.29514064 46. Tarragó MG , Chini CCS , Kanamori KS , Warner GM , Caride A , de Oliveira GC , Rud M , Samani A , Hein KZ , Huang R , (2018). A potent and specific CD38 inhibitor ameliorates age-related metabolic dysfunction by reversing tissue NAD(+) decline. Cell Metab. 27 , 1081–1095.e10. 10.1016/j.cmet.2018.03.016.29719225 47. Drew L (2017). Fighting the fatty liver. Nature 550 , S102–S103. 10.1038/550S102a.29019968 48. Loomba R , Friedman SL , and Shulman GI (2021). Mechanisms and disease consequences of nonalcoholic fatty liver disease. Cell 184 , 2537–2564. 10.1016/j.cell.2021.04.015.33989548 49. Perry RJ , Samuel VT , Petersen KF , and Shulman GI (2014). The role of hepatic lipids in hepatic insulin resistance and type 2 diabetes. Nature 510 , 84–91. 10.1038/nature13478.24899308 50. Woo Baidal JA , and Lavine JE (2016). The intersection of nonalcoholic fatty liver disease and obesity. Sci. Transl. Med. 8 , 323rv1. 10.1126/scitranslmed.aad8390. 51. Chen D , Steele AD , Lindquist S , and Guarente L (2005). Increase in activity during calorie restriction requires Sirt1. Science 310 , 1641. 10.1126/science.1118357.16339438 52. Cohen HY , Miller C , Bitterman KJ , Wall NR , Hekking B , Kessler B , Howitz KT , Gorospe M , de Cabo R , and Sinclair DA (2004). Calorie restriction promotes mammalian cell survival by inducing the SIRT1 deacetylase. Science 305 , 390–392. 10.1126/science.1099196.15205477 53. Mohrin M , Shin J , Liu Y , Brown K , Luo H , Xi Y , Haynes CM , and Chen D (2015). Stem cell aging. A mitochondrial UPR-mediated metabolic checkpoint regulates hematopoietic stem cell aging. Science 347 , 1374–1377. 10.1126/science.aaa2361.25792330 54. Qiu X , Brown K , Hirschey MD , Verdin E , and Chen D (2010). Calorie restriction reduces oxidative stress by SIRT3-mediated SOD2 activation. Cell Metab. 12 , 662–667. 10.1016/j.cmet.2010.11.015.21109198 55. Someya S , Yu W , Hallows WC , Xu J , Vann JM , Leeuwenburgh C , Tanokura M , Denu JM , and Prolla TA (2010). Sirt3 mediates reduction of oxidative damage and prevention of age-related hearing loss under caloric restriction. Cell 143 , 802–812. 10.1016/j.cell.2010.10.002.21094524 56. Wu T , Liu YH , Fu YC , Liu XM , and Zhou XH (2014). Direct evidence of sirtuin downregulation in the liver of non-alcoholic fatty liver disease patients. Ann. Clin. Lab. Sci. 44 , 410–418.25361925 57. Xu X , Deng X , Chen Y , Xu W , Xu F , and Liang H (2022). SIRT1 mediates nutritional regulation of SREBP-1c-driven hepatic PNPLA3 transcription via modulation of H3k9 acetylation. Genes Environ. 44 , 18. 10.1186/s41021-022-00246-1.35624499 58. Dahl TB , Haukeland JW , Yndestad A , Ranheim T , Gladhaug IP , Damås JK , Haaland T , Løberg EM , Arntsen B , Birkeland K , (2010). Intracellular nicotinamide phosphoribosyltransferase protects against hepatocyte apoptosis and is down-regulated in nonalcoholic fatty liver disease. J. Clin. Endocrinol. Metab. 95 , 3039–3047. 10.1210/jc.2009-2148.20392873 59. Parker R , Schmidt MS , Cain O , Gunson B , and Brenner C (2020). Nicotinamide adenine dinucleotide metabolome is functionally depressed in patients undergoing liver transplantation for alcohol-related liver disease. Hepatol. Commun. 4 , 1183–1192. 10.1002/hep4.1530.32766477 60. Wolf FA , Angerer P , and Theis FJ (2018). SCANPY: large-scale single-cell gene expression data analysis. Genome Biol. 19 , 15. 10.1186/s13059-017-1382-0.29409532 61. Polański K , Young MD , Miao Z , Meyer KB , Teichmann SA , and Park JE (2020). BBKNN: fast batch alignment of single cell transcriptomes. Bioinformatics 36 , 964–965. 10.1093/bioinformatics/btz625.31400197 62. Traag VA , Waltman L , and van Eck NJ (2019). From Louvain to Leiden: guaranteeing well-connected communities. Sci. Rep. 9 , 5233. 10.1038/s41598-019-41695-z.30914743 63. Finak G , McDavid A , Yajima M , Deng J , Gersuk V , Shalek AK , Slichter CK , Miller HW , McElrath MJ , Prlic M , (2015). MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data. Genome Biol. 16 , 278. 10.1186/s13059-015-0844-5.26653891 64. Luo H , Mu WC , Karki R , Chiang HH , Mohrin M , Shin JJ , Ohkubo R , Ito K , Kanneganti TD , and Chen D (2019). Mitochondrial stress-initiated aberrant activation of the NLRP3 inflammasome regulates the functional deterioration of hematopoietic stem cell aging. Cell Rep. 26 , 945–954.e4. 10.1016/j.celrep.2018.12.101.30673616 65. Dahl JA , and Collas P (2007). A quick and quantitative chromatin immunoprecipitation assay for small cell samples. Front. Biosci. 12 , 4925–4931.17569620 66. Tabula Muris Consortium; Supplemental text writing group; Principal investigators; Overall coordination; Logistical coordination; Organ collection and processing; Library preparation and sequencing; Computational data analysis; Cell type annotation; Writing group (2018). Single-cell transcriptomics of 20 mouse organs creates a Tabula Muris. Nature 562 , 367–372. 10.1038/s41586-018-0590-4.30283141 67. Gonçalves LA , Vigário AM , and Penha-Gonçalves C (2007). Improved isolation of murine hepatocytes for in vitro malaria liver stage studies. Malar. J. 6 , 169. 10.1186/1475-2875-6-169.18096071 68. Hoh CH , Tan YH , and Gan YH (2019). Protective role of kupffer cells and macrophages in Klebsiella pneumoniae-induced liver abscess disease. Infect. Immun. 87 , e00369–19-e003719. 10.1128/iai.00369-19.31285251 69. Zadravec D , Brolinson A , Fisher RM , Carneheim C , Csikasz RI , Bertrand-Michel J , Born J , Guillou H , Rudling M , and Jacobsson A (2010). Ablation of the very-long-chain fatty acid elongase ELOVL3 in mice leads to constrained lipid storage and resistance to diet-induced obesity. FASEB J 24 , 4366–4377. 10.1096/fj.09-152298.20605947

PMC009xxxxxx/PMC9825627.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7613461 6050 Neurochem Res Neurochem Res Neurochemical research 0364-3190 1573-6903 36063295 9825627 10.1007/s11064-022-03739-1 NIHMS1835773 Article Sodium para-aminosalicylic acid inhibits lead-induced neuroinflammation in brain cortex of rats by modulating SIRT1/HMGB1/NF-κB pathway Zhao Yue-song ab1 Li Jun-yan ac1 Li Zhao-cong ab1 Wang Lei-lei ab Gan Cui-liu ab Chen Jing ab Jiang Si-yang ab Aschner Michael d Ou Shi-yan ab** Jiang Yue-ming ab* a Department of Toxicology, School of Public Health, Guangxi Medical University, Nanning 530021, China b Guangxi Colleges and Universities Key Laboratory of Prevention and Control of Highly Prevalent Diseases, Guangxi Medical University, Nanning, China c Hengyang Center for Disease Control and Prevention, Hengyang, China d Department of Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, New York 10461, USA 1 These authors have contributed equally to this article. Authors’ contributions YSZ: Investigation, Formal analysis, Writing original draft, Writing - review & editing. JYL: Investigation, Methodology, Formal analysis, Data curation. ZCL: Investigation, Writing original draft, Writing - review & editing. LLW: Methodology, Formal analysis. CLG, JC and SYJ: Methodology, Validation. MA: Language polishing. SYO: Investigation, Supervision. YMJ: Supervision, Project administration, Funding acquisition. * Corresponding author: Y.-M. Jiang, ymjianggxmu@163.com, Address: Department of Toxicology, School of Public Health, Guangxi Medical University, No. 22, Shuang-yong Rd., Nanning 530021, Guangxi, China. ** Co-corresponding author：S.-Y. Ou, ayin.ou@163.com. Address: Department of Toxicology, School of Public Health, Guangxi Medical University, No. 22, Shuang-yong Rd., Nanning 530021, Guangxi, China. 19 9 2022 1 2023 05 9 2022 08 1 2023 48 1 238249 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Lead (Pb) is considered to be a major environmental pollutant and occupational health hazard worldwide which may lead to neuroinflammation. However, an effective treatment for Pb-induced neuroinflammation remains elusive. The aim of this study was to investigate the mechanisms of Pb-induced neuroinflammation, and the therapeutic effect of sodium para-aminosalicylic acid (PAS-Na, a non-steroidal anti-inflammatory drug) in rat cerebral cortex. The results indicated that Pb exposure induced pathological damage in cerebral cortex, accompanied by increased levels of inflammatory factors tumor necrosis factor-alpha (TNF-α) and interleukin-1 beta (IL-1β). Moreover, Pb decreased the expression of silencing information regulator 2 related enzyme 1 (SIRT1) and brain-derived neurotrophic factor (BDNF), and increased the levels of high mobile group box 1 (HMGB1) expression and p65 nuclear factor-κB (NF-κB) phosphorylation. PAS-Na treatment ameliorated Pb-induced histopathological changes in rat cerebral cortex. Moreover, PAS-Na reduced the Pb-induced increase of TNF-α and IL-1β levels concomitant with a significant increase in SIRT1 and BDNF levels, and a decrease in HMGB1 and the phosphorylation of p65 NF-κB expression. Thus, PAS-Na may exert anti-inflammatory effects by mediating the SIRT1/HMGB1/NF-κB pathway and BDNF expression. In conclusion, in this novel study PAS-Na was shown to possess an anti-inflammatory effect on cortical neuroinflammation, establishing its efficacy as a potential treatment for Pb exposures. Lead PAS-Na SIRT1/HMGB1/NF-κB pathway BDNF Neuroinflammation pmc1. Introduction Lead (Pb) is one of the most common non-essential toxic heavy metals in global environmental pollutants and can enter the human body through various sources and pathways, such as the air, food, dust, soil and water, etc [1]. Pb toxicity can affect multiple organ systems, such as the cardiovascular, urinary, nerve, bone, blood, immune, respiratory, gastrointestinal, reproductive and endocrine systems [2]. In particular, Pb readily crosses the blood–brain barrier and accumulates in neurons and glial cells, causing central nervous system (CNS) damage [3]. Occupational exposure to Pb has been shown to be associated with longitudinal decline in cognitive function and may increase the risk for developing Alzheimer’s disease (AD) and Parkinson’s disease (PD) [4]. Moreover, low-level environmental Pb exposure can lead to intellectual deficits in children [5]. The current safe reference level of blood Pb set by the Centers for Disease Control and Prevention is 5 μg/dL, while levels of blood Pb below 3 μg/dL have been shown to elicit diminished cognitive function and maladaptive behavior in humans and animal models [6]. Moreover, Pb can accumulate and concentrate in the brain[7–9]. Studies have shown that the accumulation of Pb in cerebral cortex impairs cortical neuronal function and then affects cognitive function [10]. Although studies on Pb neurotoxicity have been carried out for decades, the precise mechanisms of Pb neurotoxicity have yet to be fully characterized. Currently, the mechanism of Pb-induced neurotoxicity has been considered to be related to oxidative stress [11], apoptosis [12], autophagy dysregulation [11], epigenetic changes [13], neuroinflammation [14], to name a few. In particular, neuroinflammation causes neural cell death and irreversible damage to peripheral neurons, which is inseparable from the occurrence and development of AD [15]. Several lines of evidence using genetic and pharmacological manipulations indicate that the potent inflammatory cytokine like interleukin-1 beta (IL-1β) and tumor necrosis factor-alpha (TNF-α) exacerbates both amyloid beta (Aβ) and tau pathologies in AD [15, 16]. According to epidemiological investigations, serum levels of inflammatory factors such as IL-1β, TNF-α, IL-6 and granulocyte colony-stimulating factor in individuals exposed to Pb are higher than those in individuals without exposure [17, 18]. Moreover, the increase in intracellular inflammatory factors IL-1β and TNF-α induced by Pb exposure were thought to be related to Pb-induced cognitive impairment in animals [14, 19]. Therefore, inhibiting excessive neuroinflammation caused by Pb exposure might afford an effective means for antagonizing Pb neurotoxicity. It is well known that silencing information regulator 2 related enzyme 1 (sirtuin1, SIRT1), a nicotinamide adenine dinucleotide (NAD+)-dependent enzymes that plays a critical regulatory role in a variety of physiological and pathological processes, including autophagy, energy metabolism, apoptosis, oxidative stress, inflammatory response and cellular aging [20, 21]. Recent studies have established the protective effects of SIRT1 in neuroinflammation-related CNS diseases [21]. Of note, SIRT1 mediates the deacetylation of high mobile group box 1 (HMGB1), which is considered to be the central component of late inflammatory response proteins. The release of HMGB1 is triggered directly by the nuclear factor-κB (NF-κB) pathway upon its nuclear translocation. In response, cells release large numbers of inflammatory cytokines initiating a cascade amplification of inflammatory responses [22]. Pb exposure has been shown to down-regulate the expression of SIRT1 phosphorylation, and reduce the expression of brain-derived neurotrophic factor (BDNF), which plays an important role in regulating neuronal synapses growth, development, differentiation and plasticity in the cerebral hippocampus, leading to cognitive impairment in animals [23, 24]. However, whether Pb-induced neurotoxicity is caused by the SIRT1-mediated HMGB1/NF-κB inflammation pathway has yet to be determined. Sodium para-aminosalicylic acid (PAS-Na) is a non-steroidal anti-inflammatory drug used as an anti-tuberculosis drug in the early clinical stage, which can cross the blood-brain barrier to exert therapeutic effects. After the 1970s, in vivo studies and clinical studies have shown that para-aminosalicylic acid (PAS) as well as its salicylates PAS-Na has the effect of preventing and treating manganese poisoning in animals and manganesm patients [25–27]. In vivo findings suggest that PAS-Na alleviated Mn-induced neurotoxicity by inhibiting inflammation, oxidative stress, cells apoptosis and restoration of amino acid neurotransmitter homeostasis [25, 28, 29]. Furthermore, results of in vitro study showed that PAS-Na can inhibit manganese-induced pyroptosis, inflammatory response and oxidative stress in cells [25, 30]. In addition, we found that PAS-Na shows efficacy against Pb-induced pathological changes in hippocampal ultrastructure, showing for the first time, the potential of PAS-Na to afford protective effect against Pb-induced neuronal damage [31]. Moreover, PAS-Na has been found to ameliorate Pb-induced hippocampal neurons and PC12 cells apoptosis by inhibiting the activation of IP3R-Ca2+-ASK1-p38 signaling pathway and enhancing glutathione levels, respectively [32, 33]. In our recently in vivo study, we found that PAS-Na ameliorated Pb-induced cognitive impairment of rats by reducing inflammatory factor IL-1β levels through the ERK1/2-p90RSK/NF-κB pathway in the hippocampus [14]. However, whether PAS-Na affects Pb-induced neuroinflammation in the cerebral cortex has yet to be determined. Therefore, this study employed both in vivo and in vitro experiments to investigate whether PAS-Na can suppress Pb-induced inflammation in the cerebral cortex through the SIRT1/HMGB1/NF-κB pathway. 2. Materials and methods 2.1 Animals and experimental design Eight-week-old specific pathogen-free grade male Sprague-Dawley (SD) rats were purchased from Experimental Animal Center of Guangxi Medical University. All animal procedures were performed strictly accordance with international animal care guidelines standards. All experimental operations have been authorized by the Animal Care and Use Committee of Guangxi Medical University [Approval ID: SCXK (Gui) 2014–0002]. All rats were housed under a constant environment with adequate temperature (22 ± 2 °C), 45–65% humidity and a light/dark cycle of 12h/12h. Animals access to water and food ad libitum. After adaption for one week, the animals were randomly divided into 6 groups (n=12): control, Pb-treated, Pb+ Low (L)-PAS, Pb+ Medium (M)-PAS, Pb+ High (H)-PAS and PAS-Na control (C-PAS) groups. The rats in Pb-treated, Pb+ (L, M, H)-PAS groups received intraperitoneally (i.p.) injection with 6 mg/kg lead acetate (Sigma, USA) once a day, 5 days per week for 4 weeks, while rats in the control and C-PAS group were i.p. injected with sterile physiological saline. Next, rats in Pb+ (L, M, H)-PAS and C-PAS) groups received back subcutaneous(s.c.) injection of 100, 200, 300 and 300 mg/kg PAS-Na (Sigma, USA), respectively, once a day, 5 days per week for 3 weeks, while control and Pb-treated groups received back s.c. injection with the same volume of sterile physiological saline. The Pb exposure dose selected for the in vivo experiments was based on a previous study that intraperitoneal injection of 6 mg/kg lead acetate in rats for 5 weeks resulted in ultrastructural damage in the hippocampus [31]. The treatment dose and duration of PAS-Na were chosen because 100, 200 or 300 mg/kg PAS-Na treatment for 3 or 6 weeks significantly attenuated Pb-induced hippocampal injury and reduced manganese-induced NLRP3 inflammasome-dependent pyroptosis [25, 31]. At the end of the experiment, the rats were anesthetized by i.p. injection of 10% chloral hydrate (3.5 ml/kg) and sacrificed. The brain tissue was dissected on ice, and the cerebral cortices were collected and stored at − 80 °C. 2.2 Determination of Pb levels in cortex Approximately 50 mg of the cortex was weighed and digested in 4 mL of 65% nitric acid (Merck ppb, USA) in 180 ◦ C for 15 min using a high-throughput closed microwave digestion instrument (CEM, USA). After the digested samples were cooled to room temperature, they were placed on a graphite heating plate (Botong Chemical Technology Co., Ltd., Shanghai, China) to evaporate until about 0.5–1 mL of liquid remained. Then the samples were made up to 5ml with double distilled water. Pb concentrations of cerebral cortex were determined by electrothermal atomic absorption spectrometry (AAnalyst800, PerkinElmer, USA) with ICE 3000 spectrometers (Thermo Fisher, USA) using a Zeeman background correction. The parameters were set as follows: analysis wavelength 283.3 nm, slit 0.7 nm and carrier gas argon (Ar). 2.3 Primary cortical neurons culture and treatments Primary cortical neurons were cultured based on a previously published protocol [34]. Briefly, cerebral hemispheres were removed from rat fetuses (embryonic day 18). The superficial vessels and the meninges of cortical tissues were carefully isolated. Then, the cortex was chopped and digested in papain (Worthington, USA) for 30 min in 37°C. The dissociated neurons were collected and resuspended in the culture medium containing 84% MEM (Gibco, USA), 10% horse serum (Hyclone, USA), 5% glucose (Sigma-Aldrich, USA), 1% Glutamax (Gibco, USA). The resuspended cell were plated in poly-L-lysine-coated culture flasks at a density of 3~5 × 105 cells/cm2 and cultured at 5% CO2 cell incubator in 37°C. After 2–4h, the medium was replaced as the neurobasal medium (Gibco, USA) supplememted with 2% B-27 (Gibco, USA), 1% Glutamax (Gibco, USA) and 0.5% 100 U/ml penicillin/streptomycin (Solarbio, Beijing, China). Routinely, half of the volume of the medium was changed every 3 days. The cortical neurons were mature at 7 days and used for experiments. To obtain PAS-Na treatment models, the primary cortical neurons were exposed to 50 μM lead acetate for 24 h, then the neurons were treated with 100, 200 or 400 μM PAS-Na for another 24 h. The Pb exposure dose and duration was selected according to our previous work that 50 μM Pb exposure for 24 h significantly increased hippocampal neuronal apoptosis in vitro [32]. The SIRT1 inhibitor EX527 (MCE, USA) was added into the wells 2 h before Pb treatment. 2.4 Histopathological staining Hematoxylin and eosin (H&E) staining procedure was performed to observe morphological and cellular changes in cerebral cortex. The specific operation steps of histopathological staining refer to our previous research [14]. 2.5 Immunohistochemistry The cortical tissue sections were placed in 60 °C oven for 2 h and then deparaffinized in xylene and dehydrated through graded concentrations of ethanol. Sections antigen retrieval in 0.01 M citrate buffer for 15 min. After blocking of endogenous peroxidase activity with 3% hydrogen peroxide, the non-specific binding was blocked with 5% goat serum for 1 h at room temperature. Then the slices incubated with primary antibodies of SIRT1(1:400, CST, USA, #9475) and NF-κB (1:1600, CST, USA, #3033) overnight at 4 °C. Sections were washed with TBST three times and incubated with an anti-rabbit polyclonal antibody for 20 min at room temperature. Next, the sections were stained with diaminobenzidine (DAB) and counterstained with hematoxylin. Finally, the sections were sealed by neutral gum. Images were captured with an EVOS Cell Imaging System. 2.6 Immunofluorescence Immunofluorescence staining was used for the assessment of the changes of BDNF expression in cortical neurons. The immunofluorescence staining method and sample processing described previously were used [25]. For primary cultured cortical neurons, anti-BDNF (1:500, Abcam, UK, ab108319) and Alexa Fluor 488-conjugated anti-rabbit IgG (1:1000, CST, USA, #4412) was used to mark neurons. To determine the levels of BDNF expression in cortical tissue sections, we double-labeled sections of cerebral cortex. The slices were incubated with primary antibodies of BDNF (1:200, Servicebio, China, GB11559) at 4 °C overnight. Then the sections were incubated with Cy3- conjugated anti-rabbit IgG (1:300, Servicebio, China, GB21303) for 1 h at room temperature. Finally, nuclei were counterstained with DAPI. Images were obtained with an EVOS Cell Imaging System. 2.7 Enzyme-linked immunosorbent assay (ELISA) For the primary cortical neurons, cell lysates were mechanically collected and stored at −80°C. Total protein of cortical tissues was extracted according to previous studies [25]. The levels of IL-1β and TNF-α were measured by ELISA kits (Elabscience, China) according to the manufacturer’s instructions and then normalized by total protein with BCA Protein Assay Kit (Beyotime, China). 2.8 Western blotting Total proteins of primary cortical neurons and cortex tissues were extracted using RIPA lyses buffer (CWBIO, China) containing protease inhibitor (Roche, USA) and phosphatase inhibitor (Roche, USA). Protein quantification was performed with an BCA Protein Assay kit (Beyotime, China).The protein samples (40 μg) were separated on 8% or 10% sodium dodecyl sulfate-polyacrylamide gels and electrophoretically transferred into polyvinylidene difluoride membranes (PVDF, 0.22 μm, Roche, USA). After blocking with 5% bovine serum albumin (Beyotime, China) for 1h at room temperature, the membranes were incubated with primary antibodies overnight at 4 °C, including rabbit anti-SIRT1 (1:1000, CST, USA, #9475), HMGB1 (1:10000, Abcam, UK, ab79823), NF-κB P65 (1:1000, CST, USA, #8242), NF-κB P-P65 (1:1000 CST, USA, #3033), GAPDH (1:1000, CST, USA, #5174). Then, the membranes were incubated with anti- rabbit IgG (1:5000, CST, USA, #7074) for 1h at room temperature. Membranes were detected using the super enhanced chemiluminescent detection kit (Millipore, USA) and quantified with Image J software. 2.9 Statistical analysis All data were presented as mean ± SD, and statistical analysis was carried out by SPSS 23.0. First, homogeneity analysis of variance was performed on the data. When the data were consistent with homogeneity of variance, one-way analysis of variance (ANOVA) was used, and (LSD) tests was carried out for multiple comparisons. If the data does not conform to homogeneity of variance, Welch test was used, and games-Howell test was used for subsequent multiple comparisons. All experiments were independently repeated for a minimum of three times. P values < 0.05 was considered statistically significant. 3. Results 3.1 Effect of PAS-Na on the Pb levels in the cortex of rats After Pb exposure for 4 weeks, the concentrations of Pb were significantly increased in cerebral cortex of Pb-exposed rats compared to the control group (p<0.01, Fig. 1). Notably, compared with the Pb-exposed group, PAS-Na treatment for 3 weeks reduced the cortical Pb levels in a dose-dependent manner. Both 200 and 300 mg/kg PAS-Na treatment significantly decreased the elevated levels of Pb in Pb-exposed rats (Fig. 1). PAS-Na treatment alone had no effect on Pb levels in cortex. 3.2 PAS-Na ameliorated histopathological changes in the cortex of Pb-exposed rat The pathological changes in cortical tissues upon Pb exposure are shown in Fig. 2. In the control and C-PAS groups, the morphology of neurons was normal, absent nuclear pyknosis and degeneration, and the tissue was intact and regular. However, in the Pb-treated group, the neuronal morphology was disordered and loosely arranged, concomitant with densely stained nuclei, pyknosis, and triangular, rhombic or irregular nuclear morphology. In the L-, M-, H-PAS-Na treatment groups, cell morphology and arrangement were regular, and the number of cells with pyknosis was lesser than that in the Pb-treated group. 3.3 Effects of PAS-Na on SIRT1/HMGB1/NF-κB signaling pathway in Pb-exposed rats cortex As shown in Fig. 3a, compared with the control group, the level of SIRT1 was significantly decreased, while the levels of HMGB1 and p-p65 were significantly increased in the Pb-exposed cortical neurons (p<0.05). To examine whether the reduction in SIRT1 affects the expression of HMGB1 and p65, we pretreated primary cortical neurons with EX527, a specific inhibitor of SIRT1. As shown in Fig. 3a, EX527 significantly reduced SIRT1 expression and enhanced HMGB1 and p-p65 levels compared to the control group, suggesting that SIRT1 mediated the expression of HMGB1 and p65 NF-κB in cortical neurons. Importantly, we found that PAS-Na treatment significantly increased protein levels of SIRT1 and reduced the expressions of HMGB1 and p-p65 in Pb-treated cortical neurons (p<0.05, Fig. 3a). Next, we tested the levels of SIRT1, HMGB1 and p-p65 in the cerebral cortex of Pb-exposed rats. Western blotting analyses showed that the expression of SIRT1 was significantly decreased in Pb-exposed rats’ cortex, while HMGB1 and p-p65 expression levels were increased compared to the control group (p<0.05, Fig, 3b). After PAS-Na treatment for 3 weeks, 300 mg/kg PAS-Na significantly increased the expression of SIRT1 and decreased the levels of HMGB1 compared to the Pb-exposed group (p<0.05). Moreover, 200 and 300 mg/kg PAS-Na significantly reduced the Pb-induced phosphorylation of p65 (p<0.05, Fig, 3b). In addition, immunohistochemical staining showed that the number of SIRT1 positive cells in the cortex of Pb-exposed rats was significantly lower than in the control group (p<0.05, Fig. 3c). After 3 weeks of PAS-Na treatment, the number of SIRT1 positive cells was increased in Pb-exposed rat cortex (p<0.05). As shown in Fig. 3d, the number of NF-κB positive cells in Pb-exposed rat cortex was higher than the control group(p<0.05), and PAS-Na treatment significantly reduced the number of NF-κB positive cells in the cortex of Pb-exposed rats. 3.4 PAS-Na ameliorated Pb-induced reduction of BDNF in cortex of rats BDNF, a conserved member of the neurotrophin family, is known to play critical roles in nervous system development [35, 36]. It has been reported that SIRT1 deficiency downregulated the expression of BDNF by limiting the level of microRNA-134 to modulate neurons synaptic plasticity [36]. To investigated the neurotoxic effect of Pb on BDNF and the intervention effect of PAS-Na, we used immunofluorescence to detect BDNF levels in cerebral cortex both in vitro and in vivo. As shown in Fig. 4a, Pb exposure significantly decreased BDNF expression in cortical neurons, which was significantly enhanced by PAS-Na treatment (p<0.05). In vivo, we observed that BDNF levels in Pb-exposed cortex was significantly reduced compared to the control group (Fig. 4b). However, the expressions of BDNF in L, M-PAS-Na treatment group were significantly increased in the cortex compared with the Pb-exposed group (Fig. 4b). 3.5 PAS-Na attenuated Pb-induced increase of inflammatory cytokines in cerebral cortex of rats. To evaluate whether PAS-Na can reduce the levels of inflammatory cytokines in Pb-exposed rat cortex, we measured the levels of TNF-α and IL-1β using ELISA. Pb exposure up-regulated the levels of TNF-α and IL-1β in cortical neurons compare to the control group (p<0.05, Fig. 5a, b). After treatment with 100, 200 and 400 μM PAS-Na for 24 h, the levels of TNF-α were significantly decreased in cortical neurons compared to the Pb-exposed group (p<0.05). Concomitantly, the levels of TNF-α and IL-1β were significantly increased in cerebral cortex of Pb-exposed rats (p<0.05, Fig. 5c, d). Treatment with 100, 200 and 300 mg/kg PAS-Na significantly reduced inflammatory cytokines TNF-α and IL-1β levels in Pb-exposed rat cortex (p<0.05, Fig. 5c, d). 4. Discussion Environmental heavy metal contamination, including Pb, continues to pose persistent human health risk [3]. Numerous studies have shown that Pb, even at low blood concentrations causes damage to cortical neurons and affect cerebral cortical-related functions, including deficits in cognitive, attention, intelligence and motor skills [7, 9, 37]. MRI data have shown that blood Pb concentration was inversely correlated with cerebral cortical volume [7–9]. Children with higher blood Pb levels exhibited smaller cortical volumes and cortical surface areas, displaying lower cognitive test scores [9]. Rodent studies have found that Pb exposure can induce oxidative stress, energy metabolism disturbance and abnormal inflammatory response in cerebral cortex [19, 38, 39]. In this study, our results in behavior (data not shown) were in accordance with the previous study that Pb exposure caused rats learning and memory impairments in the Morris water maze test, which was improved by PAS-NA treatment [14]. Furthermore, we found that Pb exposure induced cerebral cortex pathological damages and inflammatory cytokines TNF-α and IL-1β releases. PAS-Na treatment, however, ameliorated the Pb-induced the pathological changes in cortical tissues and the increased levels of TNF-α and IL-1β through SIRT1/HMGB1/NF-κB pathway. Neuroinflammation generally refers to an inflammatory response within the CNS that can be caused by various pathological insults, including infection, trauma, ischemia and various toxins [40]. Pb has been shown to induce neuroinflammation by stimulating nerve cells to release pro-inflammatory cytokines, including IL-1β, IL-6 and TNF-α [17, 39]. Individuals exposed to Pb exhibit higher serum levels of TNF-α and granulocyte colony-stimulating factor than non-exposed humans [18]. Studies in rodent mammals discovered that Pb can accelerate the brain inflammatory process by affecting the expression and activity of enzymes involved in the inflammatory process (eg, cyclooxygenase 2, caspase 1, nitrogen oxide synthase), increasing expressions of inflammatory cytokines (IL-6, TGF-β1, IL-16, IL-18, and IL-10) in the brain [41]. Furthermore, Pb exposure induced a significant increase in microglial activation, upregulating the release of cytokines including TNF-a, IL-1β, iNOS in microglial cultures alone as well as neuronal injury in the co-culture with hippocampal neurons [42]. Although microglia are considered to be a major source of proinflammatory cytokines [43], our experiments found that the levels of potent proinflammatory cytokines TNF-α and IL-1β increased sharply in primary cortical neurons of Pb-exposure rats, demonstrating that Pb can induce neuronal inflammation. Importantly, PAS-Na treatment significantly reduced TNF-α and IL-1β levels in cortical tissues, demonstrating the potential of PAS-Na for the treatment of brain inflammation. In our previous animal experiments, PAS-Na has been found to mitigate manganese-induced increases in IL-1β, IL-6, TNF-α and prostaglandins E2 levels in the brain [26]. In addition, PAS-Na was found to ameliorate the pathological changes of hippocampus and the high levels of IL-1β induced by Pb [14]. Collectively, PAS-Na, a non-steroidal anti-inflammatory drug, merits further investigations into its efficacy and potential to mitigate neurodegeneration associated with neuroinflammation. SIRT1 plays an important regulatory role in neuroinflammation. For example, an in vivo study demonstrated that SIRT1 downregulated the level of NF-κB expression and then decreased the expression of pro-inflammatory cytokines TNF-α and IL-1β, while SIRT1 inhibitor EX527 abolished the anti-inflammatory effect of BML-111 (a drug that can inhibit neutrophil recruitment and peripheral inflammation) on sepsis [44]. Activation of SIRT1 by melatonin has been shown to prevent lipopolysaccharide (LPS)-induced NLRP3 inflammasome expressions in mouse N9 microglial cells and hippocampus of depressive-like behavior mice [45]. Notably, SIRT1 may have a protective effect against Pb-induced toxicity. Li et al. [46] have demonstrated that allicin exerts protective effects against Pb-induced bone loss by activating SIRT1/FoxO1 pathway and anti-oxidative stress. The upregulation of SIRT1 by SRT1720 (a SIRT1 activator) has also been shown to alleviate Pb-induced hepatic lipid accumulation in HepG2 cells [47]. However, few studies have addressed the relationship between SIRT1 and Pb-induced neuroinflammatory responses. Our experiments found that Pb exposure inhibited SIRT1 expression and elevated TNF-α and IL-1β levels in the rat cerebral cortex both in vivo and in vitro. In vitro experiments, western blot analysis found that the expression of SIRT1 was significantly decreased, while the expressions of P-P65 and HMGB1 were elevated in the EX527 (SIRT1 inhibitor) group, suggesting that the expression of HMGB1 and P-P65 in rat cerebral cortex was regulated by SIRT1. HMGB1 plays an important role in the regulation of inflammation in the brain [48]. Under pathological conditions, HMGB1 is released into the extracellular space from cellular nuclei, causing activation of microglia, thereby participating in the neuroinflammation [49]. The level of HMGB1 release is regulated by SIRT1, and HMGB1 leads to the activation of NF-κB, which augments the release of various pro-inflammatory cytokines (TNF-α and IL-1β), ultimately leading to CNS inflammation [22, 50]. Herein, our results have established that Pb exposure inhibited the expression of SIRT1 protein and promoted the expression of HMGB1 and P-P65 NF-κB, leading to increased levels of inflammatory mediators such as TNF-α and IL-1β in primary cortical neurons and rats cerebral cortex. These results demonstrate that the SIRT1/HMGB1/NF-κB pathway plays a pivotal mediating role in Pb-induced neuroinflammation in rat cortex. BDNF is a critical factor for neuronal development and synaptic plasticity and hopes to achieve as potential treatments in CNS disorders [51]. A number of studies have reported that Pb exposure reduces BDNF gene and protein expression, and it may also alter the transport of BDNF vesicles to the releasing site, resulting in reduced extracellular mature BDNF concentrations [24, 52]. BDNF may serve as a conduit between inflammation and Pb-induced neurodegeneration. It has been reported that maternal Pb exposure decreases the levels of BDNF and tyrosine receptor-kinase protein B in the offspring rat pups and retards brain development with concomitant increase in oxidative stress, and inflammatory response [24]. The relationship between BDNF and the inflammatory response is thought to be associated with increased NF-κB expression [53]. Reduction of pre-synaptic BDNF signaling activates the kinases IKKα and IKKβ, which then phosphorylates and degrades IκBα (the NF-κB inhibitory unit), and induces NF-κB release [53]. In the present study, Pb exposure significantly decreased the expression of BDNF in rat cortical tissue and primary neurons, accompanied by an increase in the expression of p-p65 NF-κB, suggesting that BDNF/NF-κB may serve as a signaling pathway in mediating Pb-induced cortical inflammation. In addition, several studies found that SIRT1 can mediate the expression of BDNF in neurons [23, 36]. Chen et al. [23] have reported that Pb exposure downregulated the expression of SIRT1, CREB phosphorylation and BDNF to impair cognitive function in rats. In agreement, our study found that Pb exposure reduced SIRT1 and BDNF expression. However, whether Pb affects the inflammatory process by mediating SIRT1 to regulate BDNF expression needs further study. CaNa2-EDTA is often used as a therapeutic agent for Pb poisoning in clinical treatment, owing its chelating metal properties. However, CaNa2-EDTA cannot cross the blood-brain barrier, limiting its therapeutic effect in the brain [54]. PAS-Na and its major metabolite N-acetyl-para-aminosalicylic acid can readily cross the blood-brain barrier into brain [55–57], which make they may be able to chelate Pb in the brain because their structures contain carboxyl and hydroxyl groups. In addition, PAS-NA has anti-inflammatory properties and has been shown to reduce levels of inflammatory factors in several in vitro and in vivo studies. Li et al. [26] have reported that PAS-Na decreased manganese-induced high levels of IL-1β, IL-6, TNF-α and IL-1β and prostaglandins E2 through MAPK pathway and COX-2 in the brain. PAS-Na was also found to inhibit manganese-induced inflammation by inhibiting NLRP3-CASP1 inflammasome pathway, NF-κB activation and oxidative stress [25]. Recently, PAS-Na has been shown to reduce Pb-induced elevation of IL-1β levels in rat hippocampus through ERK1/2-p90RSK/NF-κB pathway [14]. In the present study, we found that PAS-Na reduced the Pb-induced increase of TNF-α and IL-1β levels in the cerebral cortex by regulating the SIRT1/HMGB1/NF-κB pathway and BDNF levels. Overall, PAS-Na shows efficacy in treating brain inflammation, but further research is needed. 5. Conclusion The present study demonstrated that Pb exposure leads to cerebral cortex pathological damage, which may be due to the neuroinflammation induced by Pb. Pb decreased the levels of SIRT1 and BDNF, while increasing the levels of HMGB1 and NF-κB, accompanied with increased levels of TNF-α and IL-1β in the cortex. PAS-Na treatment ameliorated Pb-induced cortical pathological damage, exerting anti-inflammatory effects by mediating the SIRT1/HMGB1/NF-κB pathway and BDNF expression. Nonetheless, the therapeutic effect of PAS-Na on brain neuroinflammation requires additional experimentation. Acknowledgements The author thanked Dr. Yan Li from Guangxi Zhuang Autonomous Region Institute for the Prevention and Treatment of Occupational Disease for her help in the detection of cerebral cortex Pb levels. Funding Funding support was provided by grants from the National Natural Science Foundation of China (NSFC 81773476). Data Availability All data generated or analyzed during this study are included in this published article. Fig. 1 Effect of PAS-Na on the Pb levels in the cortex of rats. The Pb concentrations in cortex were determined by electrothermal atomic absorption spectrometry (n = 4 per group). Values are presented as mean ± SD. *p < 0.05 and **p < 0.01: compared to the control group; #p < 0.05 and ##p < 0.01: compared to Pb-treated group. Fig. 2 PAS-Na ameliorated histopathological changes in Pb-exposed rat cortex. The cortical sections were stained by HE. Arrow indicates degenerating neuron with nuclei densely stained, pyknosis, and showing triangular, rhombic or irregular nuclei morphology. Scale bar: 50 μm. Fig. 3 Effects of PAS-Na on SIRT1/HMGB1/NF-κB signaling pathway in Pb-exposed rat cortex. a Primary cortical neurons were exposed to 50 μM lead acetate for 24 h, follow by 100, 200 or 400 μM PAS-Na treatment for 24 h. b-d Rats were treated with 6 mg/kg lead acetate for 4 weeks, followed by 100, 200 or 300 mg/kg PAS-Na treatment for 3 weeks. The protein levels of SIRT1, HMGB1, p65 NF-κB were detected by western blotting (a, b). Immunohistochemical results of SIRT1 (c) and NF-κB (d) in the cerebral cortex of rats. Scale bar: 50 μm. Values are presented as mean ± SD. *p < 0.05 and **p < 0.01: compared to the control group; #p < 0.05 and ##p < 0.01: compared to Pb-treated group. Fig. 4 PAS-Na ameliorated Pb-induced reduction of BDNF in cortex of rats. a BDNF immunofluorescence (green) in cortical neurons. b BDNF immunofluorescence (red) in cortex tissues counterstained with DAPI (blue). Scale bar: 50 μm. Values are presented as mean ± SD. *p < 0.05 and **p < 0.01: compared to the control group; #p < 0.05 and ##p < 0.01: compared to Pb-treated group. Fig. 5 PAS-Na attenuated Pb-induced increase of inflammatory cytokines in cerebral cortex of rats. a, b Inflammatory cytokines TNF-α (a) and IL-1β (b) levels in cortical neurons. c, d Inflammatory cytokines TNF-α (c) and IL-1β (d) levels in cortex tissues. Values are presented as mean ± SD. *p < 0.05 and **p < 0.01: compared to the control group; #p < 0.05 and ##p < 0.01: compared to Pb-treated group. Fig. 6 Schema of the mechanisms by which PAS-Na protects Pb-induced neuroinflammation. Competing Interests The authors have no relevant financial or non-financial interests to disclose. Ethics Approval and Consent to participate All animal procedures performed in this study were performed strictly according to the international standards of animal care guidelines and have been approved by the Animal Care and Use Committee of Guangxi Medical University [Approval ID: SCXK (Gui) 2014–0002]. Consent for publish The paper has been approved by all authors for publication. References Uncategorized References 1. Mitra A , Chatterjee S , Kataki S , Rastogi RP , and Gupta DK , (2021) Bacterial Tolerance Strategies against Lead Toxicity and Their Relevance in Bioremediation Application. Environ Sci Pollut Res Int. 28 (12 ): p. 14271–14284.10.1007/s11356-021-12583-9.33528774 2. Mitra P , Sharma S , Purohit P , and Sharma P , (2017) Clinical and Molecular Aspects of Lead Toxicity: An Update. Crit Rev Clin Lab Sci. 54 (7–8 ): p. 506–528.10.1080/10408363.2017.1408562.29214886 3. Hon KL , Fung CK , and Leung AK , (2017) Childhood Lead Poisoning: An Overview. Hong Kong Med J. 23 (6 ): p. 616–21.10.12809/hkmj176214.29026055 4. Schwartz BS , Stewart WF , Bolla KI , Simon PD , Bandeen-Roche K , Gordon PB , Links JM , and Todd AC , (2000) Past Adult Lead Exposure Is Associated with Longitudinal Decline in Cognitive Function. Neurology. 55 (8 ): p. 1144–50.10.1212/wnl.55.8.1144.11071492 5. Lanphear BP , Hornung R , Khoury J , Yolton K , Baghurst P , Bellinger DC , Canfield RL , Dietrich KN , Bornschein R , Greene T , Rothenberg SJ , Needleman HL , Schnaas L , Wasserman G , Graziano J , and Roberts R , (2005) Low-Level Environmental Lead Exposure and Children’s Intellectual Function: An International Pooled Analysis. Environ Health Perspect. 113 (7 ): p. 894–9.10.1289/ehp.7688.16002379 6. Rocha A and Trujillo KA , (2019) Neurotoxicity of Low-Level Lead Exposure: History, Mechanisms of Action, and Behavioral Effects in Humans and Preclinical Models. Neurotoxicology. 73 : p. 58–80.10.1016/j.neuro.2019.02.021.30836127 7. Cecil KM , Brubaker CJ , Adler CM , Dietrich KN , Altaye M , Egelhoff JC , Wessel S , Elangovan I , Hornung R , Jarvis K , and Lanphear BP , (2008) Decreased Brain Volume in Adults with Childhood Lead Exposure. PLoS Med. 5 (5 ): p. e112.10.1371/journal.pmed.0050112.18507499 8. Reuben A , Elliott ML , Abraham WC , Broadbent J , Houts RM , Ireland D , Knodt AR , Poulton R , Ramrakha S , Hariri AR , Caspi A , and Moffitt TE , (2020) Association of Childhood Lead Exposure with Mri Measurements of Structural Brain Integrity in Midlife. Jama. 324 (19 ): p. 1970–1979.10.1001/jama.2020.19998.33201203 9. Marshall AT , Betts S , Kan EC , McConnell R , Lanphear BP , and Sowell ER , (2020) Association of Lead-Exposure Risk and Family Income with Childhood Brain Outcomes. Nat Med. 26 (1 ): p. 91–97.10.1038/s41591-019-0713-y.31932788 10. Leão LKR , Bittencourt LO , Oliveira ACA , Nascimento PC , Ferreira MKM , Miranda GHN , Ferreira RO , Eiró-Quirino L , Puty B , Dionizio A , Cartágenes SC , Freire MAM , Buzalaf MAR , Crespo-Lopez ME , Maia CSF , and Lima RR , (2021) Lead-Induced Motor Dysfunction Is Associated with Oxidative Stress, Proteome Modulation, and Neurodegeneration in Motor Cortex of Rats. Oxid Med Cell Longev. 2021 : p. 5595047.10.1155/2021/5595047.34659634 11. Ji X , Wang B , Paudel YN , Li Z , Zhang S , Mou L , Liu K , and Jin M , (2021) Protective Effect of Chlorogenic Acid and Its Analogues on Lead-Induced Developmental Neurotoxicity through Modulating Oxidative Stress and Autophagy. Front Mol Biosci. 8 : p. 655549.10.3389/fmolb.2021.655549.34179077 12. Liu F , Wang Z , Wei Y , Liu R , Jiang C , Gong C , Liu Y , and Yan B , (2021) The Leading Role of Adsorbed Lead in Pm(2.5)-Induced Hippocampal Neuronal Apoptosis and Synaptic Damage. J Hazard Mater. 416 : p. 125867.10.1016/j.jhazmat.2021.125867.34492814 13. Hernández-Coro A , Sánchez-Hernández BE , Montes S , Martínez-Lazcano JC , González-Guevara E , and Pérez-Severiano F , (2021) Alterations in Gene Expression Due to Chronic Lead Exposure Induce Behavioral Changes. Neurosci Biobehav Rev. 126 : p. 361–367.10.1016/j.neubiorev.2021.03.031.33819547 14. Lu LL , Zhang YW , Li ZC , Fang YY , Wang LL , Zhao YS , Li SJ , Ou SY , Aschner M , and Jiang YM , (2021) Therapeutic Effects of Sodium Para-Aminosalicylic Acid on Cognitive Deficits and Activated Erk1/2-P90(Rsk)/Nf-Kb Inflammatory Pathway in Pb-Exposed Rats. Biol Trace Elem Res.10.1007/s12011-021-02874-0. 15. Hou Y , Wei Y , Lautrup S , Yang B , Wang Y , Cordonnier S , Mattson MP , Croteau DL , and Bohr VA , (2021) Nad(+) Supplementation Reduces Neuroinflammation and Cell Senescence in a Transgenic Mouse Model of Alzheimer’s Disease Via Cgas-Sting. Proc Natl Acad Sci U S A. 118 (37 ).10.1073/pnas.2011226118. 16. Ghosh S , Wu MD , Shaftel SS , Kyrkanides S , LaFerla FM , Olschowka JA , and O’Banion MK , (2013) Sustained Interleukin-1β Overexpression Exacerbates Tau Pathology Despite Reduced Amyloid Burden in an Alzheimer’s Mouse Model. J Neurosci. 33 (11 ): p. 5053–64.10.1523/jneurosci.4361-12.2013.23486975 17. Machoń-Grecka A , Dobrakowski M , Boroń M , Lisowska G , Kasperczyk A , and Kasperczyk S , (2017) The Influence of Occupational Chronic Lead Exposure on the Levels of Selected Pro-Inflammatory Cytokines and Angiogenic Factors. Hum Exp Toxicol. 36 (5 ): p. 467–473.10.1177/0960327117703688.28395533 18. Di Lorenzo L , Vacca A , Corfiati M , Lovreglio P , and Soleo L , (2007) Evaluation of Tumor Necrosis Factor-Alpha and Granulocyte Colony-Stimulating Factor Serum Levels in Lead-Exposed Smoker Workers. Int J Immunopathol Pharmacol. 20 (2 ): p. 239–47.10.1177/039463200702000204.17624258 19. Li N , Liu F , Song L , Zhang P , Qiao M , Zhao Q , and Li W , (2014) The Effects of Early Life Pb Exposure on the Expression of Il1-B, Tnf-A and Aβ in Cerebral Cortex of Mouse Pups. J Trace Elem Med Biol. 28 (1 ): p. 100–4.10.1016/j.jtemb.2013.07.003.23999228 20. Chen C , Zhou M , Ge Y , and Wang X , (2020) Sirt1 and Aging Related Signaling Pathways. Mech Ageing Dev. 187 : p. 111215.10.1016/j.mad.2020.111215.32084459 21. Jiao F and Gong Z , (2020) The Beneficial Roles of Sirt1 in Neuroinflammation-Related Diseases. Oxid Med Cell Longev. 2020 : p. 6782872.10.1155/2020/6782872.33014276 22. Chen X , Chen C , Fan S , Wu S , Yang F , Fang Z , Fu H , and Li Y , (2018) Omega-3 Polyunsaturated Fatty Acid Attenuates the Inflammatory Response by Modulating Microglia Polarization through Sirt1-Mediated Deacetylation of the Hmgb1/Nf-Kb Pathway Following Experimental Traumatic Brain Injury. J Neuroinflammation. 15 (1 ): p. 116.10.1186/s12974-018-1151-3.29678169 23. Chen F , Zhou CC , Yang Y , Liu JW , and Yan CH , (2019) Gm1 Ameliorates Lead-Induced Cognitive Deficits and Brain Damage through Activating the Sirt1/Creb/Bdnf Pathway in the Developing Male Rat Hippocampus. Biol Trace Elem Res. 190 (2 ): p. 425–436.10.1007/s12011-018-1569-6.30414004 24. Hossain S , Bhowmick S , Jahan S , Rozario L , Sarkar M , Islam S , Basunia MA , Rahman A , Choudhury BK , and Shahjalal H , (2016) Maternal Lead Exposure Decreases the Levels of Brain Development and Cognition-Related Proteins with Concomitant Upsurges of Oxidative Stress, Inflammatory Response and Apoptosis in the Offspring Rats. Neurotoxicology. 56 : p. 150–158.10.1016/j.neuro.2016.07.013.27492863 25. Peng DJ , Li J , Deng Y , Zhu X , Zhao L , Zhang Y , Li Z , Ou S , Li S , and Jiang Y , (2020) Sodium Para-Aminosalicylic Acid Inhibits Manganese-Induced Nlrp3 Inflammasome-Dependent Pyroptosis by Inhibiting Nf-Kb Pathway Activation and Oxidative Stress. J Neuroinflammation. 17 (1 ): p. 343.10.1186/s12974-020-02018-6.33203418 26. Li SJ , Qin WX , Peng DJ , Yuan ZX , He SN , Luo YN , Aschner M , Jiang YM , Liang DY , Xie BY , and Xu F , (2018) Sodium P-Aminosalicylic Acid Inhibits Sub-Chronic Manganese-Induced Neuroinflammation in Rats by Modulating Mapk and Cox-2. Neurotoxicology. 64 : p. 219–229.10.1016/j.neuro.2017.06.012.28651968 27. Ky SQ , Deng HS , Xie PY , and Hu W , (1992) A Report of Two Cases of Chronic Serious Manganese Poisoning Treated with Sodium Para-Aminosalicylic Acid. Br J Ind Med. 49 (1 ): p. 66–9.10.1136/oem.49.1.66.1733459 28. Deng Y , Peng D , Yang C , Zhao L , Li J , Lu L , Zhu X , Li S , Aschner M , and Jiang Y , (2021) Preventive Treatment with Sodium Para-Aminosalicylic Acid Inhibits Manganese-Induced Apoptosis and Inflammation Via the Mapk Pathway in Rat Thalamus. Drug Chem Toxicol: p. 1–10.10.1080/01480545.2021.2008127. 29. Li ZC , Wang F , Li SJ , Zhao L , Li JY , Deng Y , Zhu XJ , Zhang YW , Peng DJ , and Jiang YM , (2020) Sodium Para-Aminosalicylic Acid Reverses Changes of Glutamate Turnover in Manganese-Exposed Rats. Biol Trace Elem Res. 197 (2 ): p. 544–554.10.1007/s12011-019-02001-0.31838737 30. Li J , Deng Y , Peng D , Zhao L , Fang Y , Zhu X , Li S , Aschner M , Ou S , and Jiang Y , (2021) Sodium P-Aminosalicylic Acid Attenuates Manganese-Induced Neuroinflammation in Bv2 Microglia by Modulating Nf-Kb Pathway. Biol Trace Elem Res. 199 (12 ): p. 4688–4699.10.1007/s12011-021-02581-w.33447908 31. Deng XF . ea, (2009) Efects of Sodium Para-Aminosalicylic Acid on Hippocampal Ultramicro-Structure of Subchronic Lead-Exposed Rats. J Toxicol. 23 (03 ): p. 213–216. (in Chinese). 32. Li Z-c , Wang L-l , Zhao Y-s , Peng D-j , Chen J , Jiang S-y , Zhao L , Aschner M , Li S-j , and Jiang Y-m , (2022) Sodium Para-Aminosalicylic Acid Ameliorates Lead-Induced Hippocampal Neuronal Apoptosis by Suppressing the Activation of the Ip3r-Ca2+-Ask1-P38 Signaling Pathway. Ecotoxicology and Environmental Safety. 241 : p. 113829.10.1016/j.ecoenv.2022.113829.36068756 33. al He , (2017) Effects of Sodium P-Aminosalicylate on Lead-Induced Apoptosis of Pc12 Cells in Vitro. Chin J Pharmacol Toxicol. 31 (02 ): p. 159–164. (in Chinese). 34. Roppongi RT , Champagne-Jorgensen KP , and Siddiqui TJ , (2017) Low-Density Primary Hippocampal Neuron Culture. J Vis Exp, (122 ).10.3791/55000. 35. Zhao J , Zhang Q , Zhang B , Xu T , Yin D , Gu W , and Bai J , (2020) Developmental Exposure to Lead at Environmentally Relevant Concentrations Impaired Neurobehavior and Nmdar-Dependent Bdnf Signaling in Zebrafish Larvae. Environ Pollut. 257 : p. 113627.10.1016/j.envpol.2019.113627.31796321 36. Gao J , Wang WY , Mao YW , Gräff J , Guan JS , Pan L , Mak G , Kim D , Su SC , and Tsai LH , (2010) A Novel Pathway Regulates Memory and Plasticity Via Sirt1 and Mir-134. Nature. 466 (7310 ): p. 1105–9.10.1038/nature09271.20622856 37. Wang T , Zhang J , and Xu Y , (2020) Epigenetic Basis of Lead-Induced Neurological Disorders. Int J Environ Res Public Health. 17 (13 ).10.3390/ijerph17134878. 38. Atuadu V , Benneth BA , Oyem J , Esom E , Mba C , Nebo K , Ezemeka G , and Anibeze C , (2020) Adansonia Digitata L. Leaf Extract Attenuates Lead-Induced Cortical Histoarchitectural Changes and Oxidative Stress in the Prefrontal Cortex of Adult Male Wistar Rats. Drug Metab Pers Ther.10.1515/dmdi-2020-0116. 39. Chibowska K , Korbecki J , Gutowska I , Metryka E , Tarnowski M , Goschorska M , Barczak K , Chlubek D , and Baranowska-Bosiacka I , (2020) Pre- and Neonatal Exposure to Lead (Pb) Induces Neuroinflammation in the Forebrain Cortex, Hippocampus and Cerebellum of Rat Pups. Int J Mol Sci. 21 (3 ).10.3390/ijms21031083. 40. Leng F and Edison P , (2021) Neuroinflammation and Microglial Activation in Alzheimer Disease: Where Do We Go from Here? Nat Rev Neurol. 17 (3 ): p. 157–172.10.1038/s41582-020-00435-y.33318676 41. Chibowska K , Baranowska-Bosiacka I , Falkowska A , Gutowska I , Goschorska M , and Chlubek D , (2016) Effect of Lead (Pb) on Inflammatory Processes in the Brain. Int J Mol Sci. 17 (12 ).10.3390/ijms17122140. 42. Liu MC , Liu XQ , Wang W , Shen XF , Che HL , Guo YY , Zhao MG , Chen JY , and Luo WJ , (2012) Involvement of Microglia Activation in the Lead Induced Long-Term Potentiation Impairment. PLoS One. 7 (8 ): p. e43924.10.1371/journal.pone.0043924.22952811 43. Xu L , He D , and Bai Y , (2016) Microglia-Mediated Inflammation and Neurodegenerative Disease. Mol Neurobiol. 53 (10 ): p. 6709–6715.10.1007/s12035-015-9593-4.26659872 44. Pan S , Wu Y , Pei L , Li S , Song L , Xia H , Wang Y , Yu Y , Yang X , Shu H , Zhang J , Yuan S , and Shang Y , (2018) Bml-111 Reduces Neuroinflammation and Cognitive Impairment in Mice with Sepsis Via the Sirt1/Nf-Kb Signaling Pathway. Front Cell Neurosci. 12 : p. 267.10.3389/fncel.2018.00267.30186119 45. Arioz BI , Tastan B , Tarakcioglu E , Tufekci KU , Olcum M , Ersoy N , Bagriyanik A , Genc K , and Genc S , (2019) Melatonin Attenuates Lps-Induced Acute Depressive-Like Behaviors and Microglial Nlrp3 Inflammasome Activation through the Sirt1/Nrf2 Pathway. Front Immunol. 10: p. 1511.10.3389/fimmu.2019.01511.30723470 46. Li D , Liang H , Li Y , Zhang J , Qiao L , and Luo H , (2021) Allicin Alleviates Lead-Induced Bone Loss by Preventing Oxidative Stress and Osteoclastogenesis Via Sirt1/Foxo1 Pathway in Mice. Biol Trace Elem Res. 199 (1 ): p. 237–243.10.1007/s12011-020-02136-5.32314144 47. Zhao Y , Mao A , Zhang R , Guan S , and Lu J , (2022) Sirt1/Mtor Pathway-Mediated Autophagy Dysregulation Promotes Pb-Induced Hepatic Lipid Accumulation in Hepg2 Cells. Environ Toxicol. 37 (3 ): p. 549–563.10.1002/tox.23420.34842334 48. Okuma Y , Liu K , Wake H , Zhang J , Maruo T , Date I , Yoshino T , Ohtsuka A , Otani N , Tomura S , Shima K , Yamamoto Y , Yamamoto H , Takahashi HK , Mori S , and Nishibori M , (2012) Anti-High Mobility Group Box-1 Antibody Therapy for Traumatic Brain Injury. Ann Neurol. 72 (3 ): p. 373–84.10.1002/ana.23602.22915134 49. Nakamura Y , Fukuta A , Miyashita K , Zhang FF , Wang D , Liu K , Wake H , Hisaoka-Nakashima K , Nishibori M , and Morioka N , (2021) Perineural High-Mobility Group Box 1 Induces Mechanical Hypersensitivity through Activation of Spinal Microglia: Involvement of Glutamate-Nmda Receptor Dependent Mechanism in Spinal Dorsal Horn. Biochem Pharmacol. 186: p. 114496.10.1016/j.bcp.2021.114496. 50. Liu X , Lu B , Fu J , Zhu X , Song E , and Song Y , (2021) Amorphous Silica Nanoparticles Induce Inflammation Via Activation of Nlrp3 Inflammasome and Hmgb1/Tlr4/Myd88/Nf-Kb Signaling Pathway in Huvec Cells. J Hazard Mater. 404 (Pt B ): p. 124050.10.1016/j.jhazmat.2020.124050.33053467 51. Wang CS , Kavalali ET , and Monteggia LM , (2022) Bdnf Signaling in Context: From Synaptic Regulation to Psychiatric Disorders. Cell. 185 (1 ): p. 62–76.10.1016/j.cell.2021.12.003.34963057 52. Neal AP , Stansfield KH , Worley PF , Thompson RE , and Guilarte TR , (2010) Lead Exposure During Synaptogenesis Alters Vesicular Proteins and Impairs Vesicular Release: Potential Role of Nmda Receptor-Dependent Bdnf Signaling. Toxicol Sci. 116 (1 ): p. 249–63.10.1093/toxsci/kfq111.20375082 53. Lima Giacobbo B , Doorduin J , Klein HC , Dierckx R , Bromberg E , and de Vries EFJ , (2019) Brain-Derived Neurotrophic Factor in Brain Disorders: Focus on Neuroinflammation. Mol Neurobiol. 56 (5 ): p. 3295–3312.10.1007/s12035-018-1283-6.30117106 54. Patra RC , Swarup D , and Dwivedi SK , (2001) Antioxidant Effects of Alpha Tocopherol, Ascorbic Acid and L-Methionine on Lead Induced Oxidative Stress to the Liver, Kidney and Brain in Rats. Toxicology. 162 (2 ): p. 81–8.10.1016/s0300-483x(01)00345-6.11337108 55. Hong L , Jiang W , Zheng W , and Zeng S , (2011) Hplc Analysis of Para-Aminosalicylic Acid and Its Metabolite in Plasma, Cerebrospinal Fluid and Brain Tissues. J Pharm Biomed Anal. 54 (5 ): p. 1101–9.10.1016/j.jpba.2010.11.031.21159459 56. Hong L , Jiang W , Pan H , Jiang Y , Zeng S , and Zheng W , (2011) Brain Regional Pharmacokinetics of P-Aminosalicylic Acid and Its N-Acetylated Metabolite: Effectiveness in Chelating Brain Manganese. Drug Metab Dispos. 39 (10 ): p. 1904–9.10.1124/dmd.111.040915.21768272 57. Zheng W , Jiang YM , Zhang Y , Jiang W , Wang X , and Cowan DM , (2009) Chelation Therapy of Manganese Intoxication with Para-Aminosalicylic Acid (Pas) in Sprague-Dawley Rats. Neurotoxicology. 30 (2 ): p. 240–8.10.1016/j.neuro.2008.12.007.19150464

PMC009xxxxxx/PMC9826717.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101729147 47701 ACS Appl Bio Mater ACS Appl Bio Mater ACS applied bio materials 2576-6422 35019636 9826717 10.1021/acsabm.0c01050 NIHMS1859023 Article Off-Peak Near-Infrared-II (NIR-II) Bioimaging of an Immunoconjugate Having Peak Fluorescence Emission in the NIR-I Spectral Region for Improving Tumor Margin Delineation http://orcid.org/0000-0002-0607-7944 Hettie Kenneth S. Department of Radiology, Stanford University School of Medicine, Stanford, California 94305, United States; Department of Otolaryngology - Head & Neck Surgery, Stanford University, Stanford, California 94305, United States Teraphongphom Nutte Tarn Department of Otolaryngology - Head & Neck Surgery, Stanford University, Stanford, California 94305, United States Ertsey Robert Department of Otolaryngology - Head & Neck Surgery, Stanford University, Stanford, California 94305, United States Chin Frederick T. Department of Radiology, Stanford University School of Medicine, Stanford, California 94305, United States Author Contributions The project was managed by K.S.H. Imaging was performed by and data acquired by N.T.T. and R.E. Data were processed and analyzed by K.S.H. and N.T.T. All renditions of the manuscript were written by K.S.H. The final manuscript was reviewed by all parties. Corresponding Authors: Kenneth S. Hettie – Department of Radiology, Stanford University School of Medicine, Stanford, California 94305, United States; Department of Otolaryngology - Head & Neck Surgery, Stanford University, Stanford, California 94305, United States; Phone: 650-725-8172; khettie@stanford.edu, Frederick T. Chin – Department of Radiology, Stanford University School of Medicine, Stanford, California 94305, United States; Phone: 650-725-4182; chinf@stanford.edu 29 12 2022 21 12 2020 15 10 2020 08 1 2023 3 12 86588666 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. The primary treatment for malignant tumors remains to be resection. The strongest predictor of recurrence and postoperative prognosis is whether diseased tissue/cells remain(s) at the surgical margin. Cancer surgery entails surgeons having the capability to visually distinguish between subtle shades of color in attempts of differentiating between diseased tissue and healthy tissue under standard white-light illumination, as such tissue states appear identical at the meso-/macroscopic level. Accordingly, enhancing the capability of surgeons to do so such that they can accurately delineate the tumor margin is of paramount importance. Fluorescence-guided surgery facilitates in enhancing such capability by color-coding the surgical field with overlaid contrasting pseudo-colors from real-time intraoperative fluorescence emission via utilizing fluorescent constructs in tandem. Constructs undergoing clinical trials or that are FDA-approved provide peak fluorescence emission in the visible (405 – 700 nm) or near-infrared-I (NIR-I) spectral region (700–900 nm), whereby differentiation between tissue states progressively improves in sync with using constructs that emit longer wavelengths of light. Here, we repurpose the usage of such fluorescent constructs by establishing feasibility of a tumor-targeting immunoconjugate (cetuximab-IRDye800) having peak fluorescence emission at the NIR-I spectral region to provide improved tumor margin delineation by affording higher tumor-to-background ratios (TBRs) when measuring its off-peak fluorescence emission at the near-infrared-II (NIR-II) spectral region (1000–1700 nm) in in vivo applications. We prepared murine tumor models, administered such immunoconjugate, and imaged such models pre-/post-administration via utilizing imaging systems that separately afforded acquisition of fluorescence emission in the NIR-I or NIR-II spectral region. On doing so, we determined in vivo TBRs, ex vivo TBRs with/-out skin, and ex vivo biodistribution, all via measuring the fluorescence emission of the immunoconjugate at tumor site(s) at both spectral regions. Collectively, we established feasibility of using the immunoconjugate to afford improved tumor margin delineation by providing 2-fold higher TBRs via utilizing the NIR-II spectral region to capture off-peak fluorescence emission from a fluorescent construct having NIR-I peak fluorescence emission. Graphical Abstract bioimaging near-infrared-II tumor margin delineation fluorescence-guided surgery immunoconjguate pmc1. INTRODUCTION The primary modality for treating most malignant tumors continues to entail maximal surgical resection of the diseased tissue, wherein the approach is to remove all cancer tissue/cells such that the patient would be cured of cancer.1–5 The presence or absence of residual tumor tissue/cells in the 2–10 mm area beyond that which is resected is referred to as the surgical tumor margin.2,6,7 The presence or absence of residual tumor tissue/cells at the tumor margin is considered to be one of the strongest predictors of recurrence and survival.3,8–10 Positive margins are the result of tumor tissue/cells remaining at part or all of such perimeter after surgery, wherein the presence of residual diseased tissue/cells is associated with increased local recurrence and provides poor prognoses for the cancer patient.11–13 Any postoperative salvage surgery in attempts to remove residual diseased tissue yields poor outcomes.14 Although the primary goal of cancer surgery is to cure the patient, preservation of important anatomical structures is very important as well for achieving optimal patient outcome. However, cancer surgery primarily remains based on tissue palpation and the surgeon visualizing the anatomy under standard white-light illumination.15,16 Very subtle progressive shades of white to pink tissue (e.g., nerves, connective tissue) or red to deep-red anatomical structures (e.g., blood vessels, liver, spleen, kidney) substantially limit and prevent visual distinction of tissue types and their boundaries.17,18 Accordingly, the extraordinary difficulty in visually distinguishing malignant tissue among healthy tissue in general as well as their margins under standard white-light illumination is even further confounded because both the diseased and healthy tissues visually appear identical at the meso-/macroscopic level.19,20 As such, enhancing the capability for the operating surgeon to visually distinguish between the two tissue states and the margin is of paramount importance. Current standard-of-care for preoperative tumor tissue delineation entails the use of non-optical imaging modalities such as computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET).21,22 Such modalities when separately used or configured for intraoperative use so as to enable image-guided surgery have not improved surgical margin positivity rates.7 To overcome such hurdles, optical imaging modalities such as fluorescence imaging have afforded the recent advent of fluorescence-guided surgery (FGS), which is a surgical technique that is utilized in conjunction with a non-/targeted fluorescent construct (e.g., indocyanine green (ICG) or 5-aminolevulinic acid (5-ALA)) to enhance the capability of the operating surgeon to visually discriminate between anatomical structures as well as between diseased and healthy tissues during surgery by color-coding the surgical field and the underlying tissue planes with contrasting overlaid pseudo-colors. Here, acquiring a higher tumor-to-background ratio (TBR) affords improved tumor margin delineation because diseased cells in lower populations can be observed at such sites.23–25 As such, both anatomical structure discernment and diseased tissue mapping via implementing FGS afford decreased patient co-/morbidity and surgical duration. Moreover, FGS has the potential to be separately integrated into other imaging modalities, such as photoacoustic imaging (PA), ultrasound imaging (US), MRI, and PET, to provide a multimodal imaging platform that complements the strengths of solely utilizing FGS alone.26–28 Current fluorescent imaging constructs used for FGS span a continuum of wavelengths across the electromagnetic spectrum whereby their fluorescence emission peaks are distributed between 512 and 840 nm with a subset of fluorescent constructs emitting in the near-infrared I region (NIR-I, ~700–900 nm), which provide better light penetration depth (1–10 mm) and lower signal-to-background ratios in tissue due to favorable tissue properties when obtaining such longer wavelengths than the shorter wavelengths of the visible spectral region (405–700 nm) that the human eye can see.29 In addition, low-toxicity fluorescent constructs afford repeated extended durations of imaging, which ultimately provides for greater resolution due to the ability to acquire multiple scans when imaging.30 Imaging cameras having Si-based charge-coupled device (CCD) detectors provide optimal detection sensitivity (i.e., quantum efficiency) for acquiring fluorescence emission across the NIR-I spectral region.28 While promising, fluorescence imaging of tissue in the NIR-I spectral region is still hampered by drawbacks such as high photon attenuation, tissue autofluorescence, and light scattering.31 As such, fluorescence imaging has expanded to the development of fluorescent imaging constructs with emission peaks at near-infrared-II wavelengths (NIR-II, ~1000–1700 nm; NIR-IIa, ~1000–1500 nm; NIR-IIb, 1500–1700 nm) as imaging at such longer wavelengths affords reduced scattering, minimal autofluorescence, and deeper penetration depths (10–30 mm) in comparison to the few well-established FDA-approved fluorophores and the limited fluorescent constructs undergoing clinical trials or FDA approval that also have peak fluorescence emission in the NIR-I spectral region only.32–37 Accordingly, in theory, capturing the NIR-IIb peak fluorescence emission from a fluorescent construct that provides maximal emission in such a region could be capable of providing even greater visual distinction between diseased tissue and healthy tissue, when compared to acquiring the off-peak fluorescence emission of a construct that has peak fluorescence in the NIR-I spectral region, due to affording higher TBRs arising in part from achieving extremely low autofluorescence, provided that such a NIR-IIb fluorescence-emitting construct also maintains a sufficient quantum yield (in water).37,38 For visualizing fluorescence emission in the NIR-II optical window, imaging cameras with InGaAs array detectors allow for such by providing optimal detection sensitivity across such a spectral region. However, there is extremely limited availability of molecular fluorescent constructs that emit fluorescence in the NIR-II spectral window in general or that (i) are FDA-approved, (ii) are being used in clinical trials, or (iii) are currently undergoing FDA approval, if even any at all.23 Accordingly, repurposing the fluorescent constructs that provide peak fluorescence emission in the NIR-I spectral region, which have already undergone FDA approval, are undergoing clinical trials, or are currently undergoing FDA approval, can fast track the clinical translation of utilizing the NIR-II spectral region for measuring their off-peak fluorescence emission when performing FGS such that significantly better outcomes and survival rates are achieved.39 Due to the aforementioned drawbacks of imaging at the NIR-I spectral region and the current status of a limited number of fluorescent constructs with peak emission in the NIR-I spectral region that are FDA-approved or in the process of gaining such approval, herein, we sought to repurpose their usage by establishing the feasibility of utilizing a fluorescent construct in the form of an immunoconjugate (cetuximab-IRDye800) that affords peak fluorescence emission in the NIR-I spectral region (~795 nm) to provide improved tumor margin delineation by affording higher tumor-to-background ratios (TBRs) when measuring its off-peak fluorescence emission in the NIR-II optical window (1000–1700 nm) in in vivo preclinical applications. More specifically, we sought to reinvent the usage of FDA-approved fluorescent constructs or those that are currently undergoing clinical translation by simply turning our attention to the NIR-II spectral region for acquiring fluorescence emission for FGS in order to ultimately afford improved outcomes and survival rates. In doing so, very few regulatory obstacles are encountered when merely seeking the use of a new imaging device (i.e., an InGaAs camera) to use a fluorescent construct that has already gained FDA approval for human studies/clinical trials in comparison to the many years and considerable cost that would be entailed to have a newly-established fluorescent construct that provides peak fluorescence emission in the NIR-II spectral region undergo clinical translation from the lab bench into the clinic. In addition, we chose to use the cetuximab-IRDye800 immunoconjugate because it has been utilized for numerous clinical trials, displayed a safe and efficacious profile in humans, and has gained FDA approval toward treating various cancers in various applications, whereas the limited developed organic dyes providing peak fluorescence emission in the NIR-II spectral region (used either in the fluorophore-only or immunoconjugate form) have not undergone such scrutiny especially in humans.40,41 To demonstrate a proof of concept, we prepared xenograft murine tumor models using a high EGFR-expressing cancer cell line (HCT 116) and administered the immunoconjugate to such models with the exception of a negative control, whereby we administered only saline to such. We imaged the xenograft murine HCT 116 tumor models pre- and post-administration of the immunoconjugate daily via utilizing separate small animal fluorescence imaging systems that afford the acquisition of fluorescence emission in either the NIR-I or NIR-II spectral region. On doing so, after defining the region-of-interest circumscribing each tumor (ROI), we determined the (i) in vivo TBRs, (ii) ex vivo TBRs both without and with the skin, and (iii) ex vivo biodistribution, all via measuring the fluorescence emission intensities at the site(s) of interest in the NIR-I and NIR-II spectral regions. Impressively, we obtained an average TBR of 12.1 and effectively a 2-fold higher average A TBR of 22.2 upon analyzing the average fluorescence emission intensities at the NIR-I and off-peak NIR-II spectral regions, respectively. Next, we measured the ex vivo TBR both without and with the skin attached, wherein the former better mimics intraoperative conditions. We obtained an average TBR of 5.2 (with skin) and an 11.2-fold higher average TBR of 58.0 (without skin) upon analysis when the two values were compared with both having been determined from using the acquired fluorescence emission intensities from only the NIR-I spectral region. We obtained an average TBR of 4.4 (with skin) and a 10.1-fold higher average TBR of 44.5 (without skin) upon analysis when the two values are compared with both having been determined from using the acquired fluorescence emission intensities at only the NIR-II spectral region. Lastly, in assessing the accumulation or clearance profile of the immunoconjugate of each organ, we obtained an average organ-to-background (OBR) of 27.8 for the kidney, which is nearly 2-fold higher than the liver when comparing the acquired fluorescence emission intensities from the NIR-I spectral region. On the other hand, we obtained an average OBR of 3.4 for the liver, which is 3-fold higher than the kidney when comparing the acquired fluorescence emission intensities from the NIR-II spectral region. In all, we established the feasibility of a tumor-targeting fluorescent construct in the form of the cetuximab-IRDye800 immunoconjugate, which has a peak fluorescence emission in the NIR-I spectral region, to provide higher TBRs that facilitate in improving tumor margin delineation when measuring its off-peak emission in the NIR-II spectral window in in vivo preclinical applications. Potentially, in theory, the practice of capturing off-peak fluorescence emission within the NIR-II spectral region from fluorescent constructs that provide peak fluorescence emission in the NIR-I spectral region could be universally applied towards other ex vivo and in vivo applications and be done so using other fluorescent constructs, including others that have undergone clinical trials or that are FDA-approved, so long as such fluorescent construct(s) provide(s) sufficient off-peak fluorescence emission in the NIR-II spectral region. Thus, the FDA-approved immunoconjugate, panitumumab-IRDy800, could be utilized in the same or a different application. It is important to note that the number of IRDye800 fluorophores conjugated to the antibody (cetuximab) as well as the net charge of the fluorophore plays a considerable role in determining the overall fluorescence emission intensity of the immunoconjugate, the extent of receptor binding at the tumor site, and the biodistribution of the immunoconjugate, wherein such an extent of bioconjugation of IRDye800 to cetuximab is affected by temperature, dye and antibody concentration, their duration to react, and other parameters due to the nature of its protocol not being necessarily site-specific given that cetuximab has approximately 92 exterior lysine residues with free side chains to which IRDye800 could bind (Scheme 1).42–44 The molecular structure, physical, and photophysical profile of IRDye800 are provided in Table S1. 2. RESULTS AND DISCUSSION Here, we acquired and analyzed the fluorescence emission intensities from the (i) tumor in vivo, (ii) tumor ex vivo without and with the skin, and (iii) ex vivo biodistribution of the immunoconjugate that are in the NIR-I and off-peak NIR-II separate spectral regions to evaluate the feasibility of utilizing the off-peak NIR-II fluorescence emission from a fluorescent construct that has a peak fluorescence emission in the NIR-I spectral region to improve tumor margin delineation. 2.1. In Vivo Fluorescence Imaging. We utilized fluorescence imaging toward the xenograft murine HCT 116 tumor models with one murine tumor model serving as a negative control by acquiring the average fluorescence emission intensities of the immunoconjugate (cetuximab-IRDye800) from the NIR-I and off-peak NIR-II spectral regions over a duration of 8 days (Figure 1). On observation, there appeared to be an increase in fluorescence emission from the tumor site in the NIR-I spectral region (Figure 1A–D) as well as in the NIR-II spectral region (Figure 1E–H). 2.2. In Vivo Tumor NIR-I Fluorescence Imaging Analysis. Initially, we determined that the in vivo TBR from each xenograft murine HCT 116 tumor model successively increased each day beginning with a TBR of 1.1 on post-administration day 0 to a TBR of 12.9 on post-administration day 8, which is an 11.7-fold fluorescence enhancement over the duration of 8 days (Figure 2). We obtained an average TBR of 12.1 upon analyzing the in vivo tumor images acquired when separately measuring the average fluorescence emission intensities (at days 7 and 8) acquired from the NIR-I spectral region. Interestingly, the TBRs appeared to almost double in fluorescence intensity every 2 days, with a given prior day demonstrating a TBR that is effectively comparable to that of the next day for select pairs of days. More explicitly, the TBRs determined between day 2 and day 3 were rather comparable and so were those when determined between day 4 and day 6 as well as between day 7 and day 8. 2.3. In Vivo Tumor NIR-II Fluorescence Imaging Analysis. Between day 1 and day 8 post-administration of the immunoconjugate, we obtained an average TBR of 1.0 and a TBR of 21.4, respectively, which resulted in a 21.4-fluorescence enhancement throughout the duration of 8 days (Figure 3). We obtained a TBR of 22.2 upon analyzing the in vivo tumor images acquired when separately measuring the average fluorescence emission intensities (at days 7 and 8) acquired from the off-peak NIR-II spectral region. Unlike the near-daily successive increases we observed for the TBRs when analyzing the in vivo tumor images acquired when separately measuring the average fluorescence emission intensities acquired from the NIR-I spectral region, the progression in the fluorescence intensity of the TBRs was more disjointed for the in vivo tumor images when separately measuring the average fluorescence emission intensities collected from the off-peak NIR-II spectral region. For example, post-administration, between day 0, which had a TBR of 1.0, and day 2, which had a TBR of 11.8, and thus resulted in an 11.8-fold fluorescence enhancement over the duration of 2 days, we obtained either comparable or moderately lower TBRs between days 2 and 6. Between days 6 and 7, we obtained a 2.7-fold enhancement to the TBR that was relatively sustained throughout day 8. We observed a TBR of 22.2 that is nearly 2-fold greater than the TBR of 12.1 that is obtained for the in vivo tumor images when separately measuring the average fluorescence emission intensities (at days 7 and 8) collected from the off-peak NIR-II and NIR-I spectral regions, respectively. 2.4. Ex Vivo Tumor Fluorescence Imaging Analysis. We next determined the TBRs without and with the skin when analyzing the ex vivo tumor images obtained upon separately measuring the average fluorescence emission intensities that were acquired from the NIR-I and off-peak NIR-II spectral regions (Figure 4). Unexpectedly, with the skin removed from the excised tumor (day 8 post-administration), we observed a TBR of 138.0 that is 3.1-fold greater than the TBR of 44.5 that we determined from the ex vivo tumor images obtained upon separately measuring the average fluorescence emission intensities in the NIR-I and off-peak NIR-II spectral regions, respectively. To account for such unexpected results, we presumed that differing internal environmental conditions coupled with external conditions when the skinless tumors are measured for their ex vivo fluorescence emission intensities result in a narrowing or blue-shifting of the fluorescence profile of the cetuximab-IRDye800 immunoconjugate such that there now exists less of an off-peak fluorescence emission tail. Such an explanation would indicate that the fluorescence emission intensities acquired from the NIR-I spectral region are artificially heightened when the skinless tumors are ex vivo, whereas the fluorescence emission intensities acquired from the off-peak NIR-II spectral region are artificially depressed when the skinless tumors are ex vivo. Interestingly, with the skin remaining attached to the excised tumor, we observed a TBR of 5.2 which is only 1.4-fold higher than the TBR of 4.4 that we obtained from the ex vivo tumor-with-skin images obtained upon separately measuring the average fluorescence emission intensities acquired from the NIR-I and off-peak NIR-II spectral regions, respectively. Hence, the above-mentioned results from the ex vivo tumor-with-skin images are effectively comparable. Taken together, despite the ex vivo measured fluorescence emission intensities acquired from the NIR-I spectral region presumably being artificially heightened and those acquired from the NIR-II spectral region presumably being artificially reduced, the results from both the NIR-I and NIR-II spectral regions demonstrated that the measured fluorescence emission intensities from the skinless tumors are significantly greater than those of the tumors with skin. 2.5. Ex Vivo Biodistribution Fluorescence Imaging Analysis. Lastly, we performed ex vivo biodistribution studies using the respective resected organs of a xenograft murine HCT 116 tumor model that was administered only with saline as a negative control as well as for providing the background measured fluorescence emission intensities for both the NIR-I and NIR-II spectral regions. We determined the organ-to-background ratios (OBRs) when analyzing the ex vivo organ images that we obtained upon separately measuring the average fluorescence emission intensities acquired at the NIR-I and off-peak NIR-II spectral regions, respectively (Figure 5). We obtained OBRs of 14.4, 1.7, and 27.8 for the liver, spleen, and kidney, respectively, upon separately measuring the average fluorescence emission intensities acquired from the NIR-I spectral region for each resected organ. On the other hand, we obtained OBRs of 3.4, 0.7, and 1.1 for the liver, spleen, and kidney, respectively, upon separately measuring the average fluorescence emission intensities acquired from the off-peak NIR-II spectral region for each resected organ. We determined that the TBRs were at least 2-fold higher for all organs whose fluorescence emission intensities were acquired from the NIRI spectral region when compared to those acquired from the off-peak NIR-II spectral regions; such results were unexpected. However, consistent with the plausible explanation we provided for the processed data from the skinless tumors when imaged ex vivo, we believe that differing internal environmental conditions coupled with external conditions when the organs are harvested and measured ex vivo for their fluorescence emission intensities result in a narrowing or blue-shifting of the fluorescence profile of the cetuximab-IRDye800 immunoconjugate such that there now exists a considerably reduced off-peak fluorescence emission tail. Such an explanation would indicate that the fluorescence emission intensities acquired from the NIR-I spectral region are artificially heightened when the organs are excised for ex vivo imaging, whereas the fluorescence emission intensities acquired from the off-peak NIR-II spectral region are artificially depressed when the organs are resected for ex vivo imaging. The kidney OBR compared to the liver OBR was nearly 2-fold higher when comparing the fluorescence emission intensities of the two clearance organs acquired from the NIR-I spectral region. At first glance, interestingly, it appeared that the immunoconjugate was removed from the xenograft murine HCT 116 murine tumor model by renal clearance in lieu of the intrinsic hepatic clearance mechanism that the liver provides for large biomolecules, such as for antibodies (~152,000 kDa), since the kidney glomuleri pore diameter size (diameter) is ~5.5–8 nm, which is a diameter size that is far smaller than the hydrodynamic diameter of such large biomolecules.28,45,46 However, we obtained the opposite, and the anticipated results for the fluorescence emission from the two organs that were acquired at the off-peak NIR-II spectral region, as hepatic clearance is the typical route for excretion of such large biomolecules including immunoconjugates (as opposed to small-molecule fluorophores) and is consistent with the results found in the literature.42,45 From such measured fluorescence emission, we determined that the immunoconjugate was primarily cleared by the liver as opposed to the kidney, which was found to have an OBR that is 3-fold less than that of the liver. The unanticipated results we obtained when comparing the two clearance mechanisms using NIR-I fluorescence imaging, especially when compared to the opposite of those found using NIR-II imaging, could be partly accounted for when taking into consideration the two standard deviations of the mean (i.e., error), which could at best put the results on par with each other. However, at a time when permitted to return to the laboratory (i.e., a point in time that is safe from COVID-19), we anticipate pursuing a separate larger-scale study in attempts to determine if such current results are consistent with later findings. 3. CONCLUSION As the primary modality for treating most malignant tumors continues to entail surgery whereby the outcome is strongly predicted by positive tumor margin rates, it was incumbent upon us to ascertain approaches that can improve tumor margin delineation via obtaining higher TBRs. FGS facilitates the operating surgeon by overlaying contrasting pseudo-colors onto the surgical field such that the surgeon can better distinguish between diseased tissue and healthy tissue as well as to separate anatomical structures. However, only a small handful of fluorescent constructs or the separate fluorescent agent thereof (i) have been FDA-approved, (ii) are currently undergoing clinical trials, or (iii) are undergoing the regulatory approval process. Accordingly, all such fluorescent constructs display fluorescence emission in the visible or NIR-I spectral region. The subset of those that demonstrate peak fluorescence emission in the NIR-I spectral region have improved outcomes and survival rates via facilitating in providing better delineating tumor margins. However, such fluorescent constructs are limited in the TBRs that they can afford because of the high noise level (i.e., background) associated with using the NIR-I spectral region to obtain fluorescence emission. As such, delineating tumor margins continues to remain a major obstacle. Accordingly, we sought to establish the feasibility of utilizing a fluorescent construct in the form of an immunoconjugate that has a peak fluorescence emission in the NIR-I spectral region to provide improved tumor margin delineation via affording higher TBRs when imaging the off-peak fluorescence emission in the NIR-II spectral region. By doing so, we were able to successfully establish that by acquiring the off-peak fluorescence emission in the NIR-II spectral region via utilizing a fluorescent construct (which is in clinical trials for NIR-I use only) that has peak fluorescence emission in the NIR-I spectral window can facilitate in providing improved tumor margin delineation via affording higher TBRs in in vivo preclinical applications. Indeed, we were able to achieve a 2-fold higher TBR when acquiring the fluorescence emission at the NIR-II spectral region when compared to that we obtained when using the NIR-I spectral region. As such in vivo results appear promising for clinical use, we look forward toward future studies that utilize resected tumor tissue slices to confirm and correlate tumor margin delineation at the microscopic level to such imaging pseudo-colors via histological staining. 4. EXPERIMENTAL SECTION 4.1. Cell Culture. The HCT 116 cell line was used to prepare our xenograft murine tumor models. HCT 116 cell cultures were grown in McCoy’s 5a Medium (Gibco #16600) + 10% FBS (Gibco #31605) + 100 units/mL penicillin + 100 μg/mL streptomycin (Gibco #15140–122) in a 37 °C incubator at 5% CO2. Cells were sub-cultured when they reached 80–90% confluency. The cell layer was rinsed with PBS (Gibco #10010049) and 5 mL of 0.25% (w/v) Trypsin + 0.5 mM EDTA (Gibco #25200056) solution was added to the attached cells until the cell layer was dispersed (usually within 5 min). The reaction was terminated with 5 mL of complete growth medium, and cells were collected by gently pipetting such. Cells were either used accordingly or were split 1:5 by adding about 5 × 106 cells per 75 cm2 flask in 15 mL of growth medium for further propagation. 4.2. In Vivo Tumor Models. The Administrative Panel on Laboratory Animal Care (APLAC) at Stanford University approved all animal procedures. All animals were anesthetized with inhaled 2–3% isoflurane for surgical and imaging procedures and recovered with free access to food and water. An eye lubricant and a heating pad were used during anesthetization. For xenograft tumor models, HCT 116 tumor cells (0.5 × 106 cells; 150 μL of serum-free media) that were transfected with GFP and luciferase were injected subcutaneously into female Nu/nu mice (aged 17–18 weeks; Charles River Laboratories) and allowed to grow for 2–3 weeks. Tumor growth was monitored by calipers and firefly luciferase bioluminescence imaging. The bioluminescence imaging data is provided in Figure S1. 4.3. Bioluminescence Imaging. Bioluminescence imaging was performed for ~4 weeks post-inoculation on an IVIS Spectrum (Caliper Life Sciences). Luciferase activity was visualized by injecting mice intraperitoneally with a prepared (15 mg/mL in PBS) D-Luciferin potassium salt (Biosynth International, Inc.) solution (150 μL). The luminescence was monitored after 10 min by placing the mice under nose-cone anesthesia immediately prior to imaging using an exposure time of 0.3 s, the emitted total flux (photons per second) was measured by imaging the mice until peak radiance was achieved. Peak radiance was quantified with Living Image 4.0 software (PerkinElmer; Waltham, MA). Bioluminescence imaging was routinely repeated and daily immediately prior to and after administration of either saline or cetuximab-IRDye800. 4.4. Cetuximab-IRDye800. The cetuximab-IRDye800 immunoconjugate was produced under cGMP conditions at the University of Alabama (UAB) Vector Production Facility before shipment to Stanford University Hospital Pharmacy, which is nearly identical to that found in the literature.47 In brief, cetuximab (ImClone LLC, Eli Lilly and Company) was provided as a 2 mg/mL solution and concentrated, and the pH was adjusted by buffer exchange to a 10 mg/mL solution in 50 mmol/L potassium phosphate, pH 8.5. IRDye 800CW NHS ester (LI-COR Biosciences, Lincoln, NE) underwent conjugation to cetuximab for 2 h at 20 °C in the dark, whereby an average molar ratio of 1:2.3 cetuximab:IRDye800 was attained. After column filtration to remove unconjugated dye and exchange buffer to phosphate-buffered saline (PBS), pH 7, final protein concentration adjusted to 2 mg/mL, the product was sterilized by filtration and placed into single-use vials. After conjugation, the agent was shipped to Stanford University Hospital Pharmacy. During shipping of the immunoconjugate, the temperature was stable at 4 °C. The immunoconjugate was stored at 4 °C until used upon its receipt. The same lot was used for the cohort of animals in the treatment group. 4.5. NIR-I Imaging. All NIR-I images pre- and post-injection of the immunoconjugate were captured using a LI-COR Pearl Trilogy Small Animal Imaging System, wherein a thermoelectrically cooled Si-based charged coupled device (CCD) camera was pre-installed by the manufacturer. The excitation source was a 785 nm solid-state laser diode. Fluorescence emission in the NIRI spectral region was collected at 820 ± 10 nm using a bandpass filter. The scan time was 500 ms, and the resolution was set at 170 μm. 4.6. NIR-II Imaging. All NIR-II images pre- and post-injection of the immunoconjugate were captured using a 640 × 512 pixel two-dimensional InGaAs array camera (Princeton Instruments) outfitted inside a UniNano Imaging System box (Uninano Spectral). A lens set provided tunable magnification (1× for whole body) by changing the relative position of two NIR achromats (75 and 200 mm, Thorlabs). A binning of one and a 500 ms exposure time were utilized. The excitation laser was an 808 nm laser diode at a power density of 140 mW/cm2. Fluorescence emission in the NIR-II spectral region was collected using a 1000 LP (long-pass) filter. 4.7. Statistical Analysis. Unless otherwise noted, data were expressed as mean ± standard deviation of the mean and analyzed using one-way analysis of variance followed by Tukey post hoc tests or unpaired t test with Welch’s correction from GraphPad Prism 6 (GraphPad Software, La Jolla, CA). Supplementary Material SI ACKNOWLEDGMENTS We thank Dr. Zhen Cheng at Stanford University for allowing us to use his small animal NIR-II fluorescence imaging system. We thank Dr. Eben L. Rosenthal at Stanford University for providing cGMP cetuximab-IRDye800 and allowing us to use his small animal NIR-I fluorescence imaging system. F.T.C.’s contribution to this work was supported by the Department of Energy (DOE DE-SC0008397), the Ben and Catherine Ivy Foundation, and the National Institute of Health (NIH/NCI R21 CA205564). K.S.H.’s contribution to this work was supported, in part, by NIH/NCI fellowship: F32 CA213620. Figure 1. In vivo NIR-I and NIR-II fluorescence imaging. Representative images of xenograft murine HCT 116 tumor models from 0, 2, 4, and 7 days post-administration of the immunoconjugate (200 μL, 200 mg/mL solution, 400 μg of IRDye800, cetuximab:IRDye800 average molar ratio = 1:2.3), which correspond to the subset panels of (A), (B), (C), and (D) and the subset panels of (E), (F), (G), and (H), respectively. Panels (A) and (E) are images that represent a xenograft murine HCT 116 tumor model whereby the immunoconjugate was not yet administered (i.e., pre-injection) such that it could serve as a negative control for NIR-I and NIR-II imaging, respectively. The set of panels (A), (B), (C), and (D) correspond to images obtained from acquiring the fluorescence emission of the immunoconjugate at the NIR-I spectral region. The set of panels of (E), (F), (G), and (H) correspond to images obtained from acquiring the off-peak fluorescence emission of the immunoconjugate at the NIR-II spectral region. Figure 2. Average in vivo TBR determined upon measuring and analyzing the fluorescence emission intensity of the tumor after acquiring such from the NIR-I spectral region. Immunoconjugate cetuximab-IRDye800 (200 μL, 200 mg/mL solution, 400 μg of IRDye800, cetuximab:IRDye800 average molar ratio = 1:2.3) was administered intravenously via tail-vein injection. Xenograft murine HCT 116 tumor models were imaged over a duration of 8 days post-administration of the immunoconjugate. Figure 3. Average in vivo TBR determined upon measuring and analyzing the fluorescence emission intensity after acquiring such from the NIR-II spectral region. Immunoconjugate cetuximab-IRDye800 (200 μL, 200 mg/mL solution, 400 μg of IRDye800, cetuximab:IRDye800 average molar ratio = 1:2.3) was administered intravenously via tail-vein injection. Xenograft murine HCT 116 tumor models were imaged over a duration of 8 days post-administration. Figure 4. Average ex vivo TBR determined upon measuring and analyzing the fluorescence emission intensity after acquiring such from the NIR-I and NIR-II spectral regions, respectively. Immunoconjugate cetuximab-IRDye800 (200 μL, 200 mg/mL solution, 400 μg of IRDye800, cetuximab:IRDye800 average molar ratio = 1:2.3) was administered intravenously via tail-vein injection. Xenograft murine HCT 116 tumor models were imaged over a duration of 8 days post-administration. Figure 5. Average ex vivo biodistribution analysis determined upon measuring and analyzing the fluorescence emission intensity after acquiring such in the NIR-I and NIR-II spectral regions, respectively. The respective resected organs of a xenograft murine HCT 116 tumor model that was administered with saline were used as the background for determining OBRs. Immunoconjugate cetuximab-IRDye800 (200 μL, 200 mg/mL solution, 400 μg of IRDye800, cetuximab:IRDye800 average molar ratio = 1:2.3) was administered intravenously via tail-vein injection. Xenograft murine HCT 116 tumor models were imaged over a duration of 8 days post-administration. Scheme 1. Generalized Reaction for Bioconjugation of the IRDye800 NHS Ester Fluorophore to Cetuximaba aThe nitrogen lone pair electrons of the free side chain of one of the many (~92) exterior lysine residues attack the electropositive carbonyl carbon atom that maintains the N-hydroxysuccinimide (NHS) ester functional group of IRDye800 NHS, which ultimately results in the NHS leaving from the fluorophore and concurrently such a residue establishing a covalent amide bond to IRDye800. Supporting Information The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsabm.0c01050. Structure, physical, and photophysical information on IRDye 800CW and bioluminescence imaging data (PDF) Complete contact information is available at: https://pubs.acs.org/10.1021/acsabm.0c01050 The authors declare no competing financial interest. REFERENCES (1) Zheng Y ; Yang H ; Wang H ; Kang K ; Zhang W ; Ma G ; Du S Fluorescence-Guided Surgery in Cancer Treatment: Current Status and Future Perspectives. Ann. Transl. Med 2019, 7 , S6.31032287 (2) Nguyen QT ; Tsien RY Fluorescence-Guided Surgery with Live Molecular Navigation — a New Cutting Edge. Nat. Rev. Cancer 2013, 13 , 653–662.23924645 (3) . (4) DeLong JC ; Hoffman RM ; Bouvet M Current Status and Future Perspectives of Fluorescence-Guided Surgery for Cancer. Expert Rev. Anticancer Ther 2016, 16 , 71–81.26567611 (5) Kelderhouse LE ; Chelvam V ; Wayua C ; Mahalingam S ; Poh S ; Kularatne SA ; Low PS Development of Tumor-Targeted Near Infrared Probes for Fluorescence Guided Surgery. Bioconjugate Chem. 2013, 24 , 1075–1080. (6) Fakurnejad S ; Krishnan G ; van Keulen S ; Nishio N ; Birkeland AC ; Baik FM ; Kaplan MJ ; Colevas AD ; van den Berg NS ; Rosenthal EL ; Martin BA Intraoperative Molecular Imaging for Ex Vivo Assessment of Peripheral Margins in Oral Squamous Cell Carcinoma. Front. Oncol 2020, 9 , 1476.31998640 (7) van Keulen S ; Nishio N ; Fakurnejad S ; Birkeland A ; Martin BA ; Lu G ; Zhou Q ; Chirita SU ; Forouzanfar T ; Colevas AD ; van den Berg NS ; Rosenthal EL The Clinical Application of Fluorescence-Guided Surgery in Head and Neck Cancer. J Nucl Med 2019, 60 , 758–763.30733319 (8) Tipirneni KE ; Warram JM ; Moore LS ; Prince AC ; de Boer E ; Jani AH ; Wapnir IL ; Liao JC ; Bouvet M ; Behnke NK ; Hawn MT ; Poultsides GA ; Vahrmeijer AL ; Carroll WR ; Zinn KR ; Rosenthal E Oncologic Procedures Amenable to Fluorescence-Guided Surgery. Ann. Surg 2017, 266 , 36–47.28045715 (9) Jiang Y ; Girard EJ ; Pakiam F ; Seibel EJ Calibration of Fluorescence Imaging for Tumor Surgical Margin Delineation: Multistep Registration of Fluorescence and Histological Images. J. Med. Imaging 2019, 6 , No. 025005. (10) Pop CF ; Barbieux R ; Moreau M ; Noterman D ; de Neubourg F ; Chintinne M ; Liberale G ; Larsimont D ; Nogaret JM ; Bourgeois P ; Veys I Tumor Localization and Surgical Margins Identification in Breast Cancer Using ICG-Fluorescence Imaging. Eur.J. Surg. Oncol 2019, 45 , No. e101. (11) Gao RW ; Teraphongphom NT ; van den Berg NS ; Martin BA ; Oberhelman NJ ; Divi V ; Kaplan MJ ; Hong SS ; Lu G ; Ertsey R ; Tummers WSFJ ; Gomez AJ ; Holsinger FC ; Kong CS ; Colevas AD ; Warram JM ; Rosenthal EL Determination of Tumor Margins with Surgical Specimen Mapping Using Near-Infrared Fluorescence. Cancer Res. 2018, 78 , 5144–5154.29967260 (12) Gioux S Chapter 9 - Fluorescence-Guided Surgery Imaging Systems: Basics and Advanced Concepts. In Strategies for Curative Fluorescence-Guided Surgery of Cancer; Hoffman RM , Bouvet M , Eds.; Academic Press, 2020; pp. 141–160. (13) van Keulen S ; Nishio N ; Birkeland A ; Fakurnejad S ; Martin B ; Forouzanfar T ; Cunanan K ; Colevas AD ; van den Berg NS ; Rosenthal E The Sentinel Margin: Intraoperative Ex Vivo Specimen Mapping Using Relative Fluorescence Intensity. Clin. Cancer Res 2019, 25 , 4656–4662.31142505 (14) Hamoir M ; Schmitz S ; Suarez C ; Strojan P ; Hutcheson K ; Rodrigo JP ; Mendenhall W ; Simo R ; Saba NF ; D’Cruz AK ; Haigentz M Jr. ; Bradford C ; Genden E ; Rinaldo A ; Ferlito A The Current Role of Salvage Surgery in Recurrent Head and Neck Squamous Cell Carcinoma. Cancers 2018, 10 , 267.30103407 (15) Pan J ; Deng H ; Hu S ; Xia C ; Chen Y ; Wang J ; Wang Y Real-Time Surveillance of Surgical Margins via ICG-Based near-Infrared Fluorescence Imaging in Patients with OSCC. World J. Surg. Oncol 2020, 18 , 96.32414418 (16) van Keulen S ; Nishio N ; Fakurnejad S ; van den Berg NS ; Lu G ; Birkeland A ; Martin BA ; Forouzanfar T ; Colevas AD ; Rosenthal EL Intraoperative Tumor Assessment Using Real-Time Molecular Imaging in Head and Neck Cancer Patients. J. Am. College Surg 2019, 229 , 560–567.e1. (17) Sardar HS Improving Visualization of Tumor in Fluorescence Guided Surgery. M.S., Dartmouth College, United States: New Hampshire, 2020. (18) Unger J ; Hebisch C ; Phipps JE ; Lagarto JL ; Kim H ; Darrow MA ; Bold RJ ; Marcu L Real-Time Diagnosis and Visualization of Tumor Margins in Excised Breast Specimens Using Fluorescence Lifetime Imaging and Machine Learning. Biomed. Opt. Express 2020, 11 , 1216–1230.32206404 (19) Quan YH ; Oh CH ; Jung D ; Lim J-Y ; Choi BH ; Rho J ; Choi Y ; Han KN ; Kim B-M ; Kim C ; Park J-H ; Kim HK Evaluation of Intraoperative Near-Infrared Fluorescence Visualization of the Lung Tumor Margin With Indocyanine Green Inhalation. JAMA Surg. 2020, 155 , 732–740.32579150 (20) Judy RP ; Keating JJ ; DeJesus EM ; Jiang JX ; Okusanya OT ; Nie S ; Holt DE ; Arlauckas SP ; Low PS ; Delikatny EJ ; Singhal S Quantification of Tumor Fluorescence during Intraoperative Optical Cancer Imaging. Sci. Rep 2015, 5 , 16208.26563091 (21) Haque A ; Faizi MSH ; Rather JA ; Khan MS Next Generation NIR Fluorophores for Tumor Imaging and Fluorescence-Guided Surgery: A Review. Bioorg. Med. Chem 2017, 25 , 2017–2034.28284863 (22) Low PS ; Singhal S ; Srinivasarao M Fluorescence-Guided Surgery of Cancer: Applications, Tools and Perspectives. Curr. Opin. Chem. Biol 2018, 45 , 64–72.29579618 (23) Cao J ; Zhu B ; Zheng K ; He S ; Meng L ; Song J ; Yang H Recent Progress in NIR-II Contrast Agent for Biological Imaging. Front. Bioeng. Biotechnol 2020, 7 , 487.32083067 (24) DSouza AV ; Lin H ; Henderson ER ; Samkoe KS ; Pogue B. W.s. Review of Fluorescence Guided Surgery Systems: Identification of Key Performance Capabilities beyond Indocyanine Green Imaging. J. Biomed. Optics 2016, 21 , No. 080901. (25) Harmsen S ; Teraphongphom N ; Tweedle MF ; Basilion JP ; Rosenthal EL Optical Surgical Navigation for Precision in Tumor Resections. Mol Imaging Biol 2017, 19 , 357–362.28271367 (26) Brouwer OR ; Buckle T ; Bunschoten A ; Kuil J ; Vahrmeijer AL ; Wendler T ; Valdés-Olmos RA ; van der Poel HG ; van Leeuwen FWB Image Navigation as a Means to Expand the Boundaries of Fluorescence-Guided Surgery. Phys. Med. Biol 2012, 57 , 3123–3136.22547491 (27) Baranski A-C ; Schäfer M ; Bauder-Wüst U ; Roscher M ; Schmidt J ; Stenau E ; Simpfendörfer T ; Teber D ; Maier-Hein L ; Hadaschik B ; Haberkorn U ; Eder M ; Kopka K PSMA-11–Derived Dual-Labeled PSMA Inhibitors for Preoperative PET Imaging and Precise Fluorescence-Guided Surgery of Prostate Cancer. J Nucl Med 2018, 59 , 639–645.29191856 (28) Kobayashi H ; Longmire MR ; Ogawa M ; Choyke PL Rational Chemical Design of the next Generation of Molecular Imaging Probes Based on Physics and Biology: Mixing Modalities. Colors. Signals. Chem. Soc. Rev 2011, 40 , 4626–4648.21607237 (29) Senders JT ; Muskens IS ; Schnoor R ; Karhade AV ; Cote DJ ; Smith TR ; Broekman MLD Agents for Fluorescence-Guided Glioma Surgery: A Systematic Review of Preclinical and Clinical Results. Acta Neurochir. 2017, 159 , 151–167.27878374 (30) Tian R ; Zeng Q ; Zhu S ; Lau J ; Chandra S ; Ertsey R ; Hettie KS ; Teraphongphom T ; Hu Z ; Niu G ; Kiesewetter DO ; Sun H ; Zhang X ; Antaris AL ; Brooks BR ; Chen X Albumin-Chaperoned Cyanine Dye Yields Superbright NIR-II Fluorophore with Enhanced Pharmacokinetics. Sci. Adv 2019, 5 , No. eaaw0672.31548981 (31) Zhu S ; Yung BC ; Chandra S ; Niu G ; Antaris AL ; Chen X Near-Infrared-II (NIR-II) Bioimaging via Off-Peak NIR-I Fluorescence Emission. Theranostics 2018, 8 , 4141–4151.30128042 (32) Antaris AL ; Chen H ; Cheng K ; Sun Y ; Hong G ; Qu C ; Diao S ; Deng Z ; Hu X ; Zhang B ; Zhang X ; Yaghi OK ; Alamparambil ZR ; Hong X ; Cheng Z ; Dai H A Small-Molecule Dye for NIR-II Imaging. Nat. Mater 2016, 15 , 235–242.26595119 (33) Zhang L ; Zhang Z ; Liu C ; Zhang X ; Fan Q ; Wu W ; Jiang X NIR-II Dye-Labeled Cylindrical Polymer Brushes for in Vivo Imaging. ACS Macro Lett. 2019, 8 , 1623–1628.35619383 (34) Tian R ; Ma H ; Yang Q ; Wan H ; Zhu S ; Chandra S ; Sun H ; Kiesewetter DO ; Niu G ; Liang Y ; Chen X Rational Design of a Super-Contrast NIR-II Fluorophore Affords High-Performance NIR-II Molecular Imaging Guided Microsurgery. Chem. Sci 2019, 10 , 326–332.30713641 (35) Zhu S ; Tian R ; Antaris AL ; Chen X ; Dai H Near-Infrared-II Molecular Dyes for Cancer Imaging and Surgery. Adv. Mater. Weinheim 2019, 31 , No. e1900321. (36) Wang W ; Ma Z ; Zhu S ; Wan H ; Yue J ; Ma H ; Ma R ; Yang Q ; Wang Z ; Li Q ; Qian Y ; Yue C ; Wang Y ; Fan L ; Zhong Y ; Zhou Y ; Gao H ; Ruan J ; Hu Z ; Liang Y ; Dai H Molecular Cancer Imaging in the Second Near-Infrared Window Using a Renal-Excreted NIR-II Fluorophore-Peptide Probe. Adv. Mater. Weinheim 2018, 30 , No. e1800106. (37) Li Y ; Liu Y ; Li Q ; Zeng X ; Tian T ; Zhou W ; Cui Y ; Wang X ; Cheng X ; Ding Q ; Wang X ; Wu J ; Deng H ; Li Y ; Meng X ; Deng Z ; Hong X ; Xiao Y Novel NIR-II Organic Fluorophores for Bioimaging beyond 1550 Nm. Chem. Sci 2020, 11 , 2621–2626. (38) Sun C ; Li B ; Zhao M ; Wang S ; Lei Z ; Lu L ; Zhang H ; Feng L ; Dou C ; Yin D ; Xu H ; Cheng Y ; Zhang F J-Aggregates of Cyanine Dye for NIR-II in Vivo Dynamic Vascular Imaging beyond 1500 Nm. J. Am. Chem. Soc 2019, 141 , 19221–19225.31746598 (39) Predina JD ; Newton AD ; Keating J ; Dunbar A ; Connolly C ; Baldassari M ; Mizelle J ; Xia L ; Deshpande C ; Kucharczuk J ; Low PS ; Singhal S A Phase I Clinical Trial of Targeted Intraoperative Molecular Imaging for Pulmonary Adeno-carcinomas. The Annals of Thoracic Surgery 2018, 105 , 901–908.29397932 (40) Zinn KR ; Korb M ; Samuel S ; Warram JM ; Dion D ; Killingsworth C ; Fan J ; Schoeb T ; Strong TV ; Rosenthal EL IND-Directed Safety and Biodistribution Study of Intravenously Injected Cetuximab-IRDye800 in Cynomolgus Macaques. Mol Imaging Biol 2015, 17 , 49–57.25080323 (41) Miller SE ; Tummers WS ; Teraphongphom N ; van den Berg NS ; Hasan A ; Ertsey RD ; Nagpal S ; Recht LD ; Plowey ED ; Vogel H ; Harsh GR ; Grant GA ; Li GH ; Rosenthal EL First-in-Human Intraoperative near-Infrared Fluorescence Imaging of Glioblastoma Using Cetuximab-IRDye800. J Neurooncol 2018, 139 , 135–143.29623552 (42) Choi HS ; Nasr K ; Alyabyev S ; Feith D ; Lee JH ; Kim SH ; Ashitate Y ; Hyun H ; Patonay G ; Strekowski L ; Henary M ; Frangioni JV Synthesis and In Vivo Fate of Zwitterionic Near-Infrared Fluorophores. Angewandte Chemie International Edition 2011, 50 , 6258–6263.21656624 (43) Njiojob CN ; Owens EA ; Narayana L ; Hyun H ; Choi HS ; Henary M Tailored Near-Infrared Contrast Agents for Image Guided Surgery. J. Med. Chem 2015, 58 , 2845–2854.25711712 (44) Sato K ; Gorka AP ; Nagaya T ; Michie MS ; Nani RR ; Nakamura Y ; Coble VL ; Vasalatiy OV ; Swenson RE ; Choyke PL ; Schnermann MJ ; Kobayashi H Role of Fluorophore Charge on the In Vivo Optical Imaging Properties of Near-Infrared Cyanine Dye/Monoclonal Antibody Conjugates. Bioconjugate Chem. 2016, 27 , 404–413. (45) Datta-Mannan A Mechanisms Influencing the Pharmacokinetics and Disposition of Monoclonal Antibodies and Peptides. Drug Metab. Dispos 2019, 47 , 1100–1110.31043438 (46) Vincenzi B ; Schiavon G ; Silletta M ; Santini D ; Tonini G The Biological Properties of Cetuximab. Crit. Rev. Oncol. Hematol 2008, 68 , 93–106.18676156 (47) Tummers WS ; Miller SE ; Teraphongphom NT ; Gomez A ; Steinberg I ; Huland DM ; Hong S ; Kothapalli S-R ; Hasan A ; Ertsey R ; Bonsing BA ; Vahrmeijer AL ; Swijnenburg R-J ; Longacre TA ; Fisher GA ; Gambhir SS ; Poultsides GA ; Rosenthal EL Intraoperative Pancreatic Cancer Detection Using Tumor-Specific Multimodality Molecular Imaging. Ann. Surg. Oncol 2018, 25 , 1880–1888.29667116

PMC009xxxxxx/PMC9828159.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0244160 5342 J Wildl Dis J Wildl Dis Journal of wildlife diseases 0090-3558 1943-3700 21719815 9828159 10.7589/0090-3558-47.3.501 NIHMS1858619 Article DNA VACCINATION OF BISON TO BRUCELLAR ANTIGENS ELICITS ELEVATED ANTIBODY AND IFN-γ RESPONSES Clapp Beata 1 Walters Nancy 1 Thornburg Theresa 1 Hoyt Teri 1 Yang Xinghong 1 Pascual David W. 12 1 Department of Immunology and Infectious Diseases, PO Box 173610, Montana State University, Bozeman, Montana 59717-3610, USA 2 Corresponding author (dpascual@montana.edu) 29 12 2022 7 2011 09 1 2023 47 3 501510 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Brucella abortus remains a threat to the health and well-being of livestock in states bordering the Greater Yellowstone Area. During the past several years, cohabitation of infected wildlife with cattle has jeopardized the brucellosis-free status of Idaho, USA; Wyoming, USA; and Montana, USA. Current livestock B. abortus vaccines have not proven to be efficacious in bison (Bison bison) or elk (Cervus elaphus nelsoni). One problem with the lack of vaccine efficacy may stem from the failure to understand wildlife immune responses to vaccines. In an attempt to understand their immune responses, bison were vaccinated with eukaryotic DNA expression vectors encoding the Brucella periplasmic protein, bp26, and the chaperone protein, trigger factor (TF). These DNA vaccines have previously been shown to be protective against Brucella infection in mice. Bison were immunized intramuscularly at weeks 0, 2, and 4 with bp26 and TF DNA vaccines plus CpG adjuvant or empty vector (control) plus CpG. Blood samples were collected before vaccination and at 8, 10, and 12 wk after primary vaccination. The results showed that bison immunized with bp26 and TF DNA vaccines developed enhanced antibody, proliferative T cell, and interferon-gamma (IFN-γ) responses upon in vitro restimulation with purified recombinant bp26 or TF antigens, unlike bison immunized with empty vector. Flow cytometric analysis revealed that the percentages of CD4+ and CD8+ T lymphocytes from the DNA-vaccinated groups were significantly greater than they were for those bison given empty vector. These data suggest that DNA vaccination of bison may elicit strong cellular immune responses and serve as an alternative for vaccination of bison for brucellosis. Bison brucellosis DNA vaccine immunity interferon-gamma T cells pmcINTRODUCTION Brucellae are Gram-negative facultative intracellular bacteria endemic in many areas of the world. Ten species of Brucella are recognized and classified based mainly on their preferred hosts and pathogenicity (Godfroid, 2005; Perkins et al., 2010). Animals, including humans, become infected with Brucella when mucosal membranes, open wounds, or skin abrasions come in contact with infected secretions (milk, blood, uterine discharge) or aborted fetuses (Ko and Splitter, 2003). The most common clinical manifestation of animal brucellosis is reproductive loss resulting from abortion, birth of weak offspring, or infertility (Seleem et al., 2010). In humans, brucellosis usually is associated with nonspecific flu-type symptoms, such as malaise, undulant fever, and joint aches (Olsen and Tatum, 2010). The enormous cost of brucellosis to the livestock industry, as well as its effect on public health, has prompted many countries to adopt brucellosis control and eradication programs (Olsen and Stoffregen, 2005). In the United States, a brucellosis eradication program was established in 1954 with the aim of eliminating Brucella abortus infections from cattle. The program was successful, and currently, US cattle are essentially brucellosis-free. However, the threat from brucellosis remains; infected wild animals continue to serve as reservoirs, and B. abortus–infected bison (Bison bison) and elk (Cervus elaphus nelsoni) still pose a threat in the Greater Yellowstone Area (GYA; Olsen et al., 2009). Outbreaks of brucellosis in Idaho, Wyoming, and Montana in vaccinated cattle that coexist with infected wildlife have resulted in all three states temporarily losing their brucellosis-free status (Van Campen and Rhyan, 2010). Aside from being problematic for livestock and wildlife, free-ranging wildlife remains a concern as a source of emerging human pathogens. Because humans are susceptible to zoonotic exposure through infected livestock, the coexistence of wildlife with livestock can increase disease frequency and transmission potential to humans (Rhyan and Spraker, 2010). Vaccination with live, attenuated B. abortus strains 19 and RB51 has been used to control brucellosis in cattle (Schurig et al., 2002). In most instances, the use of Brucella vaccines in wildlife species has been problematic. Vaccination with B. abortus strain RB51 has had little efficacy in bison (Davis and Elzer, 2002; Olsen et al., 2003). Strain 19 has been associated with chronic infections and abortions in bison and has been found to be ineffective as a calfhood vaccine for bison (Davis et al., 1991). Thus, the development of a more effective vaccine to protect susceptible wildlife and livestock is warranted. Immunization with naked DNA is an attractive alternative approach for immunizing against infectious diseases. Intramuscular (IM) delivery of DNA permits host synthesis of vaccines, stimulating both humoral and cellular immune responses specific to the expressed proteins (Robinson and Torres, 1997). Furthermore, DNA vaccines may have many advantages over traditional vaccines, including induction of long-lived immune responses, better stability, ease of preparation, and low cost (Oñate et al., 2003). Previous studies have proven that DNA vaccination with sodC (Oñate et al., 2003), lumazine synthase gene (Velikovsky et al., 2002), and P39 (Al-Mariri et al., 2001) can elicit partial protection against Brucella challenge in the mouse. In our previous work, we applied a search strategy to screen the Brucella melitensis 16M genome for potential vaccine candidates. We found that the periplasmic protein, bp26, and the chaperone protein, trigger factor (TF), are protective antigens when delivered as peripheral DNA vaccines (Yang et al., 2005). Because most efforts have relied mostly on small-animal laboratory models, methods remain to be shown as translatable to wildlife. We evaluated the immunogenicity of plasmid DNA carrying the Brucella bp26 and TF genes as a possible vaccine candidate for use in bison. The construction and preparation for vaccination of pcDNA3.1-bp26 and pcDNA3.1-TF vaccines have been described (Yang et al., 2005) as has the production of recombinant bp26 and TF in Pichia pastoris (Yang et al., 2007). MATERIALS AND METHODS Animal vaccination and blood collections Eight 10-mo-old bison heifers were obtained from a brucellosis-free herd not previously vaccinated with RB51. After acclimation for 4 wk, bison were randomly assigned to two groups (n=4 animals/group) for IM vaccination with 600 μg empty pcDNA3.1 vector (negative control) or a combination of pCMVbp26 (300 μg) and pCMVTF (300 μg), coadministered with 50 μg of CpG as adjuvant. The sequence of CpG (the same as in the mouse studies; Yang et al., 2005) was 5’-TCCATGACGTTCCTGACGTT-3’. Vaccination was performed at wk 0, 2, and 4. Blood samples were collected by jugular venipuncture into 10-ml syringes preloaded with 0.2 ml of sodium heparin before vaccination and at 8, 10, and 12 wk postvaccination. For serologic evaluation, blood was allowed to clot for 12 hr at 4 C and then centrifuged. Serum was aliquoted, frozen, and stored at –20 C. Antibody enzyme-linked immunosorbent assay (ELISA) To determine induced antibodies, bp26-and TF-specific ELISAs (Yang et al., 2005) were used to measure immune serum immunoglobulin (Ig) G, IgG1, and IgG2 levels. Purified bp26 and TF proteins (5 μg/ml) were used to coat Maxisorp microtiter plates (Nunc, Roskilde, Denmark) at 100 μl/well overnight at 4 C. Plates were washed four times in a wash buffer (Tris-buffered saline [pH 7.4] with 0.05% Tween 20) and blocked with 1% albumin from chicken egg Grade II (Sigma-Aldrich, St. Louis, Missouri, USA) in Trisbuffered saline for 2 hr at 37 C, incubated with serial dilutions of the sera from bison for 3 hr at room temperature (RT), and washed three times. Horseradish peroxidase–conjugated mouse anti-bovine IgG monoclonal antibody (mAb; clone IL-AR; Serotec, Raleigh, North Carolina, USA) and sheep anti-bovine IgG1 and IgG2 polyclonal antibodies (Novus Biologicals, Littleton, Colorado, USA) were used for detection. After a 90 min incubation at 37 C and washing, specific reactivity was determined by the addition of an enzyme substrate, ABTS [2,2_azinobis(3-ethylbenzthiazolinesulfonic acid)] diammonium (Moss, Inc., Pasadena, California, USA) at 100 ml/well. The absorbance was measured at 415 nm on a Kinetics Reader model ELx808 (Bio-Tek Instruments, Winooski, Vermont, USA). Endpoint titers were defined as the reciprocal of the highest dilution of a sample giving an optical density at 415 nm of 0.100 U greater than negative controls after 1 hr of incubation at 25 C. Peripheral blood mononuclear cells (PBMCs) for lymphocyte proliferation assays Subsequent to vaccination, blood was obtained from the jugular vein of all bison and placed into an acid-citrate dextrose solution. Peripheral blood mononuclear cells were enriched by density centrifugation using a Histopaq gradient (Sigma Diagnostics, Inc., St. Louis, Missouri, USA). The PBMCs were cultured at 37 C with 5% CO2 in 96-well flat-bottom plates at a concentration of 5×105 viable cells per 200 μl/well in the presence of 20 μg/ml of bp26, TF, or both; and 5.0 μg/ml concanavalin A (Sigma-Aldrich), 1.0 mg/ml ovalbumin (OVA), or no additives (unstimulated control). RPMI 1640 medium (GIBCO BRL, Grand Island, New York, USA) was supplemented with 2 mM l-glutamine, 10% heat-inactivated horse serum (GIBCO BRL), and 100 U/ml penicillin, 100 μg/ml streptomycin, referred to as complete medium (CM), for culturing PBMCs. The cells were cultured for 4 days and pulsed for 12 hr with 0.5 μCi of thymidine (50 Ci/mmol; NEN-Dupont, Boston, Massachusetts, USA) per well, and the radioactivity incorporated in the DNA was measured as mean counts per minute and obtained from triplicate cultures from each bison. In addition, a stimulation index was calculated for each animal by dividing the counts per minute of cells with antigen by the counts per minute of cells without antigen. Interferon-gamma (IFN-γ) production In vitro production of IFN-γ by PBMCs was measured at all sampling times after immunization. Briefly, PBMCs from each animal were isolated and resuspended in CM. Cells were cultured in 24-well tissue plates at 5×106 cells/ml in CM alone or with purified recombinant bp26 or TF (20 μg/ml) for 3 days at 37 C. The supernatants were collected by centrifugation and stored at −80 C. A bovine-capture ELISA was used to quantify the levels of IFN-γ from triplicate sets of samples. Microtiter wells were coated with 1 μg/ml of purified mouse anti-bovine IFN-γ mAb (clone CC302; Serotec), which we found recognized recombinant bison IFN-γ that shares >96% gene homology with bovine IFN-γ (data not shown). After blocking with PBS +1% bovine serum albumin for 2 hr at 37 C, washed wells were incubated with cell culture supernatants at 4 C for 24 hr. After washing, 0.5 μg/ml biotinylated mouse anti-bovine IFN-γ mAb (clone CC302; Serotec) was added for 90 min at 37 C. Following washing, 1:500 HRP-goat anti-biotin antibody (Vector Laboratories, Burlingame, California, USA) was added for 1 hr at RT. After washing, ABTS peroxidase substrate (Moss, Inc. Pasadena, California, USA) was added to develop the reaction. Production of IFN-γ by unstimulated cells set as a background was subtracted from all measurements. Flow cytometry Peripheral blood mononuclear cells were isolated from each blood sample by Ficoll-Hypaque density gradient centrifugation, adjusted to 1×107 cells/ml and were then fluorescently labeled with fluorochrome-conjugated mAbs for bovine γδ T cells (GD3.8) and CD4 (clone CC30; Serotec, and CD8 T cells (clone CC63; Serotec). These mAbs have been widely used in bovine studies (Smyth et al., 2001; Wilson et al., 1998). To assess whether these mAbs would recognize bison T cells, cell surface staining was compared with mAbs previously shown to react against bison T cells: anti-bovine CD4 (clone ILA11A; VMRD, Pullman, Washington, USA), -CD8 (clone CACT80C; VMRD; Simon et al., 2003; Nelson et al., 2010), and anti-bovine γδ T cells (clone ILA29; VMRD; Nelson et al., 2010), and thus, GD3.8, CC30, and CC63 are cross-reactive for bison T cells. Fluorescence was acquired on an LSR II flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA) using BD FACSDiva software, and samples were analyzed using Flowjo (TreeStar Inc., Ashland, Oregon, USA) software. Statistical analysis Analysis of variance, followed by Tukey’s method, was used to evaluate differences among antibody titers, IFN-γ production, T-cell proliferative responses, and the percentages of T-cell subsets between immunized and control bison discerned to the 95% confidence interval. RESULTS Immunization with DNA vaccines induces elevated IgG and IgG1 anti-bp26 and anti-TF antibody titers To determine whether pCMVbp26 and pCMVTF DNA vaccines could stimulate bison antibody responses, bison were immunized with IM pCMVbp26 and pCMVTF with CpG, and control group bison were immunized with pcDNA3.1 (empty vector) plus CpG three times at 2 wk intervals. Beginning at 8 wk, after primary immunization, serum samples were obtained to measure antigen-specific endpoint antibody titers (Figs. 1A, B). The bp26- and TF-specific IgG and IgG1 were detected in the immunized bison with a peak IgG titer of 210 and IgG1 titer of 29 at wk 12. The IgG2 antibodies were not detected throughout the experiment. No detectable IgG and low IgG1 antibody titers were observed for the vector-immunized bison (Figs. 1A, B). DNA-immunized bison show enhanced PBMC proliferative and IFN-γ responses To assess their cellular responses, PBMC proliferative assay was performed. Lymphocytes from pCMVbp26- plus pCMVTF-vaccinated bison showed greater proliferative responses (P<0.001; P<0.05) to purified recombinant bp26, TF, or their combination than did lymphocytes from animals immunized with pcDNA3.1 vector only at 8, 10, and 12 wk after primary immunization (Fig. 2). Only low levels of spontaneous proliferation occurred in cultures with no antigen (medium control) or irrelevant antigen (OVA control) stimulation. Throughout the study, PBMCs from bison showed elevated proliferative responses to the T-cell mitogen, ConA (Fig. 2). Because cell-mediated immunity, particularly Th1 cell-dependent, is considered crucial in protection against B. abortus infection, IFN-γ production by antigen-restimulation was measured. The PBMCs were cultured with either bp26 or TF for 3 days and were then evaluated for IFN-γ production by cytokine ELISA. Upon restimulation with bp26 (Fig. 3A) or TF (Fig. 3B), the PBMCs from pCMVbp26 plus pCMVTF-vaccinated bison showed significantly greater levels of IFN-γ (P≤0.05) than those lymphocytes from pcDNA3.1 vector-vaccinated bison beginning at 10 wk postimmunization. In samples obtained on wk 6 and 8 postimmunization, mean IFN-γ production by either immunization group did not significantly differ (Fig. 3A, B). DNA immunization enhances the numbers of CD4+ and CD8+ T cells The CD4+, CD8+, and γδ T-cell subsets were evaluated for their changes as a consequence of vaccination. The PBMCs from immunized bison were analyzed by flow cytometry at 8, 10, and 12 wk after primary immunization. The percentages of CD4+ and CD8+ T cells from the pCMVbp26 plus pCMVTF-vaccinated group were significantly greater than those bison immunized with pcDNA3.1 vector at wk 8, 10, and 12 postimmunization (Table 1). The percentage of γδT cells in the bison immunized with pCMVbp26 plus pCMVTF was significantly reduced in the peripheral blood compared with that of the bison immunized with pcDNA3.1 vector at 8 and 12 wk postimmunization (Table 1). DISCUSSION Because infected wildlife in the GYA serves as a reservoir for B. abortus, the presence of infected wildlife impedes efforts to eliminate brucellosis in the United States and will continue to be problematic in the affected areas of Wyoming, USA; Idaho, USA; and Montana, USA. Although numerous studies evaluating brucellosis vaccines in livestock have been conducted during the past decades, only a limited number of vaccines have been studied in wildlife (Davis and Elzer, 2002; Olsen and Tatum, 2010). Additionally, the work has largely been done with live S19 and RB51 vaccines (Davis et al., 1991; Olsen et al., 1997, 2003, 2009), which were originally developed for livestock. However, neither of these vaccines conferred complete protection against abortion and infection in wildlife, as evidenced by the incidences of infected livestock (Olsen et al., 2009). Therefore, a safe and effective brucellosis vaccine for free-ranging bison in the GYA might be beneficial in reducing the risk of transmission and in increasing herd protection against infection from bison as a reservoir for brucellosis. Cellular immune responses play a major role in protection against Brucella (Gonzalez-Smith et al., 2006), and for this reason, an effective vaccine against brucellosis must be based on its capacity to generate strong, cell-mediated immunity. Immunization with DNA vaccines encoding protective epitopes represents a means to generate and test various vaccines (Gurunathan et al., 2000). Because DNA vaccines can stimulate both cellular and humoral immunity (Liu, 2003), we have recently reported that, when bp26 is combined with TF, partial protection is obtained (Yang et al., 2005), suggesting that a subunit vaccine approach against brucellosis may be feasible. The combination of bp26 and TF shows significant protection against B. melitensis 16M challenge, either as a DNA vaccine (Yang et al., 2005) or as recombinant proteins (Yang et al., 2007) in BALB/c mice. Despite a significant effort to identify a single protein that confers protection (Oñate et al., 1999; Baloglu et al., 2000; Al-Mariri et al., 2001), no such candidate has been described. Therefore, we hypothesize that perhaps a combination of various antigens is required for successful protection. The use of recombinant protein technology and monoclonal antibodies has shown that the major outer membrane proteins (OMPs) appear to be of limited use as vaccines against smooth B. abortus or B. melitensis infections (Cloeckaetr et al., 2002). Recently, however, Omp25 has been shown to be involved in the virulence of B. melitensis, B. abortus, and B. ovis, and mutants lacking Omp25 are indeed attenuated in animal-infection models (Edmonds et al., 2001, 2002a, b; Jubier-Maurin et al., 2001). Pasquevich et al. have shown that immunization with OMPs of recombinant Brucella species omp16 or omp19 induces protection against B. abortus infection (Pasquevich et al., 2009). Brucella heatshock proteins appear to be ineffective as recombinant protein vaccines. HtrA and GroEL + GroES + HtrA, formulated in Ribi adjuvant, fail to protect against B. abortus 2308 despite evidence that those proteins are able to elicit antigen-specific immunity (Bae et al., 2002). Until suitable, protective vaccine candidates are identified, we have elected to test the immunogenicity of our bp26 and TF DNA vaccines in bison. Protection against brucellosis is thought to be dependent on cell-mediated immunity and, to a lesser extent, on antibodies specific to membrane proteins (Velikovsky et al., 2002). As with immunity to other intracellular pathogens, immunity to B. abortus depends on antigen-specific T-cell–mediated activation of macrophages, which are the major effectors facilitating killing and inhibiting replication of Brucellae. The Th1 cell-induced cytokines, like IFN-γ, are important for the activation of macrophages and in resistance to in vivo and in vitro Brucella infections (Zhan et al., 1993). In the present study, the induction of T-cell immunity following DNA immunization was evaluated by measuring T-lymphocyte proliferation and IFN-γ production following in vitro stimulation with purified recombinant bp26 or TF. Both proteins induced elevated T-cell proliferative responses and elevated levels of IFN-γ. Such potent IFN-γ responses suggest that DNA immunization stimulates polarization toward a Th1-type phenotype, which is typically correlated with protection against intracellular pathogens, such as Brucella (Yingst and Hoover, 2003). Flow cytometry analysis revealed an increase in the percentages of CD4+ and CD8+ T cells in immunized bison compared with that of control bison. The percentage of γδ T cells was significantly lower in DNA-vaccinated bison. To our knowledge, this study is the first to evaluate γδ T-cells responses subsequent to DNA immunization. The γδ T cells represent a major lymphocyte subset in cattle and bison and can constitute up to 60–70% of the circulating T-cell pool in calves (Jutila et al., 2008). The γδ T cells have also been shown to respond and participate in host defense responses in a variety of pathogen-induced diseases, including brucellosis (Hayday, 2000). However, the decreased percentage of γδ T cells in vaccinated animals was surprising. The γδ T cells have long been documented to recognize unprocessed or nonpeptide antigens, independent of antigen processing and presentation through major histocompatibility complex molecules (Sireci et al., 1997). This means, unlike the CD4 or CD8 T cells, γδ T cells can recognize bacterial antigens directly in the absence of classic antigen presentation. In addition, γδ T cells express innate pathogen-recognition receptors and respond directly to pathogen-associated molecular patterns (Hedges et al., 2005; Jutila et al., 2008). Additional studies will elucidate the importance and role of γδ T cells subsequent to DNA vaccination. Studies also revealed DNA immunization can elicit modest levels of IgG and IgG1 anti-bp26 and anti-TF responses. This does not necessarily suggest that pCMVbp26 and pCMVTF vaccines are poor immunogens because many investigators have also demonstrated weak antibody responses following DNA vaccination of mice (Cassataro et al., 2005; Gonzalez-Smith et al., 2006). Rather, antibody responses are generally suboptimal following DNA vaccination, often requiring a protein boost or, alternatively, they may be in part attributed to the dampening effect by Th1 cell bias used by this immunization method (Commander et al., 2007). Interestingly, IgG1 subclass responses are induced in response to both bp26 and TF, which suggest preferential bias by the DNA vaccines toward Th2-type immune response. In summary, bison are responsive to DNA immunization, as demonstrated here using pCMVbp26 and pCMVTF DNA vaccines. Importantly, this method allows the stimulation of a potent Th1 cell bias evident by the elevated IFN-γ production following antigen restimulation of bison PBMCs. Although additional vaccine candidates need to be identified, this work demonstrates the feasibility of using a DNA vaccine approach to stimulate Brucella-specific immunity. ACKNOWLEDGMENTS We thank Mark Jutila, Department of Immunology and Infectious Diseases, Montana State University, for providing us the GD3.8 anti-bovine γδ T-cell mAb and Nancy Kommers for assistance in preparing this manuscript. This work is supported by grants from USDA (2009-34397-20133), Montana Agricultural Station, US Department of Agriculture Formula Funds, National Institutes of Health Grant (P20 RR020185), and an equipment grant from the M. J. Murdock Charitable Trust. Figure 1. Immunization with DNA vaccines encoding for bp26 and trigger factor (TF) elicits serum immunoglobulin G (IgG) antibody responses. Bison (four/group) were immunized intramuscularly with pcDNA3.1 vector (600 μg) or a combination of pCMVbp26 (300 μg) and pCMVTF (300 μg), together with 50 μg CpG as adjuvant on days 0, 14, and 28. Serum IgG anti-bp26 and anti-TF titers were measured by enzyme-linked immunosorbent assay on wk 8, 10, and 12 after primary immunization. The IgG, IgG1 and IgG2 (A) anti-bp26 and (B) anti-TF endpoint titers peaked between wk 8 and 12 after primary immunization. Control bison immunized with control vector, pcDNA3.1 plus CpG failed to elicit antibodies to bp26 and TF. Asterisks represent significant differences in IgG responses versus pcDNA3.1 vector-immunized bison (P≤0.05). Figure 2. Immunization with DNA vaccines encoding for bp26 and trigger factor (TF) enhances antigen-specific bison T cell proliferative responses. Bison were immunized, as described in Fig. 1. T-cell proliferation responses were measured at 8, 10, and 12 wk after the primary immunization. Peripheral blood mononuclear cells (PBMCs) from each bison (5×105 cells/well) were prepared and stimulated in vitro with purified 20 μg/ml of recombinant bp26, TF, both, 1 mg/ml ovalbumin (OVA) or 5 μg/ml of concanavalin A (Con-A). Results are expressed as mean±SE of triplicate cultures of cells obtained from each bison. ** P≤0.001 and * P≤0.05 represent the significant differences within the sampling time in proliferative T-cell responses between bison immunized with pcDNA3.1 vector versus bison immunized with bp26 and TF DNA vaccines. Figure 3. Immunization with bp26 and trigger factor (TF) DNA vaccines show enhanced interferon (IFN)-γ production by bison peripheral blood mononuclear cells (PBMCs). At 8, 10, and 12 wk after initial immunization, PBMCs from bison immunized with bp26 and TF DNA vaccines or with pcDNA3.1 vector were incubated for 72 hr in the presence or absence of 20 γg/ml (A) bp26 or (B) TF and were evaluated for IFN-γ production by cytokine enzyme-linked immunosorbent assay. Results are expressed as net mean±SE IFN-γ production (production in wells containing bp26 or TF minus production in wells without antigen) obtained from each bison. * P≤0.05 represents significant differences within the sampling time in IFN-γ production between bison immunized with pcDNA3.1 versus those immunized with pCMVbp26 plus pCMVTF. Table 1. Flow cytometric analysis of T-cell responses to recombinant bp26 and trigger factor (TF) after DNA immunization. Peripheral blood mononuclear cells (PBMCs) from bison immunized intramuscularly with pcDNA3.1, or with a combination of pCMVbp26 plus pCMVTF, in presence of CpG were labeled with monoclonal antibodies specific for CD4+ T, CD8+ T, and TCRγδ+ cells and were analyzed by flow cytometry. Time postimmunization and immunization group % PBMC subpopulations CD4+ T cells CD8+ T cells TCRγδ+ cells     8 wk     Control 22.86±1.4 9.35±0.3   33.0±2.5     bp26+TF 33.86±0.3*    26±1.5* 22.03±0.98*     10 wk     Control 21.78±2.5   8.6±1.04   43.1±3.6     bp26+TF   32.1±0.23* 19.7±0.91*   41.2±1.1     12 wk     Control   24.8±0.50   8.2±1.39   39.1±1.1     p26+TF   39.8±3.9* 18.1±2.1*   23.2±4.4** * P≤0.001 and ** P≤0.05 indicate a significant difference within the sampling time in percentages of T-cell subsets in bison immunized with pcDNA3.1 versus the combination of the two DNA vaccines. LITERATURE CITED Al-Mariri A , Tibor A , Mertens P , De Bolle X , Michael p. , Godfroid J , Walravens K , and Letesson JJ . 2001. Induction of immune response in BALB/c mice with a DNA vaccine encoding bacterioferritin or P39 of Brucella spp. Infection and Immunity 69 : 6264–6270.11553569 Bae JE , Schurig GG , and Toth TE . 2002. Mice immune responses to Brucella abortus heat shock proteins: Use of baculovirus recombinantexpressing whole insect cells, purified Brucella abortus recombinant proteins, and a vaccinia virus recombinant as immunogens. Veterinary Microbiology 88 : 189–202.12135637 Baloglu S , Toth TE , Schurig GG , Sriranganathan N , and Boyle SM . 2000. Humoral immune response of BALB/c mice to a vaccinia virus recombinant expressing Brucella abortus GroEL does not correlate with protection against a B. abortus challenge. Veterinary Microbiology 76 : 193–199.10946149 Cassataro J , Velikovsky CA , De La Barrera S , Estein SM , Bruno L , Bowden R , Pasquevich KA , Fossati CA , and Giambartolomei GH . 2005. A DNA vaccine coding for the Brucella outer membrane protein 31 confers protection against B. melitensis and B. ovis infection by eliciting a specific cytotoxic response. Infection and Immunity 73 : 6537–6546.16177328 Cloeckaert A , Vizcaino N , Paquet JY , Bowden RA , and Elzer PH . 2002. Major outer membrane proteins of Brucella spp.: Past, present and future. Veterinary Microbiology 90 : 229–247.12414146 Commander NJ , Stephen SA , Wren BW , and Macmillan AP . 2007. The identification of two protective DNA vaccines from a panel of five plasmid constructs encoding Brucella melitensis 16M genes. Vaccine 25 : 43–54.17049676 Davis DS , and Elzer PH . 2002. Brucella vaccines in wildlife. Veterinary Microbiology 90 : 533–544.12414169 ———, Templeton JW , Ficht TA , Huber JD , Angus RD , and Adams LG . 1991. Brucella abortus in bison, II: Evaluation of strain 19 vaccination of pregnant cows. Journal of Wildlife Diseases 27 : 258–264.1906114 Edmonds MD , Cloeckaert A , Booth NJ , Fulton WT , Haqius SD , Walker JV , and Elzer PH . 2001. Attenuation of a Brucella abortus mutant lacking a major 25 kDa outer membrane protein in cattle. American Journal of Veterinary Research 62 : 1461–1466.11560278 ———, ———, and Elzer PH 2002a. Brucella species lacking the major outer membrane protein omp25 are attenuated in mice and protect against Brucella melitensis and Brucella ovis. Veterinary Microbiology 88 : 205–221.12151196 ———, ———, Hagius SD , Samartino LE , Fulton WT , Walker JV , Enright FM , Booth NJ , and Elzer PH . 2002b. Pathogenicity and protective activity in pregnant goats of a Brucella melitensis Δomp25 deletion mutant. Research in Veterinary Science 72 : 235–239.12076120 Godfroid JA , Cloeckaert A , Liautard JP , Kohler S , Fretin D , Walravens K , GarinBastuji B , and Letesson JJ . 2005. From the discovery of the Malta fever’s agent to the discovery of a marine mammal reservoir, brucellosis has continuously been a re-emerging zoonosis. Veterinary Research 36 : 313–326.15845228 Gonzalez-Smith A , Vemulapalli R Andrews E , and O~ate A . 2006. Evaluation of Brucella abortus DNA vaccine by expression of Cu-Zn superoxide dismutase antigen fused to IL-2. Immunobiology 211 : 65–74.16446171 Gurunathan S , Wu CY , Freidag BL , and Seder RA . 2000. DNA vaccines: A key for inducing long-term cellular immunity. Current Opinion in Immunology 12 : 442–447.10899026 Hayday AC 2000. γδ cells: A right time and a right place for a conserved third way of protection. Annual Reviews in Immunology 18 : 957–1026. Hedges JF , Lubick KJ , and Jutila MA . 2005. γδ T cells respond directly to pathogen associated molecular patterns. Journal of Immunology 174 : 6045–6053. Jubier-Maurin V , Boigegrain RA , Cloeckaert A , Gross A , Alvarez-Martinez MT , Terraza A , Liautard J , Köhler S , Rouot B , Dornand J , and Liautard JP . 2001. Major outer membrane protein Omp25 of Brucella suis is involved in inhibition of tumor necrosis factor-α production during infection of human macrophages. Infection and Immunity 69 : 4823–4830.11447156 Jutila MA , Holderness J , Graff JC , and Hedges JF . 2008. Antigen-independent priming: A transitional response of bovine γδ T-cells to infection. Animal Health Research Review 9 : 47–57. Ko J , and Splitter GA . 2003. Molecular host-pathogen interaction in brucellosis: Current understanding and future approaches to vaccine development for mice and humans. Clinical Microbiology Reviews 16 : 65–78.12525425 Liu MA 2003. DNA vaccines: A review. Journal of internal Medicine 253 : 402–410.12653868 Nelson DD , Davis WC , Brown WC , Li H , O’toole D , and Oaks JL . 2010. CD8+/perforin+/WC1− γδ T cells, not CD8+ αβ T cells, infiltrate vasculitis lesions of American bison (Bison bison) with experimental sheep-associated malignant catarrhal fever. Veterinary immunology and Immunopathology 136 : 284–291.20413164 Olsen SC , and Stoffregen WS . 2005. Essential role of vaccines in brucellosis control and eradication programs for livestock. Expert Review of Vaccines 4 : 915–928.16372886 ———, and Tatum F 2010. Bovine brucellosis. Veterinary Clinics of North America: Food Animal Practice 26 : 15–27.20117540 ———, Boyle SM , Schurig GG , and Sriranganathan NN . 2009. Immune responses and protection against experimental challenge after vaccination of bison with Brucella abortus strain RB51 or RB51 overexpressing superoxide dismutase and glycosyltransferase genes. Clinical and Vaccine immunology 4 : 535–540. ———, Cheville NF , Kunkle RA , Palmer MV , and Jensen AE . 1997. Bacterial survival, lymph node pathology, and serological responses of bison (Bison bison) vaccinated with Brucella abortus strain RB51 or strain 19. Journal of Wildlife Diseases 33 : 146–151.9027704 ———, Jensen AE , Stoffregen WC , and Palmer MV . 2003. Efficacy of calfhood vaccination with Brucella abortus strain RB51 in protecting bison against brucellosis. Research in Veterinary Science 74 : 17–22.12507562 Oñate AA , Cespedes S , Cabrera A , Rivers R , Gonzalez A , Munoz C , Folch H , and Andrews E . 2003. A DNA vaccine encoding Cu, Zn superoxide dismutase of Brucella abortus induces protective immunity in BALB/c mice. Infection and Immunity 71 : 4857–4861.12933826 ———, Vemulapalli E , Andrews E , Schurig GG , Boyle S , and Folch H . 1999. Vaccination with live Escherichia coli expressing Brucella abortus Cu/Zn superoxide dismutase protects mice against virulent B. abortus. Infection and Immunity 67 : 986–988.9916121 Pasquevich KA , Estein SM , Garcia Samartino C , Zwerdling A , Coria LM , Barrionuevo P , Fossati CA , Giambartolomei GH , and Cassataro J . 2009. Immunization with recombinant Brucella species outer membrane protein omp16 or omp19 in adjuvant induces specific CD4+ and CD8+ T cells as well as systemic and oral protection against Brucella abortus infection. Infection and Immunity 77 : 436–445.18981242 Perkins SD , Smither SJ , and Atkins HS . 2010. Towards a Brucella vaccine for humans. FEMS Microbiology Reviews 34 : 379–394.20180858 Rhyan JC , and Spraker TR . 2010. Emergence of diseases from wildlife reservoirs. Veterinary Pathology 47 : 34–39.20080482 Robinson HL , and Torres CA . 1997. DNA vaccines. Immunology 9 : 271–283. Schurig GG , Sriranganathan N , and Corbel MJ . 2002. Brucellosis vaccines: Past, present and future. Veterinary Microbiology 90 : 479–496.12414166 Seleem MN , Boyle SM , and Sriranganathan N . 2010. Brucellosis: Re-emerging zoonosis. Veterinary Microbiology 140 : 392–398.19604656 Simon S , Li H , O’toole D , Crawford TB , and Oaks JL . 2003. The vascular lesions of a cow and bison with sheep-associated malignant catarrhal fever contain ovine herpesvirus 2-infected CD8+ T lymphocytes. Journal of Virology 84: 2009–2013. Sireci G , Champagne E , Fournié JJ , Dieli F , and Salerno A . 1997. Patterns of phosphoantigen stimulation of human Vγ9/Vδ2 T cell clones include Th0 cytokines. Human Immunology 58 : 70–82.9475336 Smyth AJ , Welsh MD , Girvin RM , and Pollock JM . 2001. In vitro responsiveness of γδ T cell from Mycobacterium bovis-infected cattle to mycobacterial antigens: Predominant involvement of WC1+ cells. Infection and Immunity 69 : 89–96.11119493 Van Campen H , and Rhyan J . 2010. The role of wildlife in diseases of cattle. Veterinary Clinics of North America: Food Animal Practice 26 : 147–161.20117548 Velikovsky CA , Cassataro J , Giambartolomei GH , Goldbaum FA , Estein S , Bowden RA , Bruno L , Fossati CA , and Spitz M . 2002. A DNA vaccine encoding luminaze synthase from Brucella abortus induces protective immunity in BALB/c mice. Infection and Immunity 70 : 2507–2511.11953389 Wilson E , Walcheck B , Davis WC , and Jutila MA . 1998. Preferential tissue localization of bovine γδ T cell subsets defined by anti-T cell receptor for antigen antibodies. Immunology Letters 64 : 39–44.9865600 Yang X , Hudson M , Walters N , Bargatze RF , and Pascual DW . 2005. Selection of protective epitopes for Brucella melitensis by DNA vaccination. Infection and Immunity 73 : 7297–7303.16239526 ———, Walters N , Robison A , Trunkle T , and Pascual DW . 2007. Nasal immunization with recombinant Brucella melitensis bp26 and trigger factor with cholera toxin reduces B. melitensis colonization. Vaccine 25 : 2261–2268.17239499 Yingst S , and Hoover DL . 2003. T cell immunity to brucellosis. Critical Reviews in Microbiology 4 : 313–331. Zhan Y , Kelso A , and Cheers C . 1993. Endogenous gamma interferon mediates resistance to Brucella abortus infection. Infection and Immunity 61 : 4899–4901.8406893

PMC009xxxxxx/PMC9828956.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9009458 1183 Semin Immunol Semin Immunol Seminars in immunology 1044-5323 1096-3618 36306662 9828956 10.1016/j.smim.2022.101656 NIHMS1859789 Article The role of unconventional T cells in maintaining tissue homeostasis LeBlanc Gabrielle 12 Kreissl Felix 12 Melamed Jonathan 12 Sobel Adam L. 12 Constantinides Michael G. 1* 1 Department of Immunology & Microbiology, Scripps Research, La Jolla, CA 92037, USA 2 These authors contributed equally * Correspondence: constantinides@scripps.edu 4 1 2023 11 2022 25 10 2022 09 1 2023 61-64 101656101656 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. pmcIntroduction In addition to providing antimicrobial defense, the immune system also maintains homeostasis, in part through tissue lymphocytes. While conventional CD4+ and CD8+ T cells recognize peptides presented by polymorphic major histocompatibility complex (MHC) proteins, tissues contain an abundance of T cells specific for lipids, metabolites, and modified peptides [1]. These unconventional T cells are restricted by monomorphic MHC class Ib (MHC-Ib) molecules and include CD1d-restricted natural killer T (NKT) cells, MR1-restricted mucosal-associated invariant T (MAIT) cells, and CD1a/c/d-restricted γδ T cells [1]. Invariant NKT (iNKT) cells, MAIT cells, and subsets of γδ T cells express semi-invariant T cell receptors (TCRs) that limit their antigenic range analogously to innate immune receptors [1]. Selection of innate-like T cells on hematopoietic cells imbues them with effector functions prior to thymic egress [2, 3], including rapid cytokine release and expression of chemokine receptors and integrins, which enable these cells to accumulate within tissues early in life [4–6]. Here we summarize how unconventional T cells are uniquely positioned to maintain tissue homeostasis from birth and discuss how exogenous factors impact their development and function, causing deleterious effects to tissue physiology. Dependence of unconventional T cells on the microbiota From early life, the presence of intestinal microbiota is critical for iNKT cell development. Conventionally housed mice delivered via cesarean section have more splenic iNKT cells, caused by a microbiota-dependent increase in Il15 and Il4 gene expression [7]. Conversely, in the absence of commensal microbes, fewer iNKT cells accumulate in the thymus, spleen, and liver, and those that do are hyporesponsive [8, 9]. However, germ-free (GF) mice harbor more iNKT cells in their colonic lamina propria (LP) and lungs, which exacerbate airway responses and colonic inflammation (Figure 1) [10]. Without commensals, a region 5′ of the Cxcl16 gene undergoes CpG hypermethylation, which is associated with increased CXCL16 expression and recruitment of iNKT cells to the colonic mucosa. Colonization of GF neonatal mice with intestinal microbes restored normal levels of iNKT cells, but the same effect was not observed when colonized after weaning [10]. In addition to altering the migration of iNKT cells, the microbiota also modulates their development [11]. Within the thymus, iNKT cells are positively selected on glycosphingolipids presented by CD1d [12]. Sphingomonas bacteria contain cell wall glycosphingolipids that stimulate iNKT cells in a CD1d-dependent manner [13]. Colonization of GF mice with Sphingomonas yanoikuyae was sufficient to restore iNKT expression of the activation marker CD69, which did not occur in mice colonized with Escherichia coli that lack agonistic glycosphingolipids [9]. Another intestinal commensal, the bacterium Bacteroides fragilis, regulates colonic iNKT cell abundance through inhibitory sphingolipids that bind to CD1d and prevent presentation of endogenous ligands to iNKT cells (Figure 1) [14]. Monocolonization of neonatal GF mice with B. fragilis impedes iNKT proliferation within the colon, but not the lungs, demonstrating that bacterial sphingolipids inhibit local iNKT activation [14]. The diminished accumulation of iNKT cells protects these animals from oxazolone-induced colitis later in life, which is exacerbated by iNKT-derived cytokines [14]. However, delayed colonization of GF mice as adults is insufficient to prevent iNKT cell proliferation and the resulting increase in disease severity, indicating that early-life microbial interactions are essential for maintaining tissue homeostasis [14]. Early exposure to microbiota also regulates the development of other unconventional T cells [11]. In GF mice, most thymic MAIT cells are arrested at the immature CD24+ CD44− developmental stage, which precedes expression of promyelocytic leukemia zinc finger (PLZF) and the chemokine receptors and integrins regulated by this transcription factor [15, 16]. Consequently, MAIT cells are unable to traffic to tissues, resulting in a precipitous loss in the periphery (Figure 1) [15–18]. Colonization with riboflavin-synthesizing commensals during the first weeks of life restores MAIT cells, as does administration of the riboflavin derivative 5-(2-oxopropylideneamino)-6-d-ribitylaminouracil (5-OP-RU), indicating that MAIT cell development is dependent on TCR signaling [17]. While commensals and riboflavin derivatives are sufficient to rescue thymic development of mature CD24− CD44+ MAIT cells in adult GF mice, neither restores peripheral MAIT cells [16, 17], imprinting tissues with a lower abundance of these lymphocytes. Although it remains unclear why early-life microbial exposure is necessary for the development of peripheral MAIT cells, recent work has revealed competition between unconventional T cells for niche occupancy [19], suggesting that expansion of iNKT cells and γδ T cells following birth may limit subsequent accumulation of MAIT cells within tissues. While MAIT cells accumulate in murine tissues after birth [17], T cells expressing the invariant Vα7.2 TCR gene segment and CD161 were initially reported in second trimester human fetal tissues [5]. However, the proportion of CD161+ Vα7.2+ T cells that recognize 5-OP-RU presented by MR1 is only ~3% in cord blood and less than 20% in neonatal blood [20, 21], indicating that MR1-restricted MAIT cells are scarce at birth. The few MAIT cells present in human neonates lack cytokine functionality and expression of the activation marker CD45RO [20, 21]. Human MAIT cells acquire effector characteristics concomitantly with a dramatic increase in their abundance during the first weeks of life [20, 21]. This follows the initial colonization of the newborn intestine with facultative anaerobes [22], such as members of the Enterobacteriaceae family, many of which synthesize riboflavin [17]. While MAIT cell abundance varies widely between neonates from different geographic regions [21], frequencies between mono- and dizygotic twins are highly correlated [20], indicating that environmental factors in early life determine human MAIT cell frequencies. Children that receive a hematopoietic stem cell transplant recover their conventional T and B cells within a year yet exhibit 10-fold fewer MAIT cells [20], suggesting that human MAIT cell abundance may also be imprinted by microbial interactions in early life. The dependence of γδ T cells on microbial signals varies by anatomical site, even for similar subsets. In the lungs, meninges, and uterus, where the majority of tissue-resident γδ T cells are Vγ6+ (according to the nomenclature of Heilig and Tonegawa [23]), GF and specific-pathogen-free (SPF) mice have comparable frequencies of γδ T cells [24–27]. However, Vγ6+ T cells also predominate in the oral mucosa, yet GF and antibiotic-treated mice exhibit fewer IL-17-producing γδ T (γδT17) cells compared to specific-pathogen-free (SPF) animals, demonstrating that the distinct microbial communities colonizing these tissues have disparate effects on the development and function of Vγ6+ T cells [28, 29]. Characterization of intraepithelial lymphocytes (IELs) in GF mice revealed development of γδ T cells is comparable to SPF animals [30]. In contrast to IFN-γ+ γδ IELs, γδT17 cells within the lamina propria (LP) are decreased in GF mice or antibiotic treated mice [25, 31]. TCR signal transduction from the guanine nucleotide exchange factor VAV1 is essential for the production of these LP γδ T cells, suggesting TCR engagement is necessary for development (Figure 1) [25]. Whether specific commensals are responsible for inducing these γδT17 cells, and whether their induction is entirely TCR-dependent remains unclear [24]. While germ-free mice have normal frequencies of dendritic epidermal T cells (DETCs), they exhibit fewer dermal γδT17 cells [32, 33], indicating that the microbiota is necessary for these lymphocytes. Species of the Corynebacterium genus colonize human skin and, when topically applied to mice, stimulate dermal Vγ4+ T cells, but not dermal Vγ6+ T cells [34]. Since both populations of dermal γδT17 cells express IL-1R and IL-23R [35], the differential responses are due to recognition of distinct microbial antigens. While these have yet to be determined, dendritic cells (DCs) loaded with Corynebacterium cell envelope extracts stimulate Vγ4+ T cells, while DCs pulsed with strains that lack expression of mycolic acids fail to do so [34], suggesting that mycolic acids or the glycolipids they support are recognized. Thus, microbial antigens can differentially stimulate functionally similar subsets of γδ T cells. In humans, the ontogeny of γδ T cells precedes αβ T cells, with embryonic γδ TCR gene rearrangement detected by the eighth week of gestation and canonical subsets detected extrathymically during fetal development [36, 37]. After birth, members of the skin and gut microbiota have been implicated as sources of phosphoantigens that shape the T cell compartment during childhood [38, 39]. TCR sequencing analysis of γ and δ chains of children in European and African cohorts showed comparable polyclonal expansion in early life [40]. Notably, there was greater diversity in the δ chain from a European child hospitalized with a febrile event, correlating this repertoire expansion with microbial exposure [40]. Thus, evidence from both humans and animals indicate that early-life microbial interactions play a significant role in the development of γδ T cells [41]. Antibiotics affect the development and function of unconventional T cells As the role of commensal microbes becomes increasingly clear in the development of unconventional T cells, drugs that impact the microbiota would be expected affect these lymphocytes, with consequences for host physiology. Indeed, early-life exposure to antibiotics correlates with a variety of chronic pathologies, including atopic dermatitis, psoriasis, allergies, obesity, and neurodevelopmental disorders [42–44]. Administration of broad-spectrum antibiotics to neonatal mice reduced the abundance of thymic PLZF+ innate-like T cells, which persisted into adulthood, rendering the animals more susceptible to dextran sulfate sodium (DSS)-induced colitis [45]. Neonatal antibiotic treatment also increased the susceptibility of adult mice to an experimental model of psoriasis, which was attributed to an increase in IL-22-producing γδ T cells [46, 47]. Notably, cohousing antibiotic-treated mice with untreated controls did not completely ameliorate pathology, suggesting the effects of dysbiosis were persistent despite subsequent acquisition of diverse microbiota. Alternatively, administration of erythromycin increased abundance of Clostridium families XIVa and XIVab, Bacteroidetes, and Bifidobacterium and decreased segmented filamentous bacteria, resulting in an increase in regulatory IL-10-secreting γδ T cells and a reduction in contact sensitivity [48]. These contrary findings suggest that antibiotic induced microbiome-immune axis perturbations are context-dependent with respect to microbiome composition, antibiotic selectivity, and dosing regimen. While exposure to antibiotics in early life may impact host susceptibility to disease, antibiotic administration during advanced pathology may temporarily ameliorate diseases that are sustained by microbial antigens (Figure 2). Antibiotics reduce IL-17 and IL-22 production from γδ T cells, which reduces inflammation in an experimental model of psoriasis [47]. Two subsequent studies have supported the treatment of psoriasis symptoms via antibiotic-induced reduction in IL-17A secreting skin-resident γδ T cells [49, 50]. Similarly, in a murine model of cystic fibrosis (CF), streptomycin treatment reduced intestinal bacterial overgrowth and intestinal Lactobacilli, which correlated with decreased CF-specific γδT17 cells and ameliorated airway hyperresponsiveness [51]. In a murine lung cancer model, antibiotics were found to suppress adenocarcinoma tumorigenesis by arresting an inflammatory feedback mechanism that was sustained by microbial antigens and IL-17 produced by γδ T cells [52]. Broad spectrum antibiotics have been shown to increase colonic iNKT cells in adult mice. iNKT cells of antibiotic-treated mice were predisposed towards an inflammatory state following DSS-induced colitis or transfer of dysbiotic donor microbiota (Figure 2) [53]. However, individual antibiotics have differential effects on iNKT cells. Mice that received microbiota from streptomycin- or vancomycin-treated donors developed worse colitis following DSS treatment, while mice that received microbiota from donors administered metronidazole exhibited less pathology, which correlated with an enrichment of Lactobacilli and increased IL-10 secretion by iNKT cells [54]. Thus, the impact of antibiotics on iNKT cells is highly dependent on which microbial commensals are targeted. While MAIT cells require microbial metabolites for their development, whether individual antibiotics differentially regulate their development or function has not been established. Impact of non-antibiotic drugs on unconventional T cells Non-antibiotic drugs and their metabolites have recently been shown to bind the MHC-Ib molecules recognized by unconventional T cells, resulting in a competition between these compounds and their archetypal antigens [55]. Small molecules with drug-like characteristics are capable of binding CD1d and activating type II NKT cells, which also recognize lipid antigens but lack the invariant TCRα chain expressed by type I iNKT cells [56]. Non-lipidic small molecules resembling sulfa drugs were shown to be antigenic to human type II, CD1d restricted Vα24− T cells (Figure 2) [57]. Synthetic analogs of the most antigenic compound identified, phenyl 2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonate (PPBF), are not recognized by iNKT cells, yet compete with the lipid agonist α-GalCer in a dose dependent manner [57]. Whether PPBF stimulates these type II NKTs in a TCR-CD1d dependent manner was inconclusive due to the inability to stain the clone with CD1d-PPBF tetramers and has been suggested that its drug like structure may activate these NKTs via direct pharmacological action [58]. A recent study characterized the polyclonal population of type II NKT cells capable of TCR dependent activation of CD1d complexed with exogenous, sulfa drug-like, pentamethylbenzofuransulfonate (PBF) small molecules and an endogenous self-lipid [58]. Staining CD1d-restricted NKT cells from donor blood samples with a PBF-treated tetramer identified populations expressing both αβ and γδ TCRs [58]. Both PPBF and PBF compounds share structural moieties with commonly used sulfa drugs known to generate hypersensitivity reactions in humans such as sulfadiazine, sulfasalazine, and celecoxib [59, 60]. These findings underscore the importance of determining the off-target effects of PBF-like sulfa drugs on CD1 molecules and establishing whether type II NKT cells contribute to hypersensitivity [58]. T cells within human skin were recently shown to recognize CD1a without lipid and hydrophobic small molecules within cosmetics could displace CD1a-bound lipids and cause contact dermatitis, demonstrating that CD1 molecules can activate T cells independently of lipids [61, 62]. Whether endogenous or microbial compounds could analogously alter the development or functions of NKT cells remains to be determined. The ability to exogenously prime MAIT cells remains of great therapeutic interest because of their role as a first line of defense in an array of diseases [63–65]. The ability of MR1 to bind non-antibiotic drugs was confirmed after a virtual screening campaign identified drugs including diclofenac and salicylate metabolites that could alter MR1 expression and modulate MAIT cell activity [66]. However, only one drug analogue, 3-formyl salicylic acid (3-F-SA), has been tested for its ability to modulate MAIT cell activity in vivo [66]. 3-F-SA was administered concomitantly with 5-OP-RU and an intestinal pathogen deficient in riboflavin synthesis to mimic an infectious challenge and was found to dampen MAIT cell activation by competing for MR1 binding (Figure 2) [66]. Several other MR1 binding drugs shown to be non-agonists, or capable of downregulating MR1 surface expression, have not yet been tested for their capacity to inhibit MAIT cell activation in vivo [66]. Additionally, whether these drugs can outcompete microbially derived metabolites and interfere with MAIT cell development remains to be determined. Diet alters unconventional T cells Changes in diet can alter unconventional T cell function, both directly by delivery of antigens, and indirectly via remodeling of microbiota composition and metabolic output. A Western diet (WD), comprised of excess fat, protein, and carbohydrates, correlates with several chronic diseases, including obesity, metabolic disease, type II diabetes, and cardiovascular disease, coinciding with a general increase in systemic inflammation [67]. WD was recently shown to induce an inflammatory profile of γδ T cells, increasing IL17 and IL-22 production, which contributed to susceptibility to both psoriatic arthritis and joint inflammation [68, 69]. A ketogenic diet (KD) has also been shown to remodel the γδ T cell profile. Mice fed a KD for seven days were protected from lethal influenza virus inoculation, which coincided with an increase in lung-resident γδT17 cells [70]. Interestingly, while WD also increased lung-resident γδT17 cells, this was not protective against influenza infection, suggesting a ketogenesis-specific transcriptional remodeling. A subsequent study evaluated the impact of ketogenic diet on γδ T cells in adipose tissue [70, 71]. While long term KD in mice impaired metabolic health, promoted adipose tissue inflammation and decreased γδ T cells in adipose tissue, short term ketogenesis expanded adipose-resident CD44+ CD27− γδ T cells that exhibited increased transcription of adipose tissue remodeling and repair genes. γδ T cells have been identified as participants in intestinal nutrient sensing. In a comparison between high carbohydrate and high protein diets in mice, carbohydrates induced γδ T cells to suppress IL-23 production by type-3 innate lymphoid cells, thus promoting intestinal epithelial transcription towards carbohydrate utilization [72]. Microbial synthesis of α-GalCer by colonic Bacteroidetes is dependent upon dietary intake of amino acids [73]. As such, presence of these antigens in the colon is subject to modulation of the microbiome composition and metabolic output, including diet, and changes to these variables can remodel iNKT activation state. WD has been shown to significantly decrease α-GalCer in the cecum of mice [74]. Indirect activation of iNKT cells via alternative metabolites has also been observed. Aryl hydrocarbon receptor (AhR) ligands, namely indole derivatives such as oxazole acquired from dietary, microbial, and industrial sources, can stimulate AhR on intestinal epithelial cells and modulate iNKT activation. Oxazoles limit CD1d-restricted IL-10 secretion, which increased inflammatory activation of colonic iNKTs and promote susceptibility to experimental colitis [75]. Conversely, inhibition of iNKT cells has been observed following increased intake of dietary fiber. Butyrate produced by microbial fermentation of dietary fiber directly inhibits iNKT cell activation and proinflammatory cytokine production via inhibition of class I histone deacetylase (HDAC). Preincubation of iNKT cells with butyrate prior to adoptive transfer into iNKT cell-deficient Jα18−/− mice prevented antibody-induced arthritis [76]. Mice fed a high fat diet (HFD) exhibit decreased frequency of MAIT cells in blood, ileum, and adipose tissue [77]. Remaining MAIT cells in epididymal adipose tissue and ileum exhibited an activated proinflammatory phenotype characterized by increase in IL-17A and CD44 expression, suggesting potential diet-induced MAIT cell contribution to obesity-related chronic inflammation and metabolic disease. Microbiota from the cecum of HFD mice exhibited decreased riboflavin synthesis capacity. MAIT cell-deficient Mr1−/− mice on HFD exhibit increased glucose tolerance compared to wild type controls, suggesting a role for MAIT cells in the development of type II diabetes in the context of obesity. Indeed, MAIT cells producing IL-17, a cytokine implicated in insulin resistance, are elevated in humans with obesity and type II diabetes [78, 79]. In these patients, MAIT cell abundance increased in adipose tissue and decreased in circulation, with the remaining circulatory MAIT cells exhibiting enhanced potential to produce IL-17. Notably, bariatric surgery restored circulating MAIT cells and reduced their cytokine secretion within three months of surgery [79], strongly corroborating the evidence that MAIT cells are subject to dietary and metabolic modulation in a manner relevant to disease. The role of unconventional T cells in tissue repair and maintenance of homeostasis In addition to their recognition of microbial products, unconventional T cells respond to molecules induced by tissue damage. During cutaneous injury, keratinocytes surrounding the wound upregulate expression of butyrophilin-like proteins encoded by Skint genes [80], which are recognized by murine Vγ5+ Vδ1+ DETCs in a TCR-dependent manner [81]. Upon activation, DETCs along the wound edge lose their characteristic dendritic morphology and express fibroblast growth factors and insulin like growth factor 1 (IGF-1), which reduce keratinocyte apoptosis and promote proliferation and migration (Figure 3) [82, 83]. Reduced expression of Skint genes in the skin of aged mice correlates with fewer DETCs upon injury and delayed wound closure [80]. In addition to the epidermal DETCs, murine CCR6+ Vγ4+ T cells are recruited to the dermis following injury and, together with dermal-resident Vγ6+ T cells, contribute to cutaneous repair by secreting fibroblast growth factors, including FGF-9, which induces hair follicle neogenesis (Figure 3) [84, 85]. Both epidermal and dermal γδ T cells also promote cutaneous repair through IL-17A, which causes keratinocyte differentiation and proliferation [86, 87]. Release of IL-17A also enhances epithelial cell glycolysis through activation of hypoxia-inducible factor 1 alpha (HIF-1α), fueling the migration of these cells to the wound edge (Figure 3) [88]. IL-17A can be induced by activation of TRPA1 sensory neurons, which promote IL-23 production by dermal dendritic cells, thus stimulating γδ T cells [87]. However, an epidermal subset of γδ T cells analogous to DETCs has not been described in humans and γδ T cells sparsely populate the human dermis, suggesting that these cells may be transiently recruited following injury or that repair of human skin relies on alternate cell types [85, 89]. Despite the paucity of γδ T cells in human skin, these unconventional T cells present in most human and murine tissues, where they also promote repair. Following bone injury, murine Vγ6+ T cells rapidly accumulate within the damaged tissue and produce IL-17A, which stimulates the proliferation of mesenchymal progenitor cells and their differentiation into osteoblasts [90]. IL-17A released by Vγ6+ T cells also enhances repair of skeletal muscle by inducing proliferation of muscle stem cells and promoting expression of transcripts associated with regeneration rather than inflammation [91]. Interestingly, accumulation of IL-17A+ Vγ6+ T cells within injured muscle requires the microbiota and does not occur in GF mice or animals treated with broad-spectrum antibiotics, resulting in the formation of smaller muscle fibers [91]. While production of IL-17A by γδ T cells enhances tissue repair, their release of IFN-γ delays recovery. Following a spinal cord injury, Vγ4+ T cells are recruited to the wound site, where their production of IFN-γ activates macrophages, resulting in elevated levels of proinflammatory cytokines [92]. Absence of γδ T cells in Tcrd−/− mice accelerates the recovery of hind limb movement, which could be replicated by anti-Vγ4 antibody-mediated depletion, indicating that the role of γδ T cells in tissue repair is context dependent [92]. γδ T cells also contribute to the restoration of tissue homeostasis following a severe antimicrobial response. During a neonatal influenza infection, γδ T cells rapidly accumulate within the lungs where they produce IL-17A, which causes lung epithelial cells to release IL-33 [93]. This cytokine subsequently enhances production of amphiregulin by group 2 innate lymphoid cells (ILC2s) and regulatory T cells (Tregs) [93]. In the absence of γδ T cells, lung amphiregulin levels are reduced, resulting in decreased survival, demonstrating the importance of γδ T cells in this reparative cascade [93]. Additionally, γδ T cells within the oral gingiva, but not the intestinal epithelium or spleen, express tissue repair genes, including amphiregulin, which minimizes periodontal bone loss in mice [94]. Together, these findings demonstrate that γδ T cells utilize multiple complementary mechanisms to promote repair of tissues across the body, leading to the restoration of homeostasis. While murine DETCs are restricted to the epidermis and Vγ4+ and Vγ6+ T cells reside in the dermis with conventional αβ T cells [95], cutaneous MAIT cells are positioned along the basement membrane that separates the dermis and epidermis [17]. MAIT cells have also been observed near the basement membrane of the human colon [96], suggesting that this unique localization may be characteristic of MAIT cells within epithelial tissues. Since the basement membrane serves as a scaffold for the migration of progenitor cells during wound healing [97], MAIT cells are positioned to regulate the repair process. Indeed, both human and murine MAIT cells express transcriptional signatures associated with tissue repair [17, 96, 98]. TCR signaling in the absence of stimulatory cytokines promotes expression of growth factors, including FGF-9, vascular endothelial growth factor (VEGF), and platelet-derived growth factor (PDGF) [96, 98]. However, activation with the cytokines IL-12, IL-15, IL-18, and tumor necrosis factor-like protein 1A (TL1A) resulted in expression of effector cytokines and concomitant downregulation of some tissue repair factors [96]. Topical application of the riboflavin-synthesizing skin commensal Staphylococcus epidermidis provides MAIT cells with TCR stimulation in the absence of inflammation, resulting in expression of genes associated with angiogenesis and tissue repair, including angiogenin, insulin-like growth factor 1 (IGF-1), and hepatocyte growth factor (HGF) (Figure 3) [17]. While soluble factors from stimulated MAIT cells promote wound healing of intestinal epithelial cells in vitro, abrogation of TCR signaling with an anti-MR1 antibody inhibits repair [96]. Topical application of the stimulatory riboflavin derivative 5-OP-RU is also sufficient to increase re-epithelialization of cutaneous excision wounds in mice [17], demonstrating that TCR stimulation is both necessary and sufficient for the tissue repair program of MAIT cells. Additionally, MAIT-cell deficient mice exhibit impaired wound healing, indicating that MAIT cells are integral to the repair and maintenance of tissues [17]. In contrast to γδ T cells and MAIT cells, iNKT cells were initially found to inhibit tissue repair. Closure of cutaneous wounds was accelerated in iNKT cell-deficient Balb/c Traj18−/− mice compared to wildtype controls and anti-CD1d antibody enhanced wound healing in wildtype Balb/c mice [99, 100]. However, subsequent research demonstrated that wound healing was delayed in C57BL/6 Traj18−/− mice and administration of the iNKT cell agonist α-GalCer promoted wound closure in wildtype C57BL/6 mice [101]. Although these studies concluded opposing roles for iNKT cells in cutaneous wound healing, use of different murine strains may explain the disparate results. While most iNKT cells in Balb/c mice are type-2 (iNKT2), C57BL/6 animals have more type-1 (iNKT1) cells and a predominance of type-17 (iNKT17) cells in the skin and skin-draining lymph nodes [102–104]. Since other type-17 innate-like T cells promote cutaneous wound healing [17, 84], the greater abundance of iNKT17 cells in C57BL/6 mice may explain their beneficial contribution to tissue repair in these animals. Among murine tissues, iNKT cells are most abundant in the liver, where they have recently been shown to promote repair. Within 8 hours following injury, hepatic iNKT cells accumulate around the lesion, where they recognize endogenous ligands presented by CD1d [105]. Activated iNKT cells release IL-4, which induces hepatocyte proliferation and monocyte differentiation, resulting in enhanced wound healing [105]. iNKT cell-derived IL-4 also promotes thymic regeneration following irradiation [106]. IL-4 triggers thymic epithelial cells (TECs) to release the chemokine CCL11 and the resulting recruitment of CCR3+ peripheral eosinophils reestablishes thymic cellularity [106]. In the absence of iNKT cells, TECs do not recover, preventing the return of hematopoiesis [106]. Together, these recent studies demonstrate an emerging role for iNKT cells in tissue repair. Unconventional CD8+ T cells that recognize N-formylated peptides presented by the MHC-Ib molecule H2-M3 have recently been shown to promote tissue repair [107]. These cells respond to commensal S. epidermidis, which does not induce inflammation nor stimulatory cytokines when applied topically [108]. In this context, H2-M3-restricted RORγt+ CD8+ T cells preferentially accumulate in the epidermis [108], where they express immunoregulatory and tissue repair transcriptional signatures and enhance re-epithelialization of cutaneous wounds (Figure 3) [107]. The transcriptome of H2-M3-restricted RORγt+ CD8+ T cells resembles that of MAIT cells following the resolution of a Legionella infection [98], suggesting that TCR activation in the absence of stimulatory cytokines favors the expression of tissue repair genes in both populations. Conversely, intradermal inoculation of S. epidermidis causes an inflammatory response resulting in H2-M3-restricted CD8+ T cells that express T-bet and lack the reparative characteristics of the RORγt+ cells [107, 108]. However, the H2-M3-restricted Tbet+ CD8+ T cells that result from topical S. epidermidis also exhibit lower expression of tissue repair genes than their RORγt+ counterparts [107], suggesting that the polarization into either type-1 or type-17 is the primary determinant of whether these cells promote tissue repair. H2-M3-restricted CD8+ T cells can be positively selected on either thymic epithelial cells or hematopoietic cells, with the latter inducing effector characteristics, including enhanced IFN-γ production [2]. Whether the reparative functions of these cells depend on their method of selection remains to be determined. Although unconventional T cells recognize distinct antigens, similarly polarized subsets exhibit shared transcriptional programs [109]. Expression of identical cytokine receptors results in competition between unconventional T cells, enabling these populations to compensate for each other [19]. While this redundancy promotes tissue resilience by maintaining an abundance of effector cells, it can also obfuscate the roles of unconventional T cells. Recent analysis of MAIT cell-deficient Mr1−/− mice suggested that MAIT cells are not necessary for cutaneous wound healing [88], yet comparison of Mr1−/− Tcrd−/− and Tcrd−/− mice demonstrated that MAIT cells promote tissue repair in the absence of compensatory effects from more prevalent γδ T cells [17]. Conversely, use of a tamoxifen-inducible TcrdCreER Rorcf/f strain to ablate type-17 γδ T cells reduced the developmental compensation by MAIT cells that was previously observed in Tcrd−/− mice [17], resulting in decreased wound healing [88]. Future examination of unconventional T cell functions will necessitate use of compound mutations or inducible deletion to minimize redundancy. The beneficial and detrimental effects of unconventional T cells in inflammation Unconventional T cells have been implicated in numerous inflammatory conditions due to their rapid expression of cytokines following either TCR-dependent or cytokine-mediated activation [19, 110, 111]. Emerging evidence shows that unconventional T cells contribute to or attenuate inflammatory disorders in barrier tissues where they are abundant, including the skin, intestines, and genitourinary tract, as well as the liver and blood, and have been detected in synovial fluid (Figure 4) [19, 112, 113]. It is well reported that aberrations of the host microbiota in early life impact the development and abundance of unconventional T cells, leaving the host prone to inflammatory conditions [19]. Without the appropriate developmental signals from microbial stimuli, an opportunity is created for alternative cell subsets to fill the unconventional T cell niche in tissues where they are not normally abundant. These atypical immune cells are able to produce a range of cytokines, which can lead to abnormal responses to pathogens or other stimuli and cause the onset, prolongation or aggravation of inflammation. Studies of GF or antibiotic-treated mice in early life correlated the abundance of iNKT cells in tissues with increased susceptibility to inflammation in the colon and lungs [19]. Additionally, dysbiosis of gut microbiota caused by antibiotic treatment in early life increases the severity of psoriasis due to an enrichment of IL-22-producing Vγ4+ T cells [46], highlighting how impaired development of unconventional T cells can affect inflammation. Inflammatory bowel diseases (IBD), such as Crohn’s disease (CD) and ulcerative colitis (UC), are characterized by an improper T cell response to antigens presented from inflamed intestinal tissue [114]. While patients with IBD have a lower frequency of MAIT cells in their blood, they exhibit an enrichment of IL-17+ MAIT cells within their intestinal mucosa, suggesting that MAIT cells traffic to the site of inflammation [115–117]. Additionally, an increase in IL-22+ MAIT cells isolated from patients with UC has been reported [115]. However, the established roles of IL-17 and IL-22 as pro-inflammatory cytokines that can also promote tissue repair obscure the contribution of MAIT cells in IBD [118]. An in vivo model of oxazolone-induced colitis showed that Mr1−/− mice and inhibition of MAIT cell activation increased rates of survival in these mice, and ameliorated colonic inflammation[119]. The inhibition of MR1-dependent MAIT cell activation did not change IL-17 production but did reduce TNF-α and IFN-γ. It also led to reduced cytokine production from other unconventional T cells, including IL-17A from iNKT cells, as well as TNF-α and IFN-γ from iNKT, γδ T cells, and CD4+ T cells [119]. This suggests that MAIT cell activation, including production of cytokines, and subsequent activation of other cells by MAIT cells, contribute to IBD. The networks of inflammatory effects from unconventional T cells requires further study to clarify their involvement in IBD. In different models of intestinal inflammation, other sets of unconventional T cells serve a protective role. Following DSS-induced colitis, murine γδ IELs are more frequent at sites of damage and their absence correlates with longer recovery times [120]. These cells upregulate the expression of keratinocyte growth factor (KGF), which promotes the repair of the epithelial layer [120]. While αβ T cells do not express KGF in this context, mice lacking γδ T cells show some delayed repair after DSS, suggesting other involved repair mechanisms [120]. An analysis of IEL T cells isolated from the ileum of patients with severe CD detected a reduction of γδ T cells. One of the γδ T cell subsets that remain is proposed to have a protective role due to producing IL-26 [121]. Additionally, iNKTs may also contribute to inflammatory phenotypes in the intestines, however their ability to express a variety of cytokines and growth factors, makes it difficult to decipher if they are protective or pathogenic in patients and models of IBD [122]. A novel cell subtype, Crohn-associated invariant T (CAIT), has been identified in patients with CD [123]. CAIT cells are present in the mesenteric lymph nodes and intestines of CD patients and, contrary to MAIT cells, their abundance is increased in the blood during disease [123]. Transcriptional analysis has revealed similarities between CAIT, MAIT, and iNKT cells, indicating that CAIT cells are likely unconventional T cells [123]. Their expression of IFNg and TNF transcripts suggests they may contribute to inflammation [123]. In UC, a cluster of γδ T cells expressing high levels of CCR7 and low levels of NKG2A was enriched in peripheral blood [124]. The functional profiles of this cluster of cells and CAIT cells have yet to be characterized. Nonetheless, their presence in patients with IBD suggests novel unconventional T cell populations and subgroups may have an impact on inflammation in these diseases. Unconventional T cells are also reported to provide protection against airway inflammation due to allergies. A reduction of MAIT cell frequency was observed in the lungs and blood of patients with severe asthma, although this may be confounded by treatment with inhaled corticosteroids [125]. Higher MAIT cell frequency in the blood of infants and increased antigenic activity of iNKT cells against house dust extracts (HDEs) are both linked to lower risk of asthma development [126]. It is hypothesized that this protection is due to a correlation with IFN-γ produced by CD4+ T cells, as well as a promotion of Th1 cytokine production, which restricts a Th2 response [126]. Conversely, another study has implicated a higher frequency of IL-17-producing MAIT cells with severe exacerbations of asthma in older children [127]. A direct investigation of the role of MAIT cells in airway inflammation using Mr1−/− mice showed that lack of MAIT cells led to increased inflammation mediated by an ILC2 response. IL4I1, which can be produced by MAIT cells, inhibits expression of IL-5 and IL-13 by ILC2 cells in vitro and in response to the fungal allergen Alternaria in vivo [128]. While increased activity of iNKTs early in life has been correlated with a lower risk of asthma [126], stimulation of iNKTs by antigens in HDEs leads to greater production of cytokines, including IL-17, and more severe airway inflammation in an in vivo model [129]. In addition, studies of asthmatic patients have found an enrichment of iNKT cells in the lungs and are likely pathogenic, rather than protective [130]. Unconventional T cells have also been associated with inflammatory skin conditions. MAIT cells were found to be enriched in the skin of patients with dermatitis herpetiformis, but not psoriasis [131], while another study found IL-17A-producing MAIT cells in psoriatic skin [132]. Other populations of innate-like T cells, such as γδ T cells and iNKT cells, have been more directly implicated in skin inflammation [133]. γδ T cells have been found to be a major source of IL-17 in both humans and mouse models of psoriasis, in response to activation by IL-23 [95]. Other studies also implicate cytokine production by γδ T cells in psoriasis by using imiquimod (IMQ)-induced models in mice. An increase in expression of IL-17 and IL-22 further exacerbated inflammation [46, 134]. This response has also been tied to a downregulation of IL-38, which has been observed in psoriasis [134, 135]. IL-38 can regulate IL-17 production by γδ T cells by blocking X-linked IL-1 receptor accessory protein-like 1 (IL1RAPL1), which is upregulated on activated γδ T cells. Without IL-38 to restrict expression of IL-17, resolution of psoriatic inflammation is delayed [134]. iNKT cells are rarer in skin from psoriasis patients, but their ability to recruit IL-17 producing cells can contribute to inflammation [133]. Further investigation is necessary to determine their specific involvement in psoriasis. In patients with arthritis, joint inflammation can be accompanied by gut or skin inflammation, which sparked an investigation into the involvement of unconventional T cells in arthritis given their presence at these barrier tissues [136]. As described above, diet-induced changes to the immune profile of unconventional T cell subsets can lead to susceptibility to arthritis. Other studies have observed that MAIT cell frequency is decreased in the blood of patients but can be increased in synovial fluid (SF) [112, 113, 136–138]. Their migration to SF is mediated by TNF-α and IL-1β expressed by inflamed blood vessels [113]. MAIT cells isolated from these patients can produce more IL-17 and IL-22 and therefore may contribute to disease pathogenesis or exacerbation [112, 137, 138]. This was also observed in mouse models of arthritis in which MAIT cells were observed to have a pathogenic role [139]. Additionally, IL-17+ γδ T cells were detected only in SF from mice in a model of collagen-induced arthritis, but not in other models or in patients with rheumatoid arthritis (RA) [140]. The IL-17+ γδ T cells aggravated disease in this model but were not necessary to initially induce arthritis [140]. While additional studies are needed to firmly elucidate the mechanisms behind MAIT and γδ T cell contributions to arthritis, much of the current evidence points to them having a pathogenic role. The role of iNKT cells in arthritis has been more challenging to determine, as there are conflicting reports of evidence from human studies and mouse models. Similar to MAIT cell deficiency, iNKT cell-deficient mice were protected from the development of arthritis [141, 142]. Other studies have suggested that iNKT cell activation was sufficient to reduce arthritis severity [143]. This finding was supported by another study that showed in the absence of iNKT cells, arthritis symptoms worsen and that transfer of iNKT cells can slow disease progression in mice [144]. These differing outcomes may be dependent on the stimulating ligand and selected model of arthritis [136]. With the link between the liver and the gastrointestinal tract, the role of unconventional T cells in inflammatory diseases in the liver has also been explored. The three main classes of innate-like T cells have been concerned in liver inflammation, but contradicting evidence supports a protective role, as well [145–147]. The impact of each cell type is dependent on the context of inflammation. In chronic liver disease, iNKT and MAIT cells increase inflammation due to their production of IFN-γ [146]. One study observed an accumulation of iNKT cells in the livers of patients with non-alcoholic steatohepatitis (NASH)-related fibrosis, suggesting these cells promote disease progression [148]. In addition, they noted iNKT cell-deficient mice are protected from fibrosis [148]. In other models of liver inflammation, the contribution of γδ T cells is dependent on the subset and the produced cytokines [149]. While IL-17 has been linked to the pathogenesis of liver fibrosis in mice and patients [145, 150, 151], γδ T cells have been connected with a protective role against inflammation, independent of their IL-17-producing potential [149, 152]. Similar to other inflammatory conditions, MAIT cells were found to be decreased in the blood and accumulated in fibrotic livers [153]. These MAIT cells are activated and increase inflammation by promoting the production of inflammatory cytokines by myofibroblasts and macrophages in the liver [153]. Interactions between unconventional T cells and non-immune cells While iNKT cells comprise up to 40% of hepatic lymphocytes in mice, they are significantly less abundant in human liver [154]. Nevertheless, iNKT cells are thought to fulfill critical functions in liver homeostasis and disease. They emerged as key regulators of liver inflammation and are associated with a variety of pathologies ranging from hepatitis B and C virus infections to autoimmune diseases, metabolic liver diseases, and hepatocellular carcinoma [155]. Both hepatic immune cells such as Kupffer cells and dendritic cells [156, 157], and non-immune cells such as hepatocytes [158, 159], sinusoid-lining endothelial cells [160], and cholangiocytes in the biliary epithelium [161] express CD1d and are able to modulate iNKT cell responses (Table I). iNKT cells accumulate in the liver early following tissue damage and promote inflammation and hepatic fibrogenesis. This homing process is mediated by hepatic secretion of the chemokine CXCL16, presumably by liver sinusoidal endothelial cells [160, 162]. Alcoholic liver disease is characterized by increased hepatic iNKT cell numbers and activation. In contrast, during nonalcoholic fatty liver diseases, iNKT cell numbers deteriorate. This is likely caused by impaired presentation of endogenous antigens [163] and an increase in Kupffer cell derived IL-12 [164]. Different claims exist as to whether CD1d expression by hepatocytes is beneficial [158] or detrimental [159] for hepatic iNKT cell numbers, suggesting a critical role for co-stimulatory factors such as the local cytokine environment. A study on leptin-deficient ob/ob mice revealed that low norepinephrine (NE) levels correlated with decreased hepatic iNKT cell levels. In the absence of NE, iNKT cells underwent apoptosis which promoted a proinflammatory environment. Treatment with exogenous NE rescued hepatic iNKT levels and attenuated inflammation [165]. These findings add to a recent study that found that NE acts on hepatic immune cells and shapes the inflammatory environment, thus regulating insulin sensitivity and metabolic rate in hepatocytes [166]. Taken together, these suggest the presence of a novel axis by which iNKT cell activity and function can be regulated by neuronal factors. In both mice and humans, iNKT cells are enriched in visceral adipose tissue, where they constitute 15-20% of T cells and are heavily involved in regulating metabolism [154]. Activation of iNKT cells by α-GalCer or the Glp1 agonist liraglutide induced fibroblast growth factor 21 (FGF21) production in inguinal adipose tissue and promoted browning of white adipose tissue, thermogenesis, and weight loss [167]. The FGF21 producing cell type remains to be identified. Further investigation into this iNKT cell-dependent pathway may harbor great therapeutic potential as FGF21 has emerged as a key regulator of metabolic processes in adipose tissue [168]. While the previous study utilized exogenous ligands to activate iNKT cells, adipocytes are able to directly activate iNKT cells via presentation of endogenous lipid antigens on CD1d molecules [169]. Different accounts exist as to whether recognition of CD1d-antigen complexes on adipocytes results in pro- or anti-inflammatory iNKT cell responses. In the adipose tissue of mice fed a HFD, IFN-γ-producing iNKT cells accumulate in high numbers. IFN-γ production by iNKT cells upregulated CD1d expression, stimulated secretion of the chemokines CCL2 and CXCL16, and decreased secretion of the anti-inflammatory factor adiponectin in 3T3-L1 adipocytes. The same study showed that adipocyte-specific deletion of CD1d reduced weight gain in HFD mice and conveyed partial protection from an obese phenotype [170]. Contrasting these results, a different study with a similar experimental design reported increased adipose tissue inflammation and insulin resistance in HFD mice depleted of CD1d expression in adipocytes [171]. Similarly, the role of iNKT cells in the development of diet-induced obesity and insulin resistance is highly contested and independent studies have reported contradictory results that found protective, disease-promoting, or no roles for iNKT cells in obesity [172, 173]. The explanation for these contradicting data may come from a recent study that suggests that the cytokine output of adipocyte-activated iNKT cells may be altered by changes in the lipid microenvironment and, presumably, the nature of the lipids presented. Following co-culture with adipocytes that were previously exposed to a lipid-rich environment, iNKT cells adopted a pro-inflammatory profile characterized by decreased IL-4 and increased IFN-γ production [174]. iNKT cells fulfill crucial roles in regulating both immunity to infections and metabolic homeostasis. While they are highly abundant in mice, they are significantly less frequent in humans, and it remains to be investigated to what extent functional overlap exists between mouse and human iNKT cells. Although hepatocytes and adipocytes appear to be key iNKT cell activators via their capacity to present antigen on CD1d molecules, it remains to be determined whether these cells fulfill a similar function in humans. γδ T cells are rarely found in circulation and inside secondary organs but are abundant in peripheral tissues. It may therefore be assumed that they fulfill important, specialized functions in regulating tissue homeostasis. Finding the interaction partners of γδ T cells has proven to be a difficult task, however, as neither the antigens that are recognized by γδ T cells, nor their way of presentation is well characterized. The first γδ T cells to develop in mice are dendritic epidermal T cells (DETCs) [175]. DETCs are characterized by their expression of an invariant Vγ5+ Vδ1+ TCR (according to Tonegawa’s nomenclature) and their distinct morphology that includes many dendrites. They play important roles in regulating epithelial tissue homeostasis and repair [176]. In the skin, DETCs form clusters and extend dendrites towards the stratum corneum were they constantly sense ligands expressed during steady state and stress [33]. As previously discussed, they maintain tissue homeostasis, epithelial barrier integrity, and respond to wounding via the secretion of cytokines and growth factors. DETC-derived IGF-1 protects keratinocytes from apoptosis and is crucial for epithelial barrier integrity [82]. IL-13 controls the migration of epithelial cells through the epidermis, supports tissue repair, and protects against cutaneous carcinogenesis [177]. In the intestinal epithelium, activated DETCs secrete KGF1 and KGF2 [178, 179]. KGF induces migration and proliferation programs in keratinocytes and aid wound repair. In return, keratinocytes regulate the growth of resident DETCs via the secretion of IL-7 [180], and epithelial cell-derived IL15 was found to be essential for the maturation and proliferation of DETCs in the skin [181]. In the thymus, DETCs are selected on thymic epithelial cells in a manner dependent on the expression of selection and upkeep of intraepithelial T cells 1 (SKINT1), a member of the butyrophilin-like (BTNL) family [81, 182]. SKINT1 is also expressed by keratinocytes where its recognition by DETCs has recently been found to play important roles in tissue homeostasis. Disrupting the SKINT1-TCR interaction resulted in DETC-mediated inflammation and increased susceptibility to ultraviolet B radiation (UVR)-induced DNA damage [183]. These results add to a previous study that found that keratinocytes upregulate yet to be characterized DETC ligands in response to cutaneous injury and thereby initiate a DETC-mediated repair pathway [184]. BTNL family genes seem to be key mediators of keratinocyte-to-γδ T cell crosstalk as their expression in murine enterocytes and human gut epithelial cells elicit TCR-dependent responses in mouse intestinal Vγ7+ cells and human colonic Vγ4+ cells, respectively [185]. Expression of BTNL1 confined to the gut epithelium was able to select and expand Vγ7+ intraepithelial lymphocytes in BTNL1−/− mice during an early developmental window [185]. These results suggest that γδ T cells subsets could be selected outside the thymus presumably in their tissue of residence by local cell types. Further research is needed to identify the molecules and cell types that mediate this selection process. While the interaction of γδ T cells with keratinocytes has been well characterized, significantly less is known about their functions and interaction partners in other tissues. γδ T cell can stimulate hair follicle stem cells to proliferate following their activation by keratinocyte-derived IL-1α and IL-7 [186]. Furthermore, following wounding, γδ T cells secrete fibroblast growth factor 9 (FGF9) which triggers Wnt signaling in fibroblasts and induces hair follicle neogenesis [85]. Recent evidence supports a role for γδ T cells in bone homeostasis. γδ T cell-derived IL-17A stimulates proliferation and osteoblastic differentiation of mesenchymal progenitor cells, thereby promoting the bone healing process [90]. These findings are supported by a recent study that found that in a polytrauma rat model γδ T cells accumulate in the fracture site and appear to support the healing process. The supportive role of γδ T cells, however, was dependent on the neutralization of the damage associated molecular pattern high mobility group box protein 1 (HMGB1) [187]. γδ T cells have emerged as regulators of thermogenesis in adipose tissue. One proposed mechanism involves the activation of IL-33 secretion by stromal cells via secretion of TNF-α and IL-17A [188]. Another study suggests, however, that γδ T cells secrete the cytokine IL-17F that activates IL-17RC-expressing adipocytes to express TGFβ1 [189]. Lastly, the role of γδ T cell-to-neuron crosstalk has recently gained increased attention with γδ T cells attributed roles both in homeostasis and disease. One study found that short-term memory development in mice is dependent on a fetal-derived meningeal-resident γδ T cell subset that secretes IL-17. γδ T cell-derived IL-17 production sustains plasticity of glutamatergic synapses and enhanced production of brain-derived neurotropic factor by glial cells [190]. CXCR6+ γδ T cells that home to the meninges shortly after birth produce IL-17A under homeostatic conditions. The IL-17A act on IL-17A receptor-expressing neurons and induces anxiety-like behavior in mice [26]. In the periphery, γδ T cells and neurons reside in close proximity and neuronal signaling regulates γδ T cell responses. Neuronal signals activate resident dendritic cells which in return secrete IL-23 to activate that mediate both inflammatory responses [191] and systemic tissue regeneration in the skin [87]. Neuronal signaling following a pulmonary infection with Staphylococcus aureus dampens the γδ T cell-mediated immune response and results in increased morbidity in mice [192]. As the interplay between neurons and γδ T cells in tissue homeostasis, infection, and inflammation becomes increasingly apparent, it remains to be investigated whether neuronal factors can directly act on receptors expressed by γδ T cells, or whether the presence of mediators such as tissue-resident dendritic cells. γδ T cells constitute an intriguing yet little investigated T cell lineage. γδ T cell populations are highly heterogeneous in that they recognize a great variety of antigens, take different developmental paths, home to distinct tissues, and fulfill highly specialized functions. While they have been implicated in immune responses to infection, γδ T cells appear to be key regulators of tissue homeostasis and maintain close communication networks with non-immune cells. Their implication in supporting thermogenesis in and browning of white adipose tissue makes them attractive targets in the treatment of obesity. Furthermore, their proposed role as modulators of cognitive development and behavior demands a more careful examination of the molecular signals that mediate this neuro-immune crosstalk. MAIT cells are abundant in epithelial barrier tissues where they fulfill crucial functions in tissue homeostasis. Extensive research has been conducted on how MAIT cells contribute to wound healing and tissue repair, as described above. At steady state MAIT cells possess low cytotoxicity, however, the expression of perforin and granzyme may be upregulated following TCR stimulation [193, 194]. Similar to cytotoxic CD8+ T cells, they are able to kill bacteria-infected cells via the release of lytic granules [195, 196]. Whether MAIT cells are important in the killing of virus-infected cells has not been fully established. However, MAIT cells with increased granzyme B secretion phenotype have been isolated from patients infected with the human immunodeficiency virus (HIV) [197] and hepatitis C virus (HCV) [198]. While this killing of body cells is beneficial in the context of an infection, the cytotoxic properties of MAIT cells may contribute to disease progression in type 1 diabetes (T1D). Granzyme B secreting MAIT cells are elevated in the blood of T1D patients, and the frequency of these cells was negatively associated with children’s age at T1D diagnosis [199]. Furthermore, the same study was able to show that granzyme B expressing MAIT cells accumulate in the pancreas of non-obese diabetic (NOD) mice and the level of granzyme B in pancreatic islets correlated with disease progression. A recent study found that in SARS-CoV-2 infected patients, MAIT cells exit the circulation and accumulate in the lungs where they produce granzyme B and may cause damage to the epithelium [200]. A detrimental role for MAIT cells has also been identified in alcoholic liver disease (ALD). The data suggests that MAIT cells exit the circulation of ALD patients and accumulate in liver fibrotic septa. In a mouse model of chronic liver injury, MAIT cells facilitated the transformation of hepatic myofibroblasts to a proinflammatory phenotype [153]. As sensors of the microbiota, MAIT cells play a crucial role in maintaining organismal homeostasis. Their functions have primarily been characterized in epithelial barrier tissues where they promote barrier integrity and participate in the immune response to infection. It has been shown that MAIT cell driven inflammation contributes to metabolic pathologies such as obesity [201] and T1D [202]. To our knowledge, however, little is known about whether and how MAIT cells communicate with non-immune cells such as neurons, stem cells, or stromal cells in the absence of inflammatory stimuli. Conclusion By recognizing conserved ligands, including metabolites, lipids, and modified peptides, unconventional T cells increase the diversity of microbial antigens detected by the immune system. Their restriction to monomorphic MHC-Ib molecules, many of which are conserved across mammals [203], suggests that these T cells have been essential for millions of years. Due to their abundance within barrier tissues, unconventional T cells are uniquely positioned to provide antimicrobial defense and well as maintain tissue homeostasis. This duality is demonstrated by the effector molecules they rapidly release. Both IL-4 and IL-17/22 promote tissue repair, yet also mediate protection against helminths and extracellular bacteria and fungi, respectively. However, their proximity to bodily surfaces renders unconventional T cells particularly sensitive to exogenous factors, including the microbiota, diet, and drugs. For example, antibiotic-induced dysbiosis can impact the development and function of unconventional T cells, rendering the tissue susceptible to inflammation. While such responsiveness necessitates care, it provides therapeutic opportunity to leverage unconventional T cells for maintaining tissue homeostasis. Acknowledgements The authors thank Dr. Timothy Hinks for helpful discussion. This work was supported by the National Institute of Allergy and Infectious Diseases (K22-AI146217 and R21-AI171697 to M.G.C.). Figure 1: Dependence of unconventional T cells on the microbiota In germ-free mice (left), iNKT cells are selected on endogenous ligands, which imbues them with diminished effector functions, while MAIT cells are developmentally arrested within the thymus, resulting in as absence of MAIT cells within barrier tissues. Increased expression of the chemokine CXCL16 within the colon recruits iNKT cells, which predispose the tissue to inflammation. In colonized animals (right), iNKT cells are selected on microbial glycosphingolipids and MAIT cells recognize microbial metabolites, promoting their acquisition of effector functions. In the colon, inhibitory sphingolipids from B. fragilis bacteria outcompete endogenous ligands for CD1d presentation, preventing inflammation. Furthermore, derivatives of riboflavin synthesis stimulate MAIT cell proliferation within the tissue. Colonization promotes the accumulation of Vγ6+ T cells, but not γδ intraepithelial lymphocytes (IELs), which are not dependent on microbial signals. Figure 2: How drugs affect the functions of unconventional T cells Sulfa-like drugs stimulate type II NKT (NKT-II) cells, exacerbating inflammation, while drug metabolites can competitively inhibit MAIT cell recognition of riboflavin derivatives, dampening their activation. Antibiotics reduce the abundance of microbial antigens recognized by MAIT and type-17 γδ T (γδT17) cells, diminishing their effector functions and minimizing inflammation. However, certain antibiotics increase the abundance of iNKT cells, which can promote inflammation. Figure 3: Unconventional T cells promote tissue repair Commensal S. epidermidis bacteria stimulate H2-M3 restricted T cells and MAIT cells through N-formyl peptides and riboflavin derivatives, respectively, resulting in their release of reparative cytokines. Following injury, keratinocytes express butyrophilin-like proteins, which activate DETCs, causing them to lose their characteristic dendritic morphology. Both DETCs and dermal type-17 γδ T (γδT17) cells, release growth factors and IL-17, the latter of which enables epithelial cell migration by enhancing glycolysis through HIF-1α upregulation. Figure 4: The role of unconventional T cells in inflammation Unconventional T cells are associated with inflammation (denoted in red) or resolution/health (denoted in green) in the indicated organs. *iNKT cells have been reported in both inflamed and healthy joints, liver, and lungs. Table I: Interactions between unconventional T cells and non-immune cells Target Factor Origin Effects Reference iNKT cells Norepinephrine Nervous system ↑ cell survival ↓ inflammation [165] CXCL16 Liver ↑NKT cell recruitment [160, 162] IL-12 Liver ↓ NKT cell numbers [164] γδ T cells IL-7 Keratinocytes ↑ cell growth [180] IL-15 Epithelium ↑ proliferation [181] SKINT1 Keratinocytes ↑ tissue homeostasis ↑ inflammation control [183] IL-1A, IL-7 Keratinocytes ↑ hair follicle stem cell proliferation [186] Origin Factor Target Effects Reference iNKT cells IFN-γ Adipocytes ↑ CD1d expression ↑ CCl2 ↑ CXCl16 ↓ adiponectin [170] γδ T cell IL-17 Brain ↑ neuronal plasticity [190] IL-17A Mesenchymal progenitors ↑ proliferation ↑ differentiation ↑ bone healing [90] Stromal cells ↑ IL-33 secretion [188] IL-17F Adipocytes ↑ TGFβ1 [189] IGF-1 Keratinocytes ↑ cell survival ↑ epithelial barrier integrity [82] KGF1, KGF2 Keratinocytes ↑ migration ↑ proliferation ↑ wound repair [178, 179] IL-13 Epithelial cells ↑ cell migration ↑ tissue repair [177] FGF-2, FGF-7, FGF-9, FGF-10 Fibroblasts, Keratinocytes ↑ wound healing ↑ hair follicle neogenesis [84, 85] TNF-α Stromal cells ↑ IL-33 secretion [188] MAIT cells Granzyme B Various cell types ↑ apoptosis of target cells [193, 194, 200] References [1] Godfrey DI , Uldrich AP , McCluskey J , Rossjohn J , Moody DB , The burgeoning family of unconventional T cells, Nat Immunol 16 (11 ) (2015) 1114–23.26482978 [2] Cho H , Bediako Y , Xu H , Choi HJ , Wang CR , Positive selecting cell type determines the phenotype of MHC class Ib-restricted CD8+ T cells, Proc Natl Acad Sci U S A 108 (32 ) (2011) 13241–6.21788511 [3] Lee YJ , Jeon YK , Kang BH , Chung DH , Park CG , Shin HY , Jung KC , Park SH , Generation of PLZF+ CD4+ T cells via MHC class II-dependent thymocyte-thymocyte interaction is a physiological process in humans, J Exp Med 207 (1 ) (2010) 237–46.20038602 [4] Dimova T , Brouwer M , Gosselin F , Tassignon J , Leo O , Donner C , Marchant A , Vermijlen D , Effector Vgamma9Vdelta2 T cells dominate the human fetal gammadelta T-cell repertoire, Proc Natl Acad Sci U S A 112 (6 ) (2015) E556–65.25617367 [5] Leeansyah E , Loh L , Nixon DF , Sandberg JK , Acquisition of innate-like microbial reactivity in mucosal tissues during human fetal MAIT-cell development, Nat Commun 5 (2014) 3143.24452018 [6] Loh L , Ivarsson MA , Michaelsson J , Sandberg JK , Nixon DF , Invariant natural killer T cells developing in the human fetus accumulate and mature in the small intestine, Mucosal Immunol 7 (5 ) (2014) 1233–43.24646938 [7] Zachariassen LF , Krych L , Rasmussen SH , Nielsen DS , Kot W , Holm TL , Hansen AK , Hansen CHF , Cesarean Section Induces Microbiota-Regulated Immune Disturbances in C57BL/6 Mice, J Immunol 202 (1 ) (2019) 142–150.30487172 [8] Wei B , Wingender G , Fujiwara D , Chen DY , McPherson M , Brewer S , Borneman J , Kronenberg M , Braun J , Commensal microbiota and CD8+ T cells shape the formation of invariant NKT cells, J Immunol 184 (3 ) (2010) 1218–26.20048124 [9] Wingender G , Stepniak D , Krebs P , Lin L , McBride S , Wei B , Braun J , Mazmanian SK , Kronenberg M , Intestinal microbes affect phenotypes and functions of invariant natural killer T cells in mice, Gastroenterology 143 (2 ) (2012) 418–28.22522092 [10] Olszak T , An D , Zeissig S , Vera MP , Richter J , Franke A , Glickman JN , Siebert R , Baron RM , Kasper DL , Blumberg RS , Microbial exposure during early life has persistent effects on natural killer T cell function, Science 336 (6080 ) (2012) 489–93.22442383 [11] Constantinides MG , Interactions between the microbiota and innate and innate-like lymphocytes, J Leukoc Biol 103 (3 ) (2018) 409–419.29345366 [12] Bendelac A , Positive selection of mouse NK1+ T cells by CD1-expressing cortical thymocytes, J Exp Med 182 (6 ) (1995) 2091–6.7500054 [13] Sriram V , Du W , Gervay-Hague J , Brutkiewicz RR , Cell wall glycosphingolipids of Sphingomonas paucimobilis are CD1d-specific ligands for NKT cells, Eur J Immunol 35 (6 ) (2005) 1692–701.15915536 [14] An D , Oh SF , Olszak T , Neves JF , Avci FY , Erturk-Hasdemir D , Lu X , Zeissig S , Blumberg RS , Kasper DL , Sphingolipids from a symbiotic microbe regulate homeostasis of host intestinal natural killer T cells, Cell 156 (1-2 ) (2014) 123–33.24439373 [15] Koay HF , Gherardin NA , Enders A , Loh L , Mackay LK , Almeida CF , Russ BE , Nold-Petry CA , Nold MF , Bedoui S , Chen Z , Corbett AJ , Eckle SB , Meehan B , d’Udekem Y , Konstantinov IE , Lappas M , Liu L , Goodnow CC , Fairlie DP , Rossjohn J , Chong MM , Kedzierska K , Berzins SP , Belz GT , McCluskey J , Uldrich AP , Godfrey DI , Pellicci DG , A three-stage intrathymic development pathway for the mucosal-associated invariant T cell lineage, Nat Immunol 17 (11 ) (2016) 1300–1311.27668799 [16] Legoux F , Bellet D , Daviaud C , El Morr Y , Darbois A , Niort K , Procopio E , Salou M , Gilet J , Ryffel B , Balvay A , Foussier A , Sarkis M , El Marjou A , Schmidt F , Rabot S , Lantz O , Microbial metabolites control the thymic development of mucosal-associated invariant T cells, Science 366 (6464 ) (2019) 494–499.31467190 [17] Constantinides MG , Link VM , Tamoutounour S , Wong AC , Perez-Chaparro PJ , Han SJ , Chen YE , Li K , Farhat S , Weckel A , Krishnamurthy SR , Vujkovic-Cvijin I , Linehan JL , Bouladoux N , Merrill ED , Roy S , Cua DJ , Adams EJ , Bhandoola A , Scharschmidt TC , Aube J , Fischbach MA , Belkaid Y , MAIT cells are imprinted by the microbiota in early life and promote tissue repair, Science 366 (6464 ) (2019) eaax6624.31649166 [18] Treiner E , Duban L , Bahram S , Radosavljevic M , Wanner V , Tilloy F , Affaticati P , Gilfillan S , Lantz O , Selection of evolutionarily conserved mucosal-associated invariant T cells by MR1, Nature 422 (6928 ) (2003) 164–9.12634786 [19] Constantinides MG , Belkaid Y , Early-life imprinting of unconventional T cells and tissue homeostasis, Science 374 (6573 ) (2021) eabf0095.34882451 [20] Ben Youssef G , Tourret M , Salou M , Ghazarian L , Houdouin V , Mondot S , Mburu Y , Lambert M , Azarnoush S , Diana JS , Virlouvet AL , Peuchmaur M , Schmitz T , Dalle JH , Lantz O , Biran V , Caillat-Zucman S , Ontogeny of human mucosal-associated invariant T cells and related T cell subsets, J Exp Med 215 (2 ) (2018) 459–479.29339446 [21] Swarbrick GM , Gela A , Cansler ME , Null MD , Duncan RB , Nemes E , Shey M , Nsereko M , Mayanja-Kizza H , Kiguli S , Koh J , Hanekom WA , Hatherill M , Lancioni C , Lewinsohn DM , Scriba TJ , Lewinsohn DA , Postnatal Expansion, Maturation, and Functionality of MR1T Cells in Humans, Front Immunol 11 (2020) 556695.33042140 [22] Arrieta MC , Stiemsma LT , Amenyogbe N , Brown EM , Finlay B , The intestinal microbiome in early life: health and disease, Front Immunol 5 (2014) 427.25250028 [23] Heilig JS , Tonegawa S , Diversity of murine gamma genes and expression in fetal and adult T lymphocytes, Nature 322 (6082 ) (1986) 836–40.2943999 [24] Papotto PH , Yilmaz B , Silva-Santos B , Crosstalk between gammadelta T cells and the microbiota, Nat Microbiol 6 (9 ) (2021) 1110–1117.34341528 [25] Duan J , Chung H , Troy E , Kasper DL , Microbial colonization drives expansion of IL-1 receptor 1-expressing and IL-17-producing gamma/delta T cells, Cell Host Microbe 7 (2 ) (2010) 140–50.20159619 [26] Alves de Lima K , Rustenhoven J , Da Mesquita S , Wall M , Salvador AF , Smirnov I , Martelossi Cebinelli G , Mamuladze T , Baker W , Papadopoulos Z , Lopes MB , Cao WS , Xie XS , Herz J , Kipnis J , Meningeal gammadelta T cells regulate anxiety-like behavior via IL-17a signaling in neurons, Nat Immunol 21 (11 ) (2020) 1421–1429.32929273 [27] Monin L , Ushakov DS , Arnesen H , Bah N , Jandke A , Munoz-Ruiz M , Carvalho J , Joseph S , Almeida BC , Green MJ , Nye E , Hatano S , Yoshikai Y , Curtis M , Carlsen H , Steinhoff U , Boysen P , Hayday A , gammadelta T cells compose a developmentally regulated intrauterine population and protect against vaginal candidiasis, Mucosal Immunol 13 (6 ) (2020) 969–981.32472066 [28] Wilharm A , Tabib Y , Nassar M , Reinhardt A , Mizraji G , Sandrock I , Heyman O , Barros-Martins J , Aizenbud Y , Khalaileh A , Eli-Berchoer L , Elinav E , Wilensky A , Forster R , Bercovier H , Prinz I , Hovav AH , Mutual interplay between IL-17-producing gammadeltaT cells and microbiota orchestrates oral mucosal homeostasis, Proc Natl Acad Sci U S A 116 (7 ) (2019) 2652–2661.30692259 [29] Koren N , Zubeidat K , Saba Y , Horev Y , Barel O , Wilharm A , Heyman O , Wald S , Eli-Berchoer L , Shapiro H , Nadler C , Elinav E , Wilensky A , Prinz I , Bercovier H , Hovav AH , Maturation of the neonatal oral mucosa involves unique epithelium-microbiota interactions, Cell Host Microbe 29 (2 ) (2021) 197–209 e5.33412104 [30] Bandeira A , Mota-Santos T , Itohara S , Degermann S , Heusser C , Tonegawa S , Coutinho A , Localization of gamma/delta T cells to the intestinal epithelium is independent of normal microbial colonization, J Exp Med 172 (1 ) (1990) 239–44.2141628 [31] Malinarich FH , Grabski E , Worbs T , Chennupati V , Haas JD , Schmitz S , Candia E , Quera R , Malissen B , Forster R , Hermoso M , Prinz I , Constant TCR triggering suggests that the TCR expressed on intestinal intraepithelial gammadelta T cells is functional in vivo, Eur J Immunol 40 (12 ) (2010) 3378–88.21108461 [32] Naik S , Bouladoux N , Wilhelm C , Molloy MJ , Salcedo R , Kastenmuller W , Deming C , Quinones M , Koo L , Conlan S , Spencer S , Hall JA , Dzutsev A , Kong H , Campbell DJ , Trinchieri G , Segre JA , Belkaid Y , Compartmentalized control of skin immunity by resident commensals, Science 337 (6098 ) (2012) 1115–9.22837383 [33] Chodaczek G , Papanna V , Zal MA , Zal T , Body-barrier surveillance by epidermal gammadelta TCRs, Nat Immunol 13 (3 ) (2012) 272–82.22327568 [34] Ridaura VK , Bouladoux N , Claesen J , Chen YE , Byrd AL , Constantinides MG , Merrill ED , Tamoutounour S , Fischbach MA , Belkaid Y , Contextual control of skin immunity and inflammation by Corynebacterium, J Exp Med 215 (3 ) (2018) 785–799.29382696 [35] Cai Y , Xue F , Qin H , Chen X , Liu N , Fleming C , Hu X , Zhang HG , Chen F , Zheng J , Yan J , Differential Roles of the mTOR-STAT3 Signaling in Dermal gammadelta T Cell Effector Function in Skin Inflammation, Cell Rep 27 (10 ) (2019) 3034–3048 e5.31167146 [36] McVay LD , Carding SR , Generation of human gammadelta T-cell repertoires, Crit Rev Immunol 19 (5-6 ) (1999) 431–60.10647745 [37] McVay LD , Carding SR , Extrathymic origin of human gamma delta T cells during fetal development, J Immunol 157 (7 ) (1996) 2873–82.8816392 [38] Robertson RC , Manges AR , Finlay BB , Prendergast AJ , The Human Microbiome and Child Growth - First 1000 Days and Beyond, Trends Microbiol 27 (2 ) (2019) 131–147.30529020 [39] Parker CM , Groh V , Band H , Porcelli SA , Morita C , Fabbi M , Glass D , Strominger JL , Brenner MB , Evidence for extrathymic changes in the T cell receptor gamma/delta repertoire, J Exp Med 171 (5 ) (1990) 1597–612.2185330 [40] Ravens S , Fichtner AS , Willers M , Torkornoo D , Pirr S , Schoning J , Deseke M , Sandrock I , Bubke A , Wilharm A , Dodoo D , Egyir B , Flanagan KL , Steinbruck L , Dickinson P , Ghazal P , Adu B , Viemann D , Prinz I , Microbial exposure drives polyclonal expansion of innate gammadelta T cells immediately after birth, Proc Natl Acad Sci U S A 117 (31 ) (2020) 18649–18660.32690687 [41] Esin S , Shigematsu M , Nagai S , Eklund A , Wigzell H , Grunewald J , Different percentages of peripheral blood gamma delta + T cells in healthy individuals from different areas of the world, Scand J Immunol 43 (5 ) (1996) 593–6.8633219 [42] Amenyogbe N , Kollmann TR , Ben-Othman R , Early-Life Host-Microbiome Interphase: The Key Frontier for Immune Development, Front Pediatr 5 (2017) 111.28596951 [43] Duong QA , Pittet LF , Curtis N , Zimmermann P , Antibiotic exposure and adverse long-term health outcomes in children: A systematic review and meta-analysis, J Infect (2022). [44] Gensollen T , Blumberg RS , Correlation between early-life regulation of the immune system by microbiota and allergy development, J Allergy Clin Immunol 139 (4 ) (2017) 1084–1091.28390575 [45] Ennamorati M , Vasudevan C , Clerkin K , Halvorsen S , Verma S , Ibrahim S , Prosper S , Porter C , Yeliseyev V , Kim M , Gardecki J , Sassi S , Tearney G , Cherayil BJ , Bry L , Seed B , Jain N , Intestinal microbes influence development of thymic lymphocytes in early life, Proc Natl Acad Sci U S A 117 (5 ) (2020) 2570–2578.31964813 [46] Zanvit P , Konkel JE , Jiao X , Kasagi S , Zhang D , Wu R , Chia C , Ajami NJ , Smith DP , Petrosino JF , Abbatiello B , Nakatsukasa H , Chen Q , Belkaid Y , Chen ZJ , Chen W , Antibiotics in neonatal life increase murine susceptibility to experimental psoriasis, Nat Commun 6 (2015) 8424.26416167 [47] Zanvit P , Konkel JE , Jiao X , Kasagi S , Zhang DF , Wu RQ , Chia C , Ajami NJ , Smith DP , Petrosino JF , Abbatiello B , Nakatsukasa H , Chen QM , Belkaid Y , Chen ZJ , Chen WJ , Antibiotics in neonatal life increase murine susceptibility to experimental psoriasis, Nat Commun 6 (2015). [48] Strzepa A , Majewska-Szczepanik M , Lobo FM , Wen L , Szczepanik M , Broad spectrum antibiotic enrofloxacin modulates contact sensitivity through gut microbiota in a murine model, J Allergy Clin Immun 140 (1 ) (2017) 121-+.28130148 [49] Huang SW , Chen YJ , Wang ST , Ho LW , Kao JK , Narita M , Takahashi M , Wu CY , Cheng HY , Shieh JJ , Azithromycin impairs TLR7 signaling in dendritic cells and improves the severity of imiquimod-induced psoriasis-like skin inflammation in mice, J Dermatol Sci 84 (1 ) (2016) 59–70.27449383 [50] Kim Y , Lee YS , Yang JY , Lee SH , Park YY , Kweon MN , The resident pathobiont Staphylococcus xylosus in Nfkbiz-deficient skin accelerates spontaneous skin inflammation, Sci Rep-Uk 7 (2017). [51] Bazett M , Bergeron ME , Haston CK , Streptomycin treatment alters the intestinal microbiome, pulmonary T cell profile and airway hyperresponsiveness in a cystic fibrosis mouse model, Sci Rep-Uk 6 (2016). [52] Jin CC , Lagoudas GK , Zhao C , Bullman S , Bhutkar A , Hu B , Ameh S , Sandel D , Liang XS , Mazzilli S , Whary MT , Meyerson M , Germain R , Blainey PC , Fox JG , Jacks T , Commensal Microbiota Promote Lung Cancer Development via gamma delta T Cells, Cell 176 (5 ) (2019) 998-+.30712876 [53] Burrello C , Garavaglia F , Cribiu FM , Ercoli G , Bosari S , Caprioli F , Facciotti F , Short-term Oral Antibiotics Treatment Promotes Inflammatory Activation of Colonic invariant Natural killer T and Converntional CD4(+) t Cells, Front Med-Lausanne 5 (2018). [54] Strati F , Pujolassos M , Burrello C , Giuffre MR , Lattanzi G , Caprioli F , Troisi J , Facciotti F , Antibiotic-associated dysbiosis affects the ability of the gut microbiota to control intestinal inflammation upon fecal microbiota transplantation in experimental colitis models, Microbiome 9 (1 ) (2021). [55] Maier L , Pruteanu M , Kuhn M , Zeller G , Telzerow A , Anderson EE , Brochado AR , Fernandez KC , Dose H , Mori H , Patil KR , Bork P , Typas A , Extensive impact of non-antibiotic drugs on human gut bacteria, Nature 555 (7698 ) (2018) 623–628.29555994 [56] Dasgupta S , Kumar V , Type II NKT cells: a distinct CD1d-restricted immune regulatory NKT cell subset, Immunogenetics 68 (8 ) (2016) 665–76.27405300 [57] Van Rhijn I , Young DC , Im JS , Levery SB , Illarionov PA , Besra GS , Porcelli SA , Gumperz J , Cheng TY , Moody DB , CD1d-restricted T cell activation by nonlipidic small molecules, Proc Natl Acad Sci U S A 101 (37 ) (2004) 13578–83.15342907 [58] Almeida CF , Smith DGM , Cheng TY , Harpur CM , Batleska E , Nguyen-Robertson CV , Nguyen T , Thelemann T , Reddiex SJJ , Li S , Eckle SBG , Van Rhijn I , Rossjohn J , Uldrich AP , Moody DB , Williams SJ , Pellicci DG , Godfrey DI , Benzofuran sulfonates and small self-lipid antigens activate type II NKT cells via CD1d, Proc Natl Acad Sci U S A 118 (34 ) (2021). [59] Giles A , Foushee J , Lantz E , Gumina G , Sulfonamide Allergies, Pharmacy (Basel) 7 (3 ) (2019). [60] Khan DA , Knowles SR , Shear NH , Sulfonamide Hypersensitivity: Fact and Fiction, J Allergy Clin Immunol Pract 7 (7 ) (2019) 2116–2123.31495421 [61] Nicolai S , Wegrecki M , Cheng TY , Bourgeois EA , Cotton RN , Mayfield JA , Monnot GC , Le Nours J , Van Rhijn I , Rossjohn J , Moody DB , de Jong A , Human T cell response to CD1a and contact dermatitis allergens in botanical extracts and commercial skin care products, Sci Immunol 5 (43 ) (2020). [62] Cotton RN , Cheng TY , Wegrecki M , Le Nours J , Orgill DP , Pomahac B , Talbot SG , Willis RA , Altman JD , de Jong A , Ogg G , Van Rhijn I , Rossjohn J , Clark RA , Moody DB , Human skin is colonized by T cells that recognize CD1a independently of lipid, J Clin Invest 131 (1 ) (2021). [63] Greene JM , Dash P , Roy S , McMurtrey C , Awad W , Reed JS , Hammond KB , Abdulhaqq S , Wu HL , Burwitz BJ , Roth BF , Morrow DW , Ford JC , Xu G , Bae JY , Crank H , Legasse AW , Dang TH , Greenaway HY , Kurniawan M , Gold MC , Harriff MJ , Lewinsohn DA , Park BS , Axthelm MK , Stanton JJ , Hansen SG , Picker LJ , Venturi V , Hildebrand W , Thomas PG , Lewinsohn DM , Adams EJ , Sacha JB , MR1-restricted mucosal-associated invariant T (MAIT) cells respond to mycobacterial vaccination and infection in nonhuman primates, Mucosal Immunol 10 (3 ) (2017) 802–813.27759023 [64] Harriff MJ , McMurtrey C , Froyd CA , Jin H , Cansler M , Null M , Worley A , Meermeier EW , Swarbrick G , Nilsen A , Lewinsohn DA , Hildebrand W , Adams EJ , Lewinsohn DM , MR1 displays the microbial metabolome driving selective MR1-restricted T cell receptor usage, Sci Immunol 3 (25 ) (2018). [65] Wang H , D’Souza C , Lim XY , Kostenko L , Pediongco TJ , Eckle SBG , Meehan BS , Shi M , Wang N , Li S , Liu L , Mak JYW , Fairlie DP , Iwakura Y , Gunnersen JM , Stent AW , Godfrey DI , Rossjohn J , Westall GP , Kjer-Nielsen L , Strugnell RA , McCluskey J , Corbett AJ , Hinks TSC , Chen Z , MAIT cells protect against pulmonary Legionella longbeachae infection, Nat Commun 9 (1 ) (2018) 3350.30135490 [66] Keller AN , Eckle SB , Xu W , Liu L , Hughes VA , Mak JY , Meehan BS , Pediongco T , Birkinshaw RW , Chen Z , Wang H , D’Souza C , Kjer-Nielsen L , Gherardin NA , Godfrey DI , Kostenko L , Corbett AJ , Purcell AW , Fairlie DP , McCluskey J , Rossjohn J , Drugs and drug-like molecules can modulate the function of mucosal-associated invariant T cells, Nat Immunol 18 (4 ) (2017) 402–411.28166217 [67] Christ A , Lauterbach M , Latz E , Western Diet and the Immune System: An Inflammatory Connection, Immunity 51 (5 ) (2019) 794–811.31747581 [68] Shi ZR , Wu XS , Yu S , Huynh M , Jena PK , Nguyen M , Wan YJY , Hwang ST , Short-Term Exposure to a Western Diet Induces Psoriasiform Dermatitis by Promoting Accumulation of IL-17A-Producing gamma delta T Cells, J Invest Dermatol 140 (9 ) (2020) 1815–1823.32057839 [69] Shi ZR , Wu XS , Rocha CS , Rolston M , Garcia-Melchor E , Huynh M , Nguyen M , Law T , Haas KN , Yamada D , Millar NL , Wan YJY , Dandekar S , Hwang ST , Short-Term Western Diet Intake Promotes IL-23-Mediated Skin and Joint Inflammation Accompanied by Changes to the Gut Microbiota in Mice, J Invest Dermatol 141 (7 ) (2021) 1780–1791.33485880 [70] Goldberg EL , Molony RD , Kudo E , Sidorov S , Kong Y , Dixit VD , Iwasaki A , Ketogenic diet activates protective gamma delta T cell responses against influenza virus infection, Sci Immunol 4 (41 ) (2019). [71] Goldberg EL , Shchukina I , Asher JL , Sidorov S , Artyomov MN , Dixit VD , Ketogenesis activates metabolically protective gamma delta T cells in visceral adipose tissue, Nat Metab 2 (1 ) (2020) 50-+.32694683 [72] Sullivan ZA , Khoury-Hanold W , Lim J , Smillie C , Biton M , Reis BS , Zwick RK , Pope SD , Israni-Winger K , Parsa R , Philip NH , Rashed S , Palm N , Wang A , Mucida D , Regev A , Medzhitov R , gamma delta T cells regulate the intestinal response to nutrient sensing, Science 371 (6535 ) (2021) 1223-+. [73] Oh SF , Praveena T , Song H , Yoo JS , Jung DJ , Erturk-Hasdemir D , Hwang YS , Lee CWC , Le Nours JM , Kim H , Lee J , Blumberg RS , Rossjohn J , Park SB , Kasper DL , Host immunomodulatory lipids created by symbionts from dietary amino acids, Nature 600 (7888 ) (2021) 302-+.34759313 [74] von Gerichten J , Lamprecht D , Opalka L , Soulard D , Marsching C , Pilz R , Sencio V , Herzer S , Galy B , Nordstrom V , Hopf C , Grone HJ , Trottein F , Sandhoff R , Bacterial immunogenic alpha-galactosylceramide identified in the murine large intestine: dependency on diet and inflammation[S], J Lipid Res 60 (11 ) (2019) 1892–1904.31484693 [75] Iyer SS , Gensollen T , Gandhi A , Oh SF , Neves JF , Collin F , Lavin R , Serra C , Glickman J , de Silva PSA , Sartor RB , Besra G , Hauser R , Maxwell A , Llebaria A , Blumberg RS , Dietary and Microbial Oxazoles Induce Intestinal Inflammation by Modulating Aryl Hydrocarbon Receptor Responses, Cell 173 (5 ) (2018) 1123-+.29775592 [76] Lee S , Koh J , Chang Y , Kim HY , Chung DH , Invariant NKT Cells Functionally Link Microbiota-Induced Butyrate Production and Joint Inflammation, J Immunol 203 (12 ) (2019) 3199–3208.31732526 [77] Toubal A , Kiaf B , Beaudoin L , Cagninacci L , Rhimi M , Fruchet B , da Silva J , Corbett AJ , Simoni Y , Lantz O , Rossjohn J , McCluskey J , Lesnik P , Maguin E , Lehuen A , Mucosal-associated invariant T cells promote inflammation and intestinal dysbiosis leading to metabolic dysfunction during obesity, Nat Commun 11 (1 ) (2020). [78] Carolan E , Tobin LM , Mangan BA , Corrigan M , Gaoatswe G , Byrne G , Geoghegan J , Cody D , O’Connell J , Winter DC , Doherty DG , Lynch L , O’Shea D , Hogan AE , Altered Distribution and Increased IL-17 Production by Mucosal-Associated Invariant T Cells in Adult and Childhood Obesity, J Immunol 194 (12 ) (2015) 5775–5780.25980010 [79] Magalhaes I , Pingris K , Poitou C , Bessoles S , Venteclef N , Kiaf B , Beaudoin L , Da Silva J , Allatif O , Rossjohn J , Kjer-Nielsen L , McCluskey J , Ledoux S , Genser L , Torcivia A , Soudais C , Lantz O , Boitard C , Aron-Wisnewsky J , Larger E , Clement K , Lehuen A , Mucosal-associated invariant T cell alterations in obese and type 2 diabetic patients, J Clin Invest 125 (4 ) (2015) 1752–1762.25751065 [80] Keyes BE , Liu S , Asare A , Naik S , Levorse J , Polak L , Lu CP , Nikolova M , Pasolli HA , Fuchs E , Impaired Epidermal to Dendritic T Cell Signaling Slows Wound Repair in Aged Skin, Cell 167 (5 ) (2016) 1323–1338 e14.27863246 [81] Boyden LM , Lewis JM , Barbee SD , Bas A , Girardi M , Hayday AC , Tigelaar RE , Lifton RP , Skint1, the prototype of a newly identified immunoglobulin superfamily gene cluster, positively selects epidermal gammadelta T cells, Nat Genet 40 (5 ) (2008) 656–62.18408721 [82] Sharp LL , Jameson JM , Cauvi G , Havran WL , Dendritic epidermal T cells regulate skin homeostasis through local production of insulin-like growth factor 1, Nat Immunol 6 (1 ) (2005) 73–9.15592472 [83] Jameson J , Ugarte K , Chen N , Yachi P , Fuchs E , Boismenu R , Havran WL , A role for skin gammadelta T cells in wound repair, Science 296 (5568 ) (2002) 747–9.11976459 [84] Anderson LS , Yu S , Rivara KR , Reynolds MB , Hernandez AA , Wu X , Yang HY , Isseroff RR , Miller LS , Hwang ST , Simon SI , CCR6(+) gammadelta T Cells Home to Skin Wounds and Restore Normal Wound Healing in CCR6-Deficient Mice, J Invest Dermatol 139 (9 ) (2019) 2061–2064 e2.30935975 [85] Gay D , Kwon O , Zhang Z , Spata M , Plikus MV , Holler PD , Ito M , Yang Z , Treffeisen E , Kim CD , Nace A , Zhang X , Baratono S , Wang F , Ornitz DM , Millar SE , Cotsarelis G , Fgf9 from dermal gammadelta T cells induces hair follicle neogenesis after wounding, Nat Med 19 (7 ) (2013) 916–23.23727932 [86] MacLeod AS , Hemmers S , Garijo O , Chabod M , Mowen K , Witherden DA , Havran WL , Dendritic epidermal T cells regulate skin antimicrobial barrier function, J Clin Invest 123 (10 ) (2013) 4364–74.24051381 [87] Wei JJ , Kim HS , Spencer CA , Brennan-Crispi D , Zheng Y , Johnson NM , Rosenbach M , Miller C , Leung DH , Cotsarelis G , Leung TH , Activation of TRPA1 nociceptor promotes systemic adult mammalian skin regeneration, Sci Immunol 5 (50 ) (2020). [88] Konieczny P , Xing Y , Sidhu I , Subudhi I , Mansfield KP , Hsieh B , Biancur DE , Larsen SB , Cammer M , Li D , Landen NX , Loomis C , Heguy A , Tikhonova AN , Tsirigos A , Naik S , Interleukin-17 governs hypoxic adaptation of injured epithelium, Science (2022) eabg9302.35709248 [89] Elbe A , Foster CA , Stingl G , T-cell receptor alpha beta and gamma delta T cells in rat and human skin--are they equivalent?, Semin Immunol 8 (6 ) (1996) 341–9.8961385 [90] Ono T , Okamoto K , Nakashima T , Nitta T , Hori S , Iwakura Y , Takayanagi H , IL-17-producing gammadelta T cells enhance bone regeneration, Nat Commun 7 (2016) 10928.26965320 [91] Mann AO , Hanna BS , Munoz-Rojas AR , Sandrock I , Prinz I , Benoist C , Mathis D , IL-17A-producing gammadeltaT cells promote muscle regeneration in a microbiota-dependent manner, J Exp Med 219 (5 ) (2022). [92] Sun G , Yang S , Cao G , Wang Q , Hao J , Wen Q , Li Z , So KF , Liu Z , Zhou S , Zhao Y , Yang H , Zhou L , Yin Z , gammadelta T cells provide the early source of IFN-gamma to aggravate lesions in spinal cord injury, J Exp Med 215 (2 ) (2018) 521–535.29282251 [93] Guo XJ , Dash P , Crawford JC , Allen EK , Zamora AE , Boyd DF , Duan S , Bajracharya R , Awad WA , Apiwattanakul N , Vogel P , Kanneganti TD , Thomas PG , Lung gammadelta T Cells Mediate Protective Responses during Neonatal Influenza Infection that Are Associated with Type 2 Immunity, Immunity 49 (3 ) (2018) 531–544 e6.30170813 [94] Krishnan S , Prise IE , Wemyss K , Schenck LP , Bridgeman HM , McClure FA , Zangerle-Murray T , O’Boyle C , Barbera TA , Mahmood F , Bowdish DME , Zaiss DMW , Grainger JR , Konkel JE , Amphiregulin-producing gammadelta T cells are vital for safeguarding oral barrier immune homeostasis, Proc Natl Acad Sci U S A 115 (42 ) (2018) 10738–10743.30279177 [95] Cai Y , Shen X , Ding C , Qi C , Li K , Li X , Jala VR , Zhang HG , Wang T , Zheng J , Yan J , Pivotal role of dermal IL-17-producing gammadelta T cells in skin inflammation, Immunity 35 (4 ) (2011) 596–610.21982596 [96] Leng T , Akther HD , Hackstein CP , Powell K , King T , Friedrich M , Christoforidou Z , McCuaig S , Neyazi M , Arancibia-Carcamo CV , Hagel J , Powrie F , Oxford IBDI , Peres RS , Millar V , Ebner D , Lamichhane R , Ussher J , Hinks TSC , Marchi E , Willberg C , Klenerman P , TCR and Inflammatory Signals Tune Human MAIT Cells to Exert Specific Tissue Repair and Effector Functions, Cell Rep 28 (12 ) (2019) 3077–3091 e5.31533032 [97] Rognoni E , Watt FM , Skin Cell Heterogeneity in Development, Wound Healing, and Cancer, Trends Cell Biol 28 (9 ) (2018) 709–722.29807713 [98] Hinks TSC , Marchi E , Jabeen M , Olshansky M , Kurioka A , Pediongco TJ , Meehan BS , Kostenko L , Turner SJ , Corbett AJ , Chen Z , Klenerman P , McCluskey J , Activation and In Vivo Evolution of the MAIT Cell Transcriptome in Mice and Humans Reveals Tissue Repair Functionality, Cell Rep 28 (12 ) (2019) 3249–3262 e5.31533045 [99] Schneider DF , Palmer JL , Tulley JM , Speicher JT , Kovacs EJ , Gamelli RL , Faunce DE , A novel role for NKT cells in cutaneous wound repair, J Surg Res 168 (2 ) (2011) 325–33 e1.20089261 [100] Schneider DF , Palmer JL , Tulley JM , Kovacs EJ , Gamelli RL , Faunce DE , Prevention of NKT cell activation accelerates cutaneous wound closure and alters local inflammatory signals, J Surg Res 171 (1 ) (2011) 361–73.21067780 [101] Tanno H , Kawakami K , Ritsu M , Kanno E , Suzuki A , Kamimatsuno R , Takagi N , Miyasaka T , Ishii K , Imai Y , Maruyama R , Tachi M , Contribution of Invariant Natural Killer T Cells to Skin Wound Healing, Am J Pathol 185 (12 ) (2015) 3248–57.26468976 [102] Lee YJ , Wang H , Starrett GJ , Phuong V , Jameson SC , Hogquist KA , Tissue-Specific Distribution of iNKT Cells Impacts Their Cytokine Response, Immunity 43 (3 ) (2015) 566–78.26362265 [103] Doisne JM , Becourt C , Amniai L , Duarte N , Le Luduec JB , Eberl G , Benlagha K , Skin and peripheral lymph node invariant NKT cells are mainly retinoic acid receptor-related orphan receptor (gamma)t+ and respond preferentially under inflammatory conditions, J Immunol 183 (3 ) (2009) 2142–9.19587013 [104] Constantinides MG , Bendelac A , Transcriptional regulation of the NKT cell lineage, Curr Opin Immunol 25 (2 ) (2013) 161–7.23402834 [105] Liew PX , Lee WY , Kubes P , iNKT Cells Orchestrate a Switch from Inflammation to Resolution of Sterile Liver Injury, Immunity 47 (4 ) (2017) 752–765 e5.29045904 [106] Cosway EJ , White AJ , Parnell SM , Schweighoffer E , Jolin HE , Bacon A , Rodewald HR , Tybulewicz V , McKenzie ANJ , Jenkinson WE , Anderson G , Eosinophils are an essential element of a type 2 immune axis that controls thymus regeneration, Sci Immunol 7 (69 ) (2022) eabn3286.35275754 [107] Linehan JL , Harrison OJ , Han SJ , Byrd AL , Vujkovic-Cvijin I , Villarino AV , Sen SK , Shaik J , Smelkinson M , Tamoutounour S , Collins N , Bouladoux N , Dzutsev A , Rosshart SP , Arbuckle JH , Wang CR , Kristie TM , Rehermann B , Trinchieri G , Brenchley JM , O’Shea JJ , Belkaid Y , Non-classical Immunity Controls Microbiota Impact on Skin Immunity and Tissue Repair, Cell 172 (4 ) (2018) 784–796 e18.29358051 [108] Naik S , Bouladoux N , Linehan JL , Han SJ , Harrison OJ , Wilhelm C , Conlan S , Himmelfarb S , Byrd AL , Deming C , Quinones M , Brenchley JM , Kong HH , Tussiwand R , Murphy KM , Merad M , Segre JA , Belkaid Y , Commensal-dendritic-cell interaction specifies a unique protective skin immune signature, Nature 520 (7545 ) (2015) 104–8.25539086 [109] Lee M , Lee E , Han SK , Choi YH , Kwon DI , Choi H , Lee K , Park ES , Rha MS , Joo DJ , Shin EC , Kim S , Kim JK , Lee YJ , Single-cell RNA sequencing identifies shared differentiation paths of mouse thymic innate T cells, Nat Commun 11 (1 ) (2020) 4367.32868763 [110] Chandra S , Kronenberg M , Activation and Function of iNKT and MAIT Cells, Adv. Immunol 127 (2015) 145–201.26073984 [111] Godfrey DI , Koay HF , McCluskey J , Gherardin NA , The biology and functional importance of MAIT cells, Nat. Immunol 20 (9 ) (2019) 1110–1128.31406380 [112] Gracey E , Qaiyum Z , Almaghlouth I , Lawson D , Karki S , Avvaru N , Zhang Z , Yao Y , Ranganathan V , Baglaenko Y , Inman RD , IL-7 primes IL-17 in mucosal-associated invariant T (MAIT) cells, which contribute to the Th17-axis in ankylosing spondylitis, Ann. Rheum. Dis 75 (12 ) (2016) 2124–2132.27165176 [113] Kim M , Yoo SJ , Kang SW , Kwon J , Choi I , Lee CH , TNFalpha and IL-1beta in the synovial fluid facilitate mucosal-associated invariant T (MAIT) cell migration, Cytokine 99 (2017) 91–98.28756336 [114] Wallace KL , Zheng LB , Kanazawa Y , Shih DQ , Immunopathology of inflammatory bowel disease, World J. Gastroenterol 20 (1 ) (2014) 6–21.24415853 [115] Serriari NE , Eoche M , Lamotte L , Lion J , Fumery M , Marcelo P , Chatelain D , Barre A , Nguyen-Khac E , Lantz O , Dupas JL , Treiner E , Innate mucosal-associated invariant T (MAIT) cells are activated in inflammatory bowel diseases, Clin Exp Immunol 176 (2 ) (2014) 266–74.24450998 [116] Haga K , Chiba A , Shibuya T , Osada T , Ishikawa D , Kodani T , Nomura O , Watanabe S , Miyake S , MAIT cells are activated and accumulated in the inflamed mucosa of ulcerative colitis, J Gastroenterol Hepatol 31 (5 ) (2016) 965–72.26590105 [117] Tominaga K , Yamagiwa S , Setsu T , Kimura N , Honda H , Kamimura H , Honda Y , Takamura M , Yokoyama J , Suzuki K , Wakai T , Terai S , Possible involvement of mucosal-associated invariant T cells in the progression of inflammatory bowel diseases, Biomed. Res 38 (2017) 111–121.28442662 [118] Eyerich K , Dimartino V , Cavani A , IL-17 and IL-22 in immunity: Driving protection and pathology, Eur. J. Immunol 47 (4 ) (2017) 607–614.28295238 [119] Yasutomi Y , Chiba A , Haga K , Murayama G , Makiyama A , Kuga T , Watanabe M , Okamoto R , Nagahara A , Nagaishi T , Miyake S , Activated Mucosal-associated Invariant T Cells Have a Pathogenic Role in a Murine Model of Inflammatory Bowel Disease, Cell Mol Gastroenterol Hepatol 13 (1 ) (2022) 81–93.34461283 [120] Chen Y , Chou K , Fuchs E , Havran WL , Boismenu R , Protection of the intestinal mucosa by intraepithelial gamma delta T cells, Proc Natl Acad Sci U S A 99 (22 ) (2002) 14338–43.12376619 [121] Jaeger N , Gamini R , Cella M , Schettini JL , Bugatti M , Zhao S , Rosadini CV , Esaulova E , Di Luccia B , Kinnett B , Vermi W , Artyomov MN , Wynn TA , Xavier RJ , Jelinsky SA , Colonna M , Single-cell analyses of Crohn’s disease tissues reveal intestinal intraepithelial T cells heterogeneity and altered subset distributions, Nat Commun 12 (1 ) (2021) 1921.33771991 [122] Liao CM , Zimmer MI , Wang CR , The functions of type I and type II natural killer T cells in inflammatory bowel diseases, Inflamm. Bowel Dis 19 (6 ) (2013) 1330–8.23518808 [123] Rosati E , Rios Martini G , Pogorelyy MV , Minervina AA , Degenhardt F , Wendorff M , Sari S , Mayr G , Fazio A , Dowds CM , Hauser C , Tran F , von Schonfels W , Pochhammer J , Salnikova MA , Jaeckel C , Gigla JB , Sabet SS , Hubenthal M , Schiminsky E , Schreiber S , Rosenstiel PC , Scheffold A , Thomas PG , Lieb W , Bokemeyer B , Witte M , Aden K , Hendricks A , Schafmayer C , Egberts JH , Mamedov IZ , Bacher P , Franke A , A novel unconventional T cell population enriched in Crohn’s disease, Gut (2022). [124] Boland BS , He Z , Tsai MS , Olvera JG , Omilusik KD , Duong HG , Kim ES , Limary AE , Jin W , Milner JJ , Yu B , Patel SA , Louis TL , Tysl T , Kurd NS , Bortnick A , Quezada LK , Kanbar JN , Miralles A , Huylebroeck D , Valasek MA , Dulai PS , Singh S , Lu L-F , Bui JD , Murre C , Sandborn WJ , Goldrath AW , Yeo GW , Chang JT , Heterogeneity and clonal relationships of adaptive immune cells in ulcerative colitis revealed by single-cell analyses, Science Immunology 5 (50 ) (2020) eabb4432.32826341 [125] Hinks TS , Zhou X , Staples KJ , Dimitrov BD , Manta A , Petrossian T , Lum PY , Smith CG , Ward JA , Howarth PH , Walls AF , Gadola SD , Djukanovic R , Innate and adaptive T cells in asthmatic patients: Relationship to severity and disease mechanisms, J Allergy Clin Immunol 136 (2 ) (2015) 323–33.25746968 [126] Chandra S , Wingender G , Greenbaum JA , Khurana A , Gholami AM , Ganesan AP , Rosenbach M , Jaffee K , Gern JE , Wood R , O’Connor G , Sandel M , Kattan M , Bacharier L , Togias A , Horner AA , Kronenberg M , Development of Asthma in Inner-City Children: Possible Roles of MAIT Cells and Variation in the Home Environment, J Immunol 200 (6 ) (2018) 1995–2003.29431692 [127] Lezmi G , Abou Taam R , Dietrich C , Chatenoud L , de Blic J , Leite-de-Moraes M , Circulating IL-17-producing mucosal-associated invariant T cells (MAIT) are associated with symptoms in children with asthma, Clin Immunol 188 (2018) 7–11.29196148 [128] Ye L , Pan J , Pasha MA , Shen X , D’Souza SS , Fung ITH , Wang Y , Guo B , Tang DD , Yang Q , Mucosal-associated invariant T cells restrict allergic airway inflammation, J Allergy Clin Immunol 145 (5 ) (2020) 1469–1473 e4.31874183 [129] Wingender G , Rogers P , Batzer G , Lee MS , Bai D , Pei B , Khurana A , Kronenberg M , Horner AA , Invariant NKT cells are required for airway inflammation induced by environmental antigens, J. Exp. Med 208 (6 ) (2011) 1151–62.21624935 [130] Meyer EH , DeKruyff RH , Umetsu DT , T cells and NKT cells in the pathogenesis of asthma, Annu. Rev. Med 59 (2008) 281–92.17937589 [131] Li J , Reantragoon R , Kostenko L , Corbett AJ , Varigos G , Carbone FR , The frequency of mucosal-associated invariant T cells is selectively increased in dermatitis herpetiformis, Australas J Dermatol 58 (3 ) (2017) 200–204.26940855 [132] Teunissen MBM , Yeremenko NG , Baeten DLP , Chielie S , Spuls PI , de Rie MA , Lantz O , Res PCM , The IL-17A-producing CD8+ T-cell population in psoriatic lesional skin comprises mucosa-associated invariant T cells and conventional T cells, J Invest Dermatol 134 (12 ) (2014) 2898–2907.24945094 [133] Polese B , Zhang H , Thurairajah B , King IL , Innate Lymphocytes in Psoriasis, Front. Immunol 11 (2020) 242.32153574 [134] Han Y , Mora J , Huard A , da Silva P , Wiechmann S , Putyrski M , Schuster C , Elwakeel E , Lang G , Scholz A , Scholz T , Schmid T , de Bruin N , Billuart P , Sala C , Burkhardt H , Parnham MJ , Ernst A , Brune B , Weigert A , IL-38 Ameliorates Skin Inflammation and Limits IL-17 Production from gammadelta T Cells, Cell Rep. 27 (3 ) (2019) 835–846 e5.30995480 [135] Boutet MA , Bart G , Penhoat M , Amiaud J , Brulin B , Charrier C , Morel F , Lecron JC , Rolli-Derkinderen M , Bourreille A , Vigne S , Gabay C , Palmer G , Le Goff B , Blanchard F , Distinct expression of interleukin (IL)-36alpha, beta and gamma, their antagonist IL-36Ra and IL-38 in psoriasis, rheumatoid arthritis and Crohn’s disease, Clin. Exp. Immunol 184 (2 ) (2016) 159–73.26701127 [136] Mortier C , Govindarajan S , Venken K , Elewaut D , It Takes “Guts” to Cause Joint Inflammation: Role of Innate-Like T Cells, Front. Immunol 9 (2018) 1489.30008717 [137] Hayashi E , Chiba A , Tada K , Haga K , Kitagaichi M , Nakajima S , Kusaoi M , Sekiya F , Ogasawara M , Yamaji K , Tamura N , Takasaki Y , Miyake S , Involvement of Mucosal-associated Invariant T cells in Ankylosing Spondylitis, J. Rheumatol 43 (9 ) (2016) 1695–703.27370879 [138] Toussirot E , Laheurte C , Gaugler B , Gabriel D , Saas P , Increased IL-22- and IL-17A-Producing Mucosal-Associated Invariant T Cells in the Peripheral Blood of Patients With Ankylosing Spondylitis, Front. Immunol 9 (2018) 1610.30057583 [139] Chiba A , Tajima R , Tomi C , Miyazaki Y , Yamamura T , Miyake S , Mucosal-associated invariant T cells promote inflammation and exacerbate disease in murine models of arthritis, Arthritis Rheum. 64 (1 ) (2012) 153–61.21904999 [140] Ito Y , Usui T , Kobayashi S , Iguchi-Hashimoto M , Ito H , Yoshitomi H , Nakamura T , Shimizu M , Kawabata D , Yukawa N , Hashimoto M , Sakaguchi N , Sakaguchi S , Yoshifuji H , Nojima T , Ohmura K , Fujii T , Mimori T , Gamma/delta T cells are the predominant source of interleukin-17 in affected joints in collagen-induced arthritis, but not in rheumatoid arthritis, Arthritis Rheum. 60 (8 ) (2009) 2294–303.19644886 [141] Kim HY , Kim HJ , Min HS , Kim S , Park WS , Park SH , Chung DH , NKT cells promote antibody-induced joint inflammation by suppressing transforming growth factor beta1 production, J. Exp. Med 201 (1 ) (2005) 41–7.15630137 [142] Chiba A , Kaieda S , Oki S , Yamamura T , Miyake S , The involvement of V(alpha)14 natural killer T cells in the pathogenesis of arthritis in murine models, Arthritis Rheum. 52 (6 ) (2005) 1941–8.15934073 [143] Coppieters K , Van Beneden K , Jacques P , Dewint P , Vervloet A , Vander Cruyssen B , Van Calenbergh S , Chen G , Franck RW , Verbruggen G , Deforce D , Matthys P , Tsuji M , Rottiers P , Elewaut D , A single early activation of invariant NK T cells confers long-term protection against collagen-induced arthritis in a ligand-specific manner, J. Immunol 179 (4 ) (2007) 2300–9.17675491 [144] Jacques P , Venken K , Van Beneden K , Hammad H , Seeuws S , Drennan MB , Deforce D , Verbruggen G , Apostolaki M , Kollias G , Lambrecht BN , De Vos M , Elewaut D , Invariant natural killer T cells are natural regulators of murine spondylarthritis, Arthritis Rheum. 62 (4 ) (2010) 988–99.20131252 [145] Hirsova P , Bamidele AO , Wang H , Povero D , Revelo XS , Emerging Roles of T Cells in the Pathogenesis of Nonalcoholic Steatohepatitis and Hepatocellular Carcinoma, Front. Endocrinol. (Lausanne) 12 (2021) 760860.34777255 [146] Heymann F , Tacke F , Immunology in the liver--from homeostasis to disease, Nat. Rev. Gastroenterol. Hepatol 13 (2 ) (2016) 88–110.26758786 [147] Van Herck MA , Weyler J , Kwanten WJ , Dirinck EL , De Winter BY , Francque SM , Vonghia L , The Differential Roles of T Cells in Non-alcoholic Fatty Liver Disease and Obesity, Front. Immunol. 10 (2019) 82.30787925 [148] Syn WK , Oo YH , Pereira TA , Karaca GF , Jung Y , Omenetti A , Witek RP , Choi SS , Guy CD , Fearing CM , Teaberry V , Pereira FE , Adams DH , Diehl AM , Accumulation of natural killer T cells in progressive nonalcoholic fatty liver disease, Hepatology 51 (6 ) (2010) 1998–2007.20512988 [149] Hammerich L , Tacke F , Role of gamma-delta T cells in liver inflammation and fibrosis, World J. Gastrointest. Pathophysiol 5 (2 ) (2014) 107–13.24891982 [150] Meng F , Wang K , Aoyama T , Grivennikov SI , Paik Y , Scholten D , Cong M , Iwaisako K , Liu X , Zhang M , Osterreicher CH , Stickel F , Ley K , Brenner DA , Kisseleva T , Interleukin-17 signaling in inflammatory, Kupffer cells, and hepatic stellate cells exacerbates liver fibrosis in mice, Gastroenterology 143 (3 ) (2012) 765–776 e3.22687286 [151] Tan Z , Qian X , Jiang R , Liu Q , Wang Y , Chen C , Wang X , Ryffel B , Sun B , IL-17A plays a critical role in the pathogenesis of liver fibrosis through hepatic stellate cell activation, J. Immunol 191 (4 ) (2013) 1835–44.23842754 [152] Hammerich L , Bangen JM , Govaere O , Zimmermann HW , Gassler N , Huss S , Liedtke C , Prinz I , Lira SA , Luedde T , Roskams T , Trautwein C , Heymann F , Tacke F , Chemokine receptor CCR6-dependent accumulation of gammadelta T cells in injured liver restricts hepatic inflammation and fibrosis, Hepatology 59 (2 ) (2014) 630–42.23959575 [153] Hegde P , Weiss E , Paradis V , Wan J , Mabire M , Sukriti S , Rautou PE , Albuquerque M , Picq O , Gupta AC , Ferrere G , Gilgenkrantz H , Kiaf B , Toubal A , Beaudoin L , Letteron P , Moreau R , Lehuen A , Lotersztajn S , Mucosal-associated invariant T cells are a profibrogenic immune cell population in the liver, Nat Commun 9 (1 ) (2018) 2146.29858567 [154] Crosby CM , Kronenberg M , Tissue-specific functions of invariant natural killer T cells, Nat Rev Immunol 18 (9 ) (2018) 559–574.29967365 [155] Zhu S , Zhang H , Bai L , NKT cells in liver diseases, Front Med 12 (3 ) (2018) 249–261.29623561 [156] Schmieg J , Yang G , Franck RW , Van Rooijen N , Tsuji M , Glycolipid presentation to natural killer T cells differs in an organ-dependent fashion, Proc Natl Acad Sci U S A 102 (4 ) (2005) 1127–32.15644449 [157] Lee WY , Moriarty TJ , Wong CH , Zhou H , Strieter RM , van Rooijen N , Chaconas G , Kubes P , An intravascular immune response to Borrelia burgdorferi involves Kupffer cells and iNKT cells, Nat Immunol 11 (4 ) (2010) 295–302.20228796 [158] Yang L , Jhaveri R , Huang J , Qi Y , Diehl AM , Endoplasmic reticulum stress, hepatocyte CD1d and NKT cell abnormalities in murine fatty livers, Lab Invest 87 (9 ) (2007) 927–37.17607300 [159] Zeissig S , Peuker K , Iyer S , Gensollen T , Dougan SK , Olszak T , Kaser A , Blumberg RS , CD1d-Restricted pathways in hepatocytes control local natural killer T cell homeostasis and hepatic inflammation, Proc Natl Acad Sci U S A 114 (39 ) (2017) 10449–10454.28893990 [160] Geissmann F , Cameron TO , Sidobre S , Manlongat N , Kronenberg M , Briskin MJ , Dustin ML , Littman DR , Intravascular immune surveillance by CXCR6+ NKT cells patrolling liver sinusoids, PLoS Biol 3 (4 ) (2005) e113.15799695 [161] Schrumpf E , Tan C , Karlsen TH , Sponheim J , Bjorkstrom NK , Sundnes O , Alfsnes K , Kaser A , Jefferson DM , Ueno Y , Eide TJ , Haraldsen G , Zeissig S , Exley MA , Blumberg RS , Melum E , The biliary epithelium presents antigens to and activates natural killer T cells, Hepatology 62 (4 ) (2015) 1249–59.25855031 [162] Wehr A , Baeck C , Heymann F , Niemietz PM , Hammerich L , Martin C , Zimmermann HW , Pack O , Gassler N , Hittatiya K , Ludwig A , Luedde T , Trautwein C , Tacke F , Chemokine receptor CXCR6-dependent hepatic NK T Cell accumulation promotes inflammation and liver fibrosis, J Immunol 190 (10 ) (2013) 5226–36.23596313 [163] Hua J , Ma X , Webb T , Potter JJ , Oelke M , Li Z , Dietary fatty acids modulate antigen presentation to hepatic NKT cells in nonalcoholic fatty liver disease, J Lipid Res 51 (7 ) (2010) 1696–703.20185414 [164] Kremer M , Thomas E , Milton RJ , Perry AW , van Rooijen N , Wheeler MD , Zacks S , Fried M , Rippe RA , Hines IN , Kupffer cell and interleukin-12-dependent loss of natural killer T cells in hepatosteatosis, Hepatology 51 (1 ) (2010) 130–41.20034047 [165] Li Z , Oben JA , Yang S , Lin H , Stafford EA , Soloski MJ , Thomas SA , Diehl AM , Norepinephrine regulates hepatic innate immune system in leptin-deficient mice with nonalcoholic steatohepatitis, Hepatology 40 (2 ) (2004) 434–41.15368448 [166] Liu K , Yang L , Wang G , Liu J , Zhao X , Wang Y , Li J , Yang J , Metabolic stress drives sympathetic neuropathy within the liver, Cell Metab 33 (3 ) (2021) 666–675 e4.33545051 [167] Lynch L , Hogan AE , Duquette D , Lester C , Banks A , LeClair K , Cohen DE , Ghosh A , Lu B , Corrigan M , Stevanovic D , Maratos-Flier E , Drucker DJ , O’Shea D , Brenner M , iNKT Cells Induce FGF21 for Thermogenesis and Are Required for Maximal Weight Loss in GLP1 Therapy, Cell Metab 24 (3 ) (2016) 510–519.27593966 [168] Geng L , Lam KSL , Xu A , The therapeutic potential of FGF21 in metabolic diseases: from bench to clinic, Nat Rev Endocrinol 16 (11 ) (2020) 654–667.32764725 [169] Rakhshandehroo M , van Eijkeren RJ , Gabriel TL , de Haar C , Gijzel SMW , Hamers N , Ferraz MJ , Aerts J , Schipper HS , van Eijk M , Boes M , Kalkhoven E , Adipocytes harbor a glucosylceramide biosynthesis pathway involved in iNKT cell activation, Biochim Biophys Acta Mol Cell Biol Lipids 1864 (8 ) (2019) 1157–1167.31051284 [170] Satoh M , Hoshino M , Fujita K , Iizuka M , Fujii S , Clingan CS , Van Kaer L , Iwabuchi K , Adipocyte-specific CD1d-deficiency mitigates diet-induced obesity and insulin resistance in mice, Sci Rep-Uk 6 (2016). [171] Huh JY , Park J , Kim JI , Park YJ , Lee YK , Kim JB , Deletion of CD1d in Adipocytes Aggravates Adipose Tissue Inflammation and Insulin Resistance in Obesity, Diabetes 66 (4 ) (2017) 835–847.28082459 [172] Park YJ , Park J , Huh JY , Hwang I , Choe SS , Kim JB , Regulatory Roles of Invariant Natural Killer T Cells in Adipose Tissue Inflammation: Defenders Against Obesity-Induced Metabolic Complications, Frontiers in Immunology 9 (2018). [173] Satoh M , Iwabuchi K , Role of Natural Killer T Cells in the Development of Obesity and Insulin Resistance: Insights From Recent Progress, Front Immunol 9 (2018) 1314.29942311 [174] van Eijkeren RJ , Morris I , Borgman A , Markovska A , Kalkhoven E , Cytokine Output of Adipocyte-iNKT Cell Interplay Is Skewed by a Lipid-Rich Microenvironment, Front Endocrinol 11 (2020). [175] Havran WL , Allison JP , Developmentally Ordered Appearance of Thymocytes Expressing Different T-Cell Antigen Receptors, Nature 335 (6189 ) (1988) 443–445.2458531 [176] Hu W , Shang R , Yang J , Chen C , Liu Z , Liang G , He W , Luo G , Skin gammadelta T Cells and Their Function in Wound Healing, Front Immunol 13 (2022) 875076.35479079 [177] Dalessandri T , Crawford G , Hayes M , Castro Seoane R , Strid J , IL-13 from intraepithelial lymphocytes regulates tissue homeostasis and protects against carcinogenesis in the skin, Nat Commun 7 (2016) 12080.27357235 [178] Boismenu R , Havran WL , Modulation of epithelial cell growth by intraepithelial gamma delta T cells, Science 266 (5188 ) (1994) 1253–5.7973709 [179] Werner S , Peters KG , Longaker MT , Fullerpace F , Banda MJ , Williams LT , Large Induction of Keratinocyte Growth-Factor Expression in the Dermis during Wound-Healing, P Natl Acad Sci USA 89 (15 ) (1992) 6896–6900. [180] Takashima A , Matsue H , Bergstresser PR , Ariizumi K , Interleukin-7-dependent interaction of dendritic epidermal T cells with keratinocytes, J Invest Dermatol 105 (1 Suppl ) (1995) 50S–53S.7615997 [181] De Creus A , Van Beneden K , Stevenaert F , Debacker V , Plum J , Leclercq G , Developmental and functional defects of thymic and epidermal V gamma 3 cells in IL-15-deficient and IFN regulatory factor-1-deficient mice, J Immunol 168 (12 ) (2002) 6486–93.12055269 [182] Barbee SD , Woodward MJ , Turchinovich G , Mention JJ , Lewis JM , Boyden LM , Lifton RP , Tigelaar R , Hayday AC , Skint-1 is a highly specific, unique selecting component for epidermal T cells, Proc Natl Acad Sci U S A 108 (8 ) (2011) 3330–5.21300860 [183] McKenzie DR , Hart R , Bah N , Ushakov DS , Munoz-Ruiz M , Feederle R , Hayday AC , Normality sensing licenses local T cells for innate-like tissue surveillance, Nat Immunol 23 (3 ) (2022) 411–422.35165446 [184] Komori HK , Witherden DA , Kelly R , Sendaydiego K , Jameson JM , Teyton L , Havran WL , Cutting edge: dendritic epidermal gammadelta T cell ligands are rapidly and locally expressed by keratinocytes following cutaneous wounding, J Immunol 188 (7 ) (2012) 2972–6.22393149 [185] Di Marco Barros R , Roberts NA , Dart RJ , Vantourout P , Jandke A , Nussbaumer O , Deban L , Cipolat S , Hart R , Iannitto ML , Laing A , Spencer-Dene B , East P , Gibbons D , Irving PM , Pereira P , Steinhoff U , Hayday A , Epithelia Use Butyrophilin-like Molecules to Shape Organ-Specific gammadelta T Cell Compartments, Cell 167 (1 ) (2016) 203–218 e17.27641500 [186] Lee P , Gund R , Dutta A , Pincha N , Rana I , Ghosh S , Witherden D , Kandyba E , MacLeod A , Kobielak K , Havran WL , Jamora C , Stimulation of hair follicle stem cell proliferation through an IL-1 dependent activation of gammadeltaT-cells, Elife 6 (2017). [187] Muire PJ , Avila JJ , Lofgren AL , Wenke JC , Neutralization of HMGB1 improves fracture healing and gammadelta T lymphocyte counts at the fracture site in a polytrauma rat model, J Exp Orthop 9 (1 ) (2022) 21.35229226 [188] Kohlgruber AC , Gal-Oz ST , LaMarche NM , Shimazaki M , Duquette D , Koay HF , Nguyen HN , Mina AI , Paras T , Tavakkoli A , von Andrian U , Uldrich AP , Godfrey DI , Banks AS , Shay T , Brenner MB , Lynch L , gammadelta T cells producing interleukin-17A regulate adipose regulatory T cell homeostasis and thermogenesis, Nat Immunol 19 (5 ) (2018) 464–474.29670241 [189] Hu B , Jin C , Zeng X , Resch JM , Jedrychowski MP , Yang Z , Desai BN , Banks AS , Lowell BB , Mathis D , Spiegelman BM , gammadelta T cells and adipocyte IL-17RC control fat innervation and thermogenesis, Nature 578 (7796 ) (2020) 610–614.32076265 [190] Ribeiro M , Brigas HC , Temido-Ferreira M , Pousinha PA , Regen T , Santa C , Coelho JE , Marques-Morgado I , Valente CA , Omenetti S , Stockinger B , Waisman A , Manadas B , Lopes LV , Silva-Santos B , Ribot JC , Meningeal gammadelta T cell-derived IL-17 controls synaptic plasticity and short-term memory, Sci Immunol 4 (40 ) (2019). [191] Riol-Blanco L , Ordovas-Montanes J , Perro M , Naval E , Thiriot A , Alvarez D , Paust S , Wood JN , von Andrian UH , Nociceptive sensory neurons drive interleukin-23-mediated psoriasiform skin inflammation, Nature 510 (7503 ) (2014) 157–61.24759321 [192] Baral P , Umans BD , Li L , Wallrapp A , Bist M , Kirschbaum T , Wei Y , Zhou Y , Kuchroo VK , Burkett PR , Yipp BG , Liberles SD , Chiu IM , Nociceptor sensory neurons suppress neutrophil and gammadelta T cell responses in bacterial lung infections and lethal pneumonia, Nat Med 24 (4 ) (2018) 417–426.29505031 [193] Kurioka A , Ussher JE , Cosgrove C , Clough C , Fergusson JR , Smith K , Kang YH , Walker LJ , Hansen TH , Willberg CB , Klenerman P , MAIT cells are licensed through granzyme exchange to kill bacterially sensitized targets, Mucosal Immunol 8 (2 ) (2015) 429–40.25269706 [194] Lamichhane R , Schneider M , de la Harpe SM , Harrop TWR , Hannaway RF , Dearden PK , Kirman JR , Tyndall JDA , Vernall AJ , Ussher JE , TCR- or Cytokine-Activated CD8(+) Mucosal-Associated Invariant T Cells Are Rapid Polyfunctional Effectors That Can Coordinate Immune Responses, Cell Rep 28 (12 ) (2019) 3061–3076 e5.31533031 [195] Le Bourhis L , Dusseaux M , Bohineust A , Bessoles S , Martin E , Premel V , Core M , Sleurs D , Serriari NE , Treiner E , Hivroz C , Sansonetti P , Gougeon ML , Soudais C , Lantz O , MAIT cells detect and efficiently lyse bacterially-infected epithelial cells, PLoS Pathog 9 (10 ) (2013) e1003681.24130485 [196] Booth JS , Salerno-Goncalves R , Blanchard TG , Patil SA , Kader HA , Safta AM , Morningstar LM , Czinn SJ , Greenwald BD , Sztein MB , Mucosal-Associated Invariant T Cells in the Human Gastric Mucosa and Blood: Role in Helicobacter pylori Infection, Front Immunol 6 (2015) 466.26441971 [197] Phetsouphanh C , Phalora P , Hackstein CP , Thornhill J , Munier CML , Meyerowitz J , Murray L , VanVuuren C , Goedhals D , Drexhage L , Moore R , Sattentau QJ , Mak JY , Fairlie DP , Fidler S , Kelleher AD , Frater J , Klenerman P , Human MAIT cells respond to and suppress HIV-1, Elife 10 (2021). [198] Hengst J , Strunz B , Deterding K , Ljunggren HG , Leeansyah E , Manns MP , Cornberg M , Sandberg JK , Wedemeyer H , Bjorkstrom NK , Nonreversible MAIT cell-dysfunction in chronic hepatitis C virus infection despite successful interferon-free therapy, Eur J Immunol 46 (9 ) (2016) 2204–10.27296288 [199] Rouxel O , Da Silva J , Beaudoin L , Nel I , Tard C , Cagninacci L , Kiaf B , Oshima M , Diedisheim M , Salou M , Corbett A , Rossjohn J , McCluskey J , Scharfmann R , Battaglia M , Polak M , Lantz O , Beltrand J , Lehuen A , Cytotoxic and regulatory roles of mucosal-associated invariant T cells in type 1 diabetes, Nat Immunol 18 (12 ) (2017) 1321–1331.28991267 [200] Flament H , Rouland M , Beaudoin L , Toubal A , Bertrand L , Lebourgeois S , Rousseau C , Soulard P , Gouda Z , Cagninacci L , Monteiro AC , Hurtado-Nedelec M , Luce S , Bailly K , Andrieu M , Saintpierre B , Letourneur F , Jouan Y , Si-Tahar M , Baranek T , Paget C , Boitard C , Vallet-Pichard A , Gautier JF , Ajzenberg N , Terrier B , Pene F , Ghosn J , Lescure X , Yazdanpanah Y , Visseaux B , Descamps D , Timsit JF , Monteiro RC , Lehuen A , Outcome of SARS-CoV-2 infection is linked to MAIT cell activation and cytotoxicity, Nat Immunol 22 (3 ) (2021) 322–335.33531712 [201] Toubal A , Kiaf B , Beaudoin L , Cagninacci L , Rhimi M , Fruchet B , da Silva J , Corbett AJ , Simoni Y , Lantz O , Rossjohn J , McCluskey J , Lesnik P , Maguin E , Lehuen A , Mucosal-associated invariant T cells promote inflammation and intestinal dysbiosis leading to metabolic dysfunction during obesity, Nat Commun 11 (1 ) (2020) 3755.32709874 [202] Bertrand L , Lehuen A , MAIT cells in metabolic diseases, Mol Metab 27S (2019) S114–S121.31500822 [203] Boudinot P , Mondot S , Jouneau L , Teyton L , Lefranc MP , Lantz O , Restricting nonclassical MHC genes coevolve with TRAV genes used by innate-like T cells in mammals, Proc Natl Acad Sci U S A 113 (21 ) (2016) E2983–92.27170188

PMC009xxxxxx/PMC9829490.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7608054 7632 Suicide Life Threat Behav Suicide Life Threat Behav Suicide & life-threatening behavior 0363-0234 1943-278X 34254694 9829490 10.1111/sltb.12790 NIHMS1853594 Article Proximal Correlates of Suicidal Ideation among Transgender and Gender Diverse People: A Preliminary Test of the Three Step Theory Wolford-Clevenger Caitlin Ph.D *University of Tennessee-Knoxville University of Alabama at Birmingham Flores Leticia Y. Ph.D. University of Tennessee-Knoxville Stuart Gregory L. Ph.D. University of Tennessee-Knoxville * Dr. Wolford-Clevenger conducted the work while at the University of Tennessee-Knoxville at and at the University of Alabama at Birmingham She is presently affiliated with the University of Alabama at Birmingham 4 1 2023 12 2021 13 7 2021 09 1 2023 51 6 10771085 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Introduction: Transgender and gender diverse people experience higher rates of suicidal ideation than their cisgender peers; however, very little is known about factors that proximally relate to suicidal ideation in this population. This limited understanding may be due to the lack of theory-guided studies that are capable of testing proximal correlates of suicidal ideation among transgender and gender diverse people. Methods: We tested the first two steps of the three-step theory of suicide using daily survey data from a sample of 38 transgender and gender diverse people over thirty days. Results: A total of 836 daily surveys were collected (73.3% compliance). Multilevel modeling supported the first and second step of the three-step theory. Psychological pain and hopelessness interacted to predict same-day suicidal ideation, with psychological pain positively associating with ideation only at average and high levels of hopelessness. Furthermore, psychological pain that outweighed connectedness was moderately associated with suicidal ideation among those with high levels of hopelessness and psychological pain. Conclusion: The three-step theory of suicide shows promise for explaining and guiding interventions to reduce suicidal ideation in this vulnerable population. gender minority hopelessness psychological pain social connectedness pmcTransgender and gender diverse (TGD) individuals are at higher risk for suicide relative to their cisgender peers (Marshall, Claes, Bouman, Witcomb, & Arcelus, 2016). The past year prevalence of suicide attempts (11%) and suicidal ideation (51%; Adams et al., 2017) among TGD adults exceeds prevalence in the general population (3.7% and 0.5%, respectively; Crosby, Gfroerer, Han, Ortega, & Parks, 2011). These disparities highlight a need for empirical and theoretical work aimed at understanding and preventing suicide in TGD people. Ideation-to-action frameworks (Klonsky & May, 2014) are increasingly used to understand suicidal ideation and attempts by proposing distinct but related causal pathways to these phenomena. The most recent theory based in an ideation-to-action framework, the three-step theory (3ST), proposes that suicidal ideation first develops from the combination of pain (typically psychological pain) and hopelessness (Klonsky & May, 2015). Second, suicidal ideation escalates if one’s pain and hopelessness are not exceeded by one’s connectedness (e.g., to people, jobs, roles, purpose/meaning). Third, ideation then transitions into a suicide attempt if an individual has acquired (e.g., habituation to fear and pain involved in death), dispositional (e.g., genetic fearlessness/pain tolerance), or practical (e.g., access to means, knowledge of attempt lethality) capability to carry out an attempt. However, very few studies have tested the 3ST in general (Klonsky & May, 2015; Dhingra et al., 2019; Yang et al., 2019; see Klonsky, Saffer, & Bryan, 2018 for a review). Furthermore, none have tested its application to TGD people, a population which requires further understanding regarding suicide risk. A systematic review of the literature concerning suicide risk within the TGD population revealed the potential utility of this theory in explaining disparities in suicide risk (Wolford-Clevenger, et al., 2018). There is largely indirect support for the first step of 3ST that posits the combination of psychological pain with hopelessness is necessary for suicidal ideation to begin. Several minority stress experiences that could contribute to psychological pain are associated with suicidal ideation in TGD samples. These include nonaffirmation of gender identity (Parr & Howe, 2019), gender-based violence, discrimination, rejection, and internalized stigma, expectations of rejection, perceived burdensomeness, and concealment of identity (e.g., Bauer et al., 2015; Grossman et al., 2016; Kattari et al., 2016; Lehavot, et al., 2016; Moody & Smith, 2013; Testa et al., 2017). Furthermore, internal minority stress may explain the association between external minority stress and suicidal ideation (Testa et al., 2017), which is consistent with 3ST’s position that psychological pain contributes to the desire to die (Klonsky & May, 2015). Although hopelessness is ubiquitous in suicide research, very few studies have examined its relation to suicidal ideation among TGD people specifically. Literature on combined samples of sexual (e.g., lesbian, gay, bisexual) and gender minorities sheds some light on this association, revealing that hopelessness about social disconnectedness is associated with sexual and gender minority adults’ worst point of suicidal ideation (Salentine, Hilt, Muehlenkamp, & Ehlinger, 2020). No studies have tested the role of hopelessness in the context of 3ST in a sample of exclusively TGD people. This is important, as TGD people have experiences distinct from those of sexual minorities. Data have also indirectly supported the second step of the 3ST, that connectedness (most often operationalized as social connectedness) buffers the combined effects of psychological pain and hopelessness on suicidal ideation. Indeed, studies have suggested that the social marginalization of TGD people results in limited social connections, and thus protection against suicidal ideation (Bauer et al., 2015; Kattari, Walls, Speer, & Kattari, 2016; Kuper, 2015; Trujillo, Perrin, Sutter, Tabaac, & Benotsch, 2017; c.f. Lehavot et al., 2016). In more direct support of the 3ST, studies of TGD people have demonstrated that feeling disconnected from others (i.e., thwarted belongingness) is positively correlated with suicidal ideation in one’s lifetime (Grossman, Park, & Russell, 2016) and in the past year (Testa et al., 2017). Additionally, social support from friends and partners may weaken the effects of sources of psychological pain on past suicidal ideation (Trujillo et al., 2017). Finally, in a test of another ideation-to-action theory (the interpersonal-psychological theory of suicide), hopelessness about social disconnectedness mediated the association between discrimination and suicidal ideation (Salentine, Hilt, Muehlenkamp, & Ehlinger, 2020). Taken together, the literature supports the potential generalizability of 3ST to TGD people; however, three major gaps underscore the need for additional study. First, most (95.6%) of the data on correlates of suicidal ideation and behaviors in TGD people are cross-sectional (Wolford-Clevenger et al., 2018). Second, no studies examined the proximal associations of psychological pain, hopelessness, and social disconnectedness on suicidal ideation in this population. Study designs that are capable of testing proximal correlates of suicidal ideation are needed, as studies have found suicidal ideation and its risk factors (e.g., hopelessness) vary significantly within short time periods (Kleiman et al., 2017). Third, although some studies have tested ideation-to-action theories in TGD people, many of these studies use combined samples of sexual and gender minorities and none have tested the 3ST specifically. Thus, the present study aims to address these critical gaps by testing the first two steps of the 3ST using daily diary methods in a sample of exclusively TGD people. Consistent with the first step of the 3ST, we hypothesized that psychological pain and hopelessness will interact to predict same-day suicidal ideation, such that as the association between psychological pain and hopelessness strengthens, suicidal ideation increases. In line with the second step of 3ST, we hypothesized that on days when both psychological pain and hopelessness are high, the difference between social connectedness and psychological pain will be strongly associated with same-day suicidal ideation (i.e., pain exceeding connectedness will be positively and moderately associated with same-day suicidal ideation). Whereas on days when psychological pain and hopelessness are not both high, the difference between social connectedness and psychological pain will only have weak associations with same-day suicidal ideation. Methods Participants Participants were recruited from two mid-sized cities in the Southeastern United States. The number of participants recruited from each city was similar (n = 21 at City 1 and n = 17 at City 2). Participants were individuals 18 years of age or older and who identified as “Transgender, gender-diverse, of trans experience, or hav[ing] transitioned.” Sixty-eight individuals were screened for the study. Sixty (89.6%) were eligible and completed the baseline questionnaires (n = 6 were ineligible, and n = 2 did not complete the baseline questionnaires). Thirty-eight (63%) of the individuals completing the baseline phase elected to continue to the daily phase and completed at least three daily surveys to be included in the analyses. The following participant descriptive statistics pertain to this subset of the sample. Participants were given 26 non-mutually exclusive options (See Table 1 for a summary of options) for gender identity and made, on average, 3.52 selections (SD = 2.06; Mode = 2; Range 11). Participants’ gender identities were generally equally distributed across feminine, masculine, and gender nonbinary categories (See Table 1). Average age was 28.63 years. The sample was majority non-Hispanic White (84.2%), employed at least part-time (58%), reported an annual income below $50,000 (63.9%), resided with others (e.g., family, friends; 81.6%), and reported a same-sex/gender or non-monosexual orientation (e.g., lesbian, gay, other; 92.1%). The average years of education was 13.93 (SD = 4.17). See Table 1 for a full summary of participant demographics. Participants recruited from each city did not differ in these demographics (ps > .05). However, participants at City 1 reported greater psychological pain (p = .03) and less hopelessness (p < .001) than those at City 2. Participants in each city did not differ in social connectedness (p = .36) or suicidal ideation (p = .24). Procedures The institutional review boards of the first and last authors granted approval for the study procedures. A non-probability sample of self-identified TGD adults were recruited from two mid-sized cities in the Southeastern region of the United States. To be eligible for the study, participants must have been 18 years old or older and identified as transgender, gender-diverse, of trans experience, or had transitioned. Advertisements for the study were posted throughout two university campuses, including student health centers, campus pride centers, as well as at local HIV treatment centers, a university-affiliated medical center, local trans-related events, and Facebook pages for TGD people in the local areas (e.g., local TGD advocacy and support group pages). Participants were also told they could pass study information on to TGD people they knew in the community. Interested individuals were provided a hyperlink to the survey on Qualtrics.com, an online survey platform, where they were screened for eligibility (i.e., over the age of 18, identified as transgender or gender-diverse). Informed consent was obtained electronically from all participants in the study. Eligible individuals who consented were forwarded to an hour-long baseline survey that assessed demographic information and history of suicidal thoughts, behaviors, and risk factors (e.g., psychiatric symptoms). Following the baseline survey, participants were redirected to a form where they entered their e-mail addresses for the daily survey phase of the study. Beginning on the day after the baseline survey, participants received e-mails containing a link to a 5-minute survey at 6 A.M., with a reminder at 12 P.M., each day for 30 days. Participants were instructed to report on their previous day’s experience, defined as the time they woke up to the time they went to bed. The baseline and daily surveys were linked using a subject-generated identification code that was only known to the participants (Yurek et al., 2008). Thus, although participation in the study was confidential (i.e., participants’ identities could be known by their email address), their individual survey responses were anonymous. No personally identifiable information was connected to their individual baseline or daily survey data. Only the participants knew their unique identifier. Because of this, we were unable to intervene upon suicidal ideation reported in the daily surveys. Further, intervention may have influenced participant responses via strengthening of the Hawthorne effect. Branch logic was programmed in the survey platform such that any endorsement of suicidal ideation or behaviors triggered a message at the end of the survey encouraging participants to contact the investigator or other resources. A page of mental health resources was presented to all participants following each daily survey, including the first author’s phone number. We paid participants with a $5.00 Wal-Mart gift card for their baseline survey, with $0.50 added for each daily survey ($20 total possible compensation). Measures Primary Baseline Measures. A battery of baseline measures pertaining to suicide risk and its correlates in the population of interest (e.g., minority stress, psychiatric distress) was collected. For descriptive purposes of the present study aims, we report on the following baseline measures: Demographic Variables. A questionnaire developed for the study collected information on age, gender identity, sex assigned at birth (from one city), racial/ethnic identity, sexual orientation, relationship status, income, education, employment status, and other sociodemographic variables. Suicidal ideation. The 4-item Hopelessness Depression Symptom Questionnaire-Suicidality Subscale (Metalsky & Joiner, 1997) measured the frequency, planning, controllability, and impulsive nature of suicidal ideation over the past week on a 4-point Likert Scale. This questionnaire has demonstrated excellent construct validity (Metalsky & Joiner, 1997) and good internal consistency in the present sample (α = .87). The total score (a continuous score ranging from 0 to 16) was used in the analyses. Furthermore, to describe the prevalence of past-week suicidal ideation, we dichotomized this measure such that a nonzero score on this scale was indicative of the presence of suicidal ideation. Primary Daily Measures. Brief items assessing suicidal thoughts and behaviors and correlates of interest (e.g., minority stress, psychiatric symptoms) were measured each day. For the present aims, we describe the following measures: Psychological Pain. One 5-point Likert item (“My pain was making me fall apart”, 1 = Strongly Disagree to 5 = Strongly Agree) from the Psychache Scale (Holden, Mehta, Cunningham, & McLeod, 2001) assessed psychological pain the day prior. Before answering the question, participants were informed that the “following statement refers to your psychological pain, NOT your physical pain.” The scale from which this item was selected has demonstrated excellent internal consistency and construct validity (Holden et al., 2001). Hopelessness. We used the “hopeless” item of the depression-dejection subscale of the Profile of Mood States-Short Form (POMS-SF; Shacham, 1983). Participants were instructed to do the following: “Below is a list of words that describes feelings people have. Please read each one carefully. Select ONE answer to the right that best describes how you were feeling yesterday.” Participants ranked how “hopeless” they felt the day prior on a scale of 0 = “Not at all” to 4 = “Extremely”. The POMS-SF has been validated for use in daily diary studies (Cranford et al., 2006). Connectedness. Social connectedness was assessed using two items from the thwarted belongingness subscale of the Interpersonal Needs Questionnaire (Van Orden et al., 2012) that most highly correlated with the subscale in the validation study (“I felt close to others” and “I felt like I belonged”). Participants rated how they felt overall the day prior on a 7-point Likert Scale (1 = not at all true for me to 7 = very true for me). These are traditionally reverse-scored and summed to create a thwarted belongingness subscale, but given the present study’s focus on social connectedness, we did not reverse score these items. These items were summed for a total social connectedness score for that day. This scale had excellent internal consistency (α = .96). Suicidal ideation. We used the five-item Paykel Suicide Scale (PSS; Paykel, Myers, Lindenthal, & Tanner, 1974) to assess daily suicidal ideation. We selected this scale due to its demonstrated reliability and validity, as well as its brevity and amenability to altering the assessment timeframe. The first four items of the PSS assess passive and active suicidal ideation (e.g., death wish versus suicidal plans) on a 4-point scale, with the total score ranging from 0–12. This scale had acceptable internal consistency (α = .76). The fifth item assesses suicide attempts but was not included in this report due to extremely low endorsement (only one attempt reported in the daily diaries). Data Analytic Strategy Compliance rates for the daily data were computed by calculating the percentage of missed surveys in relation to total surveys sent. We also examined potential differences in participants who participated in the daily phase of the study relative to those who did not participate or completed less than three days of data. Next, for descriptive purposes, frequencies of a dichotomized version of the baseline suicidal ideation variable (0 = no suicidal ideation, 1 = some level of suicidal ideation) were computed. We used the statistical program, Hierarchical Linear Modeling (HLM) version 7, to run the following analyses. To explore potential demographic correlates of suicidal ideation in this sample, we ran separate multilevel models, using random intercepts and slopes to predict Level-1 (daily level) suicidal ideation by each Level-2 (baseline level) demographic variable (i.e., race/ethnic identity dichotomized as Non-Hispanic White/Hispanic and/or Person of Color; age, resides with people or alone, income dichotomized as less than $50,000 or more than $50,000, employed, student status, partnered, and study city). Next, we ran a null model to describe the within- and between-person variability in the Level-1 suicidal ideation variable. To test the first hypothesis, we ran a multilevel model including time, psychological pain, hopelessness, and the two-way interaction between psychological pain and hopelessness (which were converted into z-scores prior to creation of the interaction term and analysis) to predict same-day suicidal ideation. We ran the model with full maximum likelihood estimation and random intercepts and fixed slopes. Test of Level-1 homogeneity of variance was significant, and the assumption of normal distribution of residuals was not met. Therefore, the model with robust standard errors was interpreted and reported. To test the second hypothesis, we used a data analytic strategy consistent with prior tests of the 3ST (e.g., Klonsky & May, 2015). Connectedness and psychological pain scores were standardized. Then, individuals’ daily connectedness scores were subtracted from their psychological pain scores (with positive scores indicating pain was greater than connectedness). This difference score was included as Level-1 predictor of same-day suicidal ideation in the subsample of days in which both hopelessness and psychological pain were high (calculated by creating median splits of hopelessness and psychological pain). For completeness, we also ran this analysis in the remaining subsample of days in which hopelessness and psychological pain were not simultaneously high. Results A total of 836 daily surveys were collected (73.3% compliance rate). Participants who engaged in the daily phase of the study did not differ from those who did not in site, race/ethnicity, relationship status, income, employment, student status, age, education, sexual orientation, residence, city, or baseline levels of suicidal ideation (all ps > .05). Of the subsample who went on to take part in the daily phase of the study, 59.5% reported experiencing some level of past-week suicidal ideation at baseline. Regarding the daily data, suicidal ideation occurred on 25% of the completed days, with 71% of the sample experiencing suicidal ideation at least once over the thirty days. The intraclass correlation coefficient for suicidal ideation was 36%, indicating that 36% of the variance was due to between-person differences. When examining potential demographic correlates of suicidal ideation at the daily level, only non-White racial identity (coefficient = 0.12, t = 2.18, p = .04) and residing with others (coefficient = 0.11, t = 2.21, p = .03) emerged. Note that these correlates were not included in the following multilevel models due to potential for biasing effect sizes; however, secondary analyses including these correlates were not substantially different from those reported here. The model testing our first hypothesis revealed that the two-way interaction between psychological pain and hopelessness was significant (See Table 2). Decomposition of this interaction revealed that psychological pain was positively associated with suicidal ideation severity at high (+1 SD) (coefficient = 0.61, t = 11.36, p < .001) and average levels of hopelessness (coefficient = 0.25, t = 6.44, p < .001) but not low (−1 SD) levels of hopelessness (coefficient = −0.11, t = −1.86, p = .06). See Figure 1 for a visual depiction of the interaction. To get an odds ratio measure of effect, we conducted a secondary analysis. We decomposed the interaction using a dichotomous version of the suicidal ideation variable as the outcome. Results revealed that at high levels of hopelessness, psychological pain was associated with twice the odds of suicidal ideation occurring on that same day (OR = 2.42, 95% CI = [1.76, 3.33], p < .001). The analysis testing Step 2 of the 3ST revealed that the difference score between psychological pain and connectedness was moderately correlated with suicidal ideation while controlling for time (coefficient = 0.44, t = 4.16, df = 162, p < .001) on days when both psychological pain and hopelessness were high. On days when psychological pain and hopelessness were not both high, this difference score was weakly associated with suicidal ideation (coefficient = 0.11, t = 3.89, df = 590, p < .001). Discussion Studies that have tested theories to explain the higher prevalence of suicidal ideation among TGD people are scarce. This daily diary study is the first to conduct an exploratory test of the first two steps of the 3ST in a sample of exclusively TGD people. Results supported the first step of the theory, showing that psychological pain was associated with greater same-day suicidal ideation only at average and high levels of hopelessness (and was associated with over twice the odds of suicidal ideation). Our findings also supported the second step of the theory which hypothesizes that if one’s pain exceeds one’s connectedness, suicidal ideation escalates. Our findings showed the difference score between pain and social connectedness was moderately associated with same-day suicidal ideation on days when both hopelessness and psychological pain were high. Although preliminary given the small sample size, these findings corroborate the few studies that have also tested the 3ST. Our finding that psychological pain is associated with suicidal ideation at high levels of hopelessness replicates prior work in adults in the United States (Klonsky & May, 2015), university students in the United Kingdom (Dhingra, Klonsky, & Tapola, 2019), and university students in China (Yang, Liu, Chen, & Li, 2019). Our findings also converge with prior work (Klonsky & May, 2015; Yang et al., 2019) in that the difference score between pain and social connectedness was moderately associated with suicidal ideation on days when both psychological pain and hopelessness were high. These findings provide more nuance to prior cross-sectional work examining various forms of social connection as it relates to suicidal ideation in this population. Prior studies have overwhelmingly shown that social connection is negatively associated with suicidal ideation (Bauer et al., 2015; Grossman et al., 2016; Kattari et al., 2016; Kuper, 2015; Testa et al., 2017; c.f., Lehavot et al., 2016) and may buffer the effects of discrimination (a potential source of psychological pain) on suicidal ideation (Trujillo et al., 2017). The current findings suggest that particularly on days when both psychological pain and hopelessness are high, pain that exceeds social connection may escalate suicidal ideation in this population. Our findings are also unique in that we demonstrate a proximal association among these constructs. Additional consideration of the role of social connection in the context of hopelessness and psychological pain appears important, as a systematic review of a similar ideation-to-action theory (the interpersonal theory of suicide) showed that social disconnection was one of the lesser supported constructs (Ma et al., 2016). Furthermore, investigations capable of testing the temporal and proximal associations among these constructs are needed to explore these questions. Future Directions Much more work is also needed to shed light on what theories, or reconceptualizations thereof, best describe suicide risk development in this population. First and foremost, more research is needed that focuses exclusively on TGD people. Most studies that examine TGD people have combined them with cisgender, sexual minorities who have different experiences that might affect suicide risk development. Second, daily diary or ecological momentary assessment work is particularly needed, as suicidal ideation and its risk factors vary significantly in short time periods (Kleiman et al., 2017). This is the first study to examine suicidal ideation via daily diary methods in TGD people, and 64% of the variance in suicidal ideation was due to within person differences. The present and future work will help inform suicide prevention and intervention programs tailored to the TGD community. Strengths and Limitations The current study has many strengths. It employed a daily diary design, capable of elucidating the proximal associations among suicidal ideation and its theorized risk factors. It focused on TGD people exclusively. Finally, it is the first to test the 3ST using a daily diary design and with TGD people. Limitations of the current study include a small sample that was homogenous in race/ethnicity and based in the Southeastern United States. A larger sample with greater racial/ethnic and geographic diversity is needed for these findings to have broader implications for the TGD community in the United States. Further, to reduce participant burden, few items were used at the daily level to capture the theory’s constructs and their psychometric validity is unknown. However, prior research using brief items with similar content to measure social connection, hopelessness, and suicidal ideation has shown reliability and convergent validity (Forkmann et al., 2018). To improve upon these measurement issues, future studies could collect data on solely on the 3ST constructs, therefore allowing for more comprehensive measurement (rather than including additional constructs as did this study). Additionally, our study examined social connectedness rather than broad connectedness (e.g., to purpose/meaning, responsibilities) as theorized by 3ST. Future work should use a broader operationalization of connectedness. Finally, the present study did not test the third step of the 3ST regarding the transition from suicidal ideation to suicide attempts. Conclusions The limited understanding of the high rates of suicidal ideation among TGD people may be due to the lack of theory-guided studies that are capable of testing proximal correlates of suicidal ideation and behavior in this population. We tested the first two steps of the 3ST using daily survey data from a sample of 38 TGD people over thirty days. Results supported both steps of the 3ST. Psychological pain and hopelessness interacted to predict suicidal ideation occurring on the same day, with psychological pain positively associating with ideation at average and high, but not low, levels of hopelessness. Psychological pain that outweighed social connectedness was moderately associated with suicidal ideation on days when both hopelessness and pain were high (and weakly associated on days when hopelessness and pain were not both high). The 3ST shows promise for explaining and guiding interventions to reduce suicidal ideation in this vulnerable population. Acknowledgements: We would like to acknowledge Dr. Rebecca Morgan and her staff at the University of Tennessee-Knoxville for her assistance in recruitment. Funding Statement: This work was supported, in part, by the Malyon-Smith Scholarship Award sponsored by Division 44 of the American Psychological Association (APA), grant F31AA024685 from the National Institute on Alcohol Abuse and Alcoholism (NIAAA), and the Thomas Fellowship awarded to the first author. The content is solely the responsibility of the authors and does not necessarily represent the official views of APA Division 44, the NIAAA, National Institutes of Health, or the Thomas family. Figure 1. Interactive Effect of Psychological Pain and Hopelessness on Suicidal Ideation Table 1. Sample Characteristics and Baseline and Daily Variable Descriptives Baseline Variables n % Gender identity (n = 38; non-mutually exclusive categories)  Female, MTF, on MTF spectrum, transfeminine, trans woman, woman of trans experience 14 36.8  Male, FTM, on FTM spectrum, transmasculine, trans man, man of trans experience 18 47.4  Genderqueer/fluid/diverse, bigender, agender, intersex, pan/polygender, two-spirit, androgyne 15 39.5 Sex assigned at birth (n = 16, only collected at one site)  Male 4 23.5  Female 12 70.6 Sexual orientation (n = 38)  Bisexual 14 36.8  Gay 6 15.8  Heterosexual 3 7.9  Other (pansexual; polyamorous; asexual) 15 39.5 Employed at least part time (n = 38) 22 57.9 Degree-seeking status (n =38, Student status) 16 42.1 Race (n = 38)  African American/Black 2 5.3  Multiracial 3 7.9  Other race 1 2.6  White/Non-Hispanic 32 84.2 Yearly Income less than $50,000 (n = 38) 23 63.9 Relationship status (n = 38)  Single, not dating anyone 18 47.4  Dating 11 28.9  Engaged to be married 2 5.3  Married 5 13.2  Divorced 2 5.3 Baseline Variables M SD Age (range: 18–64; n = 38) 28.63 11.62 Education years (range: 2–22; n = 38) 13.93 4.17 Baseline suicidal ideation (range: 0–7; n = 37) 2.16 2.44 Daily Variables M SD Psychological pain (range: 1–5, n = 831) 2.33 1.53 Hopelessness (range: 0–4, n = 832) 1.09 1.26 Social connectedness (range: 2–14, n = 833) 8.20 3.20 Suicidal ideation (range: 0–4, n = 820) 0.52 1.00 Note: MTF = male-to-female, FTM = female-to-male, total sample size displayed next to variable name. Gender identities were combined to create the gender categories and are not mutually exclusive. Table 2. Parameters for Multilevel Model Testing Step 1 of 3ST Predicting Suicidal Ideation Level-1 Variables B SE t df p Intercept 0.29 0.07 4.46 37 <.001 Day 0.004 0.003 1.44 778 .15 Psychological Pain 0.25 0.04 6.44 778 <.001 Hopelessness 0.26 0.07 3.48 778 <.001 Psychological Pain X Hopelessness 0.36 0.04 8.49 778 <.001 Conflict of Interest Disclosure: The authors have no conflicts of interest to disclose. Ethics Approval Statement: The research was approved by institutional ethics boards of the first and last author. References Adams N , Hitomi M , & Moody C (2017). Varied reports of adult transgender suicidality: synthesizing and describing the peer-reviewed and gray literature. Transgender Health, 2 (1 ), 60–75.28861548 Bauer GR , Scheim AI , Pyne J , Travers R , & Hammond R (2015). Intervenable factors associated with suicide risk in transgender persons: a respondent driven sampling study in Ontario, Canada. BMC Public Health, 15 (1 ), 525.26032733 Cranford JA , Shrout PE , Iida M , Rafaeli E , Yip T , & Bolger N (2006). A procedure for evaluating sensitivity to within-person change: can mood measures in diary studies detect change reliably?. Personality and Social Psychology Bulletin, 32 (7 ), 917–929.16738025 Crosby A , Gfroerer J , Han B , Ortega L , & Parks SE (2011). Suicidal thoughts and behaviors among adults aged ≥ 18 Years-United States, 2008–2009. US Department of Health and Human Services, Centers for Disease Control and Prevention. Dhingra K , Klonsky ED , & Tapola V (2019). An empirical test of the Three‐Step Theory of suicide in UK University students. Suicide and Life‐Threatening Behavior, 49 (2 ), 478–487.29604120 Forkmann T , Spangenberg L , Rath D , Hallensleben N , Hegerl U , Kersting A , & Glaesmer H (2018). Assessing suicidality in real time: A psychometric evaluation of self-report items for the assessment of suicidal ideation and its proximal risk factors using ecological momentary assessments. Journal of Abnormal Psychology, 127 (8 ), 758.30299116 Grossman AH , Park JY , & Russell ST (2016). Transgender youth and suicidal behaviors: Applying the interpersonal psychological theory of suicide. Journal of Gay & Lesbian Mental Health, 20 (4 ), 329–349.28344728 Holden RR , Mehta K , Cunningham EJ , & McLeod LD (2001). Development and preliminary validation of a scale of psychache. Canadian Journal of Behavioural Science, 33 (4 ), 224. Husky M , Olié E , Guillaume S , Genty C , Swendsen J , & Courtet P (2014). Feasibility and validity of ecological momentary assessment in the investigation of suicide risk. Psychiatry Research, 220 (1 ), 564–570.25155939 Kattari SK , Walls NE , Speer SR , & Kattari L (2016). Exploring the relationship between transgender-inclusive providers and mental health outcomes among transgender/gender variant people. Social Work in Health Care, 55 (8 ), 635–650.27351890 Kessler RC , Borges G , & Walters EE (1999). Prevalence of and risk factors for lifetime suicide attempts in the National Comorbidity Survey. Archives of General Psychiatry, 56 (7 ), 617–626.10401507 Kleiman EM , Turner BJ , Fedor S , Beale EE , Huffman JC , & Nock MK (2017). Examination of real-time fluctuations in suicidal ideation and its risk factors: Results from two ecological momentary assessment studies. Journal of Abnormal Psychology, 126 , 726–738.28481571 Klonsky ED , & May AM (2014). Differentiating suicide attempters from suicide ideators: A critical frontier for suicidology research. Suicide and Life-Threatening Behavior, 44 (1 ), 1–5.24313594 Klonsky ED , & May AM (2015). The three-step theory (3ST): A new theory of suicide rooted in the “ideation-to-action” framework. International Journal of Cognitive Therapy, 8 (2 ), 114–129. Klonsky ED , Saffer BY , & Bryan CJ (2018). Ideation-to-action theories of suicide: a conceptual and empirical update. Current Opinion in Psychology, 22 , 38–43.30122276 Lehavot K , Simpson TL , & Shipherd JC (2016). Factors associated with suicidality among a national sample of transgender veterans. Suicide and Life-Threatening Behavior, 46 (5 ), 507–524.26878597 Ma J , Batterham PJ , Calear AL , & Han J (2016). A systematic review of the predictions of the Interpersonal–Psychological Theory of Suicidal Behavior. Clinical Psychology Review, 46 , 34–45.27155061 Marshall BD , Socías ME , Kerr T , Zalazar V , Sued O , & Arístegui I (2016). Prevalence and correlates of lifetime suicide attempts among transgender persons in Argentina. Journal of Homosexuality, 63 (7 ), 955–967.26566683 Metalsky GI , & Joiner TE Jr (1997). The hopelessness depression symptom questionnaire. Cognitive Therapy and Research, 21 (3 ), 359–384. Moody C , & Smith NG (2013). Suicide protective factors among trans adults. Archives of Sexual Behavior, 42 (5 ), 739–752.23613139 Parr NJ , & Howe BG (2019). Heterogeneity of transgender identity nonaffirmation microaggressions and their association with depression symptoms and suicidality among transgender persons. Psychology of Sexual Orientation and Gender Diversity, 6 (4 ), 461. Paykel ES , Myers JK , Lindenthal JJ , & Tanner J (1974). Suicidal feelings in the general population: A prevalence study. The British Journal of Psychiatry, 124 (582 ), 460–469.4836376 Salentine CM , Hilt LM , Muehlenkamp JJ , & Ehlinger PP (2020). The link between discrimination and worst point suicidal ideation among sexual and gender minority adults. Suicide and Life‐Threatening Behavior, 50 (1 ), 19–28.31282049 Shacham S (1983). A shortened version of the Profile of Mood States. Journal of Personality Assessment, 47 (3 ), 305–306.6886962 Smith DM , Wang SB , Carter ML , Fox KR , & Hooley JM (2020). Longitudinal predictors of self-injurious thoughts and behaviors in sexual and gender minority adolescents. Journal of Abnormal Psychology, 129 (1 ), 114.31657599 Testa RJ , Michaels MS , Bliss W , Rogers ML , Balsam KF , & Joiner T (2017). Suicidal ideation in transgender people: Gender minority stress and interpersonal theory factors. Journal of Abnormal Psychology, 126 (1 ), 125.27831708 Trujillo MA , Perrin PB , Sutter M , Tabaac A , & Benotsch EG (2017). The buffering role of social support on the associations among discrimination, mental health, and suicidality in a transgender sample. International Journal of Transgenderism, 18 (1 ), 39–52.29904324 Van Orden KA , Cukrowicz KC , Witte TK , & Joiner TE Jr. (2012). Thwarted belongingness and perceived burdensomeness: Construct validity and psychometric properties of the Interpersonal Needs Questionnaire. Psychological Assessment, 24 (1 ), 197.21928908 Wolford-Clevenger C , Frantell K , Smith PN , Flores LY , & Stuart GL (2018). Correlates of suicide ideation and behaviors among transgender people: A systematic review guided by ideation-to-action theory. Clinical Psychology Review, 63 , 93–105.29960203 Yang L , Liu X , Chen W , & Li L (2019). A test of the three-step theory of suicide among Chinese people: A study based on the ideation-to-action framework. Archives of Suicide Research, 23 (4 ), 648–661.30024342 Yurek LA , Vasey J , & Sullivan Havens D (2008). The use of self-generated identification codes in longitudinal research. Evaluation Review, 32 (5 ), 435–452.18477737

PMC009xxxxxx/PMC9969793.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101313252 34584 J Vis Exp J Vis Exp Journal of visualized experiments : JoVE 1940-087X 35969093 9969793 10.3791/64264 NIHMS1867397 Article MEASURING SKELETAL MUSCLE THERMOGENESIS IN MICE AND RATS Watts Christina A. 1 Haupt Alexandra 1 Smith Jordan 2 Welch Emily 3 Malik Aalia 3 Giacomino Roman 3 Walter Dinah 3 Mavundza Nhlalala 1 Shemery Ashley 1 Caldwell Heather K. 13 Novak Colleen M. 13 1 School of Biomedical Sciences, Kent State University, Kent, OH, USA 2 College of Public Health, Kent State University, Kent, OH, USA 3 Department of Biological Sciences, Kent State University, Kent, OH, USA 7 2 2023 27 7 2022 27 7 2022 27 7 2023 185 10.3791/64264This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Skeletal muscle thermogenesis provides a potential avenue for better understanding metabolic homeostasis and mechanisms underlying energy expenditure. Surprisingly little evidence is available to link neural, myocellular, and molecular mechanisms of thermogenesis directly to measurable changes in muscle temperature. This paper describes a method in which temperature transponders are utilized to retrieve direct measurements of mouse and rat skeletal muscle temperature. Remote transponders are surgically implanted within the muscle of mice and rats, and the animals are given time to recover. Mice and rats must then be repeatedly habituated to the testing environment and procedure. Changes in muscle temperature are measured in response to pharmacological or contextual stimuli in the home cage. Muscle temperature can also be measured during prescribed physical activity (i.e., treadmill walking at a constant speed) to factor out changes in activity as contributors to the changes in muscle temperature induced by these stimuli. This method has been successfully used to elucidate mechanisms underlying muscle thermogenic control at the level of the brain, sympathetic nervous system, and skeletal muscle. Provided are demonstrations of this success using predator odor (PO; ferret odor) as a contextual stimulus and injections of oxytocin (Oxt) as a pharmacological stimulus, where predator odor induces muscle thermogenesis, and Oxt suppresses muscle temperature. Thus, these datasets display the efficacy of this method in detecting rapid changes in muscle temperature. pmcIntroduction Within metabolic research, examination of skeletal muscle thermogenesis is a promising new avenue for probing body weight homeostasis. Published literature supports the idea that thermogenic responses of one of the body’s largest organ systems—the skeletal muscle—provides an avenue for increasing energy expenditure and other metabolic effects, thereby effectively rebalancing systems within diseases such as obesity2–4. If the muscle can be considered a thermogenic organ, studies must utilize a practical methodology to study thermogenic changes within this organ. The desire to understand the endothermic impact of skeletal muscles and the utility of this methodology for studying non-shivering muscle thermogenesis is not specific to metabolic studies. Disciplines, including evolution5, comparative physiology6, and ecophysiology7,8, have shown a vested interest in understanding the ways in which muscle thermogenesis may contribute to endothermy, and how this mechanism adapts to the environment. The presented protocol provides the critical methods necessary to address these questions. The provided method can be utilized in the assessment of both contextual and pharmacological stimuli modulation of muscle temperature, including the unique technique of providing PO to shift the context to replicate predator threat. Prior reports have demonstrated the ability of PO to rapidly induce a sizable increase in muscle thermogenesis9. Moreover, pharmacological stimuli can also alter muscle temperature. This has been demonstrated in the context of PO-induced muscle thermogenesis, where pharmacological blockade of peripheral β-adrenergic receptors, using nadolol, inhibited the ability of PO to induce muscle thermogenesis without significantly affecting contractile thermogenesis during treadmill walking9. Central administration of melanocortin receptor agonists in rats has also been used to discern brain mechanisms altering thermogenesis10,11. Provided here is a preliminary investigation of the ability of the neurohormone Oxt to alter muscle thermogenesis in mice. Similar to predator threat, social encounters with a same-sex conspecific increase body temperature, a phenomenon referred to as social hyperthermia12. Given the relevance of Oxt to social behavior13, it has been speculated that Oxt is a mediator of social hyperthermia in mice. Indeed, an oxytocin receptor antagonist decreases social hyperthermia in mice12, and mouse pups lacking Oxt show deficits in behavioral and physiological aspects of thermoregulation, including thermogenesis14. Given that Harshaw et al. (2021) did not find evidence supporting β3 adrenergic receptor-dependent brown adipose tissue (BAT) thermogenesis with social hyperthermia12, it has been posited that social hyperthermia may be driven by Oxt’s induction of muscle thermogenesis. To measure skeletal muscle thermogenesis, the following protocol uses the implantation of preprogrammed IPTT-300 transponders adjacent to the muscle of interest within a mouse or rat9,11,15,16. These transponders are glass-encapsulated microchips that are read using corresponding transponder readers. Little to no research has utilized this technology in this capacity, though studies have suggested a need for the specificity provided by this method17,18. Previous investigations have shown the reliability of this method and a variety of ways in which temperature transponders can be used in comparison with other temperature-testing methods19 or in conjunction with surgical methods (e.g., cannulation20). However, studies of this nature rely on different strategic placements to measure overall body temperature21–23 or specified tissues such as BAT24–26. Rather than measuring temperature from these locations or while using ear or rectal thermometers27, the method described here provides specificity for the muscle of interest. The ability to target a site by directly implanting transponders adjacent to muscles of interest is more effective for probing muscle thermogenesis specifically. It provides a new avenue in addition to those provided by surface infrared thermometry28,29 or cutaneous temperature measurements via thermocouple30. Furthermore, the data provided through this method offer a range of avenues of research, avoiding the need for large, expensive, high-tech equipment, and software such as infrared thermography31–33. This method has been successfully used to measure temperature in the quadriceps and gastrocnemius, either unilaterally or bilaterally. This method has also been effective in conjunction with stereotaxic surgery15,16. Within ~7–10 cm of the transponder-limb, portable transponder readers (DAS-8027/DAS-7007R) are used to scan, measure, and display the temperature. This distance has been critical and valuable to prior investigations9–11 because it minimizes potential stressors and temperature-altering variables such as animal handling during the testing procedures. Using timers, measurements can then be recorded and collected over a period of without direct interaction with the animals. To further minimize the disturbance of mice during testing, this method describes the assembly and use of risers made of PVC piping to give the experimenter access to the bottom of home cages during testing. Using the risers in tandem with the digital reader, temperature measurements of the transponder-limb can be made without any animal interaction after the stimulus is placed. At a minimal cost, this method can be used in conjunction with pharmacological and contextual stimuli, making it quite accessible for researchers. Additionally, this method can be employed with a substantial number of subjects (~16 mice or ~12 rats) at a time, saving time in increasing overall throughput for any research project. Introduced in this method is a crafted mechanism for presenting odors to mice using stainless steel mesh tea infuser balls, from now on referred to as ‘tea balls.’ Though these tea balls are ideal for containing any odor material, in these studies, towels that served as in-cage bedding over 2–3 weeks for ferrets, a natural predator of mice and rats, are placed within each treatment tea ball. Each towel is cut into 5 cm × 5 cm squares. This aliquoting is also repeated with otherwise identical odorless control towels. Presenting these odors without a barrier (i.e., tea ball) led to mice shredding the fibers within their cages, increasing physical activity. This behavior was not as salient in rats. Tea balls provide a ventilated casing to the towel, giving full access to the odor while staying protected for the entirety of the experimental trial. These tea balls can be sanitized in accordance with animal use protocols, prepared, and introduced directly after surgery to begin habituating the animals to the structure along with the control stimulus. Mice can then live with the additional enrichment, decreasing the salience of the acute stimulus presentation. Habituation to the presence of the tea ball is only one aspect of habituation that is critical to this method. The described habituation protocol also consists of repeated exposure to the testing procedure to normalize the testing environment (i.e., personnel, transportation and movement to the testing location, exposure to stimulus). This extended habituation minimizes nuanced responses from the animals and focuses measurements on the desired dependent variables (e.g., pharmacological or contextual stimuli). Previous assessment of this protocol has identified four trials as the minimum number of habituations necessary before temperature testing within home cages in rats 9. If testing is separated by long periods (more than 2–3 weeks), the animals must be habituated again. For repeated habituation, a minimum of 1–2 trials are sufficient. However, if temperature tests are separated by more prolonged bouts of time, repeating more trials may be necessary. In the continued effort to accustom mice and rats to the testing procedure, an acclimation period before stimulus presentation should be included in every experimental trial. This acclimation time is critical to rebalance temperature and activity after being shifted to the testing location. Rodents tend to have sharp temperature increases due to translocation. Acclimation should consist of a minimum of 1 h without interaction from the experimenter on the day of testing before any addition of a pharmacological agent or contextual stimuli. This is necessary each day of testing. In the outlined home-cage temperature tests, mice have the free range of their home cage to roam in response to the tested stimulus. This can cause variable shifts in activity, impacting the accuracy of temperature readings and, therefore, analysis of thermogenic effects of the independent variable (e.g., pharmacological or contextual stimulus). In recognition of the potential changes in temperature due to activity level, a protocol is included below describing the use of temperature during treadmill walking. Published literature described the successful use of this procedure in rats, and it is currently being employed with mice9,11,15,16. Treadmill walking maintains a constant speed of activity for the testing subject. For this study, treadmills are strictly used to control activity level and, therefore, are set to the lowest available speed on the treadmill to promote walking for mice and a similarly low setting for rats. The following procedure is outlined for temperature measurement of unilateral gastrocnemius in mice and predator odor presentation. The design can be used in conjunction with pharmacological agents and is transferrable to rats and other skeletal muscle groups (i.e., quadriceps) in mice. For rats, transponders can be placed in the gastrocnemius bilaterally and in brown adipose tissue. Due to size and distance limitations, only one transponder can be used per mouse. Minor modifications (e.g., the removal of contextual stimuli) can be made to assess thermogenic responses to pharmacological agents. Protocol These methods can be applied to both rat and mouse models and were performed with institutional approval (Kent State University, IACUC Approval #359 and #340 CN 12-04). Prior to the implementation of protocol, animals should be housed in conformance with the Guide for the Care and Use of Laboratory Animals. 1. Prepare transponder reader NOTE: Prior to use, the transponder reader must be set; the following steps only include setting changes necessary for this study. This portion of the protocol is directly associated with the DAS-8027-IUS portable readers; other reader models should follow instructions provided by the manual to achieve programming results. 1.1. Set Audio Beep to OFF 1.1.1. Turn on the device by pressing the SCAN button and wait for the lighting to appear on the OLED display. Press and hold the BACK/MENU button to get to the menu screen. 1.1.2. Using the NEXT/ENTER button, scroll through the options until OPERATIONAL SETUP. Here, toggle the up or down arrows to turn YES and open the operational submenu. 1.1.3. Using the NEXT/ENTER button, scroll to AUDIO BEEP. As the default setting is ON, toggle the up or down arrows and change the setting to OFF. 1.1.4. Press the NEXT/ENTER button to save this setting change. 1.2. Set Vibrate Upon Read to ON 1.2.1. Follow step 1.1 through step 1.2 or complete the next step directly after step 1.4. 1.2.2. Using the NEXT/ENTER button, scroll to VIBRATE UPON READ. As the default setting is OFF, toggle the up and down arrows and change the setting to ON to feel, via vibration, when the reading has been completed regardless of being able to view the screen. 2. Program transponders NOTE: Each implanted transponder should first be programmed with an animal identification (animal ID or transponder ID). This nomenclature can be used as secondary identification for the test subject (e.g., four digits for mouse strain abbreviation, location of transponder and an additional three to four digits to indicate animal number). Programming can be completed days in advance of surgery while keeping the transponders sterile prior to surgery. 2.1. Enter ID Code on the transponder. 2.1.1. Apply booster coil to the reader head—a specific accessory for model DAS 8027-IUS, which helps in the programming procedure. 2.1.2. Using a gloved hand, place the transponder (within the applicator) into the booster coil. 2.1.3. Turn on the device by pressing the SCAN button and wait for the OLED display to light up. Press and hold the BACK/MENU button to get to the menu screen. 2.1.4. Using the NEXT/ENTER button, scroll through the options until WRITE TRANSPONDER ID. Here, toggle the up or down arrows to turn YES. 2.1.5. Using the NEXT/ENTER button, toggle to ENTER ID CODE. 2.1.6. Use the up and down arrow keys to scroll through numbers and letters. Press NEXT/ENTER after each character selection to move to the following character. 2.1.7. When the ID code is complete, press SCAN to write the transponder. 2.1.8. Remove the transponder from the booster coil and repeat as necessary. Check that the transponder reads temperature changes by warming enclosed transponders between gloved hands and measuring using the temperature scanner. NOTE: AUTO MULTI WRITE and SEQUENTIAL COUNT settings can be set to ON to allow for multiple or sequential transponder programming during a session. Each transponder should be tested during programming. 3. Prepare ‘home cage balls’ 3.1. Place 5 cm × 5 cm odorless/control towel into a tea ball 3.2. Place these home cage balls in new home cages after surgery to begin habituating the animal to the method in which the contextual stimuli will be presented during testing. Replace these home cage balls every 2 weeks. 4. Surgery and postoperative care 4.1. Weigh and record subjects’ pre-surgery body weight. Using an induction chamber, provide anesthesia (e.g., 2–5% isoflurane) to the animal. 4.2. Using electric clippers, completely shave the hind limb. Administer analgesia (e.g., 5mg/kg Ketoprofen, s.c.) in compliance with institutional guidelines. NOTE: Additional analgesia may be required if this procedure is combined with other surgical methods. 4.3. Clean the area with 70% alcohol (or commercially available sterile alcohol wipe) and povidone-iodine wash (or commercially available sterile, individually wrapped betadine swabs) alternating at least three times, ending with povidone-iodine. 4.4. Return the animal to induction chamber, and anesthetize the animal to surgical levels. Then, set up the mouse in a face mask for continued exposure to anesthesia. Apply neomycin ophthalmic ointment to the eyes of the animal to prevent dryness while under anesthesia. NOTE: The procedure should not start until the mouse shows no evidence of pain reception (i.e., corneal reflex, tail pinch response, toe pinch reflex). 4.5. Using only surgical scissors, make a shallow cut through the skin on the right hind limb. 4.6. Moving parallel to the gastrocnemius, place the sharp edge of a preprogrammed and uncapped, sterile transponder into the incision. Ensure that the green plunger faces up and is visible. Continue pushing the transponder applicator into the incision until the opening of the transponder applicator is no longer visible. NOTE: Do not accidentally press the green plunger on the transponder applicator during step 4.6. Premature discharge of the transponder will lead to improper placement. 4.7. Turn the applicator 180°, resulting in the green plunger facing down toward the mouse’s limb, no longer visible to the experimenter. Push the transponder applicator into the final location. Once in ideal placement, adjacent to or partially enclosed in the gastrocnemius, push the green plunger, allowing the applicator’s pressure to guide the investigator’s hand back away from the mouse. 4.8. Using forceps, hold together the opened skin and place a wound clip with sterile autoclip or sterile suture. If needed, use absorbable sutures prior to the sterile autoclip to close the fascia layer. Using the transponder-reader, check the temperature of the mouse muscle. 4.9. Remove the mouse from anesthesia and place it in a clean home cage placed atop a water-circulating heating pad set too low for recovery. Ensure that the home cage includes a tea ball with an odorless towel to begin habituation. NOTE: The mouse should awaken from surgery within 15 min. Food can be placed at the bottom of the cage for easy access during recovery days. 4.10. Postoperative care 4.10.1. Record mouse weights and temperatures daily using a transponder-reader for at least 2 days after surgery or until mice regain or stabilize body weight. 4.10.2. Administer non-narcotic analgesia (e.g., 5mg/kg of ketoprofen, s.c.) once daily to the mice for at least 2 days post-surgery, with additional doses provided as needed. NOTE: Mice and rats should fully recover within 5–8 days of surgery and can undergo habituation and testing procedures. 5. Testing preparation—home cage 5.1. Constructing Risers NOTE: The below step is based on 194 mm × 181 mm × 398 mm mouse filter-topped cages. To fit larger cages (e.g., a rat home-cage), the width will need to be adjusted. 5.1.1. Cut the PVC pipe with a ratcheting PVC cutter into eight sections and assemble following Figure 1C. This will give an open tabletop structure that can hold approximately four cages. Make the desired number of risers. 5.2. Room setup 5.2.1. Assign a location to each riser within the testing room. Separate risers set to receive different contextual stimuli (i.e., odors) by a minimum of 2 m to avoid confounding variables. NOTE: Each mouse should have an assigned testing location within the testing room and on the physical risers as much as feasible to avoid developing associations between different locations and thermogenic stimuli. 5.2.2. Using magnetic strips, attach surgical sheets or gowns across the risers, creating a visual barrier between the researcher and the test subjects. Set this barrier to minimize temperature changes resulting from mouse activity when viewing experimenters moving towards the cage or around the testing room. 5.2.3. (OPTIONAL) Place mirrors on the surface below the risers to ease viewing of the cage bottom during testing. NOTE: Risers can be sanitized through a cage wash system. Cloth or surgical sheets should be laundered prior to habituation and testing. 5.3. Tea ball preparation 5.3.1. Prepare tea balls with control and PO towels (approximately 5 cm × 5 cm). To avoid cross-contamination, prepare control-odor tea balls first. NOTE: Predator-odor towels should be pathogen-tested prior to use. These towels should also be contained, and materials that interact with them should be immediately sanitized (i.e., cage wash), preventing exposure of odor to other animals. 6. Temperature testing—home cage NOTE: Animals need to be habituated to the entire testing procedure excluding experimental contextual or pharmacological stimuli. This should be completed a minimum of 4x before testing. 6.1. Transfer the animals to the prepared tested room. Place the animals in a preassigned location on the riser. This location should be the same throughout all habituation and testing procedures. 6.2. Remove the ‘home cage ball’ from mouse home cage and re-cover the cages with a cloth or surgical sheet. Allow the mice to acclimate to the testing space for 1–2 h. 6.3. After acclimation is complete, use the scanner to measure and record the baseline temperature of each subject. Avoid manipulating cloth coverings during measurements. NOTE: Pharmacological agents can be applied here. Wait time post injection or application can be added as needed before testing. Recording a secondary baseline directly before testing is recommended after the addition of a pharmacological agent to monitor response to pharmacological stimuli. If odor response is not being tested, temperature measurements of the mice can begin directly after injection. Randomization should be employed when providing any stimuli. 6.4. Uncover the cage and place the tea ball (control or PO) onto the floor of the home cage. Replace the cage lid and cloth covering. 6.5. Begin the stopwatch. Measure the temperatures of the test subjects in the same order of tea ball placement. Record temperatures and clock time of measurements following the desired time points. 6.6. When the experiment is complete, remove the treatment ball. Place the mice that received PO in a new home cage with the original ‘home cage ball.’ Return the ‘home cage ball’ to the cage of the mice that received control odor. Transfer the mice to the housing location. NOTE: The above procedure can be translated to rat models within appropriately sized cages. Adjustments to the measurements suggested in Figure 1C may be required to allow for better access to the bottom of the home cage. 7. Temperature testing—treadmill walking 7.1 Assign each animal a treadmill as their assigned location for habituation and testing procedures. 7.2 Prepare the treadmills for testing, ensuring that the shockers are functional. NOTE: For treadmill walking, treadmills should be set at the lowest available pace that promotes continuous movement but not running for both habituation and testing. For the 1012M-2 Modular Enclosed Metabolic Treadmill, this is 5.2 m/min for mice and 7 m/min for rats. This pace may need to be adjusted based on the obesity of the subject. 7.3. Habituation 7.3.1. Move the mice to the testing room. Allow mice 1–2 h to acclimate to the room transfer in their home cages. 7.3.2. After acclimation, guide the mice to the opening of their assigned treadmill and close the treadmill. Start the belt, shocker, and stopwatch. 7.3.3. Allow the mice to walk on the treadmills for 15 min, using shock stimulus as motivation for movement. Stop the test immediately if an animal remains on an active shocker for an extended period. 7.3.4. After the test, remove the mice and return them to the home cages. 7.3.5. Clean the treadmills using liquid detergent and water. 7.4. Testing 7.4.1. Move the mice to the testing room. Allow the mice 1–2 h to acclimate to the room transfer in their home cages. 7.4.2. Measure and record baseline temperature prior to moving the mouse to the treadmill. NOTE: For tests including pharmacological agents, apply or inject them here, following the schematic shown in Figure 2A. Wait time after injection can be added as needed before the mice are placed on the treadmill. Randomization should be employed when providing any stimuli. 7.4.3. Place 5 cm × 5 cm squares of control or PO towels within the treadmill closest to the front of the treadmill. Adhere the towels to the ceiling of the treadmill or underneath for easy placement and removal. 7.4.4. Guide the mice into the assigned treadmill. Turn on the treadmill belt and shocker. 7.4.5. Start the stopwatch. Take measurements of the test subjects in the same order in which the mice were set up in the treadmills. Record the temperatures and clock time of the measurements following the desired time points. NOTE: Temperature measurements can reliably be measured from outside the treadmill while a mouse is inside an enclosed treadmill during walking activity. For rats, the treadmill size and transponder-reader distance limitations may require an experimenter to keep the back of the treadmill open to insert the reader inside the treadmill, closer to the subject. 7.4.6. When the test is complete, turn off the shockers and treadmills; return the mice to their home cages. Transfer the mice to the housing location. 7.4.7. Clean the treadmills using liquid detergent and water, paying specific attention to remove any residual PO. 7.4.8. When experiments are complete animals can be euthanized (e.g., using CO2 inhalation), and visual confirmation of the transponder location can be assessed. Representative Results Transponders were unilaterally implanted into the right gastrocnemius of ten 4–6-month-old, wild-type (WT) mice bred from the SF1-Cre strain (Tg(Nr5a1-cre)7Lowl/J, Strain #012462, C57BL/6J and FVB backgrounds; female N = 5; male N = 5). After recovery, the mice were habituated to a home cage temperature-testing procedure that did not include a contextual stimulus (e.g., PO). Temperature measurements using a transponder wand were recorded within their housing room and after transfer to the testing location. Mice were given 1–2 h to acclimate to the testing room and location. At the completion of acclimation, baseline and consecutive measurements for 1 h were recorded for each mouse. This procedure was completed four times. Overall, no sex differences were observed. Muscle temperatures significantly increased after the mice were moved to the testing room, then decreased by the baseline measurement after 60 min spent in the testing context. Combined-sex analysis of trial 4 showed no significant difference between ‘before move’ and ‘baseline’ temperature measurements (two-tailed, paired t-test, p > 0.10), showing the effectiveness of 1 h acclimation to the testing context. Furthermore, statistical comparison of temperatures at baseline and 60 min showed a significant decrease in temperature (two-tailed, paired t-test, p < 0.01), providing evidence of the mice habituating to the investigator’s movement during measurement. However, females (but not males) showed incremental responses where temperature measured from 5 to 15 min was lower with successive habituation trials (Figure 3). When observing acute effects of moving or rise in temperature after baseline, mice tend to respond less to transport into the testing room over successive habituation trials (Supplementary File 1, trial analysis). Habituated adult WT mice described above were tested with Oxt, a pharmacological agent. Mice were given intraperitoneal injections (i.p., 2 mg/kg) of Oxt or vehicle (sterile saline) in random order, and muscle temperatures were measured prior to movement into the testing room and after 5, 10, 15, 30, 45, 60, 90, 120, 150, and 180 min of injection. Each mouse received both treatments. Repeated measures analysis of variance (ANOVA) revealed significant main effects of Oxt and time, where Oxt decreased muscle temperature relative to the vehicle. Oxt decreased muscle temperature relative to baseline as rapidly as 5 min after injection, with a maximal decrease seen 30 min after injection (Figure 4). Muscle temperatures were normalized by 60 min after Oxt injection (two-tailed, paired t-test, p > 0.10). Adult male Sprague Dawley rats (N = 4, age ~6 months) bilaterally implanted with transponders in the gastrocnemius were habituated and then tested in a home cage setting with PO (ferret odor) stimulus. Baseline measurements were recorded, and each rat was presented with PO in the form of a towel. The odor was then removed after 10 min of exposure; consecutive measurements were taken before and after the removal of the stimulus. These preliminary data (Figure 5) suggest that PO has a continued impact on skeletal muscle thermogenesis after the removal of the stimulus. Previously published data assessed predator threat activation of skeletal muscle thermogenesis in adult male Sprague Dawley rats (age ~6 months)9. Rats with implanted bilateral gastrocnemius transponders were presented with predator (ferret) odor. Measurements were taken in a home cage setting (N = 8, Figure 6A). These data revelaed a robust increase in temperature compared to control odors. To parse out aversive or stressful thermogenic responses to ferret odor, male rats (N = 7, Figure 6B) were presented with an aversive odor (butyric acid), a novel odor (2-methylbenzoxazole), fox odor, or restrained for 1 min prior to testing (moderate stress). Measurements were taken in a home cage setting over a 2 hr period. Analysis of these data showed ferret odor to produce and maintain a strong change in thermogenesis compared to all other conditions. Together, these data provide evidence of the control odor’s minimal and transient influence on skeletal muscle thermogenesis. Discussion This temperature testing protocol provides the field with an avenue to measure skeletal muscle thermogenesis directly. This is critical as research delves into identifying mechanisms underlying muscle thermogenesis34. The method provides two cost-effective protocols for measuring skeletal muscle thermogenesis under contextual and pharmacological conditions. This protocol emphasizes the importance of both habituation and acclimation within these procedures. Habituation is used to repeatedly introduce the test subject to the testing procedure without the introduction of any pharmacological or contextual stimuli; it is a critical component of both home cage and treadmill temperature testing. This gives time for the animals to familiarize themselves with the environment while decreasing the salience of the experimental context. Omitting this step can lead to biased associations with the experimental stimulus, as well as elevated thermogenic responses to control stimuli9. Animals must learn the procedure to reduce stress responses to the general movement and manipulation required to test animals under these protocols. Example data collected provides evidence of the necessity of repeated habituation (Figure 3). In a similar effort, acclimation on the day of testing is necessary for each trial. Acclimation is a daily assimilation tool, giving the animals time to relax from the stressors of translocation to the testing room. Skipping acclimation can give inaccurate baseline temperature measurements, interreacting with any later assessment. Here, muscle thermogenic measurements were used to demonstrate the hypothermic effect of intraperitoneal Oxt on mice. This outcome was surprising considering evidence supporting central Oxt’s role in thermogenesis, and specifically in social hyperthermia12,14. Others, however, have demonstrated the ability of both Oxt and vasopressin to suppress core temperature along with heart rate in rats, effects mediated by the Avpr1a receptor35. This apparent paradox has not been reconciled. It is possible that the ability of Oxt to increase or decrease the temperature in different contexts may stem from central versus peripheral action of Oxt or from the length of exposure14,36–38. Regardless, here we demonstrate that mouse muscle temperature shows a sizable decrease in temperature rapidly after peripheral Oxt injection (Figure 4), consistent with the changes in rat core temperature reported by Hicks et al. (2014) 33. In accordance with the National Institute of Health’s (NIH) expectation that investigators factor in sex as a biological variable, muscle thermogenesis is measured in males and females, in both mice and rats. Thermogenesis data from males and females can be compared, though previous and current studies have failed to identify robust sex differences in contextual thermogenesis and variation across the estrous cycle in female rats 9. One exception is the evident sex difference in muscle temperature at baseline and after transport to the testing area, particularly prior to habituation 9. This may stem from differences in locomotion after transport, as female rats have a higher locomotor response to some stressful stimuli compared to males, separable from underlying anxiety measures39. This underscores the need for repeated habituation to the experimental context, in this case, to avoid misrepresenting a sex difference in thermogenesis that may be attributed to the experimental stimulus rather than the underlying differences in the stress response. The primary method of temperature testing within animal home cages has some limitations, one being the control of variable activity levels. This can be critical as increased activity leads to increased muscle temperature. To address this, a procedure for mice and rat treadmill walking has been outlined. Controlling the animal’s movement minimizes the potential for an activity effect on the temperature, factoring out differences in contractile thermogenesis. While treadmill walking can be completed as a solo test, this method can be used in conjunction with home-cage temperature assessment. The combined analysis provides further evidence for assertions that skeletal muscle temperature changes stem from pharmacological or contextual stimuli rather than secondarily from changes in activity resulting from these stimuli9,15,16. Additionally, this method is limited in that it is mildly invasive, which does not meet the need of some research studies. However, this method only requires a single surgery, allowing researchers to avoid continual animal manipulation during testing while maintaining the specificity of the measurements. Furthermore, the currently available size of the IPTT-300 transponder does not allow the transponder to be placed directly within the gastrocnemius of a mouse. This can be completed within rat models due to their larger size. This method provides a mechanism of measurement adjacent to the muscle of interest; nevertheless, remodeled, or smaller versions of transponders capable of measuring temperature would be a great asset to the field and future studies. The broad use of the described method in our research program has given us the opportunity to manage variance in response to transponder implantation and testing procedures9,11,15,16. After implantation of the transponder, monitoring the temperatures of the animals immediately after surgery and during recovery is recommended. While this first gives insight into the animal’s health (e.g., oddly low temperature as a sign of illness or impending fatality), it also provides evidence of the transponder still being active and secured in place. A rat or mouse may scratch at the incision location, potentially resulting in the transponder either partially or completely falling out. In compliance with institutional guidelines, this surgery is considered minor. Therefore, in cases of unilateral transponder placement, if a mouse loses their transponder or if the mouse’s transponder no longer functions, the surgery can be repeated on an alternate limb. A marking (i.e., identification of new placement, or “R” for replacement) to indicate this repeated surgery is noted during the programming of the transponder as a part of the animal identification name is recommended. Furthermore, since animals have the free range of their cage, researchers may have difficulty finding the animal to take the reading. It is suggested that researchers utilize the habituation phase to practice measurements and assess their setup. Alterations may include increasing the number of experimenters and transponder scanners or decreasing the number of risers, and therefore animals, tested in each trial. This protocol provides instruction for direct temperature measurement of muscle without additional software analysis, resulting in a feasible and relatively low-cost avenue for studies where infrared cameras are typically used. Moreover, this procedure enables the collection of data that closes the gap seen by some studies seeking to connect gene or protein changes to muscle thermogenesis38. All in all, increased interest in muscle thermogenesis and its mechanisms are facilitated by direct assessment of the heat generated in the target muscle. The described procedure directly addresses this methodological void within the field by providing a mechanism for studying the skeletal muscle of both mice and rats. Supplementary Material Supplemental file 1 Acknowledgments This work is supported by R15 DK097644, R15 DK108668, and R15 DK121246. We thank Dr. Chaitanya K Gavini and Dr. Megan Rich for prior contributions and Dr. Stanley Dannemiller for ensuring our compliance with institutional animal use guidelines. A special thank you to Dr. Tim Bartness for providing the fundamental research necessary to build this method and its associated studies. Figure 1A, C, D and Figure 2A were created using Biorender.com. Figure 1: Transponders and home cage temperature testing. (A) Diagram of unilateral transponder placement for testing temperature in a mouse gastrocnemius. Once programmed and placed, the transponder-reader (DAS-8027-IUS, shown) can be used to measure temperature. (B) Left, photo of an open mesh stainless steel tea ball and a 5 cm × 5 cm towel. Right, enclosed tea ball, used to hold habituation and odor towels in home cage testing. (C) Schematic of risers constructed with PVC piping for home cage testing. (D) Workflow of home cage testing protocol. (E) Facility images of home cage testing area. Left, four mouse cages atop a riser. Magnetic strips are located on the adjacent wall, magnets, and surgical cloth are on the table. Right, covered mouse cages on risers. A, C, and D were created with Biorender.com. Figure 2: Activity-controlled temperature testing. (A) Workflow of activity-controlled temperature testing with a pharmacological agent using treadmill walking. (B) Facility images of treadmills. Left, an image of full equipment setup. Right, a closer image of individual treadmills and shockers. A was created with Biorender.com. Figure 3: Analysis of muscle temperature during habituation for home cage temperature testing. Mice unilaterally implanted with transponders in the right gastrocnemius were habituated to the testing procedure. Mice were measured in the animal housing room, ‘Before Move,’ in the testing room, ‘After Move,’ after acclimation for 1–2 h, ‘Baseline,’ then consecutively over 1 h. All statistical comparisons shown were made between trial 1 and trial 4, * p < 0.05, ** < 0.01 (t-test, N = 10); † < 0.05, †† < 0.01, ‡ < 0.001 main effect trial (ANOVA, N [trials] = 4). Error bars shown display the standard error of the mean (SEM). Figure 4: Muscle temperature during pharmacological stimulation of oxytocin in mice. Habituated mice, unilaterally implanted with transponders, were given 2 mg/kg (i.p.) of either oxytocin or vehicle (sterile saline). Significant decreases in muscle temperature were observed at 5 min after injection of oxytocin and normalized by 60 min, ** < 0.01, *** < 0.001 (two-tailed paired t-test, N = 9). Error bars shown display the standard error of the mean. Figure 5: Predator-odor thermogenesis in rat home cage temperature testing. Temperature measurements in rats with transponders implanted bilaterally in the gastrocnemius after exposure to predator (ferret) odor for 10 min. After exposure for 10 min, towels containing stimulus were removed, as indicated by the arrow. Rats maintained increased temperature 20 min after stimulus removal. Significantly greater than baseline temperature, * p < 0.05, ** < 0.01, *** < 0.001 (t-test, N = 4). Error bars shown display the standard error of the mean. Figure 6: Ferret odor induces a rapid rise in muscle temperature compared to control. (A) Gastrocnemius temperature was significantly elevated after predator (ferret) odor compared with control exposure in male rats (two-tailed paired t-test, N = 8). (B) Novel, aversive, or fox odors did not significantly change muscle temperature compared to control. Temperature change induced by moderate stress quickly declined after 5 minutes. Ferret odor maintained a robust response, in comparison to other conditions, for the entirety of the test (ANOVA, N = 7). † p < 0.05, ferret odor > all other conditions; * p < 0.025, point comparison between ferret odor and moderate stress vs control odor. Adapted with permission from the Journal of Experimental Biology9. A complete version of this article that includes the video component is available at http://dx.doi.org/10.3791/64264. Disclosures The authors declare that they have no conflicts of interest. References 1 Watts CA Measuring Skeletal Muscle Thermogenesis in Mice and Rats. J Vis Exp. 10.3791/64264 (185 ), (2022). 2 Periasamy M , Herrera JL & Reis FCG Skeletal Muscle Thermogenesis and Its Role in Whole Body Energy Metabolism. Diabetes Metabolism Journal. 41 (5 ), 327–336, (2017).29086530 3 Rowland LA , Bal NC & Periasamy M The role of skeletal-muscle-based thermogenic mechanisms in vertebrate endothermy. Biological Reviews of the Cambridge Philosophical Society. 90 (4 ), 1279–1297, (2015).25424279 4 Maurya SK Sarcolipin Is a Key Determinant of the Basal Metabolic Rate, and Its Overexpression Enhances Energy Expenditure and Resistance against Diet-induced Obesity. Journal of Biological Chemistry. 290 (17 ), 10840–10849, (2015).25713078 5 Grigg G Whole-body endothermy: ancient, homologous and widespread among the ancestors of mammals, birds and crocodylians. Biological Reviews of the Cambridge Philosophical Society. 97 (2 ), 766–801, (2022).34894040 6 Franck JPC , Slight-Simcoe E & Wegner NC Endothermy in the smalleye opah (Lampris incognitus): A potential role for the uncoupling protein sarcolipin. Comparative Biochemistry and Physiology - Part A: Molecular & Integrative Physiology. 233 48–52, (2019). 7 Nowack J Muscle nonshivering thermogenesis in a feral mammal. Scientific Reports. 9 (1 ), 6378, (2019).31011179 8 Oliver SR , Anderson KJ , Hunstiger MM & Andrews MT Turning down the heat: Down-regulation of sarcolipin in a hibernating mammal. Neuroscience Letters. 696 13–19, (2019).30528880 9 Gorrell E Skeletal muscle thermogenesis induction by exposure to predator odor. The Journal of Experimental Biology. 223 (Pt 8 ), (2020). 10 Gavini CK Leanness and heightened nonresting energy expenditure: role of skeletal muscle activity thermogenesis. The American Journal of Physiology - Endocrinology and Metabolism. 306 (6 ), E635–647, (2014).24398400 11 Almundarij TI , Gavini CK & Novak CM Suppressed sympathetic outflow to skeletal muscle, muscle thermogenesis, and activity energy expenditure with calorie restriction. Physiological Reports. 5 (4 ), (2017). 12 Harshaw C , Lanzkowsky J , Tran AD , Bradley AR & Jaime M Oxytocin and ‘social hyperthermia’: Interaction with beta3-adrenergic receptor-mediated thermogenesis and significance for the expression of social behavior in male and female mice. Hormones and Behavior. 131 104981, (2021).33878523 13 Caldwell HK Oxytocin and Vasopressin: Powerful Regulators of Social Behavior. The Neuroscientist. 23 (5 ), 517–528, (2017).28492104 14 Harshaw C , Leffel JK & Alberts JR Oxytocin and the warm outer glow: Thermoregulatory deficits cause huddling abnormalities in oxytocin-deficient mouse pups. Hormones and Behavior. 98 145–158, (2018).29277701 15 Gavini CK , Britton SL , Koch LG & Novak CM Inherently Lean Rats Have Enhanced Activity and Skeletal Muscle Response to Central Melanocortin Receptors. Obesity (Silver Spring). 26 (5 ), 885–894, (2018).29566460 16 Gavini CK , Jones WC 2nd & Novak CM Ventromedial hypothalamic melanocortin receptor activation: regulation of activity energy expenditure and skeletal muscle thermogenesis. The Journal of Physiology. 594 (18 ), 5285–5301, (2016).27126579 17 Zaretsky DV , Romanovsky AA , Zaretskaia MV & Molkov YI Tissue oxidative metabolism can increase the difference between local temperature and arterial blood temperature by up to 1.3(o)C: Implications for brain, brown adipose tissue, and muscle physiology. Temperature (Austin, Tex.). 5 (1 ), 22–35, (2018).29687042 18 Yoo Y Exercise activates compensatory thermoregulatory reaction in rats: a modeling study. Journal of Applied Physiology. 119 (12 ), 1400–1410, (2015).26472864 19 Langer F & Fietz J Ways to measure body temperature in the field. Journal of Thermal Biology. 42 46–51, (2014).24802148 20 Pence S Central Apolipoprotein A-IV Stimulates Thermogenesis in Brown Adipose Tissue. International Journal of Molecular Sciences. 22 (3 ), (2021). 21 Li D Homeostatic disturbance of thermoregulatory functions in rats with chronic fatigue. Journal of Neuroscience Research. 165 45–50, (2021). 22 Carlier J Pharmacodynamic Effects, Pharmacokinetics, and Metabolism of the Synthetic Cannabinoid AM-2201 in Male Rats. Journal of Pharmacology and Experimental Therapeutics. 367 (3 ), 543–550, (2018).30266766 23 Pato AM , Romero DM , Sosa Holt CS , Nemirovsky SI & Wolansky MJ Use of subcutaneous transponders to monitor body temperature in laboratory rats. Journal of Pharmacological and Toxicological Methods. 114 107145, (2022).34958946 24 Almeida DL Lean in one way, in obesity another: effects of moderate exercise in brown adipose tissue of early overfed male Wistar rats. International Journal of Obesity (Lond). 46 (1 ), 137–143, (2022). 25 Brito MN , Brito NA , Baro DJ , Song CK & Bartness TJ Differential activation of the sympathetic innervation of adipose tissues by melanocortin receptor stimulation. Endocrinology. 148 (11 ), 5339–5347, (2007).17702843 26 Vaughan CH , Shrestha YB & Bartness TJ Characterization of a novel melanocortin receptor-containing node in the SNS outflow circuitry to brown adipose tissue involved in thermogenesis. Brain Research. 1411 17–27, (2011).21802070 27 Kort WJ , Hekking-Weijma JM , TenKate MT , Sorm V & VanStrik R A microchip implant system as a method to determine body temperature of terminally ill rats and mice. Laboratory Animals. 32 (3 ), 260–269, (1998).9718473 28 Mei J Body temperature measurement in mice during acute illness: implantable temperature transponder versus surface infrared thermometry. Scientific Reports. 8 (1 ), 3526, (2018).29476115 29 Warn PA Infrared body temperature measurement of mice as an early predictor of death in experimental fungal infections. Laboratory Animals. 37 (2 ), 126–131, (2003).12689423 30 Hargreaves K , Dubner R , Brown F , Flores C & Joris J A new and sensitive method for measuring thermal nociception in cutaneous hyperalgesia. Pain. 32 (1 ), 77–88, (1988).3340425 31 Fiebig K , Jourdan T , Kock MH , Merle R & Thone-Reineke C Evaluation of Infrared Thermography for Temperature Measurement in Adult Male NMRI Nude Mice. Journal of the American Association for Laboratory Animal Science. 10.30802/AALAS-JAALAS-17-000137, (2018). 32 Franco NH , Geros A , Oliveira L , Olsson IAS & Aguiar P ThermoLabAnimal - A high-throughput analysis software for non-invasive thermal assessment of laboratory mice. Physiology & Behavior. 207 113–121, (2019).31078672 33 Koganti SR Disruption of KATP channel expression in skeletal muscle by targeted oligonucleotide delivery promotes activity-linked thermogenesis. Molecular Therapy. 23 (4 ), 707–716, (2015).25648265 34 Bal NC & Periasamy M Uncoupling of sarcoendoplasmic reticulum calcium ATPase pump activity by sarcolipin as the basis for muscle non-shivering thermogenesis. Philosophical Transactions of the Royal Society B. 375 (1793 ), 20190135, (2020). 35 Hicks C Body temperature and cardiac changes induced by peripherally administered oxytocin, vasopressin and the non-peptide oxytocin receptor agonist WAY 267,464: a biotelemetry study in rats. British Journal of Pharmacology. 171 (11 ), 2868–2887, (2014).24641248 36 Kasahara Y Oxytocin receptor in the hypothalamus is sufficient to rescue normal thermoregulatory function in male oxytocin receptor knockout mice. Endocrinology. 154 (11 ), 4305–4315, (2013).24002032 37 Kasahara Y Role of the Oxytocin Receptor Expressed in the Rostral Medullary Raphe in Thermoregulation During Cold Conditions. Frontiers in Endocrinology (Lausanne). 6 180, (2015). 38 Yuan J , Zhang R , Wu R , Gu Y & Lu Y The effects of oxytocin to rectify metabolic dysfunction in obese mice are associated with increased thermogenesis. Molecular and Cellular Endocrinology. 514 110903, (2020).32531419 39 Scholl JL , Afzal A , Fox LC , Watt MJ & Forster GL Sex differences in anxiety-like behaviors in rats. Physiology & Behavior. 211 112670, (2019).31487491

PMC009xxxxxx/PMC9989844.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9802571 20730 Mol Cell Mol Cell Molecular cell 1097-2765 1097-4164 35868255 9989844 10.1016/j.molcel.2022.06.028 NIHMS1878865 Article You shall not pass! unveiling the barriers for cohesin-mediated loop extrusion Meyer-Nava Silvia 12 http://orcid.org/0000-0002-7566-3875 Rivera-Mulia Juan Carlos 123* 1 Department of Biochemistry, Molecular Biology and Biophysics, University of Minnesota Medical School, Minneapolis, MN, USA 2 Stem Cell Institute, University of Minnesota Medical School, Minneapolis, MN, USA 3 Masonic Cancer Center, University of Minnesota Medical School, Minneapolis, MN, USA * Contact information: Juan Carlos Rivera-Mulia, Department of Biochemistry, Molecular Biology and Biophysics, University of Minnesota Medical School, 6-155 Jackson Hall, 321 Church Street SE, Minneapolis, MN 55455, Lab phone number: 612-301-2572, riveramj@umn.edu 1 3 2023 21 7 2022 21 7 2023 82 14 25412543 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Dequeker et al., 2022 performed elegant in vivo, in silico and in vitro experiments to demonstrate that the MCM complex, an essential DNA replication factor, is an obstacle for the DNA loop formation by cohesin. pmcGenome organization is critical to maintain gene function and is closely coordinated with the temporal order of DNA replication (Rivera-Mulia and Gilbert, 2016). On a large scale, chromosomes are divided into active and inactive compartments that align with early and late replication (Marchal et al., 2019). At the finest scale, chromatin is organized into loops that facilitate contact between regulatory elements, such as enhancer-promoter interactions. Chromatin loops are formed by the active extrusion of chromatin by the cohesin complex (Banigan and Mirny, 2020). Until now, the CCCTC-binding factor (CTCF) was the only well characterized loop extrusion barrier in vertebrates (Li et al., 2020). In a recent study published in Nature, Dequeker and Scherr et al. demonstrated that essential factors of the DNA replication machinery restrict the loop extrusion mediated by cohesin. Cohesin, a member of the structural maintenance of chromosomes family of proteins, is a ring-shaped complex and plays a fundamental role in chromatin loop formation (Oldenkamp and Rowland, 2022). Cohesin acts as a molecular motor generating loops by reeling in the chromatin fiber (Figure 1A-B), and loop sizes are determined by the dynamic loading and release of cohesin, as well as by the presence of factors that can pause, stop or remove cohesin (Banigan and Mirny, 2020). NIPBL is a factor required for cohesin loading, WAPL is a release factor, and CTCF convergent sites stop cohesin (Popay and Dixon, 2022). However, chromatin is crowded with numerous proteins, and processes like transcription, replication, and DNA repair, that can affect the dynamics of cohesin loading, release and translocation. MCM complex is a double hexamer of the minichromosome maintenance proteins MCM2-MCM7 (referred to as MCM for simplicity hereafter), that functions as helicase during DNA replication. MCM is also a ring-like complex and is loaded onto chromatin by cdc10-dependent transcript 1 (CDT1) in G1. Although MCM remains inactive until S-phase, the complex is irreversibly wrapped around the chromatin until replication initiates (Kuipers et al., 2011). To ensure all DNA is replicated, cells assemble many more helicases than they need, with the excess rescuing replication under replicative stress conditions. In fact, MCM is loaded at ~250,000 sites but only <10% of these are activated in any given cell (Rivera-Mulia and Gilbert, 2016). Thus, given the abundance of MCM and its unusually stable association to the chromatin, Dequeker and Scherr et al. analyzed MCM as a potential obstacle for cohesin-mediated loop extrusion. First, the authors investigated whether MCM loss affects loop extrusion using the oocyte-to-zygote transition as a model, which allowed them to analyze and manipulate MCM loading onto new paternal chromatin without disrupting cell cycle progression. Using single-nucleus Hi-C analyses, and a non-degradable geminin protein (geminin sequesters CDT1inhibiting MCM loading), they found that MCM loss increases CTCF-mediated loops and TADs. These findings suggest that MCM restricts loop formation (Figure 1C). A conditional knockout of cohesin was also used to demonstrate that cohesin is responsible for the rise in CTCF-loops upon MCM loss. To test if CTCF and MCM function together to establish the strength of CTCF-mediated loops, they exploited a CTCF knockdown in MCM-deficient zygotes. Both loops and TADs were lost with or without MCM, indicating that MCM has no function in establishing loop boundaries, thus only CTCF sites act as loop anchors (Figure 1D). Another possibility could be that MCM cooperate with WAPL for cohesin unloading. To test this hypothesis, the authors analyzed combined MCM loss and WAPL knockout. They found even a stronger increase in CTCF-loops and TADs than in the individual conditions, suggesting that MCM further restricts loop extrusion when cohesin remains wrapped onto the chromatin. Although only a modest increase in CTCF-loops were observed after acute MCM depletion in HCT116 cells, their findings suggest that MCM could also act as barrier for cohesin-mediated loop extrusion in somatic cells. However, further studies are needed to demonstrate whether the helicase complex can restrict loop extrusion in differentiated cell types. Next, they performed in silico analyses of polymer models of loop extrusion to identify potential mechanisms of MCM impediment for cohesin function. The simulation parameters predicted that MCM is a “semi-permeable” barrier to loop extrusion, where cohesin can bypass CTCF at ~50% and MCM at ~20% of encounters. This predicted semi-permeability of MCM explain why CTCF loops can be formed even in the presence of MCM. Moreover, the simulation also explains why MCM can reduce the frequency of CTCF-loops without significantly changing the size of the loops: since MCM is positioned stochastically in the chromatin, it acts as a random barrier with marginal effects on average loop sizes. Finally, the authors exploited a single-molecule assay to measure MCM interference with cohesin loop extrusion. Briefly, MCM was loaded onto tethered DNA molecules by the licensing factors ORC, Cdc6, and Cdt1, followed by cohesin loading. Then, total internal reflection fluorescence (TIRF) microscopy was used to measure cohesin-MCM interactions. Cohesin binding to DNA did not change with the presence of MCM; however, MCM reduced cohesin translocation by four-fold. Moreover, a fraction of loaded cohesin did not pass MCM complexes. Since this assay was performed with yeast MCM (due to the sequence specificity of yeast replication origins), the authors also tested a “humanized” MCM containing an additional motif reported to interact with cohesin (Li et al., 2020). They found that the humanized MCM are even stronger barriers for cohesin-mediated loop extrusion. Overall, these findings demonstrate that MCM can restrict loop extrusion by cohesin in G1. This study introduced a novel class of loop formation barriers linking elements of the DNA replication machinery to the loop formation by cohesin. An interesting remaining question brought by the authors, given the evolutionary conservation of MCM, is whether this helicase could be an ancestral barrier for loop formation in organisms with CTCF-independent loop anchors. Moreover, a parallel study showed that loop extrusion by cohesin confines the location of replication origins in human cells (Emerson et al., 2022). Thus, it remains unclear if cohesin can bypass only a fraction of MCM complexes while others are pushed towards the loop boundaries to fire origins in S-phase. Dissecting the links between structure and function would require further studies, but novel tools allowing manipulation of DNA loop formation and replication control will reveal insights into the molecular rules governing genome architecture and duplication. Acknowledgments J.C.R.M. is supported by funding from the National Institute of General Medical Sciences of the National Institutes of Health under award number R35GM137950. Figure 1. Cohesin-loop extrusion barriers. A) Cohesin complex is loaded onto chromatin. B) Loop extrusion is mediated by cohesin ring and can proceed when no barriers are present (green traffic light). C) The inactive helicase MCM loaded onto chromatin in G1, functions as a semi-permeable barrier for the loop extrusion. Dequeker et al., showed that when cohesin encounters MCM, loop extrusion is paused (yellow traffic light). D) Loop extrusion is stopped when cohesin encounters convergent CTCF sites (red traffic light). Declaration of Interests The authors declare no competing interests. References Banigan EJ , and Mirny LA (2020). Loop extrusion: theory meets single-molecule experiments. Curr. Opin. Cell Biol 64 , 124–138. 10.1016/j.ceb.2020.04.011.32534241 Dequeker BJH , Scherr MJ , Brandão HB , Gassler J , Powell S , Gaspar I , Flyamer IM , Lalic A , Tang W , Stocsits R , (2022). MCM complexes are barriers that restrict cohesin-mediated loop extrusion. Nature 606 , 197–203. 10.1038/s41586-022-04730-0.35585235 Emerson DJ , Zhao PA , Cook AL , Barnett RJ , Klein KN , Saulebekova D , Ge C , Zhou L , Simandi Z , Minsk MK , (2022). Cohesin-mediated loop anchors confine the locations of human replication origins. In press. Nature. Published online 08 June 2022. 10.1038/s41586-022-04803-0. Kuipers MA , Stasevich TJ , Sasaki T , Wilson KA , Hazelwood KL , McNally JG , Davidson MW , and Gilbert DM (2011). Highly stable loading of Mcm proteins onto chromatin in living cells requires replication to unload. J. Cell Biol 192 , 29–41. 10.1083/jcb.201007111.21220507 Li Y , Haarhuis JHI , Sedeño Cacciatore Á , Oldenkamp R , van Ruiten MS , Willems L , Teunissen H , Muir KW , de Wit E , Rowland BD , (2020). The structural basis for cohesin-CTCF-anchored loops. Nature 578 , 472–476. 10.1038/s41586-019-1910-z.31905366 Marchal C , Sima J , and Gilbert DM (2019). Control of DNA replication timing in the 3D genome. Nat. Rev. Mol. Cell Biol 7 , 455. 10.1038/s41580-019-0162-y. Oldenkamp R , and Rowland BD (2022). A walk through the SMC cycle: From catching DNAs to shaping the genome. Mol. Cell 82 , 1616–1630. 10.1016/j.molcel.2022.04.006.35477004 Popay TM , and Dixon JR (2022). Coming full circle: on the origin and evolution of the looping model for enhancer-promoter communication. J. Biol. Chem 102117. 10.1016/j.jbc.2022.102117.35691341 Rivera-Mulia JC , and Gilbert DM (2016). Replicating Large Genomes: Divide and Conquer. Mol. Cell 62 , 756–765. 10.1016/j.molcel.2016.05.007.27259206

PMC010xxxxxx/PMC10028731.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0404511 7473 Science Science Science (New York, N.Y.) 0036-8075 1095-9203 36423276 10028731 10.1126/science.add7450 NIHMS1863051 Article RNA-activated protein cleavage with a CRISPR-associated endopeptidase Strecker Jonathan 12345†* Demircioglu F. Esra 12345† Li David 123456 Faure Guilhem 12345 Wilkinson Max E. 12345 Gootenberg Jonathan S. 3 Abudayyeh Omar O. 3 Nishimasu Hiroshi 78910 Macrae Rhiannon K. 12345 Zhang Feng 12345* 1 Howard Hughes Medical Institute, Cambridge, MA 02139, USA 2 Broad Institute of MIT and Harvard, Cambridge, MA 02142, USA 3 McGovern Institute for Brain Research, Cambridge, MA 02139, USA 4 Department of Brain and Cognitive Sciences, Cambridge, MA 02139, USA 5 Department of Biological Engineering, Cambridge, MA 02139, USA 6 Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 7 Structural Biology Division, Research Center for Advanced Science and Technology, The University of Tokyo, Tokyo 153-8904, Japan 8 Department of Chemistry and Biotechnology, Graduate School of Engineering, The University of Tokyo, Tokyo 113-8656, Japan 9 Department of Biological Sciences, Graduate School of Science, The University of Tokyo, Tokyo 113-0033, Japan 10 Inamori Research Institute for Science, 620 Suiginya-cho, Kyoto 600-8411, Japan † These authors contributed equally to this work. Author contributions: J.S. designed and performed experiments. F.E.D. solved cryo-EM structures with input from M.E.W.. J.S. and D.L. analyzed data. G.F. performed bioinformatic analysis and structural predictions. F.Z. supervised the research and experimental design with support from R.K.M.. J.S.G., O.O.A., and H.N. provided input and discussed unpublished work. J.S., F.E.D., and D.L. wrote the manuscript with input from all authors. * Corresponding authors: zhang@broadinstitute.org (F.Z.) and strecker@broadinstitute.org (J.S.) 17 1 2023 25 11 2022 03 11 2022 21 3 2023 378 6622 874881 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. CRISPR-Cas systems provide adaptive immune responses in prokaryotes against foreign genetic elements through RNA-guided nuclease activity. Recently, additional genes with non-nuclease functions have been found in genetic association with CRISPR systems, suggesting there may be other RNA-guided non-nucleolytic enzymes. One such gene encodes the TPR-CHAT protease Csx29, which is associated with the CRISPR effector Cas7–11. Here, we demonstrate that this CRISPR-associated protease (CASP) exhibits programmable RNA-activated endopeptidase activity against a sigma factor inhibitor to regulate a transcriptional response. Cryo–electron microscopy of an active and substrate-bound CASP complex reveals an allosteric activation mechanism that reorganizes Csx29 catalytic residues upon target RNA binding. This work reveals an RNA-guided function in nature which can be leveraged for RNA sensing applications in vitro and in human cells. pmcProkaryotes possess a multitude of defense systems against foreign genetic elements, including clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated proteins (Cas) systems (1–3). While the predominant function of CRISPR-Cas systems is to provide adaptive immunity via RNA-guided DNA or RNA nuclease activity, additional proteins have been identified in genetic association with CRISPR loci (3–5). One example is that of the CRISPR-associated transposase (CAST) systems (6, 7), which perform RNA-guided DNA insertion whereby nuclease inactive CRISPR effectors guide Tn7-like mobile genetic elements to specific DNA sequences (8, 9). CAST systems have evolved on at least three separate occasions (10), highlighting the ability of diverse CRISPR effectors to acquire, or be acquired by, other bacterial enzymes. Beyond CAST systems, additional functions genetically linked to CRISPR-Cas systems are beginning to emerge, and more likely remain to be discovered and characterized. Previous work has uncovered several RNA-targeting type III CRISPR-associated protease (CASP) systems (3, 4), including a Lon protease that responds to cyclic oligoadenylate second messengers (cA4) to cleave the CRISPR-T protein (11). A recently characterized subtype III-E effector Cas7–11 (12, 13) (also referred to as gRAMP) is likewise associated with a protease, a CHAT family member containing tetratricopeptide repeats (TPR-CHAT, or Csx29). In contrast to prototypical type III CRISPR systems consisting of multi-subunit Csm/Cmr complexes (14), Cas7–11 effectors contain naturally fused Cas7 and Cas11 domains (3). Members of the CHAT family of proteases harbor catalytic cysteine residues and include eukaryotic caspases involved in programmed cell death (15), and Cas7–11-Csx29 was previously hypothesized to act as a bacterial caspase and support viral immunity (12, 13). Notably, Cas7–11 and Csx29 from Candidatus Scalindua brodae were shown to form a stable protein complex (13), but the substrate and function of the associated protease is unknown. Here, we determine the protein substrate, structure, and mechanism of a type III-E CRISPR-associated protease (CASP) from the marine anaerobe Desulfonema ishimotonii, reveal insight into its natural function in coordinating a transcriptional response to foreign genetic material, and engineer it for novel RNA sensing applications in vitro and in human cells. A Cas7–11-Csx29 complex cleaves the Csx30 protein The reported cleavage of CRISPR-T by the neighboring Lon protease (11) inspired us to look more closely at type III-E loci for potential substrates. In addition to the associated Csx29 protease, these loci frequently contain three additional genes (csx30, csx31, and a predicted sigma factor (3), hereafter CASP-σ) that we hypothesized were prime candidates (Fig. 1A, fig. S1). Starting from a system found in D. ishimotonii (DiCASP) (12), we purified a stable Cas7–11-Csx29-crRNA complex (as previously reported for Candidatus S. brodae (13)) (fig. S2A) and performed in vitro reactions by adding the proteins expressed from the three upstream genes in the presence or absence of a target RNA complementary to the crRNA. We identified that the largest protein, Csx30, is specifically cleaved in response to a target RNA (Fig. 1, B and C). Moreover, in vitro reactions yielded two precise protein products indicating a single cleavage event within Csx30 as opposed to processive protein degradation. We determined the requirements of Csx30 cleavage and found that while mutating the catalytic residues of the Csx29 protease (H615A/C658A) abolished activity, disrupting the catalytic sites of the Cas7–11 endonuclease (D429A/D654A) (12) did not (Fig. 1D, and fig. S2B). This result indicates that target RNA binding alone is sufficient for Csx29 activation, and that RNA cleavage is dispensable. In vitro characterization revealed that DiCASP is a highly active ATP-independent protease cleaving 100-fold molar excess of Csx30 substrate in minutes, with an optimal activity at 37–45°C (fig. S2, C–F). Full Csx30 cleavage activity required 22 nucleotides of complementarity between the crRNA and target RNA, and we detected low tolerance to base pair mismatches, particularly at the 5’ end of the target RNA (fig. S3). Characterization of Csx30 proteolytic processing Structural prediction of the Csx30 protein revealed two domains separated by a flexible linker (Fig. 1, E and F) which we hypothesized to be the site of cleavage. However, mass spectrometry analysis (and the estimated 48 kDa and 16 kDa gel products) indicate that Csx30 is cleaved further downstream between residues 427 and 429 (fig. S4), placing the cleavage site within a small flexible loop (residues 423–437) in the C-terminal domain of the structural model. By generating truncation mutations of Csx30, we determined that the N-terminal domain is dispensable for processing by Cas7–11-Csx29 as Csx30 fragments containing residues 396–565 were efficiently cleaved in vitro (Fig. 1G, and fig. S5). By contrast, we observed that Csx30 C-terminal residues are strictly required and that even a twenty amino acid truncation (Csx301–544) abolished cleavage activity (Fig. 1G). Mutational analysis by alanine substitutions revealed no Csx30 residues that are essential for cleavage, although some reduced the efficiency (Fig. 1H, and fig. S6). Instead, the size of the cleaved loop appears important for processing. We observed that truncating the loop by four residues, or deleting M427 alone, prevented Csx30 cleavage, while the deletion of D430 had no effect (Fig. 1H). Using an uncleavable Csx30Δloop mutant as bait, we pulled down Cas7–11-Csx29 complex both in the presence and absence of target RNA, suggesting that Csx30 binding to Cas7–11-Csx29 is not regulated by target RNA recognition or activation of the protease (fig. S7). In contrast, we did not detect Cas7–11-Csx29 binding using a truncated Csx301–544 mutant, revealing that an intact C-terminal domain is required for substrate binding (fig. S7). Allosteric activation of Csx29 upon target RNA binding To gain insight into the activation mechanism of Cas7–11-Csx29 and substrate recognition of Csx30 we solved single particle cryo-electron microscopy (cryo-EM) structures of Csx30Δloop bound to Cas7–11-Csx29 with target RNA, and an inactive complex of Cas7–11-Csx29 alone, at 2.5-Å and 3.0-Å resolution respectively (Fig. 2, A–C, figs. S8 to S10, and table S1). The overall architecture of Cas7–11 in both complexes resembles the reported DiCas7–11 structure (16), in which the Cas7.1-Cas7.4 domains organize into a filament around the crRNA core with Cas11 at the midpoint. The insertion (INS) domain within Cas7.4 was visible only in the active state (Fig. 2, B and C). Csx29 consists of a three-helix bundle N-terminal domain (NTD), a TPR domain with eight repeats, and a protease region containing a pseudo-caspase (CHAT1) and active-caspase (CHAT2) domain that resembles separases (17, 18). In both complexes, Cas7.2-Cas7.4 interface with the NTD, TPR and CHAT1 domains of Csx29. Although the overall organization of Cas7–11 remains the same upon Csx29 binding, linker L2 and the Cas7.4 zinc-finger loop undergo structural changes which look similar in both active and inactive states (fig. S11). In the inactive state, the catalytic residues of CHAT2 are improperly positioned; C658 is turned downward away from the catalytic H615, and the catalytic histidine is instead positioned toward D661 (fig. S12). However, they are repositioned upon target RNA binding to resemble the geometry of active caspases (Fig. 2, D–F, fig. S12, and fig. S13). As CHAT2 makes no direct contact with Cas7–11 or target RNA, we hypothesized that conformational changes likely occur in other regions of Csx29 and transduce an allosteric signal to the catalytic core. By comparing the inactive and active complexes we observed a major structural change within the eighth repeat of the TPR domain, which we term the activation region (AR). The AR is bipartite, composed of AR1 (aa 313–325) and AR2 (aa 356–411), which stack with each other in the inactive state (Fig. 2C). In the active complex, AR1 senses the 3’ end of target RNA (position −4 and −5) through base stacking interactions and pushes the AR2 helices away, preventing a steric clash (Fig. 2C). The target RNA in our active complex is non-complementary to the direct repeat (DR) and the structure reveals that this is an important feature. In this state, the 3’ portion of the target RNA is separated from the crRNA, and it makes a sharp kink at position −2, enabling it to traverse the TPR domain of Csx29 and reach AR1 (fig. S14A). This observation suggests that a DR-matched RNA might not activate Csx29 as it could stay hybridized with the crRNA at position −2 and beyond. Supporting this model, a target RNA fully matching the DR strongly reduced Csx30 cleavage (fig. S14, B and C). Mismatches at position −1 and −2 alone were only able to partially activate Csx29, and mismatches at −1 to −4 were required to restore full Csx30 cleavage (fig. S14C). Eliminating base pairing between the DR and the target RNA is therefore crucial for CASP activation and highlights the importance of the AR1-target RNA interaction. Of note, non-complementarity between the DR and target RNA also plays an important role in type III-A and III-B CRISPR systems to suppress the response against host derived transcripts (19, 20), and thus is a generalized component of signal transduction in type III systems. In addition to target RNA sensing by Csx29 AR1, we identified contacts between Cas7–11 and target RNA at the DR-mismatched site. In addition to Y718 which base-stacks with the nucleotide at position −2, we identified K182, R375, and E717 contacting the nucleotide at position −1 (Fig. 2G, and fig. S13). To better understand CASP activation and the AR-induced signal transduction, we examined downstream allosteric events in Csx29. In the active complex, the kinked target RNA site at position −2 is stabilized by base stacking interactions, provided by both Cas7–11-Y718 and Csx29-Y398 within AR2. Adjacent residues at the tip of the AR2 helix, E390, N391, R394, and D395, initiate a network of electrostatic and hydrogen bonded contacts extending all the way to the CHAT2 active site (Fig. 2H, and fig. S13). Prominent salt bridges formed between R394-E672 and D395-R625 help position the loop containing the catalytic C658, and the strand containing the catalytic H615, respectively. Further down, the active site H615 is positioned by E617 contacts, whereas the active site C658 is kept in place by E659-Y478 and D661-R744. In the inactive state, these same residues positioning C658 in the active complex make entirely different contacts, E659 forms hydrogen bonds with S675 and S677, and D661 instead bonds with S660 (Fig. 2, D and H, and fig. S13). We note the similarity of this mechanism to eukaryotic caspases which are also thought to be regulated by the conformation of the L4 loop containing their catalytic cysteine (21). Together, these structures reveal an allosteric cascade initiated by the 3’ end of DR-mismatched target RNA, triggering the AR within the Csx29 TPR domain, and transducing structural changes to the Csx29 CHAT2 domain to coordinate active site residues. To test this model, we made mutations in the allosteric network. A Csx29-R394A/D395A double mutant within AR2 formed a stable Cas7–11-Csx29 complex, but Csx30 cleavage was significantly impaired (Fig. 2I, and fig. S14D). Further down the allosteric cascade, mutating Csx29-E659 and D661 in the vicinity of the catalytic C658 likely disrupted Csx29 folding and we were unable to purify a Cas7–11-Csx29 complex. Finally, we tested the importance of contacts between Cas7–11 and target RNA at the DR-mismatched site. Mutating Cas7–11-K182, E717, R375, and Y718 into alanines did not impair Cas7–11-Csx29 complex assembly, however, strongly reduced CASP activation upon target RNA binding (Fig. 2I, and fig. S14D). Thus, target RNA stabilization by Cas7–11 on the DR-mismatched end is also critical for protease activation. Csx30 recognition by Cas7–11-Csx29 In addition to revealing insight into CASP activation, our active complex also provides structural details regarding the interaction with Csx30. Despite using a full-length Csx30Δloop mutant for complex assembly, only a small portion (aa 407–560) is visible in our structure (Fig. 3A, fig. S15A), and the remaining residues must therefore be flexible with respect to Cas7–11-Csx29. This region mirrors the minimal substrate we identified via truncation experiments and confirms that recognition of Csx30 is mediated through its C-terminal domain. In our structure, Csx30 is bound only to the Csx29 CHAT2 domain and does not interact with Cas7–11. There is striking charge complementarity at the Csx29-Csx30 interface, and substrate recognition is likely electrostatically driven through the negatively charged surface of Csx29 and positively charged surface of Csx30 (fig. S15B). Detailed analysis of the interface reveals that Csx30 polar and positively charged residues (N482, S526, Q531, K551, and K553) make contact with the Csx29 CHAT2 domain (Fig. 3A, and fig. S16). In addition, Csx30-M527 is enclosed in a tight hydrophobic pocket lined with Csx29’s Y706, W720, and A723. The major determinant of Csx30 engagement is likely a cumulative effect of these interactions, as mutating individual regions of the Csx29-Csx30 interface did not significantly affect Csx30 cleavage (fig. S15C). Consistent with our ability to pulldown a Cas7–11-Csx29-Csx30Δloop complex in the presence and absence of target RNA (fig. S7), the interfacing residues of Csx29 adopt a similar organization in both the active and inactive complexes, and therefore we conclude that Csx30 binding is not allosterically regulated. We also examined the position of the Csx30 cleavage site within the active complex. One limitation of our structure is that the cleavage loop is mutated (and slightly shortened), and thus, we cannot observe substrate engagement in the active site in detail. As the loop is also flexible, it is not well resolved, but its density places it near the active site of Csx29 positioning it for cleavage (Fig. 3B). Csx30 binds and inhibits the transcription factor CASP-σ We next sought to explore the biological function of Csx30 and understand how cleavage might regulate its activity. As the Cas7–11 effector alone provides defense against phage (12), we reasoned that additional functions of DiCASP would similarly be involved in the immune response. One possibility is that processed Csx30 fragments, Csx30-N (residues 1–428) or Csx30-C (residues 429–565), promote cell death or an abortive infection response to prevent phage propagation. However, we did not observe defense against three tested phage (fig. S17A). Homology searches revealed a match of Csx30-C to a peptidoglycan N-acetylglucosamine deacetylase (HHpred probability: 92.85%, e-value: 0.56), but we did not detect modification of peptidoglycan or its components with cleaved Csx30 in vitro (fig. S17B). Overexpression of Csx30 fragments was not toxic in E. coli, and we only observed a slight growth defect in cells expressing full-length Csx30, which was temperature dependent and suppressed by the addition of Csx31 and CASP-σ (fig. S18). We next turned to the other proteins encoded in the locus to gain insight into Csx30 function. We predicted a strong binding interaction between the N-terminal domain of Csx30 and CASP-σ, which strongly resembles an extracytoplasmic function (ECF) sigma factor (3) (HHpred probability 100%, e-value 3.4e-31) (Fig. 4, A and B, and fig. S19). Sigma factors are transcription initiation proteins that bind DNA and recruit the RNA polymerase catalytic core to specific promoters (22), hinting that Csx30 might be involved in regulating a transcriptional response. Consistent with our computational prediction, purification of CASP-σ in the presence of Csx30 yielded a Csx30-CASP-σ complex, in which Csx30 could still be cleaved by Cas7–11-Csx29 (Fig. 4C). Csx30-N was sufficient for the interaction with CASP-σ, although at considerably lower yield (fig. S20). Although D. ishimotonii CASP-σ is unlikely to regulate its target genes heterologously in E. coli, we reasoned that the identification of putative CASP-σ binding sites might yield insight into its preferred sequence motif and function in the natural host. We performed ChIP-seq in E. coli with HA-tagged CASP-σ and identified 13 high confidence peaks compared to input and mock IP controls (Fig. 4D, and fig. S21A). Motif analysis of ChIP-seq peaks yielded a clear hit (Fig. 4E, and fig. S21B), which was similar to a de novo predicted motif (fig. S21C) (23). Sigma factors are frequently regulated by inhibitors (anti-sigma factors), and there are examples in bacteria in which a protease cleaves an anti-sigma factor to activate a transcriptional stress response including the anti-sigma factors RseA in E. coli (24) and RsiW in B. subtilis (25). In E. coli, the DegS protease senses cell envelope stress and cleaves a transmembrane segment of RseA (26), resulting in the eventual release of the sequestered sigma factor RpoE. Based on our structural model, we predict that the Csx30-CASP-σ interaction would block CASP-σ DNA binding based on steric clashes to sigma factor-bound DNA in experimental structures (27) (fig. S22). To test whether Csx30 inhibits CASP-σ, we repeated ChIP experiments in E. coli co-expressing Csx30 and found that CASP-σ DNA binding was blocked at all four tested loci (Fig. 4F). This inhibition was dependent on full-length Csx30 as both Csx30-N and Csx30-C fragments were unable to antagonize CASP-σ binding (Fig. 4F). Together our results suggest that Csx30 is an inhibitor of CASP-σ, and that processing by Cas7–11-Csx29 alleviates this inhibition. Csx30 processing regulates CASP-σ transcriptional activity We next sought to identify potential CASP-σ targets in the natural host D. ishimotonii. As many ECF sigma factors autoregulate their own expression (28), we first searched the DiCASP locus. We identified three strong matches in the promoters of cas1 and two genes of unknown function (Fig. 4G, and table S2), indicating that CASP-σ likely coordinates additional defense functions including CRISPR spacer acquisition. Genome-wide searches for motifs in D. ishimotonii promoter regions yielded several candidates although only one site, upstream of the nhaA gene, was below a q-value of 0.6 (table S3 and S4). To test these predictions, we constructed transcriptional reporters by placing putative CASP-σ promoters upstream of green fluorescent protein (GFP) and measured the resulting fluorescence in E. coli (Fig. 4H, and fig. S23, A and B). We observed GFP expression with both tested promoters compared to a random DNA control and found that fluorescence was fully dependent on CASP-σ expression (Fig. 4I). Consistent with our previous results, co-expression of full-length Csx30 was able to completely inhibit CASP-σ-mediated GFP expression whereas processed Csx30 fragments had no effect (Fig. 4I). Supporting a role in the immune response, we could computationally identify one of the two unknown ORFs, a predicted membrane protein, in other CRISPR and defense loci (fig. S23C). RNA sensing applications with DiCASP The high proteolytic activity of Cas7–11-Csx29 in response to a target RNA enables numerous biological applications. In addition, the ability to uncouple RNA cleavage from activation of the Csx29 protease allows for non-destructive sensing of RNA. While the collateral nuclease activity of CRISPR effectors has been used to cleave nucleic acid-based reporters for diagnostic applications (29), CASP systems allow for a new modality of substrates using engineered Csx30 proteins. As a proof of concept, we generated a fluorescently labeled engineered variant of Csx30 and demonstrated its ability to detect RNA in vitro down to 250 femtomolar without nucleic acid amplification (Fig. 5, A and B, and fig. S24). We also sought to apply DiCASP for RNA transcript sensing in live cells. To determine if DiCASP can mediate RNA-activated proteolytic cleavage in human cells, we transfected plasmids expressing Cas7–11, Csx29, crRNA, a synthetic target RNA, and Csx30 fused to an HA epitope tag into HEK293T cells. Immunoblots of cell lysate revealed processing of Csx30 that was dependent on a targeting crRNA and the catalytic residues of the Csx29 protease (Fig. 5C, and fig. S25, A and B). Testing DiCASP activity across a panel of endogenous transcripts revealed Csx30 cleavage efficiencies ranging from 2 to 20% (fig. S25, C and D). To convert RNA sensing with DiCASP into a discrete and readily detectable signal we sought to design reporters containing effector domains that could be activated by Csx30 cleavage. We transfected plasmids encoding a fusion protein in which Cre recombinase is tethered to membrane anchors (e.g. the cholinergic receptor, muscarinic 3 (Chrm3) GPCR) via a Csx30-derived linker, sequestering Cre from the nucleus (Fig. 5D). Mouse Neuro-2A cells harboring an inactive loxP-GFP reporter cassette were transfected with DiCASP components and synthetic target RNA. Flow cytometry analysis revealed crRNA-dependent GFP expression in 10% of cells, and a 15-fold increase over non-targeting crRNA controls under optimal conditions (Fig. 5E, and fig. S25, E and F). Discussion Here we demonstrate that the Csx29 protease associated with the type III-E effector Cas7–11 mediates RNA-activated endopeptidase activity and elucidate its substrate, structure, and mechanism. Although the full biological consequence of Csx30 processing in D. ishimotonii is unknown, our work supports a model in which Csx30 inhibits the sigma factor CASP-σ, and that proteolytic cleavage by the Csx29 protease acts to relieve this inhibition. The parallels between DiCASP and other protease-regulated anti-sigma factors, like DegS and RseA (26), reveal convergent mechanisms for modulating gene expression in response to cellular threats. The N-terminal domain of Csx30 is sufficient for binding to CASP-σ and it is therefore unclear how proteolytic cleavage within the Csx30 C-terminal domain would release CASP-σ, or why expression of Csx30-N is unable to inhibit CASP-σ. One possibility is that the processed Csx30 fragments are unstable and that the exposed termini are subject to further degradation by host proteins. Consistent with this hypothesis, immunoblots of E. coli cell lysates harboring HA-tagged isoforms of Csx30 revealed expression of full-length Csx30 and Csx30-C, but not Csx30-N, and that blocking the “cleaved” termini with an epitope tag increased expression (fig. S26). We note potential similarities to other protease-regulated anti-sigma factor systems; DegS cleavage of RseA is insufficient to release the sigma factor RpoE and the remaining RseA fragment is further processed by the RseP (30, 31) and ClpXP proteases (32) to liberate RpoE. Our identification of three CASP-σ binding motifs within the CASP locus points to the positive autoregulation of defense genes, including cas1, which may be a mechanism to acquire new spacers during active infection and to safeguard against the acquisition of self-targeting spacers during normal growth. This result is consistent with the reported upregulation of cas1 in Pseudomonas aeruginosa by the ECF sigma factor PvdS (33). The functions of the two other predicted upregulated genes in the locus are unknown, although one has strong homology to a membrane transporter component EcsC (HHpred probability 99.9, e-value 3.1e-22). Interestingly, the top motif match outside of the CASP locus is upstream of nhaA (table S3), a Na+/H+ antiporter known to be upregulated during phage infection (34), indicating that CASP-σ may also regulate targets elsewhere in the genome. Together, our results suggest the subtype III-E CASP systems use a three-pronged strategy to defend against foreign genetic material: (1) targeted RNA cleavage via the RNA endonuclease Cas7–11, (2) a Csx30-CASP-σ regulated transcriptional response that leads to, amongst other possibilities, spacer acquisition, and (3) a potential third arm mediated by Csx31 and possibly Csx30-C (Fig. 5F). The clear conservation of Csx31 (fig. S1) is a strong indication of its biological importance and future work will be required to determine its role in the immune response. We predict similar interactions between Csx30 and CASP-σ in other type III-E systems as well as putative CASP-σ binding motifs at cas1 within the Candidatus S. brodae locus (fig. S27). There may also be parallels between DiCASP and the type III CRISPR-associated Lon protease (11). We note that CRISPR-T is also associated with a neighboring sigma factor and is predicted to physically interact (fig. S28). We hypothesize that cleavage of CRISPR-T could similarly trigger transcriptional changes and may reflect a common functional theme across diverse CASP families. This work reveals an example of CRISPR systems coordinating a wider cellular response beyond nuclease activity, and we expect that the continued investigation of CRISPR-associated enzymes will uncover many interesting, and potentially useful, RNA-activated biological processes. Material and Methods Gene synthesis and cloning The TPR-CHAT protease and csx30, csx31, and CASP-σ genes from D. ishimotonii were codon optimized for human cell expression (GenScript) and synthesized and assembled from gene fragments. Additional materials were cloned by Gibson Assembly (New England Biolabs). pDF0159 (pCMV - huDisCas7–11, Addgene # 172507), pDF0118 (TwinStrp-SUMO-DisCas7–11, Addgene #172503), and pDF0114 (pU6-crRNA, Addgene #172508) were gifts from Omar Abudayyeh & Jonathan Gootenberg. In vitro RNA synthesis In vitro transcribed RNA was generated by annealing a DNA oligonucleotide containing the reverse complement of the desired RNA with a short T7 oligonucleotide. In vitro transcription reactions were performed using the HiScribe T7 High Yield RNA synthesis kit (NEB) at 37°C for 8–12 h and RNA was purified using Agencourt AMPure RNA Clean beads (Beckman Coulter). Cell-free transcription-translation 3xHA tagged forms of Csx30–3 were cloned into pCDNA3.1 vectors and amplified by PCR using oligos containing the T7 promoter and terminator. Cell-free transcription-translation was performed using PURExpress (New England Biolabs) in 5 μL reactions containing 2 μL buffer A, 1.5 μL buffer B, 0.25 μL of Superase RNAse Inhibitor (Invitrogen), and 50–100 ng of PCR template. Reactions were incubated for 2 h at 37°C and directly transferred to in vitro reactions. Protein purification All proteins were expressed in BL21 E. coli (Sigma Aldrich, CMC0016). Cells were grown in Terrific Broth (TB) to mid-log phase and the temperature was lowered to 18°C. Expression was induced at OD600 0.6 with 0.25 mM IPTG for 16–20 h before harvesting and freezing cells at −80°C. The gRAMP-CHAT complex was purified following co-expression of plasmids containing TwinStrep-SUMO-gRAMP and a mature crRNA, and pCDF-6xHis-CHAT. Cell paste was resuspended in lysis buffer (50 mM Tris pH 7.5, 250 mM NaCl, and 5% glycerol). Cells were lysed using a LM20 microfluidizer (Microfluidics) and cleared lysate was bound to Strep-Tactin Superflow Plus (Qiagen) using the gRAMP affinity tag. The resin was extensively washed and bound protein was eluted by cleaving the TwinStrep-SUMO tag with 10 μg Ulp1 SUMO protease overnight at 4°C. The eluted protein was bound to Ni-NTA Superflow (Qiagen) in 15 mM imidazole using the CHAT affinity tag, the resin was extensively washed with lysis buffer plus 40 mM imidazole, and the complex was eluted with 300 mM imidazole buffer. The eluted complex was diluted to 100 mM NaCl and purified on a HiTrap Heparin (Cytiva) column with a 100 mM to 1 M NaCl gradient. Fractions containing the gRAMP-CHAT complex were pooled, concentrated, and run on a Superose 6 Increase column (Cytiva) with a final storage buffer of 25 mM Tris pH 7.5, 250 mM NaCl, 10% glycerol, 1 mM DTT. All purified proteins were flash frozen in liquid nitrogen and stored at −80°C until use. Csx30 was purified using a TwinStrep-SUMO tag and lysis buffer containing 50 mM Tris pH 7.5, 250 mM NaCl, and 5% glycerol. Following Ulp1 SUMO protease digestion and elution from Strep-Tacin beads, Csx30 protein was diluted to 100 mM NaCl and purified using a Resource Q anion exchange column (Cytiva) with a 100 mM to 1 M NaCl gradient before gel filtration chromatography on a Superose 6 Increase column (Cytiva) with a final storage buffer of 25 mM Tris pH 7.5, 250 mM NaCl, 10% glycerol, 1 mM DTT. For pulldown experiments, Csx30 protein was eluted with 5 μM desthiobiotin instead of Ulp1 SUMO protease cleavage before ion exchange chromatography to retain the TwinStrep-SUMO tag. CASP-σ was purified using a pCDF-6xHis-Csx30 plasmid and Ni-NTA Superflow resin (Qiagen) in lysis buffer containing 50 mM Tris pH 7.5, 250 mM NaCl, 1 mM MgCl2, 5% glycerol and 15 mM imidazole. The resin was extensively washed with lysis buffer plus 40 mM imidazole, and CASP-σ eluted with 300 mM imidazole buffer. The Csx30-CASP-σ complex was purified in a similar way with the addition of a pUC19 plasmid containing untagged Csx30. The complex was purified using a Resource Q anion exchange column (Cytiva) following CASP-σ elution and moved to storage buffer (25 mM Tris pH 7.5, 250 mM NaCl, 10% glycerol, 1 mM DTT). Csx30 in vitro reactions Typical in vitro reactions were performed in 20 μL containing 4 μL of 5x reaction buffer (100 mM HEPES pH 7.5, 500 mM NaCl, 5 mM DTT, 25% glycerol), 0.5 μL of 150 mM MgCl2, 1 μL of Csx30 substrate (2.5 uM final concentration), 2 μL of gRAMP-CHAT-crRNA complex (25 nM final concentration), and 2 μL of purified target RNA (250 nM final concentration) unless otherwise noted. Reactions were incubated at 37°C for 1 hour before the addition of Laemmli buffer. Samples were boiled for 5 minutes and run on 12-well Nupage 4–12% Bis-Tris gels (Invitrogen) and stained with Coomassie dye before imaging on a Chemi-Doc (Bio-Rad). Biochemical experiments were typically performed with two independent replicates and a representative gel image shown. Mass spectrometry analysis Gel bands were excised from Coomassie stained SDS-PAGE gels following analysis of in vitro reactions and analyzed by the Whitehead Proteomics Core Facility using trypsin and chymotrypsin digests. CASP complex formation for cryo-EM Protein purification for the inactive CASP complex was performed as described above with the following modifications: (1) A pETDuet-1 derived plasmid containing His14-TwinStrep- bdSUMO-Cas7–11 with D429A/D654A mutations and a mature crRNA, and a pCDF-6xHis-Csx29 plasmid were used for co-expression; (2) bdSENP protease was used to cleave the His14-TwinStrep-bdSUMO tag from the Cas7–11-crRNA-Csx29 complex on Strep-Tactin resin; (3) after performing Heparin column purification, the complex was dialysed against a final storage buffer containing 20 mM Tris pH 8.0, 250 mM NaCl, 2.5% glycerol, concentrated, flash frozen in liquid nitrogen and stored at −80°C until use. For the active CASP complex, purification was carried out similarly, and Csx30Δloop retaining the TwinStrep-SUMO tag was purified separately. After Heparin column purification, the Cas7–11-crRNA-Csx29 complex was mixed with target RNA and TwinStrep-SUMO-Csx30Δloop in 1:10:10 molar ratio, in a buffer condition containing 20 mM Tris pH 8.0, 100 mM NaCl, 5% glycerol, and incubated at 37°C for 30 min. The mixture was then bound to Strep-Tactin resin, and the TwinStrep-SUMO tag was cleaved with SUMO protease Ulp1 to elute the Cas7–11-crRNA-target RNA-Csx29-Csx30 complex. The complex was run on a Superose 6 Increase column (Cytiva) with a final storage buffer of 20 mM Tris pH 7.5, 100 mM NaCl, 1% glycerol, concentrated, flash frozen in liquid nitrogen and stored at −80°C until use. Cryo-EM sample preparation For cryo-EM, the inactive CASP complex was diluted to 1 μM in a final buffer containing 20 mM Tris pH 7.5, 100 mM NaCl, 0.5% glycerol, and the active CASP complex was used at 1.6 μM in its final storage buffer. Quantifoil R1.2/1.3 300 mesh Cu holey carbon grids (Quantifoil, Germany) were glow-discharged (EMS 100, ElectronMicroscopy Sciences) at 25 mA for 1 min. 3 μl of each sample was applied to glow-discharged grids, blotted for 5 s using Standard Vitrobot Filter Paper (Ted Pella), and plunge-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific) at 4°C and 100% humidity. Cryo-EM data collection All data were collected at liquid nitrogen temperature on a Titan Krios G3i microscope (Thermo Scientific), equipped with a K3 direct detector (Gatan), operated at an accelerating voltage of 300 kV, and an energy filter with slit width of 20 eV. Movies were recorded in super-resolution mode with twofold binning at 130,000× magnification giving a physical pixel size of 0.6632 Å, with a 0.5–2.0 μm defocus range, at an electron exposure rate of 25.5 e−/pix/s for 0.69 s, fractionated into 30 frames, resulting in an accumulated fluence of 40 e−/Å2 per micrograph. 16,553 movies for the inactive complex, and 10,963 movies for the active complex were collected. Cryo-EM data processing All cryo-EM data were processed using RELION-4.0 (36), compiled and configured by SBGRid (37). Movies were corrected for motion using the RELION implementation of MotionCor2, with 5-by-5 patches and dose-weighting, and Contrast Transfer Function (CTF) parameters were estimated using CTFFIND-4.1 (38). For both datasets, particle picking was carried out using the Topaz general model (39). All reported resolutions use the gold-standard Fourier shell correlation with a cutoff of 0.143. For the inactive complex, 877,928 particles were extracted from 16,553 micrographs, and downscaled twofold. Analysis of these particles by 2D classification (100 classes, tau_fudge = 2, 220 Å mask diameter) revealed a mixture of dimers and monomers (fig. S7), and a monomeric reference model generated using RELION on a preliminary dataset collected on a Talos Arctica microscope was used for reconstruction. After cleaning poor quality particles by 3D classification (4 classes, tau_fudge = 4, 30 Å resolution reference, 25 iterations), remaining particles were subject to CTF refinement and Bayesian polishing, and one more round of 3D classification (4 classes, tau_fudge = 4, 15 Å resolution reference, 25 iterations, soft mask with 3 pixel hard edge, 8 pixel soft edge), and refinement, producing a reconstruction from 374,026 particles at 3.2-Å resolution. Since the peripheral regions of the complex, as well as Csx29 NTD, and the NTD-proximal parts within the TPR domain were flexible, focused refinement was performed to improve the EM density in those regions. A mask encompassing Csx29 NTD, as well as the well-ordered core region of Cas7–11, including crRNA was generated, and 3D classification without alignment (4 classes, tau_fudge = 100, 6 Å resolution reference, 30 iterations), showed that 71% of particles did not have strong density within this masked region. After removing these particles, the remaining particles were focus-refined by performing local angular searches starting at 0.9 degree sampling, first using the classification mask, and then using a mask encompassing the entirety of Cas7–11 and Csx29 NTD, producing a reconstruction at 3.0-Å resolution. Focused refinement efforts on the Cas7–11 INS domain were not successful. To improve the density for Csx29 TPR and CHAT, a mask encompassing only these two domains was produced, and 3D classification without alignment (4 classes, tau_fudge = 100, 6 Å resolution reference, 30 iterations), showed that 76% of particles did not have strong density within the masked region. After removing these particles, the remaining particles were focus-refined by performing local angular searches starting at 0.9 degree sampling, and using the classification mask, producing a reconstruction at 3.2-Å resolution. For the active complex, 2,143,080 particles were extracted from 10,963 micrographs, and downscaled twofold. Unlike the inactive complex, 2D classification analysis (200 classes, tau_fudge = 2, 220 Å mask diameter) revealed only monomers (fig. S8). After cleaning poor quality particles by 3D classification (4 classes, tau_fudge = 4, 30 Å resolution reference, 25 iterations), remaining particles were subject to CTF refinement and Bayesian polishing, and one more round of 3D classification (4 classes, tau_fudge = 100, 10 Å resolution reference, 30 iterations, soft mask with 3 pixel hard edge, 8 pixel soft edge), and refinement, producing a reconstruction from 187,426 particles at 2.4-Å resolution. Similar to the inactive complex, the peripheral regions of the overall refined active complex had weaker EM density compared to the core, and the density for the Cas7–11 INS domain, and Csx30 was mostly blurred, so focused refinement was performed to improve the map in those regions. A mask encompassing only the Cas7–11 INS domain was generated, and 3D classification without alignment (4 classes, tau_fudge = 200, 10 Å resolution reference, 30 iterations), showed that 65% of particles did not have strong density within this masked region. After removing these particles, the remaining particles were focus-refined by performing local angular searches starting at 0.5 degree sampling, using the classification mask, producing a reconstruction at 2.8-Å resolution. The same particles were further focus-refined afterwards, by performing local angular searches starting at 0.9 degree sampling, and using a mask encompassing the entirety of Cas7–11, producing a reconstruction at 2.5-Å resolution. To improve the density for Csx29 and Csx30, a mask encompassing only the Csx29 CHAT domain, and Csx30 was produced, and 3D classification without alignment (4 classes, tau_fudge = 100, 10 Å resolution reference, 30 iterations), showed that 65% of particles did not have strong density within the masked region. After removing these particles, the remaining particles were focus-refined by performing local angular searches starting at 0.5 degree sampling, using the classification mask, producing a reconstruction at 2.7-Å resolution. The same particles were further focus-refined afterwards, by performing local angular searches starting at 0.5 degree sampling, and using a mask encompassing the entirety of Csx29 and Csx30, producing a reconstruction at 2.6-Å resolution. Model building Initial protein models were generated using AlphaFold2 (40) and fit into the cryo-EM maps, and then manually edited using Coot (41), while RNA molecules were entirely de novo built in Coot. All models were further refined in ISOLDE (42). Coordinates were refined in real space using PHENIX (43), performing one macrocycle of global minimization and atomic displacement parameter (ADP) refinement and skipping local grid searches. Statistical validation for the final models was performed using PHENIX, RNA geometry was checked using the MolProbity server (44), and 3D-FSC sphericity values were calculated using 3D-FSC server (45). Phage plaque assays E. coli strains containing CASP expression plasmids were grown overnight at 37°C in LB with the appropriate antibiotic. 500 μL of each culture was diluted in 10 ml of molten top agar (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, 7 g/L agar) and poured onto LB plates containing the appropriate antibiotic. Phage were diluted ten-fold in phosphate-buffered saline (PBS) and spotted onto dried top agar plates. Plates were incubated overnight at 37°C and imaged in a dark room with a white backlight. Thin layer chromatography Uridine 5′-diphospho-N-acetylglucosamine (UDP-GlcNAc, Sigma Aldrich U4375), N-acetylemuramic acid (MurNAc, Sigma Aldrich A3007), and peptidoglycan from Bacillus subtilis (Sigma Aldrich, 69554) were resuspended in dimethyl sulfoxide at 10 mg/mL. Full-length or cleaved Csx30 protein was added and the reactions incubated at 37°C for 2 hours in the presence of 1 mM MgCl2, 1 mM ZnCl2, and 5 mM DTT. Oligosaccharides were separated by thin layer chromatography on silica gel 60 F254 LuxPlates (Millipore Sigma) in 30% propanol for 1 hour, and charred with 30% ammonium bisulfate at 150°C for 15 min for visualization. UDP-GlcNAc was visualized under 254 nm UV light. E. coli growth experiments Stbl3 (Thermo Fisher Scientific, C737303) and TOP10 cells (Thermo Fisher Scientific, C404010) were transformed with pUC19 and pBAD derived plasmids respectively. Cells were grown overnight in LB with the appropriate antibiotic to stationary phase. For liquid culture experiments, 3 μL was used to inoculate 150 μL cultures in clear 96-well plates. Plates were sealed with clear optical film and two holes were punched for aeration using a 28 gauge needle. Plates were incubated in a Synergy Neo2 plate reader (BioTek) at the indicated temperature with constant orbital shaking and the optical density at 600 nm read every 5 minutes. Plate-based growth assays were performed by normalizing the input density of overnight cultures and performing 10-fold dilutions. 5 μL of each dilution was dropped onto agar plates and grown at the indicated temperature for 16 hours. Plates were imaged using a Chemi-Doc (Bio-Rad). Csx30 labeling and in vitro diagnostics To prevent labeling of Csx30-N amine side chains, we mutated eight lysine residues to arginine, and four lysines within the cleavage loop to alanine. Mutated and truncated Csx30 was purified as previously described except with HEPES buffer in all steps instead of Tris. Csx30 was biotinylated in vitro using the BirA biotin ligase (Avidity). Csx30 was incubated with NHS-Fluorescein (Thermo Fisher Scientific, #46409) on ice for 1 h before quenching with 200 mM Tris pH 7.5. Labeled Csx30 was purified using a Resource Q anion exchange column as before. Purified biotin-Csx30-FAM substrate was bound to MyOne Streptavidin T1 dynabeads (Thermo Fisher Scientific) in phosphate buffered saline (PBS) for 30 min at room temperature. The beads were washed 10 times with PBS supplemented with 0.1% bovine serum albumin and resuspended in PBS. In vitro reactions were performed as before and Dyneabeads were removed from the reaction using a magnetic stand. The supernatant, containing cleaved Csx30C, was transferred to 96-well plates and fluorescence measured using a Synergy Neo2 plate reader (BioTek) and subtracting the background signal from a well with no target RNA. ChIP-seq library preparation BL21 cells (Sigma Aldrich, CMC0016) expressing HA-CASP-σ were grown in 25 mL cultures in LB to mid-log phase and induced with 0.25 mM IPTG for 3 h at 37°C. Formaldehyde was added (1% final concentration) and cells incubated for 5 min before quenching with 275 mM glycine pH at 4°C for 20 min. Cells were washed in ice-cold Tris buffer saline and stored at −80°C until processing. Pellets were resuspended in 500 μL lysis buffer (10 mM Tris pH 8.0, 20% sucrose, 50 mM NaCl, 10 mM EDTA, 10 mg/mL lysozyme) and sonicated with a microtip probe (QSonica) to shear DNA. Lysates were spun for 15 min at 4°C at 21,000 g and 2 mL of immunoprecipitation buffer was added (50 mM HEPES pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% Sodium deoxycholate) with a sample taken as an input control. HA-CASP-σ immunoprecipitation was performed by adding 50 μL of washed Pierce Anti-HA Magnetic Beads (Thermo Fisher Scientific) and incubating at 4°C for 4 hours. Beads were washed 3 times with immunoprecipitation buffer, 3 times with wash buffer (10 mM Tris pH 8, 250 mM LiCl, 1 mM EDTA, 0.5% NP-40, 0.5% Sodium deoxycholate), and 2 times with TE (10 mM Tris pH 8, 1 mM EDTA). DNA was eluted with 100 μL TE supplemented with 1% SDS and a 65°C incubation for 10 min. 340 μL of TE with 40 μg RNAse A was added and samples incubated at 37°C for 2 hours. Formaldehyde cross-links were reversed by overnight incubation at 65°C and DNA was purified using Qiagen PCR Purification columns. DNA was sequenced using the NEBNext Ultra II DNA Library Prep Kit for Illumina (New England Biolabs) and an Illumina MiSeq. ChIP-seq analysis Reads were mapped as .fastq files to E. coli K12 MG1655 (NC_000913.3) using http://browsergenome.org (46) with mapping parameters: no read filter, forward mapping start = 0 bp, forward mapping length = 25 bop, reverse mapping length = 15 bp, max forward/reverse span = 1000 bp, discard ambiguous hits. Mapped reads were exported as .SAM files and imported into Geneious (v2022.1.1) where coverage tables were extracted. Reads mapping to LacI (NC_000913.3:366000–368000) were filtered out due to the presence of the LacI on a plasmid used for ChIP. Remaining reads were normalized to the median per base coverage as there is a long right tail in the reads per base distribution. Putative peaks were identified as regions where the normalized coverage was greater than 4 in the CASP-σ IP samples and less than 3 in the control IP samples using Python. Peaks were then visually examined to ensure that their shape matched the expected triangular structure of a localized ChIP-seq peak. The 60 bps centered at the max coverage position of the 13 remaining peaks were aggregated and fed into MEME (https://meme-suite.org/meme/tools/meme, version 5.4.1) (47), producing a single strong hit based on 12 of the 13 loci. A putative binding site was identified manually in the remaining sequence (NC_000913.3:3880776–3880799) and logos were generated from all 13 loci using LogoMaker (48) in a Jupyter Notebook. Scripts for analysis and generating figures and tables can be found in the Zenodo repository. ChIP-qPCR BL21 cells (Sigma Aldrich, CMC0016) co-transformed with plasmids expressing HA-CASP-σ and Csx30 isoforms were grown, formaldehyde fixed, and frozen as previously described for ChIP-seq analysis. Cell pellets were resuspended in 500 μL lysis buffer and sonicated with a Bioruptor sonication device (Diagenode) at 4°C with 30s on/off cycles at high intensity for 15 min. Three independent immunoprecipitations were performed for each sample as previously described and eluted DNA was purified using Qiagen PCR Purification columns. DNA quantification performed with custom primers and hydrolysis probes containing 5’ 6-FAM labels and ZEN (internal) and Iowa Black (3’) fluorescent quenchers (Integrated DNA Technologies) (table S4). qPCR was performed with two technical replicates for each sample and run on a LightCycler 480 (Roche) using TaqMan Universal PCR Master Mix (Thermo Fisher Scientific). Fold enrichment at four separate loci was determined using the delta-delta CT method by normalizing to a dinG control sequence (where CASP-σ does not bind) and to input DNA. De novo CASP-σ motif prediction CASP-σ from the Csx30-CASP-σ structure predicted from Colabfold was structurally aligned in PyMol (Schrödinger) separately to the σ2 and σ4 domains of E. coli RpoE (PDB code: 1OR7) (49). Using the E. coli structure as a guide, sequence alignments to other ECF sigma factors were generated and used as an input for binding motifs prediction using predictECF (https://github.com/horiatodor/predictECF) (23) in R. Scripts for analysis and generating figures can be found in the Zenodo repository. CASP-σ motif scanning Motifs for scanning the DiCASP loci (NZ_BEXT01000001:1,366,660–1,387,005), promoters from the D. ishimotonii genome, and the full D. ishimotonii genome (NZ_BEXT01000001) for putative CASP-σ binding sites were based on the position probability matrix created from the 13 peaks from ChIP-seq. Promoters were extracted by taking the 100 bps upstream of each annotated CDS in a Jupyter Notebook. Positions with Rseq ≤ 1 were masked and replaced with the average background nucleotide frequencies of each query sequence to avoid spurious sequence preferences in the motif due to potential undersampling of ChIP-seq hits (50, 51). Query sequences and motifs were analyzed using FIMO (https://meme-suite.org/meme/tools/fimo, version 5.4.1) (52). Scripts for analysis and generating tables as well as the query motifs in simple MEME format and the query sequences in .fasta format can be found in the Zenodo repository. Bacterial transcriptional reporters Fluorescent transcriptional reporters were constructed by placing putative CASP-σ promoters upstream of msGFP in low copy pACYC plasmids. BL21 cells (Sigma Aldrich, CMC0016) were co-transformed with reporters and plasmids expressing CASP-σ, Csx30 isoforms, or empty controls and grown overnight in Terrific Broth. Cultures were diluted 1:10 in fresh media and GFP fluorescence measured in a Synergy Neo2 plate reader (BioTek, 488/528nm filter). The optical density at 600 nm was also read for each well and GFP levels normalized to cell density. Experiments were performed with 3 independent cultures for each condition. Structural predictions and homology searches Csx30 and Csx30-CASP-σ structures were predicted using Colabfold (53), an interface for Alphafold2 (40) and MMSeqs2 (UniRef + environmental). Protein homology was determined using HHpred (54). Cell culture and transfection HEK293T and Neuro2A cells were cultured in Dulbecco’s modified Eagle medium with high glucose, sodium pyruvate, and GlutaMAX (Thermo Fisher Scientific), 1× penicillin–streptomycin (Thermo Fisher Scientific), and 10% fetal bovine serum (Seradigm). Cells were maintained at a confluency below 90%. For immunoblot analysis, 24-well plates were seeded with 87,500 cells/well approximately 16 h before transfection. Cell were typically transfected with 50 ng of 3xHA-Csx30, 400 ng gRAMP, 400 ng CHAT, 100 ng target, and 500 ng crRNA in Opti-MEM (Thermo Fisher Scientific) with 4.5 μL TransIt-LT1 transfection reagent (Mirus). Spacer sequences for transcripts are listed in table S5. For flow cytometry experiments, 96-well plates were seeded with 17,500 cells/well. Cell were typically transfected with 60 ng gRAMP, 60 ng CHAT, 20 ng target, 60 ng crRNA, and 0.5–5 ng of Cre constructs in Opti-MEM (Thermo Fisher Scientific) with 0.6 μL TransIt-LT1 transfection reagent (Mirus). Western blot and flow cytometry Cells were typically harvested 96 h post-transfection. Cells were washed with ice-cold PBS and lysed in 75 μL of NP-40 lysis buffer (50 mM Tris pH 8, 150 mM NaCl, 1% NP-40). Cell suspensions were kept on ice for 10 min and cleared by centrifugation at 4C for 10 min at 21,000g. Lysates were stored at −80 before western blot analysis. Lysates were mixed with 4x Lammli buffer (Bio-Rad) run on 12 -well Nupage 4–12% Bis-Tris gels (Invitrogen). Proteins were transferred to PDVF membranes using an iBlot2 at 23V for 6 min. Membranes were blocked for 30 min at room temperature with TBST (Tris-buffer saline with 0.1% Tween 20) with 5% bovine serum albumin (Rockland). anti-HA:HRP (Cell Signaling Technologies, #2999) and anti-GAPDH:HRP (Cell Signaling Technologies #3683) were added at 1:5000 dilution and incubated for 30–60 min at room temperature. Membranes were washed 5x with TBST, incubated with Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific) and imaged using a Chemi-Doc (Bio-Rad). Immunoblots of E. coli cell lysates were performed in a similar manner. Cell input was normalized using optical density at 600 nm, and cell pellets were resuspended and lysed directly in Laemmli buffer. Csx30 cleavage efficiency in immunoblots was estimated using image analysis in FIJI (55). The average signal intensity of each band was determined using a constant area selection and the lane background subtracted. Csx30 cleavage for each guide was determined as Csx30cleaved/(Csx30cleaved +Csx30full-length in three independent experiments. Expression levels of endogenous transcripts were determined from available HEK293T RNA-seq data (NCBI GEO database (56), accession GSE204833). For flow cytometry analysis, cells were trypsinized 96 h post-transfection and resuspended in PBS supplemented with 5% FBS. Cells were analyzed using a CytoFLEX S flow cytometer (Beckman Coulter). Supplementary Material Tables S1 - S7 Figures S1 - S28 Acknowledgments We thank Linyi Gao, Jonathan Schmid-Burgk, and Daniel Strebinger for valuable discussions; Blake Lash for providing Neuro2A cells; Eric Spooner and the Whitehead Institute Proteomics Core Facility for mass spectrometry analysis; Edward Brignole and Christopher Borsa at the MIT.nano CryoEM Center, staff at the Arnold and Mabel Beckman Foundation for funding the MIT.nano CryoEM Center, and the entire Zhang lab for support and advice. Funding: F.Z. is supported by NIH grants (1DP1-HL141201 and 2R01HG009761–05); the Howard Hughes Medical Institute; the Yang-Tan Molecular Therapeutics Center at McGovern, and by the Phillips family and J. and P. Poitras. J.S. is a Charles A. King Trust Postdoctoral Research Fellow. Data and materials availability: ChIP-seq data is available at the NCBI Sequence Read Archive (BioProject ID: PRJNA888197, SAMN31210314–21). Code and processed data is available at Zenodo (35). Plasmids are available from Addgene. The cryo-EM maps have been deposited in the Electron Microscopy Data Bank with the following codes: EMD-28064 (inactive CASP, focused refinement of Cas7–11, crRNA, and Csx29 NTD), EMD-28070 (inactive CASP, focused refinement of Csx29 TPR-CHAT), EMD-28065 (active CASP, focused refinement of Cas7–11, except INS, crRNA, and target RNA), EMD-28071 (active CASP, focused refinement of Cas7–11 INS), EMD-28072 (active CASP, focused refinement of Csx29 NTD and TPR), and EMD-28073 (active CASP, focused refinement of Csx29 CHAT and Csx30). The coordinates for the composite atomic models have been deposited in the Protein Data Bank under accession codes 8EEX (inactive CASP) and 8EEY (active CASP). Fig. 1. The type III-E CRISPR-associated protease Csx29 cleaves Csx30. (A) Schematic of selected CRISPR-associated protease (CASP) loci and three additional conserved genes in type III-E loci. (B) Immunoblot analysis of in vitro reactions with Cas7–11-Csx29 and HA-tagged Csx30, Csx31, and CASP-σ produced by cell-free transcription-translation. (C) A Cas7–11-Csx29-crRNA complex cleaves Csx30 protein in response to target RNA. (D) Csx30 cleavage requires target RNA and the Csx29 protease catalytic residues, but not the catalytic residues of Cas7–11. (E) Schematic of Csx30 highlighting the cleavage site (aa 427–429), linker (aa 377–406), and a potential effector domain annotated from HHpred (aa 452–545). (F) AlphaFold2 prediction of Csx30 (G) Analysis of dCas7–11-Csx29 proteolytic activity on truncated Csx30 proteins. (H) Immunoblot analysis of HA-tagged Csx30 mutants produced by cell free transcription-translation. Panels C, D, and G are SDS-PAGE gels stained with Coomassie. Fig. 2. Allosteric activation of Csx29 upon RNA binding. (A) Schematic of Cas7–11, Csx29, and Csx30 proteins domains, and the crRNA and target RNA used in structural studies. (B) Structures of the inactive (Cas7–11-Csx29-crRNA) and active (Cas7–11-Csx29-crRNA-target RNA-Csx30) CASP complexes. (C) Structural organization of the Csx29 AR in inactive and active CASP complexes. (D) Electrostatic and hydrogen bonded network within the Csx29 catalytic site in the inactive state. (E and F) Catalytic H615 and C658 residues in inactive and active Csx29 shown with EM density. (G) Contacts between Cas7–11 and the DR-mismatched portion of the target RNA in the active state. (H) Electrostatic and hydrogen bonded network extending from the AR to the Csx29 catalytic site in the active state. (I) Mutations disrupting allosteric activation residues impair Csx30 cleavage by Cas7–11-Csx29. SDS-PAGE gel stained with Coomassie. Fig. 3. Csx30 substrate recognition by Csx29. (A) Csx29-Csx30 interface in the active CASP structure. Electrostatic interactions and hydrogen bonds are drawn as dashed lines, and the hydrophobic pocket as a dashed oval. (B) Close-up of the Csx29-Csx30 interface near the catalytic H615 and C658 residues. Fig. 4. Csx30 binds and inhibits the transcription factor CASP-σ. (A) Schematic of Csx30 and CASP-σ proteins. (B) AlphaFold2 prediction of a Csx30-CASP-σ interaction. (C) Purification of a Csx30-CASP-σ complex that is cleaved by dCas7–11-Csx29. SDS-PAGE gel stained with Coomassie. (D) Representative CASP-σ ChIP-seq peaks in E. coli with a 1 kb window, input coverage shown in gray. (E) Identification of a CASP-σ binding motif from ChIP-seq peaks. (F) Enrichment of CASP-σ at four E. coli peaks by ChIP-qPCR. n = 3 replicates. (G) Predicted CASP-σ targets in the D. ishimotonii CASP locus. (H) Schematic of a CASP-σ transcriptional reporter assay. (I) CASP-σ-mediated transcriptional activity in E. coli. GFP expression was normalized to cells with a scrambled promoter sequence. n = 3 replicates. ** denotes p < 0.01, Student’s t-test. Error bars represent standard deviation from the mean in all panels. Fig. 5. RNA sensing applications and a proposed model for CASP systems. (A) Schematic of in vitro RNA detection using CASP systems and fluorescent Csx30 reporters. (B) In vitro detection of RNA as measured by released fluorescence. n = 3 replicates. (C) Immunoblot analysis of HA-tagged Csx30 in HEK293T human cells transfected with DiCASP components. (D) Schematic of engineered proteins containing a cell membrane tether, a Csx30 linker, and an effector domain. (E) Flow cytometry of DiCASP activity in mouse Neuro2A loxP:GFP cells using a Chrm3-Csx30250–565−Cre reporter. n = 3–6 replicates. (F) Model for a three-pronged strategy of CASP systems in the defense against foreign genetic elements including Cas7–11 mediated RNA endonuclease activity, a Csx30-regulated CASP-σ transcriptional response, and a possible third arm involving Csx31. Error bars represent standard deviation from the mean in all panels. Competing interests: F.Z. is a scientific advisor for Editas Medicine, Beam Therapeutics, Pairwise Plants, Arbor Biotechnologies, Proof Diagnostics, and Octant. J.S. and F.Z. are inventors on US patent application (63/337,386) submitted by The Broad Institute that covers programmable protease compositions. References 1. Bernheim A , Sorek R , The pan-immune system of bacteria: antiviral defence as a community resource. Nat. Rev. Microbiol. 18 , 113–119 (2020).31695182 2. Gao L , Altae-Tran H , Böhning F , Makarova KS , Segel M , Schmid-Burgk JL , Koob J , Wolf YI , Koonin EV , Zhang F , Diverse enzymatic activities mediate antiviral immunity in prokaryotes. Science. 369 , 1077–1084 (2020).32855333 3. Makarova KS , Wolf YI , Iranzo J , Shmakov SA , Alkhnbashi OS , Brouns SJJ , Charpentier E , Cheng D , Haft DH , Horvath P , Moineau S , Mojica FJM , Scott D , Shah SA , Siksnys V , Terns MP , Venclovas Č , White MF , Yakunin AF , Yan W , Zhang F , Garrett RA , Backofen R , van der Oost J , Barrangou R , Koonin EV , Evolutionary classification of CRISPR-Cas systems: a burst of class 2 and derived variants. Nat. Rev. Microbiol. 18 , 67–83 (2020).31857715 4. Shmakov SA , Makarova KS , Wolf YI , Severinov KV , Koonin EV , Systematic prediction of genes functionally linked to CRISPR-Cas systems by gene neighborhood analysis. Proc. Natl. Acad. Sci. U. S. A. 115 , E5307–E5316 (2018).29784811 5. Shah SA , Alkhnbashi OS , Behler J , Han W , She Q , Hess WR , Garrett RA , Backofen R , Comprehensive search for accessory proteins encoded with archaeal and bacterial type III CRISPR-cas gene cassettes reveals 39 new cas gene families. RNA Biol. 16 , 530–542 (2019).29911924 6. Peters JE , Makarova KS , Shmakov S , Koonin EV , Recruitment of CRISPR-Cas systems by Tn7-like transposons. Proc. Natl. Acad. Sci. U. S. A. 114 , E7358–E7366 (2017).28811374 7. Faure G , Shmakov SA , Yan WX , Cheng DR , Scott DA , Peters JE , Makarova KS , Koonin EV , CRISPR-Cas in mobile genetic elements: counter-defence and beyond. Nat. Rev. Microbiol. 17 , 513–525 (2019).31165781 8. Strecker J , Ladha A , Gardner Z , Schmid-Burgk JL , Makarova KS , Koonin EV , Zhang F , RNA-guided DNA insertion with CRISPR-associated transposases. Science. 365 , 48–53 (2019).31171706 9. Klompe SE , Vo PLH , Halpin-Healy TS , Sternberg SH , Transposon-encoded CRISPR-Cas systems direct RNA-guided DNA integration. Nature. 571 , 219–225 (2019).31189177 10. Koonin EV , Makarova KS , Evolutionary plasticity and functional versatility of CRISPR systems. PLoS Biol. 20 , e3001481 (2022).34986140 11. Rouillon C , Schneberger N , Chi H , Peter MF , Geyer M , Boenigk W , Seifert R , White MF , Hagelueken G , SAVED by a toxin: Structure and function of the CRISPR Lon protease. bioRxiv. (2021), p. 2021.12.06.471393. 12. Özcan A , Krajeski R , Ioannidi E , Lee B , Gardner A , Makarova KS , Koonin EV , Abudayyeh OO , Gootenberg JS , Programmable RNA targeting with the single-protein CRISPR effector Cas7–11. Nature. 597 , 720–725 (2021).34489594 13. van Beljouw SPB , Haagsma AC , Rodríguez-Molina A , van den Berg DF , Vink JNA , Brouns SJJ , The gRAMP CRISPR-Cas effector is an RNA endonuclease complexed with a caspase-like peptidase. Science. 373 , 1349–1353 (2021).34446442 14. van der Oost J , van der Oost J , Westra ER , Jackson RN , Wiedenheft B , Unravelling the structural and mechanistic basis of CRISPR–Cas systems. Nature Reviews Microbiology. 12 (2014), pp. 479–492.24909109 15. Aravind L , Koonin EV , Classification of the caspase-hemoglobinase fold: detection of new families and implications for the origin of the eukaryotic separins. Proteins. 46 , 355–367 (2002).11835511 16. Kato K , Zhou W , Okazaki S , Isayama Y , Nishizawa T , Gootenberg JS , Abudayyeh OO , Nishimasu H , Structure and engineering of the type III-E CRISPR-Cas7–11 effector complex. Cell (2022), doi:10.1016/j.cell.2022.05.003. 17. Boland A , Martin TG , Zhang Z , Yang J , Bai X-C , Chang L , Scheres SHW , Barford D , Cryo-EM structure of a metazoan separase–securin complex at near-atomic resolution. Nature Structural & Molecular Biology. 24 (2017), pp. 414–418. 18. Lin Z , Luo X , Yu H , Structural basis of cohesin cleavage by separase. Nature. 532 , 131–134 (2016).27027290 19. You L , Ma J , Wang J , Artamonova D , Wang M , Liu L , Xiang H , Severinov K , Zhang X , Wang Y , Structure Studies of the CRISPR-Csm Complex Reveal Mechanism of Co-transcriptional Interference. Cell. 176 , 239–253.e16 (2019).30503210 20. Sofos N , Feng M , Stella S , Pape T , Fuglsang A , Lin J , Huang Q , Li Y , She Q , Montoya G , Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism of Type III-B CRISPR-Cas. Mol. Cell. 79 , 741–757.e7 (2020).32730741 21. McLuskey K , Mottram JC , Comparative structural analysis of the caspase family with other clan CD cysteine peptidases. Biochem. J. 466 , 219–232 (2015).25697094 22. Feklístov A , Sharon BD , Darst SA , Gross CA , Bacterial sigma factors: a historical, structural, and genomic perspective. Annu. Rev. Microbiol. 68 , 357–376 (2014).25002089 23. Todor H , Osadnik H , Campbell EA , Myers KS , Li H , Donohue TJ , Gross CA , Rewiring the specificity of extracytoplasmic function sigma factors. Proc. Natl. Acad. Sci. U. S. A. 117 , 33496–33506 (2020).33318184 24. OMP Peptide Signals Initiate the Envelope-Stress Response by Activating DegS Protease via Relief of Inhibition Mediated by Its PDZ Domain. Cell. 113 , 61–71 (2003).12679035 25. Schöbel S , Zellmeier S , Schumann W , Wiegert T , The Bacillus subtilis sigmaW anti-sigma factor RsiW is degraded by intramembrane proteolysis through YluC. Mol. Microbiol. 52 , 1091–1105 (2004).15130127 26. Ades SE , Connolly LE , Alba BM , Gross CA , The Escherichia coli sigma(E)-dependent extracytoplasmic stress response is controlled by the regulated proteolysis of an anti-sigma factor. Genes Dev. 13 , 2449–2461 (1999).10500101 27. Lane WJ , Darst SA , The Structural Basis for Promoter −35 Element Recognition by the Group IV σ Factors. PLoS Biology. 4 (2006), p. e269.16903784 28. Casas-Pastor D , Müller RR , Jaenicke S , Brinkrolf K , Becker A , Buttner MJ , Gross CA , Mascher T , Goesmann A , Fritz G , Expansion and re-classification of the extracytoplasmic function (ECF) σ factor family. Nucleic Acids Res. 49 , 986–1005 (2021).33398323 29. Gootenberg JS , Abudayyeh OO , Lee JW , Essletzbichler P , Dy AJ , Joung J , Verdine V , Donghia N , Daringer NM , Freije CA , Myhrvold C , Bhattacharyya RP , Livny J , Regev A , Koonin EV , Hung DT , Sabeti PC , Collins JJ , Zhang F , Nucleic acid detection with CRISPR-Cas13a/C2c2. Science. 356 , 438–442 (2017).28408723 30. Alba BM , Leeds JA , Onufryk C , Lu CZ , Gross CA , DegS and YaeL participate sequentially in the cleavage of RseA to activate the ςE-dependent extracytoplasmic stress response. Genes & Development. 16 (2002), pp. 2156–2168.12183369 31. Kanehara K , Ito K , Akiyama Y , YaeL (EcfE) activates the ςE pathway of stress response through a site-2 cleavage of anti-ςE, RseA. Genes & Development. 16 (2002), pp. 2147–2155.12183368 32. Flynn JM , Levchenko I , Sauer RT , Baker TA , Modulating substrate choice: the SspB adaptor delivers a regulator of the extracytoplasmic-stress response to the AAA+ protease ClpXP for degradation. Genes Dev. 18 , 2292–2301 (2004).15371343 33. Ahator SD , Jianhe W , Zhang L-H , The ECF sigma factor PvdS regulates the type I-F CRISPR-Cas system in Pseudomonas aeruginosa. bioRxiv (2020), p. 2020.01.31.929752. 34. Malone LM , Hampton HG , Morgan XC , Fineran PC , Type I CRISPR-Cas provides robust immunity but incomplete attenuation of phage-induced cellular stress. Nucleic Acids Res. 50 , 160–174 (2022).34928385 35. Strecker J , Li D , Zhang F Code and processed data for: RNA-activated protein cleavage with a CRISPR-associated endopeptidase (Version 1.0). Zenodo 10.5281/zenodo.7221526. 36. Kimanius D , Dong L , Sharov G , Nakane T , Scheres SHW , New tools for automated cryo-EM single-particle analysis in RELION-4.0. Biochem J. 478 , 4169–4185 (2021).34783343 37. Morin A , Eisenbraun B , Key J , Sanschagrin PC , Timony MA , Ottaviano M , Sliz P , Collaboration gets the most out of software. Elife. 2 , e01456 (2013).24040512 38. Rohou A , Grigorieff N , CTFFIND4: Fast and accurate defocus estimation from electron micrographs. J. Struct. Biol. 192 , 216–221 (2015).26278980 39. Bepler T , Morin A , Rapp M , Brasch J , Shapiro L , Noble AJ , Berger B , Positive-unlabeled convolutional neural networks for particle picking in cryo-electron micrographs. Nat. Methods. 16 , 1153–1160 (2019).31591578 40. Jumper J , Evans R , Pritzel A , Green T , Figurnov M , Ronneberger O , Tunyasuvunakool K , Bates R , Žídek A , Potapenko A , Bridgland A , Meyer C , Kohl SAA , Ballard AJ , Cowie A , Romera-Paredes B , Nikolov S , Jain R , Adler J , Back T , Petersen S , Reiman D , Clancy E , Zielinski M , Steinegger M , Pacholska M , Berghammer T , Bodenstein S , Silver D , Vinyals O , Senior AW , Kavukcuoglu K , Kohli P , Hassabis D , Highly accurate protein structure prediction with AlphaFold. Nature. 596 , 583–589 (2021).34265844 41. Casañal A , Lohkamp B , Emsley P , Current developments in Coot for macromolecular model building of Electron Cryo-microscopy and Crystallographic Data. Protein Sci. 29 , 1069–1078 (2020).31730249 42. Croll TI , ISOLDE: a physically realistic environment for model building into low-resolution electron-density maps. Acta Crystallogr D Struct Biol. 74 , 519–530 (2018).29872003 43. Liebschner D , Afonine PV , Baker ML , Bunkóczi G , Chen VB , Croll TI , Hintze B , Hung LW , Jain S , McCoy AJ , Moriarty NW , Oeffner RD , Poon BK , Prisant MG , Read RJ , Richardson JS , Richardson DC , Sammito MD , Sobolev OV , Stockwell DH , Terwilliger TC , Urzhumtsev AG , Videau LL , Williams CJ , Adams PD , Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix. Acta Crystallogr D Struct Biol. 75 , 861–877 (2019).31588918 44. Williams CJ , Headd JJ , Moriarty NW , Prisant MG , Videau LL , Deis LN , Verma V , Keedy DA , Hintze BJ , Chen VB , Jain S , Lewis SM , Arendall WB 3rd , Snoeyink J , Adams PD , Lovell SC , Richardson JS , Richardson DC , MolProbity: More and better reference data for improved all-atom structure validation. Protein Sci. 27 , 293–315 (2018).29067766 45. Tan YZ , Baldwin PR , Davis JH , Williamson JR , Potter CS , Carragher B , Lyumkis D , Addressing preferred specimen orientation in single-particle cryo-EM through tilting. Nat. Methods. 14 , 793–796 (2017).28671674 46. Schmid-Burgk JL , Hornung V , BrowserGenome.org: web-based RNA-seq data analysis and visualization. Nat. Methods. 12 , 1001 (2015).26513548 47. Bailey TL , Johnson J , Grant CE , Noble WS , The MEME Suite. Nucleic Acids Res. 43 , W39–49 (2015).25953851 48. Tareen A , Kinney JB , Logomaker: beautiful sequence logos in Python. Bioinformatics. 36 , 2272–2274 (2020).31821414 49. Campbell EA , Tupy JL , Gruber TM , Wang S , Sharp MM , Gross CA , Darst SA , Crystal structure of Escherichia coli sigmaE with the cytoplasmic domain of its anti-sigma RseA. Mol. Cell. 11 , 1067–1078 (2003).12718891 50. Crooks GE , Hon G , Chandonia J-M , Brenner SE , WebLogo: a sequence logo generator. Genome Res. 14 , 1188–1190 (2004).15173120 51. Schneider TD , Stephens RM , Sequence logos: a new way to display consensus sequences. Nucleic Acids Res. 18 , 6097–6100 (1990).2172928 52. Grant CE , Bailey TL , Noble WS , FIMO: scanning for occurrences of a given motif. Bioinformatics. 27 , 1017–1018 (2011).21330290 53. Mirdita M , Schütze K , Moriwaki Y , Heo L , Ovchinnikov S , Steinegger M , ColabFold: making protein folding accessible to all. Nat. Methods. 19 , 679–682 (2022).35637307 54. Zimmermann L , Stephens A , Nam S-Z , Rau D , Kübler J , Lozajic M , Gabler F , Söding J , Lupas AN , Alva V , A Completely Reimplemented MPI Bioinformatics Toolkit with a New HHpred Server at its Core. J. Mol. Biol. 430 , 2237–2243 (2018).29258817 55. Schindelin J , Arganda-Carreras I , Frise E , Kaynig V , Longair M , Pietzsch T , Preibisch S , Rueden C , Saalfeld S , Schmid B , Tinevez J-Y , White DJ , Hartenstein V , Eliceiri K , Tomancak P , Cardona A , Fiji: an open-source platform for biological-image analysis. Nature Methods. 9 (2012), pp. 676–682.22743772 56. Lim CKW , McCallister TX , Saporito-Magriña C , McPheron GD , Krishnan R , Zeballos C MA , Powell JE , Clark LV , Perez-Pinera P , Gaj T , CRISPR base editing of cis-regulatory elements enables the perturbation of neurodegeneration-linked genes. Mol. Ther. (2022), doi:10.1016/j.ymthe.2022.08.008.

PMC010xxxxxx/PMC10035603.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 7600130 5844 Neurosci Lett Neurosci Lett Neuroscience letters 0304-3940 1872-7972 35753613 10035603 10.1016/j.neulet.2022.136753 NIHMS1874820 Article Paired Associative Stimulation Applied to the Cortex can Increase Resting-state Functional Connectivity: A Proof-of-Principle Study Hooyman Andrew 1 Garbin Alexander 23 Fisher Beth E. 45 Kutch Jason J. 4 Winstein Carolee 45 1. School of Biological and Health Systems Engineering, Arizona State University, Tempe, AZ. 2. Department of Physical Medicine and Rehabilitation, University of Colorado Anschutz Medical Campus, Aurora, Colorado, USA 3. Geriatric Research Education and Clinical Center, VA Eastern Colorado Health Care System, Aurora, CO, USA 4. Division of Biokinesiology and Physical Therapy, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, CA. 5. Department of Neurology, Keck School of Medicine, University of Southern California, Los Angeles, CA. Correspondence concerning this article should be addressed to Andrew Hooyman, Contact: Andrew.hooyman@asu.edu 4 3 2023 27 7 2022 23 6 2022 27 7 2023 784 136753136753 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Introduction: There is emerging evidence that high Beta coherence (hBc) between prefrontal and motor corticies, measured with resting-state electroencephalography (rs-EEG), can be an accurate predictor of motor skill learning and stroke recovery. However, it remains unknown whether and how intracortical connectivity may be influenced using neuromodulation. Therefore, a cortico-cortico PAS (ccPAS) paradigm may be used to increase resting-state intracortical connectivity (rs-IC) within a targeted neural circuit. Purpose: Our purpose is to demonstrate proof of principle that ccPAS can be used to increase rs-IC between a prefrontal and motor cortical region. Methods: Eleven non-disabled adults were recruited (mean age 26.4, sd 5.6, 5 female). Each participant underwent a double baseline measurement, followed by a real and control ccPAS condition, counter-balanced for order. Control and ccPAS conditions were performed over electrodes of the right prefrontal and motor cortex. Both ccPAS conditions were identical apart from the inter-stimulus interval (i.e ISI 5 ms: real ccPAS and 500 ms: control ccPAS). Whole brain rs-EEG of high Beta coherence (hBc) was acquired before and after each ccPAS condition and then analyzed for changes in rs-IC along the targeted circuit. Results: Compared to ccPAS500 and baseline, ccPAS5 induced a significant increase in rs-IC, measured as coherence between electrodes over right prefrontal and motor cortex, (p < .05). Conclusion: These findings demonstrate proof of principle that ccPAS with an STDP derived ISI, can effectively increase hBc along a targeted circuit. pmcIntroduction Recent research in stroke rehabilitation and motor skill learning have demonstrated how resting state connectivity between specific cortical circuits, i.e. strength of communication between distinct brain areas in the cortex, can predict both learning of a complex motor skill and motor recovery from stroke [1-3]. These findings provide potential neural substrates that may be modified with the goal of improving recovery of motor learning capability in stroke survivors. However, there is limited evidence that resting-state connectivity can be reliably modified even in non-disabled adults. One method to measure resting-state connectivity is resting-state electroencephalography (rs-EEG), which has been shown to measure the aggregate firing of large neuronal pools, represented as neuronal oscillations [4], within the resting cortex. The strength of oscillatory coupling between neuronal pools, measured as coherence [5], between distinct cortical areas represents information exchange between these regions. Intracortical coherence can be measured across different frequencies, representing a faster or slower rate of oscillation between neuronal pools. Interestingly, resting state coherence has been found to be predictive of different types of behaviors [6]. Specifically, a measure of resting-state intracortical coherence within the Beta band (12 – 30 Hz), has been shown to be a reliable predictor of individual motor learning capability and stroke recovery, i.e. greater resting-state Beta coherence (hBc) predicts greater learning and motor recovery [2,7]. Specifically, resting-state hBc between the prefrontal and motor cortices has been particularly robust to motor skill acquisition [8], which is consistent with previous research that demonstrates flow of information between these cortices does occur during motor task performance [9]. Additionally, previous non-invasive brain stimulation studies have demonstrated that stimulation of these areas has some beneficial impact on motor skill learning, although no direct measure of change in connectivity between the prefrontal and motor cortices was recorded [10,11]. This research provides a viable mechanism of action, hBc between prefrontal and motor cortex, to shift our understanding of rs-EEG from correlation to causality [12], to improve motor learning, enhance stroke motor recovery and potentially treat a wide array of other neurologic disorders [13]. Specifically, if non-invasive modulation of critical resting-state intracortical coherence can be reliably achieved, then its future use as an intervention could be tested. Given that we know the underlying biology believed to drive resting-state intracortical connectivity, coupled neuronal firing between regions of the cortex, we can hypothesize that the precise mechanism by which connectivity would be modified is through spike timing dependent plasticity (STDP), neurons that wire together fire together [14]. This has led to the utilization of cortico-cortico Paired Associative Stimulation (ccPAS) as a method to implement STDP-like processes to facilitate connectivity between distinct cortical regions [15]. The PAS protocol is the application of repetitive stimulation of timed paired pulses, which are delivered via transcranial magnetic stimulation (TMS), to induce changes in excitability via a STDP like mechanism [14], with the timing between pulses eliciting either an excitatory or inhibitory effect [16]. However, few studies have investigated the efficacy of ccPAS to significantly increase resting-state intracortical coherence [17-19], there has yet to be a rigorous test that STDP is the probable mechanism of action [20]. Previous research has measured the effect of ccPAS in terms of motor cortical excitability through measuring changes in motor evoked potentials [21-23], estimates of global cortical excitability [18,20] or local cortical excitability [18,24]. Additionally, few studies have used control conditions similar to the ccPAS primary method, i.e. variation in the interstimulus interval or TMS intensity, to determine how ccPAS changes intracortical coherence and the few that did, have yet to demonstrate efficacy in resting-state [19,25] or in the brain areas proposed here [26,27]. Therefore, this study was designed to answer the question: Can intracortical paired associative stimulation (ccPAS) be used to increase resting-state intracortical coherence between prefrontal and motor cortices? However, to appropriately address this question the critical barrier of individual responsiveness to brain stimulation must be adequately addressed. Previous research has demonstrated that the application of a variety of different NIBS protocols (iTBS, rTMS, and TDCS) have a roughly 50% response rate among young non-disabled adults [28-30]. Additionally, there is other evidence that the application of a TMS control, one in which it is believed that no meaningful level of electromagnetic energy is being passed into the brain, may still have some neuromodulatory effect [31]. Therefore, we argue, that the testing for efficacy of any NIBS protocol should: 1) utilize a within subject design to account for variability and 2) utilize a control rather than sham condition for comparison to the treatment of interest. Here, we examine short versus long ISI to determine ccPAS efficacy and whether the mechanism of action is the result of a STDP-like mechanism. Materials and Methods Experimental Protocol Eleven non-disabled right hand dominant adults (5 female, mean age 26.4 +/− 5.6) consented to participate and were assessed along recommended guidelines for safety when using Transcranial Magnetic Stimulation (TMS) [32]. Hand dominance was assessed using the Edinburgh Handedness Inventory [33]. All study protocols were approved by the appropriate ethical review board of the University of Southern California. This experiment employed a combined Transcranial Magnetic Stimulation and Electroencephalography (TMS/EEG) methodology and a crossover within-subject design. To focus our proof of principle study, we chose to investigate changes in resting state high Beta coherence (hBc) (21 – 30 Hz) only, because of its reported importance in motor behavior [8]. Although we also examined changes in Gamma and Alpha band frequencies along our targeted circuit as well which are reported in supplemental material. Prior to any collection of EEG data, TMS stimulator intensity (Magstim 200, monophasic stimulator) for each individual was determined by identifying the resting motor threshold of the right abductor pollicus brevis with the EEG (AntNeuro) cap on. Participants wore the EEG cap during this step to control for the distance the cap creates between the skull and the coil which results in reduced transmission of the TMS pulse to the brain. The resulting motor threshold was then multiplied by 120%. This stimulator intensity is consistent with previous ccPAS research and was applied to each ccPAS condition [20,26,34]. Afterward, the EEG cap is oriented and calibrated a saline gel to achieve an impedance level of < 20 kOhms at each electrode. It is important to note that motor thresholding was applied to the left side of the brain and ccPAS was applied to the right side of the brain. This was done to eliminate any effect that thresholding may have on the ccPAS targeted neural circuit as previous research has shown that single pulse TMS can have an effect on EEG oscillatory activity [35]. To account for individual variance in high Beta coherence over time, each participant then underwent two baseline resting-state EEG (rs-EEG) measurements to first determine initial reliability of our rs-EEG measure. This is an important first step to identify how any observed change in rs-EEG coherence was due to noise and/or normal fluctuations in brain activity. We then calculated a standard error of measurement (SEM) from the two initial baseline trials to serve as our relative noise threshold to better confirm if changes in rs-EEG were due to ccPAS. To determine efficacy of ccPAS we employed an active condition, ccPAS5, with a 5 ms inter-stimulus delay and a control condition, ccPAS500, with a 500 ms inter-stimulus delay. The motivation to use 5 ms as the active ccPAS condition was informed by Veniero and colleagues 2013, which used a similar study design with brain areas a similar distance to those investigated in this study. There was a 5 second interval between every pair of pulses. This resulted in each condition being approximately 10 minutes in length. Figure-8 TMS coils were placed over EEG electrodes representing the right prefrontal, AF8, and motor cortex, C6, regions. The first pulse is delivered to AF8 and the second pulse is delivered to C6. Number of paired-pulses (100) and stimulator intensity were equivalent between ccPAS conditions. Figure 1 displays the complete EEG electrode layout as well as the TMS/EEG set-up during the experiment. It is important to note that although participants may have perceived the difference in timing between the real and control ccPAS conditions, they remained naïve to which delay was the control or treatment condition. Therefore, participants did not have any expectations about which condition would be more or less effective. After baseline collections, participants were split into two groups, each receiving both real and control ccPAS in a counterbalanced order. Prior to and immediately after each condition five minutes of rs-EEG was recorded with the primary outcome measure of change in high Beta coherence (hBc) between conditions. During each condition each participant was tasked with counting the number of paired pulses received during each condition. Additionally, different brain states can impact the effect of ccPAS [36], therefore participants were focused on the same thing during each condition and thus any confounding of different cognitive or mental states were controlled between participants and conditions. Between conditions participants had a 30-minute break to minimize any after-effects of either condition. One participant’s data was excluded due to artifacts in baseline measurements. This resulted in 10 participants total, five receiving the ccPAS5-ccPAS500 order and five receiving the ccPAS500-ccPAS5 order. Data Processing After data collection, Independent Component Analysis (ICA) was applied to each rs-EEG measure for purposes of removal of muscle and eye artifacts, no more than 2 components were removed per data collection. After ICA was performed the rs-EEG data were zero phase, band-passed filtered with cutoff frequencies between 10 and 60 Hz (Gustafsson, 1996). We chose this frequency range due to previous research investigating the association of the beta band (13 – 32 Hz) with the motor system [2,8]. Each electrode pair (1830 combinations total) is measured for their specific intracortical connectivity using the mscohere function from the Signal Processing Toolbox within MATLAB (version 2013a) with a 1 second Hanning window. The data was not re-referenced and coherence was calculated on the entire epoch. Previous research has shown that this method of calculating coherence does not impact the consistency of frequency based neural correlates [37]. The resulting coherence value from each electrode pair varies in value between 0 to 1 with a value closer to 1 representing stronger resting-state intracortical connectivity. Computer scripts on data processing and analysis related to this project, as well as, extended methods on TMS/EEG protocols like the one proposed here have been previously published to encourage replication [38]. Statistical Analysis Our primary outcome measure is the change in rs-EEG hBc of AF8 and C6 between conditions pre/post rest, pre/post ccPAS5 and pre/post ccPAS500. We also analyzed changes in hBc among two control circuits (AF7-C6 and O2-P6) to assess how condition may have indirectly influenced strength of coherence in non-targeted neural circuits. Changes in resting state connectivity between baseline, ccPAS5 and ccPAS500 were calculated using one-way analysis of variance (ANOVA) with a Tukey HSD to determine if changes in coherence were statistically different between each condition. Order effect between ccPAS conditions was determined using a two-way analysis of variance with change in coherence as the outcome and an interaction between order conditions received (ccPAS5-ccPAS500 vs ccPAS500-ccPAS5) and condition as independent variables. Analyses were performed in MATLAB version 2017a. However, to further estimate the level of response within each individual we also compared how change in each condition compares to a standard error of measurement derived from the two baseline measures. Standard error of measurement (SEM) was calculated from the intraclass correlation coefficient (ICC) from the two baseline rs-EEG measures, i.e. hBc between our targeted ROIs: AF8 to C6. To determine ICC from the baseline measurement we used a two-way mixed-effects model, where the raters, EEG device, is considered fixed and participant is a random effect. Calculating the SEM allows us to compare number of individual responders and non-responders in a more rigorous way, i.e. is change in resting state intracortical connectivity higher than our predetermined noise threshold, SEM. This gives us more sensitivity to ccPAS response beyond if change is greater than 0. Additionally, having a derived SEM allows us to identify the locality of ccPAS5 versus ccPAS500 or the rest condition across the entire scalp. It is important to note that our SEM is specific to our experiment and our participants. The calculated SEM should not be generalized to other experiments. Alternatively, we propose adopting a similar study design that we have generated here and which has been previously published [38]. Results Changes in Target Circuit Connectivity Due to ccPAS Changes in coherence between the double baseline, and the two pre-post ccPAS conditions were statistically significant (F(2,7) = 14.23, p<.01, Figure 2). Post-hoc analysis with Tukey HSD to adjust for multiple comparisons revealed that beta coherence increased in ccPAS5 to a greater degree than baseline (mean difference = 0.21, p < .001) and ccPAS500 (mean difference = 0.18, p < .001). Using a meaningful threshold (i.e. above SEM (+/−0.096) – dashed line, Figure 2) for change in coherence between AF8 and C6, we compared the response rate in each ccPAS condition. Seven out of 10 participants demonstrated an increase in coherence after the ccPAS5 condition, while only 1 out of 10 participants demonstrated an increase after the ccPAS500 condition (Figure 2). We also observed that ccPAS5 increased coherence in the Gamma band but not Alpha Band compared to baseline and ccPAS500 (Figure 3S and 4S). We also analyzed changes in hBc along a control circuit (O2-P6) and observed no changes between conditions (Figure 6S). We also visualized if the effect of ccPAS was local or diffuse to the site of stimulation across our targeted circuit and two control circuits (Figure 1S, 2S and 5S). Order Effect Between ccPAS Conditions To determine a possible order effect on coherence, we assessed any differential response in coherence change that depended on the initial condition (i.e., ccPAS5 or ccPAS500). Statistically, there was no interaction effect between order of condition received and condition on change in coherence (F(1,9) = .2, p = .65). The change in coherence for ccPAS5 was similar between groups (Figure 3 - compare the slopes of the blue and red segments). Figure 8s illustrates the comparison between each condition and condition order. We also demonstrate that lack of change in coherence across a control circuit was also not due to any order effect (p > .05, Figure 7s). Reliability of Coherence Measurement Across Baseline A unique feature of this study design is the use of two sequential baseline measures to determine normal fluctuations in coherence as a function of EEG device noise and normal fluctuations in individual brain rhythms. Results from our two-way mixed-effect model were statistically significant (p = .0021, 95% CI = [.62, .96]) with an ICC of .88 which indicates good reliability of coherence measurement across each baseline. A scatterplot of the two measures can be visualized in Figure 4 where comparison between each baseline is consistent and linear. Due to the reliability of coherence measurement across baseline we can then determine that changes observed during each ccPAS condition were due to stimulation and not due to factors that may be present because of normal device noise or biological fluctuations. Discussion These findings demonstrate proof of principle that ccPAS can increase hBc along a targeted circuit that has previously been associated with motor skill acquisition and learning [2,8]. We also observed changes in Gamma but not Alpha band coherence possibly due to the Alpha band being more related to cortical to subcortical coherence [37]. Further, no change in the ccPAS500 condition suggests that an STDP-like mechanism is driving the change in resting-state coherence in the ccPAS5 condition. Finally, stimulation order did not impact the efficacy of ccPAS5 to modulate rs-IC along the targeted circuit. We utilized a counterbalanced within-subject design to identify the unique effects of ccPAS within an individual and demonstrated that changes in rs-IC are contingent upon inter-stimulus interval. We believe this within-subject approach is superior to a between-subject approach in light of the high variability in responsiveness seen in previous neuromodulatory studies which have shown how age and even time of day can influence outcomes [29,39]. The application of SEM as a response threshold allows us to determine if changes in resting-state coherence are meaningful versus simply statistically significant, which is arguably a more important outcome for scientific research [40]. We encourage the application of a double baseline condition in future studies on brain stimulation to establish normal changes in brain connectivity prior to stimulation [38]. This work provides additional evidence that measured functional connectivity is likely representative of underlying structural connectivity [41]. Furthermore, results of this study support a potential intervention for individuals with weak coherence within the resting motor network [13]. Specifically, previous research has shown that individuals with stroke that have weak resting-state functional connectivity within the motor network exhibit motor deficits [42]. One may theorize that this technique applied prior to motor skill learning or rehab may improve performance as ccPAS may boost interconnectivity among the neural circuitry that governs motor learning. Furthermore, results from this preliminary study demonstrate that changes in connectivity in non-targeted circuits, see Figure 2S, may occur as a result of ccPAS as a possible side-effect of inducing current into the brain. Given that the brain is a system with individual brain areas that possess multiple bi-directional connections to other brain regions it is not unreasonable to conclude that stimulation in one area may have indirect effects in other areas [9]. As previous TMS-EEG experiments have shown using iTBS, application of stimulation in one area can have ripple effects to other areas [43-45]. Although response to ccPAS is consistent between individual participants within this study, see Figure 2, the level of change is still variable and what may predict this individual variance is still unknown. Limitations Although participants in this study would be able to discern between the two conditions due to their differences in ISI, 5 ms versus 500 ms, we believe they were naïve to which condition should have a neuromodulatory effect. Although we did not ask them which condition they believe would have a neuromodulatory effect, thus expectancy effects by condition were not controlled and future research should focus on controlling for participant bias [46]. Additionally, the identifiable difference in ccPAS conditions used in this study is similar to previous ccPAS studies that either have a longer ISI [25] or differing TMS coil orientation [47]. Our sample size is small; however, our study purpose was to establish proof of principle. Additionally, our data indicated that the 30-minute wash-out period between conditions was insufficient for rs-IC to return to baseline levels following ccPAS5 (Figure 3). Future research should extend the wash-out period to mitigate after-effects of ccPAS. However, given the lack of an order effect, we do not believe the length of the wash-out period confounded our results. Furthermore, we cannot be completely certain that thresholding performed on the left hemisphere is transferable for stimulation on the right hemisphere. Although, as we have shown this method does not seem to have negatively impacted our ability to elicit changes in coherence along our targeted circuit. Given the temporary change in hBc, future research should focus on ccPAS paradigms that induce a long-lasting change in hBc, i.e. repeated sessions within and across multiple days and with a larger sample. Conclusions Our findings demonstrate that ccPAS can modulate a specific resting-state circuit, likely through an STDP-like mechanism (i.e. change in connectivity dependent on inter-stimulus interval: 5 ms vs 500 ms). Based on these findings, ccPAS may have the potential to identify the casual relationship between rs-IC and motor behavior. Supplementary Material 1 Acknowledgements: The authors would like to thank the participants who participated in this research and the division of Biokinesiology for its support. Funding: JJK was supported, in part, by NIH grants DK121724 and DK110669. CJW was supported, in part by NIH grants HD059783 and HD104296. Figure 1: A. The electrode layout of our EEG cap following the 10-20 system. The blue electrode, GND, is the ground. The Green electrode, CPZ, is the reference electrode. The red electrode, AF8, is where the first TMS pulse is delivered. The yellow electrode, C6, is where the second TMS pulse is delivered. The dark arrow represents the direction of synaptic activity between the two electrodes. B. Participants were instructed to maintain their gaze on the red fixation cross during baseline and stimulation sessions. C. Orientation of TMS coils reference to target electrodes in A. Figure 2: Boxplots of Beta coherence change in AF8-C6 for baseline and the two ccPAS conditions pre-post (N = 10). Dashed lines represent +/− SEM threshold. Numbered boxes represent each participant across condition. Figure 3: Changes in the 5 rs-EEG sessions for condition order. CcPAS5 resulted in an increase in coherence in AF8-C6 compared to ccPAS500, regardless of condition order. Two-way dashed arrow points to each ccPAS5 slope of change in coherence. Error bars are standard error of each resting-state measurement. Figure 4: Scatterplot of individual high Beta coherence between the right prefrontal (AF8) and right motor (C6) cortex across the two baseline sessions. Individual coherence measures were found to be reliable across each baseline and allowed us to establish a standard error of measurement specific to this device and cohort. Highlights Reliability of baseline resting-state coherence established levels of meaningful change in coherence due to paired associative stimulation. Meaningful increases to targeted resting-state functional connectivity are timing dependent. Seventy percent of participants experienced a significant increase in targeted coherence compared to a control condition. Declaration of Interest: CW serves as a member of the DSMB for Enspire DBS Therapy, Inc.; a consultant for MicroTransponder, Inc. and receives royalty payments from Human Kinetics, Inc. (for 6th edition of Motor Control and Learning), and Demos Medical Publishers (for 2nd edition of Stroke Recovery and Rehabilitation). Remaining authors have nothing to declare. CRediT Author Statement Andrew Hooyman: Conceptualization, Methodology, Formal Analysis, Investigation, Data Curation, Writing – Original Draft, Writing – Review and Editing, Visualization. Alexander Gabrin: Conceptualization, Methodology, Investigation, Software. Beth E. Fisher: Conceptualization, Methodology, Resources. Jason J. Kutch: Resources, Writing – Original Draft, Writing – Review and Editing. Carolee Winstein: Supervision, Writing – Original Draft, Writing – Review and Editing References [1] Stewart JC , Dewanjee P , Shariff U , Cramer SC , Dorsal premotor activity and connectivity relate to action selection performance after stroke, Hum. Brain Mapp 37 (2016) 1816–1830. 10.1002/hbm.23138.26876608 [2] Wu J , Knapp F , Cramer SC , Srinivasan R , Electroencephalographic connectivity measures predict learning of a motor sequencing task, J. Neurophysiol 119 (2018) 490–498. 10.1152/jn.00580.2017.29093171 [3] Wu J , Quinlan EB , Dodakian L , McKenzie A , Kathuria N , Zhou RJ , Augsburger R , See J , Le VH , Srinivasan R , Cramer SC , Connectivity measures are robust biomarkers of cortical function and plasticity after stroke, Brain. 138 (2015) 2359–2369. 10.1093/brain/awv156.26070983 [4] Steriade M , Dossi R , Nunez A , Network modulation of a slow intrinsic oscillation of cat thalamocortical neurons implicated in sleep delta waves: cortically induced synchronization and brainstem cholinergic suppression, J. Neurosci 11 (1991) 3200–3217. 10.1523/JNEUROSCI.11-10-03200.1991.1941080 [5] Sakkalis V , Review of advanced techniques for the estimation of brain connectivity measured with EEG/MEG, Comput. Biol. Med 41 (2011) 1110–1117. 10.1016/j.compbiomed.2011.06.020.21794851 [6] Sugata H , Yagi K , Yazawa S , Nagase Y , Tsuruta K , Ikeda T , Nojima I , Hara M , Matsushita K , Kawakami K , Kawakami K , Role of beta-band resting-state functional connectivity as a predictor of motor learning ability, Neuroimage. 210 (2020) 116562. 10.1016/J.NEUROIMAGE.2020.116562.31972278 [7] Wu J , Srinivasan R , Burke Quinlan E , Solodkin A , Small SL , Cramer SC , Utility of EEG measures of brain function in patients with acute stroke, J. Neurophysiol 115 (2016) 2399–2405. 10.1152/jn.00978.2015.26936984 [8] Wu J , Srinivasan R , Kaur A , Cramer SC , Resting-state cortical connectivity predicts motor skill acquisition, Neuroimage. 91 (2014) 84–90. 10.1016/j.neuroimage.2014.01.026.24473097 [9] Anwar AR , Muthalib M , Perrey S , Galka A , Granert O , Wolff S , Heute U , Geuschl G , Raethjen J , Methuraman M , Effective Connectivity of Cortical Sensorimotor Networks During Finger Movement Tasks: A Simultaneous fNIRS, fMRI, EEG Study, Brain Topogr. 29 (2016) 645–660. 10.1007/s10548-016-0507-1.27438589 [10] Kaminski E , Hoff M , Sehm B , Taubert M , Conde V , Steele CJ , Villringer A , Ragert P , Effect of transcranial direct current stimulation (tDCS) during complex whole body motor skill learning., Neurosci. Lett 552 (2013) 76–80. 10.1016/j.neulet.2013.07.034.23933205 [11] Foerster A , Rocha S , Wiesiolek C , Chagas AP , Machado G , Silva E , Fregni F , Monte-Silva K , Site-specific effects of mental practice combined with transcranial direct current stimulation on motor learning., Eur. J. Neurosci 37 (2013) 786–94. 10.1111/ejn.12079.23279569 [12] Herrmann CS , Strüber D , Helfrich RF , Engel AK , EEG oscillations: From correlation to causality, Int. J. Psychophysiol 103 (2016) 12–21. 10.1016/j.ijpsycho.2015.02.003.25659527 [13] Koch G , Cortico-cortical connectivity: the road from basic neurophysiological interactions to therapeutic applications, Exp. Brain Res 238 (2020) 1677–1684. 10.1007/s00221-020-05844-5.32705294 [14] Stefan K , Kunesch E , Cohen LG , Benecke R , Classen J , Induction of plasticity in the human motor cortex by paired associative stimulation, Brain. 123 (2000) 572–584. http://www.tmslab.org/tms articles research/022.pdf (accessed May 1, 2017).10686179 [15] Koch G , Ruge D , Cheeran B , Fernandez Del Olmo M , Pecchioli C , Marconi B , Versace V , Lo Gerfo E , Torriero S , Oliveri M , Caltagirone C , Rothwell JC , TMS activation of interhemispheric pathways between the posterior parietal cortex and the contralateral motor cortex, J. Physiol 587 (2009) 4281–4292. 10.1113/jphysiol.2009.174086.19622612 [16] Bi GQ , Poo MM , Synaptic modifications in cultured hippocampal neurons: dependence on spike timing, synaptic strength, and postsynaptic cell type., J. Neurosci 18 (1998) 10464–72. 10.1523/JNEUROSCI.18-24-10464.1998.9852584 [17] Veniero D , Ponzo V , Koch G , Paired Associative Stimulation Enforces the Communication between Interconnected Areas, J. Neurosci 33 (2013) 13773–13783. 10.1523/JNEUROSCI.1777-13.2013.23966698 [18] Casula EP , Rocchi L , Hannah R , Rothwell JC , Effects of pulse width, waveform and current direction in the cortex: A combined cTMS-EEG study, Brain Stimul. 11 (2018) 1063–1070. 10.1016/j.brs.2018.04.015.29709505 [19] Kumar S , Zomorrodi R , Ghazala Z , Goodman MS , Blumberger DM , Daskalakis ZJ , Fischer CE , Mulsant BH , Pollock BG , Rajji TK , Effects of repetitive paired associative stimulation on brain plasticity and working memory in Alzheimer’s disease: a pilot randomized double-blind-controlled trial., Int. Psychogeriatrics (2020) 1–13. 10.1017/S1041610220003518. [20] Casula EP , Pellicciari MC , Picazio S , Caltagirone C , Koch G , Spike-timing-dependent plasticity in the human dorso-lateral prefrontal cortex, Neuroimage. 143 (2016) 204–213. 10.1016/j.neuroimage.2016.08.060.27591116 [21] Buch ER , Johnen VM , Nelissen N , O’Shea J , Rushworth MFS , Noninvasive Associative Plasticity Induction in a Corticocortical Pathway of the Human Brain, J. Neurosci 31 (2011) 17669–17679. 10.1523/JNEUROSCI.1513-11.2011.22131427 [22] Arai N , Muller-Dahlhaus F , Murakami T , Bliem B , Lu M-K , Ugawa Y , Ziemann U , State-Dependent and Timing-Dependent Bidirectional Associative Plasticity in the Human SMA-M1 Network, J. Neurosci 31 (2011) 15376–15383. 10.1523/JNEUROSCI.2271-11.2011.22031883 [23] Lu M-K , Tsai C-H , Ziemann U , Cerebellum to motor cortex paired associative stimulation induces bidirectional STDP-like plasticity in human motor cortex, Front. Hum. Neurosci 6 (2012) 260. 10.3389/fnhum.2012.00260.23049508 [24] Huber R , Maatta S , Esser SK , Sarasso S , Ferrarelli F , Watson A , Ferreri F , Peterson MJ , Tononi G , Measures of Cortical Plasticity after Transcranial Paired Associative Stimulation Predict Changes in Electroencephalogram Slow-Wave Activity during Subsequent Sleep, J. Neurosci 28 (2008) 7911–7918. 10.1523/JNEUROSCI.1636-08.2008.18667623 [25] Johnen VM , Neubert F-X , Buch ER , Verhagen L , O’Reilly JX , Mars RB , Rushworth MFS , Causal manipulation of functional connectivity in a specific neural pathway during behaviour and at rest, Elife. 4 (2015). 10.7554/eLife.04585. [26] Santarnecchi E , Momi D , Sprugnoli G , Neri F , Pascual-Leone A , Rossi A , Rossi S , Modulation of network-to-network connectivity via spike-timing-dependent noninvasive brain stimulation, Hum. Brain Mapp 39 (2018) 4870–4883. 10.1002/hbm.24329.30113111 [27] Veniero D , Ponzo V , Koch G , Paired Associative Stimulation Enforces the Communication between Interconnected Areas, J. Neurosci 33 (2013) 13773–13783. 10.1523/JNEUROSCI.1777-13.2013.23966698 [28] Hamada M , Murase N , Hasan A , Balaratnam M , Rothwell JC , The Role of Interneuron Networks in Driving Human Motor Cortical Plasticity, Cereb. Cortex 23 (2013) 1593–1605. 10.1093/cercor/bhs147.22661405 [29] Hamada M , Galea JM , Di Lazzaro V , Mazzone P , Ziemann U , Rothwell JC , Two distinct interneuron circuits in human motor cortex are linked to different subsets of physiological and behavioral plasticity., J. Neurosci 34 (2014) 12837–49. 10.1523/JNEUROSCI.1960-14.2014.25232119 [30] Wiethoff S , Hamada M , Rothwell JC , Variability in Response to Transcranial Direct Current Stimulation of the Motor Cortex, Brain Stimul. 7 (2014) 468–475. 10.1016/j.brs.2014.02.003.24630848 [31] Harvey RL , Edwards D , Dunning K , Fregni F , Stein J , Laine J , Rogers LM , Vox F , Durand-Sanchez A , Bockbrader M , Goldstein LB , Francisco GE , Kinney CL , Liu CY , Randomized Sham-Controlled Trial of Navigated Repetitive Transcranial Magnetic Stimulation for Motor Recovery in Stroke, Stroke. 49 (2018) 2138–2146. 10.1161/STROKEAHA.117.020607.30354990 [32] Rossini PM , Burke D , Chen R , Cohen LG , Daskalakis Z , Di Iorio R , Di Lazzaro V , Ferreri F , Fitzgerald PB , George MS , Hallett M , Lefaucheur JP , Langguth B , Matsumoto H , Miniussi C , Nitsche M. a. , Pascual-Leone a. , Paulus W , Rossi S , Rothwell JC , Siebner HR , Ugawa Y , Walsh V , Ziemann U , Non-invasive electrical and magnetic stimulation of the brain, spinal cord, roots and peripheral nerves: Basic principles and procedures for routine clinical and research application: An updated report from an I.F.C.N. Committee, Clin. Neurophysiol 126 (2015) 1071–1107. 10.1016/j.clinph.2015.02.001.25797650 [33] Oldfield RC , The assessment and analysis of handedness: The Edinburgh inventory, Neuropsychologia. 9 (1971) 97–113. 10.1016/0028-3932(71)90067-4.5146491 [34] Rizzo V , Siebner HS , Morgante F , Mastroeni C , Girlanda P , Quartarone A , Paired Associative Stimulation of Left and Right Human Motor Cortex Shapes Interhemispheric Motor Inhibition based on a Hebbian Mechanism, Cereb. Cortex 19 (2009) 907–915. 10.1093/cercor/bhn144.18791179 [35] Fuggetta G , Fiaschi A , Manganotti P , Modulation of cortical oscillatory activities induced by varying single-pulse transcranial magnetic stimulation intensity over the left primary motor area: A combined EEG and TMS study, Neuroimage. 27 (2005) 896–908. 10.1016/j.neuroimage.2005.05.013.16054397 [36] Chiappini E , Silvanto J , Hibbard PB , Avenanti A , Romei V , Strengthening functionally specific neural pathways with transcranial brain stimulation, Curr. Biol 28 (2018) R735–R736. 10.1016/j.cub.2018.05.083.29990453 [37] McLain NJ , Yani MS , Kutch JJ , Analytic consistency and neural correlates of peak alpha frequency in the study of pain, J. Neurosci. Methods 368 (2022) 109460. https://doi.org/10.1016/j.jneumeth.2021.109460.34958820 [38] Hooyman A , Garbin A , Fisher B , Paired Associative Stimulation Rewired: A Novel Paradigm to Modulate Resting-State Intracortical Connectivity, J. Mot. Learn. Dev 1 (2019) 1–20. 10.1123/jmld.2018-0054. [39] Corp DT , Bereznicki HGK , Clark GM , Youssef GJ , Fried PJ , Jannati A , Davies CB , Gomes-Osman J , Stamm J , Chung SW , Bowe SJ , Rogasch NC , Fitzgerald PB , Koch G , Di Lazzaro V , Pascual-Leone A , Enticott PG , Large-scale analysis of interindividual variability in theta-burst stimulation data: Results from the ‘Big TMS Data Collaboration,’ Brain Stimul. 13 (2020) 1476–1488. 10.1016/j.brs.2020.07.018.32758665 [40] McShane BB , Gal D , Gelman A , Robert C , Tackett JL , Abandon Statistical Significance, Am. Stat 73 (2019) 235–245. 10.1080/00031305.2018.1527253. [41] van den Heuvel MP , Mandl RCW , Kahn RS , Hulshoff Pol HE , Functionally linked resting-state networks reflect the underlying structural connectivity architecture of the human brain, Hum. Brain Mapp 30 (2009) 3127–3141. 10.1002/hbm.20737.19235882 [42] Golestani A-M , Tymchuk S , Demchuk A , Goodyear BG , Longitudinal Evaluation of Resting-State fMRI After Acute Stroke With Hemiparesis, Neurorehabil. Neural Repair 27 (2013) 153–163. 10.1177/1545968312457827.22995440 [43] Chung SW , Lewis BP , Rogasch NC , Saeki T , Thomson RH , Hoy KE , Bailey NW , Fitzgerald PB , Demonstration of short-term plasticity in the dorsolateral prefrontal cortex with theta burst stimulation: A TMS-EEG study, Clin. Neurophysiol 128 (2017) 1117–1126. 10.1016/j.clinph.2017.04.005.28511124 [44] Cárdenas-Morales L , Volz LJ , Michely J , Rehme AK , Pool E-M , Nettekoven C , Eickhoff SB , Fink GR , Grefkes C , Network connectivity and individual responses to brain stimulation in the human motor system., Cereb. Cortex 24 (2014) 1697–707. 10.1093/cercor/bht023.23395849 [45] Rocchi L , Spampinato D , Pezzopane V , Orth M , Bisiacchi P , Rothwell J , Casula E , Cerebellar noninvasive neuromodulation influences the reactivity of the contralateral primary motor cortex and surrounding areas: a TMS-EMG-EEG study, The Cerebellum. (2022). 10.1007/s12311-022-01398-0. [46] Wang P , Hooyman A , Schambra HM , Lohse KR , Schaefer SY , Expectations from the general public about the efficacy of transcranial direct current stimulation for improving motor performance, Brain Stimul. (2021). 10.1016/j.brs.2021.03.005. [47] Koch G , Ponzo V , Di Lorenzo F , Caltagirone C , Veniero D , Hebbian and Anti-Hebbian Spike-Timing-Dependent Plasticity of Human Cortico-Cortical Connections, J. Neurosci 33 (2013). http://www.jneurosci.org.libproxy2.usc.edu/content/33/23/9725/tab-figures-data (accessed September 15, 2017).

PMC010xxxxxx/PMC10097464.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0372354 646 Ann Surg Ann Surg Annals of surgery 0003-4932 1528-1140 35129464 10097464 10.1097/SLA.0000000000005370 NIHMS1877005 Article International Center-Level Variation in Utilization of Completion Lymph Node Dissection and Adjuvant Systemic Therapy for Sentinel Lymph Node Positive Melanoma Broman Kristy K. MD, MPH 123 Hughes Tasha M. MD, MPH 4 Bredbeck Brooke C. MD 4 Sun James MD 1 Kirichenko Dennis MS 2 Carr Michael J. MD 1 Sharma Avinash MD 5 Bartlett Edmund K. MD 5 Nijhuis Amanda A. G. MD, PhD 6 Thompson John F. MD 6 Hieken Tina J. MD 7 Kottschade Lisa CNP 7 Downs Jennifer MD 8 Gyorki David E. MBBS, MD 8 Stahlie Emma MD 9 van Akkooi Alexander MD, PhD 9 Ollila David W. MD 10 O’shea Kristin MEd 10 Song Yun MD 11 Karakousis Giorgos MD 11 Moncrieff Marc MD 12 Nobes Jenny FRCR 12 Vetto John MD 13 Han Dale MD 13 Hotz Meghan MD 14 Farma Jeffrey M. MD 14 Deneve Jeremiah L. DO 15 Fleming Martin D. MD 15 Perez Matthew MD 16 Baecher Kirsten MD 16 Lowe Michael MD 16 Bagge Roger Olofsson MD, PhD 17 Mattsson Jan MD, PhD 17 Lee Ann Y. MD 18 Berman Russell S. MD 18 Chai Harvey 19 Kroon Hidde M. MD, PhD 19 Teras Juri MD, PhD 20 Teras Roland M. MD 20 Farrow Norma E. MD, MHS 21 Beasley Georgia M. MD, MHS 21 Hui Jane Yuet Ching MD, MS 22 Been Lukas MD, PhD 23 Kruijff Schelto MD, PhD 23 Sinco Brandy MS 4 Sarnaik Amod A. MD 12 Sondak Vernon K. MD 12 Zager Jonathan S. MD 12* Dossett Lesly A. MD, MPH 4* International High Risk Melanoma Consortium 1 Moffitt Cancer Center, Tampa, FL, USA 2 University of South Florida Morsani College of Medicine, Tampa, FL, USA 3 University of Alabama at Birmingham, Birmingham, AL, USA 4 University of Michigan, Ann Arbor, MI, USA 5 Memorial Sloan Kettering Cancer Center, New York, NY, USA 6 Melanoma Institute Australia, The University of Sydney, Sydney, New South Wales, Australia 7 Mayo Clinic, Rochester, MN, USA 8 Peter MacCallum Cancer Center, Melbourne, Australia 9 Netherlands Cancer Institute, Amsterdam, The Netherlands 10 University of North Carolina, Chapel Hill, NC, USA 11 University of Pennsylvania, Philadelphia, PA, USA 12 Norfolk and Norwich University Hospital, Norwich, United Kingdom 13 Oregon Health & Science University, Portland, OR, USA 14 Fox Chase Cancer Center, Philadelphia, PA, USA 15 University of Tennessee, Memphis, TN, USA 16 Emory University, Atlanta, GA, USA 17 University of Gothenburg, Gothenburg, Sweden 18 NYU Langone Health, New York, NY, USA 19 Royal Adelaide Hospital, University of Adelaide, Adelaide, Australia 20 North Estonia Medical Centre Foundation, Tallinn, Estonia 21 Duke University, Durham, NC, USA 22 University of Minnesota, Minneapolis, MN, USA 23 University Medical Center, Groningen, Netherlands * Drs. Dossett and Zager served as co-senior authors Correspondences and Requests for Reprints: Kristy Kummerow Broman, MD, MPH, Assistant Professor, Division of Surgical Oncology, University of Alabama at Birmingham, Boshell Diabetes Building #575, 1808 7th Avenue South, Birmingham, Alabama 35233, kristybroman@uabmc.edu 10 3 2023 27 1 2022 27 7 2023 10.1097/SLA.0000000000005370This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. pmcIntroduction Advances in melanoma management have introduced new treatment paradigms for patients with sentinel lymph node (SLN) metastases. Two randomized surgical trials, the German Cooperative Dermatologic Oncology Group study (DeCOG-SLT) published in 2016 and the Second Multicenter Selective Lymphadenectomy Trial (MSLT-II) published in 2017, demonstrated the survival equivalence of nodal observation to routine completion lymph node dissection (CLND), prompting surgeons to reconsider the necessity of regional surgery for SLN-positive disease.1–3 Simultaneous publication of positive adjuvant systemic therapy trials showed that anti-CTLA4, anti-PD1, and BRAF/MEK inhibitors are more effective and less toxic than historic alternatives, providing additional treatment options for surgically-resected melanoma patients at high risk of recurrence and death.4–7 Based on these findings, the FDA approved ipilimumab in 2015, nivolumab in 2017, dabrafenib/trametinib in 2018, and pembrolizumab in 2019 for adjuvant treatment of resected stage III melanoma, with subsequent approvals by the corresponding regulatory bodies in Europe, the United Kingdom, and Australia (Figure 1).8 These landmark trials provide evidence to omit (de-implement) regional surgery for SLN-positive patients while simultaneously administering (implementing) new medical therapies to high-risk patients in the adjuvant setting.9 While there is some evidence to suggest a long average time to implementation for most practices, less is known about how quickly practices are de-implemented or reasons for variation in de-implementation practices.10,11 Further, it is unknown how implementation of systemic therapies might influence de-implementation of local or regional treatments such as CLND. While nodal observation and adjuvant systemic therapy trials were performed in parallel, neither were studied in combination, leaving patients and physicians with four potential treatment strategies – nodal observation alone, nodal observation with adjuvant systemic therapy, CLND alone, or CLND with adjuvant systemic therapy – with widely ranging treatment intensity, morbidity, and cost. There are currently no comparative data to discern which option is optimal for each individual patient. This unique scenario provides an opportunity to understand how new results are incorporated into practice for a single disease site and to study the dynamics of concurrent de-implementation of surgical treatment and implementation of adjuvant systemic therapy. As results from large oncologic databases are not yet mature, we used the database from the International High-Risk Melanoma Consortium consisting of 21 melanoma referral centers.12 Our objectives were to evaluate overall trends and center-level variation in de-implementation of CLND and implementation of adjuvant systemic therapy for SLN-positive melanoma. Methods The International High-Risk Melanoma Consortium was established in 2017 and includes a geographically diverse network of 21 melanoma referral centers from Australia, Europe (including the United Kingdom), and the United States (US).12 In this retrospective cohort study, each participating center provided data on adult patients with SLN-positive cutaneous melanoma who were treated from July 1, 2017-June 30, 2019. Requirements for center participation included having a nodal surveillance protocol in place before study initiation, attainment of institutional ethics/review board approval, negotiation of a data use agreement with the coordinating center, Moffitt Cancer Center, and provision of de-identified patient data by established deadlines. There was no designated funding source for this study. Reporting is in accordance with EQUATOR guidelines (Supplemental File). Data was collected during routine care of patients with clinically node negative melanoma who had metastatic melanoma in at least one SLN. Included patients were required to have margin-negative resection of the primary tumor and no evidence of distant metastases on staging studies performed either before or after positive SLN biopsy but prior to further treatment planning. Performance of nodal observation versus CLND and use of adjuvant systemic therapy (i.e., anti-PD1 or anti-CTLA4 immunotherapy, BRAF/MEK inhibitor) were determined by treating physicians and patients. Unlike prior adjuvant systemic therapy studies, patients were not required to undergo CLND before receiving adjuvant systemic treatment. We determined center-level rates of de-implementation of CLND and implementation of adjuvant systemic therapy for each 3-month period (quarter) over the two years of study to describe change over time. We also described variation in comprehensive management for SLN-positive patients treated at each center including the four possible treatment strategies – nodal observation alone, nodal observation with adjuvant systemic therapy, CLND alone, or CLND with adjuvant systemic therapy. We performed a comparative analysis of treatment strategies by the following center-level characteristics: geographic region (Australia, Europe, or US), whether the center previously participated in the MLST-II trial (no DeCOG-SLT sites were included in this study), designation as a cancer center by the National Cancer Institute, the European Society of Medical Oncology, or self-designated for Australian centers, and number of SLN-positive patients treated (volume reported by tertile). Center-level adoption rates were compared using Wilcoxon rank-sum tests for variables with two categories and Kruskal Wallis tests for variables with more than two categories. For findings of significant association by Kruskal Wallis tests (alpha <0.05), Dunn’s tests with Bonferroni correction for multiple comparisons were used to determine which specific elements of each variable were associated with the treatment outcome. To adjust for patient- and disease-specific characteristics, generalized linear mixed models with random intercepts for each center were used to assess variation in de-implementation of CLND and implementation of adjuvant systemic treatment. Separate models were created for each outcome (CLND and adjuvant systemic treatment). Models were adjusted for the following factors: primary site (head/neck, trunk, extremity), tumor ulceration, presence of microsatellitosis, American Joint Committee on Cancer (AJCC) 8th edition stage, size of largest nodal metastasis (<1 mm (millimeter) or ≥1 mm), and extranodal tumor extension. Values are reported as odds ratios with 95% Confidence Intervals (CI). To demonstrate center-level variation not explained by disease-specific factors, the models were used to determine the adjusted probability of CLND and adjuvant systemic treatment, respectively, by treating center. We evaluated the relative importance of each covariate by computing the Nagelkerke pseudo r-squares for each covariate alone and for all covariates except the covariate of focus. This enabled us to evaluate the contribution of the covariate by itself and its incremental effect (eTable 1).13 We also performed clinically relevant sensitivity analyses based on eligibility criteria for adjuvant therapy trials and pertinent treatment guidelines. As some guidelines do not recommend adjuvant systemic therapy for stage IIIA patients, we separately evaluated center-level variation in use of adjuvant systemic therapy for this group.14 Likewise, we examined differences in provision of adjuvant systemic therapy for patients with nodal tumor deposits <1 mm because eligibility criteria for clinical trials of adjuvant systemic therapy required a minimum nodal tumor deposit of 1 mm.4–7 Results Temporal trends in de-implementation of CLND and implementation of adjuvant systemic therapy Participating centers collectively treated 1,109 SLN-positive patients (Table 1). In the earliest quarter of study, which was concurrent with MSLT-II publication, 28% of patients underwent CLND. This was lower than previously published rates and decreased to 8% by the last quarter of the two-year study period. At the same time adjuvant systemic therapy use increased from 29% to 60% over the two-year period (Figure 1). Center-level variation Combining nodal management and adjuvant systemic treatment strategies, patients were managed with nodal observation alone (n=519, 47%), nodal observation with adjuvant systemic therapy (n=411, 37%), CLND alone (n=102, 9%), or CLND with adjuvant systemic therapy (n=77, 7%) (Figure 2). US centers treated more patients with adjuvant therapy than European centers during the period of study, whether doing nodal observation (p=0.01) or CLND (p=0.04), while adjuvant systemic therapy use at Australian centers was not significantly different from US or European centers (Table 2). At the center level there were no significant associations between performance of CLND or use of adjuvant systemic therapy and melanoma patient volume, region, cancer center designation, or prior participation in MSLT-II (Table 2). Multi-level models In the multilevel models, higher odds of CLND were associated with head and neck primary site (relative to extremity) and nodal tumor deposit of ≥1 mm (Table 3). Accounting for disease-specific factors, the adjusted probability of CLND based on treating center ranged from 1% to 83% (median 10%) (Figure 3). Odds of adjuvant systemic therapy increased for nodal tumor deposit of ≥1 mm and decreased for patients with stage IIIA disease relative to IIIC or IIID (Table 3). Adjusted probabilities of adjuvant systemic therapy ranged from 9% to 87% by treating center (median 46%) (Figure 3). For both CLND and adjuvant systemic therapy, the most influential covariates in explaining observed variation were treating center, tumor size, and stage. Sensitivity analyses By stage, the proportion of patients receiving adjuvant systemic therapy was IIIA 28%, IIIB 44%, IIIC/D 55%. There were differences in adjuvant treatment for Stage IIIA versus Stage IIIB-D disease at the regional and center levels (Figure 4). By center, the proportion of stage IIIA patients who received adjuvant systemic therapy ranged from 0% to 88% with five centers not treating any stage IIIA patient with adjuvant systemic therapy and ten centers using adjuvant therapy in more than one-quarter of patients with stage IIIA disease. Center-level variation was similarly observed when patients were stratified by size of largest nodal tumor deposit. Median rates of adjuvant systemic therapy use for patients with nodal tumor deposits <1 mm ranged by center from 0 to 100% with 10 of 21 centers using adjuvant systemic therapy for more than one-quarter of their patients with nodal tumor deposits <1 mm. Discussion This study has three main findings. First, at major melanoma centers world-wide there has been rapid but varied incorporation of surgical trial findings into routine care for SLN-positive patients. Second, there has been a simultaneous increase in use of adjuvant therapy in SLN-positive patients. Third, while performance of CLND and administration of adjuvant systemic treatment were associated with disease-specific factors including primary tumor features and burden of SLN-positive disease, there was also significant variation in CLND and adjuvant systemic treatment patterns based on the center where patients received care. Our data demonstrate the pace of CLND de-implementation was swift over the ensuing two years of study, corroborating the findings of single institution studies and demonstrating a much shorter time to practice change than the average 17 years often cited in implementation research.15–17 One contributing factor may have been a pre-existing acknowledgement of the limitations of CLND. Prior to MSLT-II publication, several large retrospective cohort studies already suggested limited benefit of CLND, with most patients having no additional positive (non-sentinel) nodes in CLND specimens.18–20 At the time, several risk prediction tools for non-sentinel node positivity were available to support a decision not to perform CLND.18,21–24 Rates of CLND at this study initiation and in prior studies demonstrate that CLND was already being performed selectively prior to publication or presentation of DeCOG-SLT or MSLT-II results. At American College of Surgeons Commission on Cancer (CoC) participating centers in the US, the rate of CLND was 63% in 2012 and 50% in 2016.25 Even in the MSLT-II trial, more patients who were randomized to CLND did not undergo the prescribed intervention. Patients randomized to CLND who did not undergo node dissection had significantly lower nodal disease burdens, suggesting that the perceived risk of positive non-sentinel nodes influenced patients and/or physicians in their decision to accept the assigned treatment.1 Additional explanations for low baseline performance of CLND and swift de-implementation might include surgeons’ level of comfort with performing the procedure and the risk of potentially life-altering lymphedema.26 Further, MLST-II trial results were well-disseminated, with a recent survey of the Society of Surgical Oncology membership finding that 98% of respondents were aware of its findings.27 Research findings that are particularly impactful to a highly specialized provider group may disseminate more quickly due to the close-knit nature of subspecialty practitioners who routinely seek information and colleagues’ advice from outside their immediate practice environment. Finally, it is notable that the curves for de-implementation of CLND and implementation of adjuvant systemic therapy have an inverse relationship. While there is no available evidence to suggest that adjuvant systemic treatment is an effective replacement for CLND or that it confers additional regional control, patients and physicians may have been more comfortable forgoing additional surgery when alternative treatments were available to mitigate recurrence.9 Similar to trends in de-implementation of CLND, the implementation of adjuvant systemic therapy for SLN-positive melanoma began before the start of this study but rapidly escalated in a comparable timeframe. In our 21 melanoma referral centers, adjuvant systemic treatment increased from 29% in July 2017 to 60% in June 2019. Centers with high adoption used adjuvant systemic therapy in up to 92% of SLN-positive patients, including large proportions of patients with stage IIIA disease and nodal tumor deposits <1 mm. Other reports from single institution cohorts of SLN-positive patients not undergoing CLND have reported use of adjuvant systemic therapy in 69–75%.16,28 There are several potential reasons for the accelerated implementation of adjuvant systemic therapy in SLN-positive melanoma. Historically, regionally metastatic melanoma carried a poor prognosis, with only 28–44% of patients having recurrence-free survival at 5 years. Effective, well-tolerated adjuvant treatments represented a significant therapeutic advance.29 These agents had previously been tested in the setting of stage IV and unresectable stage III disease, demonstrating often dramatic response rates and significant improvements in progression-free survival.30–32 Finally, concurrent trials of immunotherapy in other solid tumors increased widespread knowledge within and outside the medical community, with drug companies broadly disseminating information about the medications, including direct to consumer advertising in the US.33 Despite overall adoption of these evidence-based practices, there remained variation both in de-implementation of CLND and implementation of adjuvant systemic treatment based on where patients were treated. Several possible reasons exist. First, unmeasured patient factors such as travel time to the treating center may have influenced patients’ preferences for both nodal observation and receiving a year of adjuvant systemic treatment. Secondly, while FDA approval came during the study period for several of the contemporary adjuvant systemic therapies, regulatory approvals were later in Europe and Australia. While the payer mix in US centers is quite heterogenous, all participating centers in Europe and Australia have some form of universal, government-run healthcare, which initially might delay or limit access to new, expensive adjuvant systemic treatments. Still, even within the studied US centers there was profound variation in both nodal observation and adjuvant systemic treatment rates for SLN-positive patients. A final potential contributor to the observed variation in both CLND and adjuvant systemic treatment is physicians’ interpretation and application of available evidence. A recent survey demonstrated that most SLN-positive melanoma patients prefer to follow their physicians’ recommendations regarding CLND, highlighting the importance of the local context in which patients receive care and the constitution of patients’ treatment teams.9,34 Evidence from randomized trials of nodal observation and adjuvant systemic treatment have been informative, but several knowledge gaps remain. Adjuvant trials, for instance, mandated CLND prior to systemic treatment, while under 10% of nodal observation trial participants received adjuvant therapy. As a result, high-level evidence is lacking on outcomes of nodal observation in adjuvant systemic therapy recipients.1,2,5,6,35,36 Also, certain populations of SLN-positive patients were underrepresented in these trials. While adjuvant systemic therapy trials required a minimal nodal tumor deposit dimension of 1 mm, patients with low nodal tumor burden constituted the majority of participants in the randomized surgical trials of nodal observation.1,2,5,6,35,36 Despite these significant differences in the study populations, our data demonstrate that many treatment teams have readily integrated the two contemporary strategies, offering patients nodal observation with adjuvant systemic treatment despite a lack of randomized evidence and only limited survival data from observational cohorts, even in patients whose tumor and nodal burdens were not represented in the randomized trials.16,28,37,38 In certain cases and centers, interpretations of available evidence may have resulted in overuse of adjuvant systemic therapy or non-evidence-based de-implementation of CLND in patients who were not represented in the nodal management trials.12 For example, a sizeable proportion of stage IIIA patients with nodal tumor deposits of <1 mm received adjuvant systemic therapy despite the absence of efficacy data for patients with low nodal tumor burdens.14 For such patients, the risk of adjuvant systemic treatment-related adverse events may exceed potential benefits. The observed variation in treatment intensity for SLN-positive melanoma, from nodal observation alone to CLND with adjuvant treatment, is associated significant differences in patient morbidity, travel burden, anxiety, and cost. Hence, it is critical to develop indications and understand outcomes for each of these combined treatment strategies. Until national datasets mature, the experience of this multi-institutional international collaborative represents the best available data on de-implementation of CLND and implementation of adjuvant systemic therapy for SLN-positive melanoma. One limitation of the study is reliance on data from melanoma referral centers which may not reflect management in other patient populations. As location of care and specifically treatment at a cancer center has been found to significantly impact implementation of evidence-based care, trends in implementation of nodal observation and adjuvant systemic treatment at non-referral centers may differ.39–44 In addition, while our international collaborative represents countries with some of the highest worldwide incidences of melanoma, it was limited to higher income countries with populations of predominantly European ancestry, limiting generalizability to other populations.45 With this retrospective study using clinical data, we were not positioned to study the specific reasons for CLND or adjuvant systemic therapy use at each center, nor could we evaluate potentially time-variant changes in barriers to or facilitators of implementation such as availability of adjuvant systemic treatments or high-quality ultrasound to perform nodal basin surveillance. Conclusions In an evolving treatment landscape for SLN-positive melanoma, fewer patients are undergoing CLND and more are receiving adjuvant systemic therapy. These changes in practice began prior to the publication of landmark trials of nodal observation and adjuvant immunotherapy and targeted therapy but accelerated dramatically at the included melanoma referral centers over a two-year time period post-publication. Location of care contributed significantly to the observed variation in de-implementation of CLND and implementation of adjuvant systemic treatment and was not explained by differences in patient mix. As there are significant differences in potential morbidity and cost of available treatment strategies, future work should explore how the context of care delivery and interprofessional interactions impact the incorporation of evidence-based findings into clinical care. Supplementary Material Supplemental Table Acknowledgements and Disclosures Dr. Beasley is a consultant for Regeneron and receives clinical trial funding from Istari Oncology. Dr. Bredbeck is supported by the Ruth L. Kirschstein Research Service Award from the National Cancer Institute (NCI; T32 CA009672). Dr. Farma is a speaker for Novartis and serves on the Data Safety Monitoring Board for Delcath. Dr. Gyorki has received honoraria from Amgen, Bayer, and Qbiotics. He has served on the advisory board for Amgen. Dr. Hieken has received research funding from Genentech and the Breast Cancer Research Foundation. Ms. Kottschade was received research funding from Bristol-Meyers Squibb and serves on the advisory board for Novartis. Dr. Olofsson Bagge has received institutional research grants from Astra Zeneca, Bristol-Myers Squibb (BMS) and SkyLineDx, speaker honorarium from Roche and Pfizer and has served on advisory boards for Amgen, BD/BARD, Bristol-Myers Squibb (BMS), Merck Sharp & Dohme (MSD), Novartis, Roche and Sanofi Genzyme. Dr. Sarnaik is a consultant to Iovance Biotherapeutics, Guidepoint, Defined Health, and Gerson Lehman group. His institution has received research funding from Iovance Biotherapeutics and Provectus Inc. He has received speaker fees from Physicians’ Education Resource and Medscape. Dr. Sondak is a paid consultant to Merck, Bristol Myers Squibb, Novartis, Regeneron, Array, Replimune, Pfizer, Genentech/Roche, Eisai, Aduro, Amgen, TRM Oncology, and Polynoma. Dr. Thompson is supported by the National Health and Medical Research Council (Grant #APP1093017). He has received honoraria for advisory board participation from Bristol Meyers Squibb Australia, Glaxo Smith Kline, Merck Sharp & Dohme Australia, Provectus. He has received travel support from Glaxo Smith Kline and Provectus. Dr. van Akkooi had received honoraria and/or served on advisory boards for Amgen, Bristol Myers Squibb, Novartis, MSD-Merck, Merck-Pfizer, Sanofi, and 4SC. He has received research grants from Amgen and Merck-Pfizer. Dr. Vetto is a consultant for Castle Biosciences. Dr. Zager has received research funding from Novartis, Philogen, Delcath Systems, Amgen, Provectus, Novartis and Castle Biosciences. He has served on Advisory Boards for Merck, Novartis and Pfizer and on the Speakers Bureau for Pfizer, Sun Pharma, and Castle Biosciences. He provides expert testimony to McGowan Hood and Bubalo Law. All other authors reported no disclosures. Figure 1. Proportion of patients who underwent completion lymph node dissection and received adjuvant systemic therapy before and after DeCOG-SLT and MSLT-2 publication and region-specific regulatory approvals of adjuvant immunotherapies and BRAF/MEK inhibitor therapy *Historical rates of CLND for years 2012 and 2016 and of adjuvant systemic therapy for 2016 for resected stage III melanoma were obtained from the National Cancer Database; historical rate of adjuvant systemic therapy from 2012 derived from MSLT-II and DeCOG-SLT publications Abbreviations: DeCOG-SLT = German Dermatologic Oncology Group Trial; MSLT-II = Second Multicenter Selective Lymphadenectomy Trial; FDA = United States Food and Drug Administration; EMA = European Medicines Agency (Europe); NICE = National Institute for Healthcare Excellence (United Kingdom); PBAC = Pharmaceutical Benefits Advisory Committee (Australia); Dab/tram = dabrafenib/trametinib Figure 2. Nodal management with observation versus completion lymph node dissection (CLND) and adjuvant systemic therapy use for patients with melanoma treated at twenty-one participating institutions in Australia, Europe, and the United States Figure 3: Probability of completion lymph node dissection (CLND) and adjuvant systemic treatment by treating centera aAdjusted for primary tumor site, ulceration, AJCC 8th edition stage, size of largest nodal tumor deposit, microsatellitosis, and extranodal tumor extension; models contained random intercept to account for clustering of patients within facility. Figure 4. Nodal management and adjuvant systemic therapy for AJCC 8th Edition Stage IIIA (A) versus Stage IIIB-D (B) melanoma patients based on region of treating center AJCC = American Joint Committee on Cancer; CLND=completion lymph node dissection; USA =United States of America Table 1. Characteristics of treating centers and SLN-positive melanoma patients TREATING CENTERS PATIENTS Number of centers 21 Number of patients 1,109 Region Male gender, N (%) 672 (61%)  Australia 3 (14%) Age, years, median (25th–75th %ile) 61 (49–71)  Europe 5 (24%) Tumor location, N (%)  United States 13 (62%) Head and Neck 144 (13%) Volume tertile (# SLN+ pts/yr) Trunk 428 (39%)  Low (6–15) 8 (38%) Extremity 537 (48%)  Middle (16–27) 6 (29%) Breslow depth, mm, median (25th–75th %ile) 2.5 (1.5–4.2)  High (28–90) 7 (33%) Tumor ulceration, N (%) 453 (41%) Cancer center a Microsatellites, N (%) 95 (9%)  No 5 (24%) Number of positive SLN, N (%)  Yes 16 (76%) 1 842 (76%) MSLT-2 trial participant 2–3 247 (22%)  No 14 (67%) 4 or more 20 (2%)  Yes 7 (33%) Size of largest nodal tumor deposit, N (%) <1 mm 508 (46%) ≥1 mm 475 (43%) Unknown 126 (11%) Extranodal extension N (%) 71 (6%) AJCC 8th edition stage, N (%) IIIA 333 (30%) IIIB 242 (22%) IIIC 490 (44%) IIID 21 (2%) III, not specified 23 (2%) a National Cancer Institute-designated, European Society of Medical Oncology-designated, or self-designated cancer centers for those outside NCI or ESMO jurisdiction SLN=sentinel lymph node; pts=patients; yr=year; MSLT-2=Second Multicenter Selective Lymphadenectomy Trial; mm=millimeters; AJCC=American Joint Committee on Cancer Table 2. Institutional factors associated with nodal management and adjuvant treatment Nodal observationa P-Value Nodal observation with adjuvanta P-Value CLND alonea P-Value CLND with adjuvanta P-Value All Centers 47% (0–92%) 43% (0–77%) 5% (0–83%) 7% (0–33%) N/A Region  Australia 50% (35–56%) 0.13 44% (18–62%) 0.03c 1% (0–18%) 0.47 1% (0–14%) 0.08d  Europe 73% (11–92%) 17% (0–27%) 8% (0–83%) 0% (0–7%)  US 41% (0–67%) 44% (2–77%) 5% (0–23%) 9% (0–33%) Volume tertile (# SLN+ pts/yr)  Low (6–15) 27% (0–92%) 0.42 38% (0–77%) 0.95 8% (0–83%) 0.77 7% (0–33%) 0.65  Mid (16–27) 49% (21–67%) 41% (2–66%) 6% (0–22%) 10% (0–11%)  High (28–90) 50% (34–73%) 43% (15–62%) 1% (0–23%) 1% (0–19%) Cancer centerb  Yes 43% (0–73%) 0.46 44% (0–77%) 0.37 4% (0–83%) 0.75 7% (0–19%) 0.66  No 51% (10–92%) 18% (0–57%) 8% (0–18%) 7% (0–33%) MSLT-2 trial participant  Yes 49% (21–56%) 0.60 33% (0–77%) 0.06 1% (0–83%) 0.16 7% (0–33%) 0.50  No 45% (0–92%) 49% (24–71%) 8% (0–17%) 9% (0–11%) a Analysis by treating center reported as the proportion of patients at each treating center who received each treatment category with values representing center-level medians and ranges b National Cancer Institute-designated, European Society of Medical Oncology-designated, or self-designated cancer centers for those outside NCI or ESMO jurisdiction; CLND=completion lymph node dissection; US=United States; SLN+=sentinel lymph node positive; pts=patients; yr=year; MSLT-2=Second Multicenter Selective Lymphadenectomy Trial C p=0.01 for comparison of rates in US versus Europe, 1.00 for US versus Australia, 0.14 for Europe versus Australia d p=0.04 for comparison of rates in US versus Europe, 0.55 for US versus Australia, 0.64 for Europe versus Australia Table 3. Probability of completion lymph node dissection and adjuvant systemic therapy based on patient, disease, and treating center factorsa Completion Lymph Node Dissection Adjuvant Systemic Therapy Odds Ratio (95% Confidence Interval) P-Value Odds Ratio (95% Confidence Interval) P-Value Primary site  Head and neck 2.19 (1.26, 3.80) 0.006 1.04 (0.66, 1.61) 0.877  Trunk 1.10 (0.73, 1.66) 0.657 1.06 (0.77, 1.47) 0.705  Extremity Reference Reference Tumor ulceration 0.88 (0.54, 1.43) 0.594 1.20 (0.82, 1.75) 0.349 Microsatellites 0.84 (0.45, 1.56) 0.582 0.92 (0.54, 1.55) 0.746 AJCC 8th edition stage  IIIA 0.70 (0.39, 1.26) 0.234 0.37 (0.23, 0.59) <0.001  IIIB 0.57 (0.32, 1.02) 0.060 0.71 (0.47, 1.08) 0.108  IIIC/D Reference Reference Nodal tumor ≥1 millimeter 3.62 (2.33, 5.63) <0.001 1.65 (1.20, 2.27) 0.002 Extranodal tumor extension 1.70 (0.89, 3.24) 0.106 1.52 (0.84, 2.76) 0.165 a Models contained random intercept to account for clustering of patients within facility References 1. Faries MB , Thompson JF , Cochran AJ , Completion Dissection or Observation for Sentinel-Node Metastasis in Melanoma. N Engl J Med. 2017;376 (23 ):2211–2222. doi:10.1056/NEJMoa1613210 28591523 2. Leiter U , Stadler R , Mauch C , Complete lymph node dissection versus no dissection in patients with sentinel lymph node biopsy positive melanoma (DeCOG-SLT): a multicentre, randomised, phase 3 trial. Lancet Oncol. 2016;17 (6 ):757–767. doi:10.1016/S1470-2045(16)00141-8 27161539 3. Leiter U , Stadler R , Mauch C , Final Analysis of DeCOG-SLT Trial: No Survival Benefit for Complete Lymph Node Dissection in Patients With Melanoma With Positive Sentinel Node. J Clin Oncol Off J Am Soc Clin Oncol. 2019;37 (32 ):3000–3008. doi:10.1200/JCO.18.02306 4. Eggermont AMM , Chiarion-Sileni V , Grob J-J , Adjuvant ipilimumab versus placebo after complete resection of high-risk stage III melanoma (EORTC 18071): a randomised, double-blind, phase 3 trial. Lancet Oncol. 2015;16 (5 ):522–530. doi:10.1016/S1470-2045(15)70122-1 25840693 5. Eggermont AMM , Blank CU , Mandala M , Adjuvant Pembrolizumab versus Placebo in Resected Stage III Melanoma. N Engl J Med. 2018;378 (19 ):1789–1801. doi:10.1056/NEJMoa1802357 29658430 6. Weber J , Mandala M , Del Vecchio M , Adjuvant Nivolumab versus Ipilimumab in Resected Stage III or IV Melanoma. N Engl J Med. 2017;377 (19 ):1824–1835. doi:10.1056/NEJMoa1709030 28891423 7. Long GV , Hauschild A , Santinami M , Adjuvant Dabrafenib plus Trametinib in Stage III BRAF-Mutated Melanoma. N Engl J Med. 2017;377 (19 ):1813–1823. doi:10.1056/NEJMoa1708539 28891408 8. Kwak M , Farrow NE , Salama AKS , Updates in adjuvant systemic therapy for melanoma. J Surg Oncol. 2019;119 (2 ):222–231. doi:10.1002/jso.25298 30481375 9. Norton WE , Chambers DA . Unpacking the complexities of de-implementing inappropriate health interventions. Implement Sci IS. 2020;15 (1 ):2. doi:10.1186/s13012-019-0960-9 31915032 10. Nilsen. Theories, models, and frameworks for de-implementation of low value care: A scoping review of the literature. Implement Res Pract. 2020;1 :1–15. 11. Wang T , Bredbeck BC , Sinco B , Variations in Persistent Use of Low-Value Breast Cancer Surgery. JAMA Surg. Published online February 3, 2021. doi:10.1001/jamasurg.2020.6942 12. Broman KK , Hughes TM , Dossett LA , Surveillance of Sentinel Node-Positive Melanoma Patients with Reasons for Exclusion from MLST-II: Multi-Institutional Propensity Score Matched Analysis. J Am Coll Surg. Published online December 13, 2020. doi:10.1016/j.jamcollsurg.2020.11.014 13. Nagelkerke NJD . A note on a general definition of the coefficient of determination. Biometrika. 1991;78 (3 ):691–692. 14. Michielin O , van Akkooi A , Lorigan P , ESMO consensus conference recommendations on the management of locoregional melanoma: under the auspices of the ESMO Guidelines Committee. Ann Oncol Off J Eur Soc Med Oncol. 2020;31 (11 ):1449–1461. doi:10.1016/j.annonc.2020.07.005 15. Bredbeck BC , Mubarak E , Zubieta DG , Management of the positive sentinel lymph node in the post-MSLT-II era. J Surg Oncol. Published online September 6, 2020. doi:10.1002/jso.26200 16. Farrow NE , Raman V , Williams TP , Nguyen KY , Tyler DS , Beasley GM . Adjuvant Therapy is Effective for Melanoma Patients with a Positive Sentinel Lymph Node Biopsy Who Forego Completion Lymphadenectomy. Ann Surg Oncol. Published online April 20, 2020. doi:10.1245/s10434-020-08478-7 17. Green LW . Closing the chasm between research and practice: evidence of and for change. Health Promot J Aust Off J Aust Assoc Health Promot Prof. 2014;25 (1 ):25–29. doi:10.1071/HE13101 18. Cochran AJ , Wen D-R , Huang R-R , Wang H-J , Elashoff R , Morton DL . Prediction of metastatic melanoma in nonsentinel nodes and clinical outcome based on the primary melanoma and the sentinel node. Mod Pathol Off J U S Can Acad Pathol Inc. 2004;17 (7 ):747–755. doi:10.1038/modpathol.3800117 19. Morton DL , Thompson JF , Cochran AJ , Sentinel-node biopsy or nodal observation in melanoma. N Engl J Med. 2006;355 (13 ):1307–1317. doi:10.1056/NEJMoa060992 17005948 20. Bamboat ZM , Konstantinidis IT , Kuk D , Ariyan CE , Brady MS , Coit DG . Observation after a positive sentinel lymph node biopsy in patients with melanoma. Ann Surg Oncol. 2014;21 (9 ):3117–3123. doi:10.1245/s10434-014-3758-7 24833100 21. Bertolli E , Franke V , Calsavara VF , Validation of a Nomogram for Non-sentinel Node Positivity in Melanoma Patients, and Its Clinical Implications: A Brazilian-Dutch Study. Ann Surg Oncol. 2019;26 (2 ):395–405. doi:10.1245/s10434-018-7038-9 30456681 22. Rossi CR , Mocellin S , Campana LG , Prediction of Non-sentinel Node Status in Patients with Melanoma and Positive Sentinel Node Biopsy: An Italian Melanoma Intergroup (IMI) Study. Ann Surg Oncol. 2018;25 (1 ):271–279. doi:10.1245/s10434-017-6143-5 29067603 23. Lee JH , Essner R , Torisu-Itakura H , Wanek L , Wang H , Morton DL . Factors predictive of tumor-positive nonsentinel lymph nodes after tumor-positive sentinel lymph node dissection for melanoma. J Clin Oncol Off J Am Soc Clin Oncol. 2004;22 (18 ):3677–3684. doi:10.1200/JCO.2004.01.012 24. MacDonald S , Siever J , Baliski C . Performance of models predicting residual lymph node disease in melanoma patients following sentinel lymph node biopsy. Am J Surg. 2020;219 (5 ):750–755. doi:10.1016/j.amjsurg.2020.02.059 32222274 25. Herb JN , Dunham LN , Ollila DW , Stitzenberg KB , Meyers MO . Use of Completion Lymph Node Dissection for Sentinel Lymph Node-Positive Melanoma. J Am Coll Surg. 2020;230 (4 ):515–524. doi:10.1016/j.jamcollsurg.2019.12.010 31954818 26. Faries MB , Thompson JF , Cochran A , The impact on morbidity and length of stay of early versus delayed complete lymphadenectomy in melanoma: results of the Multicenter Selective Lymphadenectomy Trial (I). Ann Surg Oncol. 2010;17 (12 ):3324–3329. doi:10.1245/s10434-010-1203-0 20614193 27. Hui J , E Burke , K Broman , S Marmo , E Jensen , T Tuttle , Zager J . Surgeon Decision-Making for Management of Positive Sentinel Lymph Nodes in the Post-Multicenter Selective Lymphadenectomy Trial II Era: A Survey Study. J Surg Oncol. In press. 28. Rauwerdink DJW , Molina G , Frederick DT , Sharova T , Carmichael H , Boland GM . Adjuvant Therapy Failure Patterns in the Modern Era of Melanoma Management. Ann Surg Oncol. Published online May 23, 2020. doi:10.1245/s10434-020-08631-2 29. Svedman FC , Pillas D , Taylor A , Kaur M , Linder R , Hansson J . Stage-specific survival and recurrence in patients with cutaneous malignant melanoma in Europe - a systematic review of the literature. Clin Epidemiol. 2016;8 :109–122. doi:10.2147/CLEP.S99021 27307765 30. Hodi FS , O’Day SJ , McDermott DF , Improved survival with ipilimumab in patients with metastatic melanoma. N Engl J Med. 2010;363 (8 ):711–723. doi:10.1056/NEJMoa1003466 20525992 31. Robert C , Grob JJ , Stroyakovskiy D , Five-Year Outcomes with Dabrafenib plus Trametinib in Metastatic Melanoma. N Engl J Med. 2019;381 (7 ):626–636. doi:10.1056/NEJMoa1904059 31166680 32. Robert C , Thomas L , Bondarenko I , Ipilimumab plus dacarbazine for previously untreated metastatic melanoma. N Engl J Med. 2011;364 (26 ):2517–2526. doi:10.1056/NEJMoa1104621 21639810 33. Applequist J , Ball JG . An Updated Analysis of Direct-to-Consumer Television Advertisements for Prescription Drugs. Ann Fam Med. 2018;16 (3 ):211–216. doi:10.1370/afm.2220 29760024 34. Mott N , Wang T , Miller J , Medical Maximizing-Minimizing Preferences in Relation to Low-Value Services for Older Women with Hormone Receptor-Positive Breast Cancer: A Qualitative Study. Ann Surg Oncol. 2021;28 (2 ):941–949. doi:10.1245/s10434-020-08924-6 32720038 35. Long GV , Flaherty KT , Stroyakovskiy D , Dabrafenib plus trametinib versus dabrafenib monotherapy in patients with metastatic BRAF V600E/K-mutant melanoma: long-term survival and safety analysis of a phase 3 study. Ann Oncol Off J Eur Soc Med Oncol. 2019;30 (11 ):1848. doi:10.1093/annonc/mdz221 36. Eggermont AMM , Chiarion-Sileni V , Grob J-J , Prolonged Survival in Stage III Melanoma with Ipilimumab Adjuvant Therapy. N Engl J Med. 2016;375 (19 ):1845–1855. doi:10.1056/NEJMoa1611299 27717298 37. Broman K , Hughes T , Dossett L , Active Surveillance of Melanoma Patients with Sentinel Node Metastasis: An International Multi-Institution Evaluation of Post-MSLT-2 Adoption and Early Outcomes. Cancer. In press. 38. Owen CN , Shoushtari AN , Chauhan D , Management of early melanoma recurrence despite adjuvant anti-PD-1 antibody therapy☆. Ann Oncol Off J Eur Soc Med Oncol. 2020;31 (8 ):1075–1082. doi:10.1016/j.annonc.2020.04.471 39. Howard DH , Soulos PR , Chagpar AB , Mougalian S , Killelea B , Gross CP . Contrary To Conventional Wisdom, Physicians Abandoned A Breast Cancer Treatment After A Trial Concluded It Was Ineffective. Health Aff Proj Hope. 2016;35 (7 ):1309–1315. doi:10.1377/hlthaff.2015.1490 40. Hennigs A , Köpke M , Feißt M , Which patients with sentinel node-positive breast cancer after breast conservation still receive completion axillary lymph node dissection in routine clinical practice? Breast Cancer Res Treat. 2019;173 (2 ):429–438. doi:10.1007/s10549-018-5009-2 30315437 41. Bilimoria KY , Balch CM , Wayne JD , Health care system and socioeconomic factors associated with variance in use of sentinel lymph node biopsy for melanoma in the United States. J Clin Oncol Off J Am Soc Clin Oncol. 2009;27 (11 ):1857–1863. doi:10.1200/JCO.2008.18.7567 42. Minami CA , Wayne JD , Yang AD , National Evaluation of Hospital Performance on the New Commission on Cancer Melanoma Quality Measures. Ann Surg Oncol. 2016;23 (11 ):3548–3557. doi:10.1245/s10434-016-5302-4 27278202 43. Tucker TC , Charlton ME , Schroeder MC , Improving the Quality of Cancer Care in Community Hospitals. Ann Surg Oncol. 2021;28 (2 ):632–638. doi:10.1245/s10434-020-08867-y 32712893 44. Shulman LN , Browner AE , Palis BE , Compliance with Cancer Quality Measures Over Time and Their Association with Survival Outcomes: The Commission on Cancer’s Experience with the Quality Measure Requiring at Least 12 Regional Lymph Nodes to be Removed and Analyzed with Colon Cancer Resections. Ann Surg Oncol. 2019;26 (6 ):1613–1621. doi:10.1245/s10434-019-07323-w 30927195 45. Matthews NH , Li W-Q , Qureshi AA , Weinstock MA , Cho E . Epidemiology of Melanoma. In: Ward WH , Farma JM , eds. Cutaneous Melanoma: Etiology and Therapy. Codon Publications; 2017. Accessed March 10, 2021. http://www.ncbi.nlm.nih.gov/books/NBK481862/

PMC010xxxxxx/PMC10111249.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0374755 4421 J Agric Food Chem J Agric Food Chem Journal of agricultural and food chemistry 0021-8561 1520-5118 35819336 10111249 10.1021/acs.jafc.2c01872 NIHMS1890005 Article Detection of Acrylamide in Foodstuffs by Nanobody-Based Immunoassays Liang Yifan 1† Zeng Yuyao 1† Luo Lin 1 Xu Zhenlin 12 Shen Yudong 1 Wang Hong 12* Hammock Bruce D 3 1 Guangdong Provincial Key Laboratory of Food Quality and Safety, College of Food Science, South China Agricultural University, Guangzhou 510642, China. 2 Guangdong Laboratory for Lingnan Modern Agriculture, Guangzhou 510642, China. 3 Department of Entomology and Nematology, UCD Comprehensive Cancer Center, University of California, Davis, California 95616, USA. † Equal contribution. * Corresponding author: Hong Wang, gzwhongd@163.com; Tel: +86-20-85283448; Fax: +86-20-85280270 8 4 2023 27 7 2022 12 7 2022 27 7 2023 70 29 91799186 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Acrylamide is toxicant aliphatic amide formed via the Maillard reaction between asparagines and reducing sugars during food heat processing. Herein, a specific nanobody termed Nb-7E against acrylamide derivative xanthyl acrylamide (XAA) was isolated from an immunized phage displayed library and confirmed to be able to detect acrylamide. Firstly, an ic-ELISA was established for acrylamide with the limit of detection (LOD) of 0.089 μg/mL and working range from 0.23 μg/mL to 5.6 μg/mL. Furthermore, an enhanced electrochemical immunoassay (ECIA) was developed based on the optimized reaction conditions. The LOD was low to 0.033 μg/mL, 3-fold improved than that of ic-ELISA and a wider linear detection range from 0.39 μg/mL to 50.0 μg/mL was achieved. The average recoveries ranged from 88.29% to 111.76% in spiked baked biscuits and potato crisps. Finally, the analytical performance of ECIA was validated by the standard ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS). Graphical Abstract acrylamide nanobody ELISA electrochemical immunoassay pmcIntroduction Acrylamide (AA, Mw 71.08 Da) is a vinylic hazardous chemical, which has inherently toxic properties, including neurotoxicity, reproductive toxicity, and carcinogenicity1. Therefore, it was listed by the International Agency for Research on Cancer as a probable human carcinogen in group 2A2. Since it was found and firstly reported in fired or oven-cooked carbohydrate-rich foods in 20023–4, food-borne AA contamination has been getting more and more public concern. The survey by Konings et al. showed that the mean amounts of AA in the cocktail snacks were the highest among the detected food samples and up to 1060 μg/kg, then gingerbread with 890 μg/kg and chips with 351 μg/kg5. In 2015, European Food Safety Authority (EFSA) regulated that the amount of AA in potato crisps should be below the indicative value of 1000 μg/kg6. However, the reference level for AA in potato crisps was monitored as 750 μg/kg by the European Union (EU)7. Therefore, given its toxicity and high pollution levels, a rapid, portable and highly sensitive detection method for AA would be helpful in monitoring the safety of foodstuffs more conveniently and making dietary exposure assessment with more extensive coverage. So far, some methods have been used to quantify AA in foodstuffs. Most of them suffer significantly from high instruments cost and complex operations, such as high-performance liquid chromatography (HPLC), gas chromatography (GC) and capillary electrophoresis (CE) separations8–10. Alternatively, immunoassays based on the specificity of antibodies to recognize target antigens are paid more attention for their excellent sensitivity, high-throughput screening and ease of combination with other detection formats, especially in detection of small molecule compounds in foodstuffs. Till now, several immunoassays have been reported for AA11–13. For example, Zhu et al. prepared a monoclonal antibody (mAb) against the derivative of AA (4-mercaptophenylacetic acid derivatized AA, AA-4-MPA) and developed an indirect competitive enzyme-linked immunosorbent assay (ic-ELISA) for AA determination14. Furthermore, an electrochemical immunosensor was established with a polyclonal antibody (pAb) specific for AA-4-MPA as recognition element, achieving a 100-fold improvement in sensitivity and decreasing analysis time in comparison to the original ic-ELISA15. Meanwhile, using 9-xanthydrol as derivative, Luo et al. also prepared the specific antibodies for 9-xanthydrol derivatized AA (XAA) and developed the fluorescence immunoassay for AA with the limit of detection low to 0.16 ng/mL16. However, the traditional antibodies have limitations including the unstable quality of pAbs caused by the batch-to-batch variation of immunized animals or the specificity loss of mAbs during thawing of hybridoma when it expressed additional functional variable regions17. So, a novel type of antibody with more excellent characteristics would be urgently demanded. Nanobodies (Nbs), termed variable heavy chain (VHH) domains, are derived from the heavy-chain-only antibody in Camelidae18. Compared with traditional pAbs or mAbs, Nbs have unique physiochemical advantages in terms of high robustness to harsh conditions, ease of production by prokaryotic expression system19–20. Due to its small size, Nb binds to a higher density on the surface of immunosensors, which can increase the signal-to-noise ratio, resulting in the enhanced sensitivity. For instance, Liu et al. described a nanobody-based ECIA for ultra-sensitive detection of AFB1, allowing two orders of magnitude improvement in LOD over mAb-based ECIA21. This suggests that Nbs might circumvent problems encountered with classical antibodies to detect small molecule contaminants, thus creating a novel detection system. In this study, a phage displayed library derived from an immunized Bactrian camel was constructed with the designed XAA-BSA as immunogen, then the specific Nbs were screened out and the characteristics were confirmed. Meanwhile, an ic-ELISA was established for AA detection based on the optimal Nb. To further improve the assay performance, an electrochemical immunoassay was developed under the optimized reaction conditions for detecting AA in potato crisps and biscuits samples (Scheme 1). Finally, the method was validated by the standard UPLC-MS/MS. Experimental methods Materials and reagents Acrylamide was supplied by Aladdin Chemical Technology Co., Ltd (Shanghai, China). Complete and incomplete Freund’s adjuvants, bovine serum albumin (BSA), ovalbumin (OVA) and keyhole limpet hemocyanin (KLH) were purchased from Sigma Aldrich (St. Louis, MO, USA). Xanthyl acrylamide derivatives (XAA), xanthyl methyl carbamate (XMC) and xanthyl ethyl carbamate (XEC)16, XAA-BSA as immunogen and XAA-OVA as coating antigen, and anti-XAA mAb were prepared previously in our lab22. The total RNA extraction kit was from Guangzhou Gbcbio Technologies Inc. (Guangzhou, China). The first strand cDNA synthesis kit was purchased from Takara (Dalian, China). The gel extraction and PCR purification kit were from Tiangen Biotech Co., Ltd. (Beijing, China). Helper phage M13KO7, restriction enzyme, T4 DNA ligase and others, were obtained from New England Biolabs (MA, USA). 9-xanthydrol, other organic solvents and chemicals were of analytical grade from Sigma (St. Louis, USA) and Thermo Fisher Scientific (Thermo, USA). Construction and identification of phage displayed nanobody library A three-year-old male Bactrian camel was immunized subcutaneously with 500 μg of XAA-BSA and Freund’s adjuvant mixture biweekly. The nanobody phage displayed library was constructed by using the method described previously18, 23. Briefly, after sixth immunization, fresh blood was collected and used for lymphocyte isolation, followed by RNA extraction, cDNA synthesis and VHH gene amplification by two-nested PCR. The purified PCR product was ligated into pComb3xss phagemid vector digested by restriction enzyme Sfi I and then transformed into E.coli TG1 competent cells. All cells plated on LBA agar (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, supplemented with 100 μg/L ampicillin and 15 g/L agar) were collected and rescued by helper phage M13KO7 to prepare an immunized VHH phage displayed library. The capacity and diversity were evaluated by sequence analysis of randomly 10 clones. Selection and characterization of nanobody against XAA The procedure of phage biopanning was performed as described before with some modifications23. Namely, three wells of microplate were coated overnight with 100 μL of XAA-OVA (10 μg/mL) in PBS (0.01 M), and additional one well with 100 μL of mixture of 2% KLH, OVA and BSA solution. After washing 5 times with PBST (PBS with 0.1% Tween-20), all wells were blocked with 1% fish gelatin at 37 °C for 2 h. Then, 100 μL/well of nanobody phage library was added into well coated with mixture of KLH, OVA and BSA at 37 °C for 1 h. The unbound phages were dispensed into three antigen-coated wells (100 μL/well). After incubating for 1 h, the three wells were washed 5 times with 0.1% PBST and 15 times with PBS respectively. The bound phages were eluted with 100 μL/well of 0.2 M Gly-HCl solution (pH 2.2) at 37 °C for 10 min and then immediately neutralized with 12 μL/well of 1 M Tris-HCl (pH 9.1). The eluent (10 μL) in each round were diluted to determine the output titer, and other was amplified for the next round of biopanning. In the following second, third and fourth round, competitive elution method was carried out with different concentration of XAA (1000, 500 and 100 ng/mL). The concentration of XAA-OVA was decreased to 2, 0.4, 0.08 μg/mL, and the content of Tween-20 in PBST increased from 0.2 to 0.3, and 0.4% respectively. To decrease the nonspecific binding, three kinds of blocking buffers including 3% skimmed milk, 1% fish gelatin and 3% skimmed milk were utilized. After the fourth-round panning was finished, several clones were randomly picked up from the output plates to test their binding ability with XAA by ic-ELISA23. The positive clones were sequenced for further analysis. Expression, purification and identification of anti-XAA Nbs The vector pComb3xss-Nb was transformed into competent cells E.coli BL21 (DE3). After sequencing confirmation, individual clone was selected and cultivated in LBA medium (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, supplemented with 100 μg/L ampicillin) with shaking at 250 rpm overnight. A 6 mL-aliquot of overnight culture was inoculated to 600 mL of LBA medium and shaken at 37 °C until OD600 value reached around 0.6, followed by the addition of 1 mM of isopropyl β-D-thiogalactopyranoside (IPTG) for production of nanobody. The supernatant was collected by Sorvall Lynx 4000 centrifuges (Thermo Fisher Scientific, USA), and then purified on a 1-mL Ni-NTA resin column. The purified Nb was dialyzed in PBS buffer and identified by SDS-PAGE and western blotting according to the standard protocols24, and the concentration was determined by using NanoDrop 2000C system (Thermo Fisher Scientific, USA). Procedure of Nb-based ic-ELISA The ic-ELISA protocol was performed as described before with minor modifications23. As shown in more detail below, 100 μL per well of XAA-OVA in PBS was used to coat a 96-well microplate overnight at 37 °C. The plate was washed twice with PBST (0.01 M PBS containing 0.05% Tween-20) and then blocked with 2% skimmed milk in PBS at 37 °C for 2 h. After discarding the blocking solution, the plate was dried at 37 °C for 30 min and stored at 4 °C until use. A serial of AA standards with different concentrations (0–50 μg/mL in PBS) were prepared via the following derivatization reaction. First, AA standard (0.6 mL) was mixed with 0.4 mL of 9-xanthydrol (4 mg/mL), then 0.1 mL of HCl (0.5 M) was added to trigger the reaction. After 30 min, the reaction was finished by adding 0.1 mL of NaOH (0.5 M). The mixture solution was used as competitor.. Each well was added with 50 μL of competitor and 50 μL of Nb solution at optimal concentration. After incubated at 37 °C for 30 min and washed five times with PBST, the diluted HRP-goat anti-VHH IgG (Abcam, Guangzhou) was added to the wells with 30 min incubation at 37 °C, followed by a washing step with PBST (five times). Finally, 100 μL/well of TMBZ solution was added and the color was stopped after 10 min incubation by addition of 50 μL of 10% H2SO4. The absorbance value at 450 nm was measured using a Multiskan MK3 microplate reader (Themo Labsystems, USA). Furthermore, the standard curve was obtained with a four-parameter fitting module of Origin 9.0. The limit of detection (LOD) was determined as AA concentration at 10% of maximum value calculated from standard curve, and the detectable concentration range was defined as the AA concentration that inhibited 20–80% of maximum value. The rate of cross reactivity was calculated as follows23: CR (%) = IC50 (XAA, μg/mL) / IC50 (cross-reactant, μg/mL). Construction and development of nanobody-based ECIA for AA Prior to modification, a portable commercial screen-printed carbon electrode (SPCE, Zensor R&D, Taiwan) was electrochemically activated by performing a 15-segment cyclic voltammetry scan in 0.2 M KNO3 with a potential range from −0.1 to +0.1 V at a scan rate of 50 mV/s, and then the electrode was washed with ultrapure water and dried. Prussian blue-chitosan-nanoparticle (PB-CS-NP) film was electrodeposited on SPCE electrode in a fresh solution including 2.5 mM FeCl3, 2.5 mM K3[Fe (CN)6], 1 mM KCl, 1 mM HCl and 0.05% chitosan with potential range of −1 to +1 at a scan rate of 50 mV/s. After rinsed with ultrapure water and dried, 5 μL of XAA-OVA was dispersed on the working surface of the PB-CS-NP/SPCE and incubated at 4 °C overnight, and the unbound was removed by washing with 0.01 M PBS. Subsequently, 5% skimmed milk (5 μL) was coated on the electrode surface at 37 °C for 1 h. Finally, the immunosensor was washed thoroughly with 0.01 M PBS and stored at 4 °C until used. The diluted Nb solution together mixed with different concentration of competitor was spotted onto the modified SPCE surface at 37 °C for 30 min. After the washing steps, 5 μL of HRP-goat anti-VHH IgG was added for another 30 min reaction. As for electrochemical detection, the modified immunosensor was immersed in 0.01 M PBS containing 1 mM HQ and 6% H2O2 and performed on a CHI660D electrochemical work station (CH Instruments, Shanghai Chenhua Instrument Corporation, Shanghai, China). The change in the electrochemical cathodic current before and after the addition of H2O2 was used as the signal response (ΔI), and then we could obtain the standard curve between the concentrations of AA and ΔI value. Consequently, the limit of detection (LOD) was defined as the equation: LOD = Y + 3SD, where Y is the mean signal of the blank measurements, and SD is the standard deviation of blank measurements, and a value of 3 refers to the equation constant. The lower and upper limits of quantitative concentration were defined as linear range25. Analysis of spiked samples based on ECIA Biscuits and potato crisps samples were purchased from a local supermarket. The samples were first analyzed to obtain the background level of AA and then used to perform the recovery test. Briefly, 10 g of sample spiked with different concentration of AA was mixed with 40 mL of methanol and homogenized, followed by the defatting step using 5 mL of n-hexane and centrifugation at 4500 rpm for 10 min. The supernatants were evaporated to dryness and dissolved in 5 mL of distill water. The following derivatizaiton as above described and the mixture solution was used for further analysis. Assay validation by UPLC-MS/MS The nanobody-based ECIA was validated by UPLC-MS/MS operated at the Guangzhou Institute for Food Control, China. The conditions were as follows: the C18 column (150 mm × 4.6 mm, 5 μm) was used for chromatographic separation at 40 °C. The mobile phases were 0.3% acetic acid and acetonitrile at the ratio of 95: 5 (v/v). The flow rate was 0.5 mL/min, and the injection volume of each sample was 20 μL. An AB TRIPLE QUAD 4500 Mass Spectrometer (AB, USA) was operated in ESI mode set as positive ionization multiple reactions monitoring (MRM) scanning mode. Other ionization source parameters were also optimized and set as follows: source cone temperature: 450 °C, drying gas flow: 6 L/min, detection voltage: 5.5 kV, nebulizer gas: 40 psi. The parent and daughter ion of AA were m/z 72 as well as m/z 55 and 44, respectively. Results and discussion Library construction, biopanning and expression For the molecular weight of AA is too low, until now, the immunoassays for AA mostly utilized the antibodies specific for its different derivatives12–13. In the previous study of our group, 9-xanthydrol was demonstrated to be efficient to convert AA into XAA with simple procedures and the prepared anti-XAA pAb showed the high affinity to XAA16. Therefore, XAA-BSA (Figure S1) was still used as immunogen in this work. After the sixth immunization to the Bactrian camel, the inhibition rate of the serum for XAA was more than 85%, so the peripheral blood lymphocyte was collected and cDNA was extracted then to construct the phage displayed library with the reported methods as reference18, 23. Considering that an immunized antibody library might possess higher possibility to contain the target Nbs, the constructed library with a little low capacity of 1.0×105 cfu/mL was still used to select the target Nbs. Actually, after four rounds of panning, 10 positive clones were obtained and exhibited binding activity to coating antigen with the inhibition rates for XAA ranged from 46.00% to 77.53% (Figure 1A). By the sequence alignments analysis, all Nbs clones have the same length but are slightly different in framework region 1 (FR1) and framework region 4 (FR4) (Figure 1B). Therefore, they were comfirmed as the same group of Nbs and Nb-7E with the highest inhibition rate was chosen for further analysis. After expressed in E.coli BL21 (DE3) host strains, the supernatant containing Nb-7E was purified with Ni-NTA affinity columns, and then identified by SDS-PAGE and western blotting. As shown in Figure 1C, the size of Nb-7E was approximately 18 kDa, which is consistent with the theoretical molecular weight of Nb and the yield was about 7.45 mg/L with the purity higher than 90%. Stability of anti-XAA Nb-7E Generally, Nbs possess great thermostability and the tolerance to organic solvent. Herein, the stability of the obtained Nb-7E was confirmed with the anti-XAA mAb as control. It was found that Nb-7E was able to bind to antigen at temperature as high as 95 °C, while mAb lost its activity very quickly after 5 min incubation at 80 °C. Even more, Nb-7E retained about 100% of binding activity after incubation for 1 h at 85 °C, however, mAb could only endure 10 min (Figure 2A–B). Normally, the additional disulfide bonds formed by cysteine (Cys) residues existed in Nbs were thought to contribute to reversibly refold and retain binding activity when encountering heat-induced denaturation26. According to the sequence of Nb-7E, four Cys might form two disulfide bonds, resulting in the robustness to harsh conditions. Meanwhile, methanol and acetonitrile, two kinds of organic solvents commonly used in pretreatment of samples were chosen to test the tolerance of Nb-7E. As shown in Figure 2C, with increase of methanol concentration, both Nb-7E and mAb gradually lost the activities. When the concentration of methanol was up to 70%, Nb-7E could still maintain over 70% of binding activity, whereas mAb lost almost all bioactivity. In the contrary, mAb as a whole had superior performance than did Nb-7E in tolerance for acetonitrile (Figure 2D). It has been reported that some amino acids including Met4, Pro43 and Gly66 of VL or Ile2, Leu37, Ile48 and Gly49 of VH might improve the antibodies’ tolerance to acetonitrile27. By alignment sequence analysis, we found the mAb just exactly has Met4, Pro44 and Gly66 in VL, Gly49 in VH (Table S2) yet Nb-7E contains no these residues at the same sites (Val2/37/48 and Ala49), therefore exhibited poorer resistance to acetonitrile. Establishment of ic-ELISA based on Nb-7E Considering the influence of assay parameters, several factors were optimized to develop an ic-ELISA with Nb-7E for AA. After the checkerboard procedure, 2 μg/mL of XAA-OVA and 3.4 μg/mL of Nb-7E were selected as the optimal concentrations. Other assay parameters including an incubation time of 30 min and a dilution of 1: 2500 of HRP-goat anti-VHH IgG as well as dilution buffer with 0.01 M PBS were determined (Figure S2). Under the optimal reaction conditions, an ic-ELISA was established and the limit of detection (LOD) for AA was 0.089 μg/mL with working range from 0.23 to 5.6 μg/mL (Figure 2E), which could meet the regulation standard by the EU. Compared to the previous reported ic-ELISA based on the other derivative such AA-4-MPA14, the established ic-ELISA in this work was more rapid and simpler because the derivatization by 9-xanthydrol was more gentle under the room temperature (25 °C) with less time (＜30 min). Additionally, the cross-reactivity (CR) test showed that except for the analog XMC, there was negligible CR to XEC and 9-xanthydrol (Figure 2F). By homology modeling of Nb-7E and docking with XAA or its analogs, it was found that Nb-7E formed a pocket to recognize only XAA and XMC but not XEC or 9-xanthydrol. The interaction forces between Nb-7E and XAA or XMC were both mainly induced by the formed three hydrogen bonds (Figure S3), however, further bonds distance between Nb-7E and XMC might lead to weaker interaction therefore 15.7% of CR. Enhanced Nb-7E-based ECIA for AA To further improve the sensitivity of the assay, the enhanced ECIA based on Nb-7E was developed. Several analytical parameters were confirmed including XAA-OVA concentration, Nb-7E dilution and H2O2 concentration. The higher current response, the higher sensitivity of assay. As shown in Figure 3A, the current response increased gradually with the increment of XAA-OVA concentration, and the maximum response at a concentration of 10 μg/mL was observed. Nevertheless, the higher concentrations of 10 μg/mL prompted a decline in the current. This may be because steric hindrance of the film obstructs the Nb-7E to reach the binding sites and the moving of electrons. Therefore, 10 μg/mL of XAA-OVA was chosen for subsequent study. Similarly, the other optimized conditions included a dilution of 1: 40 of Nb-7E (Figure 3B), and a concentration of 6% H2O2 (Figure 3C). Under the optimum conditions, ΔI values of different concentration of AA were recorded and a linear calibration curve was obtained in the working range from 0.39 to 50 μg/mL represented by ΔI (μA) = (4.86 ± 0.01) - (1.4 ± 0.01) × lg CAA (μg/mL) with a correlation coefficient of 0.99 (Figure 3D–E). The LOD value of the ECIA was 0.033 μg/mL, 3-fold lower than that of ic-ELISA, and a wider linear range was also observed (Table 1). Besides, the ECIA based on Nb-7E has the advantages over that on traditional mAb or pAb in antibodies production, cost and physiochemical stability. Analysis of spiked samples by ECIA and UPLC-MS/MS Biscuits (background level of AA was 289.50 ng/g) and potato crisps (background level of AA was 74.03 ng/g) were analyzed after spiking with different concentration of AA (25, 50, and 75 ng/g). Results showed that average recoveries in spiked samples for ECIA were 88.29% to 111.76% and coefficient of variations (CVs) were all less than 5% (Table 2). The detection results were also validated by UPLC-MS/MS and the correlation coefficient between ECIA and UPLC-MS/MS was up to 0.997 (Figure 4). It means the proposed ECIA based on Nb-7E had a good accuracy for quantitative analysis of AA in foodstuffs. In this study, specific Nbs against acrylamide derivative xanthyl acrylamide (XAA) were for the first time isolated from a camel immunized nanobody library. Based on the specific Nb-7E, an ic-ELISA for AA was established, furthermore, an enhanced ECIA biosensor with a wider linear range and improved sensitivity was constructed. The LODs of these two methods were far below 750 μg/kg, the regulation standard by the European Union for AA. The accuracy of the proposed ECIA was also testified by the standard UPLC-MS/MS, which suggested the generation of Nbs in this work can be used as a novel reagent in immunoassays and the established methods were proved to be effective and prospective for AA detection in foodstuffs. Supplementary Material Supporting Information Funding This work was supported by Lingnan Modern Agricultural Science And Technology Guangdong Laboratory Independent Scientific Research Project (NZ2021032), Natural Science Foundation of China (31972157), National Key R&D Program of China (2019YFE0116600), Key-Area Research and Development Program of Guangdong Province (2019B020211002), Guangdong Province Universities and Colleges Pearl River Scholar Funded Scheme (2017), Guangdong Provincial Key Laboratory of Food Quality and Safety (2020B1212060059). Supporting information Synthetic route to complete antigens, parameter optimization of ic-ELISA, docking of Nb-7E and XAA/XMC, primers used to amplify VHH gene and sequence of variable regions of XAA-mAb. Figure 1 (A) The positive clones identified by ic-ELISA through biopanning. (B) Sequence alignment of the selected Nbs, the framework regions (FRs) and complementary termination regions (CDRs) were determined by the IMGT database. (C) SDS-PAGE and western blotting analysis of purified Nb-7E. Figure 2 (A) The retention of binding activity of Nb-7E for incubation at different temperatures for 5 min. (B) The thermostability of Nb-7E after being incubated at 85 °C for 10, 20, 30, 40, 50 and 60 min. Activity analysis of Nbs incubated with 10%−70% of methanol (C) and acetonitrile (D). (E) A standard curve of ic-ELISA based on Nb-7E. (F) Cross-reactivity determination of related analogs by ic-ELISA (n = 3). Figure 3 Optimization and establishment of Nb-7E based ECIA. (A) XAA-OVA concentration, (B) Nb-7E dilution and (C) H2O2 concentration. (D) Electrocatalytic current responses of Nb-based ECIA for the detection of different concentrations of AA: 0.39 μg/mL (a), 0.78 μg/mL (b), 1.56 μg/mL (c), 3.125 μg/mL (d), 12.5 μg/mL (e), 25 μg/mL (f), 50 μg/mL (g). (E) Calibration curve of Nb-based ECIA between the different concentrations of AA and ΔI values (n = 5). Figure 4 Correlations analysis between Nb-7E based ECIA and UPLC-MS/MS for AA in spiked samples. Scheme 1 Workflow of selection of nti-XAA Nb from an immunized Bactrian camel and development of ic-ELISA and ECIA based on anti-XAA Nb. Table 1 Analytical characteristic of the ic-ELISA and ECIA based on Nb-7E for AA. Parameters ic-ELISA ECIA Coating antigen (ng) 200/well 50 Competition time (min) 30 30 Dilution of Nb-7E 1: 250 1: 40 LOD (μg/mL) 0.089 0.033 Linear range (μg/mL) 0.23–5.6 0.39–50 Assay time (min) 70 60 Table 2 Recovery analysis of AA in spiked samples by Nb-based ECIA (n = 3). Samples Background level (ng/g) Spiked level (ng/g) Founda ± SDb (ng/g) Recoveryc ± CVd (%) Biscuits 289.50 25 22.07 ± 0.69 88.29 ± 3.13 50 55.88 ±1.11 111.76 ± 1.99 75 75.66 ± 1.15 100.88 ± 1.52 Potato crisps 74.03 25 25.91 ± 0.11 103. 64 ± 0.42 50 50.79 ± 2.43 101.58 ± 4.78 75 74.66 ± 3.48 99.55 ± 4.66 a Found value is calculated using measurement value to subtract background level b SD: standard deviation c Recovery = Found / Spiked level d CV: coefficient of variation = SD / Found The authors declare no competing financial interest. All procedures involving camels were performed in accordance with the relevant protective and administrative guidelines for laboratory animals of China. References (1) Zhang Y ; Ren YP ; Zhang Y New Research Developments on Acrylamide: Analytical Chemistry, Formation Mechanism, and Mitigation Recipes. Chemical Reviews. 2009, 109 , 4375–4397.19663380 (2) International Agency for Research on Cancer, Lyon, France, 1994, pp 270. (3) Authority EFS . Results on acrylamide levels in food from monitoring year 2008. EFSA Journal. 2010, 8. (4) Svensson K ; Abramsson L ; Becker W ; Glynn A ; Hellenas KE ; Lind Y ; Rosen J Dietary intake of acrylamide in Sweden. Food and Chemical Toxicology. 2003, 41 , 1581–1586.12963011 (5) Konings EJM ; Baars AJ ; van Klaveren JD ; Spanjer MC ; Rensen PM ; Hiemstra M ; van Kooij JA ; Peters PWJ Acrylamide exposure from foods of the Dutch population and an assessment of the consequent risks. Food and Chemical Toxicology. 2003, 41 , 1569–1579.12963010 (6) EFSA CONTAM Panel. Scientific opinion on acrylamide in food. EFSA Journal. 2015, 13 , 4104. (7) Rayappa MK ; Viswanathan PA ; Rattu G ; Krishna PM Nanomaterials Enabled and Bio/Chemical Analytical Sensors for Acrylamide Detection in Thermally Processed Foods: Advances and Outlook. Journal of Agricultural and Food Chemistry. 2021, 69 , 4578–4603.33851531 (8) Hoenicke K ; Gatermann R ; Harder W ; Hartig L Analysis of acrylamide in different foodstuffs using liquid chromatography−tandem mass spectrometry and gas chromatography–tandem mass spectrometry. Analytica Chimica Acta. 2004, 520 , 207–215. (9) Zhang Y ; Dong Y ; Ren YP ; Zhang Y Rapid determination of acrylamide contaminant in conventional fried foods by gas chromatography with electron capture detector. Journal of Chromatography A. 2004, 1116 , 209–216. (10) Elisabet B ; Oscar N ; Luis P ; Maria TG Analysis of acrylamide in food products by in-line preconcentration capillary zone electrophoresis. Journal of Chromatography A. 2006, 1129 , 129–134.16843477 (11) Zhou S ; Zhang C ; Wang D ; Zhao MP Antigen synthetic strategy and immunoassay development for detection of acrylamide in foods. Analyst. 2008, 133 , 903–909.18575643 (12) Quan Y ; Chen ML ; Zhan YH ; Zhang GH Development of an Enhanced Chemiluminescence ELISA for the Rapid Detection of Acrylamide in Food Products. Journal of Agricultural and Food Chemistry. 2011, 59 , 6895–6899.21639145 (13) Wu J ; Shen YD ; Lei HT ; Sun YM ; Yang JY ; Xiao ZL ; Wang H ; Xu ZL Hapten Synthesis and Development of a Competitive Indirect Enzyme-Linked Immunosorbent Assay for Acrylamide in Food Samples. Journal of Agricultural and Food Chemistry. 2014, 62 , 7078–7084.24998485 (14) Zhu YT ; Song SS ; Liu LQ ; Kuang H ; Xu CL An indirect competitive enzyme-linked immunosorbent assay for acrylamide detection based on a monoclonal antibody. Food and Agricultural Immunology. 2016, 27 , 796–805. (15) Wu MF ; Wang Y ; Li S ; Dong XX ; Yang JY ; Shen YD ; Wang H ; Sun YM ; Lei HT ; Xu ZL Ultrasensitive immunosensor for acrylamide based on chitosan/SnO2-SiC hollow sphere nanochains/gold nanomaterial as signal amplification. Analytica Chimica Acta. 2019, 1049 , 188–195.30612650 (16) Moraes JZ ; Hamaguchi B ; Braggion C ; Speciale ER ; Aguiar RB Hybridoma technology: is it still useful? Current Research in Immunology. 2021, 2 , 32–40.35492397 (17) Zhang YQ ; Xu ZL ; Wang F ; Cai J ; Dong JX ; Zhang JR ; Si R ; Wang CL ; Wang Y ; Shen YD ; Sun YM ; Wang H Isolation of Bactrian Camel Single Domain Antibody for Parathion and Development of One-Step dc-FEIA Method Using VHH-Alkaline Phosphatase Fusion Protein. Analytical Chemistry. 2018, 90 , 12886–12892.30256086 (18) van der Linden RH ; Frenken LG ; de Geus B ; Harmsen MM ; Ruuls RC ; Stok W ; de Ron L ; Wilson S ; Davis P ; Verrips CT Comparison of Physical Chemical Properties of Llama VHH Antibody Fragments and Mouse Monoclonal Antibodies. Biochimica et Biophysica Acta. 1999, 1431 , 37–46.10209277 (19) Liu Y ; Huang H Expression of Single-Domain Antibody in Different Systems. Applied Microbiology and Biotechnology. 2018, 102 , 539–551.29177623 (20) Liu X ; Wen YP ; Wang WJ ; Zhao ZT ; Han Y ; Tang KJ ; Wang D Nanobody-based electrochemical competitive immunosensor for the detection of AFB1 through AFB1-HCR as signal amplifier. Microchimica Acta. 2020, 187 , 352.32462392 (21) Luo L ; Lei HT ; Yang JY ; Liu GL ; Sun YM ; Bai WD ; Wang H ; Shen YD ; Chen S ; Xu ZL Development of an indirect ELISA for the determination of ethyl carbamate in Chinese rice wine. Analytica Chimica Acta. 2017, 950 , 162–169.27916121 (22) Luo L ; Jia BZ ; Wei XQ ; Xiao ZL ; Wang H ; Sun YM ; Shen YD ; Lei HT ; Xu ZL Development of an inner filter effect-based fluorescence immunoassay for the detection of acrylamide using 9-xanthydrol derivatization. Sensors and Actuators B: Chemical. 2021, 332 , 129561. (23) Wang F ; Li ZF ; Yang YY ; Wan DB ; Vasylieva N ; Zhang YQ ; Cai J ; Wang H ; Shen YD ; Xu ZL ; Hammock BD Chemiluminescent Enzyme Immunoassay and Bioluminescent Enzyme Immunoassay for Tenuazonic Acid Mycotoxin by Exploitation of Nanobody and Nanobody-Nanoluciferase Fusion. Analytical Chemistry. 2020, 92 , 11935–11942.32702970 (24) Liang YF ; Wang Y ; Wang F ; Li JD ; Wang CL ; Dong JH ; Ueda H ; Xiao ZL ; Shen YD ; Xu ZL ; Wang H An enhanced open sandwich immunoassay by molecular evolution for noncompetitive detection of Alternaria mycotoxin tenuazonic acid. Food Chemistry. 2021, 361 , 130103.34082388 (25) Dong XX ; Yang JY ; Luo L ; Zhang YF ; Mao CB ; Sun YM ; Lei HT ; Shen YD ; Beier RC ; Xu ZL Portable amperometric immunosensor for histamine detection using Prussian blue-chitosan-gold nanoparticle nanocomposite films. Biosensors and Bioelectronics. 2017, 98 , 305–309.28697442 (26) Kunz P ; Zinner K ; Mücke N ; Bartoschik T ; Muyldermans S ; Hoheisel JD The structural basis of nanobody unfolding reversibility and thermoresistance. Scientific Reports. 2018, 8 , 7934.29784954 (27) Li CL ; Yang HJ ; Yu WB ; Yu XZ ; Wen K ; Shen JZ ; Wang ZH Engineering of Organic Solvent-Tolerant Antibody to Sulfonamides by CDR Grafting for Analytical Purposes. Analytical Chemistry. 2021, 93 , 6008–6012.33728902

PMC010xxxxxx/PMC10270648.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101299742 Mucosal Immunol Mucosal Immunol Mucosal immunology 1933-0219 1935-3456 36623588 10270648 10.1016/j.mucimm.2023.01.001 EMS181071 Article CD200R1 promotes IL-17 production by ILC3s, by enhancing STAT3 activation Linley Holly 12 Ogden Alice 12 Jaigirdar Shafqat 12 Buckingham Lucy 12 Cox Joshua 12 Priestley Megan 12 Saunders Amy 12* 1 Manchester Collaborative Centre for Inflammation Research 2 Lydia Becker Institute of Immunology and Inflammation, Division of Infection, Immunity and Respiratory Medicine, School of Biological Science, Faculty of Biology, Medicine and Health, University of Manchester, Manchester Academic Health Science Centre, Manchester M13 9PL, UK * corresponding author: Dr Amy Saunders, University of Manchester, 2nd Floor CTF Building, 46 Grafton Street, Manchester M13 9NT, UK. Tel: +44 161 2751690. amy.saunders@manchester.ac.uk 01 4 2023 07 1 2023 18 7 2023 26 7 2023 16 2 167179 This file is available to download for the purposes of text mining, consistent with the principles of UK copyright law. Psoriasis is a common chronic inflammatory skin disease with no cure. It is driven by the IL-23/IL-17A axis and Th17 cells but, recently group 3 innate lymphoid cells (ILC3s) have also been implicated. Despite being the focus of much research, factors regulating the activity of ILC3s remain incompletely understood. Immune regulatory pathways are particularly important at barrier sites such as the skin, gut and lung, which are exposed to environmental substances and microbes. CD200R1 is an immune regulatory cell surface receptor which inhibits proinflammatory cytokine production in myeloid cells. CD200R1 is also highly expressed on ILCs, where its function remains largely unexplored. We previously observed reduced CD200R1 signalling in psoriasis skin, suggesting that dysregulation may promote disease. Here we show that contrary to this, psoriasis models are less severe in CD200R1-deficient mice due to reduced IL-17 production. Here we uncover a key cell-intrinsic role for CD200R1 in promoting IL-23-driven IL-17A production by ILC3s, by promoting STAT3 activation. Therefore, contrary to its inhibitory role in myeloid cells, CD200R1 is required on ILC3 to promote IL-23-stimulated STAT3 activation, triggering optimal IL-17 production. pmcIntroduction Barrier sites such as the skin, are enriched for immune cells that can rapidly respond to stimulation by producing large quantities of proinflammatory cytokines. Innate lymphoid cells (ILCs) are one such cell type, with group 3 ILCs (ILC3s) responding to IL-23 and IL-1β by producing IL-17A and IL-22 (1). These cytokines cause changes in epidermal cells including hyperproliferation, dysregulated maturation, and epidermal production of cytokines, chemokines and antimicrobial peptides, which are critical for protection against extracellular bacterial and fungal infection (2, 3). However, the IL-23/IL-17 axis can also drive psoriasis, a common, incurable chronic inflammatory skin disease, as demonstrated by the efficacy of biologics targeting this pathway (4). ILC3s are also implicated in psoriasis as they are a potent source of IL-17 and are increased in lesions (5, 6). Therefore, it is crucial to understand how ILC3 activity is regulated to prevent inappropriate activation. Immunosuppressive pathways are key to maintaining immune homeostasis and are particularly important in barrier tissues where there is constant exposure to environmental stimuli and commensal microbes. A key immunosuppressive pathway in myeloid cells is CD200R1 signalling, which has been shown to suppress proinflammatory cytokine production, reducing immune responses against self-antigens (7–10) allergens (11), infectious agents (12) and cancer (13). CD200R1 is a cell surface receptor which on engagement with its ligand, CD200, transmits a signalling cascade inhibiting MAP kinases and suppressing myeloid cell activation. CD200R1 is expressed on many immune cell types, and in addition to myeloid cells, is particularly highly expressed on ILC (14–17), however its function on these cells remains underexplored. We have previously shown that the ligand, CD200 is reduced in non-lesional psoriatic skin (17) and we therefore hypothesized that in healthy skin, CD200R1 signalling prevents inappropriate inflammation, but in psoriasis this is reduced, contributing to psoriasis susceptibility. To examine this hypothesis, here we compare the severity of inflammation in psoriasis models in WT and CD200R1-deficient mice. This shows that surprisingly, global CD200R1 deficiency reduces the severity of psoriasis-like skin inflammation due to reduced IL-17A production. CD200R1 is highly expressed by ILCs and here we show that CD200R1-deficient ILC3s are less able to produce IL-17A. This is associated with a vastly different transcriptomic profile despite the surface phenotype being largely unchanged. We also demonstrate that the requirement for CD200R1 in ILC3 is cell intrinsic. To understand why CD200R1KO ILC3s are unable to produce IL-17A optimally, signalling downstream of IL-23 stimulation was investigated, demonstrating that CD200R1 is required for optimal STAT3 activation in response to IL-23. Therefore, we describe a critical role for CD200R1 in promoting IL-17 production by ILC3 via enhancing IL-23-driven STAT3 activation. Results Global CD200R1 deficiency impairs IL-17 production in psoriasis models As we (17) and others (18) have demonstrated reduced CD200 levels in psoriatic patient samples, and CD200R1 signalling has been shown to dampen immune responses (7–13), we hypothesized that the absence of CD200R1 would enhance inflammation in psoriasis models. Contrary to this, global CD200R1KO deficiency protected mice from skin thickening in a model induced by repeated intradermal injection with IL-23 (19) (Figure 1A and B). Total immune cells (CD45+) were increased in IL-23 treated WT mouse skin, but this increase was inhibited in CD200R1KO mice (Figure 1C). To determine how CD200R1 promotes inflammation, skin IL-17A levels were measured, as this is a critical cytokine driving inflammation. ELISA and flow cytometry demonstrated reduced IL-17A overall in skin, and reduced production by both ILCs and CD3low (dermal) γδ T cells in CD200R1KO mice (Figure 1D-F and gating shown in Figure 2C). To investigate this further, a more complex psoriasis model was used in which Aldara cream, (containing the TLR7-agonist, imiquimod, and inflammasome activating isosteric acid) induces inflammation (20). After 4 days of treatment skin thickening was similar in WT and CD200R1KO mice, but redness and scaling were slightly reduced in the absence of CD200R1 (Figure 1G-I). The absence of CD200R1 reduced the accumulation of immune cells, in particular neutrophils, within skin (Fig 1J and K) and reduced the production of IL-17A by ILCs and γδ T cells within skin and LN (Figures 1L and M). IL-17 is critical to combatting fungal infections therefore, to examine if CD200R1KO mice also have altered fungal immune responses in skin, a C. albicans infection model was performed which showed CD200R1KO mice produce less IL-17 in response to fungal skin infection (Figure S1). CD200R1 is highly expressed by ILCs Given the reduction in IL-17A production in psoriasis-like skin inflammation models, the expression of CD200R1 on skin cells was examined. As anticipated from the literature, CD200R1 was expressed on skin macrophage and dendritic cell subsets (gated as described previously (21)) but was absent from non-immune cells and dendritic epidermal T cells (DETC) (Figure 2A-D). CD200R1 was also expressed on a proportion of dermal γδ T cells and on a larger proportion of skin ILCs (Figure 2D) as previously described (17). On splitting the ILCs into subsets based on their expression of lineage defining transcription factors, CD200R1 was shown to be expressed most highly on ILC2 but, was also expressed on ILC3s and GATA3- RORγt- ILCs (which contain ILC1 as well as ILCs which have lost expression of lineage defining transcription factors (22)) in skin (Figure 2E). Although dermal γδ T cells are thought to be the dominant drivers of inflammation in mouse models, via their production of IL-17A, ILCs also produce IL-17A in these models and, have been shown to be important contributors to inflammation (23). Given this, the relatively low level of expression of CD200R1 on dermal γδ T cells, the higher expression on ILCs, and the evidence that ILC3 contribute to human psoriasis (5, 6), the effect of CD200R1 on ILCs was examined. CD200R1 promotes IL-17A production by ILCs To determine if CD200R1 regulates IL-17 production by ILCs in a more direct manner, outside of the context of skin inflammation, dorsal skin cells were isolated and stimulated with IL-23, or PMA and Ionomycin, demonstrating that CD200R1-deficient ILCs are less able to produce IL-17A than WT, regardless of the stimulatory agents used (Figure 3A-C). IL-22 production was also examined and may be reduced in CD200R1KO ILCs, but this was not found to be statistically significantly different, potentially due to the relatively low levels of production (Figure S2A-B). Levels of IL-22 were reduced in the culture supernatant from CD200R1KO skin cells relative to WT (Figure S2B), but it is unclear if this is due to a reduction in IL-22 production by ILCs or another cell type present. To determine the ILC subtype producing IL-17A, IL-17A production was examined in RORγt+ and RORγt- cells. This showed that it is the RORγt+ ILCs which are responsible for producing IL-17A in response to IL-23 stimulation (Figure 3D). Similarly, to determine if the IL-17A producing cells could be inflammatory ILC2 which co-express GATA3 and RORγt, IL-17A production was examined on GATA3+ and GATA3- ILC populations. This showed that the IL-17A producing cells are predominantly GATA3- RORγt+ (Figure 3D) and are therefore ILC3 rather than inflammatory ILC2. Similar to skin, gut ILCs from CD200R1-deficient mice were less able to produce IL-17A after stimulation with IL-23, or with PMA, Ionomycin and a cocktail of cytokines, regardless of whether they are cultured unseparated, or cultured after flow cytometric sorting to obtain a purified ILC population (Figure 3E). This reduction in IL-17A production by CD200R1 deficient ILCs was found to be independent of secreted inhibitory factors in the CD200R1-deficient cell cultures, as co-culture of WT and CD200R1-deficient cells maintained the impaired IL-17A production by CD200R1KO ILCs (Figure 3F). There was also no increase in known ILC3 inhibitory cytokines, IL-10, IFNγ and IL-25 in the CD200R1-deficient cell cultures relative to WT cultures (Figure S2C), suggesting that CD200R1 directly affects IL-17 production rather than modulating the activity of an intermediary cell type in this in vitro model. CD200R1 does not affect ILC3 subsets We hypothesized that the reduced IL-17 production seen in CD200R1KO ILCs may be due to either a reduction in ILC3s numbers, or the lack of an ILC3 subset. To test this hypothesis, firstly the proportion and numbers of ILC2, ILC3 and ILC lacking GATA3 and RORγt (22) were examined in skin and small intestinal lamina propria. This showed there was not a reduction in ILC3 in CD200R1KO which could account for the reduced IL-17 production (Figure 4A-B). ILC3 are heterogeneous, particularly in the small intestine where 4 main subsets of ILC3s are seen (24). NCR+ (NKp46+) ILC3s are specialised in interacting with myeloid and stromal cells, CD4+ and CD4- Lti-like (CCR6+) ILC3s have roles in modulating innate and adaptive immune responses, and NCR- ILC3s are precursors of the NCR+ ILC3 subset. All of these subsets were observed in CD200R1KO small intestine at similar frequencies to WT mice (Figure 4C-D), showing the reduced IL-17 production in CD200R1KO ILCs is not due to a missing major ILC3 subset. Less is known about the subsets of ILC3 present in skin but, applying similar gating as that used for the small intestine allowed the presence of CD4- Lti-like ILC3s and NCR- ILC3s to be observed. Similar to the small intestine, CD200R1 deficiency did not affect these skin ILC3 populations (Figure 4E-F). CD200R1-deficient ILC3s are transcriptomically very different to WT To determine why ILC3s from CD200R1-deficient mice are less able to produce IL-17A, the cell surface phenotype was analysed by flow cytometry. Skin ILC3s showed undetectable CD4, IL-23R and ckit and only low levels of CD25 and CD1d expression. They expressed CCR6, Sca-1 and MHC class II, but CD200R1 deficiency does not appear to affect the expression of any of these markers (Figure 5A). In the gut, expression of all of the markers was observed, but again, CD200R1 deficiency did not affect the levels of these markers (Figure 5B), demonstrating that ILC3 from CD200R1-deficient mice have a similar cell surface phenotype to WT. IL-23R was expressed at an extremely low level due to a presumed combination of low expression level and the lack of appropriate reagents (25), making interpretation of the effect of CD200R1 on IL-23R expression problematic. Gut ILC3 are highly heterogeneous, so there may be differences in the expression of some of these markers on a specific ILC3 subsets. However, as the main ILC3 subsets are in equal proportions in CD200R1KO and WT mice (Figure 4B-D), this seems unlikely. To determine how CD200R1 affects ILC3 function, ILC3s from WT (RORc eGFP) and CD200R1-deficient (CD200R1KO RORc eGFP) mice were sorted from small intestine (total ILC3s isolated, gating and purity shown in Figure S3) and RNASeq was performed. Small intestinal cells were used to provide a large enough cell population to be isolated. ILC3s from CD200R1KO mice were found to have 3285 differentially regulated genes, with 2252 of those upregulated, and 1033 down-regulated relative to WT (Figure 5C). Despite there being more upregulated than down regulated genes in the ILC3s from CD200R1-deficient mice, there are less highly upregulated genes (with a log2 fold change of >10) than there are highly down regulated genes (with a log2 fold change <-10) (Figure 5D). Pathway analysis revealed many changes in ILC3s from CD200R1-deficient mice, but the most highly enriched pathway was the nuclear factor-erythroid 2–related factor 2 (NRF2)-mediated oxidative stress response (Figure 5E-F), which is known to protect against oxidative stress and reduce ROS production but also inhibits Th17 differentiation, STAT3 activation and innate immune cell activation (26, 27). Indeed, other pathways enriched in CD200R1KO ILC3s included STAT3 signalling as well as others of potential interest including TGFβ signalling, and embryonic stem cell pluripotency (Figure 5E-F). A small number of pathways were down-regulated in ILC3 from CD200R1KO mice, the most significantly altered being interferon signalling (Figure 5G-H). This may reflect an inability to signal in response to interferons, or a reduction in interferons in CD200R1KO mice. Overall these data suggest that ILC3s from CD200R1-deficient mice are transcriptomically very different from WT and may have changes in cytokine signalling pathways, however, as these data are from global CD200R1-deficient mice, they do not distinguish between a role for CD200R1 on the ILC3s themselves, versus CD200R1 affecting ILC3s via another cell type. CD200R1 plays a cell intrinsic role in promoting IL-17A production by ILC3s In Figure 3E we showed that sorted CD200R1KO gut ILC3 are deficient at IL-17A production and, in Figure 3F we showed that when CD200R1KO skin cells were co-cultured with WT skin cells, the CD200R1KO ILCs were deficient in IL-17A production, ruling out effects of CD200R1 deficiency on other cell types in the cultures. This suggests that the effects of CD200R1 on IL-17A production by ILC may be cell intrinsic but, these data do not exclude effects of CD200R1 on non-ILC influencing ILC activity long term. These findings only rule out the effect of CD200R1 being mediated by non-ILCs within the time frame of the experiment (24 hrs). Therefore, to determine if CD200R1 affects ILC3 activity in a cell intrinsic or extrinsic manner, chimeric mice were generated with a mixture of WT and CD200R1KO bone marrow, and skin inflammation was induced. This revealed that CD200R1KO ILC are less able to produce IL-17A than WT ILCs in skin (Figure 6A-C). The overall number of ILCs in skin is unaffected by the genotype of the host mice, or the genotype of the bone marrow from which the ILC are derived. However, the number of IL-17A producing ILCs derived from CD200R1KO bone marrow is reduced relative to those derived from WT bone marrow, in untreated WT and Aldara treated CD200R1KO hosts (Figure 6C). Together, this shows that CD200R1 promotes IL-17 production by skin ILC3s in a cell intrinsic manner. CD200R1 promotes IL-17 production by ILC3s via enhancing STAT3 activation in response to IL-23 As CD200R1KO ILC3s are less able to produce IL-17A in response to IL-23, and detecting IL-23R using antibodies is problematic (25), we investigated the Il-23r transcript level in our RNAseq data. This showed that CD200R1KO ILC3 do not have reduced Il-23r transcript levels (Figure 6D). STAT3 is a crucial mediator of IL-23 signalling, and the STAT3 signalling pathway was found to be enriched in CD200R1KO ILC3 (Figure 5E-F). This enrichment in signalling components may suggest there is an enhanced response to IL-23, or alternatively it may reflect the cell compensating for a deficiency in signalling. The RNAseq data were again examined to determine if Stat3 expression was affected by CD200R1. Stat3 levels were also found to be independent of CD200R1 but, in contrast the CD200R1-deficient ILC3s do have reduced Il-17a transcript levels as expected (Figure 6D). As IL-23R levels appear to be unchanged in the absence of CD200R1 (Figure 6D), we investigated signalling downstream of IL-23R by examining the phosphorylation of STAT3 (pSTAT3). This demonstrated that the proportion of pSTAT3 positive cells and the level of pSTAT3 is reduced in CD200R1-deficient ILC3s (Figure 6E), showing that signalling in response to IL-23 is impaired in the absence of CD200R1. The RNAseq data suggest that STAT3 signalling may be affected by CD200R1 (Figure 5E-F), which coincides with the reduction in pSTAT3 levels in stimulated ILC3s (Figure 6E). To understand this further, expression of components of the IL-23R signalling pathway were determined in the RNAseq data. On engagement of IL-23R, JAK2 and TYK2 become activated and phosphorylate STAT3 allowing its dimerization and translocation to the nucleus where it can modulate transcription of target genes (28). Jak2 expression was not affected by CD200R1, however, Tyk2 which is also activated rapidly in response to IL-23R engagement was increased in CD200R1KO ILC3s (Figure 6F). TYK2 has been shown to negatively regulate STAT3 activation (29), perhaps accounting for the reduced STAT3 phosphorylation observed in these cells. NRAS and RRAS mRNA levels were also reduced in CD200R1KO ILC3s again showing that IL-23R signalling is reduced in the absence of CD200R1 (Figure 6F) and providing an explanation for the reduced IL-17A production in the absence of CD200R1. Discussion Here we show that CD200R1 promotes IL-17A production by ILCs in a cell intrinsic manner and via the promotion of STAT3 phosphorylation. This is surprising as the CD200:CD200R1 pathway has been shown to inhibit proinflammatory cytokine production by myeloid cells (30, 31), and suppress a variety of immune responses (7–13). However, most studies examining the role of CD200:CD200R1 signalling used exogenous ligand or blocked CD200R1 using antibodies. Therefore, the role of this pathway in ubiquitous knockout animals lacking CD200R1 in all cells, has not previously been fully addressed. We recently examined psoriasis-like skin inflammation in mice treated with a CD200R1 blocking antibody and found that neutrophil accumulation was increased, demonstrating that CD200R1 limits neutrophil recruitment (17). Here, we did not observe effects on neutrophil recruitment in the IL-23 injection model (data not shown), and only saw a slight decrease in neutrophils in the Aldara cream-induced psoriasis model. However, neutrophil recruitment effects may be masked in CD200R1-deficient mice due to compensation between the increase in neutrophil recruitment and the reduction in IL-17A production. Indeed, the skin thickening in response to psoriasis-like inflammation was not changed in CD200R1-deficient mice despite the reduction in IL-17A, which suggests that there may be an increase in an alternative inflammatory response in these mice. The fact that we did not observe a decrease in IL-17A production when CD200R1 was blocked in an in vitro ILC3 activation model using adult skin cells (17), may suggest that CD200R1 promotes IL-17A production via effects on the development or differentiation of ILC3s, or CD200R1 may be required over a longer time period to affect IL-17A production. Other work has shown that mice genetically lacking the ligand, CD200, have reduced suppression, including in influenza and mouse coronavirus infections (12, 32, 33), in an experimental meningococcal septicaemia model (34) and in experimental autoimmune encephalitis and collagen induced arthritis models (35). This suggests that endogenous CD200 does indeed inhibit diverse immune responses, which is opposite to the effects that we describe here for its receptor, CD200R1. However, recent papers suggest that the CD200:CD200R1 signalling pathway may not act as a straightforward ligand-receptor pair. The effect of exogenous CD200 is not mirrored by the effect of agonistic antibodies against CD200R1 (36), suggesting that either the agonistic antibodies do not activate CD200R1 in the same way as CD200, or that CD200 has effects independent from CD200R1. Although there exists CD200R-like proteins which appear to have activating functions and could account for differential effects of CD200, these do not appear to be capable of CD200 binding (37, 38), so probably do not account for these effects. Similarly, viruses are known to express CD200 homologues which are proposed to allow viral evasion of the immune response (39), however it has been shown that some of these effects are independent of CD200R1 (40). This again suggests that CD200 has effects independent of CD200R1. Conversely, in the gut, other ligands for CD200R1 have been identified (41), and it is not yet clear if these ligands play similar, or distinct roles to CD200. In support of our finding here that CD200R1 can promote inflammation, a recent report, suggests that CD200R1 switches from being suppressive to being stimulatory in the presence of IFNa (42). Therefore, it is clear that we do not yet fully understand the different roles played by CD200 versus CD200R1, and how these effects are mediated. Gaining a better understanding of this pathway is crucial, particularly as therapeutics are being developed to target this pathway for the treatment of cancers (43). Although here we have shown that CD200R1 promotes IL-17A production by ILC3s in a cell intrinsic manner, and that CD200R1 promotes STAT3 activation downstream of IL-23 stimulation, here we have not shown that the reduced STAT3 activation in ILC3s in CD200R1KO mice is due to the lack of CD200R1 on the ILC3s themselves. Therefore, we cannot rule out the possibility that the effects of CD200R1 on STAT3 activation are mediated by another cell type, and CD200R1 on ILC3s affects IL-17A production via another mechanism. One interesting finding here, is that CD200R1 promotes ILC3 IL-17A production in response to PMA/Ionomycin stimulation, which bypasses the requirement for STAT3 activation. Whilst we do not yet understand the mechanism by which CD200R1 acts, given the large transcriptomic differences, it seems likely that CD200R1-deficiency dysregulates signalling pathway components more widely than just the STAT3 signalling pathway. In Fig.3 we show 20% of WT ILCs produce IL-17A in response to PMA/Ionomycin stimulation. In the BM chimeras a similar proportion (15-20%) of the WT bone marrow derived cells produce IL-17A whereas, lower proportions of the surviving host ILCs, and the CD200R1KO bone marrow-derived ILCs produce IL-17A, indicating reduced activity. The CD200R1KO bone marrow derived ILCs are less able to produce IL-17A due to their deficiency in CD200R1, as this is the only difference between these cells and the WT bone marrow derived ILCs. The residual host ILC’s reduced ability to produce IL-17A was unexpected, but may be due to radiation effects or the age of the mice (at analysis they are 20-25 weeks old). In contrast to the idea that this is age-related, transcripts associated with IL-17 signalling have been shown to be enriched with age in ILCs (44). High dose localised irradiation has been shown to induce IL-17 production by γδ T cells and ILCs (45), rather than inhibit IL-17A production however, differences in the radiation dose and body site targeted may explain this difference. In the case of targeted radiation, ILCs could be recruited from other skin or body sites in contrast, here, where whole-body irradiation was used, ILCs are depleted globally, so host ILCs cannot be recruited from other tissue sites. Therefore, it seems likely that whole body irradiation is responsible for the reduced ability of host ILC to produce IL-17A either by directly affecting ILC subsets or activity, or indirectly via effects on other cell types. Here, the IL-17 production capacity of ILC was investigated in lymph nodes, in addition to skin. Clearly, the skin is the crucial tissue to investigate when studying skin inflammation models, but it has the disadvantage of requiring lengthy enzymatic digestion procedures to allow the isolation of cells, which can impact on the cell activity and phenotype. Conversely, isolating cells from lymph nodes is much more straightforward, however, the contribution of lymph node derived IL-17 to skin inflammation is not well understood. Evidence shows that there is very little trafficking of ILCs from skin to the LN either in other inflammatory skin models or under homeostatic conditions (22). However, the Aldara model of psoriasis does involve systemic inflammation, including the induction of IL-17 in secondary lymphoid organs (20). Also, other inflammatory skin models have increased ILC numbers in skin draining lymph nodes due to increased proliferation and recruitment from the circulation (22). Therefore, an increase in IL-17A production in lymph nodes is likely to reflect ongoing inflammation systemically and probably also cutaneously. In conclusion, here we show a crucial novel role for CD200R1 in promoting IL-17A production by ILC3s showing that CD200:CD200R1 effects are more complex than just inhibiting immune responses. Materials and Methods Mice All animal experiments were locally ethically approved and performed in accordance with the UK Home Office Animals (Scientific Procedures) Act 1986. C57BL/6 mice were obtained from Charles River Laboratories. CD200R1KO mice (46), (provided by Prof. Tracy Hussell), RORc eGFP(47), (provided by Prof. Gerard Eberl), CD45.1 (48) (B.6SJL-Ptprca Pep3b/BoyJ–Ly5.1 mice, provided by Prof. Andrew MacDonald), CD200R1KO CD45.1 and CD200R1KO RORc eGFP mice (generated by crossing CD200R1KO with CD45.1 or RORc eGFP mice respectively) were bred and maintained in SPF conditions in house. Mice were 7 to 14 weeks old at the start of procedures, unless otherwise stated. Skin inflammation models Mice were anaesthetised and ear thickness was measured daily using a digital micrometer (Mitutoyo). Redness and scaling were scored daily. For the IL-23 intradermal injection model, ears were injected intradermally with 1 μg IL-23 (eBioscience) or PBS, daily for five consecutive days (19). For the Aldara cream model (20), ears were treated topically with 20 mg Aldara cream (Meda Pharmaceuticals), (containing imiquimod and isosteric acid) daily, for 4 days. H&E staining Ear tissue was fixed in 10% neutral buffered formalin then embedded in paraffin and cut to 5 µm. Haematoxylin and eosin staining was carried out using an automated Shandon Varistain V24-4. Images were acquired using a 3D-Histech Pannoramic-250 microscope slide-scanner. Snapshots were taken with Case Viewer software (3D-Histech). Cell isolation Skin and LN Ears were split in half and floated on 0.8% w/v Trypsin (Sigma) at 37 °C for 30 min, then epidermis was removed and both epidermis and dermis were chopped and digested in 0.1 mg/ml (0.5 Wunch units/ml) Liberase TM (Roche) at 37 °C for 1 hr. Subcutaneous fat was removed from dorsal skin before floating on Trypsin as above. Tissue was chopped and digested with 1 mg/ml Dispase II (Roche) at 37 °C for 1 hr. Digested skin tissue suspensions, or inguinal, brachial and axillary lymph nodes were passed through 70 µm cell strainers and cells were counted. Small intestinal lamina propria cells Cells were isolated as detailed previously (49). Briefly, small intestines were excised and Peyer’s patches were removed. Intestines were sliced open, washed with PBS and cut into segments and shaken in RPMI with 3% FCS, 20mM HEPES and 1% penicillin streptomycin. The tissue was transferred to pre-warmed RPMI supplemented with 3% FCS, 20mM HEPES, 1% penicillin streptomycin, 2mM EDTA (Fischer Scientific) and 14.5 mg/mL dithiothreitol (DTT; Sigma) and incubated at 37°C for 20 minutes with agitation. The tissue was shaken 3 times in pre-warmed RPMI supplemented with 20 mM HEPES, 1% penicillin streptomycin and 2mM EDTA then was washed with PBS. Tissues were minced then digested with 0.1 mg/mL Liberase TL (Roche) and 0.5 mg/mL DNaseI in RPMI for 30 minutes at 37°C. Cell suspensions were passed through 70 μm, and 40 μm cell strainers (Fischer Scientific) then washed in complete RPMI (RPMI supplemented with 10% heat inactivated FBS, 1% penicillin streptomycin solution, 2 mM L-glutamine, 1 mM sodium pyruvate, 20 mM HEPES, 1X non-essential amino acid solution, 25 nM 2-mercaptoethanol (Sigma)). Bone marrow cells Fibulas and tibias were removed, and bone marrow cells were isolated using a needle and syringe and PBS. Red blood cells were lysed using ACK lysis buffer (Lonza) and cells were washed and counted. Flow cytometric analysis Cells were incubated with 0.5 µg/ml anti-CD16/32 (2.4G2, BD Bioscience) and Near IR Dead cell stain (Invitrogen) prior to staining with fluorescently labelled antibodies. Cells were fixed with Foxp3/Transcription Factor Buffer Staining Set (eBioscience) for between 30 min and 16 hr at 4°C. For cytokine analysis, 10 µM Brefeldin A was added to cell cultures for 4 hr prior to staining for cell surface markers as described above. After overnight fixation, cells were permeabilized with Foxp3/Transcription Factor Buffer Staining Set (eBioscience) and were stained with intracellular markers or cytokines. Cells were analysed on a BD Fortessa or LSRII flow cytometer. Data were analysed using FlowJo (TreeStar). Antibodies are detailed in Table S1. For pSTAT3 staining, inguinal, axillary and brachial LN cells were stained for surface markers, then stimulated with 100 ng/ml IL-23 (Biolegend) for 15 min. Cells were fixed with Phosflow Fix buffer I (BD Biosciences) at 37 °C for 10 min, before permeabilisation at 4 °C in Phosflow Perm Buffer III (BD Biosciences) for 30 min, staining with PE conjugated pSTAT3 (pY705) (BD Bioscience clone 4/P-STAT3) at room temperature for 30 min. Data were normalized to the mean stimulated WT data. ILCs were gated as live, single cell, CD45+, Lin- (Ter119, F4/80, CD11b, CD11c, FceRIa, Gr1, CD19), CD3-, TCRb-, TCRgd-, CD90.2+ CD127+. In vitro ILC activation assay Mouse dorsal skin cells were stimulated in culture with 40 ng/mL IL-23 (Biolegend) alone, or in combination with 40 ng/mL IL-1β (eBioscience) overnight. Supernatants were retained for cytokine analysis and cells were cultured with 10 µM Brefeldin A (Sigma) for 4-5 hr before staining for flow cytometric analysis. Alternatively, cells were stimulated with 50 ng/mL PMA and 500 ng/mL ionomycin (Sigma) and 10 mM Brefeldin A for 4 hr before staining for flow cytometric analysis. Similarly, Small intestinal lamina propria cells were stimulated with 20 ng/mL IL-23, IL-6, IL-2 and IL-1β in the presence of 10 μM Brefeldin A for 2 hours, before PMA (50 ng/mL) and ionomycin (500 ng/mL) was added for a further 3 hours. For co-culture of WT and CD200R1KO cells, dorsal skin cells were isolated and either the WT or CD200R1KO cells were labelled with 10 μM eF450-conjugated cell proliferation dye (Fischer Scientific) for 10 minutes in the dark at 37°C then were washed and co-cultured with the unlabelled cells. ELISA Levels of IL-17A, IL-10, IFN-γ, and IL-25 were measured using Ready-SET-Go! ELISA kits (eBioscience) following the manufacturer’s instructions. Data plotted are the mean of at least 2 technical replicates for each individual sample. Flow cytometric cell sorting Cells were isolated from the small intestinal lamina propria and labelled with fluorescent antibodies as detailed above, before sorting on a BD FACS Aria cell sorter to a typical purity of >95%. RNAseq ILC3s (Live, single cell, CD45+, Lin- (Ter119, F4/80, CD11b, CD11c, FceRIa, Gr1, CD19), CD3-, TCRb-, TCRgd-, CD90.2+, CD127+, GFP+) were sorted from RORc eGFP (WT) and CD200R1KO RORc eGFP (KO) small intestinal lamina propria keeping cells from each mouse separate. RNA was isolated using the RNeasy Micro kit (QIAGEN) following the manufacturer’s instructions. Contaminating genomic DNA was removed using gDNA Eliminator columns. Quality and integrity of the RNA samples were assessed using a 2200 TapeStation (Agilent Technologies, Cheadle, UK) and libraries were generated using the TruSeq® Stranded mRNA assay (Illumina, Inc., California, USA) according to the manufacturer’s protocol before sequencing with an Illumina HiSeq4000 instrument. The output data was demultiplexed (allowing one mismatch) and BCL-to-Fastq conversion was performed using Illumina’s bcl2fastq software, version 2.17.1.14. Reads were quality trimmed using Trimmomatic_0.36 (PMID: 24695404) then mapped against the reference mouse genome, version mm10/GRCm38 using STAR_2.4.2 (PMID: 23104886). Counts per gene were calculated with HTSeq_0.6.1 (PMID:25260700) using annotation from GENCODE M12 (http://www.gencodegenes.org/). Normalisation and differential expression was calculated with DESeq2_1.12.3 (PMID:25516281) and edgeR_3.12.1 (PMID:19910308). Ingenuity Pathways Analysis was used to determine differentially regulated pathways. Generating bone marrow chimeric mice Male WT (C57BL/6) and CD200R1KO mice (both CD45.2) were exposed to a split dose of 10.5-11Gy irradiation. Bone marrow was isolated from male WT (CD45.1 or CD45.1 x C57BL/6 F1) and CD200R1KO (either CD45.1 or CD45.1 x CD200R1KO F1) mice and donor T cells were removed using CD90.2 microbeads and Miltenyi LS MACS columns. Cells were counted and mixed at a 1:1 ratio before being adoptively transferred intravenously into the irradiated hosts. 10-12 weeks post reconstitution, skin inflammation was induced by topical Aldara cream application daily (as described above) for 6 days. A day after the final treatment, mice were euthanised and cells analysed by flow cytometry after stimulating with PMA and Ionomycin to ensure cytokine production was detectable. Statistics Data were analysed for normality by Shapiro-Wilk test. When comparing two independent groups a Student’s t test (for normally distributed data) was used. For three or more groups, data were analysed by One-way ANOVA (as normally distributed) or to determine the effects of two independent variables on multiple groups a Two-way ANOVA was used followed by a Bonferroni post hoc test. All statistical tests were performed using Prism Software (GraphPad Software Inc., USA). Values of less than p<0.05 were considered significant. Error bars show standard deviation. All experiments were performed at least twice (except for the RNAseq analysis), with at least 3 independent samples per group. Each data point represents an individual animal. Supplementary Material Supplementary Information Acknowledgments This research was funded by a pre-competitive, open innovation award to the Manchester Collaborative Centre for Inflammation Research, by University of Manchester, AstraZeneca and GSK, and a Wellcome Trust and Royal Society, Sir Henry Dale Fellowship to AS (109375/Z/15/Z). The University of Manchester Bioimaging Facility microscopes used in this study were purchased with grants from BBSRC, Wellcome and the University of Manchester Strategic Fund. We acknowledge assistance from Peter Walker, Roger Meadows and Gareth Howell, and Leo Zeef and the use of the University of Manchester Histology, Flow Cytometry, Genomic Technologies, Bioinformatics and Biological Services facilities. We also acknowledge technical help from Peter Warn and Andrew Sharp on skin infection models and Tovah Shaw on intestinal cell isolation. We acknowledge intellectual expertise on ILC3s from Matthew Hepworth and on CD200R1 from Tracy Hussell. This work was funded by a pre-competitive, open innovation award to the Manchester Collaborative Centre for Inflammation Research, at the University of Manchester, AstraZeneca and GSK, and a Wellcome Trust and Royal Society, Sir Henry Dale Fellowship to AS (109375/Z/15/Z). Abbreviations DETC dendritic epidermal T cell IL interleukin ILC innate lymphoid cell ILC3s group 3 innate lymphoid cells Lti lymphoid tissue inducer cell, NCR Natural cytotoxicity receptor ROS Reactive oxygen species TLR7 Toll-like receptor 7 Th17 Type 17 T helper cell Data availability statement The datasets generated during this study are available from the corresponding author on reasonable request. Figure 1 CD200R1 promotes IL-17A production in murine models of psoriasis A-F WT and CD200R1KO (KO) Ears were intradermally injected with 1µg IL-23, or PBS daily for 5 consecutive days. A. Skin thickening measured using a digital micrometer. B. H&E stained skin sections. C. Proportion of live, single cells in skin that are CD45+. D. IL-17A level in media from overnight cultured skin cells. E. Proportion of skin ILCs (CD45+ Lin- CD3- TCRβ- γδTCR- Thy1+ CD127+) producing IL-17A. F. Proportion of skin CD3low γδ T cells producing IL-17A. G-M. Inflammation was induced by topically treating WT and CD200R1KO (KO) littermate ear skin with Aldara cream daily. G. Skin thickening measured using a digital micrometer. H. Skin redness and scaling scores. I. H&E stained skin sections. J. Proportion of live, single cells in skin that are CD45+. K. Proportion of skin CD45+ cells that are neutrophils (CD45+ CD11b+ Ly6G+). L. Proportion and number of skin CD3low γδ T cells or ILCs producing IL-17A. M. Proportion and number of lymph node γδ T cells or ILCs producing IL-17A. Histology scale bars show 100 µm. Data shown are pooled from 2 independent experiments except for B and I showing representative histology, D showing representative ELISA, H, J, K and M showing representative data from one experiment. A, C-F were analysed by two-way ANOVA followed by a Bonferroni post hoc test. H, J-M were analysed by Student’s t test. A, C, E, F n=9, D n=5. G, H, J-M n=5-7. Figure 2 CD200R1 is expressed by skin ILCs Dorsal skin cells were isolated and analysed by flow cytometry using fluorescence minus one (FMO) controls to set the gates. A. Dermal live CD45+ single cells were gated into various monocyte, dendritic cell and macrophage populations. B. Whole skin live CD45+ single cells were gated for Langerhans cells. C. Whole skin live CD45+ single cells were gated for T cell subsets and ILCs. D. CD200R1 expression on each cell subset. E. Skin ILCs (CD45+ Lin- CD3- TCRβ- γδTCR- Thy1+ CD127+) were gated for ILC2 (GATA3+), ILC3 (RORγt+) and non-ILC2/3 (GATA3- RORγt-), and CD200R1 expression was analysed within each subset. Filled histograms show CD200R1KO data, lines show WT data. All data are representative of 2 replicate experiments except E showing data from 2 experiments which was analysed by One-way ANOVA. Figure 3 CD200R1 is required for efficient IL-17A production by ILCs Dorsal skin or intestinal lamina propria cells from WT or CD200R1KO (KO) mice were stimulated in culture and IL-17A production by ILC was measured by flow cytometry. A. IL-17 production by dorsal skin ILCs (CD45+ Lin- CD3- TCRβ- γδTCR- Thy1+ CD127+) stimulated with IL-23, or with PMA and ionomycin for 4 hrs. B. IL-17A levels in cell supernatant from IL-23-stimulated skin cells. C. IL-17 production by dorsal skin ILCs from littermates in response to IL-23. D. IL-17A production in RORγt+ and RORγt- ILCs, and in GATA3+ and GATA3- ILCs in response to IL-23 stimulation. E. IL-17 production by intestinal lamina propria cells, or flow cytometric sorted intestinal lamina propria ILCs (CD45+ Lin- CD3- TCRβ- γδTCR- Thy1+ CD127+) stimulated with IL-23 alone or PMA, Ionomycin, IL-23, IL-6, IL-2 and IL-1β for 5 hrs. F. CD200R1KO dorsal skin cells were labelled with a fluorescent dye before being co-cultured in a 1:1 ratio with unlabelled WT cells. IL-17A production was analysed in co-cultured cells stimulated with IL-23 overnight. A (PMA/Ionomycin stim) (n=4), B (n=5), E (stimulation cocktail) (n=3) data shown are from one representative experiment. C (n=6), E (IL-23 stim) (n=8-9), F (n=7) data from two independent experiments. A (IL-23 stim) (n=14) data shown are from 3 independent experiments. Data were analysed by Student’s t test. Figure 4 CD200R1KO mice do not lack any major subsets of ILC3s A. The proportion and number of WT and CD200R1KO (KO) skin ILCs (CD45+ Lin- CD3- TCRβ- γδTCR- Thy1+ CD127+) within each subset (ILC2, GATA3+; ILC3 RORγt+; non ILC2/3, GATA3- RORγt-). B. The proportion and number of small intestinal lamina propria ILCs within each subset. C. The proportion of ILC3s in each subset in WT and CD200R1KO small intestinal lamina propria. D. The proportion of ILC3s in each subset in WT and CD200R1KO skin. A data shown from 2 independent experiments (n= 7-9). B-F data are representative of 2 independent experiments, B n= 3, D and F n=4. Figure 5 CD200R1KO ILC3s are deficient at activating STAT3 WT and CD200R1KO (KO) ILC3s (CD45+ Lin- CD3- TCRβ- γδTCR- Thy1+ CD127+ RORγt+) cell surface marker expression in A. dorsal skin, B. small intestinal lamina propria. C. RNAseq analysis of flow cytometrically sorted ILC3s from WT (RORγt eGFP) and CD200R1KO (CD200R1KO RORγeGFP) lamina propria. Log2fold change values are shown for all differentially regulated genes. D. Volcano plot of the RNAseq data. E. Pathways predicted to be upregulated in KO ILC3 by pathway analysis of the RNAseq data. F. Heat maps showing z scores for genes in pathways predicted to be enriched in KO ILC3. G. Pathways predicted to be reduced in KO ILC3 by pathway analysis of the RNAseq data. H. Heat maps showing z scores for genes in pathways predicted to be downregulated in KO ILC3. A, B. data are from 2 independent experiments n= 4-8. Data were analysed by Student’s t test. Figure 6 CD200R1 is required in a cell intrinsic manner to promote IL-17A production by ILC3 A. Representative plots showing IL-17 production by ILC in mixed bone marrow (WT and CD200R1KO) chimeric mice which were stimulated in culture with PMA and Ionomycin. B-C. IL-17 production by skin ILC in mixed bone marrow (WT and CD200R1KO) chimeric mice, B. numbers and proportion of ILC and ILC3s producing IL-17A in skin. In the bottom left bar chart, the background FMO value (1.81) was subtracted from each data point. Grey data points show data less than 2x the background (FMO) which were set to zero. C. Flow cytometry showing the ILC subset producing IL-17. D. Il-23r, Stat3 and Il-17a expression in gut WT or CD200R1KO ILC3 measured by RNAseq. E. Representative pSTAT3 staining in inguinal, axillary and brachial lymph node ILCs either unstimulated, IL-23-stimulated WT ILCs or IL-23-stimulated CD200R1KO ILCs relative to the pSTAT3 FMO control. Gated on ILCs. Lower numbers show the % of ILCs which are RORγt+. Upper numbers show the proportion of ILC3s staining for pSTAT3. Bar charts show the normalised % of ILC3s which are pSTAT3+ and the MFI (median fluorescence intensity) for pSTAT3 in ILC3s. F. Expression of components of the IL-23R signalling pathway in gut WT or CD200R1KO ILC3 measured by RNAseq. B, E data are from 2 independent experiments. B n= 4-7, E n= 4-6. B, C data were analysed by two-way ANOVA followed by a Bonferroni post hoc test. D, F data were analysed by Student’s t test. E data were analysed by one-way ANOVA followed by a Tukey’s post hoc test. Author contributions HL, SJ, LB, JC, MP and AS performed experiments. HL, AO, SJ and AS analyzed the data. AS designed the project and wrote the manuscript. Competing interests The authors declare no competing interests. 1 Cua DJ Tato CM Innate IL-17-producing cells: the sentinels of the immune system Nat Rev Immunol 2010 10 7 479 89 20559326 2 Iwakura Y Ishigame H Saijo S Nakae S Functional specialization of interleukin-17 family members Immunity 2011 34 2 149 62 21349428 3 Nograles KE Zaba LC Guttman-Yassky E Fuentes-Duculan J Suarez-Farinas M Cardinale I Th17 cytokines interleukin (IL)-17 and IL-22 modulate distinct inflammatory and keratinocyte-response pathways Br J Dermatol 2008 159 5 1092 102 18684158 4 Ghoreschi K Balato A Enerback C Sabat R Therapeutics targeting the IL-23 and IL-17 pathway in psoriasis Lancet 2021 397 10275 754 66 33515492 5 Teunissen MBM Munneke JM Bernink JH Spuls PI Res PCM Te Velde A Composition of innate lymphoid cell subsets in the human skin: enrichment of NCR(+) ILC3 in lesional skin and blood of psoriasis patients J Invest Dermatol 2014 134 9 2351 60 24658504 6 Villanova F Flutter B Tosi I Grys K Sreeneebus H Perera GK Characterization of innate lymphoid cells in human skin and blood demonstrates increase of NKp44+ ILC3 in psoriasis J Invest Dermatol 2014 134 4 984 91 24352038 7 Rosenblum MD Yancey KB Olasz EB Truitt RL CD200, a “no danger” signal for hair follicles J Dermatol Sci 2006 41 3 165 74 16386879 8 Liu Y Bando Y Vargas-Lowy D Elyaman W Khoury SJ Huang T CD200R1 agonist attenuates mechanisms of chronic disease in a murine model of multiple sclerosis J Neurosci 2010 30 6 2025 38 20147531 9 Gorczynski RM Chen Z Lee L Yu K Hu J Anti-CD200R ameliorates collagen-induced arthritis in mice Clin Immunol 2002 104 3 256 64 12217336 10 Gorczynski RM Chen Z Yu K Hu J CD200 immunoadhesin suppresses collagen-induced arthritis in mice Clin Immunol 2001 101 3 328 34 11726225 11 Rosenblum MD Olasz E Woodliff JE Johnson BD Konkol MC Gerber KA CD200 is a novel p53-target gene involved in apoptosis-associated immune tolerance Blood 2004 103 7 2691 8 14644999 12 Snelgrove RJ Goulding J Didierlaurent AM Lyonga D Vekaria S Edwards L A critical function for CD200 in lung immune homeostasis and the severity of influenza infection Nat Immunol 2008 9 9 1074 83 18660812 13 Kretz-Rommel A Qin F Dakappagari N Ravey EP McWhirter J Oltean D CD200 expression on tumor cells suppresses antitumor immunity: new approaches to cancer immunotherapy J Immunol 2007 178 9 5595 605 17442942 14 Blom LH Martel BC Larsen LF Hansen CV Christensen MP Juel-Berg N The immunoglobulin superfamily member CD200R identifies cells involved in type 2 immune responses Allergy 2017 72 7 1081 90 28106273 15 Weizman OE Adams NM Schuster IS Krishna C Pritykin Y Lau C ILC1 Confer Early Host Protection at Initial Sites of Viral Infection Cell 2017 171 4 795 808 e12 29056343 16 Nagasawa M Heesters BA Kradolfer CMA Krabbendam L Martinez-Gonzalez I de Bruijn MJW KLRG1 and NKp46 discriminate subpopulations of human CD117(+)CRTH2(-) ILCs biased toward ILC2 or ILC3 J Exp Med 2019 216 8 1762 76 31201208 17 Linley H Jaigirdar S Mohamed K Griffiths CEM Saunders A Reduced cutaneous CD200:CD200R1 signaling in psoriasis enhances neutrophil recruitment to skin Immun Inflamm Dis 2022 10 7 e648 35759230 18 Li B Tsoi LC Swindell WR Gudjonsson JE Tejasvi T Johnston A Transcriptome analysis of psoriasis in a large case-control sample: RNA-seq provides insights into disease mechanisms J Invest Dermatol 2014 134 7 1828 38 24441097 19 Chan JR Blumenschein W Murphy E Diveu C Wiekowski M Abbondanzo S IL-23 stimulates epidermal hyperplasia via TNF and IL-20R2-dependent mechanisms with implications for psoriasis pathogenesis J Exp Med 2006 203 12 2577 87 17074928 20 van der Fits L Mourits S Voerman JS Kant M Boon L Laman JD Imiquimod-induced psoriasis-like skin inflammation in mice is mediated via the IL-23/IL-17 axis J Immunol 2009 182 9 5836 45 19380832 21 Tamoutounour S Guilliams M Montanana Sanchis F Liu H Terhorst D Malosse C Origins and functional specialization of macrophages and of conventional and monocyte-derived dendritic cells in mouse skin Immunity 2013 39 5 925 38 24184057 22 Dutton EE Camelo A Sleeman M Herbst R Carlesso G Belz GT Characterisation of innate lymphoid cell populations at different sites in mice with defective T cell immunity Wellcome Open Res 2017 2 117 29588921 23 Pantelyushin S Haak S Ingold B Kulig P Heppner FL Navarini AA Rorgammat+ innate lymphocytes and gammadelta T cells initiate psoriasiform plaque formation in mice J Clin Invest 2012 122 6 2252 6 22546855 24 Valle-Noguera A Gomez-Sanchez MJ Girard-Madoux MJH Cruz-Adalia A Optimized Protocol for Characterization of Mouse Gut Innate Lymphoid Cells Front Immunol 2020 11 563414 33329525 25 Awasthi A Riol-Blanco L Jager A Korn T Pot C Galileos G Cutting edge: IL-23 receptor gfp reporter mice reveal distinct populations of IL-17-producing cells J Immunol 2009 182 10 5904 8 19414740 26 Zhao M Chen H Ding Q Xu X Yu B Huang Z Nuclear Factor Erythroid 2-related Factor 2 Deficiency Exacerbates Lupus Nephritis in B6/lpr mice by Regulating Th17 Cell Function Sci Rep 2016 6 38619 27941837 27 Thimmulappa RK Lee H Rangasamy T Reddy SP Yamamoto M Kensler TW Nrf2 is a critical regulator of the innate immune response and survival during experimental sepsis J Clin Invest 2006 116 4 984 95 16585964 28 Pastor-Fernandez G Mariblanca IR Navarro MN Decoding IL-23 Signaling Cascade for New Therapeutic Opportunities Cells 2020 9 9 29 Mori R Wauman J Icardi L Van der Heyden J De Cauwer L Peelman F TYK2-induced phosphorylation of Y640 suppresses STAT3 transcriptional activity Sci Rep 2017 7 1 15919 29162862 30 Cherwinski HM Murphy CA Joyce BL Bigler ME Song YS Zurawski SM The CD200 receptor is a novel and potent regulator of murine and human mast cell function J Immunol 2005 174 3 1348 56 15661892 31 Jenmalm MC Cherwinski H Bowman EP Phillips JH Sedgwick JD Regulation of myeloid cell function through the CD200 receptor J Immunol 2006 176 1 191 9 16365410 32 Karnam G Rygiel TP Raaben M Grinwis GC Coenjaerts FE Ressing ME CD200 receptor controls sex-specific TLR7 responses to viral infection PLoS Pathog 2012 8 5 e1002710 22615569 33 Rygiel TP Rijkers ES de Ruiter T Stolte EH van der Valk M Rimmelzwaan GF Lack of CD200 enhances pathological T cell responses during influenza infection J Immunol 2009 183 3 1990 6 19587022 34 Mukhopadhyay S Pluddemann A Hoe JC Williams KJ Varin A Makepeace K Immune inhibitory ligand CD200 induction by TLRs and NLRs limits macrophage activation to protect the host from meningococcal septicemia Cell Host Microbe 2010 8 3 236 47 20833375 35 Hoek RM Ruuls SR Murphy CA Wright GJ Goddard R Zurawski SM Down-regulation of the macrophage lineage through interaction with OX2 (CD200) Science 2000 290 5497 1768 71 11099416 36 Pilch Z Tonecka K Skorzynski M Sas Z Braniewska A Kryczka T The pro-tumor effect of CD200 expression is not mimicked by agonistic CD200R antibodies PLoS One 2019 14 1 e0210796 30653571 37 Hatherley D Cherwinski HM Moshref M Barclay AN Recombinant CD200 protein does not bind activating proteins closely related to CD200 receptor J Immunol 2005 175 4 2469 74 16081818 38 Wright GJ Cherwinski H Foster-Cuevas M Brooke G Puklavec MJ Bigler M Characterization of the CD200 receptor family in mice and humans and their interactions with CD200 J Immunol 2003 171 6 3034 46 12960329 39 Vaine CA Soberman RJ The CD200-CD200R1 inhibitory signaling pathway: immune regulation and host-pathogen interactions Adv Immunol 2014 121 191 211 24388216 40 Akkaya M Kwong LS Akkaya E Hatherley D Barclay AN Rabbit CD200R binds host CD200 but not CD200-like proteins from poxviruses Virology 2016 488 1 8 26590792 41 Kojima T Tsuchiya K Ikemizu S Yoshikawa S Yamanishi Y Watanabe M Novel CD200 homologues iSEC1 and iSEC2 are gastrointestinal secretory cell-specific ligands of inhibitory receptor CD200R Sci Rep 2016 6 36457 27819346 42 van der Vlist M Ramos MIP van den Hoogen LL Hiddingh S Timmerman LM de Hond TAP Signaling by the inhibitory receptor CD200R is rewired by type I interferon Sci Signal 2021 14 704 eabb4324 34637328 43 Mahadevan D Lanasa MC Farber C Pandey M Whelden M Faas SJ Phase I study of samalizumab in chronic lymphocytic leukemia and multiple myeloma: blockade of the immune checkpoint CD200 J Immunother Cancer 2019 7 1 227 31443741 44 Solá P Mereu E Bonjoch J Casado M Reina O Blanco E Esteller M DiCroce L Heyn H Solanas G Aznar Benitah S Local IL-17 orchestrates skin aging bioRxiv 2022 45 Liao W Hei TK Cheng SK Radiation-Induced Dermatitis is Mediated by IL17-Expressing gammadelta T Cells Radiat Res 2017 187 4 454 64 28406748 46 Boudakov I Liu J Fan N Gulay P Wong K Gorczynski RM Mice lacking CD200R1 show absence of suppression of lipopolysaccharide-induced tumor necrosis factor-alpha and mixed leukocyte culture responses by CD200 Transplantation 2007 84 2 251 7 17667818 47 Lochner M Peduto L Cherrier M Sawa S Langa F Varona R In vivo equilibrium of proinflammatory IL-17+ and regulatory IL-10+ Foxp3+ RORgamma t+ T cells J Exp Med 2008 205 6 1381 93 18504307 48 Shen FW Saga Y Litman G Freeman G Tung JS Cantor H Cloning of Ly-5 cDNA Proc Natl Acad Sci U S A 1985 82 21 7360 3 3864163 49 Shaw TN Houston SA Wemyss K Bridgeman HM Barbera TA Zangerle-Murray T Tissue-resident macrophages in the intestine are long lived and defined by Tim-4 and CD4 expression J Exp Med 2018 215 6 1507 18 29789388

PMC010xxxxxx/PMC10286627.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9100846 1173 Cancer Causes Control Cancer Causes Control Cancer causes & control : CCC 0957-5243 1573-7225 35235084 10286627 10.1007/s10552-022-01558-x NIHMS1899441 Article Healthy lifestyle index and risk of pancreatic cancer in the Women’s Health Initiative Peila Rita 1 Coday Mace 2 Crane Tracy E. 3 Saquib Nazmus 4 Shadyab Aladdin H. 5 Tabung Fred K. 6 Zhang Xiaochen 7 Wactawski-Wende Jean 8 Rohan Thomas E. 1 1 Department of Epidemiology and Population Health, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461, USA 2 Department of Preventive Medicine, College of Medicine, University of Tennessee Health Science Center, Memphis, TN, USA 3 Behavioral Measurement and Interventions Cancer Prevention and Control Program, University of Arizona, Tucson, AZ, USA 4 College of Medicine at Sulaiman, Al Rajhi University, Al Bukayriyah, Saudi Arabia 5 Herbert Wertheim School of Public Health and Human Longevity Science, University of California, San Diego, La Jolla, CA, USA 6 Internal Medicine, Division of Medical Oncology, The Ohio State University College of Medicine and Comprehensive Cancer Center, Columbus, OH, USA 7 Division of Epidemiology, College of Public Health, Comprehensive Cancer Center, Ohio State University, Columbus, OH, USA 8 Department of Epidemiology and Environmental Health, School of Public Health, University of Buffalo, Buffalo, NY, USA Author contributions Conceptualization—TER. Data curation, formal analysis, and investigation RP. Methodology—RP: writing—original draft—RP, TER. Writing—rewriting and editing—all authors. ✉ Rita Peila, rita.peila@einsteinmed.org, Thomas E. Rohan, thomas.rohan@einsteinmed.org 9 6 2023 5 2022 02 3 2022 22 6 2023 33 5 737747 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Purpose Lifestyle factors such as smoking, alcohol, body weight, physical activity, and diet quality have been associated with the risk of pancreatic cancer. However, studies of their combined association in women are limited. Methods Data on smoking habits, alcohol intake, diet composition, recreational physical activity, body weight, and waist circumference, obtained at recruitment for 136,945 postmenopausal women (aged 50–79 years) participating in the Women’s Health Initiative study, were categorized separately, with higher scores for each variable assigned to the categories representing healthier behaviors. The combined healthy lifestyle index (HLI) score, created by summing the scores for each risk factor, was grouped into quartiles. We used multivariable-adjusted Cox regression to estimate hazard ratios (HR) and 95% confidence intervals (CI) for pancreatic cancer risk in association with the HLI. Results Over an average follow-up period of approximately 16.0 years, 1,119 incident cases of pancreatic cancer were ascertained. Compared to women in the lowest HLI quartile, those in the upper quartiles (qt) had a reduced risk of pancreatic cancer (multivariable-adjusted HRqt3rd 0.83, 95% CI 0.74–0.99; and HRqt4th 0.74, 95% CI 0.62–0.88, respectively, p trend = 0.001). Use of waist circumference instead of BMI in the HLI score yielded similar results. Among women who were either non-diabetic or non-smokers, high HLI was also associated with reduced risk HRqt4th 0.78, 95% CI 0.65–0.85 and HRqt4th 0.80, 95% CI 0.66–0.97, respectively). Stratification by BMI categories (18.5− < 25.0, 25.0− < 30.0 and > 30.0 kg/m2) showed similar results in all groups. Conclusions Our findings suggest that in postmenopausal women, a healthy lifestyle is associated with reduced risk of pancreatic cancer. Healthy lifestyle index Pancreatic cancer Postmenopausal women Propspective cohort study pmcIntroduction Pancreatic ductal adenocarcinoma is the 7th most common type of cancer in the US and the 14th worldwide [1, 2]. In its early stages, the disease is often asymptomatic, so that in the majority of cases, detection and diagnosis occur at an advanced stage, when resection and treatment options are limited [3, 4], resulting in a high fatality rate associated with this malignancy [2]. Increasing age, family history, diabetes, and chronic pancreatitis are major contributors to pancreatic cancer incidence [5, 6]. In addition, several potentially modifiable risk factors such as tobacco smoking, obesity, heavy alcohol consumption, and certain dietary patterns have also been linked to increased pancreatic cancer occurrence [6, 7]. Tobacco smoking has been consistently linked with pancreatic cancer, with increased risk in current smokers, in former smokers up to 20 years after quitting [8, 9], and in those exposed to second-hand smoke [10]. Obesity and overweight in childhood and throughout adulthood have been associated with increased risk of pancreatic precancerous lesions [11] and pancreatic cancer [12]. Dietary patterns, including relatively high intake of red and processed meat, fat, and sugar, have been linked to elevated risk of pancreatic cancer in some studies [13–15], while there are indications that whole-plant foods may inhibit pancreatic carcinogenesis and reduce the risk of pancreatic cancer [16, 17]. Despite some inconsistencies, several prospective cohort studies have found a positive dose-dependent association between alcohol intake, particularly distilled spirits such as liquor, and increased risk of pancreatic cancer [18]. The association between physical activity and pancreatic cancer has been the focus of several case–control and cohort studies, which have shown a potential protective effects of physical activity if practiced consistently over time [19]. Both the International Agency for Research on Cancer and the World Cancer Research Fund (WCRF)/American Institute for Cancer Research (AICR) have identified these factors as targets to improve primary prevention and reduce the burden of pancreatic cancer [20]. A study focused on the population attributable fraction (PAF) of pancreatic cancer based on nationally representative data on exposures and outcome prevalence estimated a PAF for pancreatic cancer of approximately 25% due to these modifiable risk factors, in the US population [21]. Similar estimates were found in the analysis of two distinct cohorts with a total of 140,000 individuals with detailed information on these factors and pancreatic cancer occurrence [22]. The healthy lifestyle index (HLI) is a composite score which combines information on body mass index (BMI), smoking, alcohol intake, diet quality, and physical activity and has been linked to the incidence of and mortality from several chronic diseases including cardiovascular disease and some types of cancer, as well as to life expectancy and quality of life [23–27]. However, only two studies have used a combined index score of these modifiable risk factors to assess the impact of a healthy lifestyle on the risk of pancreatic cancer [28, 29]. Therefore, given the paucity of studies to date, we evaluated the association of this lifestyle-related composite index with risk of pancreatic cancer in the Women’s Health Initiative (WHI) cohort. Methods The WHI study is a large prospective study designed to advance understanding of the determinants of major chronic disease in postmenopausal women. The study includes 161,808 postmenopausal women aged 50–79 at enrollment, representing major racial/ethnic groups recruited from the general population at 40 clinical centers throughout the United States between 1993 and 1998. The WHI comprises an observational component and 4 overlapping clinical trials, including 2 hormone therapy trials, a dietary modification trial, and a calcium plus vitamin D supplementation trial. Details of the selection criteria, study design, and trial interventions have been extensively reported [30, 31]. The original WHI study was completed in 2005 and was followed by two extension studies (2005–2010 and 2010–2020) with the purpose of gathering additional outcome information. Regular contact (through visits, phone calls, or mailings) with participants occurred throughout the follow-up period to oversee the study participation and the occurrence of health-related outcomes. The study was approved by institutional review boards at all 40 clinical centers and at the coordinating center. All participants in the WHI gave written-informed consent. Exposure information and covariates Information on demographic, health and reproductive characteristics, medication use, and lifestyle factors was collected at enrollment. Selection and categorization of the variables included in the HLI was based on WCRF/AIRC recommendation and on previous studies in this cohort [24, 32]. Dietary intake was assessed using a validated food frequency questionnaire administrated at the baseline examination [33] and was used to calculate the Alternate Healthy Eating Index (AHEI-2010), a diet quality index previously used by this and other studies [34, 35]. The AHEI is a composite score based on the intake of vegetables, fruit, nuts and soy protein, red and processed meat, trans fat, the ratio of polyunsaturated to saturated fat, and multivitamin use, with higher score values assigned to more healthy dietary habits [36]. The total AHEI-2010 score was categorized into quintiles. A self-administered questionnaire on frequency and duration of mild, moderate, and vigorous intensity recreational physical activity was completed at baseline by all study participants. A metabolic equivalent (MET) value was assigned to each category of physical activity, and the total weekly MET hours of activity was computed by multiplying the MET of each category by the duration (hours) and then summing the values (MET hours/week (h/wk)) [37]; quintiles of MET-h/wk were created. Smoking status was categorized based on smoking history and cigarette amount (never, former ≤ 15 pack years (number of packs of cigarettes smoked per day by the number of years the person had smoked), former > 15 pack years, current ≤ 15 pack years, and current > 15 pack years). Alcohol intake was calculated from the food frequency questionnaire for beer, wine, and liquor, converted to total grams (g) (12.8 g of ethanol for 360 ml (12 oz) of regular beer, 11.3 g for 360 ml (12 oz) of light beer, 11.0 g for 120 ml (4 oz) of wine, and 14.0 g for 45 ml (1.5 oz) of liquor) and categorized as < 6.0, 6.0– < 12.0, 12.0– < 24.0, 24.0– < 60.0, and ≥ 60.0 g a day (g/d). Measurements of weight (kg), height (cm), and waist circumference (cm) were obtained at baseline by trained personal. BMI was computed as measured weight divided by the square of height (kg/m2) and categorized into three groups (normal, 18.5– < 25.0; overweight, 25.0– < 30.0; and obese ≥ 30.0 kg/m2); women with a BMI < 18.5 kg/m2 were excluded, to reduce the likelihood of reverse causality due to an effect of pancreatic cancer on BMI. Waist circumference was categorized into three groups (< 80 cm, 80– < 88 cm, and ≥ 88 cm). The combined HLI score included diet quality, level of physical activity, smoking, alcohol intake, and either body mass index or, alternatively, waist circumference, as a marker of abdominal obesity [38]. A description of each component of the HLI and the HLIWST (which included waist circumference instead of BMI) and their distribution is presented in Supplementary Table 1. The total HLI score ranged from 0 to 18 with the highest score representing the healthiest combination of factors (highest quintiles of AHEI and level of physical activity, never smoked, < 6 g/d of alcohol consumption, and normal BMI, or waist circumference < 80 cm). Information on education, race and ethnicity, marital status, total non-alcohol energy intake, history of type 2 diabetes, history of pancreatitis, and use of hormone replacement therapy was collected at baseline. Type 2 diabetes status was updated throughout the follow-up based on information collected at each visit or via phone contact regarding receipt of a diagnosis of diabetes, use of oral glucose lowering medications or insulin, report of diabetes-related diet and exercise modifications, and hospitalization due to diabetes complications [39]. Analytic sample For the present study, we included all participants in the observational and clinical trial components, except women randomized to the dietary intervention arm (n = 19,541) of the dietary modification trial, as well as those missing follow-up information (n = 450). In addition, we excluded underweight women (BMI < 18.5 kg/m2, n = 1,328). Some of the participants had missing data on various components of the HLI and HLIWST (Supplementary Table 1), leaving a total of 136,945 women with complete data on the HLI and 139,556 on the HLIWST. Case ascertainment Information of cancer occurrence was collected every 6 to 12 months by mailed or telephone questionnaires. Self-reported cancer cases were verified by centralized review of medical records and pathology reports when a biopsy or resection were available [40]. Final adjudication and histology coding were used to identify primary pancreatic adenocarcinoma cases using the National Cancer Institute’s Surveillance, Epidemiology, and End Results coding system (including codes C25.0–C25.3 and C25.7–C25.9—International Classification of Diseases for Oncology (ICD)-2, and ICD-O-3 histology codes: 8000, 8010, 8021, 8140, 8144, 8150, 8240, 8246, 8260, 8323, 8440, 8470, 8480, 8481, 8490, 8500, 8503, 8560, 8570, 8800, and 8801). Statistical analysis The study population was categorized into quartiles of the HLI score based on its distribution among non-cases. Descriptive statistical analyses were conducted to compare the study population baseline characteristics by HLI quartiles using Wilcoxon rank-sum tests for continuous variables and χ2 tests for categorical variables. Cox-proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for the associations between the HLI quartiles and risk of pancreatic cancer using the lowest quartile as referent group; a similar approach was used for analyses based on the HLIwst. Follow-up time was used as the timescale and was calculated from the date of study entry until the date of cancer diagnosis for the cases, and until the date of death (as reported during the study follow-up and confirmed using the Vital Status from the National Death Index), withdrawal from the study, loss to follow-up, or the end of follow-up (02/28/2020) for the non-cases. No violation of the proportional hazards assumption for HLI quartiles was found on the basis of examination of Schoenfeld residuals (p value = 0.08). To test for linear trend in the risk of pancreatic cancer across the quartiles of the HLI or HLIwst, the quartile levels were included in the model as an ordinal variable and Wald test p values were reported. To evaluate whether HLIwst provided similar results to those obtained using BMI in the HLI, we compared the HR estimates associated with increments of 1 standard deviation of the HLI and HLIWST. Selection of the variables included in the multivariable model was based on their association with pancreatic cancer and whether these factors acted as confounders by altering the estimates of association for the main exposure by more than 10%. The models were adjusted for education, race/ethnicity, marital status, total non-alcohol energy intake, hormone replacement therapy, participation in the calcium and vitamin D trial, the dietary modification trial (no-intervention arm only), and/or the hormone replacement therapy trials, diabetes status and use of diabetes medications, and history of pancreatitis. In addition, when the HLIwst score was used, height was also included in the model. We conducted several additional analyses. First, we conducted a sensitivity analysis excluding women with less than two years of follow-up to reduce the possibility of an effect of an undiagnosed pancreatic cancer on risk factors such as BMI and waist circumference. Second, type 2 diabetes is an established risk factor for pancreatic cancer [41] and is often associated with elevated BMI [42], and a diet high in fat and low in whole grain, fruits, and vegetables [43]. Therefore, we carried out analyses excluding women with type 2 diabetes to test whether the association between HLI and pancreatic cancer was mainly driven by diabetes. All diabetes-free women at baseline (n = 135,264) were included in this analysis and censored at the time they reported having diabetes, if their diabetes status changed [44]. Third, analyses were also conducted excluding women who were current smokers at baseline (n = 6,733). Fourth, to test if the association between HLI and pancreatic cancer was modified by BMI level, we carried out analyses stratified by BMI categories using HRTWST quartiles. The p value for the interaction term between HLIWST quartiles and BMI categories was obtained by comparing the likelihood ratio of the full regression model, including the cross-product of the HLIWST and BMI terms, versus the reduced model without this term. Finally, we evaluated the association of the individual components of HLI with risk of pancreatic cancer by removing one component at a time from the composite score and adjusting for it in the model. Women with missing values for any of the HLI components were excluded from the main analysis, but they were included in the analyses of the individual components for which they did not have missing values. The analyses were conducted using STATA version 17 (Stata Corp LP, College Station, TX). All p values were two sided. Results Over an average follow-up period of 16.2 years (standard deviation, 7.0 years), a total of 1,119 confirmed cases of incident pancreatic cancer were ascertained among women with complete information on all the HLI components and 1,141 cases were ascertained among women with complete information on HLIWST. Baseline characteristics examined by HLI quartiles showed that women with higher HLI scores were more likely to be older, to be White, to have a higher level of education, to be married or in a marriage-like relationship, to be enrolled in the observational study component of WHI, and to have a lower total non-alcohol energy intake, and that they were less likely to have diabetes and to take diabetes medication at baseline (Table 1). Compared to women in the lowest quartile of the HLI, those in the upper quartiles had lower risk of pancreatic cancer (multivariable-adjusted HRqt3rd 0.83, 95% CI 0.74–0.99; and HRqt4th 0.74, 95% CI 0.62–0.88, respectively, p trend = 0.001). There was an 11% reduction in risk of pancreatic cancer per standard deviation increase in the HLI (SD = 2.9 HLI) (HR 0.89, 95% CI 0.84–0.95) (Table 2). Excluding women with less than two years of follow-up from the analysis did not change the risk estimates (HRqt4th 0.73, 95% CI 0.60–0.84). Use of waist circumference instead of BMI in the HLI score yielded similar results, with a lower risk of pancreatic cancer for those in the higher HLIWST quartiles compared to the lowest quartile (HRqt2nd 0.84, 95% CI 0.70–0.99; HRqt3rd 0.80, 95% CI 0.68–0.95; and HRqt4th 0.72, 95% CI 0.61–0.85, respectively, p trend = 0.001), and a similar reduction in risk associated with a one SD (3.0 HLI WST) increment (HR 0.91, 95% CI 0.85–0.99) (Table 2). A total of 20,994 (15.5%) women developed type 2 diabetes over an average period of 6.1 years. Analysis restricted to women who remained diabetes-free during the follow-up period yielded similar results (HRqt4th 0.78, 95% CI 0.64–0.95). Also, the results were similar when the analysis was restricted to non-current smokers (HRqt4th 0.80, 95% CI 0.66–0.97). Analyses stratified by categories of BMI showed that among women with normal BMI, a relatively high HLIWST was associated with a reduced risk of pancreatic cancer (HRqt4th 0.65, 95% CI 0.47–0.90), while among overweight and obese women, there was also a reduction in risk, although the confidence intervals included an HR of one, presumably due to the smaller sample size in these two strata. No overall effect modification by BMI on the association of HLIWST quartiles with pancreatic cancer risk was observed (p value for interaction term = 0.739) (Table 3). Additional analyses in which each lifestyle component was excluded in turn from the overall HLI score showed that compared to women in the lowest quartile of HLI, those in the uppermost quartile had a significantly reduced risk of pancreatic cancer independently of the component excluded (Table 4). Discussion In this large multicenter population-based study of postmenopausal women, we found that a healthy lifestyle index, represented by having a normal BMI or a waist circumference less than 80 cm, no cigarette smoking, low alcohol intake, undertaking at least moderate levels of physical activity, and a diet rich in vegetables, fruits, nuts, and low in red meat, trans fat, and with a high polyunsaturated to saturated fat ratio, was associated with a reduced risk of pancreatic cancer over an average follow-up period of 16 years. The results were similar regardless of whether we used the HLI or the HLIWST scores, and when the analysis was restricted to non-diabetics, non-smokers, and women with normal BMI (Table 4). Several epidemiological studies have examined the association between modifiable risk factors and the risk of pancreatic cancer, although only two studies have evaluated a combined healthy lifestyle score. A study conducted in the National Institutes of Health-AARP Diet and Health study cohort of approximately 450,000 participants (41.5% women), aged 50–71, used dichotomized scores (healthy behavior no/yes) for each of the 5 factors. Over a follow-up period of approximately 7 years, a total of 1,057 cases (382 in women) of pancreatic cancer were identified. The study showed that compared to the lowest score (0–1 points), the highest composite score (4–5) was associated with a reduction in pancreatic cancer risk (HR 0.48, 95% CI 0.37–0.63 in men, and HR 0.64, 95% CI 0.45–0.91 and in women) [28]. A second study used data from approximately 400,000 participants (age range of 35–70 years) in the European Prospective Investigation into Cancer and Nutrition and constructed a HLI using dose-dependent scores (0–4) to quantify the level of smoking status, alcohol intake, waist-to-hip ratio, and diet based on the intake of cereal fiber, vegetables and fruits, red and processed meat, polyunsaturated to saturated fat ratio, trans fat, and glycemic load [29]. Over 15 years of follow-up, a total of 1,113 cases of pancreatic cancer (634 in women) were diagnosed, and an inverse association was found between the highest HLI score (> 15 points) and risk of pancreatic cancer (HR 0.79, 95% CI 0.73–0.86, using as the reference group the second lowest HLI category (5–9 points)). Among women, each 1 standard deviation increment in the HLI score was associated with a 21% reduction in the risk of pancreatic cancer (HR 0.79, 95% CI 0.73–0.86). The results of our study are in line with those of previous studies, indicating that a healthy lifestyle score based on several modifiable factors is associated with reduced risk of pancreatic cancer. There is a complex relationship between type 2 diabetes and pancreatic cancer. Diabetes of long duration (> 3 years) is a major risk factor for pancreatic cancer [45] and is associated with a 50–150% increased risk. However, evidence suggests that for individuals who received a diabetes diagnosis less than two years before the diagnosis of pancreatic cancer, a reverse temporal relationship between these two diseases might have occurred [46]. In the present study, we showed for the first time that a healthy lifestyle is associated with a lower risk of pancreatic cancer among non-diabetic women. Information on type 2 diabetes was available at baseline and throughout the study follow-up and was used to define time-at-risk for participants who developed diabetes subsequent to the baseline exam. This approach reduced the possibility that the observed lower risk of pancreatic cancer in women with high healthy lifestyle index was due to the association of HLI with type 2 diabetes [27]. Additional studies are needed to confirm whether a healthy lifestyle may lower the risk of pancreatic cancer in women without diabetes over a prolonged period. There were too few cases of pancreatic cancer among diabetic women to allow us to examine the association between HLI and pancreatic cancer in this sub-group. As with non-diabetics, a healthy lifestyle was associated with a reduction in risk of pancreatic cancer among women who were non-smokers. Cigarette smoking is an established risk factor for pancreatic cancer and is estimated to account for approximately 20–25% of cases [45]. In the WHI cohort, only 7% of participants were current smokers. In analyses stratified by BMI categories, we showed that among women with normal BMI, there was a statistically significant inverse association between the HLI upper quartile and pancreatic cancer risk, while in overweight and obese women, the results were not significant. These findings may be due to small sample sizes within the higher BMI categories. The results obtained with the exclusion of each individual component in turn from the HLI composite score showed that the protective effect of high HLI does not depend upon one particular factor but rather on the combination of multiple lifestyle components. However, we recognize that characteristics such as smoking and high BMI may have a stronger association than the other components of the HLI with this outcome. The present study, the largest with respect to the number of participating women and the number of incident pancreatic cancer cases, was based on a well characterized cohort with detailed information on risk factors, a prospective design, and an average follow-up period of 16 years. The relative higher incidence rate of pancreatic cancer observed in this cohort compared to other studies [28, 29] was likely due to the higher median age of this cohort at enrollment and the extended duration of follow-up. Objective anthropometric measurements were obtained during the study baseline exam for all participants, while previous studies relied on self-reported data in part or in full [28, 29]. Cancer cases were confirmed centrally by trained physicians according to standard guidelines. The healthy lifestyle index was based on a multilevel risk score, which allowed us to evaluate the association of HLI with the outcome in a dose-dependent manner. Although the estimates of association were adjusted for several risk factors, we cannot exclude the possibility of residual confounding. Furthermore, we did not have repeated measurements of lifestyle factors throughout follow-up, which would have allowed for better characterization and control for lifestyle factors changes. In conclusion, we found that in this cohort of postmenopausal women, a relatively high healthy lifestyle index is associated with reduced risk of pancreatic cancer over a follow-up period of more than a decade. Pancreatic cancer remains one of the deadliest types of malignancy in the United States and worldwide. The results of our study suggest that promoting smoking cessation and limiting alcohol consumption, along with undertaking regular physical activity, consuming a healthy diet rich in fruits, vegetables, and whole grains and low in saturated and trans fats, and maintaining a healthy body weight, may be important steps to lower the risk of this lethal cancer. Supplementary Material Supplementary Acknowledgments We thank the Women’s Health Initiative investigators, staff, and the trial participants for their outstanding dedication and commitment. Program Office: (National Heart, Lung, and Blood Institute, Bethesda, MD) Jacques Roscoe, Shari Ludlum, Dale Burden, Joan McGowan, Leslie Ford, and Nancy Geller. Clinical Coordinating Center: (Fred Hutchinson Cancer Research Center, Seattle, WA) Garnet Anderson, Ross Prentice, Andrea LaCroix, and Charles Kopperberg). Investigators and Academic Centers: (Brigham and Women’s Hospital, Harvard Medical School, Boston, MA) JoAnn E, Manson; (MedStar Health Research Institute/Howard University, Washington, DC) Barbara V Howard; (Stanford Prevention Research Center, Stanford, CA) Marcia L. Stefanick; (The Ohio State University, Columbus, OH) Rebecca Jackson; (University of Arizona, Tucson/Phoenix, AZ) Cynthia A. Thompson; (University at Buffalo, Buffalo, NY) Jean Wactawski-Wende; (University of Florida, Gainesville/Jacksonville, FL) Marian Limacher; (University of Iowa, Iowa City/Davenport, IA) Robert Wallace; (University of Pittsburgh, Pittsburgh, PA) Lewis Kuller; (City of Hope Comprehensive Cancer Center, Duarte, CA) Rowan T. Chlebowski; (Wake Forest University School of Medicine, Winston–Salem, NC) Sally Shumaker. Women’s Health Initiative Memory Study: (Wake Forest University School of Medicine, Winston Salem, NC) Sally Shumaker. A full list of all the investigators who have contributed to Women’s Health Initiative science appears at https://www.whi.org/researchers/Documents%20%20Write%20a%20Paper/WHI%20Investigator%20Long%20List.pdf Funding This work was supported by Women’s Health Initiative. The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services through contracts HHSN268201600018C, H H S N 2 6 8 2 0 1 6 0 0 0 0 1 C, H H S N 2 6 8 2 0 1 6 0 0 0 0 2 C, HHSN268201600003C, and HHSN268201600004C. This work was supported by the National Cancer Institute (F99CA253745 to X.Z.). Data availability Data are available through the WHI online resource, https://www.whi.org/researchers/data/Pages/Home.aspx, and the WHI remains funded indefinitely through BioLINCC, https://biolincc.nhlbi.nih.gov/studies/whi_ctos/. Table 1 Baseline characteristics of participants in the Women’s Health Initiative by Healthy Lifestyle Index Healthy lifestyle index quartile 1st 2nd 3rd 4th p valueb N 31,301 32,131 33,902 39,611 HLI range 0–9 10–11 12–13 ≥ 14 Characteristics Age at enrollment,a years 62.3 (7.1) 63.3 (7.2) 63.7 (7.3) 63.8 (7.3) < 0.001 Age at diagnosis,c years 77.5 (8.6) 79.3 (8.5) 80.2 (8.5) 80.7 (8.4) < 0.001 Ethnicity, % White 79.0 81.3 84.0 85.6 < 0.001 Black 13.9 10.2 7.5 4.9 Others 7.1 8.5 8.4 9.5 Education, % ≤ High school 30.9 26.1 21.1 14.7 < 0.001 Some college 41.0 39.3 37.1 33.7 ≥ College degree 26.0 32.1 38.7 47.1 Marital status, % Single, divorced, separated, widow 42.4 38.9 35.8 33.9 < 0.001 Married or marriage-like relationship 57.6 61.1 63.2 66.1 WHI study enrollment, % Observational study 54.4 60.0 67.0 76.7 < 0.001 Calcium and vitamin D trial 24.6 21.9 18.5 21.8 < 0.001 Dietary modification triald 28.3 25.2 19.8 12.9 < 0.001 Hormone therapy trial 23.4 19.3 15.9 11.8 < 0.001 Estrogen-alone intervention arm 5.0 4,0 3.0 1.9 < 0.001 Estrogen + Progesterone intervention arm 6.9 5.9 5.0 4.1 < 0.001 History of type 2 diabetes, % 2.0 1.6 1.4 1.1 < 0.001 Diabetes medications, % 1.1 0.8 0.6 0.5 < 0.001 History of pancreatitis, % 0.9 0.9 0.8 0.6 < 0.001 Total non-alcohol dietary energy,a kcal/day 1747.6 (792.0) 1618.4 (732.5) 1564.8 (699.2) 1523.4 (638.5) < 0.001 Body mass index,a kg/m2 31.2 (6.4) 29.1 (5.9) 27.3 (5.3) 25.0 (4.1) , < 0.001 Waist circumference,a cm 81.9 (17.5) 76.2 (16.0) 71.5 (14.3) 65.4 (11.4) < 0.001 Alcohol,a g/day 8.3 (16.5) 5.1 (10.6) 4.6 (8.1) 3.5 (5.5) < 0.001 Diet score (AHEI-2010), highest quintile, % 1.6 7.1 18.3 46.2 < 0.001 Never Smoker, % 26.6 49.6 55.2 67.0 < 0.001 Physical activity,a MET-hours/week 3.1 6.8 12.6 23.0 < 0.001 HLI healthy lifestyle index, MET metabolic energy equivalent a Data represent mean (standard deviation) b p value based on Wilcoxon rank-sum tests for continuous variables and χ2 tests for categorical variables c Mean age at diagnosis among pancreatic cancer cases d Among women in the control arm of the trial Table 2 Incidence rates and hazard ratios (95% confidence intervals) for risk of pancreatic cancer in association with the Healthy Lifestyle Index Healthy lifestyle index quartile 1st 2nd 3rd 4th p trend a Total study population N 31,301 32,131 33,902 39,611 Pancreatic cancer cases 267 262 285 305 Incidence rate per 1000 person-yearsb 0.56 0.51 0.51 0.45 HR (95% CI) 1.00 0.84 (0.71–0.99) 0.81 (0.68–0.96) 0.70 (0.60–0.83) < 0.001 HR (95% CI)c 1.00 0.85 (0.71–1.01) 0.83 (0.74–0.99) 0.74 (0.62–0.88) 0.001 HR (95% CI) for 1 SD increase in HLIc, d 0.89 (0.84–0.95) Using waist circumference in the HLI e N 33,995 32,551 33,761 39,249 Pancreatic cancer cases 297 268 279 297 Incidence rate per 1000 person-yearsb 0.58 0.52 0.50 0.44 HR (95% CI) 1.00 0.83 (0.71–0.98) 0.78 (0.66–0.92) 0.69 (0.58–0.81) < 0.001 HR (95% CI)c, f 1.00 0.84 (0.70–0.99) 0.80 (0.68–0.95) 0.72 (0.61–0.85) 0.001 HR (95% CI) for 1 SD increase in HLIc, d 0.91 (0.85–0.98) Excluding women with follow-up < 2 years N 30,696 31,685 33,510 39,241 Pancreatic cancer cases 248 241 275 286 Incidence rate per 1000 person-yearsb 0.52 0.47 0.50 0.43 HR (95% CI) 1.00 0.83 (0.69–0.99) 0.84 (0.70–0.99) 0.71 (0.60–0.84) < 0.001 HR (95% CI)c 1.00 0.84 (0.69–1.00) 0.86 (0.72–1.02) 0.73 (0.61–0.88) 0.002 HR (95% CI) for 1 SD increase in HLIc, d 0.90 (0.84–0.96) Non-diabetics N 30,818 31,725 33,499 39,222 Pancreatic cancer cases 190 209 240 270 Incidence rate per 1000 person-yearsb 0.47 0.46 0.48 0.43 HR (95% CI) 1.00 0.90 (0.7–1.10) 0.90 (0.74–1.09) 0.79 (0.68–0.96) 0.018 HR (95% CI)c 1.00 0.90 (0.71–1.10) 0.89 (0.73–1.08) 0.78 (0.64–0.95) 0.015 HR (95% CI) for 1 SD increase in HLIc, d 0.91 (0.85–0.97) Non-smokers N 24,383 30,635 33,328 39,615 Pancreatic cancer cases 194 250 277 304 Incidence rate per 1000 person-yearsb 0.52 0.51 0.50 0.45 HR (95% CI) 1.00 0.93 (0.77–1.12) 0.89 (0.74–1.07) 0.79 (0.66–0.94) 0.006 HR (95% CI)c 1.00 0.94 (0.78–1.13) 0.91 (0.75–1.09) 0.80 (0.66–0.97) 0.017 HR (95% CI) for 1 SD increase in HLIc, d 0.92 (0.86–0.99) HR hazard ratio, CI confidence interval, SD standard deviation Hazard ratios and 95% confidence intervals estimates adjusted for age at enrollment a p value based on Wilcoxon rank-sum tests for continuous variables and χ2 tests for categorical variables b Unadjusted incidence rate c Model adjusted for age, education, marital status, hormone replacement therapy clinical trial participation and arm, calcium plus vitamin D clinical trial participation and arm, dietary modification clinical trial non-intervention arm, diabetes status and medication, pancreatitis, and total non-alcohol dietary energy d Hazard ratio and 95% confidence interval associated with an increase in the HLI of one standard deviation e The HLI score includes waist circumference categories (score 0: ≥ 88 cm; 1: 80– < 88 cm; 2: < 80 cm) instead of BMI f Model adjusted as for modela plus height Table 3 Incidence rates and hazard ratios (95% confidence intervals) for risk of pancreatic cancer in association with the HLIwst by body mass index categories Healthy lifestyle index quartilea 1st 2nd 3rd 4th p trend BMI 18.5–<25.0 (kg/m2) N 5,729 8,421 12,650 22,437 Pancreatic cancer cases 49 70 108 164 Incidence rate per 1000 person-yearsb 0.57 0.51 0.51 0.42 HR (95% CI) 1.00 0.83 (0.58–1.20) 0.80 (0.57–1.12) 0.66 (0.48–0.91) 0.006 HR (95% CI)c 1.00 0.84 (0.58–1.20) 0.80 (0.56–1.12) 0.65 (0.47–0.90) 0.006 BMI 25.0– <30.0 (kg/m2) N 11,177 12,026 12,742 12,066 Pancreatic cancer cases 102 99 108 102 Incidence rate per 1000 person-yearsb 0.59 0.51 0.51 0.50 HR (95% CI) 1.00 0.82 (0.62–1.08) 0.79 (0.60–1.04) 0.78 (0.59–1.02) 0.083 HR (95% CI)c 1.00 0.81 (0.61–1.07) 0.79 (0.60–1.04) 0.78 (0.58–1.03) 0.091 BMI> 30.0 (kg/m2) N 16,780 11,779 7,903 3,833 Pancreatic cancer cases 141 97 52 27 Incidence rate per 1000 person-yearsb 0.57 0.54 0.42 0.44 HR (95% CI) 1.00 0.89 (0.69–1.15) 0.68 (0.50–0.94) 0.71 (0.47–1.07) 0.012 HR (95% CI)c 1.00 0.91 (0.70–1.18) 0.72 (0.52–0.99) 0.77 (0.50–1.17) 0.045 Interaction term between BMI group*HLI quartiles p value = 0.739 HLI healthy lifestyle index, wst waist circumference, HR hazard ratio, CI confidence interval, BMI body mass index All models were adjusted for age at enrollment in the study a Healthy lifestyle index calculated using categories of waist circumference (score 0: ≥ 88 cm; 1: 80− < 88 cm; 2: < 80 cm) instead of BMI b Unadjusted incidence rate c Model adjusted for age, height, education, marital status, hormone replacement therapy clinical trial participation and arm, calcium plus vitamin D clinical trial participation and arm, dietary modification clinical trial non-intervention arm, diabetes status and medication, pancreatitis, and total non-alcohol energy intake Table 4 Hazard ratios (95% confidence intervals) for risk of pancreatic cancer in association with the Healthy Lifestyle Index, excluding each lifestyle component in turn Healthy lifestyle index quartile 1st 2nd 3rd 4th p trend HLI without smoking Incidence rate per 1000 person-yearsa 0.54 0.51 0.53 0.46 HR (95% CI)b 0.89 (0.75–1.06) 0.89 (0.76–1.05) 0.78 (0.65–0.92) 0.005 HR (95% CI)c,d 1.00 0..91 (0.77–1.09) 0.93 (0.78–1.10) 0.83 (0.69–0.99) 0.059 HR (95% CI) for 1 SD increase in HLIb,c,d,e 0.94 (0.89–1.00) HLI without diet Incidence rate per 1000 person-yearsa 0.58 0.47 0.49 0.47 HR (95% CI)b 1.00 0.77 (0.63–0.92) 0.79 (0.68–0.92) 0.73 (0.62–0.86) < 0.001 HR (95% CI)c,f 1.00 0.77 (0.64–0.93) 0.81 (0.70–0.95) 0.76 (0.64–0.90) 0.002 HR (95% CI) for 1 SD increase in HLIb,d,e,f 0.88 (0.83–0.94) HLI without physical activity Incidence rate per 1000 person-yearsa 0.61 0.53 0.47 0.47 HR (95% CI)b 1.00 0.81 (0.67–0.97) 0.69 (0.58–0.83) 0.66 (0.54–0.80) < 0.001 HR (95% CI)c,g 1.00 0.81 (0.67–0.97) 0.70 (0.58–0.84) 0.67 (0.55–0.82) < 0.001 HR (95% CI) for 1 SD increase in HLIb,d,e,g 0.87 (0.82–0.93) HLI without alcohol Incidence rate per 1000 person-yearsa 0.54 0.50 0.50 0.49 HR (95% CI)b 1.00 0.85 (0.72–.1.00) 0.82 (0.70–0.97) 0.80 (0.68–0.94) 0.006 HR (95% CI)c,h 1.00 0.87 (0.74–1.02) 0.84 (0.71–0.99) 0.83 (0.70–0.99) 0.035 HR (95% CI) for 1 SD increase in HLIb,c,e,h 0.89 (0.84–0.96) HLI without body mass index Incidence rate per 1000 person-yearsa 0.54 0.54 0.50 0.46 HR (95% CI)b 1.00 0.93 (0.78–1.10) 0.83 (0.70–0.99) 0.76 (0.64–0.90) 0.001 HR (95% CI)c,i 1.00 0.91 (0.76–1.08) 0.84 (0.71–1.00) 0.78 (0.65–0.94) 0.006 HR (95% CI) for 1 SD increase in HLIb,c,e,i 0.89 (0.83–0.94) a Unadjusted incidence rate b Model adjusted for age at enrollment in the study c Model adjusted for age, education, marital status, hormone replacement therapy clinical trial participation and arm, calcium plus vitamin D clinical trial participation and arm, dietary modification clinical trial non-intervention arm, diabetes status and medication, pancreatitis, and total no-alcohol energy intake d Model additionally adjusted for smoking categories e Hazard ratio and 95% confidence interval associated with an increase in the HLI of one standard deviation f Model additionally adjusted for quintiles of diet score (AHEI-2010) g Model additionally adjusted for quintiles of physical activity h Model additionally adjusted for alcohol intake categories i Model additionally adjusted for body mass index categories Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s10552-022-01558-x. Conflict of interest The authors declare that they have no conflict of interest. Ethical approval All procedures in the study were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments. The study was approved by institutional review boards at all 40 clinical centers and at the coordinating center. Informed consent All participants in the WHI gave written-informed consent. References 1. Siegel RL , Miller KD , Jemal A (2020) Cancer statistics, 2020. CA Cancer J Clin 70 :7–30. 10.3322/caac.21590 31912902 2. McGuigan A , Kelly P , Turkington RC , Jones C , Coleman HG , McCain RS (2018) Pancreatic cancer: A review of clinical diagnosis, epidemiology, treatment and outcomes. World J Gastroenterol 24 :4846–4861. 10.3748/wjg.v24.i43.4846 30487695 3. Kleeff J , Korc M , Apte M , La Vecchia C , Johnson CD , Biankin AV (2016) Pancreatic cancer. Nat Rev Dis Primers 2 :16022. 10.1038/nrdp.2016.22 27158978 4. Ducreux M , Seufferlein T , Van Laethem JL , Laurent-Puig P , Smo-lenschi C , Malka D (2019) Systemic treatment of pancreatic cancer revisited. Semin Oncol 46 :28–38. 10.1053/j.seminoncol.2018.12.003 30638624 5. The global, regional, and national burden of pancreatic cancer and its attributable risk factors in 195 countries and territories, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet Gastroenterol Hepatol. 4 (2019) 934–947. 10.1016/s24681253(19)30347-4.31648972 6. Maisonneuve P , Lowenfels AB (2015) Risk factors for pancreatic cancer: a summary review of meta-analytical studies. Int J Epidemiol 44 :186–198. 10.1093/ije/dyu240 25502106 7. Midha S , Chawla S , Garg PK (2016) Modifiable and non-modifiable risk factors for pancreatic cancer: a review. Cancer Lett 381 :269–277. 10.1016/j.canlet.2016.07.022 27461582 8. Koyanagi YN , Ito H , Matsuo K , Sugawara Y , Hidaka A , Sawada N (2019) Smoking and pancreatic cancer incidence: a pooled analysis of 10 population-based cohort studies in Japan. Cancer Epidemiol Biomarkers Prev 28 :1370–1378. 10.1158/1055-9965.Epi-18-1327 31113869 9. Cigarette smoking and pancreatic cancer: an analysis from the International Pancreatic Cancer Case-Control Consortium (Panc4). Ann Oncol. 23 (2012) 2773. 10.1093/annonc/mds491.32937733 10. Chuang SC , Gallo V , Michaud D , Overvad K , Tjønneland A , Clavel-Chapelon F (2011) Exposure to environmental tobacco smoke in childhood and incidence of cancer in adulthood in never smokers in the European Prospective Investigation into Cancer and Nutrition. Cancer Causes Control 22 :487–494. 10.1007/s10552-010-9723-2 21279734 11. Rebours V , Gaujoux S , d’Assignies G , Sauvanet A , Ruszniewski P , Lévy P (2015) Obesity and fatty pancreatic infiltration are risk factors for pancreatic precancerous lesions (PanIN). Clin Cancer Res 21 :3522–3528. 10.1158/1078-0432.Ccr-14-2385 25700304 12. Berrington de Gonzalez A , Sweetland S , Spencer E (2003) A meta-analysis of obesity and the risk of pancreatic cancer. Br J Cancer 89 :519–523. 10.1038/sj.bjc.6601140 12888824 13. Casari I , Falasca M (2015) Diet and Pancreatic Cancer Prevention. Cancers (Basel) 7 :2309–2317. 10.3390/cancers7040892 26610570 14. Hine RJ , Srivastava S , Milner JA , Ross SA (2003) Nutritional links to plausible mechanisms underlying pancreatic cancer: a conference report. Pancreas 27 :356–366. 10.1097/00006676-200311000-00014 14576500 15. Nöthlings U , Wilkens LR , Murphy SP , Hankin JH , Henderson BE , Kolonel LN (2005) Meat and fat intake as risk factors for pancreatic cancer: the multiethnic cohort study. J Natl Cancer Inst 97 :1458–1465. 10.1093/jnci/dji292 16204695 16. Harris DM , Srihari P , Go VLW (2011) Pancreatic Cancer Prevention and the 2010 Dietary Guidelines for Americans. 40 :641–643. 10.1097/MPA.0b013e3182203e3c 17. Zheng J , Guinter MA , Merchant AT , Wirth MD , Zhang J , Stolzenberg-Solomon RZ (2017) Dietary patterns and risk of pancreatic cancer: a systematic review. Nutr Rev 75 :883–908. 10.1093/nutrit/nux038 29025004 18. Wang YT , Gou YW , Jin WW , Xiao M , Fang HY (2016) Association between alcohol intake and the risk of pancreatic cancer: a dose-response meta-analysis of cohort studies. BMC Cancer 16 :212. 10.1186/s12885-016-2241-1 26968702 19. Behrens G , Jochem C , Schmid D , Keimling M , Ricci C , Leitz-mann MF (2015) Physical activity and risk of pancreatic cancer: a systematic review and meta-analysis. Eur J Epidemiol 30 :279–298. 10.1007/s10654-015-0014-9 25773752 20. Maisonneuve P (2019) Epidemiology and burden of pancreatic cancer. Presse Med 48 :e113–e123. 10.1016/j.lpm.2019.02.030 30878335 21. Islami F , Goding Sauer A , Miller KD , Siegel RL , Fedewa SA , Jacobs EJ (2018) Proportion and number of cancer cases and deaths attributable to potentially modifiable risk factors in the United States. CA Cancer J Clin 68 :31–54. 10.3322/caac.21440 29160902 22. Song M , Giovannucci E (2020) Preventable incidence of carcinoma associated with adiposity, alcohol and physical inactivity according to smoking status in the United States. Int J Cancer 146 :2960–2967. 10.1002/ijc.32602 31369145 23. Ghosn B , Benisi-Kohansal S , Ebrahimpour-Koujan S , Azadbakht L , Esmaillzadeh A (2020) Association between healthy lifestyle score and breast cancer. Nutr J 19 :4. 10.1186/s12937-020-0520-9 31937325 24. Arthur R , Brasky TM , Crane TE , Felix AS , Kaunitz AM , Shadyab AH (2019) Associations of a healthy lifestyle index with the risks of endometrial and ovarian cancer among women in the women’s health initiative study. Am J Epidemiol 188 :261–273. 10.1093/aje/kwy249 25. Dartois L , Fagherazzi G , Boutron-Ruault MC , Mesrine S , Clavel-Chapelon F (2014) Association between five lifestyle habits and cancer risk: results from the E3N cohort. Cancer Prev Res (Phila) 7 :516–525. 10.1158/1940-6207.Capr-13-0325 24574508 26. Barbaresko J , Rienks J , Nöthlings U (2018) Lifestyle indices and cardiovascular disease risk: a meta-analysis. Am J Prev Med 55 :555–564. 10.1016/j.amepre.2018.04.046 30241617 27. Li Y , Schoufour J , Wang DD , Dhana K , Pan A , Liu X (2020) Healthy lifestyle and life expectancy free of cancer, cardiovascular disease, and type 2 diabetes: prospective cohort study. BMJ 368 :l6669. 10.1136/bmj.l6669 31915124 28. Jiao L , Mitrou PN , Reedy J , Graubard BI , Hollenbeck AR , Schatzkin A (2009) A combined healthy lifestyle score and risk of pancreatic cancer in a large cohort study. Arch Intern Med 169 :764–770. 10.1001/archinternmed.2009.46 19398688 29. Naudin S , Viallon V , Hashim D , Freisling H , Jenab M , Weiderpass E (2020) Healthy lifestyle and the risk of pancreatic cancer in the EPIC study. Eur J Epidemiol 35 :975–986. 10.1007/s10654-019-00559-6 31564045 30. (1998) Design of the Women’s Health Initiative clinical trial and observational study. The Women’s Health Initiative Study Group. Control Clin Trials. 19 :61–109. 10.1016/s0197-2456(97)00078-0.9492970 31. Langer RD , White E , Lewis CE , Kotchen JM , Hendrix SL , Trevisan M (2003) The Women’s Health Initiative Observational Study: baseline characteristics of participants and reliability of baseline measures. Ann Epidemiol 13 :S107–S121. 10.1016/s1047-2797(03)00047-4 14575943 32. Arthur R , Wassertheil-Smoller S , Manson JE , Luo J , Snetselaar L , Hastert T (2018) The combined association of modifiable risk factors with breast cancer risk in the women’s health initiative. Cancer Prev Res (Phila) 11 :317–326. 10.1158/1940-6207.Capr-17-0347 29483073 33. Patterson RE , Kristal AR , Tinker LF , Carter RA , Bolton MP , Agurs-Collins T (1999) Measurement characteristics of the Women’s Health Initiative food frequency questionnaire. Ann Epidemiol 9 :178–187. 10.1016/s1047-2797(98)00055-6 10192650 34. Belin RJ , Greenland P , Allison M , Martin L , Shikany JM , Larson J (2011) Diet quality and the risk of cardiovascular disease: the Women’s Health Initiative (WHI). Am J Clin Nutr 94 :49–57. 10.3945/ajcn.110.011221 21613562 35. McCullough ML , Feskanich D , Stampfer MJ , Giovannucci EL , Rimm EB , Hu FB (2002) Diet quality and major chronic disease risk in men and women: moving toward improved dietary guidance. Am J Clin Nutr 76 :1261–1271. 10.1093/ajcn/76.6.1261 12450892 36. Chiuve SE , Fung TT , Rimm EB , Hu FB , McCullough ML , Wang M (2012) Alternative dietary indices both strongly predict risk of chronic disease. J Nutr 142 :1009–1018. 10.3945/jn.111.157222 22513989 37. Ainsworth BE , Haskell WL , Leon AS , Jacobs DR Jr , Montoye HJ , Sallis JF (1993) Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc 25 :71–80. 10.1249/00005768-199301000-00011 8292105 38. Aune D , Greenwood DC , Chan DS , Vieira R , Vieira AR , Navarro Rosenblatt DA (2012) Body mass index, abdominal fatness and pancreatic cancer risk: a systematic review and non-linear dose-response meta-analysis of prospective studies. Ann Oncol 23 :843–852. 10.1093/annonc/mdr398 21890910 39. Liu S , Tinker L , Song Y , Rifai N , Bonds DE , Cook NR (2007) A prospective study of inflammatory cytokines and diabetes mellitus in a multiethnic cohort of postmenopausal women. Arch Intern Med 167 :1676–1685. 10.1001/archinte.167.15.1676 17698692 40. Curb JD , McTiernan A , Heckbert SR , Kooperberg C , Stanford J , Nevitt M (2003) Outcomes ascertainment and adjudication methods in the Women’s Health Initiative. Ann Epidemiol 13 :S122–S128. 10.1016/s1047-2797(03)00048-6 14575944 41. Everhart J , Wright D (1995) Diabetes mellitus as a risk factor for pancreatic cancer. A meta-analysis. JAMA 273 :1605–1609 7745774 42. Eibl G , Cruz-Monserrate Z , Korc M , Petrov MS , Goodarzi MO , Fisher WE (2018) Diabetes mellitus and obesity as risk factors for pancreatic cancer. J Acad Nutr Diet 118 :555–567. 10.1016/j.jand.2017.07.005 28919082 43. Toi PL , Anothaisintawee T , Chaikledkaew U , Briones JR , Reutrakul S , Thakkinstian A (2020) Preventive role of diet interventions and dietary factors in type 2 diabetes mellitus: an umbrella review. Nutrients. 10.3390/nu12092722 44. Jin Q , Hart PA , Shi N , Joseph JJ , Donneyong M , Conwell DL (2021) Dietary patterns of insulinemia, inflammation and glycemia, and pancreatic cancer risk: findings from the women’s health initiative. Cancer Epidemiol Biomarkers Prev 30 :1229–1240. 10.1158/1055-9965.Epi-20-1478 33827986 45. Klein AP (2021) Pancreatic cancer epidemiology: understanding the role of lifestyle and inherited risk factors. Nat Rev Gastroenterol Hepatol 18 :493–502. 10.1038/s41575-021-00457-x 34002083 46. Andersen DK , Korc M , Petersen GM , Eibl G , Li D , Rickels MR (2017) Diabetes, pancreatogenic diabetes, and pancreatic cancer. Diabetes 66 :1103–1110. 10.2337/db16-1477 28507210

PMC010xxxxxx/PMC10361418.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9918300877606676 51148 SSM Qual Res Health SSM Qual Res Health SSM. Qualitative research in health 2667-3215 37483653 10361418 10.1016/j.ssmqr.2022.100204 NIHMS1913647 Article Mechanisms to enhance racial equity in health care: Developing a model to facilitate translation of the ACCURE intervention Griesemer Ida ab* Birken Sarah A. c Rini Christine d Maman Suzanne e John Randall f Thatcher Kari b Dixon Crystal bg Yongue Christina bh Baker Stephanie bi Bosire Claire b Garikipati Aditi b Ryals Cleo A. bfjk Lightfoot Alexandra F. bel a US Department of Veterans Affairs, Center for Healthcare Organization and Implementation Research (CHOIR), VA Boston Healthcare System, 150 South Huntington Avenue (152M), Jamaica Plain Campus, Building 9, Boston, MA, 02130, USA b Greensboro Health Disparities Collaborative, 301 S. Elm Street, Suite 414, Greensboro, NC, 27401, USA c Department of Implementation Science, Wake Forest School of Medicine, 300 Medical Center Blvd, Winston-Salem, NC, 27157, USA d Department of Medical Social Sciences, Feinberg School of Medicine, Northwestern University, 420 E Superior St, Chicago, IL, 60611, USA e Department of Health Behavior, University of North Carolina, 135 Dauer Drive, Chapel Hill, NC, 27599, USA f Department of Health Policy and Management, University of North Carolina, 135 Dauer Drive, Chapel Hill, NC, 27599, USA g Department of Health and Exercise Science, Wake Forest University, 1834 Wake Forest Rd., Winston-Salem, NC, 27109, USA h Department of Public Health Education, University of North Carolina, 1408 Walker Ave # 437, Greensboro, NC, 27412, USA i Department of Public Health Studies, Elon University, 100 Campus Drive, Elon, NC, 27244, USA j Lineberger Comprehensive Cancer Center, University of North Carolina, 450 West Dr, Chapel Hill, NC, 27599, USA k Flatiron Health, 233 Spring St., New York, NY, 10013, USA l Center for Health Promotion and Disease Prevention, University of North Carolina, 1700 MLK Jr Blvd Ste 7426, Chapel Hill, NC, 27599, USA * Corresponding author. US Department of Veterans Affairs, Center for Healthcare Organization and Implementation Research (CHOIR), VA Boston Healthcare System, 150 South Huntington Avenue (152M), Jamaica Plain Campus, Building 9, Boston, MA, 02130, USA. Ida.Griesemer@va.gov (I. Griesemer). 13 7 2023 6 2023 02 12 2022 21 7 2023 3 100204This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Background: As medical and public health professional organizations call on researchers and policy makers to address structural racism in health care, guidance on evidence-based interventions to enhance health care equity is needed. The most promising organizational change interventions to reduce racial health disparities use multilevel approaches and are tailored to specific settings. This study examines the Accountability for Cancer Care through Undoing Racism and Equity (ACCURE) intervention, which changed systems of care at two U.S. cancer centers and eliminated the Black-White racial disparity in treatment completion among patients with early-stage breast and lung cancer. Purpose: We aimed to document key characteristics of ACCURE to facilitate translation of the intervention in other care settings. Methods: We conducted semi-structured interviews with participants who were involved in the design and implementation of ACCURE and analyzed their responses to identify the intervention’s mechanisms of change and key components. Results: Study participants (n = 18) described transparency and accountability as mechanisms of change that were operationalized through ACCURE’s key components. Intervention components were designed to enhance either institutional transparency (e.g., a data system that facilitated real-time reporting of quality metrics disaggregated by patient race) or accountability of the care system to community values and patient needs for minimally biased, tailored communication and support (e.g., nurse navigators with training in antiracism and proactive care protocols). Conclusions: The antiracism principles transparency and accountability may be effective change mechanisms in equity-focused health services interventions. The model presented in this study can guide future research aiming to adapt ACCURE and evaluate the intervention’s implementation and effectiveness in new settings and patient populations. Cancer care Quality improvement Intervention Planned adaptation Community-based participatory research Antiracism pmc1. Introduction Since publication of the 2002 Institute of Medicine report titled “Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care,” (Nelson et al., 2002) evidence of racial disparities in health outcomes has steadily accumulated and remains relevant in every category of illness. In the case of cancer, one of the leading causes of mortality in the U.S., death rates are higher among Black Americans than any other racial or ethnic group for most cancers (DeSantis et al., 2019). Over the last two decades, public health researchers and advocates have examined the role of racism as a social determinant of health and a driving factor in race-based health disparities (Came & Griffith, 2018; Jones, 2002, pp. 7–22; Largent, 2018). Experts define racism as a hierarchical social system in which the dominant group uses social power to systematically advantage racial in-group members and simultaneously oppress group members defined as inferior (Bailey et al., 2021; Williams et al., 2019). Institutional racism is the manifestation of a race-based system of advantage and disadvantage in the policies and practices of institutions (Bassett & Graves, 2018). Scholars have highlighted ways in which historical policies that exemplify institutional racism, such as hospital segregation, continue to have implications for the unequal treatment of people of color today (Bailey et al., 2017; Largent, 2018). Acknowledging systemic aspects of racism that are embedded in the social environment and institutional cultures, researchers have called for public health strategies that promote organizational change in order to create more equitable care outcomes (Bailey et al., 2021; Bassett & Graves, 2018; Griffith, 2010). Despite the growing literature on the need for institutional change to address racial disparities in health care, research on interventions aimed at reducing cancer disparities remains primarily focused on changing patient knowledge and behavior, as opposed to examining health care systems where patients receive care (Hardeman, 2020). A recent policy statement from the American Society of Clinical Oncology called on researchers to advance health equity by implementing evidence-based interventions that address structural barriers to accessing quality care (Patel et al., 2020). Health care settings can be observed and modified, and intervening at the organizational level, both within care systems and across systems and institutions, is an underutilized approach to address disparities in outcomes (Bassett & Graves, 2018; Griffith, Yonas, Mason, & Havens, 2010). By targeting systems of care, quality improvement interventions have the potential to address an underlying cause of racial disparities: institutional racism. In doing so, system-focused interventions are likely to be more sustainable and reach more patients than interventions targeting individuals. The most promising organizational change interventions to reduce racial health disparities use multilevel approaches and are tailored to the specific context (Chin et al., 2012; Hassen et al., 2021). The Accountability for Cancer Care through Undoing Racism and Equity (ACCURE) intervention, the parent trial for this study, was a multicomponent intervention that successfully eliminated the racial disparity in treatment completion rates between Black and White early-stage breast and lung cancer patients. The intervention took place at two U.S. cancer centers (Cone Health Cancer Center in Greensboro, NC, and Hillman Cancer Center in Pittsburgh, PA) and involved several strategies including training for providers on the root causes of health care inequities (Black et al., 2019), a data system (Real-Time Registry) that tracked patient progress in real time disaggregated by patient race (Cykert et al., 2020), and nurse navigators who supported patient engagement through data-informed care (Griesemer et al.). Results from ACCURE showed that prior to implementing the intervention, treatment completion rates in the population cohort at the two cancer centers (n = 8,945) were 79.8% for Black patients vs. 87.3% for White patients (p < 0.001). After ACCURE, the racial disparity among patients in the intervention group (n = 302) was nonsignificant (Black patients 88.4% and White patients 89.5%, p = 0.77). Multivariate analyses confirmed the intervention yielded a significant reduction in this disparity (Black-White OR 0.98, 95% CI 0.46, 2.1). Details of the trial evaluating ACCURE have been reported in a previous publication (Cykert et al., 2020). Importantly, the ACCURE study used a Community-Based Participatory Research (CBPR) approach, meaning community members affected by racial disparities in health care were involved in every phase of the project (Wallerstein & Duran, 2006). The organization that designed the intervention and guided implementation was the Greensboro Health Disparities Collaborative (GHDC), a community-academic-medical research partnership founded in 2003 and based in Greensboro, NC. GHDC was established by Greensboro community members who wanted to understand and address racial disparities in health care through applied research. They interviewed and recruited academic researchers from the University of North Carolina to partner with them in applying for research funding from the National Institutes of Health (Yonas et al., 2006). That application led to the Cancer Care and Racial Equity Study (CCARES) (Yonas et al., 2013), which in turn led a larger grant that funded the development and implementation of ACCURE. Both studies were conducted through a CBPR process of organizing stakeholders from the local community and from medical and academics institutions to collaborate on system-change health disparities research and education. GHDC’s application of CBPR, including discussion of important issues such as navigating power dynamics and resource allocation, has been described in detail in prior publications (Black et al., 2021; Eng et al., 2017; Schaal et al., 2016; Yonas et al., 2013; Yonas et al., 2006). 1.1. Theoretical frameworks 1.1.1. Undoing Racism® To design the ACCURE intervention, GHDC drew from principles described in Undoing Racism®, a resource developed by a collective of antiracist organizers and educators (The People’s Institute for Survival and Beyond, 2018). Upon joining GHDC, members are required to participate in the Racial Equity Institute’s Phase 1 workshop, a two-day antiracism training based on Undoing Racism®. The purpose of the workshop is to foster a shared analysis of the systems and structures that maintain oppression. Building a shared analysis of systemic racism is a common antiracism organizing strategy (Came & Griffith, 2018) and is foundational to GHDC’s approach to conducting antiracist research in health care systems. CCARES, GHDC’s research study prior to ACCURE, was informed by the Undoing Racism® principle of analyzing power (Table 1). The researchers studied institutional power in Cone Health by examining data sources and the flow of information regarding patient outcomes. GHDC identified a lack of data transparency in care outcomes, which in turn obscured racial inequities in treatment engagement and completion. CCARES also involved qualitative interviews with patients from Cone Health to examine barriers to quality care and racial differences in how patients experienced care (Yonas et al., 2013). Building on the foundation of this prior research, GHDC designed the components of ACCURE to: (1) enhance data transparency, and (2) apply the Undoing Racism® principle of maintaining accountability to communities to a racial equity-focused intervention in cancer care (Table 1). To our knowledge, ACCURE is the first health services intervention to use Undoing Racism® to guide intervention design. The principles of transparency and accountability are well documented in ACCURE (Black et al., 2019; Cykert et al., 2020; Eng et al., 2017; Schaal et al., 2016), but there is a lack of literature linking these principles to existing public health models. As public health leaders call on the field to dismantle structural racism built into systems of care (Choo, 2021; Crear-Perry et al., 2020; Hardeman et al., 2016), new models for intervention development that incorporate antiracism principles could advance research on effective strategies for enhancing racial equity in the delivery of health care. 1.1.2. Planned Adaptation Model The present study was guided by the Planned Adaptation Model (Lee et al., 2008), which specifies a four-step process for adapting evidence-based interventions (Table 2). This model, along with research by Kirk and colleagues that offers methodological guidelines for applying the Planned Adaptation Model (2021)1, provides a systematic approach for documenting and adapting existing interventions so they can maintain effectiveness in new environments. The objective of the present study was to complete step one in the model: examine the original intervention’s theory of change by identifying mechanisms and key activities. We used first-hand narratives from those closest to the intervention to document how ACCURE changed systems of care. 1.1.3. Social Ecological Model In the analysis stage of this study, we applied the Social Ecological Model (SEM), a widely used model in health promotion programs (Ma et al., 2017; McCormack et al., 2017), to synthesize our results (see section 3.2). The SEM describes levels at which it’s possible to intervene to address public health issues (McLeroy et al., 1988). The model is often depicted as concentric circles (Fig. 1), with the outer circles representing greater spheres of potential influence (e.g., policy, community, and organizational levels), and the inner circles representing interventions targeting individuals (interpersonal and intrapersonal levels). Interventions are more likely to be effective when targeting multiple levels of the SEM, compared to interventions focusing only on the inter or intrapersonal level (Hassen et al., 2021; Kellou et al., 2014). The SEM informed the interpretation of findings and contributed to our overarching aim: to document key characteristics of ACCURE to facilitate translation of the intervention in new settings. 2. Methods This study was a post-hoc examination of the original ACCURE intervention, which took place from 2012 to 2018. The present study was conducted in 2019–2021 and involved three phases. We first reviewed materials documenting the ACCURE intervention, including publications describing the intervention development process and outcomes (Black et al., 20211; Cykert et al., 2020; Eng et al., 2017; Schaal et al., 2016) and procedure manuals. Next, we developed a semi-structured interview guide informed by the Planned Adaptation Model and previous research on documenting key characteristics of an evidence-based intervention (Kirk et al., 2021). We then recruited interview participants, conducted interviews, and analyzed the data to identify themes. The research process was enhanced by a Community Advisory Board (CAB) made up of five members of GHDC. While the original ACCURE study was a CBPR process conducted in full partnership with GHDC, the present study was designed by the first author, a GHDC member, with advising from GDHC and her academic mentors. She formed a CAB to ensure that study design and analysis remained aligned with and accountable to the interests and values of the larger GHDC. The CAB met three times over the course of the study and members were compensated $25 per meeting. In the first meeting, which took place during study development, the CAB provided input on the interview guide. After data collection was complete, the CAB met again to review the initial findings. In the final meeting, we discussed the overall interpretation of the findings and the framing of the results. In addition, the CAB served as a bridge between the activities of present study and the larger GHDC by bringing the CAB members’ perspectives on the history of GHDC and ACCURE into discussions about the present study, and by bringing updates about the present study back to GHDC at monthly meetings. CAB members were invited to contribute to dissemination products from the study including this manuscript. 2.1. Participants and recruitment We used purposive sampling to recruit participants who were actively involved in the design and implementation of ACCURE for in-depth interviews (Table 3). We aimed to recruit participants with a range of perspectives, including longstanding GHDC members with knowledge of prior research informing ACCURE and the rationale for the intervention, community-based research assistants directly involved with data collection, academic research staff who contributed to grant writing and project management, information technology specialists who designed the intervention’s data system, and cancer center administrators and providers who worked with researchers on study implementation, patient recruitment, and intervention delivery. Patient interviews were not conducted because the present study aimed to document ACCURE from the perspective of those who designed and delivered the intervention. CAB members were ineligible to participate in interviews. The first author consulted with ACCURE investigators to identify 20 potential participants with first-hand knowledge of the intervention. This pool represented the full group of individuals who had sufficient familiarity with ACCURE to provide insight into this study’s research questions. Participants were contacted by the first author via email and invited to participate in an interview about their involvement with ACCURE. If they agreed to an interview, the first author scheduled a time to meet. Potential participants who did not respond to the first email were sent a maximum of two follow-up emails. 2.2. Data collection The first author conducted interviews (n = 18) between December 2019 and June 2020. Each interview lasted approximately one hour. The first nine interviews were conducted in-person at private locations convenient for participants. The remaining interviews were conducted via video conference due to the COVID-19 pandemic. The interviewer obtained verbal consent to participate in the research study from participants prior to each interview. Participants were compensated $20 per interview; seven participants waived compensation. Interviews were audio recorded and transcribed verbatim. All study procedures were approved by the University of North Carolina at Chapel Hill’s Institutional Review Board. 2.3. Analysis The interviewer wrote memos after each interview, noting key topics and ideas that could be explored further. Deidentified transcripts were uploaded into Atlas.ti software. Guided by the Framework Method for qualitative health research (Gale et al., 2013), the first author developed a codebook of relevant categories in the interview data. Topical codes were deductive and based on previous research applications of the Planned Adaptation Model (Kirk et al., 2021). Additional codes were developed inductively during analysis to capture recurring themes. To assess reliability in coding, 20% of transcripts were randomly selected to be coded separately by a research assistant (R.J.). The analysts met twice to discuss discrepancies in coding and refine code definitions. The first author then updated the discussed transcripts and coded the remaining transcripts. Next, she used code reports and matrices to organize the data and identify themes across the interviews (Raskind et al., 2019). She then developed a model to visually represent key findings. CAB members provided feedback on the model and results. The first author also presented study findings in a GHDC meeting and invited input on whether the visual depiction of ACCURE accurately represented GHDC’s understanding of the intervention. 3. Results The sample included 18 participants representing a range of roles in ACCURE and community, academic, and medical affiliations (see Table 3). The sample reflected the racial diversity of the ACCURE study team (participants identified as Black, White, and Asian) and included men and women. The results from the in-depth interviews are presented in two sections: (1) the motivation for ACCURE, and (2) an organizing model of ACCURE linking each intervention component to an underlying mechanism of change. 3.1. Motivation for ACCURE Participants described two primary motivations for ACCURE: gaps in data reporting that obscured racial disparities in treatment outcomes, and commitment to a system-change intervention approach. These themes are discussed below with supporting quotations from interview participants. 3.1.1. Gaps in data reporting that obscured disparities A key motivation among academic research partners was to improve data reporting and transparency in order to illuminate racial disparities, raise awareness among care system employees about treatment disparities in their organizations, and use the data to improve care quality. Participants referred to CCARES, prior research that pointed to gaps in how treatment data was reported in the Cone Health cancer registry database (Yonas et al., 2013). Patient race and treatment completion data were not consistently reported, and the information translated into the cancer registry often lagged months behind actual treatment time. ACCURE researchers viewed this time lag as a barrier to quality care because providers were unaware of which patients were falling behind on treatment until it was too late to reengage them in care. One participant explained, “We … tried to look at the cancer registry to see what were the differences between Black and White women who have breast cancer, and there was so much missing data … that was a shocking kind of wake-up call to action among [GHDC] members … if we are going to eliminate a racial disparity, you have to first document that … The medical system partner had that information in medical charts, but it wasn’t in the cancer registry.” (Project manager) This spurred the researchers to analyze cancer registry and medical records to document disparities in treatment outcomes among Cone Health patients. The process of analyzing data disaggregated by patient race illuminated racial disparities in treatment completion and links to higher mortality rates among Black patients. Another participant described cancer center providers’ reaction to the documented disparities: “… so we used their cancer registry data and we analyzed it for them by race for the previous years, so that they can see that there were African American women dying from it. The doctors just … They didn’t know. They didn’t know who was not completing the care. They didn’t know who had died from it. They just don’t follow up and so they were very astounded by the data.” (Principal Investigator) Through conversations with providers and leaders at the cancer center, the researchers learned that data were not often used for quality improvement efforts, and when they were, the usual approach was to examine outcomes across all patients in certain illness categories. One participant stated: “… the providers and administrators are not always hearing about all these kinds of discrepancies in care between different racial groups because they didn’t have the data to really see it in real time, or at all, if no one was really looking at it. Because oftentimes people are looking at things as the whole group … and not looking between groups and seeing what maybe some of the differences are. So it’s kind of bringing attention to that in a way and having the data … to show that there was a true difference going on.” (Postdoctoral fellow) This participant explained that due to gaps in data reporting, especially with regard to patient racial identity groups, there was a lack of awareness among cancer center staff about disparities in treatment outcomes. Racial disparities were present in the data, but because the data were not stratified and examined by patient race, the disparities were obscured. These findings helped motivate Cone Health to partner with GHDC in developing a grant proposal to design an intervention aimed at addressing documented disparities. 3.1.2. Commitment to a system-change intervention approach Participants also emphasized GHDC’s commitment to a system-change approach to enhance racial equity in health care organizations. In explaining this motivation, one participant referred to the qualitative interviews conducted with patients in CCARES, stating, “Well, the motivation was to actually try and intervene in the system. Recognizing, from people’s lived experience and from our data from CCARES, how do we approach a system and influence people’s outcomes within the system?” (Survey data coordinator). This participant highlighted the link between patient experiences documented in prior research (Yonas et al., 2013) and GHDC’s commitment to addressing disparities in care outcomes by implementing changes at the organizational level. Another participant discussed the antiracism training that informs GHDC’s work as an inspiration for the system-change approach. She said: “… part of the [ACCURE] interventions were inspired by … the racial equity training, which suggested ways to undo the historical impacts of racism … focusing on changing the health care system. Not trying to change the people, but trying to change the health care system … too much blame is put on communities and the people themselves, when they’re not always the decision makers for what’s happening [to] them [and] for the pressure that they experience … and so we need to change institutions.” (Project manager) This sentiment was expressed across the interviews, with many participants stating that the motivation for the intervention was to implement strategies focused on changing the health care system, as opposed to changing individuals. Another participant emphasized the importance of addressing racism from a systems perspective, stating: “When you see disparities, how do you correct those? How do you correct it from a system versus for somebody’s implicit bias or how they view a person of color? How do you change a system where, regardless of your implicit bias or your ideology or your beliefs, it won’t affect the amount of treatment? Because … everybody should get the same. How do you make the system do that versus the individual do that?” (GHDC Executive Board member) The questions raised by this participant illuminate the research team’s commitment to implementing changes that were integrated into the operations of the care system. GHDC members viewed system-change as the foundation of their approach. Intervening to address biases and beliefs held by individual providers was a complementary strategy, but, in this participant’s view, would not be sufficient to enact lasting change if implemented alone. 3.2. An organizing model of ACCURE: Transparency, accountability, and the Social Ecological Model Participants’ theory of change for how ACCURE worked to eliminate the racial disparity in treatment completion was grounded in the antiracism principles of transparency and accountability, which had informed the design of the original ACCURE intervention. While the present study was not designed specifically to examine these principles, the influence they had on ACCURE was evident in the data. Throughout the interviews, participants emphasized transparency and accountability as mechanisms that were essential to intervention success. These principles were operationalized in each intervention component, described below. Data from the interviews also indicated that participants understood ACCURE as a multilevel intervention: “a layered approach” (Survey data coordinator). We applied the Social Ecological Model (Fig. 1) to our analysis to examine the multilevel nature of participants’ descriptions of the principles of transparency and accountability. Three levels of the SEM were represented in ACCURE: the community, organizational, and interpersonal levels. Participants discussed transparency and accountability as operating across these three levels, linking the principles to themes of community involvement, organizational change, and interpersonal support for patients across the dataset. To illustrate this, we developed a model that overlays transparency and accountability with the three levels of the SEM that were key to ACCURE (Fig. 2). We then mapped the intervention components onto the levels. The components listed on the left side of the model worked to enhance transparency, while the components listed on the right worked to enhance accountability. The sections below describe how ACCURE worked at each level, using supporting data from the interviews to underscore the rationale for each component in the context of transparency and accountability. 3.3. Community-level components Community-based partnership. At the community level, the foundational component of ACCURE was a strong community-based partnership with academic, medical, and community partners (GHDC), which directed the project and maintained accountability to pre-established collective values. Participants emphasized that accountability to a community-led organization was critical to the process of designing an equity-focused intervention. One participant stated, “Well, the community accountability part … that’s a critical part … health care systems think they can just recreate the actual interventions that we did with ACCURE, and they can, but what makes it stronger is having a community-accountable partner to ask questions along the way. And the … partner needs to be a group of people who in some way represent and identify with … the racial disparity that’s going on, whether they are part of the same racial identity group or they’re part of the disease condition … or they know close loved ones to them.” (Project manager) Emphasizing the importance of community involvement from the outset and throughout the process, another participant said, “… having a community-based organization that’s grounded in the equity work. That was a key component. That had an active role in the development of that from beginning … having some community body … that’s engaged in constant conversation about what’s going on. So, you’re reporting to them and then they’re given the space and time to give input.” (Postdoctoral fellow) This quotation highlights the ongoing nature of GHDC’s involvement with ACCURE, and the importance of providing multiple opportunities for community-based research partners to provide feedback on the intervention development and implementation process. Antiracism training. Participants described antiracism training as foundational to the intervention. The training allowed community, academic, and medical research partners to establish a common understanding of the rationale for the intervention’s system-based approach. One participant stated, “I would say that having racial equity or antiracism training for essential staff is a baseline. There needs to be a common understanding of systemic inequity. And even in a highly educated people, there’s usually a major gap in their understanding of systemic inequity. And so, I think that has to be foundational.” (Survey data coordinator) This participant described the training as an important requirement for research partners involved with the intervention, regardless of their educational background. A physician involved with the ACCURE study described the transformational effect that antiracism training had on his understanding of racial disparities: “… as much as I was aware of racial disparities in health care and the outcomes, I wasn’t really aware of the institutional underpinnings of why we ended up with the system we currently have. It was helpful … I found myself at that time when I became the lung cancer champion not necessarily really accepting that there was institutional racism and it wasn’t until the mandated two-day training session that it really hit home with me that, okay, now I really understand.” (Physician) The participant recognized that the antiracism training prompted a shift in his understanding of institutional racism, which in turn strengthened his understanding of the rationale for ACCURE’s system-change approach. This was especially important, because as a physician champion for the study, the participant was tasked with galvanizing support for the intervention among cancer center staff. A cancer center administrator described the impact the historical aspect of the antiracism training on how she viewed her work: “I think knowing the history allows you to see injustice, and then to begin to say with transparency, this was unjust. How do we go back now? You can’t rewrite history and all the patients that have come through, but how do we change it so that history doesn’t repeat itself? … what’s been most helpful is to take these antiracism principles of the transparency of the data and the accountability to change.” (Administrative leadership) This quotation highlights the participant’s thought process that led them to connect the lessons from the antiracism training to the principles of transparency and accountability that were operationalized in ACCURE. Ongoing communication. GHDC met monthly over the course of the ACCURE study. This ongoing communication allowed research partners to continually discuss study design and implementation issues, express concerns, and weigh in on decisions. GHDC members led all stages of the research process, from writing the grant application, to selecting measures and developing the script for patient telephone surveys, to adapting the protocol during the study to address implementation barriers. One participant explained, “… we all worked through the plan because we were so invested in writing the application, so we knew what the plan was. And so, if there was deviation from it, we knew we would come to [GHDC] and say, ‘Okay, do you agree with this deviation? Do you agree with this budget cut?’ Those kind of things.” (Principal Investigator) At times, discussions at GHDC meetings led to interpersonal conflict among members with diverse perspectives on the best way to proceed. Participants who were longtime GHDC members viewed these conflicts, and the ability to work through them together without silencing or minimizing anyone’s perspective, as a core strength of the group. One participant said, “… it’s the consistency of the group to stay together, the consistency of the group to have transparent and real conversations … From those conversations you can … work out some stuff before you even go to Cone [Health Cancer Center]. When you go to Cone, you know what your focus is and what you’re trying to do.” (GHDC Executive Board member) This quotation lays out the participant’s perspective on how having difficult conversations internally at GHDC meetings allowed the group to strengthen their pitch to medical partners to get them to engage with the research project. 3.4. Organizational-level components The community-level research partnership was the foundation that allowed the ACCURE research partners to develop an organizational change intervention to address racial disparities in cancer treatment. The key intervention components at the organizational level have been described in previous publications (Cykert et al., 2020; Eng et al., 2017). The data from the present study reinforced the centrality of these components (the data system, training mechanism, and advocacy roles) and pointed to an additional factor at the organizational level: leadership that is open to change and committed to authentic partnership with community members. Here, we use evidence from the present study to underline the rationale for each component. Organizational leadership. In describing the relationship-building process among community, academic, and medical research partners that led up to implementation of ACCURE, participants spoke about the process of creating buy-in among leaders in the care system over several years prior to developing the ACCURE intervention. A community-based participant described some of the key questions that GHDC members had about Cone Health’s role in the project and the organization’s readiness to change: “Will Cone change? Will they want to change? How much of a change would this affect them, based on their processes, based on their finances, or just the fact that our outside community-based or ‘relationship organization’ is going to actually say that you need to change something? How will they respond?” (GHDC Executive Board member) This participant emphasized the need for openness among the care system’s leadership and an interest in facilitating an equity-focused intervention. Participants also discussed the CBPR approach of the intervention as important to the relationship-building process with Cone Health. Community research partners were actively involved with discussions with cancer center leaders, underscoring the importance of ongoing communication with and accountability to community members affected by care system disparities. Data system. Many participants described the crux of the intervention as the data system that was connected to the electronic health records. The system collected data in real-time and alerted medical staff to deviances from standard care. One participant described the data system, which was called the Real-Time Registry, as “the backbone of ACCURE.” He went on to say, “We always knew where the patient was in their care process, and if they didn’t go through the steps that were programmed in the Registry ... a warning would come up” (Principal Investigator). Another participant described the purpose of the Real-Time Registry as making sure patients did not fall through the cracks by tracking their appointment data and flagging missed milestones in their cancer treatment. The Real-Time Registry enhanced transparency so employees in the care system could see when a patient was not receiving their recommended treatment in a timely manner. A participant said, “The Real-Time Registry provided that real-time transparency, and was also the tool of accountability because a warning came up, and somebody had to deal with it” (Principal Investigator). By linking the warning flags produced by the Real-Time Registry to a designated role in the care system (in this case, the ACCURE navigators, described below) the data system leveraged transparency to prompt accountability in the care system to reach out to patients and address barriers to quality care. Training mechanism. Another intervention component that enhanced transparency at the organizational level was training sessions for cancer center staff, known as Health Equity Education Training (HEET) sessions (Black et al., 2019). The content of the sessions was informed by focus groups with Black and White cancer survivors, conducted in the intervention development stage of ACCURE, to better understand patient experiences and to identify care system-related barriers that affected treatment engagement. Focus group participants were asked to describe interactions with the care system that were “pressure points,” or times when institutional factors created barriers to remaining engaged in treatment. The focus group findings highlighted a lack of accountability of the care system to provide patients, especially those who identified as Black, with sufficient support to navigate their cancer treatment (Black et al., 2021; Eng et al., 2017). These findings, along with site-specific data on care quality metrics, disaggregated by patient race, were presented to cancer center staff during the HEET sessions. Participants said that the training sessions played an important role in raising awareness among cancer center staff about the systemic nature of health disparities and enhancing transparency regarding care inequities at the clinic level. One participant said, “… the goal was to increase awareness of the staff regarding disparities from a systemic lens to help them understand the disparities. When we talk about disparities, it’s not about individual acts of meanness, it’s about systemic barriers. And then to engage them in thinking about those systemic barriers so that not only are we trying to address the people enrolled in the study, but also helping the staff to evolve their lens about what racial equity means and how it shows up. So that hopefully future problem solving can be done from a more systemic perspective.” (Survey data coordinator) The HEET sessions increased transparency in the institution-level barriers that contribute to disparities. Another participant described the effect that increased transparency around institutional racism had on providers: “I think that in some ways the staff were awakened and had a better understanding of how patient outcomes may be driven by their skin color or by their race rather than all the treatments we’re offering. I think there was some eye-opening moments for clinicians in the program” (Physician). This quotation demonstrates the roll the HEET sessions played in raising awareness among cancer center staff about how racism can affect patient care. Advocacy roles. Accountability at the organizational level included advocacy roles (i.e., physician champions and ACCURE navigators) adapted from existing provider roles. These roles involved additional training and specialized protocols created specifically for the intervention (Black et al., 2019). Physician champions were intended to advocate for study goals from within the cancer center, support providers in adapting to and engaging with structural changes such as the Real-Time Registry and HEET sessions, and help disseminate site-specific data on treatment disparities. One participant described the role physician champions played in bringing other physicians on board with ACCURE recruitment goals: “The physician champion was the local study cheerleader. We really need to enroll these patients … one of their main jobs was to motivate others to be interested in bringing patients into the study” (Principal Investigator). Another participant emphasized the importance of the physician champion as a link from community partners to cancer center leadership: “There has to be a physician champion. There has to be a clear pathway to the top of the breast cancer center or the head of Moses Cone [Cancer Center]. There has to be this open dialogue … with those in academia, Moses Cone and whoever the community partners are.” (GHDC Executive Board member) The ACCURE navigators were also a key accountability component at the organizational level. They were integrated into the care team, meaning they attended weekly meetings among surgeons and oncologists to discuss patients’ treatment recommendations, and communicated patients’ needs and concerns with providers. Participants described the rationale for the ACCURE navigator role as the link between data transparency enabled by the Real-Time Registry and the delivery of quality care to patients. One participant explained: “Well, certainly the Real-Time Registry is important, but … without a follow up plan I don’t think [it] would be helpful … The system takes responsibility for following up with the missed milestones. The nurse navigator model worked really well for us … there has to be a mechanism of accountability to the missed milestones.” (Survey data coordinator) This participant suggested that other roles in the care system could be adapted to fulfill this aspect of the intervention. Another participant emphasized the need for accountability within the care system to follow up with patients who may need tailored support to remain engaged in treatment: “Whenever a patient goes through the system, especially when there’s a sequence of treatments, who is responsible for ensuring that both sides of the sequence happen, that the planners of the sequence do their thing, and the patient shows up and does their thing? People can look at each other, and have a bystander. ‘Oh, I thought it was you’ … no one has direct responsibility in a patient-centered fashion, then things are much more likely to fall through the cracks, especially when the person coming through the system is disadvantaged somehow.” (Principal Investigator) This participant articulated a gap in accountability in the care system that may contribute to treatment disparities. The ACCURE intervention’s solution for this gap was the interpersonal-level support provided to patients through the ACCURE navigators. 3.5. Interpersonal-level components Patient advocates. The ACCURE nurse navigators connected the organizational level, system-based aspects of the intervention to patients with cancer through individualized support. They were the patient-facing aspect of the intervention who provided care system accountability to patient needs. Prior to the intervention, the ACCURE navigators at both study sites participated in the antiracism training which reframed the responsibility for identifying and managing treatment obstacles from falling primarily on the person with cancer. Instead, ACCURE navigators shared this responsibility and were trained to proactively reach out to patients who were falling behind on treatment milestones such as scheduling surgery or attending chemotherapy appointments. Participants described the ACCURE navigators as a critical component of the intervention. One participant spoke about specific personality traits that were important for carrying out the navigator role: “The nurse navigator, she was essential … I think a lot of it also had to do with the personality and the person, and the empathy and the compassion she had. The fact that she wanted everyone to have equal footing in reaching either long term survival, cure, or whatever it may be, for their cancer. She genuinely cared.” (Administrative leadership) In an interview with the ACCURE navigator at one of the study sites, she described key qualifications for the role: “Good listening skills, good communication skills, organization skills, empathy, compassion.” Describing the impact the ACCURE navigator had on patient care at the cancer center, one physician said: “… it was a continuity of care throughout their journey of treatment … that human connection with having that patient navigator who is aware of the conversation and the treatment plan and the next steps … that agent is vitally important.” This quotation demonstrates how the ACCURE navigators enhanced two-way communication between patients and the care system, which allowed for greater transparency. For patients, there was more transparency in their treatment plan and what they could expect in terms of their treatment schedule and side effects. For providers, there was also increased transparency around obstacles patients were experiencing that interfered with their ability to remain engaged in treatment. This two-way communication allowed navigators to bridge communication gaps between patients and oncologists, and to support patients by offering resources such as transportation vouchers or referrals to specialists. The direct link between the ACCURE navigators and the Real-Time Registry allowed the navigators to see when a patient was falling behind on their treatment and use that information to reengage patients in care. 4. Discussion This study documents ACCURE’s motivation, mechanisms of change, and key components from the perspective of community, academic, and medical research partners who designed and implemented the intervention. GHDC’s emphasis on changing systems, and the desire to move beyond documenting disparities to intervention, were driving forces in ACCURE. Findings from the interviews indicate that the antiracism principles transparency and accountability were effective change mechanisms in an equity-focused health services intervention in cancer care. ACCURE embedded these principles into the routine practice of care delivery across community, organizational, and interpersonal levels to address structural factors in the care system that contribute to treatment disparities and, in turn, improve care quality for all patients (Cykert et al., 2020). The key components identified in this study (i.e., the bullet points in Fig. 2) were: a community-based partnership with antiracism training and ongoing communication among research partners, a Real-Time Registry data system, a provider training mechanism, buy-in from organizational leaders, advocates for the intervention itself (physician champions), and patient advocates (navigators). A recent systematic review synthesized commonalities among antiracism interventions in health care settings (Hassen et al., 2021). While the article search for the review was conducted in 2018, prior to publication of the study documenting ACCURE’s success in eliminating a Black-White disparity in treatment completion, the review’s findings mirror many of the strategies employed by ACCURE. Hassen and colleagues lay out a conceptual model that names six foundational elements of a health care-based antiracism intervention, all of which were applied in ACCURE: defining the problem, using shared antiracism language, establishing leadership buy-in, investing resources, partnering with experts, and establishing community partnership. The review also highlights “transparent accountability mechanisms” as a key strategy for antiracism interventions (Hassen et al., 2021, p. 12). The ACCURE intervention predates this review and yet the findings from the present study are congruent with the conceptual model presented by Hassen and colleagues. Our study adds to this body of work by offering a model of an evidence-based intervention that depicts specific strategies for operationalizing transparency and accountability at multiple levels. A methodological contribution of this study is the visual display of interview findings, an underused approach in qualitative inquiry (Kegler et al., 2019). The model presented in this manuscript (Fig. 2) provides health services researchers and practitioners with a practical resource to understand the components in the ACCURE intervention. The original study evaluated the overall impact of the intervention, and thus was not designed to tease out differential effects of the intervention components. Future research should use this new model to facilitate more complex study designs in adapted versions of ACCURE, so researchers can compare the impact of various components (e.g., cluster-randomized trials). Additionally, the original study did not include organization-level measures. Future research should consider measures at the organizational level to evaluate the degree to which various components contributed to organizational change. This could include the development of new measures to evaluate change in organizational transparency around care metrics and accountability to the population served. As the purpose of this study was to document ACCURE’s key components, an examination of the original intervention’s implementation barriers and facilitators is beyond the scope of this manuscript. Here, we will briefly highlight one challenge that came to light in the interviews regarding the implementation of the HEET sessions. At Cone Health, a cancer center in a community hospital, the training sessions were held in the evening and attendance was low. This contrasted with Hillman Cancer Center, which is an academic hospital that holds regular Grand Rounds educational sessions for cancer center staff. At Hillman Cancer Center, the HEET sessions were held at Grand Rounds during the workday. One participant pointed to this inconsistency as evidence that the HEET sessions were less essential to the intervention’s success, stating, “The lessons are important, but they didn’t drive the improved care” (Principal Investigator). However, even at Cone Health where attendance at the sessions was low, several participants noted the impact they had on the organization even beyond ACCURE’s active intervention stage. A physician stated, “We’ve brought back some of the HEET session modules … to the cancer center and it’s always eye-opening for the staff to be introduced to a new concept like that … it can be very impactful.” Future adaptations of ACCURE should consider the implementation context, including the question of whether there is an existing infrastructure for staff training. When considering implementation strategies for the overall intervention, we recommend referring to emerging guidance on applying an antiracist lens to implementation science. A foundational piece is partnering with community stakeholders whose lived experiences represent those most impacted by health care disparities when adapting evidence-based interventions (Shelton et al., 2021). These partnerships will be important in the process of adapting ACCURE to the values and needs of a specific community. For example, we can imagine that while race-specific data tracking was a strategy favored by GHDC in designing the original intervention, other community groups may find this type of data monitoring intrusive or harmful. Community-guided adaptation will allow future versions of ACCURE to be grounded in the expertise and lived experiences of local community members. 4.1. Limitations The interviews for this study were conducted three years after the ACCURE study concluded, so the data may have been subject to recall bias due to participants not correctly remembering certain details or events related to the intervention. While we were able to recruit nearly all key study personnel to participate in interviews, two were not available, which may have biased the data toward greater representation of the experiences of Cone Health employees as compared to Hillman Cancer Center employees. Additionally, due to the COVID-19 pandemic, we transitioned data collection from in-person to video conference halfway through the study. This change did not appear to affect the quality of the data. In interviews that took place during that period, participants were more likely to bring up the pandemic’s impact on the health system. 5. Conclusion Despite challenges researchers encountered when designing and implementing ACCURE, the intervention worked to improve quality and racial equity in cancer care (Cykert et al., 2020). The antiracism principles of transparency and accountability guided every aspect of the intervention. The vibrant partnership fostered by GHDC and the commitment among members to contribute to impactful racial equity research were foundational to the study’s success. One participant summarized the character of GHDC in saying: “The basic principle of [GHDC] is an understanding of racial equity principles. And I think what’s most important … is to have an enthusiasm for racial equity work … you have to believe in the importance of it and have a willingness to stretch yourself, because I think part of what makes [GHDC] work is people being deeply enough engaged that they can tolerate the discomfort of being challenged.” (Survey data coordinator) In developing a research partnership where members are open to having their views challenged and are committed to working collectively towards a common goal, future researchers can use the organizing model of ACCURE to design equity-focused health services interventions that are guided by the principles of transparency and accountability. Evidence from the ACCURE intervention suggests that future interventions are most likely to be successful if they are developed in partnership with community-based researchers, involve systematic changes at the organizational level, and are responsive to the unique characteristics of the health system and patient population. Acknowledgements This manuscript is dedicated to Claire Morse, in loving memory and faithful commitment to her legacy as an antiracist organizer. We would also like to express gratitude to Fatima Guerrab and Thomas Clodfelter for their contributions to this research. Funding This study was funded by a National Research Service Award Pre-Doctoral Traineeship (for I. Griesemer) from the Agency for Healthcare Research and Quality, sponsored by The Cecil G. Sheps Center for Health Services Research, the University of North Carolina at Chapel Hill [grant number T32-HS000032] and a NC Translational and Clinical Studies Institute (TraCS) pilot grant, National Center for Advancing Translational Sciences (NCATS, National Institutes of Health [grant number UL1TR002489]. Writing of this manuscript was also supported by the Department of Veterans Affairs Office of Academic Affiliations Advanced Fellowship Program in Health Services Research, the Center for Healthcare Organization and Implementation Research (CHOIR), Boston, MA (for I. Griesemer). Abbreviations ACCURE Accountability for Cancer Care through Undoing Racism and Equity CBPR Community-Based Participatory Research GHDC Greensboro Health Disparities Collaborative CAB Community Advisory Board CCARES Cancer Care and Racial Equity Study SEM Social Ecological Model HEET Health Equity Education Training Fig. 1. Social Ecological Model. Fig. 2. An organizing model of ACCUREa. aIn this model, ACCURE’s intervention components are mapped onto three Social Ecological levels and the antiracism principles of transparency and accountability. Table 1 Undoing Racism® principles applied in the development of ACCURE. Principle Description (from PISABa) Application Analyzing power As a society, we often believe that individuals and/or their communities are solely responsible for their conditions. Through the analysis of institutional power, we can identify and unpack the systems external to the community that create the internal realities many people experience daily. Formative research in the intervention development stage of ACCURE examined institutional power in the cancer care system. This work led GHDC to identify transparency as a key principle to guide the design of the ACCURE intervention.b Maintaining accountability Organizing with integrity requires that we be accountable to the communities struggling with racist oppression. GHDC designed ACCURE to increase cancer center accountability to Black patients.b a The People’s Institute for Survival and Beyond. b See section 3.2 for an in-depth examination of how ACCURE applied the principles of transparency and accountability. Table 2 Planned adaptation model. Step 1 Examine intervention’s theory of change Step 2 Identify differences between old and new setting Step 3 Adapt intervention to new setting Step 4 Evaluate adapted intervention Table 3 Participant roles in ACCURE (n = 18). GHDCa member Non-GHDC member Community research partners  GHDC Executive Board member 1  Survey data coordinator 1  Budget coordinator 1  Research assistant 2 Academic research partners  Principal Investigator 2  Postdoctoral fellow 1  Project manager 1  Information technology specialist 1 Medical research partners  Administrative leadership 1 2  Physician 1 2  Nurse navigator 1  Information technology specialist 1 a Greensboro Health Disparities Collaborative. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. 1 References dated 2021 were accessed in 2019–2020 prior to formal publication as online or preprint versions with permission from the authors. References Bailey ZD , Feldman JM , & Bassett MT (2021). How structural racism works—racist policies as a root cause of US racial health inequities. New England Journal of Medicine, 384 (8 ), 768–773. 10.1056/NEJMms2025396 33326717 Bailey ZD , Krieger N , Agénor M , Graves J , Linos N , & Bassett MT (2017). Structural racism and health inequities in the USA: Evidence and interventions. The Lancet, 389 (10077 ), 1453–1463. 10.1016/S0140-6736(17)30569-X Bassett MT , & Graves JD (2018). Uprooting institutionalized racism as public health practice. American Journal of Public Health, 108 (4 ), 457–458. 10.2105/AJPH.2018.304314 29513591 Black KZ , Baker SL , Robertson LB , Lightfoot AF , Alexander-Bratcher KM , Befus D , … Eng E (2019). Health care: Antiracism organizing for culture and institutional change in cancer care. In Ford CL , Griffith DM , Bruce MA , & Gilbert KL (Eds.), Racism: Science & tools for the public health professional (pp. 283–302). American Public Health Association Press. Black KZ , Lightfoot AF , Schaal JC , Mouw MS , Yongue C , Samuel CA , Faustin YF , Ackert KL , Akins B , Baker SL , Foley K , Hilton AR , Mann-Jackson L , Robertson LB , Shin JY , Yonas MA , & Eng E (2021). It’s like you don’t have a roadmap really: Using an antiracism framework to analyze patients’ encounters in the cancer system. Ethnicity and Health, 26 (5 ), 676–696. 10.1080/13557858.2018.1557114 30543116 Came H , & Griffith D (2018). Tackling racism as a “wicked” public health problem: Enabling allies in anti-racism praxis. Social Science & Medicine, 199 , 181–188. 10.1016/j.socscimed.2017.03.028 28342562 Chin MH , Clarke AR , Nocon RS , Casey AA , Goddu AP , Keesecker NM , & Cook SC (2012). A roadmap and best practices for organizations to reduce racial and ethnic disparities in health care. Journal of General Internal Medicine, 27 (8 ), 992–1000. 10.1007/s11606-012-2082-9 22798211 Choo E (2021). Easy in, tough out: The dam of health-care racism. The Lancet, 397 (10274 ), 570. 10.1016/S0140-6736(21)00299-3 Crear-Perry J , Maybank A , Keeys M , Mitchell N , & Godbolt D (2020). Moving towards anti-racist praxis in medicine. The Lancet, 396 (10249 ), 451–453. 10.1016/S0140-6736(20)31543-9 Cykert S , Eng E , Manning MA , Robertson LB , Heron DE , Jones NS , Schaal JC , Lightfoot A , Zhou H , & Yongue C (2020). A multi-faceted intervention aimed at Black-White disparities in the treatment of early stage cancers: The ACCURE Pragmatic Quality Improvement trial. Journal of the National Medical Association, 112 (5 ), 468–477. 10.1016/j.jnma.2019.03.001 30928088 DeSantis CE , Miller KD , Goding Sauer A , Jemal A , & Siegel RL (2019). Cancer statistics for african Americans, 2019. CA: A Cancer Journal for Clinicians, 69 (3 ), 211–233. 10.3322/caac.21555 30762872 Eng E , Schaal J , Baker S , Black K , Cykert S , Jones N , Lightfoot A , Robertson L , Samuel C , & Smith B (2017). Partnership, transparency, and accountability. In Duran B , Oetzel JG , Minkler M , & Wallerstein N (Eds.), Community-based participatory research for health: Advancing social and health equity (pp. 107–122). John Wiley & Sons. Gale NK , Heath G , Cameron E , Rashid S , & Redwood S (2013). Using the framework method for the analysis of qualitative data in multi-disciplinary health research. BMC Medical Research Methodology, 13 (1 ), 117. 10.1186/1471-2288-13-117 24047204 Griesemer I , Lightfoot AF , Eng E , Bosire C , Guerrab F , Kotey A , … Robertson LB Examining ACCURE’s nurse navigation through an antiracist lens: Transparency and accountability in cancer care. Health Promotion Practice. In press. Griffith DM , Yonas M , Mason M , & Havens BE (2010). Considering organizational factors in addressing health care disparities: Two case examples. Health Promotion Practice, 11 (3 ), 367–376. 10.1177/1524839908330863 19346409 Hardeman RR (2020). Examining racism in health services research: A disciplinary self-critique. Health Services Research, 55 (Suppl 2 ), 777. 10.1111/1475-6773.13558 32976632 Hardeman RR , Medina EM , & Kozhimannil KB (2016). Structural racism and supporting Black lives - the role of health professionals. New England Journal of Medicine, 375 (22 ), 2113–2115. 10.1056/NEJMp1609535 27732126 Hassen N , Lofters A , Michael S , Mall A , Pinto AD , & Rackal J (2021). Implementing anti-racism interventions in healthcare settings: A scoping review. International Journal of Environmental Research and Public Health, 18 (6 ), 2993. 10.3390/ijerph18062993 33803942 Jones CP (2002). Confronting Institutionalized Racism. Phylon (1960-), 50 (1/2 ), 7–22. 10.2307/4149999 Kegler MC , Raskind IG , Comeau DL , Griffith DM , Cooper HLF , & Shelton RC (2019). Study design and use of inquiry frameworks in qualitative research published in health education & behavior. Health Education & Behavior, 46 (1 ), 24–31. 10.1177/1090198118795018 30227081 Kellou N , Sandalinas F , Copin N , & Simon C (2014). Prevention of unhealthy weight in children by promoting physical activity using a socio-ecological approach: What can we learn from intervention studies? Diabetes and Metabolism, 40 (4 ), 258–271. 10.1016/j.diabet.2014.01.002 24698814 Kirk MA , Haines ER , Rokoske FS , Powell BJ , Weinberger M , Hanson LC , & Birken SA (2021). A case study of a theory-based method for identifying and reporting core functions and forms of evidence-based interventions. Translational Behavioral Medicine, 11 (1 ), 21–33. 10.1093/tbm/ibz178 31793635 Largent EA (2018). Public health, racism, and the lasting impact of hospital segregation. Public Health Reports, 133 (6 ), 715–720. 10.1177/0033354918795891 30223719 Lee SJ , Altschul I , & Mowbray CT (2008). Using planned adaptation to implement evidence-based programs with new populations. American Journal of Community Psychology, 41 (3–4 ), 290–303. 10.1007/s10464-008-9160-5 18307029 Ma PH , Chan ZC , & Loke AY (2017). The socio-ecological model approach to understanding barriers and facilitators to the accessing of health services by sex workers: A systematic review. AIDS and Behavior, 21 (8 ), 2412–2438. 10.1007/s10461-017-1818-2 28631228 McCormack L , Thomas V , Lewis MA , & Rudd R (2017). Improving low health literacy and patient engagement: A social ecological approach. Patient Education and Counseling, 100 (1 ), 8–13. 10.1016/j.pec.2016.07.007 27475265 McLeroy KR , Bibeau D , Steckler A , & Glanz K (1988). An ecological perspective on health promotion programs. Health Education Quarterly, 15 (4 ), 351–377. http://www.ncbi.nlm.nih.gov/pubmed/3068205.3068205 Nelson AR , Stith AY , & Smedley BD (2002). Unequal treatment: Confronting racial and ethnic disparities in health care (full printed version). National Academies Press. Patel MI , Lopez AM , Blackstock W , Reeder-Hayes K , Moushey EA , Phillips J , & Tap W (2020). Cancer disparities and health equity: A policy statement from the American society of clinical Oncology. Journal of Clinical Oncology, 38 (29 ), 3439–3448. 10.1200/JCO.20.00642 32783672 The People’s Institute for survival and beyond.(2018). PISAB. Retrieved September 26, 2019 from https://www.pisab.org/about-us/. Raskind IG , Shelton RC , Comeau DL , Cooper HLF , Griffith DM , & Kegler MC (2019). A review of qualitative data analysis practices in health education and health behavior research. Health Education & Behavior, 46 (1 ), 32–39. 10.1177/1090198118795019 30227078 Schaal JC , Lightfoot AF , Black KZ , Stein K , White SB , Cothern C , Gilbert K , Hardy CY , Jeon JY , & Mann L (2016). Community-guided focus group analysis to examine cancer disparities. Progress in community health partnerships: Research, Education, and Action, 10 (1 ), 159. 10.1353/cpr.2016.0013 27018365 Shelton RC , Adsul P , Oh A , Moise N , & Griffith DM (2021). Application of an antiracism lens in the field of implementation science (IS): Recommendations for reframing implementation research with a focus on justice and racial equity. Implementation Research and Practice, 2 . 10.1177/26334895211049482 Wallerstein NB , & Duran B (2006). Using community-based participatory research to address health disparities. Health Promotion Practice, 7 (3 ), 312–323. 10.1177/1524839906289376 16760238 Williams DR , Lawrence JA , & Davis BA (2019). Racism and health: Evidence and needed research. Annual Review of Public Health, 40 (1 ), 105–125. 10.1146/annurev-publhealth-040218-043750 Yonas MA , Aronson R , Schaal J , Eng E , Hardy C , & Jones N (2013). Critical incident technique: An innovative participatory approach to examine and document racial disparities in breast cancer healthcare services. Health Education Research, 28 (5 ), 748–759. 10.1093/her/cyt082 24000307 Yonas MA , Jones N , Eng E , Vines AI , Aronson R , Griffith DM , White B , & DuBose M (2006). The art and science of integrating undoing racism with CBPR: Challenges of pursuing NIH funding to investigate cancer care and racial equity. Journal of Urban Health, 83 (6 ), 1004–1012. 10.1007/s11524-006-9114-x 17072760

PMC010xxxxxx/PMC10361708.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9918471185006676 52004 Life Metab Life Metab Life metabolism 2755-0230 37485302 10361708 10.1093/lifemeta/load013 NIHMS1901420 Article Chowing down: diet considerations in rodent models of metabolic disease Klatt Kevin C. 1 Bass Kevin 2 Speakman John R. 3 Hall Kevin D. 4* 1 Department of Nutritional Sciences and Toxicology, University of California Berkeley, Berkeley, CA 94720, USA 2 Garrison Institute of Aging, Texas Tech University Health Science Center, Lubbock, TX 79430, USA 3 Center for Energy Metabolism and Reproduction, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, Guangdong 518055, China 4 National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892, USA Author contributions All authors collectively drafted and revised the manuscript. * Corresponding author. National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892, USA. kevinh@niddk.nih.gov 11 6 2023 6 2023 26 4 2023 21 7 2023 2 3 load013This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Diet plays a substantial role in the etiology, progression, and treatment of chronic disease and is best considered as a multifaceted set of modifiable input variables with pleiotropic effects on a variety of biological pathways spanning multiple organ systems. This brief review discusses key issues related to the design and conduct of diet interventions in rodent models of metabolic disease and their implications for interpreting experiments. We also make specific recommendations to improve rodent diet studies to help better understand the role of diet on metabolic physiology and thereby improve our understanding of metabolic disease. diet composition nutrition chow study design recommendations pmcIntroduction The 1926 Nobel Prize in Physiology and Medicine was awarded to Johannes Fibiger for his discovery that stomach cancer could be caused by a parasitic roundworm infection in rats and mice. Unfortunately, Fibiger’s rodent diets had insufficient vitamin A, and it was later found that roundworm infection did not produce stomach cancer when this dietary deficiency was removed [1]. Fibiger’s Nobel Prize was described as “one of the biggest blunders made by the Karolinska Institute” [2] and served as a warning that nutrition can seriously confound the interpretation of mechanistic studies in experimental rodents. To address concerns about potential confounding and reproducibility that may result from deficient or highly variable diet compositions across laboratories, national efforts were launched to standardize estimated nutrient requirements of laboratory rodents [3]. Nutrient requirement estimates for the mouse, last published by the National Research Council in 1995, have been established for total energy and its contributing macronutrients, protein and amino acids, minerals and vitamins. The indicator of requirement is commonly set as the dietary concentration necessary to facilitate growth, reproduction, lactation, and/or the maintenance of adult health. Prior to the 1970s, the nutrient requirements of the laboratory rodents were met using closed diet formulations (i.e. proprietary ingredient composition) that relied primarily on natural ingredients (e.g., minimally refined whole grain and fish meals). In 1974, Knapka and colleagues developed the first open-formula diet (i.e. known ingredient composition) termed the NIH-07 formulation, to serve as a cereal grain-based, non-purified standard reference diet that met the 1962 National Research Council (NRC) recommended nutrient concentrations for mice [4]. This formulation provided the foundation for the various, yet increasingly dated, rodent diet formulations employed today, the most common of which being the AIN93 series (Supplementary Table S1) [5–7]. Efforts to standardize diet formulations were ultimately derived from efforts to provide the essential nutrients, while minimizing the variability diet might introduce. Such an approach engenders a view of laboratory animal nutrition as a factor to be fixed, rather than a key aspect of experimental design to be considered in each investigation. However, diet is now increasingly recognized as much more than a nuisance variable. Rather, diet is best conceptualized as a multifaceted set of modifiable input variables with pleiotropic effects on a variety of biological pathways spanning multiple organ systems. Diet is a key modulator of metabolic physiology and plays a substantial role in influencing what we might identify as ‘normal’ physiology, as well as the etiology, progression, and treatment of chronic disease [8, 9]. The broad, complex physiological effects of diet interventions present both unique opportunities and challenges for the design of rigorous studies and their interpretation. Herein, we discuss key issues related to diet interventions in rodent models of metabolic disease and their implications for interpreting experiments. We also make specific recommendations to improve rodent diet studies to more fully leverage the powerful effects of diet on metabolic physiology to improve our understanding of metabolic disease. Compared to what? Diet design hierarchy Unlike other experimental variables, diet interventions have no placebo or other obvious control. Thus, causal inferences must be drawn from comparison diets that can differ in at least one, and often many variables (i.e., diet ingredients). Diet formulations are based on refined versus “natural” (i.e., unrefined) ingredients, which determines the capacity of investigators to manipulate diet composition in a controlled manner. Natural diet ingredients contain multiple nutrients and bioactive compounds in variable quantities thereby limiting the ability of investigators to modify components individually. Within vivariums, animals are commonly fed ‘chows’, an unstandardized term that most often refers to cereal-grain- and legume-based formulations supplemented with refined ingredients that facilitate adequate nourishment of rodent colonies in a readily affordable manner. Unlike chows, purified diets are formulated using refined ingredients (e.g., sucrose, corn starch, casein, refined oils, cellulose fiber, and micronutrient mixes), enabling precise manipulation of individual dietary components. However, this increased experimental control comes at significant financial cost and the exclusion of many dietary bioactive compounds that are otherwise present in natural diets. Ultimately, the appropriate diet formulation depends on the level of specificity desired when causally attributing observed effects to specific diet differences and identifying the underlying biological mechanisms. Below, we highlight these diet design challenges as they apply to both refined and natural diet formulations. Dietary pattern designs Often, experiments compare the effects of two diets that differ in numerous components (Fig. 1a; Supplementary Fig. S1). In such “dietary pattern designs”, causality with respect to measured phenotypes can only be attributed to some undetermined combination of diet differences. Common dietary pattern designs include the comparison of low-fat control diets to diet-induced disease models (e.g., Western or Atherogenic diets), wherein numerous ingredients and nutrients differ between formulations. Rigorous dietary pattern studies utilize purified ingredients, limiting the number of ingredient differences between the intervention and control diets and ensuring that such differences are quantifiable/known. Unfortunately, it is common to find dietary pattern studies using “low fat” control groups consuming natural foodstuffs (i.e. “chows”) that are inappropriately compared with experimental groups consuming purified diets (Fig. 2). Such experimental designs (corresponding to a “dietary pattern design”; Fig. 1A) [10, 11] maximize formulation differences and increase the potential for confounding. This common approach is problematic for causal inference, as chows lack standardization across all nutritionally relevant parameters [12]. Furthermore, rodent chow varies from lot to lot within the same product line with respect to major bioactive components, including but not limited to native factors such as fibers and plant secondary metabolites (e.g., polyphenols, carotenoids, phytate), as well as contaminants and processing related bioactives such as heavy metals, pesticides, mycotoxins, advanced glycation end products, and lipopolysaccharide. Differences are further exacerbated by the need to sterilize chows (e.g., autoclave, irradiate, etc.), introducing further compositional variability of bioactive food components [13]. Importantly, much of the variability within chows goes routinely unmeasured in standard compositional analyses, and many such factors can be profound determinants of physiology and metabolism. For example, recent investigations have detailed large variability across common vivarium chows in their Fermentable Oligo-, Di-, Mono-saccharides And Polyols (FODMAPs) contents, a class of microbiota-accessible carbohydrates that modify microbiome composition and cecal metabolite concentrations [12]. Collectively, batch-to-batch variability potentially undermines reproducibility even within identical commercial diet lines [14–18]. As an example of how using chow controls can lead to misleading results, a series of high-profile studies on so-called obesogenic microbiota attributed increased obesity in groups of mice fed purified high-fat diets to differences in microbiota composition compared to mice fed unrefined chow diets [19–21]). Unfortunately, it was later demonstrated that mice fed purified low-fat diets, like their purified high-fat diet counterparts, also developed “obesogenic microbiota” while nonetheless not gaining excess body fat, suggesting that the observed microbiota differences may have been due to the refined nature of the high-fat diet rather than causing or being caused by obesity per se [22]. Given the widespread inappropriate use of chow diets as controls for purified diets, this cautionary tale raises disconcerting questions about the number of potentially misinterpreted and misleading studies. While researchers commonly employ ‘dietary pattern designs’ to induce a phenotype (e.g., obesity and insulin resistance), attribution of specific dietary factors causing the phenotype is complicated by the numerous differences between experimental and control diets. The feeding of controls with as few composition differences as possible, can improve the likelihood of identifying specific diet factors causing the phenotype. Alternatively, using a variety of different diets to induce similar phenotypes can limit the potential that observed physiological relationships are not merely secondary to a particular choice of diet composition. For example, researchers intending to study the impact of obesity and insulin resistance on various organ systems can do so through the use of separate high fat diets rich in either monounsaturated or polyunsaturated fatty acids (compared to refined low fat controls), reducing the likelihood that observed phenotypes are secondary to alterations in specific dietary fatty acids and their impact on tissue lipid composition and related signaling (e.g., eicosanoids) as opposed to the effects of obesity and insulin resistance per se. Diet substitution designs To narrow the scope of diet differences that may be responsible for any observed effects, “diet substitution designs” specify that any diet component with substantial mass or energy content that is modified in one arm of the study must also have a corresponding component of the same mass or energy modified in the comparison diet (Fig. 1b). This avoids differential concentration or dilution of dietary components when a single component is added or subtracted from the formulation and avoids mismatching all other dietary components per unit mass or energy. Researchers often employ dietary pattern or diet substitution designs to target a specific metabolic mechanism, but two-armed designs are inherently confounded when it comes to inferring the causal contributions to specific dietary factors. Even in the simplest diet substitution design, pairwise comparisons of diets always differ in at least two variables, rendering two-arm diet investigations insufficient for causal inference about a particular diet component to a study outcome. Outcome differences may be due to the addition of an individual component, the removal of the component that was replaced, or some combination. In other words, diet substitution effects require that investigators consider potential effect modification due to the substitution component when making causal attributions. For example, studies that investigate effects of replacing, gram-for-gram, individual essential amino acids, such as methionine or leucine, with individual non-essential amino acids [23–26]. Such nutrient substitution otherwise maintains the same protein, energy, and nutrient composition of each diet and isolates the change in diet to a single amino acid, isolating the specificity of causal inference to that single substitution. This example highlights an ideal case: a precise, purified amino acid substitution. However, in less ideal circumstances, diet substitution designs can still introduce substantial uncontrolled confounding. This can occur when substituting ingredients with multiple components, for example, substituting oils containing multiple different fatty acids as well as phytosterol and vitamin E contents, in variable quantities in each oil. Such a substitution confounds causal inferences with respect to the target fatty acid substitution. Substitutions may also introduce substantial confounding when they result in divergent total food intake in terms of energy or mass, commonly observed when formulation changes modify palatability or energy density. For example, isocaloric exchange between dietary fat (~9 kcal/g) and carbohydrate (4 kcal/g) results in matching all other diet components on a per energy basis but not per unit mass. As many other elements of a diet are based on the formulation weight (e.g., energy density of the diet; vitamins, minerals, and fiber commonly added at a % w/w), such substitutions, as commonly employed in ‘High Fat Diets’, can induce differences in intake beyond the isocaloric exchange that may have independent effects on outcomes of interest. In instances where caloric intake is expected to be the same across groups despite changes in energy density, ingredients in the formula can be added on a weight per total kcal of the formulation basis, as opposed to % w/w, to avoid mismatched intakes. Food intakes often diverge when energy densities have been altered, and thus, some have proposed the addition of non-caloric ingredients to the formulation to normalize energy densities (e.g., the addition of cellulose); however, this introduces another mismatch by diluting or concentrating other diet components per unit mass and assumes a completely inert effect of the added ingredient. Attempts to match food intake between groups can induce other behavior differences that also need to be considered when interpreting outcomes. Finally, despite this tremendous advantage over the dietary pattern design, even a perfectly executed two-armed diet substitution study without the above problems cannot disentangle the effects of the component that was restricted from the effects of the component that was added to replace it; this limitation is inherent to the substitution design and cannot be overcome so long as the design is used. To overcome the limits to causal inference inherent to 2-arm designs, a “multiple diet substitution design” is needed by adding additional comparator groups, each making a different substitution, thereby producing a ranking of the relative impact of each diet component on the outcome of interest. In other words, the physiological effects of a single diet component may be isolated by making multiple substitutions and examining whether the outcomes remain robust and examine effect modifications due to the substituted components (Fig. 1c). For example, a methionine restriction study could compare multiple diets that are restricted in methionine but differ in their replacement of amino acids and thereby determine whether the observed effects are robust for all such substitutions. If dose response relationships are also of interest, then the necessary studies become increasingly more complex to instill confidence about the relative impact of individual diet components. Unfortunately, while essential for inferring causal contributions of individual diet components, such multi-arm studies are expensive and burdensome, and involve a large number of comparisons. Embracing diet complexity as a probe of physiology Conceptualizations such as the “Geometric Framework for Nutrition” [27], have encouraged the field to move away from attempts to modify a single dose of a dietary variable and to embrace nutrient substitution matrices that consider potential interactions across dose-response ranges in order to advance our understanding of nutritional phenotypes. This involves conducting large studies using a wide variety of chemically defined (i.e., purified) diets. Detailed phenotyping using physiological characterizations, isotope tracing, multi-omics technologies, and other quantitative methods, together with genetic and pharmaceutical perturbations along proposed causal pathways can be used to strengthen causal attributions and elucidate diet-induced phenotypes. Several investigators have used such a multivariable diet approach to investigate important questions regarding macronutrients and caloric intake, adiposity, reproductive function, and longevity [28–30]. For example, to address the effects of dietary macronutrient distribution on energy intake, expenditure, and adiposity, Hu et al. [31] employed 29 diets across various permutations of total fat (8.3% to 80%), carbohydrate (10% to 80%; 5% to 30% as solid sucrose), and protein (5% to 30%), maintaining the fatty acid composition of the diets in mice. Across all dietary permutations, dietary fat, provided as a single mixed oil source to match the relative saturated, monounsaturated, and polyunsaturated fatty acid composition of the Western diet, was the key driver of energy intake and adiposity through 50%–60% intake, independent of varying protein content and carbohydrate level and types (corn starch, maltodextrin, and sucrose). Using the same 29-diet matrix, Hu et al. [32] failed to detect any further impacts of diet composition on glucose tolerance once the effects of adiposity were controlled. This large undertaking highlights the strength of designs that go beyond the commonly employed two-arm design. However, despite using 29 different diets, additional factors, such as variation in fiber and micronutrient intakes, as well as fat composition and energy density, remained unexplored in these studies, limiting sole causal attribution to total fat intake per se. As researchers embrace diet complexity, it becomes apparent that one single background diet composition is not likely to meet the needs of a research community and that existing formulations may be suboptimal for understanding the complexity of both physiology and disease. There is a great need for renewed interest in modifying diet as an experimental variable in biomedical research to facilitate discovery that includes commonly modified ingredients (e.g., sources of macronutrients) as well as the vast array of bioactive compounds found in natural foodstuffs currently omitted from purified formulations. Design and measurement challenges Assessment of intake Assessing the validity of any test of a diet-related hypothesis requires the assessment of intake, though this is a challenging task. Individual measurements of food intake in group housed animals is not possible unless sophisticated hopper systems enable the intake of chipped individuals to be monitored. In the absence of such a system, reporting food intake of the group on a per animal basis by dividing by the number of animals in the cage can be misleading. However, it cannot be assumed that animals that are group-housed consume equal amounts of food. Furthermore, the standard procedure to measure food intake is by the difference in weight of food in the hopper over a given period. However, this fails to account for food losses, as not everything that leaves the hopper is consumed and bits of food in the bedding must be accounted for to get an accurate measure of intake. The extent of food loss depends on the hardness of the food with softer foods being more easily fragmented and lost [33]. In addition to wastage, all ingested food does not get absorbed. Accurate estimates of total energy absorption therefore need to account for losses in feces (i.e. assimilation efficiency). This is almost never done (but see for example in [31]), yet differences in absorption efficiency between individuals can explain relatively large changes in adiposity over periods of 3–4 weeks [34, 35]. This is a neglected area in rodent dietary studies but has significant implications for experimental interpretation with respect to the relative energy intake between different diets, coprophagy, etc. With respect to the latter, it is important to consider that the nutritional status of rodents is influenced by not only laboratory diet selection but also by consumption of feces (‘coprophagy’) and bedding. Coprophagy is a source of several micronutrients (e.g., vitamins B, K) and other postbiotic bioactives (e.g., short chain fatty acids; plant secondary metabolites) due to microbial metabolism (“postbiotics”) [36–38]. The impact of such factors on rodent nutrition and physiology are increasingly considered in the era of microbiome investigations [39, 40]. For example, choice of bedding has been shown to modify microbiome composition, metabolic outcomes, and fecal energy content [41, 42]. Despite this, few investigators have tested the interaction between dietary interventions and the presence or absence of coprophagy [40]. When food intakes are altered by diet composition, pair-feeding may be employed to separate the effects of the diet itself from its confounding secondary effects on intake and body composition. In a pair-feeding protocol, the mice exposed to the test diet are only supplied with sufficient food to match the calorie intake of the control group. This can be done either with or without adjusting for potential differences in assimilation efficiency between the groups. Pair-feeding, however, can also generate confounding effects through altering the timing of intake, depending on the exact protocol. For example, if the pair-fed group is given their ration around the time of lights out (~ZT12), the animals may consume the entire ration over a short period, thereby creating a prolonged fasting duration relative to the ad libitum control animals. As a result of the differing fasting periods, different metabolic outcomes may result despite the same calorie intake throughout the day [43]. Moreover, in such a scenario, animals sacrificed in mid-afternoon will not only differ in their diet composition but also how long since they last fed. Analogous to pair feeding, many time-restricted feeding protocols match total intakes but introduce confounding via altering the interval between feedings/degree of fasting as well as the timing of intake within the circadian cycle. This type of effect can be removed by using automated feeders that distribute the food to the pair-fed animal by temporally yoking its total intake and the time-matching feedings to that of a specific control. Collectively, total intake, duration of feeding and fasting, as well as intake patterns in relation to the circadian cycle should all be considered as potential factors that may contribute to the impact of diet composition on measured outcomes. Randomization and sample size While it is commonplace to power studies based on individual rodent number, group-housing introduces non-independence of individual-level phenotypic data (e.g., body weight, gene expression, etc.); this non-independence presents a challenge for nearly all experimental variables, but is exacerbated in the case of diet, an intervention delivered at the level of the cage. Thus, sample size and statistical power calculations need to consider diet interventions delivered at the level of the cage as cluster-randomized interventions. The analyses of such data from group housed animals should be handled by including the cage as a random effect in a mixed effects model. Unfortunately, few investigators report designing and analyzing dietary intervention studies in group housed animals appropriately, rarely providing the number of clusters (i.e. number of cages), and sample size per cluster (i.e. number of animals housed per cage). Inappropriately treating such data as independent results in imprecise estimates of statistical precision and artificially low P-values [44, 45]. Even for individually housed mice, detecting significant differences in food intake is challenging due to high intake variability within and between individuals [46, 47]. Failure to detect statistically significant intake differences is often due to underpowered studies and should not be interpreted as indicating no meaningful intake difference. For example, Fischer et al. [48] found a small difference in body fat due to knocking out the FTO gene and claimed that energy intake was not responsible because no statistically significant intake difference was found. However, the energy intake difference necessary to explain the body fat observations with a power of 80% and a significance level of 0.05 would have required a sample size of 4337 per group – compared to the actual sample of around 10. Hence the power to detect a food intake effect was only 5.1% [49]. Individual variation in daily energy intake has a coefficient of variation (CV) between 10% and 17% which itself depends on the diet composition [50]. Supplementary Table S2 shows the sample size necessary to detect different effect sizes on food intake in a standard two sample t-test with a power of 80% and alpha = 0.05. Given that most studies have sample sizes around 10 per group, the effect size that can be detected between groups is only around 15%–25% difference in daily energy intake. However, a sustained 3% increase in energy intake, with mice eating on average 65 kJ/d, would lead to a measurable gain in body weight of ~2.2 g based on the relationship of ~0.9 kJ/d per g of body weight [51], and would require a sample size between 176 and 504 mice per group to detect the energy intake difference. Averaging food intake over multiple days is a useful strategy to reduce variation and improve study power [47]. However, diminishing returns are observed beyond about 10 days of averaging (see Supplementary Fig. S1). More data are needed on temporal patterns and variation of intake in group housed animals to better inform power analyses. Generalizability and translation Age, sex, and strain of rodents are influential variables that interact with nutrition and affect the ability to generalize and translate experimental results. For example, nutrient requirements and metabolic responses to diet vary across the lifespan. Because the average age at death of mice is about 800 days while humans live on average about 80 years, the approximate equivalence between mice and humans is that 10 days for a mouse are roughly equivalent to 1 year for a human. However, the key life transition stages do not precisely line up. Mice wean at the age of 3 weeks (equivalent to ~2 years in humans) and complete linear growth around the age of 12 weeks (equivalent to ~8.5 years in humans). Experiments conducted on mice younger than 12 weeks are not readily translatable to adult humans and important differences in experimental outcomes may depend critically on this age difference. For example, Sørensen et al. [52] found that protein had a strong leveraging effect on food intake in young mice (aged 9 weeks at diet onset), but this effect was not replicated in mice aged 12 weeks at diet onset [31], potentially because protein requirements are much more important during growth than among adults. Moreover, feeding behavior is likely influenced by sex in group-housed rodents. Despite common assumptions that female mice will exhibit greater variability due to estrus cycling [53], group housing of males introduces fighting and barbering stresses that in turn introduce variability in intake and energy homeostasis across dominant and subordinate individuals. Individual females, however, are less stressed when group-housed than those kept solitarily [54]. Nutrient requirement estimates are also not tailored to rodent strains, though diet composition-by-strain interactions have been noted for decades when defining the energy, fat, and protein requirements required for growth and reproduction [55]. Recently, such interactions have been observed in a comprehensive metabolic phenotyping of 4 inbred mouse strains (A/J, B6J, FVB/NJ, NOD/ShiLtJ) fed 6 diets, including traditional mouse (purified low fat/control, Western) and human diets (American, Mediterranean, Human, Japanese) [56]. Robustness of diet effects in the context of systematically heterogenized experimental variables (e.g., strains, sex, age, and microbiome composition) increases confidence that results are generalizable and potentially translatable to humans. Conversely, the lack of consistently observed effects opens up avenues for understanding novel biological factors that are permissive to certain diet effects, an increasingly emphasized concept in the era of “precision” medicine and nutrition. Recommendations and conclusions The complexity of rodent diet intervention studies is challenging to capture in a single brief review, but Table 1 summarizes our major observations about common practices, their limitations, and our recommendations. We have emphasized the important role of diet to elicit widespread physiological changes, so designing and interpreting diet intervention studies should not be inappropriately viewed through a lens of targeting a particular biological mechanism without acknowledging alternative independent mechanisms. Clearly isolating the cause of the diet effects to a single biological mechanism will require multi-omics and quantitative phenotyping approaches, dose-response studies to assess the robustness of the phenotypes, and the incorporation of genetic and pharmacological manipulations to perturb likely causal mechanisms. With these caveats in mind, we believe that carefully designed diet interventions in rodent models hold great promise for both elucidating important metabolic physiology in health and diseases as well as modulating outcomes in ways that may better translate to humans. Supplementary Material supplementary information Acknowledgements KCK received salary support from an NIH Training Grant during the drafting of this manuscript (T32ES027801). Figure 1 Diet design hierarchy. (a) Dietary pattern design. Chow vs. purified design. All nutrients between standard chow (SC) and high-fat diet (HFD) are mismatched. An undetermined component or components within the overall dietary pattern are responsible for the study outcome. No further causal inference is possible on the basis of the experimental diets alone. Addition design. All nutrients between SC and SC + puree (SC + P) are mismatched. An undetermined component or components within the overall dietary pattern are responsible for the study outcome. No further causal inference is possible on the basis of the experimental diets alone. (b) Substitution design. Ad libitum design. Matched protein, micronutrients, and fiber content. Mismatched energy density may cause excess calorie intake in the HFD group compared to low-fat diet (LFD) group. Pair-fed. Matched protein, micronutrients, and fiber content. Mismatched energy density may cause excess calorie intake in the HFD group compared to LFD group; pair feeding can equalize for differences in energy intake. However, pair feeding may introduce differences in feeding times or lengths of feeding windows or volumetric differences with physiological import. Energy density matched. Matched protein, micronutrients, and energy density. However, mismatches in fiber content may have independent physiological import. (c) Multiple substitution design. An extension of the substitution design, a multiple substitution design creates relative rankings of all nutrients of interest with respect to an outcome or outcomes of interest. Figure 2 The prevalence of unrefined comparators for refined diets across top biomedical research journals. The inappropriate use of unrefined dietary formulations as comparators for refined dietary formulations, exhibiting numerous quantified and unquantified compositional differences, is highly prevalent. To estimate the prevalence, all papers published between 2019 and 2020 in the journals Cell Metabolism, Diabetes, Journal of Clinical Investigation, Nature, and Nature Medicine found in PubMed via the search terms “high-fat mouse” and including a diet-versus-diet comparison in at least two groups of mice, were included for analysis, for a total of 91 papers. 14 (15%) papers compared refined to refined diets; 30 (33%) papers compared unrefined to refined diets; and 47 (52%) papers did not provide enough information. These data show that, consistent with previous reports [10, 11], inappropriate dietary design and reporting remains the norm, even in papers published in the highest-impact journals. Table 1 Recommendations for authors, reviewers, and editors of nutritional investigations Domain Commonplace practice Limitation Recommendation Design Unrefined, grain-based ‘chows’ are commonly utilized as ‘control’ comparators for purified diets ‘Chows’ and ‘purified’ diets differ substantially in their composition and present an immeasurably confounded design The use of chows as comparators for purified diets should be limited to pilot investigations and eliminated as the sole source of data for final publication comparisons Various ‘control’ and ‘experimental’ purified diet formulations are purchased and utilized across separate experiments that are ultimately juxtaposed in a manuscript Diet formulations, including those differing in a dietary component of interest, can differ broadly in their overall composition and introduce unquantifiable confounding Work with commercial diet vendors and trained animal nutritionists before undertaking diet investigations to ensure the proposed study diets are feasible and the appropriate control diets are procured An experimental and a control diet are intended to be fed with the aim of targeting a specific metabolic pathway and making causal inferences about an individual ingredient/nutrient in order to inform therapeutic approaches Diets differ in at least 2 variables, limiting attribution to a single component. Effects may be due to any one variable modified, or the substitution, and such effects may be modified by interacting variables in the background diet. Food components exhibit significant pleiotropy, targeting multiple downstream mechanisms, and rarely exhibit linear dose-response relationships Employ multi-level ingredient/nutrient substitution matrix designs where possible that consider substitution effects, assess dose-response relationships, as well as effect modification by background diet composition Harmonize diet design with other elements of the experimental design (e.g., genetic, pharmacological manipulations) that can isolate the contributions of potential relevant mechanisms at play. Employ unbiased, high-throughput approaches (e.g., metabolomics) to characterize the metabolic context in which relevant diet-induced mechanisms occur Experimental diet manipulations are undertaken with limited feasibility or pilot testing and unclear rigor Experimental diet manipulations frequently have unintended consequences that compromise the ability to test the intended hypothesis. Confounded attempts may go unpublished and resources wasted, or published and the confounding impacts of the unintended consequences go unrecognized, minimized, or unreported Undertake pilot studies to confirm expected phenotypes (e.g., weight gain or maintenance), to assess for unintended consequences of dietary manipulations (e.g., food aversion and weight loss, apparent pathology) and to facilitate a priori power analyses) Analysis and reporting Deviations from protocols or in-tandem decisions can introduce biases and compromise rigor; practices may include selective reporting of assessed outcomes, unjustified and/or non-transparent removal of outliers and other protocol modifications, and reporting of spurious findings as significant Open pre-registration of studies should become commonplace, including the intended diet formulations, age at diet manipulation, sex of animal, power calculations, all outcomes to be assessed and their method of assessment, and statistical analysis plan Diet composition is assumed based on the label compositional analysis Diet manufacturing and processing, shipment and/or storage conditions can influence diet composition in unintended manners, resulting in alterations to the concentration of compounds of interest being fed. Diet contains numerous unquantified factors that may be relevant effect modifiers for the outcomes studied Pursue an independent laboratory analysis of commercially purchased diet to ensure expected concentrations of relevant food derived components, especially when conducting long-term studies with >1 lot number. It is advisable that researchers store a frozen aliquot of investigational diets for future analysis and comparison Diet information used throughout the study is not specified (e.g., ‘chow’) and/or listed throughout the text. Referenced diets may or may not be open source. Bedding type and consumable enrichment is rarely reported Without brand and catalog number information, diet composition is challenging to assess. Custom purified diets do not have full diet composition data available through vendor websites, requiring contact with vendors to retrieve such information. Bedding and enrichment can modify metabolism-related phenotypes and cannot be accounted for without reporting Transparently list the name, formulation, and known composition of all feeds used in the investigation in a main or supplementary table. Bedding type and identifier information should be reported as well as consumable enrichment use Total number of rodents is reported per diet group Rodents are group-housed and diet is delivered at the level of the cage, introducing non-independence of the individual animals. Failing to utilize a cluster analysis of such data results in artificially lower estimates of variance and lower P values Clearly report the unit of randomization within a study (cage or individual rodent) and related relevant parameters (e.g., animals per cluster). Choose the appropriate statistical approach and explicitly justify this in manuscripts Investigations modifying diet composition do not report longitudinal changes in food intake and body weight. Conclusions about a diet’s effects on food intake are made regardless of the statistical power to detect an effect. Diet composition and/or feeding protocols may change the pattern of intake (i.e. duration of fasting between feeding intervals, relation of intake to the circadian cycle) The impact of diet composition changes on outcomes may be mediated through alterations in energy balance and body composition rather than through independent effects of the diet component. Many studies are underpowered to detect food intake changes that may underlie phenotypes. Few studies assess whether changes in diet composition or feeding protocol imparts their effects through altering the pattern of intake Report the impact of diet composition modification on longitudinal measures of food intake, body weight, and body composition regardless of whether such variables are the primary outcome of the investigation. Limit conclusions about the impact of diet on components of energy balance when not explicitly powered to do so. Design experiments to manipulate food access to control for alterations in fasting duration and circadian alignment A single diet component is highlighted throughout the manuscript Diets lack a placebo, introducing an inherent relative effect of investigational diets in relation to their selected control. Investigations rarely employ multiple comparators across a dose-response relationship to confidently attribute causal effects to one diet component Titles, abstracts, and in-text descriptions should transparently report effects of the investigational diet relative to comparator diets, highlight relevant substitution effects, and make evident the degree of confidence in the dose-response relationships. Named diets and their compositions should be clearly detailed in the main manuscript leaving readers with a clear appreciation for variables differing between diets Conflict of interest John R. Speakman holds the position of editors-in-chief for Life Metabolism, and is blinded from reviewing or making decisions for the manuscript. The other authors declare that no conflict of interest exists. References 1. Hitchcock CR , Bell ET . Studies on the nematode parasite, Gongylonema neoplasticum (spiroptera neoplasticum), and avitaminosis A in the forestomach of rats: comparison with Fibiger’s results. J Natl Cancer Inst 1952;12 :1345–87.14939031 2. Erling Norrby, Nobel Prizes and Life Sciences. London and Singapore: World Scientific Press, 2010. 3. Council National Research . 1972. Nutrient Requirements of Laboratory Animals: Cat, Guinea Pig, Hamster, Monkey, Mouse, Rat: Second Revised Edition, 1972. Washington, DC: The National Academies Press, 1972. 4. Knapka JJ , Smith KP , Judge FJ . Effect of open and closed formula rations on the performance of three strains of laboratory mice. Lab Anim Sci 1974;24 :480–7.4365336 5. Rao GN . New diet (NTP-2000) for rats in the National Toxicology Program toxicity and carcinogenicity studies. Fundam Appl Toxicol 1996;32 :102–8.8812243 6. Reeves PG , Nielsen FH , Fahey GC Jr . AIN-93 purified diets for laboratory rodents: final report of the American Institute of Nutrition ad hoc writing committee on the reformulation of the AIN-76A rodent diet. J Nutr. 199;123 :1939–51. 7. Rees ML , Gioscia-Ryan RA , McCune SA The AIN-76A defined rodent diet accelerates the development of heart failure in SHHF rats: a cautionary note on its use in cardiac studies. J Anim Physiol Anim Nutr (Berl). 2014 Feb;98 :56–64.23298172 8. Diet, Nutrition and the Prevention of Chronic Diseases. 2003. World Health Organization Technical Report Series 916 : i–viii , 1–149, backcover. 9. National Academies of Sciences, Engineering, and Medicine, Health and Medicine Division, Food and Nutrition Board, and Committee on the Development of Guiding Principles for the Inclusion of Chronic Disease Endpoints in Future Dietary Reference Intakes. 2017. Guiding Principles for Developing Dietary Reference Intakes Based on Chronic Disease. Edited by Oria Maria P. and Kumanyika Shiriki . Washington, DC: National Academies Press (US), 2017. 10. Warden CH , Fisler JS . Comparisons of diets used in animal models of high-fat feeding. Cell Metab 2008;7 :277.18396128 11. Pellizzon MA , Ricci MR . The common use of improper control diets in diet-induced metabolic disease research confounds data interpretation: the fiber factor. Nutr Metab (Lond) 2018;15 :3.29371873 12. Tuck CJ , De Palma G , Takami K Nutritional profile of rodent diets impacts experimental reproducibility in microbiome preclinical research. Sci Rep 2020;10 :17784.33082369 13. Pellizzon MA , Ricci MR . Choice of laboratory rodent diet may confound data interpretation and reproducibility. Curr Dev Nutr 2020;4 :nzaa031.32258990 14. Greenman DL , Oller WL , Littlefield NA Commercial laboratory animal diets: toxicant and nutrient variability. J Toxicol Environ Health 1980;6 :235–46.7392092 15. Wise A , Gilburt DJ . The variability of dietary fibre in laboratory animal diets and its relevance to the control of experimental conditions. Food Cosmet Toxicol 1980;18 :643–8.6259033 16. Newberne PM , Fox JG . Nutritional adequacy and quality control of rodent diets. Lab Anim Sci 1980;30 :352–65.6892043 17. Newberne PM , Sotnikov AV . Diet: the neglected variable in chemical safety evaluations. Toxicol Pathol 1996;24 :746–56.8994306 18. Liu R , Cheng M , Kothapalli KSD Glycerol derived process contaminants in refined coconut oil induce cholesterol synthesis in HepG2 cells. Food Chem Toxicol 2019;127 :135–42.30878531 19. Turnbaugh PJ , Ley RE , Mahowald MA An obesity-associated gut microbiome with increased capacity for energy harvest. Nature 2006;444 :1027–31.17183312 20. Ley RE , Bäckhed F , Turnbaugh P Obesity alters gut microbial ecology. Proc Natl Acad Sci U S A 2005;102 :11070–5.16033867 21. Turnbaugh PJ , Bäckhed F , Fulton L Diet-induced obesity is linked to marked but reversible alterations in the mouse distal gut microbiome. Cell Host Microbe 2008;3 :213–23.18407065 22. Dalby MJ , Ross AW , Walker AW Dietary uncoupling of gut microbiota and energy harvesting from obesity and glucose tolerance in mice. Cell Rep 2017;21 :1521–33.29117558 23. Ables GP , Perrone CE , Orentreich D Methionine-restricted C57BL/6J mice are resistant to diet-induced obesity and insulin resistance but have low bone density. PLoS One 2012;7 :e51357.23236485 24. Plaisance EP , Henagan TM , Echlin H Role of β-adrenergic receptors in the hyperphagic and hypermetabolic responses to dietary methionine restriction. Am J Physiol Regul Integr Comp Physiol 2010;299 :R740–50.20554934 25. Hasek BE , Stewart LK , Henagan TM Dietary methionine restriction enhances metabolic flexibility and increases uncoupled respiration in both fed and fasted states. Am J Physiol Regul Integr Comp Physiol 2010;299 :R728–39.20538896 26. Zhang X , Sergin I , Evans TD High-protein diets increase cardiovascular risk by activating macrophage mTOR to suppress mitophagy. Nat Metab 2020;2 :110–25.32128508 27. Simpson SJ , Le Couteur DG , James DE The Geometric Framework for Nutrition as a tool in precision medicine. Nutr Healthy Aging 2017;4 :217–26.29276791 28. Solon-Biet SM , McMahon AC , Ballard JW The ratio of macronutrients, not caloric intake, dictates cardiometabolic health, aging, and longevity in ad libitum-fed mice. Cell Metab 2014;19 :418–30.24606899 29. Solon-Biet SM , Mitchell SJ , Coogan SC Dietary protein to carbohydrate ratio and caloric restriction: comparing metabolic outcomes in mice. Cell Rep 2015;11 :1529–34.26027933 30. Solon-Biet SM , Walters KA , Simanainen UK Macronutrient balance, reproductive function, and lifespan in aging mice. Proc Natl Acad Sci U S A 2015;112 :3481–6.25733862 31. Hu S , Wang L , Yang D Dietary fat, but not protein or carbohydrate, regulates energy intake and causes adiposity in mice. Cell Metab 2018;28 :415–31.e4.30017356 32. Hu S , Togo J , Wang L Effects of dietary macronutrients and body composition on glucose homeostasis in mice. Natl Sci Rev 2020 Aug 6;8 :nwaa177.34691555 33. Cameron KM , Speakman JR . The extent and function of ‘food grinding’ in the laboratory mouse (Mus musculus). Lab Anim 2010;44 :298–304.20457827 34. Król E , Duncan JS , Redman P Photoperiod regulates leptin sensitivity in field voles, Microtus agrestis. J Comp Physiol B 2006;176 :153–63.16402185 35. Król E , Redman P , Thomson PJ Effect of photoperiod on body mass, food intake and body composition in the field vole, Microtus agrestis. J Exp Biol 2005;208 :571–84.15671345 36. Giovannetti PM . Effect of Coprophagy on Nutrition. Nutr Res 1982;2 :335–49. 37. Giovannetti PM , Stothers SC , Parker RJ . Coprophagy preventionand availability of amino acid in wheat for the growing rat. Can J Anim Sci 1970;50 :269–77. 38. Barnes RH , Fiala G , McGehee B Prevention of coprophagy in the rat. J Nutr 1957;63 :489–98.13492082 39. Bo TB , Zhang XY , Kohl KD Coprophagy prevention alters microbiome, metabolism, neurochemistry, and cognitive behavior in a small mammal. ISME J 2020;14 :2625–45.32632263 40. Bogatyrev SR , Rolando JC , Ismagilov RF . Self-reinoculation with fecal flora changes microbiota density and composition leading to an altered bile-acid profile in the mouse small intestine. Microbiome 2020;8 :19.32051033 41. Ambery AG , Tackett L , Penque BA Effect of corncob bedding on feed conversion efficiency in a high-fat diet-induced prediabetic model in C57Bl/6J mice. J Am Assoc Lab Anim Sci 2014;53 :449–51.25255066 42. Gregor A , Fragner L , Trajanoski S Cage bedding modifies metabolic and gut microbiota profiles in mouse studies applying dietary restriction. Sci Rep 2020;10 :20835.33257713 43. Pak HH , Haws SA , Green CL Fasting drives the metabolic, molecular and geroprotective effects of a calorie-restricted diet in mice. Nat Metab 2021;3 :1327–41.34663973 44. Donner A , Klar N . Pitfalls of and controversies in cluster randomization trials. Am J Public Health 2004;94 :416–22.14998805 45. Vorland CJ , Brown AW , Dawson JA Errors in the implementation, analysis, and reporting of randomization within obesity and nutrition research: a guide to their avoidance. Int J Obes (Lond). 2021;45 :2335–46.34326476 46. Zhang LN , Morgan DG , Clapham JC Factors predicting nongenetic variability in body weight gain induced by a high-fat diet in inbred C57BL/6J mice. Obesity (Silver Spring) 2012;20 :1179–88.21720432 47. Selimkhanov J , Thompson WC , Patterson TA Evaluation of a Mathematical Model of Rat Body Weight Regulation in Application to Caloric Restriction and Drug Treatment Studies. PLoS One 2016;11 :e0155674.27227543 48. Fischer J , Koch L , Emmerling C Inactivation of the Fto gene protects from obesity. Nature 2009;458 :894–8.19234441 49. Speakman JR . FTO effect on energy demand versus food intake. Nature 2010;464 :E1; discussion E2.20360686 50. Wu Y , Hu S , Yang D Increased variation in body weight and food intake is related to increased dietary fat but not increased carbohydrate or protein in mice. Front Nutr 2022;9 :835536.35360679 51. Guo J , Jou W , Gavrilova O , Hall KD . Persistent diet-induced obesity in male C57BL/6 mice resulting from temporary obesigenic diets. PLoS One 2009;4 :e5370.19401758 52. Sørensen A , Mayntz D , Raubenheimer D Protein-leverage in mice: the geometry of macronutrient balancing and consequences for fat deposition. Obesity (Silver Spring). 2008;16 :566–71.18239565 53. Hughes RN . Sex still matters: has the prevalence of male-only studies of drug effects on rodent behaviour changed during the past decade? Behav Pharmacol 2019;30 :95–9.29847339 54. Takahashi A Toward understanding the sex differences in the biological mechanism of social stress in mouse models. Front Psychiatry 2021;12 :644161.33664686 55. National Research Council (US) Subcommittee on Laboratory Animal Nutrition. Nutrient Requirements of Laboratory Animals: Fourth Revised Edition, 1995. Washington, DC: National Academies Press (US), 1995. 56. Barrington WT , Wulfridge P , Wells AE Improving metabolic health through precision dietetics in mice. Genetics 2018;208 :399–417.29158425

PMC010xxxxxx/PMC10363995.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101135941 32136 Macromol Biosci Macromol Biosci Macromolecular bioscience 1616-5187 1616-5195 36905285 10363995 10.1002/mabi.202300011 NIHMS1883472 Article In Vitro Proof of Concept of a First-Generation Growth-Accommodating Heart Valved Conduit for Pediatric Use Li Richard L. Dr. ab Sun Mingze Dr. a Russ Jonathan B. Dr. c Pousse Pierre-Louis Dr. a Kossar Alexander P. Dr. a Gibson Isabel Dr. a Paschalides Costas b Herschman Abigail R. ab Abyaneh Maryam H. Dr. a Ferrari Giovanni Prof. a Bacha Emile Dr. a Waisman Haim Prof. c Vedula Vijay Prof. b Kysar Jeffrey W. Prof. bd Kalfa David Dr. a a Department of Surgery, Division of Cardiac, Thoracic and Vascular Surgery, Section of Pediatric and Congenital Cardiac Surgery, New-York Presbyterian - Morgan Stanley Children's Hospital, Columbia University Medical Center, 3959 Broadway, CHN-274, New York, NY 10032 USA b Department of Mechanical Engineering, Fu Foundation School of Engineering and Applied Science, Columbia University, 220 Mudd Building, 500 W. 120th Street, New York, NY 10027 USA c Department of Civil Engineering and Engineering Mechanics, Fu Foundation School of Engineering and Applied Science, Columbia University, 610 Mudd Building, 500 W. 120th Street, New York, NY 10027 USA d Department of Otolaryngology – Head and Neck Surgery, Columbia University Medical Center, 3959 Broadway, 5th Floor, New York, NY 10032 USA Corresponding authors’ information: Dr. D. Kalfa, MD, PhD, Pediatric Cardiac Surgery, New-York Presbyterian - Morgan Stanley Children's Hospital, Columbia University Medical Center, 3959 Broadway, CHN-274, New York, NY 10032 USA, Tel.: (212) 305-5975; fax: (212) 305-4408, dk2757@cumc.columbia.edu; Prof. J. W. Kysar, PhD, Department of Mechanical Engineering, Columbia University, 240 Mudd Building, Mail Code 4703, 500 W. 120th Street, New York, NY 10027 USA, Tel.: (212) 854-7432; fax: (212) 854-3304, jk2079@columbia.edu 30 3 2023 7 2023 22 3 2023 24 7 2023 23 7 e2300011e2300011 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Currently available heart valve prostheses have no growth potential, requiring children with heart valve diseases to endure multiple valve replacement surgeries with compounding risks. This study demonstrates the in vitro proof of concept of a biostable polymeric trileaflet valved conduit designed for surgical implantation and subsequent expansion via transcatheter balloon dilation to accommodate the growth of pediatric patients and delay or avoid repeated open-heart surgeries. The valved conduit is formed via dip molding using a polydimethylsiloxane-based polyurethane, a biocompatible material shown here to be capable of permanent stretching under mechanical loading. The valve leaflets are designed with an increased coaptation area to preserve valve competence at expanded diameters. Four 22 mm diameter valved conduits are tested in vitro for hydrodynamics, balloon dilated to new permanent diameters of 23.26 ± 0.38 mm, and then tested again. Upon further dilation, two valved conduits sustain leaflet tears, while the two surviving devices reach final diameters of 24.38 ± 0.19 mm. After each successful dilation, the valved conduits show increased effective orifice areas and decreased transvalvular pressure differentials while maintaining low regurgitation. These results demonstrate concept feasibility and motivate further development of a polymeric balloon-expandable device to replace valves in children and avoid reoperations. Graphical Abstract We report the in vitro proof of concept of a biostable polymeric valved conduit, designed to accommodate growth via transcatheter balloon dilation. Four devices are initially expanded from 22 to 23.26 mm in diameter. Two devices are further dilated to 24.38 mm, while the remaining two sustain leaflet tears. The surviving devices demonstrate increased orifice areas and decreased transvalvular gradients. growth accommodation valved conduit polymeric valves right ventricular outflow tract congenital heart disease pmc1 Introduction Congenital heart disease (CHD) is reported in approximately 1% of all live births, affecting ~40,000 babies per year in the U.S. and ~1.4 million worldwide.[1] Common conditions associated with CHD include atrial and ventricular septal defects (holes in the internal walls of the heart), aortic and pulmonary stenoses (narrowing of the valves and arteries), and tetralogy of Fallot (a combination of a ventricular septal defect, pulmonary stenosis, right ventricular hypertrophy, and an overriding aorta).[2] More than half of children born with CHD require open-heart surgery to correct these malformations, and more than 20% of those who require surgery will need the implantation of a valve or valved conduit to repair the right ventricular outflow tract (RVOT).[1,3,4] However, all currently available valved prostheses, including cryopreserved homografts, xenograft conduits, bioprostheses, and mechanical prostheses, have serious limitations resulting in high rates of reintervention.[5,6] Tissue valves exhibit poor durability due to structural valve degeneration (SVD), which encompasses thickening, calcification, tearing, or other disruptions to the leaflet tissue leading to stenosis or regurgitation.[7-10] Mechanical valves do not degrade substantially, but they are susceptible to obstruction from pannus formation.[11] Other drawbacks of mechanical valves include their larger size, which limits their use in younger patients,[6] and their hinged design, which creates non-physiological flow patterns that increase the risk of thrombosis and necessitate anticoagulation therapy.[12] Most critically, all these devices are designed to function at a fixed size and are constructed from non-living tissue or rigid materials that do not adapt to the patient’s somatic growth. As a result of these limitations, children with prosthetic valves require one to four reoperations to replace the valve before they reach adulthood,[11,13,14] with each additional open-heart surgery carrying a 1-15% risk of death, bleeding, infection or multi-organ dysfunction.[5,15,16] Commonly reported reasons for reoperations in pediatric patients include somatic outgrowth,[11,17] complications due to SVD in tissue valves (e.g. stenosis, regurgitation),[13,18] and in rarer cases, endocarditis and aneurysm.[11,13] Both outgrowth and SVD can occur independently, as SVD is a common occurrence even in fully grown adults.[8,9] However, SVD has been shown to be accelerated in children, with an increase in the frequency of required reoperations.[18-20] While this phenomenon has been attributed to intensified immunological responses, altered blood biochemistry, and increased calcium metabolism,[21,22] another possible mechanism driving acceleration of SVD is outgrowth. An outgrown or undersized valve (which is effectively stenotic) creates high pressure gradients and turbulent flow that can trigger a cascade effect: increased internal leaflet stresses, intimal proliferation, macrophage infiltration[22] leading to reactive-oxygen-species-mediated (ROS) oxidation,[23,24] and leaflet thickening resulting in further stress increases, calcification, degradation, and mechanical failure of the valve.[14,22,25-27] A new valved device that can accommodate a child’s growth would reduce the need for reoperations and greatly improve the standard of care for children and adults with CHD. The ideal device will last from the newborn stage well into adulthood, although the elimination of even one reoperation to upsize a valved prosthesis would have a significant clinical impact. There have been attempts to surgically implant percutaneous stented valves, such as the Melody™ transcatheter pulmonary valve,[28-32] for later intervention by transcatheter balloon dilation. However, these subsequent dilations have occasionally resulted in severe regurgitation[29] because the Melody™ was not originally designed to be competent across a large range of diameters – there is insufficient coaptation area, or leaflet contact, at larger diameters to enable valve closure – or to be expanded multiple times. There is now an increased focus on developing valved devices specifically to accommodate a child’s growth. The Autus Valve (Boston, Massachusetts) is an incrementally balloon-expandable valve mimicking the bileaflet geometry of a human venous valve.[33] Designed for pulmonary valve replacement and growth accommodation in pediatric patients, it is currently undergoing human clinical trials.[34] Meanwhile, Draper Laboratory, Inc. (Cambridge, Massachusetts) is developing the LEAP valve, which has an adaptive stent that expands without the need for balloon interventions.[35] However, none of these devices are integrated with expandable conduits designed for reconstruction of the RVOT. In more than half of the patients requiring pulmonary valve implantations and in all such neonates, infants, and children up to the age of 5-6 years, the RVOT is entirely absent or insufficiently sized.[36] Thus, a valved conduit is the major clinical need for this population, as a valve-only device would not be adequate to fully reconstruct the RVOT and properly accommodate growth. An expandable valve-only prosthesis could potentially be inserted into an expandable vascular graft. However, such a configuration has not been studied, and the effect of balloon expansion on the cohesion between these separate components is unknown. We report the first in vitro proof of concept of a growth-accommodating polymeric valve with an integrated conduit – a valved conduit – that can be implanted surgically for pediatric RVOT reconstruction and subsequently expanded via transcatheter balloon dilation to match part of the patient’s growth into adulthood, thus being suitable for adult patients as well. We hypothesize that the expandability of the device can be achieved by using a permanently deformable polymeric material, in contrast to current mechanical and tissue-based valves, and that valve competence can be maintained at expanded diameters by designing the valve leaflets with an increased coaptation area, where this increased area is obtained by increasing the leaflet coaptation height and free edge length. Our objective here was to select biostable polymers that meet the mechanical requirements, design and fabricate the first generation of a fully polymeric biostable valved conduit, and test it in vitro before and after balloon dilations. 2 Results 2.1 Materials Characterization and Selection Two commercially available, biostable, and biocompatible polymers, Carbothane™ AC-4075A (Lubrizol, Cleveland, Ohio), which is a polycarbonate urethane (PCU), and Elast-Eon™ E5-325 (Biomerics, Salt Lake City, Utah), which is a polydimethylsiloxane-based (PDMS) polyurethane (PU), were identified as potential materials for the valved conduit proof of concept due to their high compliance matching that of native heart valve tissue[37,38] and their excellent biocompatibility.[39-44] Furthermore, Carbothane™ and Elast-Eon™ are known to be processable by dip molding, a fabrication technique which has produced excellent results for polymeric valves.[38,45] These materials were further evaluated by mechanical and biocompatibility testing to determine their suitability for constructing the device. 2.1.1 Mechanical Testing The objective of the mechanical tests was to assess the capacity of the materials for permanent (inelastic) deformation as a result of stretching under uniaxial mechanical loading. Herein, we refer to the amount of deformation of a sample, whether temporary or permanent, by the stretch ratio λ = L/L0, where L is the length of the uniformly deformed sample and L0 is its initial length. First, to determine the extent to which the materials could be stretched, we obtained their uniaxial elongations at break and ultimate strengths. The Carbothane™ samples (n = 5) had an elongation at break of λult = 5.81 ± 0.08 and a nominal tensile strength of σult = 48.40 ± 2.52 MPa (mean ± s.d.). The Elast-Eon™ samples (n = 5) were stretched to the crosshead travel limit of the tensile testing machine at λ = 6, but they did not break. The corresponding nominal tensile stress at λ = 6 was σ = 13.12 ± 0.27 MPa. The time-dependent viscoelastic behaviors of Carbothane™ and Elast-Eon™ (n = 4 samples each group) were characterized by uniaxial stress relaxation tests. Averaged results in Figure 1a show that the stress decreased quickly within the first 50 seconds. By 300 seconds, the stress had begun to asymptotically approach a stable limit, indicating that most of the viscoelastic response had dissipated. The remaining stress at the stable limit represents elastic energy being stored. These results provide an estimate of the time scale of the transient viscoelastic behavior in Carbothane™ and Elast-Eon™. Individual samples of the two materials were each subjected to a single temporary stretch λtemp and allowed to recover. The amount of permanent stretch λperm remaining after 24 hours was recorded. Since the stress relaxation responses had approached a stable limit within 300 seconds, we determined that 24 hours was long enough to properly account for time-dependent viscous strain. Both Elast-Eon™ and Carbothane™ exhibited elastomeric mechanical behavior, with significant elastic and viscoelastic recovery from large deformations and relatively small amounts of permanent stretch (Figure 1b-c). Stretching Elast-Eon™ by λtemp = 5 resulted in a permanent stretch of λperm = 1.49 ± 0.03 (n = 4, mean ± s.d.) (Figure 1b), while the same test for Carbothane™ resulted in λperm = 1.36 ± 0.06 (n = 8) (Figure 1c). Stretching Elast-Eon™ by λtemp = 2, 3, and 4 resulted in permanent stretches of 1.06 ± 0.03 (n = 7), 1.14 ± 0.03 (n = 7), and 1.33 ± 0.06 (n = 5), respectively. Stretching Carbothane™ by λtemp = 2, 3, and 4 resulted in permanent stretches of 1.01 ± 0.01 (n = 4), 1.04 ± 0.03 (n = 5), and 1.31 ± 0.06 (n = 4), respectively (Figure 1d). 2.1.2 In Vivo Biocompatibility Testing The biocompatibility of the following polymers was evaluated in a rat subcutaneous model[46-49] of implantation: 1) non-stretched Carbothane™, 2) non-stretched Elast-Eon™, 3) Elast-Eon™ which had been temporarily stretched by 2x, resulting in a 1.1x permanent pre-stretch, and 4) FDA-approved expanded polytetrafluoroethylene (ePTFE) (GORE® PRECLUDE® Pericardial Membrane, W. L. Gore & Associates, Flagstaff, Arizona) as the control group since multiple ePTFE-based valvular devices are currently used in clinical practice and have shown excellent biocompatibility, biostability and durability.[50-53] Figure 1 shows histological sections of the materials upon explantation after 2 months. The pink color in Figure 1e-h indicates the formation of encapsulating tissue, which is the inevitable host response to the implantation of an artificial material. Figure 1e-l shows that the Carbothane™ and Elast-Eon™ samples had good biocompatibility, displaying a lack of cell penetration (lack of foreign body response) and lack of calcification (lack of immunological response) similar to the ePTFE control patches. There was also no visible difference between the pre-stretched and non-stretched Elast-Eon™ samples. The mechanical tests of both Elast-Eon™ and Carbothane™ showed only limited permanent stretch. Nevertheless, Elast-Eon™ retained permanent stretches that were greater than in Carbothane™ (λtemp = 2, p = 0.008; λtemp = 3, p = 5e-5; λtemp = 4, p = 0.65; λtemp = 5, p = 0.0038) and substantial enough to demonstrate the proof of concept of growth accommodation. Elast-Eon™ also demonstrated good biocompatibility and was therefore selected for fabricating the first-generation valved conduit prototype. 2.2 First-Generation Design The design of the valved conduit comprises a cylindrical conduit with a trileaflet valve located at the center (Figure 2a). The geometry of the valve leaflets is elliptical along the radial direction and hyperbolic along the circumferential direction, following equations previously described by Mackay et al.[54] To ensure the persistence of valve competence at larger diameters after the expansion, this design was modified (SolidWorks, Dassault Systèmes, Waltham, Massachusetts) to have an increased coaptation area. The associated increased coaptation height h was calculated as (1) h=l2−b2, where h is the height of a right triangle having base length b equal to the initial conduit radius and hypotenuse length l equal to the expanded conduit radius (Figure 2b). The new length of the free edge was set as 2l using a triangular profile. In this study, the valved conduit was designed to have an initial diameter of 22 mm. Since the mechanical tests for Elast-Eon™ showed that λtemp = 2 would produce λperm = 1.06 ± 0.03, it was calculated that a 44 mm temporary balloon dilation (near the limit of clinical feasibility in adolescents) of the 22 mm device would yield a final diameter of 23.32 mm (λperm = 1.06). Using b=11mm and l=12mm (23.32 mm / 2 ≈ 12 mm) in Equation (1), the increased coaptation height was estimated to be h=4.8mm. After rounding up to h=5mm, l was recalculated as l=12.1mm, and the new length of the free edge was 2l=24.2mm. The length of the conduit was ~9 cm to fit the pulse duplicator testing fixture. 2.3 Fabrication We used dip molding in Elast-Eon™ to fabricate four 22 mm diameter valved conduit prototypes (Devices #1-4) (Figure 2c). The resulting conduit wall and leaflet thicknesses are shown in Figure 3. There was some variation in these thicknesses along the length of the device, as expected from the flow of synthetic polymer solution with the dip molding technique. 2.4 In Vitro Evaluation The four valved conduits were tested in vitro for hydrodynamics in a heart valve pulse duplicator, successively dilated to larger diameters, and tested again after each dilation for hydrodynamics. The devices were dilated using a Coda balloon catheter first to a maximum diameter of 40 mm (λtemp = 1.8) (Figure 4a-b1) and then immediately released. The duration of time to fully inflate and then deflate the balloon was ~1 minute, excluding intermittent breaks to refill the syringe and monitor the diameter of inflation. After 24 hours, the devices had recovered to new permanent diameters of 23.26 ± 0.38 mm (mean ± s.d.) (λperm = 1.06 ± 0.02) (Figure 4c1). A second dilation was performed to a maximum diameter of 44 mm (λtemp = 2) (Figure 4b2). During this second dilation, two of the devices, Device #3 and Device #4, sustained tears in their leaflets. The two surviving devices, Device #1 and Device #2, recovered after the second dilation to new permanent diameters of 24.38 ± 0.19 mm (λtemp = 1.11 ± 0.01) (Figure 4c2), showing a greater amount of permanent stretch compared to the uniaxial tests of Elast-Eon™ (λtemp = 2 and λperm = 1.06 ± 0.03). Hydrodynamic testing of the four devices under adolescent/adult pulmonary conditions showed good valve function in the pre-dilated state, with effective orifice area (EOA), mean positive pressure differential (PPD), and regurgitant fraction (RF) all at acceptable levels. Device #1 had the thickest leaflets and the highest mean PPD among the four prototypes. Valve competence was maintained after the first balloon dilation, with increased EOA and decreased mean PPD for all four devices (Table 1). After the second dilation, the two surviving devices both showed additional increases in EOA and decreases in mean PPD. While there are no standard performance requirements for valved prostheses in the pulmonic position, the minimum ISO 5840-2 requirements for EOA in the aortic position were exceeded by nearly all the devices before and after the balloon dilations, except for Device #1 just before and after the first dilation.[55] RF increased after each balloon dilation, but it was maintained well below the maximum of 10% per the ISO 5840–2 requirements for aortic valves. Readings from the pulse duplicator (Figure 5a-e) showed oscillations in flow and ventricular pressure during the valve closing phase. Smaller oscillations in arterial pressure also occurred during valve opening and coincided with visible leaflet flutter (Video S1). The magnitude and frequency of these oscillations generally decreased after each successive balloon dilation. The effect of the balloon dilation on the valve coaptation height was investigated experimentally. We had modified the original leaflet design of Mackay et al.[54] to have an increased coaptation height h, as defined in Equation (1), in the pre-dilation state. Here, h′ is defined as the corresponding dimension (i.e. the remainder of the increased coaptation height) in the post-dilation state. Figure 6 schematically shows valved conduits in the pre-dilation state and after recovering from the first balloon dilation, subjected to fluid pressures of 3 mmHg and 25 mmHg from the distal end to induce valve closure. At a pressure of 3 mmHg, the observed increased coaptation height in the pre-dilation valves was h3=2.64±0.34mm (mean ± s.d.). The increased coaptation height was smaller in the post-dilation valves, with h3′=2.11±0.48mm. The same trend was observed at a pressure of 25 mmHg, with h25=2.17±0.12mm and h25′=1.83±0.32mm. 2.5 Computational Modeling of the Balloon Dilation A 22 mm valved conduit finite element model was temporarily dilated to a maximum diameter of 44 mm to mimic the balloon dilation experiment (Video S1). The resulting distribution of permanent deformation was non-uniform in the valve region, with the magnitude of deformation being greatest along the leaflet attachment areas (Figure 7). Since the final expanded geometry of the model was irregular and not cylindrical, a final diameter of 24.9 ± 1.0 mm (mean ± s.d.) was calculated from the mean value of the conduit circumference in the valve region (i.e. the section of conduit encompassing the height of the leaflets). This numerical prediction of permanent stretch (λtemp = 2 and λperm = 1.13 ± 0.05) showed good agreement with the experimental balloon dilations of the valved conduits (λtemp = 2 and λperm = 1.11 ± 0.01), but was greater than the results from uniaxial testing of Elast-Eon™ (λtemp = 2 and λperm = 1.06 ± 0.03). 3 Discussion Our ultimate goal is to develop an integrated valve and conduit device that can be expanded from 12 mm (neonatal size) to 24 mm (adult size) in diameter using multiple incremental balloon dilations. These Elast-Eon™-based prototypes represent the first proof of concept of an expandable valved conduit with balloon dilations from 22 mm (adolescent size) to over 23 mm in diameter. Additionally, we demonstrated the feasibility of subsequent dilations to reach a diameter over 24 mm, which holds clinical significance since the total expansion from 22 mm to 24 mm would potentially eliminate one reoperation to upsize a valved prosthesis while accommodating an adolescent into adulthood. The tearing observed in two of the devices during their second dilations is likely due to limitations from our manual fabrication process and could be resolved through manufacturing process improvements or with the future development of new biomaterials. In vitro testing of the fabricated devices showed excellent valve performance both pre- and post-dilation, with EOA increasing after each successful dilation. The computational model of the device expansion demonstrated good agreement with the experimental dilations while predicting a non-uniform valve expansion which must be considered in future designs. 3.1 Materials Characterization and Selection While Elast-Eon™ generally exhibited elastomeric behavior, the mechanical tests also showed that it can be permanently stretched, a property that may be used to accommodate patient growth. The resulting permanent stretches were governed by the magnitudes of the temporary stretches and could be reliably reproduced to facilitate predictable growth. The actual amounts of permanent stretch in Elast-Eon™ were small, yet they were still sufficient to demonstrate a proof of concept of growth accommodation. Elast-Eon™ also demonstrated excellent biocompatibility and biostability in a rat subcutaneous model, even after stretching. This critical result suggests that the permanent balloon dilations would not cause the device’s biological properties to substantially degrade. Ultimately, neither Elast-Eon™ nor Carbothane™ is an adequate material for achieving the desired permanent 2x expansion that would be needed long-term to expand a neonatal size device to the adult size. Temporary 2x stretches, which approach the limit of clinical feasibility, resulted in no greater than a 1.06x permanent stretch for either material during mechanical testing. Meanwhile, excessively large 5x temporary stretches, which are clinically infeasible, resulted in no greater than a ~1.5x permanent stretch. A biomaterial with a greater capacity for permanent, or plastic, deformation would be more desirable for this growth-accommodating device and is an opportunity for future polymer development. 3.2 First-Generation Design The valve was designed with an increased coaptation area to ensure competence at expanded diameters. This increased area was achieved by increasing the leaflets’ radial length and free edge length. Clinically, our experience with the Ozaki technique[56,57] for aortic valve reconstruction in children has shown that a valve designed with increased radial and free edge lengths is compatible with patient growth. However, in the Ozaki design, the increased free edge length is obtained by increasing the width of the entire leaflet. This forces the leaflets, which are cut from a sheet of autologous pericardium, to initially adopt a redundant sinusoidal curvature along the free edge. A dip-molded polymeric valve with such a sinusoidal curvature would likely have poor hydrodynamic function. By fabricating the valve via dip molding, the polymer microstructure is set to remember the geometry of the mold. If the polymeric leaflets were sinusoidally shaped, they would tend to keep their sinusoidal curvature under loading and consequently have decreased mobility and increased resistance during valve opening. Furthermore, as the valve diameter was increased, the sinusoidal leaflets would be forced to unfold and flatten, resulting in higher internal leaflet stresses and reduced durability. We avoided these potential drawbacks by lengthening the leaflets in the radial direction alone and creating a triangular profile to increase the free edge length (Figure 2b), while still obtaining a sufficiently increased coaptation area in the dip-molded valve. When the valved conduits in this study were expanded, the valve leaflets likely sustained some permanent circumferential stretching that assisted in achieving valve closure at the larger diameter. However, when determining the increased coaptation height and new free edge length required to cover the new lumen area, this circumferential stretching was excluded from the calculations. Also excluded was the potential leaflet elongation that could occur due to extended cycling (i.e. creep) under physiological conditions.[58,59] A trileaflet design was chosen to mimic the three leaflets of a native pulmonary valve and to maintain the associated physiological flow patterns.[60,61] Trileaflet polymeric valves have shown excellent hemodynamic function and durability.[38,45,52,53] In particular, we selected the ellipto-hyperbolic trileaflet geometry since previous valves with this design sustained over 500 million cycles (equivalent to ~13 years) during in vitro fatigue testing.[54] However, future work will also explore bileaflet valve designs, motivated by good clinical outcomes from bileaflet polymeric valve implantation for RVOT repair.[50,62-64] Additionally, Hofferberth et al. demonstrated excellent growth potential and valvular performance with a bileaflet venous valve geometry.[33,65] 3.3 Fabrication Dip molding is known to be a complex process due to the continuous flow of the liquid polymer and the many parameters that must be controlled, including polymer viscosity, dipping speed, and drying position and temperature.[66] Our manual dip molding process resulted in valved conduits with some variability in conduit wall and leaflet thicknesses (Figure 3). This variability was due to the downward flow of liquid Elast-Eon™ solution along the valved conduit mold, as it was held upright to avoid flow of excess liquid polymer into the void between the two mold halves during the dipping and drying processes. Nevertheless, the conduit thickness variations had little effect on the uniformity of the balloon dilations, as all the devices were expanded to similar permanent diameters (Table 1). The resulting range of leaflet thicknesses was comparable to the dip-molded polyurethane valves of Mackay et al.[54] which reached over 500 million cycles in accelerated wear testing and to drop-coated PCU valves which reached 1 billion cycles.[67] However, further optimization of the fabrication process is desired, as control of leaflet and conduit wall thickness and uniformity are imperative to long-term hydrodynamic performance and durability.[68,69] Leaflet non-uniformity and insufficient leaflet thickness may also have contributed to the tearing of Devices #3 and #4 upon the second set of balloon dilations. Manufacturing process improvements could include robotic mechanisms for controlling the dipping speed and for tumbling the mold while drying to improve uniformity of polymer distribution,[70,71] as well as spray coating.[72] 3.4 In Vitro Evaluation Four valved conduit prototypes were permanently balloon-expanded from a diameter of 22 mm to 23.26 ± 0.38 mm, and two of the prototypes were further expanded to 24.38 ± 0.19 mm. The devices demonstrated increased EOA, lowered mean PPD, and acceptable increases in RF after each successful balloon dilation when tested under pulmonary conditions (Table 1). These results show the feasibility of balloon expansion of a polymeric valved conduit while preserving valve competence. The pulse duplicator experiments also highlight the importance of considering valve performance across all stages of growth. As previously noted, there was visible fluttering of the leaflets which appeared to correspond to arterial pressure oscillations during valve opening (Video S1, Figure 5a-e). Flutter is undesirable as it is associated with increased leaflet strains, potentially leading to structural failure and decreased valve durability.[69,73] In the present case, flutter may be explained by the presence of redundant leaflet material associated with the increased coaptation height. Interestingly, the magnitudes of the pressure oscillations were noticeably reduced after each balloon dilation (Figure 5), which suggests that the redundant material became more effectively utilized at the larger diameters and the flutter was possibly reduced. This explanation is also consistent with the observed decreases in coaptation height from the pre-dilation to post-dilation states (Table 1). While previous valves constructed from Elast-Eon™ variants have already demonstrated good thrombogenic properties in a sheep model,[44] we will further assess device thrombogenicity due to valvular dynamics and shear forces in fluid-structure interaction (FSI) simulations. These simulations will also explore methods to reduce leaflet flutter and the flow and pressure oscillations observed in vitro, since such flow disturbances are known to precipitate thrombosis.[74] Approaches to reduce leaflet flutter include increasing the leaflet thickness[69,75] as well as varying the leaflet material and density. Any such changes would need to be balanced with overall valve performance – as seen in Figure 3d, thicker leaflets can lead to increased mean PPD and decreased EOA. We are also investigating new growth-accommodating valve designs that do not require redundant leaflet material and can thus function more optimally and with improved durability at each stage of growth. 3.5 Computational Modeling of the Balloon Dilation The finite element analysis of the balloon dilation predicted a final diameter in very good agreement with the experimental dilations. This result highlights the reliability and consistency of the balloon dilations and supports the eventual translation of balloon dilatable materials to the clinic. Interestingly, the model also predicted an inhomogeneous distribution of permanent deformation in the device after expansion (Figure 7). The greatest deformation occurred along the leaflet attachment areas, which contain stress concentrations that lead to higher strains and greater permanent stretching. The appearance of stress concentrations in the complex valved conduit geometry could explain why both the experimental balloon dilations (λperm = 1.11 ± 0.01) and numerical simulation (λperm = 1.13 ± 0.05) of the valved conduits showed greater stretching than in the uniaxial test strips (λperm = 1.06 ± 0.03), which had a simple rectangular geometry. These modeling results also suggest that large dilations could result in tearing along those high-stress regions, which was observed in Devices #3 and #4 during their second balloon dilations. Another implication of the non-uniform deformation predicted by the model is that the pre-dilation and post-dilation devices would not be geometrically similar, further emphasizing the need to consider valve design for all stages of growth. 3.6 Limitations The mechanical tests and the balloon dilation procedure were conducted in room air at ~25 °C. It is possible that a blood or saline environment at 37 °C would affect the polymers’ mechanical response and the resulting permanent stretch. Also, only the central valve region of the conduit was dilated. The proximal and distal ends remained at 22 mm and were not dilated due to the fixed 22 mm size of the pulse duplicator testing fixture. This may have impaired the hydrodynamic performance of the dilated devices. Additionally, the fatigue life of this device has not been tested as we only intended this design to be a proof of concept for an expandable valved conduit. While rat subcutaneous implantation is an accepted model for initial biocompatibility testing of valve components,[47,76,77] the valved conduit material should ultimately be exposed to circulation and cyclic loading in the valve position of a large animal model to assess not only biocompatibility, but also potential effects on mechanical properties, expandability, and long-term durability. Additionally, the potential for expansion after formation of surgical adhesions around the conduit must be confirmed. Although the impacts of these in vivo interactions are unknown, our clinical experience with balloon dilation and stenting in non-valved vascular locations is encouraging, as we know that some permanent deformation of Gore-Tex® grafts, patched vessels, or previously implanted stents can be achieved at different time points after the initial surgery. 3.7 Outlook Since the completion of this study, we have identified another biostable material capable of greater than 2x permanent stretch, and we are currently developing a 2nd generation prototype. Based on the results presented here, it is expected that the design of this new prototype will benefit from the optimization of hemodynamics across all stages of growth. To achieve this, we are developing computational simulations of valve kinematics and hemodynamics that will enable the study of various growth-accommodating valve geometries with different numbers of leaflets (e.g. bileaflet vs. trileaflet) and different leaflet heights, widths, and free edge lengths. These simulations will take place within a framework that incorporates the nonlinear FSI physics via coupled finite element and computational fluid dynamics analyses, while also accounting for potentially altered material properties post-dilation. The intent is then to manufacture optimized prototypes, demonstrate their in vivo performance and expandability in a large animal model of RVOT replacement in sheep, and with FDA approval, conduct a phase I clinical trial to evaluate their safety for implantation in humans. The concept of tissue engineering represents the ultimate solution for pediatric heart valves, as the intended result is an autologous organ that will grow and adapt to changes in the patient’s physiology.[78,79] The classical methodology for tissue engineering utilizes a biodegradable valve-shaped scaffold that is seeded with cells, matured in vitro in a bioreactor, and then implanted in the patient so that leaflet tissue can grow naturally.[80] However, key challenges of this method include maintaining the balance of scaffold biodegradation with extracellular matrix formation, as well as preventing leaflet shortening arising from the contractile nature of seeded cells.[81,82] Novel regenerative techniques, such as those based on in-body tissue architecture[83] or implantation of decellularized scaffolds,[84] have shown up to 1 year of good valve function in vivo. However, their long-term growth and durability has yet to be demonstrated. In contrast, the concept of growth accommodation via balloon dilation of a polymeric valve is based on technologies with greater acceptance in current clinical practice. Catheter-based interventions already represent the standard of care for treating valvular disease, such as with balloon valvuloplasty for valve stenosis[85,86] and post-implantation balloon dilation of transcatheter valves.[87] Biostable polymeric valves have also been successfully translated to the clinic – ePTFE valves are commonly assembled in the operating room and then implanted with excellent results.[50-53] Polyurethane valves manufactured by robotic dip molding (Foldax, Inc., Salt Lake City, Utah) are currently undergoing clinical trials,[88,89] as is the aforementioned Autus Valve.[33,34] Furthermore, tissue-engineered valves currently require months of preparation, while synthetic polymeric devices can be manufactured and mass-produced more quickly. Hence, as tissue-engineering technology continues to mature, balloon-expandable polymeric valves constitute a more realistic near-term solution for pediatric patients and warrant greater attention. 4 Conclusions In this paper, we have established the first proof of concept of a polymeric valved conduit that can be expanded via transcatheter balloon dilation while maintaining its valvular competence. Four 22 mm diameter valved conduits were balloon dilated to new permanent diameters of 23.26 ± 0.38 mm. Further dilation caused two of the valved conduits to sustain leaflet tears, while the two surviving devices reached final diameters of 24.38 ± 0.19 mm. The expansions in diameter were achieved via permanent deformation of the Elast-Eon™ material, while valve performance at the expanded diameters was maintained using a leaflet design with increased coaptation area. Our results have demonstrated the feasibility of this concept and provide motivation for further development of a polymeric valved conduit that can accommodate the growth of children from a neonate to adult size and reduce the need for multiple open-heart surgeries. 5 Experimental Methods Mechanical Testing: We evaluated the mechanical behavior of two commercially available, biostable and biocompatible polymers: Carbothane™ AC-4075A (Lubrizol, Cleveland, Ohio) and Elast-Eon™ E5-325 (Biomerics, Salt Lake City, Utah). To prepare material samples, pellets of either Carbothane™ (20% w/v) or Elast-Eon™ (40% w/v) were dissolved in N,N-Dimethylacetamide (99.5%, ACROS Organics, Fair Lawn, New Jersey) to create a viscous solution. After the dissipation of bubbles (~24 hours), the polymer solution was cast onto flat plates and then dried in an oven for 1 hour at 80 °C and ambient pressure. The resulting polymer films were cut into individual specimens for testing, and the thicknesses of the specimens were measured using a digital thickness gauge (Mitutoyo 547-526S, Mitutoyo Corporation, Tokyo, Japan). All mechanical tests were performed in air at ambient temperature (~25 °C) using an Instron MicroTester 5848 with a 50 N load cell (Instron, Norwood, Massachusetts), and all measurements were taken from distinct samples. Sample strain was measured using the machine crosshead displacement. To obtain the elongations at break and ultimate strengths, dog-bone samples with a 22 mm gauge length were cut with an American Society for Testing and Materials (ASTM) D1708 cutting die (Ace Steel Rule Dies, Medford, New Jersey). The samples were then uniaxially stretched at a strain rate of 0.0067 s−1 to match the strain rate used for the stretch tests. For the stress relaxation tests, 1x4 cm rectangular samples were individually mounted with a 15 mm gauge length between the Instron machine grips. Each sample was first preconditioned for 5 cycles of stretching to λ = 1.5 and unloading at a strain rate of 0.1 s−1. Then, it was stretched again at the same strain rate and held at a constant stretch of λ = 1.5 for 300 seconds while the load on the sample was monitored. For the stretch tests, 1x4 cm rectangular samples were individually mounted with a 15 mm gauge length between the Instron machine grips and then uniaxially stretched to a single predetermined stretch ratio (λ = 2, 3, 4, or 5) at a strain rate of 0.0067 s−1 (corresponding to a machine crosshead speed of 0.1 mm/s). The samples were then immediately released from this stretch by returning the machine grips to the 15 mm gauge length at the same strain rate, with the sample still held in the grips. The amount of immediately recoverable deformation was denoted by the point of return to zero stress. The samples were then removed from the machine grips and allowed to recover viscoelastically with no external loading. Prior to the start of the test, the gauge length between the grip edges was also marked with a marker. After removal from the testing machine, the amount of strain in each sample was tracked by measuring the distance between these marks. The final permanent stretch was measured 24 hours after the end of each test. In Vivo Biocompatibility Testing: 8 mm disc specimens were subcutaneously implanted[47,48,76] in 4-month-old male Sprague-Dawley rats (n = 3; Charles River Laboratories, Wilmington, Massachusetts) for a period of two months. Each animal received four specimens, with one specimen made from each polymer. Upon explantation, the material specimens were harvested, sectioned, and stained with hematoxylin and eosin (H&E) and Alizarin Red. The experimental animal protocol was approved by the Columbia University Institutional Animal Care and Use Committee (IACUC #AC-AABD5614). Animals received humane care in accordance with the “Guide for the Care and Use of Laboratory Animals” (National Research Council, Eight Edition, 2011). These animal experiments did not use a method of randomization, and the investigators were not blinded to allocation during data collection and analysis. Animals were housed in the Black Building on the campus of the Columbia University Irving Medical Center and treated under the supervision of the Columbia University Institute of Comparative Medicine (ICM). The rats were not subjected to water or food restrictions. ICM animal care staff conducted routine husbandry procedures (cage cleaning, feeding and watering). Full time ICM veterinarian staff monitored the rats at least twice a day to assess their condition. The veterinary staff was available at all times and assisted with surgical procedures, injections and sample harvesting. Rats were initially anesthetized with isofluorane inhalation (4%) via an induction chamber and then maintained with isoflurane at 1.0-1.5% via nose cone during the polymer patch implantation. After implantation, the rats were allowed to recover on a warm pad at 37 °C and returned to the cage once awake. Potential pain was assessed every 2 days and relieved by Meloxicam administration. If any of the following sign/symptoms were noted analgesic was administered: decreased activity, hunched posture, lack of grooming, abnormal gait. Lack of appetite or water consumption was noted. Rats were euthanized by isoflurane overdose, a method consistent with the recommendations of the Panel on Euthanasia of the American Veterinary Medical Association. Death was verified by cervical dislocation since the IACUC approves this method as quick and painless on small animals. Fabrication by Dip Molding: The device prototypes were formed by dip molding. First, the geometry of the valved conduit was modeled in the computer-aided design (CAD) program SolidWorks (Dassault Systèmes, Waltham, Massachusetts) and used to design a two-piece mold, which was then machined in aluminum by 5-axis computer numerical control (CNC) milling (Protolabs, Maple Plain, Minnesota). To form the leaflets, the positive end of the mold was manually dipped into a liquid solution of Elast-Eon™ (40% w/v in N,N-Dimethylacetamide) and then dried in an oven at 80 °C for >12 hours, evaporating the solvent and leaving a conformal polymer coating. Next, the negative end of the mold was fitted over the positive end with the first coating still intact, and the fully assembled mold was dipped and then dried to form the conduit. Conduit thickness was increased using additional rounds of dipping and drying – 4 rounds total for all devices. After the final round, the polymer-coated mold was removed from the oven, cooled to room temperature, and then soaked in water for 15 minutes to loosen the polymer from the mold. Finally, the polymer was carefully peeled from the mold, and the three leaflets were separated using a sharp blade. Conduit wall thickness was measured pre-dilation using a digital thickness gauge (Mitutoyo 547-526S, Mitutoyo Corporation, Tokyo, Japan) at locations proximal to the valve, distal to the valve, and in the middle of the conduit just slightly distal to the valve. At each of the three locations, three measurements were taken at equidistant sites along the circumference of the conduit. After the completion of in vitro device evaluation involving dilation as detailed below, the leaflets were excised from the conduit, and their thicknesses were mapped. The thickness of each leaflet was measured at four different sites spanning the top, middle, and bottom of the leaflet. In Vitro Evaluation: Permanent dilation of the valved conduits was performed using a 46 mm diameter Coda balloon catheter (Cook Medical, Bloomington, Indiana). The balloon was filled using a syringe to the target diameters. The balloon was inflated to a 40 mm diameter, using a 40 mm inner diameter 3D-printed ring as an indicator, and then immediately released with no holding time. The new permanent diameters of the valved conduits were measured 24 hours after the balloon dilations. The second round of balloon dilations was performed in a similar manner, but to a diameter of 44 mm. The in vitro hydrodynamic function of the prototypes was evaluated before and after the balloon dilations using a commercial heart valve pulse duplicator (HDTi 6000, BDC Laboratories, Wheat Ridge, Colorado) equipped with a flow meter (Transonic Systems, Ithaca, New York) and upstream and downstream pressure transducers (BDC Laboratories). The devices were tested using pulmonary conditions at 15 mmHg mean arterial pressure (70 bpm heart rate, 70 mL stroke volume, systole comprising 35% of the cardiac cycle, and a working fluid of 1% w/v saline solution). Regurgitant fraction (RF), mean positive pressure differential (PPD), and effective orifice area (EOA) were calculated using Statys™ software (BDC Laboratories) in accordance with ISO 5840-1:2021, and the results were averaged over ten consecutive cardiac cycles. The mean PPD, where pressure differential (PD) is defined by (2) PD=ventricularpressure−arterialpressure, was calculated over the time period ranging from the start of the positive pressure differential (ventricularpressure>arterialpressure) to the end of the positive pressure differential. The effective orifice area was calculated as (3) EOA=qvRMS51.6×Δpρ, where qvRMS is the root mean square forward flow (ml/s) during the positive differential pressure period, Δp is the mean PPD (mmHg) during the same period, and ρ is the density of the test fluid (g·cm−3). The regurgitant fraction was calculated as (4) RF=Closingvolume+LeakagevolumeForwardflowvolume. We measured the coaptation heights of each valved conduit before and after the balloon dilations. The prototypes were first mounted vertically in a test fixture (BDC Laboratories) with the distal end facing upwards and a 25.4 mm inner diameter clear polycarbonate tube affixed to the top of the test fixture. Then, tap water at ~25 °C was poured into the distal end of the tube to generate pressure heads corresponding to pressures of 3 mmHg, which was just sufficient to close the valves, and 25 mmHg, which is the typical peak diastolic pressure in the pulmonary artery. The pressure was monitored using a pressure transducer (BDC Laboratories) and Statys™ software (BDC Laboratories). Images of the valves in the closed states were recorded with a digital camera, and measurements were taken using ImageJ software (National Institutes of Health, Bethesda, Maryland). The recorded coaptation height was averaged from three measurements. Computational Modeling of the Balloon Dilation: The expansion of the valved conduit was simulated using a mechanical finite element analysis (FEA) model (Abaqus, Dassault Systèmes) with a calibrated Elast-Eon™ material model. The FEA model was constructed from the geometry of the valve using shell elements. A uniform thickness of 100 μm was assigned throughout the structure to match the measured thicknesses of the leaflets. The balloon was modeled as a rigid cylindrical structure using membrane elements that were rigidly constrained, and the dilation was simulated via a uniform radial expansion. The nonlinear elastic material response in the valved conduit was modeled using an Ogden-type hyperelastic model.[90] The permanent, inelastic deformation was approximated by a J2-plasticity model,[91,92] while Mullins effect was captured using a damage model.[93,94] Material calibration and construction of the FEA model are further described in the Supporting Information section. Statistical Analyses: Results were analyzed using OriginPro 2016 (OriginLab, Northampton, Massachusetts). Mechanical and hydrodynamic data and device thicknesses are expressed as mean ± standard deviation. Device diameters and coaptation heights are expressed as the measurement ± estimated measurement error. Mechanical data and thickness measurements were analyzed for statistical significance by the unpaired Student’s t-test (two-sided), with p < 0.05 considered statistically significant. Supplementary Material Figure S1 Figure S2 Figure S3 Video S1 Supplementary Material Table S1 Table S2 Table S3 Acknowledgements The authors would like to thank Dr. Yingfei Xue, Dr. Antonio Frasca, and Kenneth Cai for helpful discussions and for their assistance with in vitro and in vivo testing. The authors would also like to thank Andrew Weiss and Ernesto Cabello of Piper Plastics Corp. for their technical guidance on dip molding. This work was supported by the National Institutes of Health, Bethesda, Maryland [R01-HL-155381] (DK), [R01-HL-143008] (GF), [T32-HL007854] (APK); the Thoracic Surgery Foundation Research Award, Chicago, Illinois (DK); the Kibel Fund for Aortic Valve Research, Philadelphia, Pennsylvania (GF); the Congenital Heart Defect Coalition, Butler, New Jersey (DK); the Babies Heart Fund for Research in Congenital Heart Disease, New York, New York (DK); the Columbia University Provost Seed Grant, New York, New York (DK); the Columbia University Alliance Joint Projects Grant, New York, New York (DK); the Columbia Irving Scholar Award, New York, New York (DK); the Post-9/11 G.I. Bill (RLL); the Columbia University Yellow Ribbon Program, New York, New York (RLL); and the National Science Foundation Graduate Research Fellowship under Grant No. DGE-2036197 (ARH). Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request. Figure 1. Mechanical characterization and in vivo biocompatibility testing of Elast-EonTM and Carbothane™ (a) The averaged (mean) responses from stress relaxation tests of Elast-Eon™ and Carbothane™ (n = 4 samples per group) show significant dissipation of time-dependent viscous effects within the first 300 seconds of an induced strain. Standard deviation is ± 0.03 MPa for both Elast-Eon™ and Carbothane™. (b-c) Representative stress-stretch curves for distinct samples of (b) Elast-Eon™ and (c) Carbothane™ showing elastomeric mechanical behavior when stretched uniaxially to stretch ratios of λtemp = 2 (black line), λtemp = 3 (red line), λtemp = 4 (green line), and λtemp = 5 (blue line) and then unloaded. The dashed gray line indicates the amount of immediate recovery after stretching to λtemp = 5 and then unloading, and the solid gray line indicates the permanent deformation remaining after 24 hours. (d) Amount of permanent stretch resulting from different temporary stretch ratios. Blue circles and red diamonds represent the mean values, and error bars represent the standard deviation. Elast-Eon™ showed greater permanent stretch than Carbothane™ at λtemp = 2, 3, and 5. *p = 0.008; **p = 5e-5; ***p = 0.65 (n.s.); ****p = 0.0038, unpaired Student’s t-test (n = 4 to 8 samples per group). (e-l) Histological sections stained with hematoxylin and eosin (middle row, e-h) and Alizarin Red (bottom row, i-l) of ePTFE control samples (e, i), non-stretched Carbothane™ (f, j), non-stretched Elast-Eon™ (g, k), and Elast-Eon™ permanently pre-stretched by λperm = 1.1 (h, l) showing no cell penetration or calcification in a rat subcutaneous model with explantation at 2 months. Insets (f-h, j-l) show the locations of the transparent Carbothane™ and Elast-Eon™ samples. Original magnification 5x; scale bars = 500 μm. Figure 2. Design and fabrication of the growth-accommodating valved conduit (a) Design schematic of the growth-accommodating polymeric valved conduit showing a conduit with a trileaflet valve positioned in its center. (b) The coaptation area of the original leaflet design by Mackay et al.[54] is characterized by an original coaptation height c. We modified this leaflet design to have an increased coaptation area to ensure competence at an expanded valve diameter. This increased area is characterized by an increased coaptation height h which forms the side of a right triangle having base length b equal to the initial conduit radius and hypotenuse length l equal to the expanded conduit radius, as well as a new length of the free edge 2l which follows a triangular profile. (c) Fabricated valved conduit in the pre-dilation state (22 mm diameter). Grid lines are in inches. (d1) Two-piece aluminum mold for dip molding fabrication of Elast-Eon™ valved conduit prototypes. The negative end of the mold is pictured at the top, and the positive end is at the bottom. (d2) The two separate pieces of the mold. Figure 3. Thicknesses of pre-dilation conduit walls and post-dilation excised leaflets (a) Conduit wall thickness measurement sites P (proximal), M (middle), and D (distal). (b) Leaflet thickness measurement sites 1-4. (c) Measurements of pre-dilation conduit wall thickness. (d) Measurements of excised, post-dilation leaflet thickness and corresponding mean positive pressure differentials (PPD) from pre-dilation hydrodynamic testing of the valved conduits. For conduit wall and leaflet thicknesses, diamonds represent individual measurements, horizontal bars represent the mean values, and whiskers represent the minimum and maximum values. For mean PPD, circles represent the mean values, and whiskers represent ±1 s.d. *p < 0.05, **p < 0.01, ****p < 0.0001, unpaired Student’s t-test (n = 3 measurements per group). Figure 4. Transcatheter balloon dilation of a valved conduit Four 22 mm diameter devices (a) were temporarily balloon dilated to a diameter of 40 mm using a Coda balloon catheter (b1), after which they recovered to new permanent diameters of 23.26 ± 0.38 mm (c1). The devices were further temporarily dilated to a larger diameter of 44 mm (b2) Two of the four devices tore, and the two surviving devices recovered to new permanent diameters of 24.38 ± 0.19 mm (c2). Figure 5. Effects of the balloon dilations on in vitro hydrodynamic performance of the valved conduits (a-e) Representative pulse duplicator readings from a single cardiac cycle for ventricular pressure (blue), arterial pressure (red), and forward flow (green) are shown as a function of time. Pressure and flow oscillations observed in the pre-dilation valves were reduced after each dilation. (f-o) Closed and open configurations of the valved conduits before and after each dilation showing leaflet coaptation and opening. Scale bars = 10 mm. Figure 6. Effect of balloon dilation on initial increased leaflet coaptation height Images of valved conduits in the pre-dilation state (a,b) and post-dilation state (c,d) schematically illustrating changes in coaptation height after balloon dilation (a→c, b→d) and with increased pressure (a→b, c→d). At a small pressure of 3 mmHg, the increased coaptation height h3 in the pre-dilation valve (a) is greater than the corresponding coaptation height h3′ in the post-dilation valve (c). An increase in pressure from 3 mmHg to 25 mmHg also leads to apparent reductions in coaptation height h25 of a pre-dilation valve (b) and in coaptation height h25′ of a post-dilation valve (d). Scale bars = 10 mm. Figure 7. Permanent deformation in the valved conduit predicted by the simulation of the balloon dilation The device is shown in the final unpressurized, post-dilation configuration. The distribution of permanent deformation is represented by the colored equivalent plastic strain (PEEQ) contours. The colors from blue to red represent increasing amounts of permanent deformation. (a) Angled view showing the valve leaflets and deflated balloon within the conduit. (b) Side view of the conduit exterior. The distribution of permanent deformation was non-uniform throughout the valve region, with the magnitude of deformation being greatest near the commissures. Scale factor of deformations = 1. Table 1. In vitro test data. Hydrodynamic data was recorded over 10 consecutive cardiac cycles (mean ± s.d.). Provided are the maximum RF permitted and minimum EOA required for prosthetic valves in the aortic position per ISO 5840-2:2021, although there are no performance requirements for the pulmonic position. Mean arterial pressure [mmHg] Regurgitant fraction [%] Mean positive pressure differential [mmHg] Effective orifice area [cm2] Coaptation height (@3mmHg) [mm] Coaptation height (@25mmHg) [mm] Final diameter [mm] Device #1 Pre-dilation 15.0 ± 0.1 2.3 ± 0.2 15.8 ± 1.6 1.14 ± 0.02 h3=2.92 h25=2.17 22.00 Post-dilation* 15.1 ± 0.1 2.5 ± 0.1 14.8 ± 0.0 1.23 ± 0.01 h3‘=1.81 h25‘=1.68 23.00 2nd Post-dilation** 15.1 ± 0.1 4.5 ± 0.5 9.7 ± 0.2 1.66 ± 0.01 24.56 Device #2 Pre-dilation 15.1 ± 0.1 1.7 ± 0.2 12.0 ± 0.2 1.46 ± 0.01 h3=2.07 h25=1.97 22.00 Post-dilation* 15.0 ± 0.1 2.5 ± 0.1 11.0 ± 0.1 1.51 ± 0.01 h3‘=1.49 h25‘=1.38 22.77 2nd Post-dilation** 14.9 ± 0.2 4.8 ± 0.4 9.2 ± 0.1 1.64 ± 0.01 24.19 Device #3 Pre-dilation 15.0 ± 0.1 2.0 ± 0.2 9.9 ± 0.4 1.41 ± 0.01 h3=2.70 h25=2.21 22.00 Post-dilation* 14.9 ± 0.1 2.8 ± 0.3 9.4 ± 0.0 1.45 ± 0.01 h3‘=2.49 h25‘=2.06 23.54 Device #4 Pre-dilation 15.1 ± 0.1 3.0 ± 0.1 10.6 ± 0.1 1.63 ± 0.01 h3=2.85 h25=2.31 22.00 Post-dilation* 15.1 ± 0.1 3.9 ± 0.2 8.8 ± 0.2 1.66 ± 0.01 h3‘=2.65 h25‘=2.19 23.71 ISO 5840-2 standards 22 mm valve - ≤ 10 - ≥ 1.15 - - - 23 mm valve - ≤ 10 - ≥ 1.25 - - - 24 mm valve - ≤ 10 - ≥ 1.35 - - - * Post-dilation: Devices were dilated to 40 mm. ** 2nd Post-dilation: Devices were dilated to 44 mm. Supporting Information Supporting Information is available from the Wiley Online Library or from the authors. Conflict of Interest Statement The authors declare no conflicts of interest or relationships with industry. References [1] van der Linde D , Konings EEM , Slager MA , Witsenburg M , Helbing WA , Takkenberg JJM , Roos-Hesselink JW , J. Am. Coll. Cardiol 2011, 58 , 2241.22078432 [2] Roger VL , Go AS , Lloyd-Jones DM , Adams RJ , Berry JD , Brown TM , Carnethon MR , Dai S , de Simone G , Ford ES , Fox CS , Fullerton HJ , Gillespie C , Greenlund KJ , Hailpern SM , Heit JA , Ho PM , Howard VJ , Kissela BM , Kittner SJ , Lackland DT , Lichtman JH , Lisabeth LD , Makuc DM , Marcus GM , Marelli A , Matchar DB , McDermott MM , Meigs JB , Moy CS , Mozaffarian D , Mussolino ME , Nichol G , Paynter NP , Rosamond WD , Sorlie PD , Stafford RS , Turan TN , Turner MB , Wong ND , Wylie-Rosett J , Circulation 2011, 123 , 18. [3] Hoffman JIE , Kaplan S , J. Am. Coll. Cardiol 2002, 39 , 1890.12084585 [4] Reller MD , Strickland MJ , Riehle-Colarusso T , Mahle WT , Correa A , J. Pediatr 2008, 153 , 807.18657826 [5] Kaza AK , Lim HG , Dibardino DJ , Bautista-Hernandez V , Robinson J , Allan C , Laussen P , Fynn-Thompson F , Bacha E , del Nido PJ , Mayer JE , Pigula FA , J. Thorac. Cardiovasc. Surg 2009, 138 , 911.19660342 [6] Karamlou T , Jang K , Williams WG , Caldarone CA , Van Arsdell G , Coles JG , McCrindle BW , Circulation 2005, 112 , 3462.16316968 [7] Dvir D , Bourguignon T , Otto CM , Hahn RT , Rosenhek R , Webb JG , Treede H , Sarano ME , Feldman T , Wijeysundera HC , Topilsky Y , Aupart M , Reardon MJ , Mackensen GB , Szeto WY , Kornowski R , Gammie JS , Yoganathan AP , Arbel Y , Borger MA , Simonato M , Reisman M , Makkar RR , Abizaid A , McCabe JM , Dahle G , Aldea GS , Leipsic J , Pibarot P , Moat NE , Mack MJ , Kappetein AP , Leon MB , Circulation 2018, 137 , 388.29358344 [8] David TE , Armstrong S , Maganti M , Ann. Thorac. Surg 2010, 90 , 775.20732495 [9] Bourguignon T , Bouquiaux-Stablo AL , Candolfi P , Mirza A , Loardi C , May MA , El-Khoury R , Marchand M , Aupart M , Ann. Thorac. Surg 2015, 99 , 831.25583467 [10] Neves JP , Gulbenkian S , Ramos T , Martins AP , Caldas MC , Mascarenhas R , Guerreiroa M , Matoso-Ferreira A , Santos R , Monteiro C , Melo JQ , J. Thorac. Cardiovasc. Surg 1997, 113 , 1014.9202681 [11] Kanter KR , Kirshbom PM , Kogon BE , Ann. Thorac. Surg 2006, 82 , 1594.17062211 [12] Iung B , Rodes-Cabau J , Eur. Heart J 2014, 35 , 2942.25205532 [13] Nomoto R , Sleeper LA , Borisuk MJ , Bergerson L , Pigula FA , Emani S , Fynn-Thompson F , Mayer JE , del Nido PJ , Baird CW , J. Thorac. Cardiovasc. Surg 2016, 152 , 1333.27637422 [14] Karamlou T , Blackstone EH , Hawkins JA , Jacobs ML , Kanter KR , Brown JW , Mavroudis C , Caldarone CA , Williams WG , McCrindle BW , J. Thorac. Cardiovasc. Surg 2006, 132 , 829.17000294 [15] da Costa FDA , Etnel JRG , Charitos EI , Sievers HH , Stierle U , Fornazari D , Takkenberg JJM , Bogers AJJC , Mokhles MM , Ann. Thorac. Surg 2018, 105 , 1205.29307455 [16] Belli E , Salihoǧlu E , Leobon B , Roubertie F , Ly M , Roussin R , Serraf A , Ann. Thorac. Surg 2010, 89 , 152.20103226 [17] Tweddell JS , Pelech AN , Frommelt PC , Mussatto KA , Wyman JD , Fedderly RT , Berger S , Frommelt MA , Lewis DA , Friedberg DZ , Thomas JP , Sachdeva R , Litwin SB , Circulation 2000, 102 , DOI 10.1161/circ.102.suppl_3.III-130. [18] Chen PC , Sager MS , Zurakowski D , Pigula FA , Baird CW , Mayer JE , del Nido PJ , Emani SM , J. Thorac. Cardiovasc. Surg 2012, 143 , 352.22153723 [19] Saleeb SF , Newburger JW , Geva T , Baird CW , Gauvreau K , Padera RF , del Nido PJ , Borisuk MJ , Sanders SP , Mayer JE , Circulation 2014, 130 , 51.24756063 [20] Fuller SM , Borisuk MJ , Sleeper LA , Bacha E , Burchill L , Guleserian K , Ilbawi M , Razzouk A , Shinkawa T , Lu M , Baird CW , Semin. Thorac. Cardiovasc. Surg 2021, DOI 10.1053/j.semtcvs.2021.06.020. [21] Yamamoto Y , Yamagishi M , Miyazaki T , Gen. Thorac. Cardiovasc. Surg 2015, 63 , 131.25503561 [22] Kostyunin AE , Yuzhalin AE , Rezvova MA , Ovcharenko EA , Glushkova TV , Kutikhin AG , J. Am. Heart Assoc 2020, 9 , 1. [23] Christian AJ , Lin H , Alferiev IS , Connolly JM , Ferrari G , Hazen SL , Ischiropoulos H , Levy RJ , Biomaterials 2014, 35 , 2097.24360721 [24] Lee S , Levy RJ , Christian AJ , Hazen SL , Frick NE , Lai EK , Grau JB , Bavaria JE , Ferrari G , J. Am. Heart Assoc 2017, 6 , 1. [25] Gomel MA , Lee R , Grande-Allen KJ , Front. Cardiovasc. Med 2019, 5 , 1. [26] Liao KK , Li X , John R , Amatya DM , Joyce LD , Park SJ , Bianco R , Bolman RM , Ann. Thorac. Surg 2008, 86 , 491.18640322 [27] Flameng W , Herregods M-C , Vercalsteren M , Herijgers P , Bogaerts K , Meuris B , Circulation 2010, 121 , 2123.20439787 [28] Emani SM , Piekarski BL , Zurakowski D , Baird CA , Marshall AC , Lock JE , del Nido PJ , J. Thorac. Cardiovasc. Surg 2016, 152 , 1514.27692768 [29] Feins EN , Chávez M , Callahan R , del Nido PJ , Emani SM , Baird CW , Ann. Thorac. Surg 2020, DOI 10.1016/j.athoracsur.2020.05.061. [30] Pluchinotta FR , Piekarski BL , Milani V , Kretschmar O , Burch PT , Hakami L , Meyer DB , Jacques F , Ghez O , Trezzi M , Carotti A , Qureshi SA , Michel-Behnke I , Hammel JM , Chai P , McMullan D , Mettler B , Ferrer Q , Carminati M , Emani SM , Circ. Cardiovasc. Interv 2018, 11 , 1. [31] Sullivan PM , Wong PC , Kim R , Ing FF , Cardiol. Young 2019, 29 , 235.30511601 [32] Langer NB , Solowiejczyk D , Fahey JT , Torres A , Bacha E , Kalfa D , J. Thorac. Cardiovasc. Surg 2018, 156 , 1190.29709362 [33] Hofferberth SC , Saeed MY , Tomholt L , Fernandes MC , Payne CJ , Price K , Marx GR , Esch JJ , Brown DW , Brown J , Hammer PE , Bianco RW , Weaver JC , Edelman ER , del Nido PJ , Sci. Transl. Med 2020, 12 , eaay4006.32075944 [34] “Autus Size-Adjustable Valve for Surgical Pulmonary Valve Replacement Early Feasibility Study,” can be found under https://clinicaltrials.gov/ct2/show/NCT05006404, n.d. [35] Feins EN , Emani SM , Semin. Thorac. Cardiovasc. Surg. Pediatr. Card. Surg. Annu 2020, 23 , 17.32354541 [36] Manavitehrani I , Ebrahimi P , Yang I , Daly S , Schindeler A , Saxena A , Little DG , Fletcher DF , Dehghani F , Winlaw DS , Cardiovasc. Eng. Technol 2019, DOI 10.1007/s13239-019-00406-5. [37] Stradins P , Lacis R , Ozolanta I , Purina B , Ose V , Feldmane L , Kasyanov V , Eur. J. Cardio-thoracic Surg 2004, 26 , 634. [38] Li RL , Russ J , Paschalides C , Ferrari G , Waisman H , Kysar JW , Kalfa D , Biomaterials 2019, 225 , 119493.31569017 [39] Bélanger MC , Marois Y , Roy R , Mehri Y , Wagner E , Zhang Z , King MW , Yang M , Hahn C , Guidoin R , Artif. Organs 2000, 24 , 879.11119076 [40] Yang M , Zhang Z , Hahn C , Laroche G , King MW , Guidoin R , J. Biomed. Mater. Res 1999, 48 , 13.10029144 [41] Yang M , Zhang Z , Hahn C , King MW , Guidoin R , J. Biomed. Mater. Res 1999, 48 , 648.10490678 [42] Bernacca GM , Straub I , Wheatley DJ , J. Biomed. Mater. Res 2002, 61 , 138.12001256 [43] Simmons A , Hyvarinen J , Odell RA , Martin DJ , Gunatillake PA , Noble KR , Poole-Warren LA , Biomaterials 2004, 25 , 4887.15109849 [44] Wheatley DJ , Bernacca GM , Tolland MM , O’Connor B , Fisher J , Williams DF , Int. J. Artif. Organs 2001, 24 , 95.11256515 [45] Bezuidenhout D , Williams DF , Zilla P , Biomaterials 2015, 36 , 6.25443788 [46] Levy RJ , Schoen FJ , Levy JT , Nelson AC , Howard SL , Oshry LJ , Am. J. Pathol 1983, 113 , 143.6605687 [47] Bonetti A , Marchini M , Ortolani F , J. Thorac. Dis 2019, 11 , 2126.31285908 [48] Frasca A , Xue Y , Kossar AP , Keeney S , Rock C , Zakharchenko A , Streeter M , Gorman RC , Grau JB , George I , Bavaria JE , Krieger A , Spiegel DA , Levy RJ , Ferrari G , JACC Basic to Transl. Sci 2020, 5 , 755. [49] Christian AJ , Alferiev IS , Connolly JM , Ischiropoulos H , Levy RJ , J. Biomed. Mater. Res. - Part A 2015, 103 , 2441. [50] Quintessenza JA , Jacobs JP , Chai PJ , Morell VO , Lindberg H , World J. Pediatr. Congenit. Hear. Surg 2010, 1 , 20. [51] Ando M , Takahashi Y , J. Thorac. Cardiovasc. Surg 2009, 137 , 124.19154914 [52] Miyazaki T , Yamagishi M , Maeda Y , Taniguchi S , Fujita S , Hongu H , Yaku H , J. Thorac. Cardiovasc. Surg 2018, 155 , 2567.29510932 [53] Yamashita E , Yamagishi M , Miyazaki T , Maeda Y , Yamamoto Y , Kato N , Asada S , Hongu H , Yaku H , Ann. Thorac. Surg 2016, 102 , 1336.27173066 [54] Mackay TG , Wheatley DJ , Bernacca GM , Fisher AC , Hindle CS , Biomaterials 1996, 17 , 1857.8889065 [55] International Organization for Standardization, Cardiovascular Implants - Cardiac Valve Prostheses - Part 2: Surgically Implanted Heart Valve Substitutes (ISO 5840-2:2021), 2021. [56] Ozaki S , Kawase I , Yamashita H , Uchida S , Nozawa Y , Matsuyama T , Takatoh M , Hagiwara S , Interact. Cardiovasc. Thorac. Surg 2011, 12 , 550.21273254 [57] Ozaki S , Kawase I , Yamashita H , Uchida S , Takatoh M , Kiyohara N , J. Thorac. Cardiovasc. Surg 2018, 155 , 2379.29567131 [58] Bernacca GM , O’Connor B , Williams DF , Wheatley DJ , Biomaterials 2002, 23 , 45.11762853 [59] Kidane AG , Burriesci G , Edirisinghe M , Ghanbari H , Acta Biomater. 2009, 5 , 2409.19497802 [60] Sotiropoulos F , Le TB , Gilmanov A , Annu. Rev. Fluid Mech 2016, 48 , 259. [61] Daebritz SH , Fausten B , Hermanns B , Franke A , Schroeder J , Groetzner J , Autschbach R , Messmer BJ , Sachweh JS , Heart Surg. Forum 2004, 7 , 371. [62] Shinkawa T , Tang X , Gossett JM , Mustafa T , Hategekimana F , Watanabe F , Miyazaki T , Yamagishi M , Imamura M , Ann. Thorac. Surg 2015, 100 , 129.26004923 [63] Choi KH , Sung SC , Kim H , Lee HD , Kim G , Ko H , J. Card. Surg 2018, 33 , 36.29314335 [64] Lee C , Lee CH , Kwak JG , Song JY , Shim WS , Choi EY , Lee SY , Kim YM , Eur. J. Cardio-thoracic Surg 2013, 43 , 468. [65] Hammer PE , Roberts EG , Emani SM , del Nido PJ , J. Thorac. Cardiovasc. Surg 2017, 153 , 389.27665220 [66] Bezuidenhout D , Williams DF , Zilla P , Biomaterials 2015, 36 , 6.25443788 [67] Daebritz SH , Circulation 2003, 108 , 134. [68] Stasiak J , Serrani M , Biral E , Taylor J , Zaman A , Jones S , Ness T , de Gaetano F , Costantino ML , Bruno VD , Suleiman M-S , Ascione R , Moggridge G , Biomater. Sci 2020, DOI 10.1039/d0bm00412j. [69] Johnson EL , Wu MCH , Xu F , Wiese NM , Rajanna MR , Herrema AJ , Ganapathysubramanian B , Hughes TJR , Sacks MS , Hsu M-C , Proc. Natl. Acad. Sci 2020, 117 , 19007.32709744 [70] Jansen J , Reul H , J. Med. Eng. Technol 1992, 16 , 27.1640445 [71] Rahmani B , Tzamtzis S , Ghanbari H , Burriesci G , Seifalian AM , J. Biomech 2012, 45 , 1205.22336198 [72] Park KS , Appa H , Visagie C , Bezuidenhout D , Zilla PP , A Prosthetic Heart Valve, 2016, US 2016/0067038. [73] Vennemann B , Rösgen T , Heinisch PP , Obrist D , ASAIO J 2018, 64 , 651.29045279 [74] Scotten LN , Siegel R , J. Heart Valve Dis 2011, 20 , 664.22655497 [75] Lee JH , Scotten LN , Hunt R , Caranasos TG , Vavalle JP , Griffith BE , JTCVS Open 2021, 6 , 60.35211686 [76] Sun M , Elkhodiry M , Shi L , Xue Y , Abyaneh MH , Kossar AP , Giuglaris C , Carter SL , Li RL , Bacha E , Ferrari G , Kysar J , Myers K , Kalfa D , Biomaterials 2022, 288 , 121756.36041938 [77] Xue Y , Kossar AP , Abramov A , Frasca A , Sun M , Zyablitskaya M , Paik D , Kalfa D , Della Barbera M , Thiene G , Kozaki S , Kawashima T , Gorman JH III , Gorman RC , Gillespie MJ , Carreon CK , Sanders SP , Levy RJ , Ferrari G , Cardiovasc. Res 2022, cvac002. [78] Bouten CVC , Smits AIPM , Baaijens FPT , Front. Cardiovasc. Med 2018, 5 , 1.29404341 [79] Boyd R , Parisi F , Kalfa D , Semin. Thorac. Cardiovasc. Surg 2019, 31 , 807.31176798 [80] Langer R , Vacanti J , Science (80-. ). 1993, 260 , 920. [81] Driessen-Mol A , Emmert MY , Dijkman PE , Frese L , Sanders B , Weber B , Cesarovic N , Sidler M , Leenders J , Jenni R , Grünenfelder J , Falk V , Baaijens FPT , Hoerstrup SP , J. Am. Coll. Cardiol 2014, 63 , 1320.24361320 [82] Weber B , Dijkman PE , Scherman J , Sanders B , Emmert MY , Grünenfelder J , Verbeek R , Bracher M , Black M , Franz T , Kortsmit J , Modregger P , Peter S , Stampanoni M , Robert J , Kehl D , van Doeselaar M , Schweiger M , Brokopp CE , Wälchli T , Falk V , Zilla P , Driessen-Mol A , Baaijens FPT , Hoerstrup SP , Biomaterials 2013, 34 , 7269.23810254 [83] Kawashima T , Umeno T , Terazawa T , Wada T , Shuto T , Nishida H , Anai H , Nakayama Y , Miyamoto S , Interact. Cardiovasc. Thorac. Surg 2021, 32 , 969.33543242 [84] Syedain ZH , Haynie B , Johnson SL , Lahti M , Berry J , Carney JP , Li J , Hill RC , Hansen KC , Thrivikraman G , Bianco R , Tranquillo RT , Sci. Transl. Med 2021, 13 , eabb7225.33731437 [85] Otto CM , Mickel MC , Kennedy JW , Alderman EL , Bashore TM , Block PC , Brinker JA , Diver D , Ferguson J , Holmes DR , Circulation 1994, 89 , 642.8313553 [86] Alsawah GA , Hafez MM , Matter M , Abo-Haded HM , Rakha S , Almarsafawy H , Prog. Pediatr. Cardiol 2016, 43 , 127. [87] Hahn RT , Pibarot P , Webb J , Rodes-Cabau J , Herrmann HC , Williams M , Makkar R , Szeto WY , Main ML , Thourani VH , Tuzcu EM , Kapadia S , Akin J , McAndrew T , Xu K , Leon MB , Kodali SK , JACC Cardiovasc. Interv 2014, 7 , 781.25060022 [88] “Early Feasibility Study for the Foldax Tria Aortic Heart Valve,” can be found under https://clinicaltrials.gov/ct2/show/NCT03851068, n.d. [89] “Early Feasibility Study for the Foldax TRIA Mitral Heart Valve Replacement Investigational Device Exemption,” can be found under https://clinicaltrials.gov/ct2/show/NCT04717570, n.d. [90] Ogden RW , Proc. R. Soc. London. Ser. A Math. Phys. Sci 1972, 326 , 565. [91] Simo JC , Comput. Methods Appl. Mech. Eng 1992, 99 , 61. [92] Weber G , Anand L , Comput. Methods Appl. Mech. Eng 1990, 79 , 173. [93] Govindarajan SM , Hurtado J , Mars W , in Proc. 5th Eur. Conf. Const. Model. Rubber, ECCMR 2007, Paris, 2008, pp. 249–254. [94] ABAQUS User’s Manual, Dassault Systèmes Simulia Corp., Providence, RI, 2019.

PMC010xxxxxx/PMC10364060.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8710358 2405 Brain Inj Brain Inj Brain injury 0269-9052 1362-301X 35604956 10364060 10.1080/02699052.2022.2077987 NIHMS1914929 Article Risk Factors for Development of Long-Term Mood and Anxiety Disorder after Pediatric Traumatic Brain Injury: A Population-Based, Birth Cohort Analysis Esterov Dmitry DO Department of Physical Medicine and Rehabilitation, Mayo Clinic, Rochester, Minnesota Witkowski Julie MD Mayo Clinic School of Graduate Medical Education, Mayo Clinic College of Medicine and Science, Rochester, Minnesota Department of Physical Medicine and Rehabilitation, Northwestern Medicine, Wheaton, Illinois McCall Dana M. PhD Mayo Clinic School of Graduate Medical Education, Mayo Clinic College of Medicine and Science, Rochester, Minnesota Gundersen Health System, La Crosse, Wisconsin Wi Chung-Il MD Department of Pediatric and Adolescent Medicine, Mayo Clinic, Rochester, Minnesota Weaver Amy L. MS Division of Clinical Trials and Biostatistics, Mayo Clinic, Rochester, Minnesota Brown Allen W. MD Department of Physical Medicine and Rehabilitation, Mayo Clinic, Rochester, Minnesota Author Contributions Dr. Esterov had full access to all data in the study and takes responsibility for the integrity of the data and accuracy of the data analysis. Concept and design: Dr. Esterov, Dr. Brown, Dr. Wi, Dr. McCall, Ms. Amy Weaver, Dr. Witkowski Acquisition, analysis, or interpretation of data: Dr. Esterov, Dr. Brown, Dr. Wi, Dr. McCall, Ms. Amy Weaver, Dr. Witkowski Drafting of the manuscript: Dr. Esterov, Dr. Brown, Dr. Wi, Dr. McCall, Ms. Amy Weaver, Dr. Witkowski Critical revision of the manuscript for important intellectual content: Dr. Esterov, Dr. Brown, Dr. Wi, Dr. McCall, Ms. Amy Weaver, Dr. Witkowski Statistical analysis: Ms. Amy Weaver Final approval of the version to be published: Dr. Esterov, Dr. Brown, Dr. Wi, Dr. McCall, Ms. Amy Weaver, Dr. Witkowski Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: Dr. Esterov, Dr. Brown, Dr. Wi, Dr. McCall, Ms. Amy Weaver, Dr. Witkowski Reprints: Dmitry Esterov, DO, Department of Physical Medicine and Rehabilitation, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (esterov.dmitry@mayo.edu; Phone: 507-255-3116; Fax: 507-255-1625). 19 7 2023 12 5 2022 23 5 2022 24 7 2023 36 6 722732 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Objectives The objective of this study was to identify characteristics associated with an increased risk of anxiety and mood disorder prior to 25 years of age, in children who sustained a traumatic brain injury (TBI) prior to age 10. Methods This population-based study identified 562 TBI cases from a 1976–1982 birth cohort in Olmsted County, Minnesota. TBI cases were manually confirmed and classified by injury severity. Separate Cox proportional hazards regression models were fit to estimate the association of TBI and secondary non-TBI related characteristics with the risk of a subsequent clinically determined anxiety or mood disorder. Multivariable-adjusted population attributable risk (PAR) estimates were calculated for TBI characteristics. Results Older age at initial TBI and extracranial injury at time of initial TBI were significantly associated with an increased risk of anxiety (adjusted HR [95% CI]: 1.33 [1.16, 1.52] per 1-year increase and 2.41 [1.26, 4.59]), respectively. Older age at initial TBI was significantly associated with an increased risk of a mood disorder (adjusted HR 1.17 [1.08–1.27]). Conclusion In individuals sustaining a TBI prior to age 10, age at injury greater than 5 years old was the largest contributor to development of a mood or anxiety disorder. Brain Injuries Traumatic Psychiatry and Psychology brain diseases Depressive Disorders Anxiety Disorders Bipolar Disorder pmcIntroduction Pediatric traumatic brain injury (TBI) is associated with more than 60,000 hospitalizations, 600,000 Emergency Department visits, and 7000 deaths annually in the United States, thus representing an important public health concern.(1, 2) In addition to chronic cognitive and physical impairment,(3–5) associations have been observed between pediatric TBI and the development of psychiatric sequelae, including mood and anxiety disorders.(6–10) Mood and anxiety disorders in the general pediatric population can persist into adulthood and are increasing in prevalence and associated disability.(11, 12) The prevalence of mood disorders in particular increase with age between age 10 and age 18, while the prevalence of anxiety disorders remains stable across age groups.(11) Psychiatric symptoms associated with pediatric TBI have been found to place a child at risk for additional negative health outcomes, including cognitive impairment, chronic pain, and sleep disorders.(13–16) Although mood and anxiety disorders have been described after pediatric TBI, the actual incidence of psychiatric disorders in children with TBI is uncertain, with estimates ranging as wide as 10% - 100%.(7) Identifying the factors that place a child at increased risk of psychiatric sequalae after a TBI can assist in the development of both screening and treatment interventions. Several risk factors have been identified that contribute to the risk of developing a mood or anxiety disorder up to several years after pediatric TBI. These have included TBI related characteristics such as intraparenchymal brain lesions and older age at injury,(10, 17–19) as well as secondary factors such as a family history of a psychiatric disorder, low socioeconomic status, and prior psychosocial stressors.(18, 20, 21) Although associations have been observed between pediatric TBI and short-term development of mood and anxiety disorders, there is limited understanding of the risk factors for developing these disorders into young adulthood.(22) Evidence is particularly limited for individuals who sustain TBI at less than 10 years, which accounts for a high proportion of all childhood TBI, especially for those sustaining TBI between ages 0–4.(2, 6) In particular, the Centers for Disease control report that children ages 0–4 have the highest rate of TBI-related ED visits (2194 per 100,000), greater than two times the rate of those in the next highest age group (887 per 100,000) in those aged 15–24.(2) Research on anxiety disorders after childhood TBI is especially sparse, limited by small sample sizes and short follow up, thus the incidence of anxiety disorders after childhood TBI is not known, particularly in the long term.(21) The extent to which the TBI related characteristics in comparison to secondary non-TBI characteristics – such as demographic, socioeconomic, comorbid conditions, and other psychosocial factors – contributes to the risk of clinically determined psychiatric disorder by adulthood after early childhood TBI remains unclear.(6, 7) Understanding these associations over the long term is important because there is evidence that the neurobehavioral effects of pediatric TBI may not be apparent early after injury and the initial effects of early brain injury may worsen over time as social and cognitive demands increase.(23, 24) A population-based birth cohort has the potential to provide unique insights into childhood and adolescent mental health outcomes.(25) Longitudinal studies allow for the opportunity to assess whether psychiatric changes after pediatric TBI resolve over time or continue to persist.(10) This study utilized a longitudinal birth cohort to identify the risk factors associated with a long-term clinically determined mood and anxiety disorder in children who sustained a TBI prior to age 10. We further determined the attributable risk of TBI related characteristics – including age at injury, TBI severity, hospitalization for TBI, skull fracture, and extracranial injury – to the risk of developing a mood and anxiety disorder by age 25, after adjusting for known risk factors for development of mood and anxiety disorders in the general population. Materials and Methods Study Population This study was approved by the Olmsted Medical Center and Mayo Clinic Institutional Review Boards. The population for this cohort study was assembled using birth certificate data available for Olmsted County, Minnesota, obtained from the Minnesota Department of Health and using the resources of the Rochester Epidemiology Project (REP). The REP is a medical record linkage system that provides longitudinal data from all outpatient and inpatient medical professionals in the community for all individuals who resided in Olmsted County, Minnesota, from 1966 to the present.(26, 27) Diagnoses assigned at each visit are coded and maintained in continuously updated files that are automatically indexed for retrieval as part of the REP infrastructure.(27) This study utilized a birth cohort of 8,548 individuals born from January 1, 1976 through December 31, 1982 to mothers who resided in the townships comprising Minnesota Independent School District 535 in Olmsted County, Minnesota at the time of the child’s delivery.(28, 29) The birth cohort was initially developed to study the incidence of learning disabilities (LD) and Attention Deficit Hyperactivity Disorder (ADHD) and was therefore restricted to the subset of 5718 children who were still in the community after 5 years of age.(28, 30) Individuals with severe intellectual disability (n=19) or who declined research authorization in accordance with Minnesota law (n=181) were subsequently excluded, yielding a cohort of 5,518 individuals. Identification of TBI Cases The methods for identifying TBI cases and categorizing each case by injury severity have been described in detail elsewhere.(31, 32) The REP records of the 5,518 cohort members were reviewed for any Hospital Adaptation of the International Classification of Diseases, Eighth Revision (H-ICDA) or International Classification of Diseases, Ninth Revision (ICD-9) diagnosis code suggestive of TBI (Supplement 1). Medical records that contained a code suggestive of TBI prior to age 10 were screened for the following: 1) loss of consciousness, nausea, vomiting, or headache association with the injury, 2) hospital admission secondary to the injury, 3) hospitalization for rehabilitation, or 4) Computed Tomography or X-ray performed within 2 weeks of the injury. Individuals that screened positive were then fully abstracted by manual record review for case confirmation, demographics, maternal education level, medical history, and mechanism of injury. TBI was defined as a traumatically induced injury that contributed to physiological disruption of brain function, and confirmed events were those that noted documentation of concussion with loss of consciousness, traumatic amnesia, neurological signs of brain injury such as seizures, evidence of intracranial injury, skull fracture, and/or postconcussive symptoms. The Mayo Classification System for TBI Severity was then used to categorize TBI into the following categories: Definite (consistent with moderate-severe TBI), Probable (consistent with mild TBI), or Possible (consistent with concussive TBI) (Supplement 2). This classification system is valid in all age groups and has been shown to classify TBI events more accurately than single indicator systems (e.g., length of posttraumatic amnesia, initial Glasgow Coma Score, length of loss of consciousness).(31, 33) The Mayo Classification System stratifies each TBI case by injury severity based on the strength of available information in the medical record and is of particular value particularly in retrospective studies where data can be limited by variably missing indicators of injury severity.(31) The following characteristics at the time in the initial TBI were recorded: age at injury, TBI severity category, hospitalization for TBI, skull fracture at time of TBI, mechanism of injury, and extracranial injury at time of TBI. The date of the case’s TBI diagnosis was defined as the “index date.” Patient characteristics Demographic and socioeconomic characteristics obtained from the child’s birth certificate included sex and race of the child, maternal and paternal age, as well as maternal and paternal education. In addition, socioeconomic status at birth was measured by the HOUSES index, which is an individual measure of socioeconomic status derived from four items of real property data (estimated building value of unit, square footage, number of bedrooms, number of bathrooms) from the County Assessor’s office.(34) The HOUSES index has been previously validated as a marker of socioeconomic status.(34–36) The HOUSES Index was determined using data from 1985 real property data (the oldest historical data available) for the address where the child lived at birth, but only for addresses with a street address (e.g., PO Box address could not be matched to HOUSES). Each property item corresponding to individual’s address was standardized into a z-score and were aggregated into an overall z-score for the four items such that a higher HOUSES score indicated higher SES. This standardized z-score of HOUSES index was then converted to quartiles within the county population, not within the study cohort. Therefore, the HOUSES quartiles of this cohort are not equally distributed among Q1 though Q4. Birth characteristics included gestational age, birth weight, and size for gestational age. Size for gestational age during this era was defined using the 10th and 90th percentiles previously reported for each gestational age of 30,772 deliveries during 1962 to 1969 in Cleveland, Ohio of single live births as reported by Brenner et al.(37) Gestational age was confirmed by manual review of each medical record and was classified into 7 categories: extremely preterm (23–27 weeks), very preterm (28–31 weeks), moderate or late preterm (32–36 weeks), early term (37–38 weeks), full term (39–40) weeks, late term (41 weeks), and Postterm (42–43 weeks).(38) Family history characteristics included maternal depression prior to initial TBI. Maternal depression was classified by manual record review using the same criteria of identifying mood and anxiety disorders of each child with TBI (described below). Lastly, time dependent covariates included recurrent TBI prior to age 10, along with diagnosis of ADHD and history of LD. The identification of LD among the birth cohort members has been previously described.(28, 29, 39) Through a contractual agreement with the school district, the results of academic achievement tests and cognitive assessments administered to members of the birth cohort during their years in the local school district were accessed. LD in mathematics, reading, or written language was determined based on whether each individual met criteria using either of two regression-based formulas or one non-regression-based discrepancy formula using cognitive and achievement scores.(29, 40) Identification of Mood and Anxiety Disorders The methods for identifying and classifying mood and anxiety disorders were based on methodology from a prior birth cohort study.(41) Depression, anxiety, bipolar disorder, suicidal ideation, and self-injurious behavior HICDA and ICD-9 diagnosis codes that were assigned by a REP-affiliated practitioner up until the individual’s 25th birthday were electronically obtained for all TBI cases. A cut off of 25 years at time of diagnosis was chosen as this age generally corresponds to a more inclusive definition of adolescence, spanning between 10 and 24 years of age.(42) Diagnosis codes were classified into either a depressive, anxiety, or bipolar diagnosis (Supplemental 3). Cases were classified by the study statistician (A.L.W.) as having a depressive, anxiety, or bipolar diagnosis if an individual had an associated diagnosis on 2 or more visit dates greater than 30 days apart. Individuals could be classified as having more than 1 comorbid clinical diagnosis. Based on criteria determined by Kirsch et al.,(41) a subset of these medical records were then manually reviewed (D.E.) if any of the following criteria were met: 1) depression, anxiety, or bipolar disorder diagnostic code first assigned before 5 years of age; 2) only 2 visit dates greater than 30 days apart from the same diagnostic category; 3) less than 4 visit dates with bipolar – related codes; 4) classification for depression or anxiety met but individual had a single visit date for another diagnostic category (i.e., depression, anxiety, or bipolar); 5) did not meet classification criteria for either diagnostic category but had at least one suicide related code; or 6) depression/anxiety related codes only included 2 adjustment disorder codes. Of these medical records, the reviewer (D.E.) then classified an individual as having a depressive, anxiety, or bipolar related diagnosis if any of the following criteria were met prior to age 25: 1) diagnosis of a depressive, anxiety, or bipolar disorder confirmed by a psychiatrist/psychologist; 2) pharmacotherapy prescribed for associated psychiatric disorder; or 3) evidence of patient participating in counseling services/psychotherapy for the associated disorder.(43) We assessed the reliability of medical record abstraction of depressive, anxiety, and bipolar diagnoses by having a Pediatric Neuropsychologist (D.M.M.) review the complete medical record for a random sample of 30 patients. This abstractor did not have access to the results of the initial abstraction. The agreement on presence or absence of a depressive, anxiety, or bipolar related disorder was 87% (26/30; 13 both agreed on present and 13 both agreed on absent), with a kappa value of 73% (95% CI: 49 – 98), indicating adequate agreement. Statistical Analysis Data management and statistical analyses were performed using SAS version 9.4 software and RStudio version 1.3.1093.1. Because of the small number of individuals diagnosed with bipolar disorder, depression and bipolar disorder diagnoses were combined into a single category of “mood disorder”. Separate parallel analyses were performed for each psychiatric disorder of interest: a) anxiety and b) mood disorder. For each analysis, duration of follow-up was calculated from the index date (i.e. date of first TBI) until either diagnosis of the psychiatric disorder or last clinical visit date prior to their 25th birthday, whichever came first. Individuals diagnosed with the psychiatric disorder prior to their initial TBI were excluded from the analysis. Univariate Cox proportional hazards regression models were fit to evaluate the association of patient characteristics with the risk of the psychiatric disorder and the strength of the association was summarized using the hazard ratio (HR) and corresponding 95% confidence interval derived from each model. For each continuously scaled characteristic (maternal/paternal age at birth, gestational age, birth weight, and age at TBI), a penalized smoothing spline was used to model a potentially nonlinear relationship with the risk of the psychiatric disorder, before proceeding to assuming a linear relationship. Patient characteristics that were not static at the time of the TBI (e.g. diagnosis of ADHD, LD, and repeat TBI prior to age 10) were evaluated as time-dependent covariates. Five separate multivariate Cox models were fit to evaluate the association of each TBI characteristic (age at injury, injury severity, hospitalization, skull fracture, extracranial injury) with the risk of the psychiatric disorder after adjusting for the following set of potential confounders that were determined a priori from the literature: sex, size for gestational age, maternal education, and history of maternal depression at the time of the individual’s birth.(44–46) All calculated p-values were two-sided and p-values less than 0.05 were considered statistically significant. A multivariable-adjusted population attributable risk (PAR) estimate was calculated separately for each of the binary TBI characteristics using methods proposed by Chen, Lin, and Zeng(47) for cohort studies with censored event times implemented in the paf package in R using a Cox model, adjusted for the same four forementioned non-TBI covariates.(48) PAR is the proportion of disease in a population that could be prevented by elimination of an exposure or risk factor.(49) For age at initial TBI, age was dichotomized as ≥5 vs. <5 years, based on stratification for the youngest age group most commonly used by the Centers for Disease Control (0–4).(50) Reported values of PAR correspond to 10 years after the TBI. Results Demographics and Presentation Characteristics Out of 5,518 individuals in the cohort, TBI prior to 10 years of age was identified in 566 individuals. Of these 566 TBI cases, 1 had a TBI-related death at 5 years of age and 3 did not have research authorization at the time of the current study and were excluded, resulting in 562 TBI cases. There were 238 females (42.3%) and 324 males (57.7%) in the cohort, and 546 (97.2%) were white Demographic and socioeconomic characteristics of the TBI cases are summarized in the first column of Table 1. Of the 562 cases of TBI identified, 8 had Definite TBI (1.4%), 140 had Probable TBI (24.9%), and 414 had Possible TBI (73.7%). The mean age of TBI for all severities was 4.7 years (SD 2.8). Skull fracture occurred in 49 (8.7%) individuals, 63 (11.2%) individuals were hospitalized for their injury, and extracranial injury of any sort occurred in 144 (25.6%) individuals. The most common cause of injury was falls, occurring in 344/562 individuals (data not shown). There were 80 individuals (14.2 %) whose mechanism of injury was getting hit by an object, though there were no cases found of abuse / non-accidental trauma found. Anxiety and Mood Disorders Among the 562 TBI cases, just one individual was diagnosed with anxiety or mood disorders prior to the index date - a male diagnosed with depression at age 7 who experienced a TBI at 9 years of age. A total of 115 TBI cases received either an anxiety (n=47) or mood disorder (n=109) diagnosis between the initial TBI and their 25th birthday; the median age at first diagnosis was 18.1 years (interquartile range (IQR), 16.3–21.7). Among the remaining patients without a subsequent diagnosis of an anxiety or mood disorder, the median age at their last visit to REP-affiliated practitioner prior to their 25th birthday was 23.1 (IQR, 20.4–25.5) years (data not shown). Univariate Analysis The panels in Figure 1 depict the relationship between continuously scaled characteristics and the risk of subsequent anxiety. For maternal age, birth weight, and age at initial TBI it is reasonable to assume a linear relationship with the risk of anxiety, whereas for gestational age there is a potential non-linear relationship and therefore this variable was subsequently assessed using gestational age categories. Similar patterns were observed when these characteristics were evaluated for an association with the risk of a subsequent mood disorder (data not shown). Tables 1 and 2 summarize the unadjusted hazard ratios from the of the univariate analyses evaluating non-TBI characteristics of interest (Table 1) and TBI characteristics (Table 2) for an association with either a subsequent anxiety or mood disorder. Female sex, maternal education level, gestational age, maternal depression prior to initial TBI, older age at first childhood TBI, and having a Definite TBI diagnosis (vs. Possible TBI) were all significantly associated with an increased risk of anxiety disorder. Female sex and older age at first TBI were also significantly associated with an increased risk of mood disorder. Recurrent TBI prior to age 10, ADHD diagnosis, and LD diagnosis, each evaluated as time-dependent covariates, were not identified as significantly associated with an increased risk of anxiety or mood disorder (data not shown). Multivariable Analysis The 5 TBI characteristics – age at injury, injury severity, hospitalization, extracranial injury, and skull fracture at time of TBI – were each evaluated separately for associated risk of subsequent psychiatric disorders; this was done after adjusting for characteristics that are known to be associated with depression and anxiety disorders in the general pediatric and adolescent population that were found to be significant in our univariate analysis.(44–46) Upon adjusting for non-TBI characteristics, both older age at initial TBI (HR 1.33 [95% CI 1.16–1.52]), having a Definite TBI (vs. Possible TBI: HR 4.89 [95% CI 1.08–22.12]) and extracranial injury (HR 2.41 [95% CI 1.26, 4.59]) at time of initial TBI were significantly associated with an increased risk of anxiety when evaluated separately (Table 2). Only older age at initial TBI (HR 1.17 [95 CI 1.08, 1.27]) was significantly associated with an increased risk of mood disorders after adjusting for potential covariates (Table 2). Among the 5 TBI characteristics, the multivariable-adjusted population attributable risk (PAR) of subsequent anxiety was highest for older age at initial TBI (54.5% using an age cutoff of 5), followed by 21.1% for extracranial injury, 6.7% for definite vs. possible TBI, 6.8% for skull fracture, and 6.1% for hospitalization. Likewise, the multivariable-adjusted PAR of subsequent mood disorder was highest for older age at initial TBI (35.8% using an age cutoff of 5). Discussion In this longitudinal birth cohort study, older age at TBI, Definite TBI, female sex, maternal education either less than high school or some college, late term pregnancy, as well as maternal depression were univariately associated with development of anxiety disorder by age 25. After adjusting for several characteristics known to be associated with anxiety and mood disorders in the general pediatric population, older age at injury remained significantly associated with an anxiety disorder. Additionally, extracranial injury at time of initial TBI was significantly associated with an increased risk of subsequent anxiety, as was Definite TBI. Only older age at the time of the initial TBI was associated with development of a mood disorder. The multivariable-adjusted PAR for both anxiety and mood disorder was highest for older age at TBI, indicating that among TBI related factors, age at injury attributed to the greatest risk of an anxiety or mood disorder. The age at which pediatric TBI occurs has important implications as there may be a relative vulnerability to TBI at different ages depending on the stage of brain development.(3, 24) In this study, the multivariable-adjusted population attributable risk (PAR) for subsequent anxiety and mood disorder was highest for age at initial TBI between ages 5–10 (54.5%) when compared to ages 0–4. Because the greatest incidence of childhood TBI occurs from ages 0–4,(2) it is reassuring that individuals who sustained a TBI at this age were at a relatively lower risk of developing an anxiety or mood disorder by early adulthood. Our results are consistent with previous studies that have found older age at injury to be an important predictor of psychiatric outcome. In a nationwide Swedish cohort assessing a wide range of psychological long term outcomes, older age at injury was a risk factor for psychiatric inpatient hospitalization for individuals with TBI compared to siblings without TBI, with the least significant association found for individuals injured between ages 0–4.(51) Max et al. showed that age at injury was a significant predictor of both depressive and anxiety disorders 6 months after TBI, where the mean age at injury for children with a new anxiety disorder was 8.4 years and 11.9 years for development of a depressive disorder.(18, 52) A diffuse biomechanical injury has the potential to alter developmental outcomes as the injury and subsequent recovery occurs while the brain is developing.(53) While there is greater evidence describing mechanisms of cognitive outcomes after different ages of TBI,(54) the mechanisms that may contribute to psychiatric disorders during various stages of development remains to be elucidated, particularly in the long term.(24) There is evidence that TBI occurring between ages 7–9 may be associated with worse cognitive outcomes, suggesting that this time period may represent a critical period of brain development.(55) In addition, there is research showing that the neural mechanisms between cognitive and emotional outcomes are highly intertwined.(56) Studies have shown anatomical correlates including left inferior frontal and right frontal white matter lesions correlating with depression in children,(18) though age related factors associated with neuro-anatomical correlates are unknown. It is also possible that chronic behavioral and psychological conditions may develop relating to cognitive difficulties and subsequent changes in poor academic and job performance, thus resulting in behavioral/psychiatric concerns.(24) Overall, further research is needed to understand the neuroanatomic and neurophysiological correlates of psychiatric outcomes after childhood TBI during various stages of development. Given that the sample in this study was predominantly individuals who sustained mild TBI (Mayo Classification Possible and Probable TBI), it is less likely that solely TBI related pathophysiological changes attributed the findings of a differential outcome between younger and relatively older age groups. Rather, it is more likely that other factors independent of the injury – including a variety of personal, family, and psychosocial variables - contributed to outcomes found in this sample. While we accounted for several birth, family, and socioeconomic factors, there may be other factors that we did not account for that contributed to this association. While increasing injury severity has been more consistently found to be a risk factor for worse cognitive impairment after pediatric TBI,(5) TBI severity has not consistently been found to increase risk for a long term clinically diagnosed mood or anxiety disorder.(6) A recent scoping review concluded that depression is largely a secondary outcome after pediatric TBI, rather than directly relating to severity of injury.(6) Consistent with prior analyses of population based samples with TBI, injuries in this study were predominated by Possible or Probable TBI,(33, 57) with only a few individuals with Definite TBI (n = 8). While Definite TBI compared to Possible TBI was significantly associated with an anxiety disorder, (adjusted HR 4.89 [95% CI 1.08–22.12]), the importance of this finding should be interpreted with caution due to the small number of Definite TBI cases in the sample. The birth cohort sample included TBI sustained prior to the widespread implementation of computed tomography for TBI. Though 63 (11.2%) of individuals were hospitalized and 49 (8.7%) of individuals in the sample sustained a skull fracture, classification of Definite TBI cases may have been limited by a lack of characterization of intracranial hemorrhage. While several longitudinal studies have identified intracranial lesions in the frontal lobes to be associated with a mood and anxiety disorders up to several years after pediatric TBI,(17, 52) further research is needed to understand neuroimaging correlates to the development of these disorders into adulthood. Among other TBI related variables we studied, extracranial injury was significantly associated with later development of an anxiety disorder. The impact of extracranial injuries such as orthopedic trauma concomitant with TBI may be associated with immobility, hospitalization, and delayed return to functional independence, which can result in increased psychiatric sequelae.(7) In this study, extracranial injury was not quantified by a validated scoring system but rather abstracted as a dichotomous variable. Thus, conclusions about the impact of non-head injury are limited regarding the differential impact of the severity and specific type of extracranial injury. Previous literature has shown that Mean Abbreviated Injury Scale scores for non-head injuries did not significantly differ between children who developed or did not develop a new psychiatric disorder up to a year after TBI in one study,(58) and that behavioral problems at 12 months after TBI was not associated with extracranial injuries at the time of mTBI.(59) Taken as a whole, these results suggest that age at time of TBI may be an important characteristic to consider when assessing for a long-term risk of a mood or anxiety disorder after a pediatric TBI occurring before age 10. In assessing for this risk, clinicians should consider screening children who sustain TBI for known risk factors for development of mood and anxiety disorders in the general population, such as female sex, decreased maternal education, low fetal weight percentile, and family history of depression. While our study did show that extracranial injury and Definite TBI were significantly associated with an anxiety disorder, extracranial injury severity and type was not quantified, and the sample of Definite TBI was very small (n=8). Future studies may consider quantifying extracranial injury, further understanding the impact of intracranial lesions and the risk in those with severe TBI, and better understand the differential impact of pediatric TBI after age 10 to understand the impact of injury severity and relative older age at injury on the long-term risk of mood and anxiety disorders. Ultimately, future research may consider a prediction tool that identifies children at highest risk for adverse psychiatric outcomes, with prospective studies then assessing whether early intervention targeting modifiable risk factors and targeted treatment strategies for individuals at highest risk decreases the incidence of psychiatric sequelae after childhood TBI. There were several strengths to this study. We used a longitudinal birth cohort which allowed for abstraction of characteristics in multiple domains including birth related characteristics, family history of depression defined by medical review, an individual measure of socioeconomic status (HOUSES index),(34) as well as research defined and confirmed ADHD and LD diagnoses. In addition, TBI cases were manually confirmed and classified by injury severity using a validated classification system,(31) and outcomes were defined based on prior studies and with use of manual record review with adequate reliability.(41) The longitudinal design overall allowed for long-term associations to be made, which has otherwise been scantily reported in the literature.(6, 7) There were several limitations as well. While we accounted for risk factors in several domains, there may be unmeasured confounders/mediating variables that were unaccounted for, such as the impact of early life stress, family environment, or specific adverse childhood events which are known risk factors for mood and anxiety disorders.(60, 61) Though an individual measure of socioeconomic status was used, 35% of study subjects did not have HOUSES index, largely due to incomplete addresses at birth (e.g., PO Box addresses). The HOUSES index can be investigated in future studies where SES at specific periods (e.g., at birth, at time of TBI, at the time of diagnosis of depression or anxiety), cumulative SES, or change of SES affect development of depression or anxiety.(62) Although we did find female sex to be a risk factor for both a mood and anxiety disorder in our univariate analysis and adjusted for sex in our multivariate model, we did not assess for the relative risk of sex specifically in our PAR analysis. Future research should focus on the relative impact of sex on the incidence of new onset mood and anxiety disorders after pediatric TBI. There were very few cases of bipolar disorder identified in this cohort, and depressive disorders were groups with bipolar disorders in a broad category of mood disorders. While this methodology has been used in a previous study,(41) further research is needed to understand risk factors specifically for bipolar disorder after childhood TBI. Finally, while age, sex, and ethnic characteristics of Olmsted County residents are similar to those of the state of Minnesota and the Upper Midwest, the understanding of the risk in other urban as well as ethnically or socioeconomically diverse populations in the United States was limited in this study.(27, 63, 64) Supplementary Material Supplemental Acknowledgments This study used the resources of the Rochester Epidemiology Project (REP) medical records-linkage system, which is supported by the National Institute on Aging (NIA; AG 058738), by the Mayo Clinic Research Committee, and by fees paid annually by REP users. The content of this article is solely the responsibility of the authors and does not represent the official views of the National Institutes of Health (NIH) or the Mayo Clinic. Abbreviations ADHD Attention Deficit Hyperactivity Disorder aHR adjusted hazard ratio H-ICDA Hospital Adaptation of the International Classification of Diseases, Eighth Revision ICD-9 International Classification of Diseases, Ninth Revision IQR interquartile range LD learning disabilities PAR Population attributable risk REP Rochester Epidemiology Project TBI traumatic brain injury Figure 1. Graphical depiction of the association of maternal age at birth (A), gestational age (B), birth weight (C), and age at initial TBI (D), respectively, with the risk of anxiety as estimated by modelling each covariate using a smoothing spline in a separate Cox model. The dashed lines denote 95% confidence bands for the hazard ratio (HR). For illustrative purposes the values in the tails of the distribution of each covariate have been rounded prior to fitting the Cox models. Each graph was generated such that the reference point with a HR of 1.0 was set at the mean of each covariate. Table 1. Non-TBI characteristics evaluated univariately for an association with anxiety or mood disorder among 562 individuals with TBI prior to 10 years of age Characteristic Overall distribution† Outcome=Anxiety Outcome =Mood disorder No. with the outcome Unadjusted HR (95% CI) No. with the outcome Unadjusted HR (95% CI) Demographic Sex  Female 238 (42.3) 33 2.74 (1.46, 5.12) 73 2.61 (1.75, 3.90)  Male 324 (57.7) 14 Referent 36 Referent Child race  White 546 (97.2) 46 1.64 (0.23, 11.89) 106 1.28 (0.41, 4.04)  Non-white 16 (2.8) 1 Referent 3 Referent Socioeconomic at birth Maternal age (years) 26.2 (4.7) 47 1.28 (0.94, 1.73) ‡ 109 1.01 (0.83, 1.24) ‡ Paternal age (years) 28.6 (5.3) 47 1.27 (0.98, 1.64) ‡ 109 1.07 (0.89, 1.28) ‡ Maternal education  Less than high school 45 (8.0%) 7 4.66 (1.63, 13.32) 9 1.36 (0.64, 2.87)  High school graduate 169 (30.1%) 7 Referent 29 Referent  Some college 180 (32.0%) 18 2.58 (1.08, 6.18) 38 1.32 (0.81, 2.14)  College graduate 121 (21.5%) 10 2.14 (0.81, 5.62) 18 0.89 (0.50, 1.61)  Not documented 47 (8.4%) 5 -- 15 -- Paternal education  Less than high school 30 (5.3%) 1 0.47 (0.06, 3.69) 7 1.05 (0.46, 2.39)  High school graduate 150 (26.7%) 9 Referent 29 Referent  Some college 123 (21.9%) 12 1.84 (0.77, 4.37) 24 1.09 (0.63, 1.87)  College graduate 184 (32.7%) 17 1.74 (0.78, 3.91) 27 0.80 (0.48, 1.36) Not documented 75 (13.3%) 8 -- 22 -- HOUSES Index  1 (lowest SES) 90 (16.0%) 11 2.31 (0.80, 6.66) 22 1.54 (0.81, 2.94)  2 72 (12.8%) 10 2.84 (0.97, 8.32) 12 1.05 (0.50, 2.22)  3 103 (18.3%) 7 1.45 (0.46, 4.57) 22 1.49 (0.78, 2.84)  4 (highest SES) 101 (18.0%) 5 Referent 16 Referent  Unable to determine 196 (34.9%) 14 -- 37 -- Family history Maternal depression prior to initial TBI  Yes 50 (8.9%) 7 2.36 (1.05, 5.26) 10 1.31 (0.68, 2.52)  No 512 (91.1%) 40 Referent 99 Referent Birth Gestational age  Preterm (≤36 weeks) ^ 28 (5.0%) 4 2.52 (0.83, 7.66) 8 1.39 (0.66, 2.94)  Early term (37–38 weeks) 79 (14.1%) 7 2.21 (0.89, 5.49) 15 1.29 (0.72, 2.30)  Full term (39–40 weeks) 271 (48.2%) 14 Referent 49 Referent  Late term (41 weeks) 90 (16.0%) 16 3.94 (1.92, 8.07) 20 1.34 (0.80, 2.26)  Postterm (42–45 weeks) 79 (14.1%) 5 1.14 (0.41, 3.18) 16 1.06 (0.60, 1.87)  Not documented 15 (2.7%) 1 -- 1 -- Birth weight (grams) 3487 (530) 47 1.15 (0.88,1.51)‡ 109 1.08 (0.90, 1.29)‡ Size for gestational age  Small 16 (2.8%) 3 3.13 (0.96, 10.18) 4 1.78 (0.65, 4.85)  Average 414 (73.7%) 37 Referent 83 Referent  Large 117 (20.8%) 6 0.58 (0.25, 1.39) 21 0.96 (0.60, 1.56)  Unable to determine 15 (2.7%) 1 -- 1 -- Abbreviations: CI, confidence interval; HR, hazard ratio; SD, standard deviation. The HR (95% CI) are bolded if the 95% CI does not contain 1, indicating statistical significance at the 0.05 level. † Results presented as frequency and percentage (of 562 patients) for categorical variables and mean (SD) for continuous variables. ‡ Hazard ratio per a 1-year increase in the child’s age, per a 5-year increase in the age of the parent, and per 500-gram decrease in birth weight, respectively. ^ Of the 28 individuals who were born preterm, 2 were very preterm (28–31 weeks) and 26 were moderate or later preterm (32–36 weeks). Table 2. TBI characteristics evaluated for an association with anxiety or mood disorder among 562 individuals with TBI prior to 10 years of age Characteristic at time of initial TBI Overall distribution† Outcome=Anxiety Outcome =Mood disorder No. with the outcome Unadjusted HR (95% CI) Adjusted HR (95% CI)‡ No. with the outcome Unadjusted HR (95% CI) Adjusted HR (95% CI)‡ Age (years) 4.7 (2.8) 47 1.32 (1.15, 1.52) § 1.33 (1.16, 1.52) § 109 1.14 (1.06, 1.24) § 1.17 (1.08, 1.27) § TBI severity  Definite 8 (1.4%) 2 4.82 (1.14, 20.30) 4.89 (1.08, 22.12) 3 2.94 (0.92, 9.33) 2.95 (0.91, 9.59)  Probable 140 (24.9%) 14 1.22 (0.65, 2.29) 1.24 (0.65, 2.37) 27 0.92 (0.59, 1.42) 0.95 (0.61, 1.48)  Possible 414 (73.7%) 31 Referent Referent 79 Referent Referent Hospitalized  Yes 63 (11.2%) 7 1.33 (0.60, 2.97) 1.76 (0.76, 4.12) 14 1.12 (0.64, 1.96) 1.31 (0.74, 2.32)  No 499 (88.8%) 40 Referent Referent 95 Referent Referent Skull fracture  Yes 49 (8.7%) 8 1.71 (0.80, 3.68) 1.92 (0.86, 4.24) 10 0.85 (0.44, 1.64) 0.93 (0.48, 1.80)  No 513 (91.3%) 39 Referent Referent 99 Referent Referent Extracranial injury  Yes 144 (25.6%) 16 1.62 (0.88, 2.96) 2.41 (1.26, 4.59) 26 0.92 (0.59, 1.42) 1.05 (0.66, 1.65)  No 418 (74.4%) 31 Referent Referent 83 Referent Referent Abbreviations: CI, confidence interval; HR, hazard ratio; SD, standard deviation. The HR (95% CI) are bolded if the 95% CI does not contain 1, indicating statistical significance at the 0.05 level. † Results presented as frequency and percentage (of 562 patients) for categorical variables and mean (SD) for continuous variables. ‡ Adjusted for variables significant in Table 2:Each of the 5 TBI characteristics was evaluated in a separate multivariable Cox proportional hazards regression model adjusted for the following covariates: sex, size for gestational age, maternal education, and history of maternal depression at the time of the individual’s birth. § Hazard ratio per a 1-year increase in the child’s age at initial TBI Conflicts of Interest and Financial Disclosure Dr. Esterov received training or funding through the Center for Clinical and Translational Science (funded through the Small Grants Program) and assistance as part of a postdoctoral master’s degree in Clinical and Translational Research (assistance with design, methods). This publication was supported by CTSA Grant Number UL1 TR002377 from the National Center for Advancing Translational Science (NCATS). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH. References 1. Dewan MC , Mummareddy N , Wellons JC , 3rd, Bonfield CM. Epidemiology of Global Pediatric Traumatic Brain Injury: Qualitative Review. World Neurosurg. 2016;91 :497–509.e1. Epub 2016/03/29. doi: 10.1016/j.wneu.2016.03.045.27018009 2. Centers for Disease Control and Prevention. (2018). Report to Congress: The Management of Traumatic Brain Injury in Children, National Center for Injury Prevention and Control; Division of Unintentional Injury Prevention. Atlanta, GA. 3. Babikian T , Asarnow R . Neurocognitive outcomes and recovery after pediatric TBI: meta-analytic review of the literature. Neuropsychology. 2009;23 (3 ):283–96. Epub 2009/05/06. doi: 10.1037/a0015268.19413443 4. Andruszkow H , Deniz E , Urner J , Probst C , Grun O , Lohse R , Frink M , Krettek C , Zeckey C , Hildebrand F . Physical and psychological long-term outcome after traumatic brain injury in children and adult patients. Health Qual Life Outcomes 12 , 26 (2014). doi: 10.1186/1477-7525-12-26.24571742 5. Neumane S , Câmara-Costa H , Francillette L , Araujo M , Toure H , Brugel D , Laurent-Vannier A , Ewing-Cobbs L , Meyer P , Dellatolas G , Watier L , Chevignard M . Functional outcome after severe childhood traumatic brain injury: Results of the TGE prospective longitudinal study. Ann Phys Rehabil Med. 2021;64 (1 ):101375. doi: 10.1016/j.rehab.2020.01.008.32275965 6. Laliberte Durish C , Pereverseff RS , Yeates KO . Depression and Depressive Symptoms in Pediatric Traumatic Brain Injury: A Scoping Review. J Head Trauma Rehabil. 2018;33 (3 ):E18–e30. Epub 2017/09/20. doi: 10.1097/htr.0000000000000343. 7. Emery CA , Barlow KM , Brooks BL , Max JE , Villavicencio-Requis A , Gnanakumar V , Robertson HL , Schneider K , Yeates KO . A Systematic Review of Psychiatric, Psychological, and Behavioural Outcomes following Mild Traumatic Brain Injury in Children and Adolescents. The Canadian Journal of Psychiatry. 2016;61 (5 ):259–269. doi:10.1177/0706743716643741.27254800 8. Max JE , Wilde EA , Bigler ED , MacLeod M , Vasquez AC , Schmidt AT , Chapman SB , Hotz G , Yang TT , Levin HS . Psychiatric disorders after pediatric traumatic brain injury: a prospective, longitudinal, controlled study. J Neuropsychiatry Clin Neurosci. 2012;24 (4 ):427–36. Epub 2012/12/12. doi: 10.1176/appi.neuropsych.12060149.23224448 9. Albicini M , McKinlay A . Anxiety Disorders in Adults With Childhood Traumatic Brain Injury: Evidence of Difficulties More Than 10 Years Postinjury. J Head Trauma Rehabil. 2018;33 (3 ):191–9. doi: 10.1097/HTR.0000000000000312.28520662 10. Luis CA , Mittenberg W . Mood and Anxiety Disorders Following Pediatric Traumatic Brain Injury: A Prospective Study. J Clin Exp Neuropsychol. 2002;24 (3 ):270–9. doi: 10.1076/jcen.24.3.270.982.11992209 11. Merikangas KR , He JP , Burstein M , Swanson SA , Avenevoli S , Cui L , Benjet C , Georgiades K , Swendsen J . Lifetime prevalence of mental disorders in U.S. adolescents: results from the National Comorbidity Survey Replication--Adolescent Supplement (NCS-A). J Am Acad Child Adolesc Psychiatry. 2010;49 (10 ):980–9. Epub 2010/09/22. doi: 10.1016/j.jaac.2010.05.017.20855043 12. Baranne ML , Falissard B . Global burden of mental disorders among children aged 5–14 years. Child Adolesc Psychiatry Ment Health. 2018;12 (1 ):19. doi: 10.1186/s13034-018-0225-4.29682005 13. Tham SW , Fales J , Palermo TM . Subjective and objective assessment of sleep in adolescents with mild traumatic brain injury. J Neurotrauma. 2015;32 (11 ):847–52. Epub 2015/02/25. doi: 10.1089/neu.2014.3559.25707446 14. Max JE . Neuropsychiatry of pediatric traumatic brain injury. Psychiatr Clin North Am. 2014;37 (1 ):125–40. Epub 2014/01/14. doi: 10.1016/j.psc.2013.11.003.24529428 15. O’Connor SS , Zatzick DF , Wang J , Temkin N , Koepsell TD , Jaffe KM , Durbin D , Vavilala MS , Dorsch A , Rivara FP . Association between posttraumatic stress, depression, and functional impairments in adolescents 24 months after traumatic brain injury. J Trauma Stress. 2012;25 (3 ):264–71. Epub 2012/06/26. doi: 10.1002/jts.21704.22729979 16. Tham SW , Palermo TM , Wang J , Jaffe KM , Temkin N , Durbin D , Rivara FP . Persistent pain in adolescents following traumatic brain injury. J Pain. 2013;14 (10 ):1242–9. Epub 2013/08/02. doi: 10.1016/j.jpain.2013.05.007.23911979 17. Max JE , Friedman K , Wilde EA , Bigler ED , Hanten G , Schachar RJ , Saunders AE , Dennis M , Ewing-Cobbs L , Chapman SB , Psychiatric disorders in children and adolescents 24 months after mild traumatic brain injury. J Neuropsychiatry Clin Neurosci. 2015;27 (2 ):112–20. Epub 2015/04/30. doi: 10.1176/appi.neuropsych.13080190.25923850 18. Max JE , Keatley E , Wilde EA , Bigler ED , Schachar RJ , Saunders AE , Ewing-Cobbs L , Chapman SB , Dennis M , Yang TT , Depression in children and adolescents in the first 6 months after traumatic brain injury. Int J Dev Neurosci. 2012;30 (3 ):239–45. Epub 2011/12/17. doi: 10.1016/j.ijdevneu.2011.12.005.22197971 19. Vasa RA , Gerring JP , Grados M , Slomine B , Christensen JR , Rising W , Denckla MB , Riddle MA . Anxiety After Severe Pediatric Closed Head Injury. J Am Acad Child Adolesc Psychiatry. 2002;41 (2 ):148–56. doi: 10.1097/00004583-200202000-00008.11837404 20. Grados MA , Vasa RA , Riddle MA , Slomine BS , Salorio C , Christensen J , Gerring J . New onset obsessive-compulsive symptoms in children and adolescents with severe traumatic brain injury. Depression and Anxiety. 2008;25 (5 ):398–407. doi: 10.1002/da.20398.17957806 21. Albicini M , McKinlay A . A systematic review of anxiety disorders following mild, moderate and severe TBI in children and adolescents. In: Durbano F , editor. A fresh look at anxiety disorders. IntechOpen; 2015; p. 199–24. 22. Lumba-Brown A , Yeates KO , Sarmiento K , Breiding MJ , Haegerich TM , Gioia GA , Turner M , Benzel EC , Suskauer SJ , Giza CC , Diagnosis and Management of Mild Traumatic Brain Injury in Children: A Systematic Review. JAMA Pediatr. 2018;172 (11 ):e182847–e. doi: 10.1001/jamapediatrics.2018.2847.30193325 23. McKinlay A , Grace R , Horwood J , Fergusson D , MacFarlane M . Adolescent psychiatric symptoms following preschool childhood mild traumatic brain injury: evidence from a birth cohort. J Head Trauma Rehabil. 2009;24 (3 ):221–7. Epub 2009/05/23. doi: 10.1097/HTR.0b013e3181a40590.19461369 24. Babikian T , Merkley T , Savage RC , Giza CC , Levin H . Chronic Aspects of Pediatric Traumatic Brain Injury: Review of the Literature. J Neurotrauma. 2015;32 (23 ):1849–60. Epub 2015/09/29. doi: 10.1089/neu.2015.3971.26414654 25. Katusic SK , Colligan RC , Myers SM , Voigt RG , Yoshimasu K , Stoeckel RE , Weaver AL . What can large population-based birth cohort study ask about past, present and future of children with disorders of development, learning and behaviour? J Epidemiol Community Health. 2017;71 (4 ):410–6. Epub 2017/02/09. doi: 10.1136/jech-2016-208482.28167642 26. Melton LJ . History of the Rochester Epidemiology Project. Mayo Clinic Proceedings. 1996;71 (3 ):266–74. doi: 10.4065/71.3.266.8594285 27. Rocca WA , Yawn BP , St. Sauver JL , Grossardt BR , Melton LJ . History of the Rochester Epidemiology Project: Half a Century of Medical Records Linkage in a US Population. Mayo Clinic Proceedings. 2012;87 (12 ):1202–13. doi: 10.1016/j.mayocp.2012.08.012.23199802 28. Katusic SK , Colligan RC , Barbaresi WJ , Schaid DJ , Jacobsen SJ . Incidence of reading disability in a population-based birth cohort, 1976–1982, Rochester, Minn. Mayo Clin Proc. 2001;76 (11 ):1081–92. Epub 2001/11/13. doi: 10.4065/76.11.1081.11702896 29. Barbaresi WJ , Katusic SK , Colligan RC , Weaver AL , Jacobsen SJ . Math learning disorder: incidence in a population-based birth cohort, 1976–82, Rochester, Minn. Ambul Pediatr. 2005;5 (5 ):281–9. Epub 2005/09/20. doi: 10.1367/a04-209r.1.16167851 30. Barbaresi WJ , Katusic SK , Colligan RC , Pankratz VS , Weaver AL , Weber KJ , Mrazek DA , Jacobsen SJ . How common is attention-deficit/hyperactivity disorder? Incidence in a population-based birth cohort in Rochester, Minn. Arch Pediatr Adolesc Med. 2002;156 (3 ):217–24. Epub 2002/03/07. doi: 10.1001/archpedi.156.3.217.11876664 31. Malec JF , Brown AW , Leibson CL , Flaada JT , Mandrekar JN , Diehl NN , Perkins PK . The Mayo classification system for traumatic brain injury severity. Journal of Neurotrauma. 2007;24 (9 ):1417–24. doi: 10.1089/neu.2006.0245.17892404 32. Brown AW , Leibson CL , Mandrekar J , Ransom JE , Malec JF . Long-Term Survival After Traumatic Brain Injury: A Population-Based Analysis Controlled for Nonhead Trauma. The Journal of Head Trauma Rehabilitation. 2014;29 (1 ):E1–E8. doi: 10.1097/HTR.0b013e318280d3e6. 33. Leibson CL , Brown AW , Ransom JE , Diehl NN , Perkins PK , Mandrekar J , Malec JF . Incidence of traumatic brain injury across the full disease spectrum: a population-based medical record review study. Epidemiology. 2011;22 (6 ):836–44.21968774 34. Juhn YJ , Beebe TJ , Finnie DM , Sloan J , Wheeler PH , Yawn B , Williams AR . Development and Initial Testing of a New Socioeconomic Status Measure Based on Housing Data. J Urban Health. 2011;88 (5 ):933–44. doi: 10.1007/s11524-011-9572-7.21499815 35. Takahashi PY , Ryu E , Hathcock MA , Olson JE , Bielinski SJ , Cerhan JR , Rand-Weaver J , Juhn YJ . A novel housing-based socioeconomic measure predicts hospitalisation and multiple chronic conditions in a community population. J Epidemiol Community Health. 2016;70 (3 ):286. doi: 10.1136/jech-2015-205925.26458399 36. Butterfield MC , Williams AR , Beebe T , Finnie D , Liu H , Liesinger J , Sloan J , Wheeler PH , Yawn B , Juhn YJ . A two-county comparison of the HOUSES index on predicting self-rated health. J Epidemiol Community Health. 2011;65 (3 ):254. doi: 10.1136/jech.2008.084723.20439350 37. Brenner WE , Edelman DA , Hendricks CH . A standard of fetal growth for the United States of America. Am J Obstet Gynecol. 1976;126 (5 ):555–64. Epub 1976/11/01. doi: 10.1016/0002-9378(76)90748-1.984126 38. ACOG Committee Opinion No 579: Definition of term pregnancy. Obstet Gynecol. 2013;122 (5 ):1139–40. Epub 2013/10/24. doi: 10.1097/01.AOG.0000437385.88715.4a.24150030 39. Katusic SK , Colligan RC , Weaver AL , Barbaresi WJ . The forgotten learning disability: epidemiology of written-language disorder in a population-based birth cohort (1976–1982), Rochester, Minnesota. Pediatrics. 2009;123 (5 ):1306–13. Epub 2009/05/01. doi: 10.1542/peds.2008-2098.19403496 40. Katusic SK , Barbaresi WJ , Colligan RC , Weaver AL , Leibson CL , Jacobsen SJ . Case definition in epidemiologic studies of AD/HD. Ann Epidemiol. 2005;15 (6 ):430–7. Epub 2005/06/22. doi: 10.1016/j.annepidem.2004.12.004.15967390 41. Kirsch AC , Huebner ARS , Mehta SQ , Howie FR , Weaver AL , Myers SM , Voigt RG , Katusic SK . Association of Comorbid Mood and Anxiety Disorders With Autism Spectrum Disorder. JAMA Pediatr. 2020;174 (1 ):63–70. doi: 10.1001/jamapediatrics.2019.4368.31790555 42. Sawyer SM , Azzopardi PS , Wickremarathne D , Patton GC . The age of adolescence. Lancet Child Adolesc Health. 2018;2 (3 ):223–8. Epub 2018/09/01. doi: 10.1016/s2352-4642(18)30022-1.30169257 43. Handbook of DSM-5 Disorders in Children and Adolescents. Switzerland: Springer International; 2017. 44. Breslau J , Gilman SE , Stein BD , Ruder T , Gmelin T , Miller E . Sex differences in recent first-onset depression in an epidemiological sample of adolescents. Transl Psychiatry. 2017;7 (5 ):e1139–e. doi: 10.1038/tp.2017.105.28556831 45. Van Lieshout RJ , Boyle MH , Favotto L , Krzeczkowski JE , Savoy C , Saigal S , Schmidt LA . Impact of extremely low-birth-weight status on risk and resilience for depression and anxiety in adulthood. J Child Psychol Psychiatry. 2018;59 (5 ):596–603. Epub 2017/10/04. doi: 10.1111/jcpp.12826.28971484 46. Elovainio M , Pulkki-Råback L , Hakulinen C , Ferrie JE , Jokela M , Hintsanen M , Raitakari OT , Keltikangas-Järvinen L . Childhood and adolescence risk factors and development of depressive symptoms: the 32-year prospective Young Finns follow-up study. J Epidemiol Community Health. 2015;69 (11 ):1109. doi: 10.1136/jech-2014-205352.26082517 47. Chen L , Lin DY , Zeng D . Attributable fraction functions for censored event times. Biometrika. 2010;97 (3 ):713–26. doi: 10.1093/biomet/asq023.23956459 48. Chen L pAfffcsd, R package version 1.0. 2014. http://cran.r-project.org/web/packages/paf/index.html. Accessed 8 July 2021. 49. Zapata-Diomedi B , Barendregt JJ , Veerman JL . Population attributable fraction: names, types and issues with incorrect interpretation of relative risks. Br J Sports Med. 2018;52 (4 ):212–3. doi: 10.1136/bjsports-2015-095531.26964147 50. Peterson AB , Xu L , Daugherty J Surveillance Report of Traumatic Brain Injury-related Emergency Department Visits, Hospitalizations, and Deaths—United States, 2014. Centers for Disease Control and Prevention, US Department of Health and Human Services. 2019. 51. Sariaslan A , Sharp DJ , D’Onofrio BM , Larsson H , Fazel S . Long-Term Outcomes Associated with Traumatic Brain Injury in Childhood and Adolescence: A Nationwide Swedish Cohort Study of a Wide Range of Medical and Social Outcomes. PLOS Med. 2016;13 (8 ):e1002103. doi: 10.1371/journal.pmed.1002103.27552147 52. Max JE , Keatley E , Wilde EA , Bigler ED , Levin HS , Schachar RJ , Saunders A , Ewing-Cobbs L , Chapman SB , Dennis M , Yang TT . Anxiety disorders in children and adolescents in the first six months after traumatic brain injury. J Neuropsychiatry Clin Neurosci. 2011;23 (1 ):29–39. Epub 2011/02/10. doi: 10.1176/appi.neuropsych.23.1.2910.1176/jnp.23.1.jnp29.21304136 53. Giza CC , Kolb B , Harris NG , Asarnow RF , Prins ML . Hitting a moving target: Basic mechanisms of recovery from acquired developmental brain injury. Dev Neurorehabil. 2009;12 (5 ):255–68. Epub 2009/12/04. doi: 10.3109/17518420903087558.19956795 54. Taylor HG , Alden J . Age-related differences in outcomes following childhood brain insults: an introduction and overview. J Int Neuropsychol Soc. 1997;3 (6 ):555–67. Epub 1998/02/04.9448369 55. Crowe LM , Catroppa C , Babl FE , Rosenfeld JV , Anderson V . Timing of traumatic brain injury in childhood and intellectual outcome. J Pediatr Psychol. 2012;37 (7 ):745–54. Epub 2012/06/07. doi: 10.1093/jpepsy/jss070.22669504 56. Bell MA , Wolfe CD . Emotion and cognition: an intricately bound developmental process. Child Dev. 2004;75 (2 ):366–70. Epub 2004/04/02. doi: 10.1111/j.1467-8624.2004.00679.x.15056192 57. Esterov D , Bellamkonda E , Mandrekar J , Ransom JE , Brown AW . Cause of Death after Traumatic Brain Injury: A Population-Based Health Record Review Analysis Referenced for Nonhead Trauma. Neuroepidemiology. 2021 :1–8. Epub 2021/04/12. doi: 10.1159/000514807. 58. Max JE , Pardo D , Hanten G , Schachar RJ , Saunders AE , Ewing-Cobbs L , Chapman SB , Dennis M , Wilde EA , Bigler ED , Psychiatric Disorders in Children and Adolescents Six-to-Twelve Months After Mild Traumatic Brain Injury. J Nuropsychiatry Clin Neurosci. 2013;25 (4 ):272–82. doi: 10.1176/appi.neuropsych.12040078. 59. Taylor HG , Orchinik LJ , Minich N , Dietrich A , Nuss K , Wright M , Bangert B , Rusin J , Yeates KO . Symptoms of Persistent Behavior Problems in Children With Mild Traumatic Brain Injury. J Head Trauma Rehabil. 2015;30 (5 ):302–10. Epub 2015/01/30. doi: 10.1097/htr.0000000000000106.25629259 60. Gilman SE , Kawachi I , Fitzmaurice GM , Buka SL . Socio-economic status, family disruption and residential stability in childhood: relation to onset, recurrence and remission of major depression. Psychol Med. 2003;33 (8 ):1341–55. Epub 2003/10/30. doi: 10.1017/S0033291703008377.14672243 61. Chapman DP , Whitfield CL , Felitti VJ , Dube SR , Edwards VJ , Anda RF . Adverse childhood experiences and the risk of depressive disorders in adulthood. Journal of Affective Disorders. 2004;82 (2 ):217–25. doi: 10.1016/j.jad.2003.12.013.15488250 62. Chen E , Martin AD , Matthews KA . Trajectories of socioeconomic status across children’s lifetime predict health. Pediatrics. 2007;120 (2 ):e297–303. Epub 2007/07/04. doi: 10.1542/peds.2006-3098.17606533 63. St Sauver JL , Grossardt BR , Leibson CL , Yawn BP , Melton LJ , 3rd, Rocca WA. Generalizability of epidemiological findings and public health decisions: an illustration from the Rochester Epidemiology Project. Mayo Clin Proc. 2012;87 (2 ):151–60. Epub 2012/02/07. doi: 10.1016/j.mayocp.2011.11.009.22305027 64. St Sauver JL , Grossardt BR , Yawn BP , Melton LJ 3rd , Pankratz JJ , Brue SM , Rocca WA . Data resource profile: the Rochester Epidemiology Project (REP) medical records-linkage system. Int J Epi. 2012;41 (6 ):1614–24. Epub 2012/11/20. doi: 10.1093/ije/dys195.

PMC010xxxxxx/PMC10364089.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 8101368 6523 Prostate Prostate The Prostate 0270-4137 1097-0045 36842100 10364089 10.1002/pros.24500 NIHMS1916130 Article Prevalence of genotoxic bacteria in men undergoing biopsy for prostate cancer http://orcid.org/0000-0003-3699-5161 Lee John BS 1 Wickes Brian L. PhD 2 Fu Jianmin PhD 2 Brockman Nohelli E. MS 2 Garg Harshit MD 1 Jobin Christian PhD 3 Johson-Pais Teresa PhD 1 Leach Robin PhD 4 Lai Zhao PhD 5 http://orcid.org/0000-0001-6978-1026 Liss Michael A. MD, PhD 1 1 Department of Urology, University of Texas Health San Antonio, San Antonio, Texas, USA 2 Department of Microbiology, Immunology, and Molecular Genetics, University of Texas Health San Antonio, San Antonio, Texas, USA 3 Division of Gastroenterology, Hepatology, and Nutrition, University of Florida College of Medicine, Gainesville, Florida, USA 4 Department of Cell and Systems Biology, University of Texas Health San Antonio, San Antonio, Texas, USA 5 Department of Molecular Medicine, University of Texas Health San Antonio, San Antonio, Texas, USA Correspondence: John Lee, Department of Urology, University of Texas Health San Antonio, San Antonio, TX, USA. leej26@livemail.uthscsa.edu 13 7 2023 5 2023 26 2 2023 24 7 2023 83 7 663669 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Background: New evidence suggests that bacteria-produced DNA toxins may have a role in the development or progression of prostate cancer. To determine the prevalence of these genes in a noninfection (i.e., colonized) state, we screened urine specimens in men before undergoing a biopsy for prostate cancer detection. Methods: We developed a multiplex polymerase chain reaction using three of the most described bacterial genotoxin gene primers: Colibactin (polyketone synthase [pks] gene island: clbN and clbB), cytotoxic necrotizing factor (cnf1) toxin, and cytolethal distending toxin B (cdtB) represented gene islands. After calibration on Escherichia coli samples of known genotypes, we used a training and validation cohort. We performed multiplex testing on a training cohort of previously collected urine from 45 men undergoing prostate biopsy. For the validation cohort, we utilized baseline urine samples from a previous randomized clinical trial (n = 263) with known prostate cancer outcomes. Results: The prevalence of four common bacterial genotoxin genes detected in the urine before prostate biopsy for prostate cancer is 8% (25/311). The prevalence of pks island (clbN and clbB), cnf1, and cdt toxin genes are 6.1%, 2.4%, and 1.7%, respectively. We found no association between urinary genotoxins and prostate cancer (p = 0.83). We did identify a higher proportion of low-grade cancer (92% vs. 44%) in those men positive for urinary genotoxin and higher-grade cancer in those genotoxin negative (8% vs. 56%, p = 0.001). Conclusions: The prevalence of urinary genotoxins is low and does not correspond to a prostate cancer diagnosis. The urine was taken at one point in time and does not rule out the possibility of previous exposure. E. coli genotoxin prostate cancer urinary microbiome pmc1 | INTRODUCTION Microbial–human interaction leading to cancer initiation has led to profound progress and innovative prevention techniques in several cancers, yet a microbial cause for prostate cancer remains unproven.1 Identifying a specific cause of prostate cancer could focus countermeasures to favorably impact the second most common cancer in men worldwide.2 Studies have speculated a link between chronic bacterial infection to prostate cancer through subclinical inflammation or DNA damage,3,4 similar to gastric (Helicobacter pylori)5 and colon cancer (Escherichia coli [E. coli]).6 For example, colibactin is a bacterially produced toxin implicated as a cause of colon cancer. E. coli strains that produce colibactin carry a ∽50 kb polyketide synthase operon, which harbors polyketone synthase (pks+) genes.7,8 The mechanism by which colibactin causes colon cancer includes DNA cross-linking and DNA doubl-strand breaks (DSBs). Once two strands of DNA are broken, they can undergo reassembly leading to genomic rearrangements. One example is TMPRSS2:2ERG (T2:ERG) gene fusions, which occur in half of all prostate cancer patients.9,10 The colibactin concentration has been difficult to detect outside of active infection; therefore, detection relies on the identification of pks genes. Sfanos and collaborators confirmed the plausibility of pks+ E. coli to potentially initiate prostate cancer through DNA damage induction of T2:ERG rearragment.11 However, colibactin is not the only genotoxin implicated in causing genotoxic effects. Two other bacterial-produced genotoxins include cytotoxic necrotizing factor (cnf1) toxin and cytolethal distending toxin B (cdtB) commonly occur in uropathogenic E. coli.12,13 Although cdtB has not be shown to cause DNA DSBs, we have included it for completeness. The significance of the prevalence of oncogenic bacteria in the urine of asymptomatic men to determine its association with prostate cancer risk remains unknown. Herein, we developed a multiplex polymerase chain reaction (PCR) assay that was able to amplify bacterial pks+ markers (clbB and clbN), as well as genes for cdtB, and cnf1. In the same urine samples, we compared the T2:ERG gene fusion status in relation to genotoxin presence as an indicator of DNA DSBs. 2 | METHODS 2.1 | Multiplex PCR development A pilot assay was first developed to determine sensitivity. A sample of pooled urine specimens was prepared and tested to confirm the absence of the cnf1, clbN, clbB, cdtB, and uidA markers. PCR reactions consisted of using individual primer pairs (see below) for each marker in a PCR reaction containing 1 ng of template DNA, 5 μL of 2mM dNTP, 0.5 μL KOD Xtreme Hot Start DNA polymerase (Sigma-Aldrich), 0.75 μL of a 10 μM stock of each primer, 12.5 μL of 2X buffer, and distilled water to 25 μL. Thermocycler conditions consisted of an initial activation step of 94°C for 2min, followed by 35 cycles of 98°C for 10 s, 68°C for 1min, and a final step of 68°C for 5min. The multiplex PCR reaction was developed using a Multiplex PCR Kit (Qiagen Inc.). Sensitivity limits were determined using strains B1-5-1 and B3-1-4 as controls for all four markers (cnf1, clbN, clbB, and cdtB). They were grown overnight in Luria-Bertani broth and washed once with phosphate-buffered saline, then spiked into urine to give a starting OD600 = 1 (8 × 108 colony-forming unit [CFU]/mL). Serial 10-fold dilutions were made in urine followed by extraction using the QIAamp DNA Micro Kit (Qiagen) according to the manufacturer’s instructions. The final multiplex reaction mix consisted of 12.5 μL of 2X Qiagen Multiplex PCR mix, 2.5 μL of a 2 μM stock primer mix containing each primer, 1.0 μL of template DNA extract, and 9.0 μL of water for a final reaction volume of 25 μL. The primer stock was made up of 2.0 μL of a 100 μM stock of each primer in a final volume of 100 μL. The multiplex mix consisted of primers for cnf1 (forward [For]: 5′-ATCTTATACTGGATGGGATCATCTTGG-3′, reverse [Rev]: 5′-CAGAACGACGTTCTTCATAAGTATC-3′), clbN (For: 5′-GTTTTGCTCGCCAGATAGTCATTC-3′, Rev: 5′-CAGTTCG GGTATGTGTGGAAGG-3′), clbB (For: 5′-GATTTGGATACTGGCGATAACCG-3′, Rev: 5′-CCATTTCCCGTTTGAGCACAC-3′), and cdtB (For: 5′-GAAAATAAATGGAACACACATGTCCG-3′, Rev: 5′-AAATCTCCTGCAATCATCCAGTTA-3′). uidA was used as an E. coli control but was not part of the multiplex reaction since we used it to prescreen samples that would later be tested with the multiplex mix. Primers for this gene were For: 5′-GCGTCTGTTGACTGGCA GGTGGTGG-3′, Rev: 5′-GTTGCCCGCTTCGAAACCAATGCCT-3′. Primer amplicon sizes were cnf1 974 bp, clbN 733 bp, clbB 579 bp, cdtB 466 bp, and uidA 508 bp, respectively. Primers were selected from previous studies.7,14–16 In addition to the positive controls, negative control E. coli strains used for assay development included B1-7-6, JJ0055, JJ1166, and JJ1167, which were used as negative controls for all markers except uidA. All strains were obtained from James Johnson. 2.2 | Gel electrophoresis Gel Electrophoresis of PCR samples was performed using a 1.6% combination gel of NuSieve 3:1 Agarose (Lonza Bioscience) and Certified Molecular Biology Agarose (Bio-Rad Laboratories Inc.) (1.2% NuSieve, 0.4% Certified Molecular Biology Agarose). The Invitrogen KB Plus ladder (Thermo Fischer Scientific) was used for fragment size comparison. 2.3 | Urine sample processing Previously collected urine samples were stored in 5 mL aliquots. Bacterial DNA was prepared for extraction from these samples by centrifuging 1.0 mL of urine in a 1.5 mL microfuge tube (Light Labs) at 6000g for 2 min. The supernatant was discarded, and the pellet was suspended in 500 μL of Buffer AE from the QIAamp DNA Micro Kit (Qiagen Inc.). The remaining steps were according to the manufacturer’s instructions. Samples were eluted at the final step with 50 μL of Buffer AE and stored at −20°C. 2.4 | Training population To test the multiplex in urine samples, we used previously collected urine samples from 45 men who consented and enrolled in our genitourinary urinary tissue bank (IRB#: HSC20050234H). Urine samples were collected after the digital rectal exam (DRE). The DRE is known to improve DNA concentrations specifically from prostate origin.17 2.5 | Validation population Stored urine samples were identified from the finasteride challenge study (n = 263) that randomized men to receive finasteride for 3 months before prostate biopsy (IRB#: HSC20100352H).18 Men were recruited between 2011 and 2016. The eligibility criteria specified men 55 years or older with or without prior negative prostate biopsy. At 90 ± 14 days, subjects underwent a 12-core prostate biopsy, and the samples were processed for pathologic assessment. We only analyzed those patients that consented to future studies with their specimens. 2.6 | Associated data Data were collected from the database that included demographics and pathologic assessment. In the validation cohort, we obtained PCA3 and T2:ERG status reported during the initial study. The T2:ERG status was blinded to the laboratory personnel processing the urine samples for genotoxin status. 2.7 | Statistical analyses We analyzed univariant data with χ2 and Fischer’s exact for categorical variables and Student’s test for continuous variables. We performed logistic regression for multivariant analyses with a primary outcome of prostate cancer. Secondary analysis was performed for clinically significant prostate cancer and T2:ERG status. We performed the analysis using SPSS v27 (IBM Corp.). 3 | RESULTS 3.1 | Optimization of PCR Because we were interested in developing an assay that could incorporate a large number of markers and be flexible enough to switch out or add new markers, we chose the multiplex gel assay, which is qualitative due to the output being ethidium bromide-stained band patterns on an agarose gel. Quantitative real-time PCR would be a better choice for a quantitative assay, however, we only needed an assay that gave us + or − results, but could also be used to determine the presence or absence of individual markers in a single test. The overall comparisons of our PCR testing are confirmed using known samples shown in Table 1. Figure 1A shows the results of the optimized assay in which all four bands could be detected simultaneously to a level of 14 CFU. Figure 1B shows the results comparing E. coli strains for marker specificity and Figure 1C shows the results of the uidA E. coli control. Prescreening with an individual uidA PCR test was used to save the multiplex reagents rather than incorporate uidA into the multiplex reaction. 3.2 | Demographics Demographics are displayed in Table 2. We noted that the test cohort did have a higher prostate-specific antigen than the validation cohort (6.88 vs. 5.22, p < 0.001). However, there was no statistical difference in overall cancer detection (46.7% vs. 49.2%, p = 0.87) or the diagnosis of clinically significant prostate cancer (grade group ≥2, 52.4% vs. 52.7%, p = 0.98). We also noted a higher proportion of cdt genes in the test cohort compared to the validation cohort (p = 0.02, 8.9% vs. 2.3%, respectively). We tested a total of 263 unique urine samples in the validation cohort that had associated PCA3 and T2:ERG scores (Figure 2). 3.3 | Prevalence of urinary genotoxins We identified 8% (25/311) of subjects with at least one of the three bacterial genotoxin genes present in the urine. The most common genotoxin genes identified was the pks+ (clbN or clbB) gene island which produces colibactin, with a total of 19 (6.1%) samples. cnf1 toxin and cdtB were detected less frequently for a total of 7 (1.7%) and 10 (2.4%), respectively. Eight of the 25 ((8/25, 32%) samples were positive for more than one gene. The cdtB gene was commonly accompanied by a pks gene (71%, 5/7). 3.4 | Urinary genotoxin association with prostate cancer Urinary genotoxin was not associated with prostate cancer (p = 0.83, Table 3). Low-risk prostate cancer was associated with the genotoxin group, whereas clinically significant cancer was more common in the group negative for urinary genotoxins (p = 0.001). The cancer rate of pks+ and pks− specifically was 44% (8/18) and 48% (145/288), and the difference between the two cancer rates was not statistically significant (p = 0.63). 3.5 | Genotoxin associated with T2:ERG translocation Only the validation cohort has T2:ERG and PCA3 scores to compare to genotoxin status.18 The urine T2:Erg amplification assay was published to be 92% concordant with fluorescence in situ hybridization testing in tissue.19 We found no association between urinary genotoxin with PCA3 or T2:ERG scores (p = 0.56 and p = 0.76, respectively). The T2:ERG score was more common in subjects that were diagnosed with prostate cancer (p < 0.001), but showed no significant difference in clinically significant prostate cancer (p = 0.11). The PCA3 score was more common in subjects that were diagnosed with prostate cancer (p = 0.023) but showed no significant difference in clinically significant prostate cancer (p = 0.30). Additionally, only 8.23% (7/85) of the T2:ERG+ prostate cancer participants were positive for urine pks. 4 | DISCUSSION The prevalence of the pks gene that produces the genotoxin colibactin was 5.6% (n = 15/266) in the urine of asymptomatic men before a prostate biopsy within our population (i.e., validation) cohort. We conclude that colibactin-producing bacteria are not colonized in prostates of men with prostate cancer. If colibactin is a contributor to prostate cancer, it is likely an infection event then the bacteria resolves (i.e., a high-and-run event). Other DNA-damaging genotoxins such as cnf1 toxin and cdtB were rare (n = 7, 1.7%) and 10 (2.4%). In an infection study using rats, the combination of pks+ and cnf1 caused the most severe genotoxic effects on the bladder and kidneys.20 We identified 32% (n = 8/25) of the samples that were positive for more than one genotoxic gene. We also found that the cdtB gene was more commonly accompanied by a pks gene (71%, 5/7) rather than the cnf1 gene. Colibactin is known to cause DNA alkylation and crosslinking, leading to DNA DSBs.21–23 Sfanos et al. found that in prostatectomy samples with known inflammation, the presence of colibactin-producing bacteria may be associated with a common translocation (T2:ERG) in which a double stranded DNA break is requisite.11 Even though colibactin (pks genes) was implicated in the commonT2:ERG gene fusion in prostate cancer, we found that urinary pks genes were not associated with prostate cancer in this asymptomatic cohort undergoing prostate biopsy for cancer diagnosis. Although the findings were negative, we cannot conclude that there is no association between colibactin-producing E. coli and prostate cancer. The insult may be a “hit and run” mechanism associated with a brief bout of infection or inflammation that causes DNA damage, then the bacteria are treated or dissipate over time. E. coli may clear the urine yet may leave behind a DNA damage signature, similar to what has been shown in colorectal cancer, but this possibility has yet to be confirmed.24 cnf1 catalyzes the deamination of Rho family proteins inducing multinucleation in cultured eukaryotic cells.25 cnf1 is commonly associated with uropathogenic E. coli. Colibactin (produced by pks genes) and cnf1 genes are both bacterial genotoxins associated with the induction of colon cancer.26 Fabbri et al. showed that CNF1 induces mesenchymal transition in intestinal cells as a possible cancer-inducing mechanism.27 cdt also causes DNA damage throught dual DNase and phosphatase activities to induce DNA DSBs, cell cycle arrest, and apoptosis.28 Chronic exposure to sublethal doses of CDT promotes genetic instability and a defective DNA damage response.29 Both of these bacterial genes were rare in asymptomatic men. Limitations of this study include the retrospective nature; however, in a prevalence study, our target was to have a known population of men with known tissue diagnoses of prostate cancer. We also limited our detection to only three genotoxins, although they are the most common colibactin genes known to cause DNA DSBs. T2:Erg status was not available on the testing cohort. The test cohort had post-prostate exam samples, and the validation population cohort did not, which could account for minor differences between cohorts. The most significant limitation is that the samples were taken at one point in time before a prostate biopsy. Serial sampling as well as infection history (urinary tract infection or prostatitis) would have been pertinent but was not available in the datasets. In conclusion, The prevalence of the pks gene was 5.6% (n = 15/266) and not associated with a prostate cancer diagnosis. The prevalence of any genotoxin-detection bacteria in the urine of asymptomatic men is low (8%). Future studies include testing the feces as a gut reservoir for these organisms to allow intermittent infections not captured in the urine. Additionally, the DNA damage caused by bacterial genotoxins in prostate cells is unlikely to be the same as in colon cells. Therefore, investigating a gene signature related to genotoxin exposure may be helpful to explore possible previous exposures as a cause of prostate cancer. ACKNOWLEDGMENTS The authors would like to thank Eric Oswald, Université de Toulouse, France, for sharing the pks+ Escherichia coli samples as known isolates to test polymerase chain reaction testing. John Lee was supported by the American Urological Association Summer Medical Student Fellowship sponsored by the Herbert Brendler, MD Research Fund. Funding information AUA Summer Medical Student Fellowships; Los Padres DATA AVAILABILITY STATEMENT Data are available upon request. FIGURE 1 Sensitivity and specificity of the multiplex PCR test for genotoxic bacteria. Multiplex PCR assay development. (A) Assay sensitivity: lane 1 positive control containing 140 CFU/mL, lane 2 positive control containing 70 CFU/mL, lane 3 positive control containing 28 CFU/mL, lane 4 positive control containing 14 CFU/mL, lane 5 positive control containing 2 CFU/mL, and lane 6 no DNA control. (B) Multiplex assay for marker specificity: lane 1 B1-7-6, lane 2 JJ0055, lane 3 JJ1166, lane 4 JJ1167, lane 5 positive Escherichia coli control for all four markers B1-5-1 and B3-1-4, and lane 6 no DNA control. (C) uidA controls: B1-7-6, lane 2 JJ0055, lane 3 JJ1166, lane 4 JJ1167, lane 5 positive B1-5-1 and B3-1-4, and lane 6 no DNA control. CFU, colony-forming unit; PCR, polymerase chain reaction. L, size ladder. FIGURE 2 Validation consort diagram. TABLE 1 Optimization of PCR primers. Observed Expected Johnson ID clbN (733 bp) cdtB (466 bp) clbB (579 bp) cnf1 (974 bp) uidA (503 bp) clbN (733 bp) cdtB (466 bp) clbB (579 bp) cnf1 (974 bp) uidA (503 bp) 1 + + + + + + + + 2 + + + + + + + + 3 + + + + + + + + 4 + + + + + + + + 5 + + 6 + + 7 + + 8 + + + + + + 9 + + Note: Escherichia coli strains with listed genotypes, and results, were used to construct the multiplex assay. Strains were courtesy of Dr. James Johnson at the University of Minnesota. Abbreviation: PCR, polymerase chain reaction. TABLE 2 Demographics. Demographic Training set (N = 45), median (IQR) or number (%) Validation set (n = 266), median (IQR) or number (%) p Value Total (n = 311) Age 65 (61–70) 65 (61–68) 0.65 65 (61–68) Race/ethnicity 0.09  Black 11 (24.4%) 34 (12.8%) 45 (14.5%)  Hispanic 18 (40%) 100 (37.6%) 118 (37.9%)  White 16 (35%) 132 (49.6%) 148 (47.6%) Prostate-specific antigen 6.88 (5.05–10.03) 5.22 (4.30–6.63) 0.001 5.4 (4.4–6.8) Family history of prostate cancer 11 (24.4%) 47 (18%) 0.22 58 (18.6%) PCA3 score 31.1 (13.8–60.3) T2:ERG score 4.78 (0.09–32.12) Bacterial toxin genes 6 (13.3%) 18 (6.8%) 0.16 24 (7.7%)  pks (clbN or clbB) 3 (6.7%) 15 (5.6%) 0.78 18 (4.4%)  cdt 0 (0%) 7 (2.6%) 0.27 7 (1.7%)  cnf-1 4 (8.9%) 6 (2.3%) 0.02 10 (2.4%) Prostate cancer 21 (46.7%) 131 (49.2%) 0.87 152 (48.9%)  Grade group 1 10 (22.2%) 62 (23.3%) 72 (23.2%)  Grade group 2 2 (4.4%) 32 (12%) 34 (10.9%)  Grade group 3 6 (13.3%) 20 (7.5%) 26 (8.4%)  Grade group 4 2 (4.4%) 16 (6%) 18 (5.8%)  Grade group 5 1 (2.2%) 1 (0.4%) 2 (0.6%) Abbreviation: IQR, interquartile range. TABLE 3 Demographics of pks+ versus pks− subjects. Demographic pks+ (N = 15) pks− (N = 248) p Value Age 64 (62–67) 65 (61–68) 0.82 Race/ethnicity 0.23  Black 0 (0%) 34 (14%)  Non-Black 15 (100%) 217 (87%) Prostate-specific antigen 5.6 (4.5–6.5) 5.2 (4.3–6.7) 0.57 Previous prostate biopsy 3 (20%) 40 (16%) 0.71 Prostate cancer family history 2 (13%) 45 (18%) 1 Prostate cancer prevention trial risk calculator (%) 27% (24%−32%) 28% (25%−32%) 0.25 PCA3 31.1 (8.5–42.6) 31.1 (13.8–61.6) 0.36 T2:Erg 5.5 (0.9–27.7) 4.7 (0.1–34.2) 0.6 Any prostate cancer 6 (40%) 126 (51%) 0.44 Clinically significant prostate cancer (grade group >1) 1 (7%) 68 (27%) 0.1 Copresence with other genotoxins  cnfl 4 (27%) 2 (0.8%) <0.001  cdt 5 (33%) 2 (0.8%) <0.001 CONFLICT OF INTEREST STATEMENT The authors declare no conflict of interest. REFERENCES 1. Helmink BA , Khan MAW , Hermann A , Gopalakrishnan V , Wargo JA . The microbiome, cancer, and cancer therapy. Nat Med. 2019;25 : 377–388.30842679 2. Bray F , Ferlay J , Soerjomataram I , Siegel RL , Torre LA , Jemal A . Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68 :394–424.30207593 3. Gao Y , Wei L , Wang C , Chronic prostatitis alters the prostatic microenvironment and accelerates preneoplastic lesions in C57BL/6 mice. Biol Res. 2019;52 :30.31088536 4. Puhr M , De Marzo A , Isaacs W , Inflammation, microbiota, and prostate cancer. Eur Urol Focus. 2016;2 :374–382.28723469 5. Correa P , Haenszel W , Cuello C , Tannenbaum S , Archer M . A model for gastric cancer epidemiology. Lancet. 1975;2 :58–60.49653 6. Dejea CM , Fathi P , Craig JM , Patients with familial adenomatous polyposis harbor colonic biofilms containing tumorigenic bacteria. Science. 2018;359 :592–597.29420293 7. Johnson JR , Johnston B , Kuskowski MA , Nougayrede JP , Oswald E . Molecular epidemiology and phylogenetic distribution of the Escherichia coli pks genomic island. J Clin Microbiol. 2008;46 :3906–3911.18945841 8. Nougayrède JP , Homburg S , Taieb F , Escherichia coli induces DNA double-strand breaks in eukaryotic cells. Science. 2006;313 : 848–851.16902142 9. Baca SC , Prandi D , Lawrence MS , Punctuated evolution of prostate cancer genomes. Cell. 2013;153 :666–677.23622249 10. Tomlins SA , Rhodes DR , Perner S , Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005;310 :644–648.16254181 11. Shrestha E , Coulter JB , Guzman W , Oncogenic gene fusions in nonneoplastic precursors as evidence that bacterial infection can initiate prostate cancer. Proc Natl Acad Sci USA. 2021;118 :e2018976118.34341114 12. Kadhum HJ , Finlay D , Rowe MT , WILSON IG , BALL HJ . Occurrence and characteristics of cytotoxic necrotizing factors, cytolethal distending toxins and other virulence factors in Escherichia coli from human blood and faecal samples. Epidemiol Infect. 2008;136 : 752–760.17678557 13. Kurnick SA , Mannion AJ , Feng Y , Madden CM , Chamberlain P , Fox JG . Genotoxic Escherichia coli strains encoding colibactin, cytolethal distending toxin, and cytotoxic necrotizing factor in laboratory rats. Comp Med. 2019;69 :103–113.30902120 14. Tóth I , Hérault F , Beutin L , Oswald E . Production of cytolethal distending toxins by pathogenic Escherichia coli strains isolated from human and animal sources: establishment of the existence of a new cdt variant (type IV). J Clin Microbiol. 2003;41 :4285–4291.12958258 15. Chapman TA , Wu XY , Barchia I , Comparison of virulence gene profiles of Escherichia coli strains isolated from healthy and diarrheic swine. Appl Environ Microbiol. 2006;72 :4782–4795.16820472 16. Johnson JR , Johnston BD , Gordon DM . Rapid and specific detection of the Escherichia coli sequence type 648 complex within phylogroup F. J Clin Microbiol. 2017;55 :1116–1121.28100599 17. Groskopf J , Aubin SM , Deras IL , APTIMA PCA3 molecular urine test: development of a method to aid in the diagnosis of prostate cancer. Clin Chem. 2006;52 :1089–1095.16627561 18. Hernandez J , Gelfond J , Goros M , The effect of 3-month finasteride challenge on biomarkers for predicting cancer outcome on biopsy: results of a randomized trial. PLoS One. 2018; 13 :e0204823.30300367 19. Tomlins SA , Aubin SMJ , Siddiqui J , Urine TMPRSS2:ERG fusion transcript stratifies prostate cancer risk in men with elevated serum PSA. Sci Transl Med. 2011;3 :94ra72. 20. Morgan RN , Saleh SE , Farrag HA , Aboulwafa MM . Prevalence and pathologic effects of colibactin and cytotoxic necrotizing factor-1 (Cnf 1) in Escherichia coli: experimental and bioinformatics analyses. Gut Pathog. 2019;11 :22.31139264 21. Xue M , Kim CS , Healy AR , Structure elucidation of colibactin and its DNA cross-links. Science. 2019;365 :eaax2685.31395743 22. Wilson MR , Jiang Y , Villalta PW , The human gut bacterial genotoxin colibactin alkylates DNA. Science. 2019;363 :eaar7785.30765538 23. Li ZR , Li J , Cai W , Macrocyclic colibactin induces DNA double-strand breaks via copper-mediated oxidative cleavage. Nat Chem. 2019;11 :880–889.31527851 24. PleguezueloŝManzano C , Puschhof J , Rosendahl Huber A , Mutational signature in colorectal cancer caused by genotoxic pks+ E. coli. Nature. 2020;580 :269–273.32106218 25. Caprioli A , Falbo V , Ruggeri FM , Cytotoxic necrotizing factor production by hemolytic strains of Escherichia coli causing extra-intestinal infections. J Clin Microbiol. 1987;25 :146–149.3539987 26. Cougnoux A , Dalmasso G , Martinez R , Bacterial genotoxin colibactin promotes colon tumour growth by inducing a senescence-associated secretory phenotype. Gut. 2014;63 : 1932–1942.24658599 27. Fabbri A , Travaglione S , Rosadi F , The Escherichia coli protein toxin cytotoxic necrotizing factor 1 induces epithelial mesenchymal transition. Cell Microbiol. 2020;22 :e13138.31698538 28. Graillot V , Dormoy I , Dupuy J , Genotoxicity of cytolethal distending toxin (CDT) on isogenic human colorectal cell lines: potential promoting effects for colorectal carcinogenesis. Front Cell Infect Microbiol. 2016;6 :34.27047802 29. Guidi R , Guerra L , Levi L , Chronic exposure to the cytolethal distending toxins of gramn-egative bacteria promotes genomic instability and altered DNA damage response. Cell Microbiol. 2013;15 :98–113.22998585

PMC010xxxxxx/PMC10364092.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101284307 33047 Nat Protoc Nat Protoc Nature protocols 1754-2189 1750-2799 36755131 10364092 10.1038/s41596-023-00803-0 NIHMS1901256 Article Quantitative Multiple Fragment Monitoring with Enhanced In-Source Fragmentation/Annotation Mass Spectrometry Bernardo-Bermejo Samuel a* Xue Jingchuan b* Hoang Linh c Billings Elizabeth c Webb Bill c Honders M. Willy d Venneker Sanne e Heijs Bram f Castro-Puyana María a Marina María Luisa a van den Akker Erik B. ghi Griffioen Marieke d Siuzdak Gary c# Giera Martin f# Sánchez-López Elena f# a Universidad de Alcalá, Departamento de Química Analítica, Química Física e Ingeniería Química, Ctra. Madrid-Barcelona Km. 33.600, 28871 Alcalá de Henares (Madrid), Spain. b Guangdong Provincial Key Laboratory of Water Quality Improvement and Ecological Restoration for Watersheds, Institute of Environmental and Ecological Engineering, Guangdong University of Technology, Guangzhou, 510006, China. c Scripps Center for Metabolomics, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, United States. d Department of Hematology, Leiden University Medical Center, 2300RC, Leiden, The Netherlands. e Department of Pathology, Leiden University Medical Center, 2300RC, Leiden, The Netherlands. f Center for Proteomics and Metabolomics, Leiden University Medical Center, 2300RC Leiden, Netherlands. g Center for Computational Biology, Leiden University Medical Center, 2300RC, Leiden, The Netherlands. h The Delft Bioinformatics Lab, Delft University of Technology, 2628CD, Delft, The Netherlands. i Section of Molecular Epidemiology, Leiden University Medical Center, 2300RC, Leiden, The Netherlands. * Shared co-first authorship. # Corresponding authors: Gary Siuzdak. Scripps Center for Metabolomics, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, United States. siuzdak@scripps.edu, Martin Giera and Elena Sánchez-López. Center for Proteomics and Metabolomics, Leiden University Medical Center, 2300RC Leiden, Netherlands. M.A.Giera@lumc.nl and E.Sanchez_Lopez@lumc.nl 25 6 2023 4 2023 08 2 2023 24 7 2023 18 4 12961315 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Analytical techniques with high sensitivity and selectivity are essential to the quantitative analysis of clinical samples. Liquid chromatography (LC) coupled to tandem mass spectrometry (MS/MS) is the gold standard in clinical chemistry. However, tandem mass spectrometers come at high capital expenditure and maintenance costs. We recently showed that it is possible to generate very similar results using a much simpler single MS detector by performing enhanced in-source fragmentation/annotation (EISA) combined with correlated ion monitoring. Here we provide a step-by-step protocol for optimizing the analytical conditions for EISA, so anyone properly trained in LC-MS can follow and apply this technique for any given analyte. We exemplify the approach by using 2-hydroxyglutarate (2-HG) which is a clinically relevant metabolite whose D-enantiomer is considered an “oncometabolite”, characteristic to cancers associated with mutated isocitrate dehydrogenases 1 or 2 (IDH1/2). We include procedures for determining quantitative robustness, and show results of these relating to the analysis of DL-2-HG in cells, as well as in serum samples from acute myeloid leukemia patients that contain the IDH1/2 mutation. This EISA-MS protocol is a broadly applicable and low-cost approach for the quantification of small molecules that has been developed to work well for both single quadrupole and time-of-flight mass analyzers. pmcINTRODUCTION Atmospheric pressure ionization (API), especially electrospray ionization (ESI), has significantly enhanced liquid chromatography mass spectrometry (LC-MS) capabilities largely due to its ability to generate intact molecular ions with high sensitivity. LC-API-MS allows the detection of primarily intact molecular ions, and tandem MS (MS/MS) was designed to provide an added dimension of information contained in the fragmentation spectra of each molecular ion. The concept has widely become adapted and is of crucial value in a variety of MS applications, including mixture analysis, molecular structure elucidation and quantitative analysis1–4. The place where fragments are generated is called the collision cell. In the collision cell, ion fragmentation is performed by collision-induced dissociation (CID), a process in which the ion acquires internal energy by collisions with neutral molecules and/or atoms. CID is widely used in both high- and low-resolution MS instrumentation to generate fragment spectra. However, adding additional tandem MS hardware components such as the collision cell comes at the cost of ion losses during transmission in combination with increased instrumentation and maintenance costs, ultimately limiting an even broader application of MS. We recently introduced the concept of enhanced in-source fragmentation annotation (EISA) as a simple and efficient alternative approach to traditional tandem MS fragment ion generation, producing ESI in-source fragments5,6. By optimizing in-source fragmentation conditions, fragments generated using EISA on single quadrupole MS machines can mimic the tandem MS fragments produced via CID without compromising ion transmission. Moreover, EISA-generated spectra are highly comparable with CID fragment spectra and hence can be searched using MS/MS spectral libraries such as METLIN5. However, the most significant advantage that EISA allows is in using the fragment ions to perform quantitative analyses using much simpler (single quadrupole) technology. This approach is termed quantitative multiple fragment ion monitoring (Q-MFM) or alternatively Q-MRM referring to single quadrupole multiple reaction monitoring and comes at a fraction of the cost of traditional triple quadrupole (QqQ) mass spectrometers. In principle, the approach for method development and validation described in this protocol can be applied to any given analyte. As an example for this Q-MFM protocol, we chose the clinically relevant DL-2-hydroxyglutarate (DL-2-HG) enantiomers. The analysis of these enantiomers is useful because of the importance of D-2-HG as an oncometabolite. This can be challenging because enantiomers have the same physicochemical properties and therefore cannot be separated in a standard chromatographic approach. For this, the diastereomers must be firstly formed by means of derivatisation with a enantiomerically pure agent. DEVELOPMENT OF THE PROTOCOL Initially EISA was investigated as an alternative approach for the generation of fragment ions6. As a logical next step the quantitative nature and applicability of EISA was investigated for small molecule analysis and peptide fragmentation in bottom-up proteomics7. The approach is designated as Q-MFM, referring to the use of a single stage MS in combination with a software tool for correlating fragment ions (correlated ion monitoring, CIM) so that it is possible to see results similar to those obtained using a reaction monitoring approach6. Thus far quantitative EISA Q-MFM has been applied to several small molecule metabolites in diverse matrices6 and in peptide quantification7. Overall, the metrics for measuring quantitative soundness for these compounds (dynamic range, matrix effects, trueness and precision) have been adequate6. In principle this approach can be adapted for use with any kind of analyte that is ionizable using electrospray ionization. In this protocol we describe, quantitative Q-MFM analysis of the clinically relevant DL-enantiomers of 2-HG. The ratio between the D- and L-2-HG is clinically meaningful as increased levels of D-2-HG can be used as biomarker for mutated isocitrate dehydrogenases 1 or 2 (IDH1/2)8. For chromatographic separation we applied the method originally developed by Struys et al.9, that transforms the DL-2-HG enantiomers into separable diastereomers by derivatization with diacetyl-L-tartaric anhydride (L-DATAN). Subsequently, we optimized chromatographic conditions as well as Q-MFM-based analysis and validated the assay for several matrices. COMPARISON WITH OTHER METHODS The industry standard for quantitative LC-MS analysis of small molecules is isotope dilution-based MS (IDMS) executed in multiple reaction monitoring (MRM) mode on QqQ instrumentation10. As outlined above, this involves fragmenting molecules by CID in the collision cell. Our here described procedure adopts the IDMS and MRM concept, however all molecules are fragmented within the ion source allowing for the removal of the collision cell from the procedure. The selected ions specific to any particular molecule can provide for quantitative LC-MS methods using single stage MS systems (e.g. quadrupole (Q) or time-of-flight (TOF)) with sensitivities comparable to QqQ instrumentation. The fragmentation of all molecules within the ion source and the removal of an extra mass filter significantly reduces hardware costs, although it limits the selectivity of the method. Nevertheless, modern UHPLC systems and core shell columns allow for superior chromatographic resolutions and advanced separation of analytes11. In turn, Q-MFM coupled to UHPLC is a broadly applicable, cheap alternative for the quantitative analysis of small molecules in complex mixtures. It is important to keep in mind that fragmentation is dependent on the molecule chemical structure. Some molecules might be prone to fragment while others will be more challenging. However, this is not a limitation of EISA, rather a pre-requisite of an analyte chemical nature and it has been shown that EISA and MS/MS fragmentation patterns are very similar5. We have already demonstrated the applicability of EISA in a total of 50 endogenous molecules with a broad range of physicochemical properties and chemical structures, including amino acids, lipids, and fatty acids6. This method worked well for all of these compounds and all of the fragmentation patterns followed those observed in MS/MS. As collision cell is not needed, EISA method enables MRM level quantitative performance by monitoring/correlating precursor and fragment ions on existing single quadrupole instrumentation that are generally inexpensive. EXPERIMENTAL DESIGN A detailed depiction of the PROCEDURE can be found in Figure 1. Experimentally some steps might be modified or carried out in an alternative fashion, however, in our experience the described PROCEDURE is the most practical and facile approach. Using Figure 1 as guideline, we provide a step-by-step description of the entire protocol. Selection of the analyte(s) The first step is to select the metabolite(s) of interest. Once the analytes have been chosen, it is important to determine a suitable internal standard (IS). The IS plays a critical role not only in correcting for possible losses during the sample preparation but also to correct for matrix effects12. The ratio calculated by dividing the response of the targeted analyte and that of the IS can then be interpolated onto a calibration line to calculate the unknown concentration of the analyte in the sample. Ideally the IS should be a stable (heavy) isotope-labelled version of the target analyte (deuterated or 13C labelled). If using the isotopic-labeled variant is not feasible (e.g. elevated cost or not commercially available), a surrogate can be used, i.e. a similarly structured molecule. Importantly, the IS should not be present in the biological sample of interest and the retention time must be close to that of the selected analyte. Optimization of chromatographic separation Although Q-MFM analysis is highly comparable to QqQ-based MRM analysis in terms of sensitivity6, it does lack an additional mass filter, which translates into a lower selectivity when compared to “traditional” QqQ-based MRM analysis. This is the reason why in Q-MFM analysis, chromatographic separation of target analytes from overlapping components is even more important to take into account. In the here described Protocol we present such an optimization strategy for the derivatized DL-enantiomers of 2-HG. Q-MFM optimization Subsequent to the optimization of the chromatographic workflow, the Q-MFM parameters can be optimized taking the eluent composition at analyte specific retention times into account during direct-infusion experiments. As described below, the main goal is the optimization of the in-source fragmentation parameters so that recognizable fragment ions are observed without significantly compromising the intensity of the molecular ion. Several parameters have to be optimized such as the end plate offset, capillary voltage, nebulizer pressure, drying gas and in-source collision induced dissociation (isCID). Linear range assessment Once both the LC and Q-MFM methods have been optimized, the next step is to identify what is the expected concentration range for the monitored analyte(s) of interest and thereafter assess method linearity. The calibration line should contain at least six datapoints obtained from solutions with increasing concentrations of the analyte and identical IS concentration. This way, the linear relationship between the analyte concentration and the IS-normalized peak area of the analyte (areaanalyte/areaIS) can be established. Subsequently, the unknown concentration in the samples can be predicted using the least squares regression method. Since in LC-MS low concentrations are subjected to a higher variability, it is common practice is to use weighted least squares linear regression. Using a 1/x2-type of weighting results in a better accuracy for low concentration datapoints. High correlation coefficients (R2) (e.g. higher than 0.99) mean that the calibration curve approaches linearity. Another way of proving that the calibration curve is linear is to perform appropriate statistical method such as analysis of variance (ANOVA). Validation of the LC-Q-MFM method As for any quantitative LC-MS assay, a thorough validation of the final assay is mandatory. Standard analytical figures of merit include, linearity, limits of detection (LOD), and quantification (LOQ), trueness (the accuracy performance characteristic refers to systematic and random error, being trueness and precision) , recovery and precision. This should be done for each matrix under investigation. Ultimately, validation will prove the applicability of Q-MFM for a specific target analyte in a given matrix. We also performed an interlaboratory comparison with two different single stage MS instruments and found largely overlapping results. MATERIAL REAGENTS Reagents used in the UHPLC-EISA-TOF system LC-MS grade acetonitrile (Honeywell, cat. no. 34967). LC-MS grade methanol (Supelco, cat. no. 1.06035.2500). LC-MS grade 2-propanol (Honeywell, cat. no. 34965). ! CAUTION: Acetonitrile, methanol and 2-propanol are highly flammable. LC-MS grade formic acid ≥ 99% (VWR, cat. no. 84865.180). ! CAUTION: Formic acid is corrosive, flammable, and toxic. LC-MS grade water (Honeywell, cat. no. 14263). n-Hexane (Honeywell, cat. no. 32293). LC-MS calibration standard, for ESI-TOF (Agilent Technologies, cat. no. G1969–85000). LC-MS grade acetic acid ≥ 99% (Sigma Aldrich, cat. no. A6283). LC-MS grade ammonium formate (Fluka, cat. no. 55674). Disodium D-2-hydroxyglutarate (Merck, cat. no. 61382). Disodium L-2-hydroxyglutarate (Merck, cat. no. 90790). Disodium DL-2-hydroxyglutarate-d3,OD (CDN isotopes, cat. no. D-7496). Sodium lactate (Sigma Aldrich, cat. no. L7022). (+)−Diacetyl-L-tartaric anhydride (L-DATAN) (Aldrich, cat. no. 358924). Sodium azide (Merck, cat.no. 7290). Reagents used in the HPLC-EISA-Q system LC-MS grade acetonitrile (Fisher Chemical, cat. no. A955–4). LC-MS grade methanol >99.9% (Honeywell, cat. no. LC230–4). LC-MS grade 2-propanol >99.9% (Honeywell, cat. no. 34965–4X4L). ACS reagent grade Formic acid ≥ 98% (Sigma-Aldrich, cat. no. 33015–500ML). LC-MS grade water (Honeywell, cat. no. 365–4). LC-MS calibration standard, for ESI-TOF (Agilent Technologies, cat. no. G1969–85000). Acetic acid Reagent Plus ≥ 99% (Sigma-Aldrich, cat. no. A6283–500ML). Ammonium formate solution BioUltra 10 mol/L in water (Sigma-Aldrich, cat. no. 78314). Disodium D-2-hydroxyglutarate (Cayman, cat. no. 11605). Disodium L-2-hydroxyglutarate (MilliporeSigma Supelco, cat. no. 61313-10MG). Disodium DL-2-hydroxyglutarate-d3,OD (CDN isotopes, cat. no. D-7496). ACS reagent grade Lactic acid solution ≥ 85% (Sigma-Aldrich, cat. no. 252476–100G). (+)−Diacetyl-L-tartaric anhydride (L-DATAN) (Aldrich, cat. no. 358924). BIOLOGICAL MATERIAL Cell lines The K562 cell line (https://scicrunch.org/resolver/RRID:CVCL_0004, ATCC, Manassas, VA, USA) was cultured in IMDM with 10% heat-inactivated fetal bovine serum (FBS) (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) and 1.5% 200 mmol/L L-glutamine (Lonza). The chondrosarcoma cell lines CH2879 (https://scicrunch.org/resolver/RRID:CVCL_9921, IDH1 and IDH2 wildtype)13,14 and JJ012 (https://scicrunch.org/resolver/RRID:CVCL_D605, IDH1 p.R132G)13,15 were cultured in RPMI medium supplemented with 10% heat-inactivated FBS (F524, Sigma-Aldrich, Sant Louis, MO, USA). Cells were trypsinized, put on ice, and counted with a hemocytometer. Cell pellets of 1 × 106 cells were prepared, washed twice with ice-cold PBS, and stored at −80 °C until DL-2-HG extraction. Heat inactivated fetal calf serum (FCS) was obtained from Bodinco B.V. (Alkmaar, the Netherlands). ! CAUTION: The cell lines used should be regularly checked to ensure they are authentic and are not infected with mycoplasma Biological fluids Urine from a healthy male volunteer was collected as first urine of the day. To prevent bacterial growth 0.1% (vol/vol) of sodium azide in water was added upon collection. Serum samples obtained from patients at diagnosis of acute myeloid leukemia (AML) with written informed consent were selected from the Hematology Biobank of Leiden University Medical Center with approval by the institutional review board (no. B18.047). In both patients, the IDH1-R132H mutation was detected in diagnostic cell samples by RNA-Seq16. Human plasma was acquired from Sigma-Aldrich (cat. No. P9523–5ML). Identical material was used for Q-MFM analysis on TOF and single quadrupole MS systems. EQUIPMENT Equipment used in the UHPLC-EISA-TOF system Aqua C30 column (3 μm; 2 mm inner diameter (i.d.) × 150 mm length (Develosil, cat. no. RPAR320150W). HPLC guard holder for 2.1 to 4.6 mm ID columns (Phenomenex, cat. no. AJ0–9000). C8 2.1 mm ID precolumn (Phenomenex, cat. no. AJ0–8784). Dionex UltiMate 3000 UHPLC (Thermo Fisher) or generic HPLC. Bruker Impact II Q-TOF system (Bruker). Stuart Scientific Block heater (for the derivatization step; Fisher scientific). Mini Spin centrifuge (Eppendorf). Fisherbrand vortex mixer (Fisher Scientific). Centrifuge 1.5–2 mL Eppendorf tubes (Eppendorf). Heat block. pH meter (Mettler Toledo). Ultrasonic bath. Magnetic stir bar. Polypropylene microcentrifuge tubes (1.5 mL; Eppendorf, cat. no. 0030 123.328). Polypropylene microcentrifuge tubes (2.0 mL; Eppendorf, cat. no. 0030 120.094). HPLC vials with screw (Agilent, cat. no. 5182–0714). HPLC cap vials (Agilent, cat. no. 5182–0717). 250 μL inserts with polymer feet (Agilent, cat. no. 5181–1270). Solid phase extraction system (Waters). C18 Cartridges (Waters, cat. no. WAT054945). 500 μL glass syringe (Hamilton, cat. no. 81265). Equipment used in the HPLC-EISA-Q system InfinityLab Poroshell 120 EC-C18 column, 4.6 × 50 mm, 2.7 μm (Agilent, cat.no. 699975–902T). Agilent 6130 Single Quadrupole LC-MS. Agilent HPLC 1260 Infinity. pH meter (Mettler Toledo). Thermomixer R (Eppendorf). Vortex mixer (Thermo Scientific). Centrifuge 1.5–2 mL Eppendorf tubes (Eppendorf). Refrigerated CentriVap Concentrator (Labconco). Safe-Lock Eppendorf tubes, 1.5 mL (Eppendorf, cat. no. 022363204). Safe-Lock Eppendorf tubes, 2.0 mL (Eppendorf, cat. no. 022363344). HPLC glass vials with screw (Thermo Fisher, cat. no. 4000-S1W). HPLC vial caps (Agilent, cat. no. 5182–0717). Vial inserts, 250 μL, deactivated glass with polymer feet (Agilent, cat. no. 5181–8872). SOFTWARE Software used in the UHPLC-EISA-TOF system UPLC system was controlled by Chromeleon version 6.80 (Dionex, USA). Bruker Impact II Q-TOF system was controlled by otof Control version 3.4 (Bruker Daltonik GmbH, Germany). LC-MS hyphenation was managed by Compass HyStar (Bruker Daltonik GmbH, Germany). Peak integration was conducted via Compass DataAnalysis version 4.2 (Bruker Daltonik GmbH, Germany). Software used in the HPLC-EISA-Q system Agilent ChemStation Rev. C01.10. Agilent MassHunter Quantitative Analysis Version B.08.00. Correlated ion monitoring (CIM) algorithm. EQUIPMENT SETUP CIM algorithm. This algorithm was developed to perform peak picking, alignment and data analysis and can be accessed via https://github.com/ricoderks/eisaCIM (see also ref. 6). Briefly, chromatograms for each individual metabolite are constructed by using each SIM trace (for single Q analyzers) using XCMS (v3.12.0) R-package with an established intensity threshold. This is a crucial step as to obtain clean signals that do not include co-eluting peaks. Finally, SIM traces containing precursor and fragment ions can be aligned and grouped as a CIM chromatogram based on a given retention time window. REAGENT SETUP Cells. For cultured cells, use at least 1 × 106 cells. Biological fluids. For biological fluid samples, start with aliquots of 10 μL of fetal calf serum, human serum or plasma or 20 μL of human urine in 1.5-mL polypropylene microcentrifuge tubes. Methanol (80% (vol/vol)). To prepare a volume of 100 mL of a 80% (vol/vol) methanol solution for metabolite extraction and protein precipitation, mix 80 mL of LC-MS grade methanol and 20 mL of LC-MS grade water in a glass jar and gently shake for mixing. Store at −20 °C until further use. 2-propanol (50% (vol/vol)). To prepare a volume of 500 mL of a 50% (vol/vol) 2-propanol solution to clean the needle of the UHPLC system between injections, mix 250 mL of LC-MS grade 2-propanol and 250 mL of LC-MS grade water in a HPLC bottle. Cap and gently shake for mixing. Slightly open the bottle and sonicate for 10 min to eliminate oxygen. The solution can be stored for three weeks at room temperature (~ 20–25 °C). Formic Acid (50% (vol/vol)). To prepare a volume of 1 mL of a 50% (vol/vol) formic acid solution to adjust the pH buffer solution A, mix 500 μL of LC-MS grade formic acid and 500 μL of LC-MS grade water in a 1.5-mL polypropylene microcentrifuge tube. Cap and gently shake for mixing. The solution can be stored for three weeks at room temperature (~ 20–25 °C). Acetonitrile:acetic acid (4:1 (vol/vol)). To prepare a volume of 1 mL of 4:1 (vol/vol) acetonitrile:acetic acid, to dissolve the L-DATAN, mix 800 μL of LC-MS grade acetonitrile and 200 μL of acetic acid in a 1.5-mL polypropylene microcentrifuge tube. Cap and gently shake for mixing. The solution can be stored for three weeks at room temperature (~ 20–25 °C). LC-MS grade water:acetic acid (4:1 (vol/vol)). To prepare a volume of 1 mL of a 4:1 (vol/vol) LC-MS grade water:acetic acid to reconstitute the dry pellet prior LC-MS analysis, mix 800 μL of LC-MS grade water and 200 μL of acetic acid in a 1.5-mL polypropylene microcentrifuge tube. Cap and gently shake for mixing. The solution can be stored for three weeks at room temperature (~ 20–25 °C). L-DATAN solution (50 mg/mL in (4:1 (vol/vol) acetonitrile:acetic acid). To prepare 1 mL, add 50 mg of L-DATAN to a 1.5-mL polypropylene microcentrifuge tube and add 1 mL of (4:1 (vol/vol)) acetonitrile:acetic acid. Cap and gently shake for mixing. The solution must be prepared fresh directly before use and cannot be stored. Sodium L-lactate (10 mmol/L in water). To prepare 2 mL, add 2.2 mg of sodium lactate to a 2.0-mL polypropylene microcentrifuge tube and add 2 mL of LC-MS grade water. Cap and gently shake for mixing. The solution can be stored for at least six months at −20 °C. D-2-hydroxyglutarate (D-2-HG) (50 mmol/L in water). To prepare 1 mL, add 9.6 mg of D-2-HG disodium salt to a 1.5-mL polypropylene microcentrifuge tube and add 1 mL of LC-MS grade water. Cap and gently shake for mixing. The solution can be stored for at least six months at −20 °C. L-2-hydroxyglutarate (L-2-HG) (50 mmol/L in water). To prepare 1 mL, add 9.6 mg of L-2-HG disodium salt to a 1.5-mL polypropylene microcentrifuge tube and add 1 mL of LC-MS grade water. Cap and gently shake for mixing. The solution can be stored for at least six months at −20 °C. DL-2-HG (2 mmol/L in water). To prepare 1 mL, mix 20 μL of the 50 mmol/L D-2-HG solution, 20 μL of the 50 mmol/L L-2-HG solution and 960 μL of LC-MS grade water in a 1.5-mL polypropylene microcentrifuge tube. Cap and gently shake for mixing. The solution can be stored for at least six months at −20 °C. Please, note that the individual concentration per enantiomer is 1 mmol/L. DL-2-HG (200 μmol/L in water). To prepare 1 mL, mix 100 μL of the 2 mmol/L DL-2-HG solution and 900 μL of LC-MS grade water in a 1.5-mL polypropylene microcentrifuge tube. Cap and gently shake for mixing. The solution can be stored for at least six months at −20 °C. Please, note that the individual concentration per enantiomer is 100 μmol/L. DL-2-HG (2 μmol/L in water). To prepare 1 mL, mix 10 μL of 200 μmol/L DL-2-HG solution and 990 μL of LC-MS grade water in a 1.5-mL polypropylene microcentrifuge tube. Cap and gently shake for mixing. The solution can be stored for at least six months at −20 °C. Please, note that the individual concentration per enantiomer is 1 μmol/L. DL-2-HG-d3 (IS) (10 mmol/L in water). To prepare 1 mL, add 2.0 mg of disodium DL-2-HG-d3 to a 1.5-mL microcentrifuge tube and add 1 mL of LC-MS grade water. Cap and gently shake for mixing. The solution can be stored for at least six months at −20°C. Please, note that the individual concentration per enantiomer is 5 mmol/L. DL-2-HG-d3 (IS) (1 mmol/L). To prepare 1 mL, mix 100 μL of 10 mmol/L DL-2-HG-d3 solution and 900 μL of LC-MS grade water in a 1.5-mL polypropylene microcentrifuge tube. Cap and gently shake. The solution can be stored for at least six months at −20 °C. Please, note that the individual concentration per enantiomer is 0.5 mmol/L. LC mobile phase A (125 mg/L of ammonium formate in water, pH 3.6). To prepare LC mobile phase A, add 125 mg of ammonium formate to a 1-liter HPLC bottle and add 990 mL of LC-MS grade water. Shake the solution using a magnetic stirrer and add 250 μL of a 50% (vol/vol) formic acid solution. Check the pH to assure it is at 3.6 and complete the volume up to 1 L with LC-MS grade water. Sonicate for 10 min to degas the solution. The solution can be stored for three weeks at room temperature (~ 20–25 °C). LC mobile phase B (95% (vol/vol) acetonitrile in water). To prepare LC mobile phase B, mix 950 mL of acetonitrile and add 50 mL of LC-MS grade water to a 1-liter HPLC bottle. Cap and shake the solution. Sonicate for 10 min to degas the solution. The solution can be stored for three weeks at room temperature (~ 20–25 °C). PROCEDURE The overall objective of this protocol is to show a practical guide to develop a LC-Q-MFM strategy for quantitative purposes (see Figure 1). We exemplify this process by showing results for the quantification of DL-2-HG (see Figure 2). Select the target molecule/s. Prepare the standard solutions. Develop the chromatographic method. The development of the LC method for DL-2-HG is out of scope for this protocol but we have included a detailed description in the Supplementary Material (see also Supplementary Figures 1-4). <CRITICAL STEP> It is advisable that the optimization of the EISA method is conducted once a satisfactory chromatographical separation has been obtained. This way, the infused standard solution can exactly resemble subsequent mobile phase composition. (optional) Decide whether derivatisation is needed. In Box 1, we provide steps for derivatisation with L-DATAN as an example. However this is highly dependent on the analyte of interest. Importantly, although we exemplify this protocol for a chiral molecule, researchers interested in incorporating the EISA strategy to their laboratories for non-chiral analysis can skip the steps included in Box 1. The Supplementary Material also includes a more detailed description of the EISA optimization for our particular example (DL-2-HG). Optimization of the EISA method. Timing: 6 h. <CRITICAL> The procedure requires that the EISA-MS parameters are optimized for in-source fragmentation. Similar to tandem MS/MS method optimisation, different parameters have to be fine-tuned to reach the optimum values for Q-MRM experiments. The particular goal of the EISA optimization is to get as many high abundant fragment ions as possible without significantly compromising the intensity of the molecular ion (precursor). To speed up the process, we advise optimising most of these parameters using direct infusion. <CRITICAL> When multiple analytes are to be determined each optimization should be done separately. The optimum condition is one where all analytes have some in-source fragmentation while maintaining the highest intensity of both precursor and product ions. Asses the following parameters using direct infusion: End plate offset: from 0 to 2000 V. Capillary: from 3000 to 5000 V. Nebulizer: from 1 to 2.4 bar. Dry Gas: from 1 to 10 L/min. <CRITICAL STEP> Any parameters related to voltage and heat in the ion source are normally associated with the in-source fragmentation of analytes. Selection of the parameters are largely dependent on the research purpose and are instrument-dependent so instrument vendor recommendations should be checked to know lower and upper limits. Our optimization was conducted on a Bruker Impact II system. <CRITICAL STEP> In our lab, we performed step-wise optimization: we optimized the first parameter, e.g. end plate offset, we selected the optimum value (i.e. 10 V) and then continued with the next parameter using this optimized value. There are other ways that this could be done, you might consider using “design of experiments” to find the optimum value among the multiple parameters assayed here. To choose the optimum value per condition, look for the value that enables the highest intensity of the precursor ion with adequate fragmentation. Create a plot that shows the intensity of the precursor and the product ions. It is also helpful to represent the sum of the intensities for both ions. Figure 3 shows the different intensity points obtained for each of the parameters assayed. From these data we decided on the following parameters: End plate offset: 10 V Capillary: 4000 V Nebulizer: 1.6 bar Dry Gas: 6 L/min <CRITICAL STEP> Although in panel Figure 3d we found that 1 L/min was the optimum value the ionization was too low to be stable, thus, we selected 6 L/min as it also gave high intensity of precursor + product. (optional) Optimize the in-source collision induced dissociation parameters for the analytes in the conditions they are in after LC separation. Under optimum LC-MS conditions assess the isCID voltage between 0 and 80 eV. To do this, prepare your standard mix as described in Step XX and inject it at least twice per isCID voltage assayed. Extract the EIC of the product and precursor ions and integrate the resulting peaks. In the example shown in Figure 4, these are m/z 147 and m/z 363 respectively. Determine the S/N ratio using the given software. Plot the S/N of each analyte per assayed isCID condition. To choose the optimum value, select the one with the highest S/N (see Figure 4). <CRITICAL STEP> In many ESI sources, it is possible to select a parameter which is exclusively designed for controlling the in-source fragmentation energy. This is vendor dependent, e.g., isCID energy in Bruker mass spectrometers and fragmentor voltage in the Agilent mass spectrometers. In our case we could acquire fragment ions until 40 eV, although the S/N at that isCID was compromised. Complete fragmentation of the precursor ion was achieved at higher isCID energies. <CRITICAL STEP> If more than one product ion is obtained, under optimum LC-MS conditions you can always evaluate what combination of precursor + product/s ions gives the highest S/N. For the example we have selected, the precursor (m/z 363) + just one product (m/z 147) ion gave the best S/N. <CRITICAL STEP> When using a single quadrupole mass analyzer it is possible to work in so-called single ion monitoring (SIM) so that each ion (precursor and product) becomes independent channels. Each SIM channel has their own settings (including some parameters controlling the in-source fragmentation depending on the vendor), which can be optimized exclusively for the performance of the target ion. However, these ion channels share many parameters in the ion source, which need to be optimized by taking all target ions into account (see Note 6 of Box 1). By means of the CIM algorithm, it is possible to reconstruct these ion channels into a single molecular trace (https://github.com/ricoderks/eisaCIM (see also ref. 6). The CIM algorithm is designed to align and selectively compile multiple ions within one chromatogram by analyzing the SIM traces within a pre-specified RT window from mzxml files, including peak picking, alignment and data analysis. A CIM chromatogram can be created as a compilation of the individual ion signals only if each signal satisfies pre-set criteria. Preparation of an internal standard solution for calibration. Timing: 30 min. <CRITICAL> Since the overall aim is to quantify the selected metabolite (in this example, DL-2-HG) in samples of clinical interest through a simplified strategy, it is important to evaluate the dynamic range and linearity of the method. It is also important to incorporate an IS. In this particular case, we used the deuterated (d3) version of DL-2-HG. To prepare the calibration line, design a pipetting scheme like that shown in Table 1. Each solution must include the same amount of IS (for area correction), and any supporting reagents required for derivatization (e.g. sodium lactate (to facilitate the derivatization of low concentrations of DL-2-HG17 and methanol (to aid in the drying step prior derivatization)). In order to calculate the actual standard concentration, take into account the final reconstitution volume used for your samples. In our case this is 100 μL. Preparation of biological samples: DL-2-HG extraction The extraction of DL-2-HG from the different biological samples is the first step in the analysis and quantification process. Here it is important to take into account the type of sample and the needed solvent to achieve a successful extraction that will lead to protein precipitation, necessary before any LC analysis. To carry out the extraction from cells follow option A (see Figure 5a). Here we exemplify the sample preparation by using myelogenous leukemia cells (K562) and chondrosarcoma cells (CH2879-IDH1wt and JJ012-IDH1R132G), but it can be used for any other mammalian cell line. To conduct the extraction from biofluids (fetal calf serum, human serum, plasma and urine) follow option B (see Figure 5b). Make sure to keep samples on ice at all times to prevent sample degradation and minimize residual enzymatic activity. (A) Extraction from cell pellets. Timing: 1.5 h. Add 200 μL of ice-cold (−20 °C) 80% (vol/vol) methanol to a frozen cell pellet in a 1.5 or 2.0 mL-polypropylene microcentrifuge tube. We typically expect the cell pellet to contain ~ 1 × 106 cells. This number could be optimized depending on the sensitivity of the method and the expected concentration of the desired analytes. Add 2 μL of IS stock solution (1 mmol/L DL-2-HG-d3). CRITICAL STEP. You must ensure that the IS solution is added at this point to account for errors during sample preparation and analysis. (optional) If needed during method validation, add at this step the desired concentration of DL-2-HG standard according to the validation of the method section. Vortex for 15 s and sonicate for 1 min at room temperature. Keep at −20 °C for 30 min to allow for protein precipitation. Centrifuge at 16,000g for 10 min at 4 °C. Transfer the supernatant to a new 1.5 mL-polypropylene microcentrifuge tube. Note: make sure the insoluble pellet is not taken. Add an additional 100 μL of ice-cold 80% (vol/vol) methanol to the insoluble pellet to ensure everything is extracted. Centrifuge at 16,000g for 10 min, 4 °C. Transfer the supernatant to the previous 1.5 mL-polypropylene microcentrifuge tube. If you are performing derivatisation, then perform this reaction immediately (follow all steps in this Protocol) if you are not performing derivatisation, skip Steps 12–17. If needed, keep the insoluble pellet for bicinchoninic acid (BCA) protein quantification for normalization purposes. For details on protein analysis, refer to the manufacturer’s instructions. <PAUSE POINT> If needed samples can be stored at −20 °C for maximum one week. (B) Extraction from biofluids: fetal calf serum, human serum, plasma and urine. Timing: 1 h. Add 9 volumes of ice-cold LC-MS grade methanol (cooled to −20 °C) to 1 volume of biofluid. In our case, we added 90 μL of methanol to 10 μL of fetal calf serum, human serum or plasma and 180 μL of methanol to 20 μL of human urine in a 1.5 or 2 mL-polypropylene microcentrifuge tube. Add 2 μL of IS stock solution (1 mmol/L DL-2-HG-d3). (optional) If needed during method validation, add at this step the desired concentration of DL-2-HG standard according to the validation of the method section. Vortex for 15 s at room temperature. Keep at −20 °C for 30 min to allow for protein precipitation. Centrifuge at 16,000g for 10 min at 4 °C. Transfer the supernatant to a new 1.5 mL-polypropylene microcentrifuge tube. Take care to not transfer any of the insoluble pellet. If you are performing derivatisation, then perform this reaction immediately (follow all steps in this Protocol) if you are not performing derivatisation, skip Steps 12–17. (optional) Derivatization of the calibration standards and extractant samples e.g. with diacetyl-L-tartaric anhydride (L-DATAN). Timing: 4 h. <CRITICAL> In order to carry out the chiral separation of DL-2-HG, derivatization with L-DATAN is necessary to obtain the corresponding diastereomers that will be then separated on a standard reverse phase column (see Figure 6). This section can be left out if your method does not require derivatisation of the analyte compound. It can also be replaced with the derivatisation method that is most appropriate for your analyte; much of the advice we have included here will be relevant to other derivatisation methods. Evaporate samples until total dryness using a SpeedVac for 45 min at 60 °C. <PAUSE POINT> If needed samples can be stored at this point for later analysis. Dried samples could be kept at −20 °C if analyzed in the same week or at −80 °C if analysis cannot be initiated in a week. However we recommend to prepare and analyze the samples as soon as possible. Add 100 μL of LC-MS grade methanol to the microcentrifuge tubes to help any remaining droplets of water to evaporate. Dry in the SpeedVac for an additional 20 min at 60 °C. CRITICAL STEP. You must ensure that all water is evaporated and that the sample has completely dried, otherwise the derivatization will not take place. Prepare a fresh solution of 50 mg/mL L-DATAN in 4:1 (vol/vol) acetonitrile:acetic acid. Prepare immediately before use to avoid unnecessary evaporation. Add 50 μL of the L-DATAN solution. Vortex for 15 s. Heat at 80 °C in a heat block for 2 h. Depending on the concentration, samples may turn into a dark yellow or brownish color. CRITICAL STEP. Make sure that the polypropylene microcentrifuge tubes are tightly closed. The heat will generate vapors inside the tubes. Keep at room temperature for 2 min and spin-down for 15 s at room temperature. This will enable that all the content returns to the bottom of the polypropylene microcentrifuge tube and will not be lost. Dry samples using a SpeedVac for 60 min until completely dry. Redissolve the dried residue in 100 μL of a 4:1 (vol/vol) LC-MS grade water:acetic acid solution and vortex. CRITICAL STEP. Make sure that the dried pellet resulting from the drying step is completely redissolved. This might need vortexing for several minutes. If a turbid solution is observed, centrifuge at 16,000g for 10 min at room temperature. Transfer 100 μL in a HPLC vial with insert for LC-MS analysis. Preparing the UHPLC-MS for sample analysis and data acquisition. Timing: 10 min (calibration of the MS instrument) + 10 min (setting the method parameters) + 10 min (conditioning of the column) + 13 min (analysis time per sample). Calibrate the MS instrument according to the manufacturer’s instructions using a Tuning Mix solution. Create an instrument method in MS scan mode following the instructions of the acquisition software. Work in negative ion mode, use a mass range from m/z 30 to 370 and a spectra rate of 1.00 Hz. Set up the ion source conditions. These are the optimized values obtained by performing Steps 5–7. In our case these values are included in the Supplementary Material’s “EISA optimization” section. The rest of the MS-related parameters are also included in this section. Before starting the analytical run, condition the Aqua C30 column with the initial gradient conditions for 10 min until reaching a stable pressure, in our case, 100% mobile phase A at 0.4 mL/min resulted in a backpressure of ~ 260 bar. Set up the parameters and the gradient for the UHPLC system. In our case, these parameters and the gradient can be found in the Supplementary Information “Optimization of the LC conditions” section. Include necessary blanks at the beginning, at the end and during the run to monitor sample carryover. When the run has finished, wash the column following manufacturer’s recommendations. In our case we washed the Aqua C30 column with acetonitrile for 10 min, LC-MS grade water for 10 min and 60% (vol/vol) acetonitrile:water for 10 min at a flow of 0.4 mL/min. Evaluating the analytical characteristics of the method <CRITICAL> To demonstrate the suitability of the method for quantification purposes, its analytical characteristics must be validated. In this example we assess its fit-for-purpose according to linearity, LOD, LOQ, trueness (accuracy), recovery, and precision. Linearity. Timing: 5 h (analysis time) + 2 h (data analysis). Construct a calibration line for each of your chosen analytes (e.g. using the information in Table 1 where the concentration range is 0.5–100 μmol/L for each enantiomer). Refer to the Step 8. Inject the nine calibration line solutions and analyse these using the optimized EISA method. Refer to the section: “Preparing the UHPLC-MS for sample analysis and data acquisition”. Integrate the peak areas of both the external and the internal standard (e.g. for both DL-2-HG and deuterated DL-2-HG). For each enantiomer, plot the area ratio versus the nine calibration line. A weighted linear regression is often recommended due to the fact that variance over the entire concentration range is not equal, rather the variance increases as concentration increases. For this reason, 1/x2 is the most commonly used weighting factor that improves accuracy in quantifying low concentrations standards. To assess the linearity, calculate the R2 of the regression model which should be as high as possible (ideally > 0.99). As recommended by the United States Food and Drug Administration (FDA), the linearity should also be evaluated by appropriate statistical methods such as ANOVA18. Determination of the LOD. Timing: 5 min. Once the linearity is assessed, determine the LOD as 3.3 × (standard deviation of the intercept of the calibration line) / (slope of the calibration line). Determination of the LOQ. Timing: 15 min. Once the linearity is assessed, determine the LOQ as the lowest concentration from the calibration line that can be quantified with acceptable accuracy (between 70–130 %) and precision (coefficient of variation <30%). Trueness. Timing: 12.5 h (analysis time) + 2 h (data analysis). <CRITICAL> Here we asses trueness expressed in terms of “bias” by recovery experiments using spiked samples. In this context, spiked samples are biological samples to which different concentrations of external standard have been added. The external standard is added together with the IS before sample extraction (and before the derivatization when needed). The samples themselves may also contain some of the compound of interest. Thus it is important having a control sample (without adding the external standard) functioning as reference of the endogenous concentration. Considering that a sample with no detectable endogenous analyte was used, the analysis of a spiked sample should reveal a quantified concentration equal to the concentration used (within a 80–120% range). Prepare a calibration series (see e.g. Table 1) for each analyte of interest. Refer to the Step 8. Prepare the different biological samples according to section “Preparation of biological samples: DL-2-HG extraction” and on step iii) add the required volume of the external standard stock solution to assess the trueness at four different levels. For the DL-2-HG example these would be: 0 (non-spiked), 10, 50 and 70 μmol/L (see Table 2 for pipetting scheme). Inject both the calibration line solutions and the samples (spiked or non-spiked) using the EISA method as detailed in “Preparing the UHPLC-MS for sample analysis and data acquisition”. Integrate the peak areas of both the external and the internal standard (in this example, DL-2-HG and the IS). Using the calibration line to calculate the concentration of each analyte of interest in each spiked sample as well as in the non-spiked sample. Evaluate the trueness of the method by subtracting the amount of the analyte of interest (e.g. DL-2-HG) in the non-spiked solutions from the amount detected in the three different spiked levels (10, 50 and 70 μmol/L) (this needs to be done if there were endogenous levels of the compound in the sample). Divide the resulting concentration by the value of the known added concentration and multiply by 100. The values should be in the 80–120% range. Recovery. Timing: 6.5 h (analysis time) + 1 h (data analysis). Prepare the different biological samples according to section “Preparation of biological samples: DL-2-HG extraction”, but for each experimental replicate create two samples: one where the external standard is added either before and one where it is added after the extraction step. In the example shown, add the necessary volume of the stock solution to achieve a final 10 μmol/L per enantiomer (for a detailed pipetting scheme follow the same procedure as in the accuracy, corresponding to the 10 μmol/L column, see Table 2). Inject both the spiked samples using the EISA method as detailed in “Preparing the UHPLC-MS for sample analysis and data acquisition”. Integrate the peak areas of both the external and the internal standard (e.g. DL-2-HG and the IS). To calculate the recovery, use the external standard area ratio obtained for the sample where the standard was added before extraction and divide this value by the resulting area ratio obtained after extraction and multiply by 100. The resulting value, should be in the 80–120 % range. Precision. Timing: 7.5 h over a span of 3 days (2.5 h each day) + 1.5 h (data analysis). Prepare the different biological samples as described in the section section “Preparation of biological samples: DL-2-HG extraction” and add the required volume of external standard stock solution (for this example use the necessary volume of the stock solution so that that final concentration is 10 μmol/L per enantiomer). Follow the same procedure as in the accuracy study, corresponding to the 10 μmol/L column (see Table 2). The corresponding samples should be prepared in triplicate during three consecutive days to assess the precision. Inject both the spiked samples using the EISA method as described in “Preparing the UHPLC-MS for sample analysis and data acquisition”. Integrate the peak areas of both the external and the internal standard (e.g. DL-2-HG and the IS). To calculate the between-injection repeatability, use the area ratio for the external standard obtained in three consecutive injections of the same replicate and calculate the relative standard deviation (RSD) (%). To calculate the measurement repeatability, use the area ratio for the external standard obtained in three replicates originating from the same sample, each injected in triplicate in the same day and calculate the RSD (%). To calculate the intermediate measurement precision, use the area ratio for the external standard obtained in three replicates originating from the same sample, injected in triplicate on three consecutive days and calculate the RSD (%). ? TROUBLESHOOTING For a detailed troubleshooting see Table 3. TIMING Step 5–7, Optimization of an EISA method: 6 h. Step 8, Preparation of an internal standard solution for calibration line: 30 min. Step 9A, Extraction from cells pellets: 1.5 h. Step 9B, Extraction from biofluids: fetal calf serum, human serum, plasma and urine: 1 h. (optional) Step 10–20, Derivatization of the calibration standards and extractant samples e.g. with diacetyl-L-tartaric anhydride (L-DATAN). Timing: 4 h. Step 21–26, Preparing the UHPLC-MS for sample analysis and data acquisition: 10 min (calibration of the MS instrument) + 10 min (setting the method parameters) + 10 min (conditioning of the column) + 13 min (analysis time per sample). Step 27–30, Linearity. 5 h (analysis time) + 2 h (data analysis). Step 31, Determination of the LOD: 5 min. Step 32, Determination of the LOQ: 15 min. Step 33–37, Trueness: 12.5 h (analysis time) + 2 h (data analysis). Step 38–40, Recovery: 6.5 h (analysis time) + 1 h (data analysis). Step 41–45, Precision: 7.5 h in a span of 3 days (2.5 h each day) + 1.5 h (data analysis). ANTICIPATED RESULTS EISA takes advantage fragment ions produced in the ion source; these fragment ions are similar to those generated by tandem MS/MS. Among the benefits of EISA we can highlight: Selective and sensitive results using a single mass analyzer, allowing for broad applicability at low cost. Fragment ions can be produced over a wide dynamic range of ion intensities. Precursor ion remains at high intensity when producing recognizable fragments, therefore, increasing the sensitivity. Robust identification of a wide range of small molecules. The purpose of this protocol is to show the development of an EISA methodology for quantification purposes. Here we exemplified this by showing the quantitative analysis of DL-2-HG in different biological matrices. The results of the validation for each sample are illustrated in Table 4. It demonstrates that this protocol allows for the accurate and precise quantification of DL-2-HG in all cases obtaining a good recovery. Moreover, the LOD calculated was 0.08 and 0.09 μmol/L for L-2-HG and D-2-HG, respectively. Here we want to provide readers with a brief discussion of the anticipated results of our Q-MFM assessed in the clinical setting. Assessing the reproducibility of Q-MFM: benchmarking the EISA-TOF against an EISA-Q method To further assess the reproducibility of the Q-MFM method we transferred our DL-2-HG approach to a single quadrupole system located in a different laboratory. The Pearson correlation between the area ratios from the calibration line (Table 1) of both enantiomers in the Q-MFM methods (TOF and Q) revealed a high level of agreement between the two instruments (Supplementary Figure 5). In both laboratories, the method was validated in terms of trueness (accuracy), recovery and precision using the same commercially available human plasma (Table 5). As can be seen, both strategies were highly comparable, which highlights the high reproducibility of this approach even when performed at two different locations, with two different instruments, different vendors and handled by different laboratory staff. Assessing the applicability of Q-MFM in the diagnosis of IDH1 mutation in biological specimens: analysis of cell lines, and serum from acute myeloid leukemia patients To further demonstrate the applicability of the Q-MFM in a clinical setup, we analyzed several biological specimens suspected to have endogenous D-2-HG. As expected, JJ012 chondrosarcoma cells carrying a mutation in the IDH1 gene accumulated high amounts of the D-enantiomer (62 ± 0.8 μmol/L) whereas the L-counterpart was not detected (Figure 7a). On the other hand, no D-2-HG was detected in CH2879 chondrosarcoma cells carrying wildtype IDH1 and IDH2 genes (Figure 7b). Serum samples from two different AML patients carrying IDH1 mutations also showed presence of D-2-HG: 2.6 ± 0.2 and 12.0 ± 0.8 μmol/L (Figure 7c) in a volume of 20 μL whereas L-2-HG levels were below the LOD. On the contrary, serum from a healthy donor did not contain neither D- nor L-2-HG (Figure 7d). This is a clear demonstration that the developed method based on Q-MFM is able to quantify D-2-HG and correlate it to deficiencies in IDH1, being a potential tool for clinical practice and diagnosis. Assessing the applicability of the Q-MFM in the quantification of D- and L-2-HG in urine Urine samples of a healthy donor were screened for DL-2-HG analysis and results indicated that using a volume of 20 μL both enantiomers could be quantified at near 1-to-1 ratio, namely, 9.7 ± 0.2 and 6.9 ± 0.1 μmol/L for D- and L-2-HG, respectively (see Figure 7e). It is known that both enantiomers are present at low levels in urine samples from healthy individuals as previously reported9. A striking accumulation of any of these would be indicative of D- and/or L-2-hydroxyglutaric acidurias, a hallmark of encephalopathies and psychomotor retardation19. Supplementary Material Supplementary Material Acknowledgements: This work was supported by the Program for Guangdong Introducing Innovative and Entrepreneurial Teams (2019ZT08L213), the Guangdong Provincial Key Laboratory Project (2019B121203011), Key-Area Research and Development Program of Guangdong Province (2020B1111380003), the Leiden Center for Computational Oncology (LCCO), the Spanish Ministry of Science and Innovation (PID2019-104913GB-I00 project) and the Spanish Ministry of Economy and Competitiveness (S.B.B.’s predoctoral research contract (BES-2017-082458)). The authors are grateful to A. Llombart-Bosch (University of Valencia, Spain) for the CH2879 cell line and J.A. Block (Rush University Medical Centre, Chicago, IL, USA) for the JJ012 cell line. Authors would also like to thank Ieva Palubeckaitė and Judith Bovée (Department of Pathology, Leiden University Medical Center, the Netherlands) for providing chondrosarcoma cells and valuable discussion, Hans Dalebout and René van Zeijl (Center for Proteomics and Metabolomics, Leiden University Medical Center, the Netherlands) for their technical assistance and Peter van Balen (Department of Hematology, Leiden University Medical Center, The Netherlands) for providing the serum samples. Figure 1. Workflow to optimize an EISA method for the analysis of clinical samples. This figure has been created with BioRender.com. Figure 2. UHPLC-EISA-TOF separation of DL-2-HG (black) and its d3-labelled IS (blue) derivatized with L-DATAN and resulting MS spectra of L- and D-enantiomers at 3.5 and 3.7 min, respectively. Figure 3. Optimized parameters of the EISA method: a) End plate offset; b) Capillary voltage; c) Nebulizer pressure; d) Dry gas. All parameters were optimized by direct infusion as detailed in Box 1. The optimum values are the ones indicated by the arrow. Figure 4. Signal-to-noise ratio (S/N) of each 2-HG enantiomer (20 μM) per assessed isCID value. Duplicate analyses are shown. Figure 5. Extraction protocol for the analysis of DL-2-HG from cell pellets or biofluids. This figure has been created with BioRender.com. Figure 6. Derivatization reaction scheme and sample protocol to achieve the derivatization of DL-2-HG with L-DATAN. This figure has been created with BioRender.com. Figure 7. UHPLC-EISA-TOF extracted ion chromatograms of different biological samples derivatized with L-DATAN: (a) chondrosarcoma IDH1 mutant cell line; (b) chondrosarcoma IDH1 wild type cell line; (c) serum from an AML patient carrying a IDH1 mutation; (d) serum from a healthy donor; (e) Urine from a healthy donor. In all cases, the black trace corresponds to L-DATAN derivatized D- and/or L-2-HG (m/z 363.0570 + 147.0270) if present endogenously in the samples. The blue trace corresponds to the L-DATAN derivatized internal standard (IS) DL-2-HG-d3 (m/z 366.0745 + 150.0491). Both were extracted with a m/z ± 0.01 window. Table 1. Pipetting scheme to prepare the calibration line of the DL-2-HG for the EISA method. Calibration points 1 2 3 4 5 6 7 8 9 Concentration (μmol/L) per DL-2-HG enantiomer 0.5 1.25 2.5 5 7.5 10 20 50 100 Concentration (μmol/L) of the stock solutions used for spikinga 2 200 200 200 200 200 200 200 200 Volume (μL) of stock solution addedb 50 1.25 2.5 5 7.5 10 20 50 100 Volume (μL) of 1 mmol/L IS solution addeda,b 2 2 2 2 2 2 2 2 2 Volume (μL) of 10 mmol/L sodium lactate solution addedb 2 2 2 2 2 2 2 2 2 Volume (μL) of methanol added 200 200 200 200 200 200 200 200 200 a The concentration depicted here is the sum of both enantiomers. b In case volumes lower than 20 μL cannot be accurately taken, use less concentrated stock solutions so that larger volumes of the corresponding aliquots can be used. Table 2. Pipetting scheme to assess the trueness of the DL-2-HG EISA method Spiked concentration (μmol/L) per DL-2-HG enantiomer 0 (non-spiked) 10 50 70 Concentration (μmol/L) of the stock solutions used for spikinga - 2 2 2 Volume (μL) of stock solution addedb - 1 5 7 Volume (μL) of 1 mmol/L IS solution addeda,b 2 2 2 2 Volume (μL) of methanol added 90 90 90 90 a The concentration depicted here is the sum of both enantiomers. b In situations where volumes lower than 20 μL cannot be accurately taken, use less concentrated stock solutions so that larger volumes of the corresponding aliquots can be used. Table 3. Troubleshooting of the UHPLC-EISA-TOF method for the quantification of DL-2-HG in biological samples Step Problem Observation Possible reason Solution Derivatization (Step 16) Derivatization did not work Standards or samples do not turn into yellow or brownish color Not all water was evaporated. Alternatively, it could also be possible that the sample is not very concentrated and thus the color is not that noticeable. Make sure all water has been evaporated before adding L-DATAN solution. Liquid chromatography instrument (Steps 24–26) High column backpressure Pressure near 400 bar and/or low reproducibility of the retention time Column may be clogged Clean the column using 2-propanol. If the problem persists, replace the column Pressure fluctuation The pressure is not constant Valve or pump seal leaks and might be damaged Replace the check valve and the pump seals Loss of chiral separation The resolution of the peaks is lower than expected Mobile phase A and B are not prepared correctly or column is defective Make sure mobile phase A and B are well prepared and/or replace the column Mass spectrometry instrument (Steps 21–23) No peaks appear Ions of the compound are not observed Some parameters are not set correctly setup and/or the derivatization step was not successful Check the MS parameters and/or make sure the derivatization was conducted in the proper way Poor mass accuracy Peaks are not observed when they are extracted with a width of m/z ±0.01 The system is not well calibrated Calibrate the MS detector according to the manufacturer’s instructions Loss of sensitivity Ion counts are too low The ion source may be contaminated Clean the ion source, spray shield and the glass capillary following the manufacturer’s instructions Table 4. Results of assessing the trueness, precision and recovery of the different biological samples by the developed EISA-TOF method K562 cells Urine FCS Serum L-2-HG D-2-HG L-2-HG D-2-HG L-2-HG D-2-HG L-2-HG D-2-HG Trueness % Mean recovery ± SD 10 μmol/L 97 ± 4 90 ± 5 98 ± 8 105 ± 5 105 ± 8 101± 4 103 ± 2 106 ± 6 50 μmol/L 95 ± 2 87 ± 2 100 ± 8 113 ± 10 100 ± 5 96.7 ± 0.3 106 ± 3 105 ± 5 70 μmol/L 97 ± 2 83 ± 5 98 ± 3 105 ± 5 99 ± 2 96 ± 3 105 ± 1 105 ± 2 Precision RSD (%) Between-injection repeatability (n=3)a 5.0 5.0 3.6 3.9 1.1 2.2 1.3 2.2 Measurement repeatability (n=3)b 4.8 4.8 1.8 2.4 5.3 6.8 3.5 3.1 Intermediate measurement precision (n=9)c 4.9 4.9 4.6 4.4 6.7 7.3 3.8 5.2 Recovery (before and after the extraction) % Mean recovery ± SD 104 ± 8 104 ± 8 99 ± 2 94 ± 5 99 ± 3 94 ± 7 95 ± 3 99 ± 5 a Three consecutive injections of the corresponding biological sample spiked with 10 μmol/L of L-2-HG, 10 μmol/L of D-2-HG, 10 μmol/L of L-2-HG-d3 and 10 μmol/L of D-2-HG-d3. b Three replicates of the corresponding biological sample spiked with 10 μmol/L of L-2-HG, 10 μmol/L of D-2-HG, 10 μmol/L of L-2-HG-d3 and 10 μmol/L of D-2-HG-d3 injected in triplicate on the same day. c Three replicates of the corresponding biological sample spiked with 10 μmol/L of L-2-HG, 10 μmol/L of D-2-HG, 10 μmol/L of L-2-HG-d3 and 10 μmol/L of D-2-HG-d3 injected during three consecutive days. Table 5. Results of the calibration line and the analytical characteristics of the plasma analysis for the EISA-TOF and EISA-Q methods. EISA-TOF EISA-Q Internal standard calibration method (0.5–100 μmol/L)a L-2-HG D-2-HG L-2-HG D-2-HG Slope ± SD 0.131 ± 0.002 0.116 ± 0.002 0.124 ± 0.002 0.119 ± 0.003 Intercept ± SD 0.011 ± 0.003 0.021 ± 0.003 0.009 ± 0.003 0.013 ± 0.004 R2 0.998 0.997 0.998 0.995 LOQb 0.5 μmol/L 0.5 μmol/L 0.5 μmol/L 0.5 μmol/L F-value of ANOVAc 3268 2918 4046 1668 Plasma Trueness % Mean recovery ± SD 10 μmol/L 98 ± 4 100 ± 4 85 ± 6 96 ± 21 50 μmol/L 97 ± 5 98.9 ± 0.1 99 ± 3 104 ± 4 70 μmol/L 95 ± 8 99 ± 10 101 ± 4 104 ± 11 Precision RSD (%) Between injection repeatability (n=3)d 3.6 2.8 2.3 6.8 Measurement repeatability (n=3)e 2.6 5.5 5.4 5.6 Intermediate Measurement precision (n=9)f 4.1 6.0 4.7 6.7 Recovery (before and after the extraction) % Mean recovery ± SD 106 ± 5 97 ± 4 86 ± 4 88 ± 14 a Nine standard solutions at concentration levels from 0.5 to 100 μmol/L for L-2-HG and D-2-HG. Each solution contains 10 μmol/L of L-2-HG-d3 and D-2-HG-d3. The calibration line was established by plotting the area ratio of each enantiomer versus the concentration using a 1/x2 weighting. b LOQ estimated as the lowest concentration in the calibration that can be quantified with acceptable trueness (between 70–130%) and precision (coefficient of variation <30%). c F-value for ANOVA test that confirms that experimental data fit to a linear model. In all cases the p-values were lower than 0.05. d Three consecutive injections of a plasma sample spiked with 10 μmol/L of L-2-HG, 10 μmol/L of D-2-HG, 10 μmol/L of L-2-HG-d3 and 10 μmol/L of D-2-HG-d3. e Three replicates of plasma samples spiked with 10 μmol/L of L-2-HG, 10 μmol/L of D-2-HG, 10 μmol/L of L-2-HG-d3 and 10 μmol/L of D-2-HG-d3 injected in triplicate on the same day. f Three replicates of plasma samples spiked with 10 μmol/L of L-2-HG, 10 μmol/L of D-2-HG, 10 μmol/L of L-2-HG-d3 and 10 μmol/L of D-2-HG-d3 injected during three consecutive days. Box 1. Preparation of DL-2-HG standard solution including L-DATAN-derivatization and SPE purification This step is required for the optimization of the analysis of DL-2-HG standard and does not always need to be performed. This should be conducted only for the analytes of interest (DL-2-HG in our case) and not to the IS. This step is not relevant for the quantification non-chiral molecules. SPE purification is performed to minimize interference of non-derivatized L-DATAN during the EISA method optimization via direct infusion and avoid contamination of the system. This process might also be relevant if you have used a different derivatisation method. Mix 25 μL of 50 mmol/L D-2-HG solution and 25 μL of 50 mmol/L L-2-HG solution in a 1.5-mL polypropylene microcentrifuge tube. Follow the steps 10–17 from section “Derivatization of the calibration standards and extractant samples e.g. with diacetyl-L-tartaric anhydride (L-DATAN)” and after Step 17, redissolve the dried residue in 1 mL of LC-MS grade water. Purify the derivatized compound using a solid phase extraction (SPE) system and a C18 cartridge (Sep-Pak C18 200 mg) as follows: Wash with 3 mL of methanol to condition the stationary phase. Wash with 3 mL of LC-MS grade-water. Add 1 mL of the derivatized DL-2-HG solution. Wash with 3 mL of LC-MS grade-water. Discard this fraction. Wash with 3 mL of hexane. Discard this fraction. Place a glass test tube and add 3 mL of acetonitrile to the SPE cartridge to elute and collect the derivatized compound. Evaporate the derivatized compound using a N2 evaporator operated at 40 °C using a water bath. Redissolve the dried residue in 1 mL of the composition of the mobile phase where the analytes elute in the LC-MS system. In our case this was a mixture of 70:30 (v/v) mobile phase A and B, respectively. The final concentration was 1.25 mM for each 2-HG enantiomer. Load a 500 μL syringe with the SPE-purified (derivatized) compound to optimize the MS parameters by direct infusion. Competing interests: Authors declare no competing interests. References [1] Holcapek M , Jirasko R , Lisa M . Recent developments in liquid chromatography–mass spectrometry and related techniques. J. Chromatogr. A 1259 , 3–15 (2012).22959775 [2] Leung KS-Y , & Fong BM-W LC-MS/MS in the routine clinical laboratory: has its time come? Anal. Bioanal. Chem 406 , 2289–2301 (2014).24337187 [3] Wishart D . Emerging applications of metabolomics in drug discovery and precision medicine. Nat. Rev. Drug. Discov 15 , 473–484 (2016).26965202 [4] Blazenovic I , Kind T , Ji J , Fiehn O , Blaženović. Software Tools and Approaches for Compound Identification of LC-MS/MS Data in Metabolomics. Metabolites 8 , 31 (2018).29748461 [5] Xue J . Enhanced in-Source Fragmentation Annotation Enables Novel Data Independent Acquisition and Autonomous METLIN Molecular Identification. Anal. Chem 92 , 6051–6059 (2020).32242660 [6] Xue J . Single Quadrupole Multiple Fragment Ion Monitoring Quantitative Mass Spectrometry, Anal. Chem 93 , 10879–10889 (2021). [7] Xue J , Derks RJE , Hoang L , Giera M , Siuzdak G . Proteomics with Enhanced In-Source Fragmentation/Annotation: Applying XCMS-EISA Informatics and Q-MRM High-Sensitivity Quantification. J. Am. Soc. Mass Spectrom 32 , 2644–2654 (2021).34633184 [8] Dang L . Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature 462 , 739–743 (2009).19935646 [9] Struys EA , Jansen EEW , Verhoeven NM , Jakobs C . Measurement of urinary D- and L-2-hydroxyglutarate enantiomers by stable-isotope-dilution liquid chromatography-tandem mass spectrometry after derivatization with diacetyl-L-tartaric anhydride, Clin. Chem 50 , 1391–1395 (2004).15166110 [10] Zhou J . & Yin Y . Strategies for large-scale targeted metabolomics quantification by liquid chromatography-mass spectrometry. Analyst 141 , 6362 (2016).27722450 [11] González-Ruiz V , Olives AI , Martin MA Core-shell particles lead the way to renewing high-performance liquid chromatography, TrAC-Trend. Anal. Chem 64 , 17–28 (2015). [12] Ciccimaro E . & Blair IA Stable-isotope dilution LC-MS for quantitative biomarker analysis, Bioanalysis 2 , 311–341 (2010).20352077 [13] Pansuriya TC Somatic mosaic IDH1 and IDH2 mutations are associated with enchondroma and spindle cell hemangioma in Ollier disease and Maffucci syndrome. Nat. Genet 43 , 1256–1261 (2011).22057234 [14] Gil-Benso R . Establishment and characterization of a continuous human chondrosarcoma cell line, ch-2879: comparative histologic and genetic studies with its tumor of origin. Lab. Invest 83 , 877–887 (2003).12808123 [15] Scully SP Marshall Urist Award. Interstitial collagenase gene expression correlates with in vitro invasion in human chondrosarcoma. Clin. Orthop. Relat. Res 376 , 291–303 (2000). [16] Arindrarto W , Comprehensive diagnostics of acute myeloid leukemia by whole transcriptome RNA sequencing. Leukemia 35 , 47–61 (2021).32127641 [17] Oldham WM and Loscalzo J . Quantification of 2-hydroxyglutarate enantiomers by liquid chromatography-mass spectrometry. Bio. Protoc 20 , e1908 (2016). [18] FDA, editor. Analytical Procedures and Methods Validation U.S.D.o.H.a.H. Services Rockville, Maryland: Food and Drug Admistration (2000). [19] Kranendijk M , Struys EA , Salomons GS , Van der Knaap MS , Jakobs C . Progress in understanding 2-hydroxyglutaric acidurias. J. Inherit. Metab. Dis, 35 , 571–587 (2012).22391998

PMC010xxxxxx/PMC10364130.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 101217933 32339 J Neural Eng J Neural Eng Journal of neural engineering 1741-2560 1741-2552 36541546 10364130 10.1088/1741-2552/aca4fa NIHMS1916488 Article Characterizing physiological high-frequency oscillations using deep learning Zhang Yipeng MS 1 Chung Hoyoung 1 Ngo Jacquline P. MD 2 Monsoor Tonmoy MS 1 Hussain Shaun A. MD, MS 2 Matsumoto Joyce H. MD 2 Walshaw Patricia D. PhD 3 Fallah Aria MD, MS 4 Sim Myung Shin PhD 5 Asano Eishi MD, PhD, MS 6 Sankar Raman MD, PhD 2713 Staba Richard J. PhD 7 Engel Jerome Jr. MD, PhD 78910 Speier William PhD 1112 Roychowdhury Vwani PhD 1 Nariai Hiroki MD, PhD, MS 213* 1 Department of Electrical and Computer Engineering, University of California, Los Angeles, CA, USA 2 Division of Pediatric Neurology, Department of Pediatrics, UCLA Mattel Children’s Hospital, David Geffen School of Medicine, Los Angeles, CA, USA 3 Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine, Los Angeles, CA, USA 4 Department of Neurosurgery, UCLA Medical Center, David Geffen School of Medicine, Los Angeles, CA, USA 5 Department of Medicine, Statistics Core, University of California, Los Angeles, CA, USA 6 Department of Pediatrics and Neurology, Children’s Hospital of Michigan, Wayne State University School of Medicine, Detroit, MI, USA 7 Department of Neurology, UCLA Medical Center, David Geffen School of Medicine, Los Angeles, CA, USA 8 Department of Neurobiology, University of California, Los Angeles, CA, USA 9 Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles, CA, USA 10 The Brain Research Institute, University of California, Los Angeles, CA, USA 11 Department of Radiological Sciences, University of California, Los Angeles, CA, USA 12 Department of Bioengineering, University of California, Los Angeles, CA, USA 13 The UCLA Children’s Discovery and Innovation Institute, Los Angeles, CA, USA AUTHOR CONTRIBUTIONS We certify that all the authors listed made significant contributions to the work and share responsibility and accountability for the results.Conception and design of the study: YZ, WS, VR, HN Acquisition and analysis of data: YZ, HC, JN, TM, JHM, PW, MSS Drafting a significant portion of the manuscript or figures: YZ, SAH, JHM, AF, AE, RS, RJS, JE, WS, VR, HN * Corresponding author: Hiroki Nariai, MD, PhD, MS, Division of Pediatric Neurology, Department of Pediatrics, UCLA Mattel Children’s Hospital, David Geffen School of Medicine, Los Angeles, CA, USA, Address: 10833 Le Conte Ave, Room 22-474, Los Angeles, CA 90095-1752, USA, hnariai@mednet.ucla.edu 13 7 2023 07 12 2022 07 12 2022 24 7 2023 19 6 10.1088/1741-2552/aca4faThis file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Objective: Intracranially-recorded interictal high-frequency oscillations (HFOs) have been proposed as a promising spatial biomarker of the epileptogenic zone. However, HFOs can also be recorded in the healthy brain regions, which complicates the interpretation of HFOs. The present study aimed to characterize salient features of physiological HFOs using deep learning (DL). Methods: We studied children with neocortical epilepsy who underwent intracranial strip/grid evaluation. Time-series EEG data were transformed into DL training inputs. The eloquent cortex (EC) was defined by functional cortical mapping and used as a DL label. Morphological characteristics of HFOs obtained from EC (ecHFOs) were distilled and interpreted through a novel weakly supervised DL model. Results: A total of 63,379 interictal intracranially-recorded HFOs from 18 children were analyzed. The ecHFOs had lower amplitude throughout the 80–500 Hz frequency band around the HFO onset and also had a lower signal amplitude in the low frequency band throughout a one-second time window than non-ecHFOs, resembling a bell-shaped template in the time-frequency map. A minority of ecHFOs were HFOs with spikes (22.9%). Such morphological characteristics were confirmed to influence DL model prediction via perturbation analyses. Using the resection ratio (removed HFOs/detected HFOs) of non-ecHFOs, the prediction of postoperative seizure outcomes improved compared to using uncorrected HFOs (area under the ROC curve of 0.82, increased from 0.76). Interpretation: We characterized salient features of physiological HFOs using a DL algorithm. Our results suggested that this DL-based HFO classification, once trained, might help separate physiological from pathological HFOs, and efficiently guide surgical resection using HFOs. HFO physiological HFO machine learning pmcINTRODUCTION Intracranially-recorded interictal high-frequency oscillations (HFOs) have been proposed as a promising spatial biomarker of the epileptogenic zone (EZ).1, 2 Animal and human studies have demonstrated the association between HFOs and brain tissue capable of generating seizures.3–6 Several retrospective studies have linked favorable post-surgical seizure outcomes to the resection of cortical sites showing interictal HFOs.7–11 However, despite the potential, HFOs can also be recorded in healthy brain regions, which complicates the interpretation of HFOs when one attempts to guide resection using HFOs. Several recent studies, including a large multicenter prospective study, failed to correlate the removal of HFO-generating brain regions with postoperative seizure freedom; some patients became seizure-free despite part of the brain regions generating HFOs being preserved.12, 13 An ongoing clinical trial utilizing HFOs in electrocorticography to guide resection has excluded enrollment of occipital lobe epilepsy, due to abundant physiological HFOs in the visual areas.14 The current impasse is that there are no methods to separate pathological from physiological HFOs. Within the field of artificial intelligence, machine learning can bridge statistics and computer science to develop algorithms to complete tasks by exposure to meaningful clinical data without explicit instruction. Indeed, machine learning has been successfully applied to the problem of classifying HFOs based on a priori manual engineering of event-wise features, which includes: linear discriminant analysis,15 support vector machines,16, 17 decision trees,18 and clustering.19 More recently, the deep learning (DL) framework has been adopted, which directly works with raw data (avoiding any a priori feature engineering) and yields better performance in the field of neuroimaging.20 Leveraging DL’s revolutionary success in the field of computer vision using Convolutional Neural Networks (CNNs), prior studies explored the use of CNNs in EEG analysis, especially converting one-dimensional EEG signals into a two-dimensional image for CNNs input.21–23 The previous DL approaches conducted the HFO classification in a supervised manner, requiring human annotated labels which constrains the spectrum of usage of their methods, especially the needs of human expert labeling. In the context of medical image analysis, recent work has shown that optimized model architectures and loss functions could mitigate data labeling errors, thus making the DL framework even more applicable.24 Our recent work demonstrated that using the channel resection status as DL labels, a novel weakly-supervised DL algorithm characterized HFOs generated by the epileptogenic zone.25 The present study aimed to characterize physiological HFOs, using functional cortical mapping results as DL training labels. We leveraged results from functional cortical mapping by stimulation and gamma-related language mapping in children with neocortical medication-resistant epilepsy who underwent intracranial EEG monitoring with a grid/strip approach. We characterized salient features of physiological HFOs represented by the eloquent cortices (ecHFOs) through the DL algorithm. Also, we investigated if removing cortical regions with HFOs excluding ecHFOs (to mitigate the effect of physiological HFOs) correlated with postoperative seizure freedom better than using uncorrected HFOs alone. METHODS: Patient cohort: This was a retrospective cohort study. Children (below age 21) with medically refractory epilepsy (typically with monthly or greater seizure frequency and failure of more than three first-line anti-seizure medications) who had intracranial electrodes implanted for the planning of epilepsy surgery with anticipated cortical resection with the Pediatric Epilepsy Program at UCLA were consecutively recruited between August 2016 and August 2018. Diagnostic stereo-EEG evaluation (not intended for resective surgery) was excluded. Standard protocol approvals, registrations, and patient consents: The institutional review board at UCLA approved the use of human subjects and waived the need for written informed consent. All testing was deemed clinically relevant for patient care, and also all the retrospective EEG data used for this study were de-identified before data extraction and analysis. This study was not a clinical trial, and it was not registered in any public registry. Patient evaluation: All children with medically refractory epilepsy referred during the study period underwent a standardized presurgical evaluation, which—at a minimum—consisted of inpatient video-EEG monitoring, high resolution (3.0 T) brain magnetic resonance imaging (MRI), and 18 fluoro-deoxyglucose positron emission tomography (FDG-PET), with MRI-PET co-registration.26 The margins and extent of resections were determined mainly based on seizure onset zone (SOZ), clinically defined as regions initially exhibiting sustained rhythmic waveforms at the onset of habitual seizures. In some cases, the seizure onset zones were incompletely resected to prevent an unacceptable neurological deficit. Subdural electrode placement: Macroelectrodes, including platinum grid electrodes (10 mm intercontact distance) and depth electrodes (platinum, 5 mm intercontact distance), were surgically implanted. The total number of electrode contacts in each subject ranged from 40 to 128 (median 96 contacts). The placement of intracranial electrodes was mainly guided by the results of scalp video-EEG recording and neuroimaging studies. All electrode plates were stitched to adjacent plates, the edge of the dura mater, or both, to minimize the movement of subdural electrodes after placement. Acquisition of three-dimensional (3D) brain surface images: We obtained preoperative high-resolution 3D magnetization-prepared rapid acquisition with gradient echo (MPRAGE) T1-weighted image of the entire head. A FreeSurfer-based 3D surface image was created with the location of electrodes directly defined on the brain surface, using post-implant computed tomography (CT) images.27 In addition, intraoperative pictures were taken with a digital camera before dural closure to enhance the spatial accuracy of electrode localization on the 3D brain surface. Upon re-exposure for resective surgery, we visually confirmed that the electrodes had not migrated compared to the digital photo obtained during the electrode implantation surgery. Intracranial EEG (iEEG) recording: Intracranial EEG (iEEG) recording was obtained using Nihon Kohden Systems (Neurofax 1100A, Irvine, California, USA). The study recording was acquired with a digital sampling frequency of 2,000 Hz, which defaults to a proprietary Nihon Kohden setting of a low frequency filter of 0.016 Hz and a high frequency filter of 600 Hz at the time of acquisition. For each subject, a continuous 90-minute EEG segment were selected at least two hours before or after seizures, before anti-seizure medication tapering, and before cortical stimulation mapping, which typically occurred two days after the implant. All the study iEEG data were part of the clinical EEG recording. Functional cortical mapping: Cortical stimulation was performed as part of clinical management to define regions of the eloquent cortices (EC) to help guide resections. A pulse train of repetitive electrical stimuli was delivered to neighboring electrode pairs, with a stimulus frequency of 50 Hz, pulse duration of 300 μs, and train duration ranging up to 5 seconds (sensorimotor or visual mapping) or 7 seconds (language mapping). Stimulus intensity ranged from 1 mA to 13 mA. Seven out of 19 patients also underwent auditory and picture-naming tasks to supplement cortical stimulation mapping. In short, patients were instructed to overtly verbalize an answer to a given auditory question or name a picture. Each EEG trial data was transformed into the time-frequency domain using complex demodulation via BESA software (BESA GmbH, Germany). The iEEG signal at each channel was assigned an amplitude (a measure proportional to the square root of power) as a function of time and frequency (in steps of 10 ms and 5 Hz). The time-frequency transform was obtained by multiplication of the time-domain signal with a complex exponential, followed by a band-pass filter. iEEG traces were aligned to: (i) stimulus (question) onset; (ii) stimulus offset; and (iii) response (answer) onset. We determined whether the degree of such gamma-augmentation reached significance using studentized bootstrap statistics followed by Simes’ correction.28, 29 Sites surviving correction showing significant gamma-augmentation spanning (i) at least 20-Hz in width and (ii) at least 20-ms in duration were defined as ‘language-related gamma sites’.30 Further methodological details of the gamma mapping are described in the prior studies.31 Automated detection and classification of HFOs: A customized average reference was used for the HFO analysis, with the removal of electrodes containing significant artifacts.26, 32 Candidate interictal high frequency oscillations (HFOs) were identified by an automated short-term energy detector (STE).33, 34 This detector considers HFOs as oscillatory events with at least six peaks and a center frequency occurring between 80–500 Hz. The root mean square (RMS) threshold was set at five standard deviations (SD), and the peak threshold was set at three SD. The HFO events are segments of EEG signals with durations ranging from 60 to 200 ms.34 The candidate HFOs were evaluated by our previously validated algorithm (recall = 98.0%, precision = 96.1%, F1 score = 96.8% against human experts validation) to reject artifacts for further analyses.25 For simplicity, we use HFO in the later paragraphs to represent the HFO after the artifact rejection. Weakly supervised deep learning using functional cortical mapping results as labels: The general workflow of the DL training and inference was shown in the flowchart (Figure 1A). The channel-wise annotations include behavior (positive behavioral changes with cortical stimulation mapping or language-related gamma sites), SZ/AD (channels with seizures or afterdischarges with cortical stimulation mapping), spike (cortical sites showing spontaneous interictal spikes), and none (no behavioral changes with mapping, SZ/AD, and spikes). We hypothesize that a morphologically distinct class of HFOs generated by EC, representative as physiological HFOs (defined as ecHFOs). We expect a large percentage of the HFOs in the behavioral channels are ecHFOs. However, such channels could also have HFOs that are morphologically distinct from ecHFOs, since each channel also picks up neuronal activities generated by other physiological or pathological processes. We refer to such complementary HFOs as non-ecHFOs. Since the physiological processes generating ecHFOs are also present in different regions of the brain, for example, the unstimulated brain region, we expect non-behavioral channels to also have ecHFOs. However, as there is no ground truth annotation for an ecHFO event, we relied on channel-level annotations acquired from clinical experiments. Specifically, positive labels are assigned to HFOs from channels that only have behavior responses, excluding spike and SZ/AD channels, and negative labels are assigned to all other HFOs. When an HFO is from a channel with both behavior response and pathological (SZ/AD or Spike) response, we excluded them from the training set, as the distribution of ecHFOs and non-ecHFOs might be comparable. Clearly, such labeling has errors. However, if ecHFOs have distinct morphological signatures, then we could utilize the generalization power of the neural network to distill the ecHFO from this weak supervision. Model architecture and feature representation of HFOs: We adopt the same CNN architecture as the prior study because it has demonstrated excellent performance in classifying HFOs.25 A one-second window represented each HFO, centered on the HFO (0 ms). While the HFO duration is generally shorter than 200 ms, it does not necessarily mean that the discovered features representing the HFO’s characteristics are fully describable within that window. For example, spikes associated with HFOs provide distinct attributes in the time-frequency window, extending before and after the HFO detection point. Therefore, we used the longer window size to provide a more comprehensive description of an HFO event, which encompasses HFO itself and surrounding EEG waveforms.25 Then three images are constructed to capture the time-frequency domain features as well as signal morphology information of the HFO window. The time-frequency plot (scalogram) was generated by continuous Gabor Wavelets ranging from 10Hz to 500Hz. The EEG tracing plot was generated on a 2000 × 2000 image by scaling the time-series signal into the 0 to 2000 range to represent the EEG waveform’s morphology. The amplitude-coding plot was generated to represent the relative amplitude of the time-series signal: for every time point, the pixel intensity of a column of the image represented the signal’s raw value at that time. No frequency filtering was used to generate the EEG tracing and amplitude-coding plots. Then these three images were resized into the standard size (224 × 224), serving as the input to the neural network (Figure 1B). Regarding the time-frequency plot, the resizing operation is just binning the frequency bands, and the time-frequency plot still keeps the frequency information of the HFO signal. The EEG tracing plot captures the general EEG waveform of the HFO event, and the amplitude coding plot captures the HFO’s relative amplitude information. Thus, resizing the image does not lose HFO’s frequency information. This operation provides comprehensive HFO information as inputs to the neural network. Model training: We used 90 minutes of data from each patient for training. Since our task is binary classification, we use binary cross-entropy as the loss function, L=−[y⋅log(x)+(1−y)⋅log(1−x)], where y is the label (1 for ecHFO, 0 for non-ecHFO [ecHFOs were subtracted from total HFOs]), and x is the model output ecHFO probability. In training, we adopt stratified sampling to balance the data distribution in different labels. The Adam optimizer was adopted with a learning rate of 0.0001. The training was conducted using 25 epochs (training iterations), and validation loss was plotted with respect to the number of epochs completed. For selecting the best model across epochs, we picked the model corresponding to the balanced global minima over 25 iterations, i.e., the highest balanced accuracy (the averaged recall for each class). In order to fully explore the performance of the model, five-fold cross-validation was conducted using the pooled data across the full patient cohort. For each fold, 20% of the dataset was selected as the test set, 70% was selected as the training set, and the remaining 10% was used for validation. Performance analysis framework: Since the model was evaluated by using five-fold cross-validation, there were five distinctive models trained by different folds of data. For the downstream analysis, if an HFO is from a channel with both behavior response and SZ/AD response, the classification of these HFOs is determined by the mean of the probabilities from five models. Otherwise, the classification of the HFO is determined by the model where this specific HFO is in the test set. Characterization of ecHFOs: We adopted the same procedure as the prior study25 to determine which characterization in the time-frequency scalogram of ecHFOs differed from that of non-ecHFOs, that is, conducting a t-test across all pixels in the time-frequency scalogram (Figure 2A). The null hypothesis is that the values within the scalogram of ecHFO are not lower than those of non-ecHFO. We summarized the results by taking the average of the difference discovered in each patient to prove the characterization in the time-frequency plot generalizes at the population level (Figure 2B). In addition, to examine the influence of age, we divided the patient cohort into two groups (young patients [age < 10] and old patients [age ≥ 10]) and repeated the same process. Interpretability analysis of ecHFOs: Perturbation on time-frequency plot: We adopted a similar perturbation analysis as prior work25 based on the characterization of ecHFO in the time-frequency plot (Figure 2C). We measured the change in probabilities as the ratio (Poriginal – Pperturb) / Poriginal where the Poriginal is the original model confidence toward the event being ecHFO, and Pperturb is the model confidence with perturbed input for the same event being ecHFO. We summarized the change in the ratio via the histogram for each patient across all the predicted ecHFOs (Figure 2D). A one-tailed t-test was also performed on the change of that output probability score to ensure that the change was significant and generalized well at the population level. Perturbation on amplitude coding plot: We adopted a data-driven approach to discover the effect of introducing a spike in the amplitude coding plot (Figure 3). As in our prior work,25 we built a detector to capture HFO with a spike (spk-HFO) trained on expert annotations (86.5% accuracy with an F1 score of 80.8%).25 The spk-HFO detector was applied to our data and predicted whether each HFO event was an spk-HFO or not. For all detected spk-HFOs in each patient, we separated these HFOs into upward spikes and downward spikes. The upward (downward) spikes are defined as spk-HFO events, with the average amplitude in the center 20-ms duration being greater (smaller) than the average amplitude in the peripheral 250 ms. We then created upgoing and downgoing spike templates by taking the average of all tracings. The corresponding amplitude coding plots of the upgoing and downgoing spike templates are built for perturbing the amplitude coding plots for each patient (Figure 3A, C). This perturbation was performed by adding the spike template to the original amplitude coding plot, and we measured the change in model confidence. We summarized the results (ratio change) via the histogram for each patient across all the predicted ecHFOs (Figure 3B, D). A one-tailed t-test was performed on the change of output probability to evaluate whether the change was significant and generalized well at the population level. Time Domain Characteristic ecHFO: We reported three time-domain characteristics of the predicted ecHFOs and non-ecHFOs, which are amplitude, duration, and max frequency (Figure 4). The duration is defined as the time duration of the HFO predicted by the STE HFO detector. The amplitude is defined as the maximum absolute value within the HFO interval, and the max frequency is defined as the largest frequency component within a window (0.2 seconds before and after) around the center of the HFO. We plotted the normalized histogram to compare the distribution of these three characteristics. Comparison of resection ratios of HFOs to postoperative seizure outcomes: We estimated the probability of each patient’s surgical success (for the 14 patients who underwent resective surgery with known seizure outcomes at 24 months) based on the resection ratio of HFOs (number of resected HFOs/number of detected HFOs) as a classifier. We constructed the receiver operating characteristic (ROC) curve and calculated the area under the curve (AUC) values in the resection ratio of unclassified HFOs, ecHFOs, and HFOs with spikes (spk-HFOs). The classification of HFOs with spikes was performed using our prior algorithm.25 Determination of the channel resection status (resected vs. preserved) was determined based on intraoperative pictures (pre-and post-resection) and also on post-resection brain MRI, based on discussion among a clinical neurophysiologist (HN), neurosurgeon (AF), and radiologist (SN). A multiple logistic regression model incorporating the resection ratio of classified HFOs (non-ecHFOs and spk-HFOs) and complete resection of SOZ was also created. The surgical outcomes were determined 24 months after resection as either seizure-free or not seizure-free. Statistical analysis: Above mentioned statistical calculations were carried out using Python (version 3.7.3; Python Software Foundation, USA). The deep neural network was developed using PyTorch (version 1.6.0; Facebook’s AI Research lab). Quantitative measures are described by medians with interquartile or means with standard deviations. Comparisons between groups were performed using chi-square for comparing two distributions and Student’s t-test for quantitative measures (in means with standard deviations). All comparisons were two-sided and significant results were considered at p < 0.05 unless stated otherwise. Specific statistical tests performed for each experiment were described in each section. Data sharing and availability of the methods: Anonymized EEG data used in this study are available upon reasonable request to the corresponding author. The python-based code used in this study is freely available at (https://github.com/roychowdhuryresearch/HFO-Classification). One can train and test the deep learning algorithm from their data and confirm our methods’ validity and utility. RESULTS: Clinical information (patient characteristics): There were 19 patients (10 females) enrolled during the study period. The median age at surgery was 14 years (range: 3–20 years). The median electrocorticography monitoring duration was 4 days (range: 2–14 days), and the median number of seizures captured during the monitoring was 8 (IQ range: 4–25). There were 15 patients who underwent resection, and 14 patients provided postoperative seizure outcomes at 24 months (9 of 14 became seizure-free). Due to unknown postoperative seizure outcomes, patient #10 was removed from further HFO analyses. Details of patients’ clinical information are listed in Table 1. Interictal HFO detection: A total of 63379 HFO events were detected (median 2258.5 events per patient) in 90-minute EEG data from the 18 patients (patient #10 was removed for subsequent HFO analysis due to lack of seizure outcomes). There were 15111 HFOs recorded from the stimulated channels (48268 HFOs from non-stimulated channels). There were 7173 HFOs obtained from behavior channels (positive behavioral responses from cortical stimulation mapping and/or gamma mapping), and “behavior only” channels (behavior channels without spontaneous spikes or SZ/AD) exhibited 3822 HFOs. There were 5317 HFOs obtained from “none” channels (no spontaneous spikes and no behavioral responses with mapping). After the DL training, we noted that 22.9% (4547/19816) of the ecHFOs were HFOs with spikes, and 77.1% (30647/43563) of non-ecHFOs were HFOs with spikes (p < 0.0001, a chi-square test). Time-frequency plot characteristics of ecHFO and non-ecHFOs: The analysis of the time-frequency plots from the selected individual patient demonstrated that ecHFOs had lower amplitude throughout the frequency band, including both ripples (80–250 Hz) and fast ripples (250–500 Hz) around the center point (0 ms, where HFOs were detected) than non-ecHFOs (Figure 2A). We inspected the averaged overall characterization and selected regions greater than 0.7 to define the salient characteristic difference between ecHFOs and non-ecHFOs. The selected region resembled a bell-shaped (Figure 2B). We observed the similar characteristic template in the time-frequency plot regardless of whether a young or old cohort was used for the training (Suppl. Figure). Perturbation analyses: By utilizing the bell-shaped template found in the time-frequency map, we observed that the bell-shaped perturbation on the time-frequency plot significantly increased the model prediction probability towards non-ecHFOs (mean increase ratio was 0.532 [95% confidence interval: 0.528–0.535], p < 0.001) (Figure 2C, D). Furthermore, we analyzed the effect of introducing a spike-like shape in the amplitude-coding plot. By introducing a downgoing or an upgoing spike in an ecHFOs event, the model confidence increased towards a non-ecHFO event (Figure 3A, C). On the population level, the prevalent probability increase in ecHFO events among all patients (Figure 3B, D) demonstrated the non-trivial model response by introducing a spike in the time domain (mean ratio increase 0.268 for a downgoing spike introduction [95% confidence interval: 0.263–0.272], and 0.320 for an upgoing spike introduction [95% confidence interval: 0.316–0.324], both with p< 0.001). Feature characterization of ecHFOs vs. non-ecHFOs: We plotted the histogram of the peak frequency, amplitude, and duration of both ecHFO and non-ecHFO, respectively (Figure 4). The ecHFOs exhibited a smaller amplitude (p-value < 0.001), a lower max frequency (p-value < 0.001) and a trend of shorter HFO length (p-value = 0.07) than non-ecHFO. However, there was no decision boundary to clearly discriminate between ecHFO and non-ecHFO using each of these traditional features. Inference of channel characteristics: After the model was trained, it assigned one prediction label for all of the HFOs in each channel, and we plotted the ecHFO ratio (proportion of ecHFOs divided by the number of all HFOs) and the model confidence distribution towards ecHFOs in different channels (Figure 5). In channels with a physiological response (only behavior responses with functional mapping), the ecHFO ratio was high, and confidence scores were skewed towards ecHFOs. Channels with a pathological response (spike and SZ/AD with stimulation) showed a lower ecHFO ratio, and confidence scores were skewed towards non-ecHFOs. Channels with no behavioral responses showed the confidence scores skewed towards non-ecHFOs. Notably, channels with both properties, in which both behavioral responses and spikes or/and SZ/AD, model confidence showed a uniform distribution, suggesting the presence of both ecHFOs and non-ecHFOs. Clinical inference using postoperative seizure outcomes: We created the ROC curves using the HFO resection ratio to predict postoperative seizure freedom at 24 months (n = 14) (Figure 6). Using the resection ratio of HFOs showed acceptable prediction performance (AUC= 0.76; p < 0.001). The resection ratio of non-ecHFOs and spk-HFOs exhibited higher AUC values of 0.82 and 0.87 than unclassified HFOs, respectively (p < 0.001 for both). The performance was further augmented by using a multiple regression model incorporating the resection ratio of such classified HFOs (non-ecHFOs and spk-HFOs) and the complete removal of SOZ (AUC = 0.933 for both classifiers, p =0.047 and 0.043, respectively). DISCUSSION HFOs with similar frequency ranges emerge despite considerably different mechanisms.35, 36 A traditional hypothesis-driven approach to separate physiological HFOs from pathological HFOs poses challenges because numerous yet-to-be-identified features must be considered. Simple engineering features of HFOs, including amplitude, frequency, and duration, do not appear to successfully separate pathological from physiological HFOs.37, 38 Visual classification of HFOs with or without spikes along with artifact removal (such as ringing) is commonly performed because HFOs with spike-wave discharges are considered representative of pathological HFOs.39 However, this task is time-consuming and exhibits poor inter-rater reliability among human experts.40, 41 Fast ripples (250–500 Hz) might more specifically localize epileptogenic zones than ripples do (80–250 Hz), but their detection rate is much lower than ripples.13 Correcting the HFO detection rate with region-specific normative values seems a reasonable approach,32, 42–44 but this does not determine each HFO event as either pathological or physiological. In the present study, we leveraged the robust clinical definition of the EC with cortical stimulation and gamma-related language mapping to identify cortical areas generating physiological HFOs (represented by ecHFOs). Our approach using a DL-based algorithm is distinct from other approaches. Based on our hypothesis that physiological HFOs look morphologically different from pathological HFOs, leveraging DL’s ability to analyze imaging input (transformed from EEG time-series data) seemed logical. By training a DL model with labels based on functional cortical mapping results. We then investigated the salient features of the physiological HFOs through the DL algorithm. The novel findings include that we found a bell-shaped template in the time-frequency plot as the discriminating feature of ecHFOs. More specifically, ecHFOs would have lower signal amplitude at the center of the HFO onset across the frequencies (including both ripple and fast ripple band) and lower signal amplitude in the low frequency band throughout the time window than non-ecHFOs (representing pathological HFOs). The ecHFOs generally had a slower peak frequency, a smaller amplitude, and tended to have a longer duration than non-ecHFOs. However, these time-domain characteristics showed significant overlaps between ecHFOs and non-ecHFOs, and none of such simple engineering features could clearly separate them, consistent with the previous studies.37, 38 Once the DL model was trained, we verified specific morphological features of ecHFOs via perturbation analysis. Insertion of spike templates significantly decreased ecHFO probability, which confirmed the traditional knowledge that HFOs with spikes are most likely pathological HFOs. Our findings have significant clinical implications. Using the resection ratio (removed HFOs/detected HFOs) of non-ecHFOs (likely representing pathological HFOs), we demonstrated that the prediction of postoperative seizure outcomes significantly improved compared to using the uncorrected HFO resection ratio (AUC of 0.82, increased from 0.76). Combined with the current clinical standard of complete resection of SOZ, the AUC further improved to 0.93. The above findings suggest that the classification of HFOs (physiological vs. pathological) can potentially improve the clinical utility of HFOs in guiding resection, and this can still be combined with our current clinical standard of capturing habitual seizures to decide the resection margin. A similar improvement in AUC was observed using spk-HFOs (AUC of 0.87, increased from 0.76), which seemed even better than non-ecHFOs. This result suggests that spk-HFOs may better represent pathological HFOs than non-ecHFO. However, HFOs with spikes need to be evaluated by human experts, which exhibits concerns about inter-rater reliability.41 There is also no objective definition of spikes. So, analyzing “HFOs with spikes” may not be as straightforward as it sounds. At least, the determination of ecHFOs can be made by objective functional mapping results. We are planning more studies to investigate the additional clinical utility of ecHFOs, such as identifying eloquent cortices by analyzing ecHFOs, without the need for stimulation mapping. A recent study demonstrated that a cortical location could produce different types of HFOs, such as both physiological and pathological HFOs.19 We probed this point by investigating cortical sites with both EC characteristics (behavioral responses with stimulation mapping or/and gamma activity with language tasks) and pathological properties (spontaneous spikes or/and AD/SZ). The ecHFO ratio was between cortical sites with only behavioral responses and that with spikes or/and SZ/AD only. The confidence scores of HFO-type prediction are diffusely distributed from ecHFOs to non-ecHFOs, without clear bimodal distribution. If such “both” channels produce morphologically distinct physiological and pathological HFOs, one would expect to see a clear bimodal distribution. Our findings suggest that in cortical sites with both pathological and physiological properties, the morphologies of HFOs are heterogeneous. This may be a limitation of using macroelectrodes since multiple local generators might influence each HFO morphology. There are several limitations in our study. We have analyzed only 18 patients to define physiological HFOs. Also, postoperative outcomes were correlated in only 14 patients. Although we analyzed an extended EEG dataset (90 minutes from each subject) to maximize the number of HFOs for analysis, we will need to analyze more patients to have a definitive conclusion. We demonstrated the discovered ecHFOs’ characteristics were consistent even trained by different age groups. Still, a robust conclusion should be drawn from analyses of a larger number of subjects to examine the effect of age. Although we assumed that HFOs from the healthy brain regions look morphologically similar, such characteristics may be different in the different brain regions. With more subjects, we will be able to characterize the difference in ecHFOs generated in functionally distinct cortical areas, such as the somatosensory, language, and visual cortex. In addition, we analyzed only sleep EEG to maximize the number of HFOs for analysis. States of consciousness (including awake and different sleep stages) and vigilance levels may affect HFO morphology.45 Lastly, with an increasing number of diagnostic stereotactic EEG (SEEG) studies, it is of interest to investigate how HFO morphologies via SEEG would look different compared to strip/grid-sampled HFOs. In summary, we proposed to use a weakly-supervised DL algorithm to characterize physiological HFOs using robust clinical outcomes (functional cortical mapping results) as training labels. Although future work to include a larger number of subjects will be needed, this method provided salient morphological features of physiological HFOs. Our results also suggested that this DL-based HFO classification, once trained, might help separate pathological from physiological HFOs, and efficiently guide surgical resection using HFOs. Supplementary Material Suppl_data ACKNOWLEDGMENT The authors have no conflict of interest to disclose. HN is supported by the Sudha Neelakantan & Venky Harinarayan Charitable Fund, the Elsie and Isaac Fogelman Endowment, and the UCLA Children’s Discovery and Innovation Institute (CDI) Junior Faculty Career Development Grant (#CDI-SEED-010121; #CDI-TTCF-07012021). SAH has received research support from the Epilepsy Therapy Project, the Milken Family Foundation, the Hughes Family Foundation, the Elsie and Isaac Fogelman Endowment, Eisai, Lundbeck, Insys, Zogenix, GW Pharmaceuticals, UCB, and has received honoraria for service on the scientific advisory boards of Questcor, Mallinckrodt, Insys, UCB, and Upsher-Smith, for service as a consultant to Eisai, UCB, GW Pharmaceuticals, Insys, and Mallinckrodt, and for service on the speakers’ bureaus of Mallinckrodt and Greenwich Bioscience. RS serves on scientific advisory boards and speakers bureaus and has received honoraria and funding for travel from Eisai, Greenwich Biosciences, UCB Pharma, Sunovion, Supernus, Lundbeck Pharma, Liva Nova, and West Therapeutics (advisory only); receives royalties from the publication of Pellock’s Pediatric Neurology (Demos Publishing, 2016) and Epilepsy: Mechanisms, Models, and Translational Perspectives (CRC Press, 2011). RJS is supported by the National Institute of Neurological Disorders and Stroke (NINDS) R01NS106957. JEJ is supported by NINDS U54NS100064 and R01NS033310. The research described was also supported by NIH/National Center for Advancing Translational Science (NCATS) UCLA CTSI Grant Number UL1TR001881. We are indebted to Jason T. Lerner, Lekha M. Rao, Rajsekar R. Rajaraman, Maria Garcia Roca, Richard Le, Patrick Wilson, and Jimmy C Nguyen for their assistance in the study and sample acquisition. Figure 1: Processing workflow (A): The detected HFO (by STE detector) is first filtered by the artifact detector that was developed in our previous study.25 The constructed image features along with the channel-wise clinical information, are fed into the convolutional neural network to train the model (ecHFO detector). In the real application of the model (inference), an unseen EEG signal is sent into the pipeline, and the STE detector is applied to detect all HFOs. Then we construct the image feature based on the detected HFO. The same artifacts detector25 is applied to reject the artifacts, and then the real HFO is filtered by the ecHFO detector. The detector will output a score from zero to one representing the confidence of each HFO being ecHFO. HFO feature representation and model architecture (B): We captured the time-frequency domain features as well as signal morphology information of the HFO window via three images. The time-frequency plot (scalogram) was generated by continuous Gabor Wavelets ranging from 10Hz to 500Hz. The EEG tracing plot was generated on a 2000 × 2000 image by scaling the time-series signal into the 0 to 2000 range to represent the EEG waveform’s morphology. We constructed another two images from the original EEG signal serving as the additional inputs to the network. The amplitude-coding plot was generated to represent the relative amplitude of the time-series signal: for every time point, the pixel intensity of a column of the image represented the signal’s raw value at that time. These three images were resized into the standard size (224 × 224), serving as the input to the neural network. We used ResNet 18 with a modified output layer for binary classification. The weights in the convolution layers are frozen and serve as feature extractors in the convolutional neural network. Figure 2: Characteristics in the time-frequency plot of ecHFOs against non-ecHFOs. (A) The time-frequency plot characteristics of the eloquent cortex and non-eloquent cortex HFOs for Pt 4, 5, 7, and 8. The yellow-colored regions in the figure stood for the pixels, where the power spectrum of ecHFOs is statistically lower than (p-value below 0.05 from the one-tailed t-test) non-ecHFOs. The figure showed one set of clearly interpretable distinguishing features between ecHFOs and non-ecHFOs: the ecHFOs generally have lower power at lower frequencies during the HFO event (center part along the time axis), Panel (B-Right) was generated by taking the average of the individual binary images from each of the 18 patients. It showed the distinguishing features are also significant at the population level. The feature can be assembled as a “Bell-Shape” if we take the region > 0.7 in the plot averaging all of the time-frequency plot characteristics (red color). (B-Left). (C) The model’s response to the Bell-shaped perturbation on the time-frequency plot. We provide two examples of perturbation for ecHFO events in Pt 3. Each row presents one example and the first column indicates the original time-frequency plot while the second indicates the perturbed time-frequency plot based on the Bell Shape perturbation. The prediction value of the model changed from 0.780 (therefore originally predicting it as non-ecHFO) to 0.299 (thus a change of 0.481), implying that the perturbed HFO would correspond to a non-ecHFO. (D) The change in model confidence in population level. Each column (along the y-axis) is a histogram of the percentage of change in confidence for one distinct patient. It shows the frequency distribution of confidence changes after adding the bell shape perturbation to the time-frequency plot to all classified ecHFOs for the given patient. The change in confidence level is significant, with an average of 0.532 with a 95% confidence interval [0.528, 0.535] noted as the solid red line in the histogram. HFO = high-frequency oscillation; ecHFOs = eloquent cortex HFOs; non-ecHFO = non-eloquent cortex HFOs. Figure 3: The model’s responses to injecting a spike-like feature into the amplitude-coding plot. Examples of introducing a downgoing (A) and upgoing (C) spike feature into classified non-ecHFO events. These demonstrate that on the introduction of a spike-like perturbation, the model predicts higher confidence towards non-ecHFOs (B, D). Subfigure (A) shows the amplitude encoding image before perturb, spike template, and the after-perturb (top row), and the corresponding time-series signal with downgoing spike perturbation (bottom row). Similarly, subfigure (C) shows the same information on a different classified ecHFO but with upgoing spike perturbation. For each patient, we compute a histogram for the distribution of the change in confidence (B) for downgoing spike perturbation. The same steps are repeated for upgoing spike perturbation, and the results are shown in (D). The percentage of change in confidence for both up-and-downgoing spike perturbation is significantly greater than zeros, with means downgoing: 0.268 (95% confidence interval [0.263, 0.272]) and upgoing: 0.320 (95% confidence interval [0.316, 0.324]), which are noted as solid red lines in each histogram. HFO = high-frequency oscillation; ecHFOs = eloquent cortex HFOs; non-ecHFO = non-eloquent cortex HFOs. Figure 4: Traditional feature characterization of ecHFO and Non-ecHFO: The normalized histogram of peak frequency (A), amplitude (B), and duration (C) of predicted ecHFO. (A) The ecHFO generally has a lower peak frequency than the non-ecHFO (p-value < 0.001). (B) The ecHFO generally has a smaller amplitude than the non-ecHFO (p-value < 0.001). (C) The ecHFO generally has the trend of having a longer HFO than the non-ecHFO (p-value = 0.07). However, there are no clear decision boundaries that can be drawn to clearly discriminate between ecHFO and non-ecHFO using each of these traditional features. Figure 5. Inference of channel characteristics: (A) The ratio of ecHFOs (eHFOs/total HFOs) in different types (physiological: behavior or gamma only; pathological: SZ/AD or spike only; both: both physiological and pathological) of channels from all patients (n=14) is plotted in box plots. The percentage of ecHFOs was higher in physiological channels than that in pathological channels. (B) The model confidence distribution of each individual eHFOs in channels with physiological, pathological, and both categories are shown. The distribution in both channels is closer to a uniform distribution. Figure 6. The accuracy of prediction models incorporating HFO resection ratio. We constructed postoperative seizure outcome prediction models using the HFO resection ratio derived from EEG data (n = 14). Each receiver-operating characteristics (ROC) curve delineates the accuracy of seizure outcome classification of a given model, using the area under the ROC curve statistics. (A) Unclassified HFO resection ratio was used as a single classifier. (B) non-ecHFO resection ratio was used as a single classifier. (C) non-ecHFO resection ratio was used as a single classifier. (D, E) A multiple regression model incorporating the resection ratio of HFOs (non-ecHFOs or spk-HFO) and complete removal of the SOZ (yes or no) was used, which demonstrated further improved predictive value of postoperative seizure outcomes. HFO = high-frequency oscillation; non-ecHFO = non-eloquent cortex HFOs; spk-HFO = HFO with spike; SOZ = Seizure onset zone. Table 1. Cohort characteristics Pt No. Sex Age at surgery (yr) Epilepsy duration (yr) Anti-seizure medications No. of electrodes placed No. of electrodes resected % of electrodes resected Duration of EEG (days) No. of sz captured MRI lesion/FDG-PET hypometabolism Electrode coverage Surgery Pathology Functional Cortical Mapping type No. of HFOs detected (90 min) Outcome (follow-up at 24 months) 1 M 20 6 CLB, LVT, LCM 40 9 22.50% 3 21 NL/ R FP R FP R focal resection of sensorimotor cortex Gliosis Stimulation (sensorimotor) 953 Sz free 2 M 11 9 CLB, CNZ, LVT, RFD, PPN 80 18 22.50% 3 26 R PO/ R PO R FTPO R focal resection around parietal tumor Ganglioneurocytoma Stimulation (sensorimotor and language) 1493 Sz free 3 F 19 9 LVT, LCM 92 27 29.35% 5 8 R F/ R F R FTP R focal resection of frontal cortex FCD 1b Stimulation (sensorimotor) 2696 Sz free 4 F 14 10 CLB, LTG, LCM 64 28 43.75% 2 7 L F/ L F L FP L frontal lobectomy sparing sensorimotor cortex Gliosis Stimulation (sensorimotor and language) 5301 Sz free 5 M 9 7 CLB, LTG 100 83 83.00% 6 4 R TPO/ R TPO R FTPO R TPO Gliosis Stimulation (sensorimotor) 2428 Sz free 6 F 3 2 CLB, OXC 96 42 43.75% 2 18 R F/ R FP R FTP R frontal lobectomy sparing sensorimotor cortex FCD 2a Stimulation (sensorimotor) 2089 Sz recurrence after 1 day 7 M 5 3 PB, PPN, OXC 104 82 78.85% 2 22 L FP/ L FP L FTP L frontal lobectomy including resection of sensorimotor cortex FCD 2a Stimulation (sensorimotor) 8653 Sz free 8 F 19 7 LVT, LCM 104 0 0.00% 2 23 R TPO/ R TPO R FTPO No resection (RNS placement of R frontoparietal area) NA Stimulation (sensorimotor) 5176 Sz reduction by 50% 9 M 13 6 OXC, LTG, CLB 72 5 6.94% 6 7 R P/ R P R FP R focal resection of parietal cortex FCD 2a Stimulation (sensorimotor) 1253 Sz recurrence after 10 days 10 F 9 7 CLB, OXC 108 43 39.81% 8 35 L F/ L FP L FTP L frontal lobectomy sparing sensorimotor cortex FCD 1c Stimulation (sensorimotor and language) NA Sz free for 20 months, then lost follow-up 11 F 8 7 LVT, LCM, CLB 66 14 21.21% 6 1 L T/ L TP L FTP L temporal lobectomy Multinodular and vacuolating neuronal tumor (MVNT) Stimulation (sensorimotor and language), gamma mapping 1639 Sz free 12 F 18 17 CLB, LTG 84 12 14.29% 3 25 L P/ L P L FTP L focal resection of parietal cortex Gliosis Stimulation (sensorimotor and language), gamma mapping 1423 Sz recurrence after 3 days 13 F 15 3 LCM, LVT, OXC 86 9 10.47% 4 4 R F/ R F R FTP R focal resection around frontal tumor Oligodendroglioma Stimulation (sensorimotor), gamma mapping 1436 Sz free 14 F 19 18 LTG, LCM, OXC, PPN 70 0 0.00% 4 37 R FP/ R FP R FTPO No resection (RNS placement of R frontoparietal area) NA Stimulation (sensorimotor) 1900 Sz reduction by 75% 15 F 15 15 TPM, LTG 102 60 58.82% 4 4 L TPO/ L TPO L FTPO L TPO FCD 2a, Gliosis Stimulation (sensorimotor and language), gamma mapping 6366 Sz free 16 M 6 6 CLB, OXC 104 41 39.42% 11 2 L PO/ L PO L FTPO L parietooccipital resection Ulegyria, FCD 3d, gliosis Stimulation (sensorimotor), gamma mapping 14494 Sz recurrence after 23 days 17 M 20 15 LCM, BVC, FBM 118 29 24.58% 12 5 L TO/ L TO L FTPO L temporal lobectomy plus RNS Gliosis Stimulation (sensorimotor and language), gamma mapping 2700 Sz recurrence after 4 days 18 M 12 5 CNZ, CLP, ECZ, LCM, LVT 112 0 0.00% 4 100 L P/ L TP L FTPO No resection (RNS placement of L sensorimotor cortex) NA Stimulation (sensorimotor) 777 Sz reduction by 75% 19 M 14 6 ECZ, CLB 128 0 0.00% 14 8 L T/ L TP LFTPO No resection (RNS placement of L temporal, parietal, and occipital area) NA Stimulation (sensorimotor and language), gamma mapping 2602 Sz reduction by 80% M: Male; F: Female; FRs: Fast ripples; NA: Not applicable; RNS: Responsive nerve stimulator; FCD: Focal cortical dysplasia; SOZ: Seizure onset zone; Sz: Seizure. L: Left; R: Right; F: Frontal; P: Parietal; T: Temporal; O: Occipital. CLB: Clobazam; LVT: Levetiracetam; LCM: Lacosamide; CNZ: Clonazepam; RFD: Rufinamide; PPN: Perampanel; LTG: Lamotrigine; OXC: Oxcarbazepine; PB: Phenobarbital: TPM: Topiramate; BVC: Brivaracetam; FBM: Felbamate; CLP: Clorazepate; ECZ: Eslicarbazepine. CONFLICTS OF INTEREST The authors have no conflicts of interest to disclose. REFERENCES 1. Bragin A , Engel J Jr. , Wilson CL , Fried I , Buzsáki G . High-frequency oscillations in human brain. Hippocampus 1999;9 :137–142.10226774 2. Gotman J High frequency oscillations: the new EEG frontier? Epilepsia 2010;51 Suppl 1 :63–65.20331719 3. Bragin A , Engel J Jr. , Wilson CL , Fried I , Mathern GW . Hippocampal and entorhinal cortex high-frequency oscillations (100−−500 Hz) in human epileptic brain and in kainic acid--treated rats with chronic seizures. Epilepsia 1999;40 :127–137.9952257 4. Staba RJ , Wilson CL , Bragin A , Jhung D , Fried I , Engel J Jr. High-frequency oscillations recorded in human medial temporal lobe during sleep. Annals of neurology 2004;56 :108–115.15236407 5. Jirsch JD , Urrestarazu E , LeVan P , Olivier A , Dubeau F , Gotman J . High-frequency oscillations during human focal seizures. Brain 2006;129 :1593–1608.16632553 6. Worrell GA , Parish L , Cranstoun SD , Jonas R , Baltuch G , Litt B . High-frequency oscillations and seizure generation in neocortical epilepsy. Brain 2004;127 :1496–1506.15155522 7. Jacobs J , Zijlmans M , Zelmann R , High-frequency electroencephalographic oscillations correlate with outcome of epilepsy surgery. Annals of neurology 2010;67 :209–220.20225281 8. Wu JY , Sankar R , Lerner JT , Matsumoto JH , Vinters HV , Mathern GW . Removing interictal fast ripples on electrocorticography linked with seizure freedom in children. Neurology 2010;75 :1686–1694.20926787 9. Akiyama T , McCoy B , Go CY , Focal resection of fast ripples on extraoperative intracranial EEG improves seizure outcome in pediatric epilepsy. Epilepsia 2011;52 :1802–1811.21801168 10. van ‘t Klooster MA , van Klink NE , Leijten FS , Residual fast ripples in the intraoperative corticogram predict epilepsy surgery outcome. Neurology 2015;85 :120–128.26070338 11. Fedele T , van ‘t Klooster M , Burnos S , Automatic detection of high frequency oscillations during epilepsy surgery predicts seizure outcome. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology 2016;127 :3066–3074.27472542 12. Jacobs J , Wu JY , Perucca P , Removing high-frequency oscillations: A prospective multicenter study on seizure outcome. Neurology 2018;91 :e1040–e1052.30120133 13. Roehri N , Pizzo F , Lagarde S , High-frequency oscillations are not better biomarkers of epileptogenic tissues than spikes. Annals of neurology 2018;83 :84–97.29244226 14. van ‘t Klooster MA , Leijten FS , Huiskamp G , High frequency oscillations in the intra-operative ECoG to guide epilepsy surgery (“The HFO Trial”): study protocol for a randomized controlled trial. Trials 2015;16 :422.26399310 15. Jrad N , Kachenoura A , Merlet I , Nica A , Benar CG , Wendling F . Classification of high frequency oscillations in epileptic intracerebral EEG. Annu Int Conf IEEE Eng Med Biol Soc 2015;2015 :574–577.26736327 16. Amiri M , Lina JM , Pizzo F , Gotman J . High Frequency Oscillations and spikes: Separating real HFOs from false oscillations. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology 2016;127 :187–196.26100149 17. Jrad N , Kachenoura A , Merlet I , Automatic Detection and Classification of High-Frequency Oscillations in Depth-EEG Signals. IEEE Trans Biomed Eng 2017;64 :2230–2240.28113293 18. Chaibi S , Lajnef T , Samet M , Jerbi K , Kachouri A . Detection of High Frequency Oscillations (HFOs) in the 80–500 Hz range in epilepsy recordings using decision tree analysis. International Image Processing, Applications and Systems Conference 2014:1–6. 19. Liu S , Gurses C , Sha Z , Stereotyped high-frequency oscillations discriminate seizure onset zones and critical functional cortex in focal epilepsy. Brain 2018;141 :713–730.29394328 20. Lundervold AS , Lundervold A . An overview of deep learning in medical imaging focusing on MRI. Z Med Phys 2019;29 :102–127.30553609 21. Jing J , Sun H , Kim JA , Development of Expert-Level Automated Detection of Epileptiform Discharges During Electroencephalogram Interpretation. JAMA neurology 2020;77 :103–108.31633740 22. Zhao B , Hu W , Zhang C , Integrated Automatic Detection, Classification and Imaging of High Frequency Oscillations With Stereoelectroencephalography. Front Neurosci 2020;14 :546.32581688 23. Zuo R , Wei J , Li X , Automated Detection of High-Frequency Oscillations in Epilepsy Based on a Convolutional Neural Network. Front Comput Neurosci 2019;13 :6.30809142 24. Karimi D , Dou H , Warfield SK , Gholipour A . Deep learning with noisy labels: Exploring techniques and remedies in medical image analysis. Med Image Anal 2020;65 :101759.32623277 25. Zhang Y , Lu Q , Monsoor T , Refining epileptogenic high-frequency oscillations using deep learning: a reverse engineering approach. Brain Commun 2022;4 :fcab267.35169696 26. Nariai H , Hussain SA , Bernardo D , Prospective observational study: Fast ripple localization delineates the epileptogenic zone. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology 2019;130 :2144–2152.31569042 27. Nakai Y , Jeong JW , Brown EC , Three- and four-dimensional mapping of speech and language in patients with epilepsy. Brain 2017;140 :1351–1370.28334963 28. Brown EC , Rothermel R , Nishida M , In vivo animation of auditory-language-induced gamma-oscillations in children with intractable focal epilepsy. Neuroimage 2008;41 :1120–1131.18455440 29. Koga S , Rothermel R , Juhasz C , Nagasawa T , Sood S , Asano E . Electrocorticographic correlates of cognitive control in a Stroop task-intracranial recording in epileptic patients. Human brain mapping 2011;32 :1580–1591.20845393 30. Kojima K , Brown EC , Rothermel R , Clinical significance and developmental changes of auditory-language-related gamma activity. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology 2013;124 :857–869.23141882 31. Kambara T , Sood S , Alqatan Z , Presurgical language mapping using event-related high-gamma activity: The Detroit procedure. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology 2018;129 :145–154.29190521 32. Kuroda N , Sonoda M , Miyakoshi M , Objective interictal electrophysiology biomarkers optimize prediction of epilepsy surgery outcome. Brain Commun 2021;3 :fcab042.33959709 33. Staba RJ , Wilson CL , Bragin A , Fried I , Engel J Jr. Quantitative analysis of high-frequency oscillations (80–500 Hz) recorded in human epileptic hippocampus and entorhinal cortex. J Neurophysiol 2002;88 :1743–1752.12364503 34. Navarrete M , Alvarado-Rojas C , Le Van Quyen M , Valderrama M . RIPPLELAB: A Comprehensive Application for the Detection, Analysis and Classification of High Frequency Oscillations in Electroencephalographic Signals. PLoS One 2016;11 :e0158276.27341033 35. Jefferys JG , Menendez de la Prida L , Wendling F , Mechanisms of physiological and epileptic HFO generation. Prog Neurobiol 2012;98 :250–264.22420980 36. Zijlmans M , Jiruska P , Zelmann R , Leijten FS , Jefferys JG , Gotman J . High-frequency oscillations as a new biomarker in epilepsy. Annals of neurology 2012;71 :169–178.22367988 37. Matsumoto A , Brinkmann BH , Matthew Stead S , Pathological and physiological high-frequency oscillations in focal human epilepsy. J Neurophysiol 2013;110 :1958–1964.23926038 38. Cimbalnik J , Brinkmann B , Kremen V , Physiological and pathological high frequency oscillations in focal epilepsy. Ann Clin Transl Neurol 2018;5 :1062–1076.30250863 39. Weiss SA , Orosz I , Salamon N , Ripples on spikes show increased phase-amplitude coupling in mesial temporal lobe epilepsy seizure-onset zones. Epilepsia 2016;57 :1916–1930.27723936 40. Spring AM , Pittman DJ , Aghakhani Y , Interrater reliability of visually evaluated high frequency oscillations. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology 2017;128 :433–441.28160749 41. Nariai H , Wu JY , Bernardo D , Fallah A , Sankar R , Hussain SA . Interrater reliability in visual identification of interictal high-frequency oscillations on electrocorticography and scalp EEG. Epilepsia open 2018;3 :127–132. 42. Frauscher B , von Ellenrieder N , Zelmann R , High-Frequency Oscillations in the Normal Human Brain. Annals of neurology 2018;84 :374–385.30051505 43. Guragain H , Cimbalnik J , Stead M , Spatial variation in high-frequency oscillation rates and amplitudes in intracranial EEG. Neurology 2018;90 :e639–e646.29367441 44. Zweiphenning W , von Ellenrieder N , Dubeau F , Correcting for physiological ripples improves epileptic focus identification and outcome prediction. Epilepsia 2022;63 :483–496.34919741 45. von Ellenrieder N , Dubeau F , Gotman J , Frauscher B . Physiological and pathological high-frequency oscillations have distinct sleep-homeostatic properties. NeuroImage Clinical 2017;14 :566–573.28337411

PMC010xxxxxx/PMC10364141.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9422759 2553 Occup Environ Med Occup Environ Med Occupational and environmental medicine 1351-0711 1470-7926 30760604 10364141 10.1136/oemed-2018-105327 NIHMS1914831 Article Case-control investigation of occupational lead exposure and kidney cancer Callahan Catherine L 1* Friesen Melissa C 1 Locke Sarah J 1 Dopart Pamela J 1 Stewart Patricia A 2 Schwartz Kendra 3 Ruterbusch Julie J 3 Graubard Barry I 4 Chow Wong-Ho 5 Rothman Nathaniel 1 Hofmann Jonathan N 1 Purdue Mark P 1 1 Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology Genetics, National Cancer Institute, Bethesda, MD, USA 2 Stewart Exposure Assessments, LLC, Arlington, VA, USA 3 Department of Family Medicine and Public Health Sciences, Karmanos Cancer Institute, Wayne State University, Detroit, MI, USA 4 Biostatistics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD, USA 5 Department of Epidemiology, The University of Texas MD Anderson Cancer Center, Houston, Texas, United States of America * Corresponding author: Catherine.callahan@nih.gov, Phone: 240-276-5040, 9609 Medical Center Drive, Rockville, MD 20850 17 7 2023 7 2019 13 2 2019 24 7 2023 76 7 433440 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Objectives: Lead is a suspected carcinogen that has been inconsistently associated with kidney cancer. To clarify this relationship, we conducted an analysis of occupational lead exposure within a population-based study of kidney cancer using detailed exposure assessment methods. Methods: Study participants (1,217 cases and 1,235 controls), enrolled between 2002 and 2007, provided information on their occupational histories and, for selected lead-related occupations, answered questions regarding workplace tasks, and use of protective equipment. Industrial hygienists used this information to develop several estimates of occupational lead exposure, including probability, duration, and cumulative exposure. Unconditional logistic regression was used to compute odds ratios (ORs) and 95% confidence intervals (CIs) for different exposure metrics, with unexposed subjects serving as the reference group. Analyses were also conducted stratifying on several factors, including, for subjects of European ancestry only, single nucleotide polymorphisms in ALAD (rs1805313, rs1800435, rs8177796, rs2761016), a gene involved in lead toxicokinetics. Results: In our study, cumulative occupational lead exposure was not associated with kidney cancer (OR 0.9, 95% CI 0.7–1.3 for highest quartile vs. unexposed; Ptrend = 0.80). Other lead exposure metrics were similarly null. We observed no evidence of effect modification for the evaluated ALAD variants and most stratifying factors, although lead exposure was associated with increased risk among never smokers. Conclusions: The findings of this study do not offer clear support for an association between occupational lead exposure and kidney cancer. kidney cancer occupational lead exposure cancer epidemiology pmcIntroduction The elimination of leaded gasoline in the US and other countries has substantially reduced environmental lead levels. However, high exposures continue to occur in several occupations (battery-production workers, lead smelter and refinery workers, leaded-glass workers, pigment workers, construction workers, and radiator-repair workers) and some communities, such as recent concerns regarding contaminated water in Flint, Michigan.1 2 Exposure to lead has several established deleterious health effects, including neurotoxicity, kidney damage, anemia, hypertension, and miscarriage.3 Lead is also a suspected carcinogen. In 2006, the International Agency for Research on Cancer (IARC) classified inorganic lead as a probable carcinogen (group 2A), noting sufficient evidence in experimental animals for kidney tumors and other malignancies and suggestive epidemiologic evidence of associations with cancers of the brain, kidney, lung, and stomach.1 Organic lead was not classifiable with regards to carcinogenicity (group 3).1 Since the 2006 IARC review, lead exposure has been associated with kidney cancer in some additional studies,4–6 but not others.7–9 Many studies were hindered by an inability to adjust for smoking as a confounder and/or limited exposure assessment. Findings from one study suggest that kidney cancer and lead association is modified by single-nucleotide polymorphisms (SNPs; rs8177796 and rs2761016) mapping to ALAD, which encodes a heme biosynthesis enzyme (5-aminolevulinic acid dehydratase)10 implicated in lead-induced anemia toxicity.11 Another ALAD SNP, rs1805313, has been significantly associated with blood lead levels (p= 3.9 × 10−14) in a genome-wide association study (GWAS);12 this variant has not been evaluated as a possible effect modifier for lead exposure and kidney cancer. To clarify the association between occupational lead exposure and kidney cancer and further evaluate the possible interplay of four SNPs within ALAD, we conducted an analysis within the US Kidney Cancer Study (USKC), a population-based case-control study featuring detailed information on workplace exposures and GWAS data. Methods The methods of the USKC have been described previously.13 We conducted a population-based case-control study in Detroit, MI (Wayne, Oakland, and Macomb Counties) and Chicago, IL (Cook County). Cases of renal cell carcinoma, which make up over 90% of all kidney cancers, were identified in Detroit via the Metropolitan Detroit Cancer Surveillance System from February 2002 until July 2006 (white cases) or January 2007 (black cases). Chicago kidney cancer cases were identified through a review of pathology reports from 56 hospitals in Cook County. The sampling strategy was designed to efficiently recruit a sufficient number of black cases and controls for analyses stratified by race while not exceeding recruitment goals for whites. Thus, all black cases were sampled and white cases were subsampled depending on age and sex strata for recruitment. Kidney cancer diagnoses were confirmed from medical records and histology was assigned from central histopathological review if available (706 cases) or from the original diagnostic pathology reports. Controls for each center were identified through Department of Motor Vehicles (DMV) records (age 20–64) or Medicare eligibility files (age 65–79). Controls were frequency matched on sex, age (five-year intervals), and race at a 2:1 ratio for black cases and a 1:1 ratio for white cases. To attain the matching ratio for black cases, census block groups with a high density of black people were intentionally oversampled. Of the 1,571 cases we attempted to recruit, 1,217 participated (77%) and 1,235 controls of the 2,269 controls we recruited participated (54%). The study was approved by Institutional Review Boards at all institutions, and written informed consent was obtained from all participants before interview. Exposure assessment Each participant was administered a computer assisted personal interview in the home by a trained interviewer. Prior to the home visit, participants were mailed a work history calendar asking them to list each job they held for at least 12 months since the age of 16. During the interview, participants were asked open-ended questions regarding the usual number of hours worked per week, type of business or service, tasks, and a description of chemicals/materials and equipment used for each job. All jobs were then coded using the Standard Industrial Classification (SIC) and Standard Occupational Classification (SOC) systems.14 15 For selected occupations, one of 39 job- or industry-specific interview modules were administered. Metal-relevant modules queried detailed information on each job (e.g. handyman) or industry (e.g. foundry) regarding average frequency of various metal-related tasks, work practices, and in some modules, the potential for dermal exposure. A maximum of five modules were administered per participant to reduce participant burden. Modules were assigned based on length of time a job was held (longest held job to shortest). We developed a three-phase process to assign estimates of the probability, frequency, and intensity of occupational exposure to lead using information from the occupational history and module responses. This process is described in detail in Callahan et al.16 In the first phase, probability of exposure to ten lead sources (paint, gasoline, grinding metal, solder, welding, engine repair, pigments, guns, printing, and “other”) was assigned using hierarchial decision rules for each job a subject reported in their work history. The decision rules to assign probability of exposure were based on SIC/SOC codes, variables systematically extracted from free-text responses in the lifetime occupational history,17 and module responses. The source-specific lead probability for each job was assigned to one of four categories [none (0 to <5% of workers exposed), low (≥5% to <50% of workers exposed), medium (≥50 to 80% of workers exposed) or high (≥80% of workers exposed)]. In the second phase, jobs with a ≥50% probability of lead exposure were assigned source- and time-specific estimates of lead exposure frequency and intensity. Frequency estimates (number of hours performing activity) were prioritized as: 1) the number of hours the subject reported for a given lead activity from the module responses; 2) the job-means of the module responses applied to jobs with missing frequency; and 3) expert judgment when job-means were not available. The source-specific lead intensity estimates were based on a comprehensive review of published blood lead measurements, which included evaluations of weighted arithmetic means reported by 351 sets of summary statistics;18 industry-specific temporal evaluations,19 and detailed statistical analysis of data associated with activities disturbing materials painted with or containing lead.20 “Worst-case” exposure sampling scenarios (e.g. elevated blood lead levels, employee concerns, regulatory violations) were excluded from the database of published blood lead measurements. For each lead source (s), the source-specific blood lead concentration for each year, y, from occupational sources was calculated as described below in equation 1. The base intensity estimates were assigned using programmable decision rules for all jobs with medium or high source-specific ratings, with the following exceptions. Expert review was used to assign intensity and frequency estimates for jobs with medium or high probability of lead exposure from ‘printing’, ‘pigment’, and ‘other’ lead sources [122 jobs]. Additionally, an expert reviewed the base intensity and frequency estimates for 30 jobs with a module and 162 jobs without a module for which, when compared to jobs in the same occupation group, the frequency or concentration estimate were above the 75th percentile of estimates or when an atypical module response was observed. To remove assumed environmental lead exposure, the median adult population-level blood lead estimated from National Health and Nutrition Examination Survey (NHANES) data was subtracted from the occupation or activity-specific blood lead estimate. We also normalized the concentration estimate to a 2080-hour work year (=1 full time equivalent (FTE) based on 40 hours per week and 52 weeks per year) to obtain units of μg/dL-FTE. Equation 1: Concentrationsy(μg/dL–FTE)=(BaseIntensityEstimatefor1980s*TimeTrendModifiery–NHANESMedianBloodLeadEstimatey)*#hoursperformingactivity/40hoursperweekx52weeksperyear In the third phase, source-specific metrics were integrated to develop several subject-level lead exposure metrics. Overall probability of lead exposure was assigned as the highest probability rating from any source at any job. Eighty-seven participants with an incomplete occupational history were categorized as having low (≥5–50%) probability of exposure. For participants with ≥50% overall probability of lead exposure the following metrics were assigned: Duration of exposure (years): defined as the sum of the number of years worked at each job with an exposure probability to any lead source≥50% Average yearly exposure (μg/dL): defined as the sum of estimated blood lead concentrations divided by the sum of years worked at each job with an exposure probability ≥50% Maximum lead exposure (μg/dL): defined as the highest annual estimated blood lead concentration from any job with an exposure probability ≥50% Cumulative exposure (μg/dL-FTE): defined as the sum of estimated blood lead concentrations for all jobs with an exposure probability ≥50% We also calculated the exposure metrics described above based only on jobs with ≥80% probability of exposure. All exposure metrics incorporated a five-year lag, metrics incorporating a ten- and twenty-year lag were also developed. Genotyping Genomic DNA extracted from buccal cell samples (89% of participants) and blood (80%) was collected from 1,087 cases (89%) and 1,091 controls (89%). Of these, 662 cases and 561 controls of European ancestry were genotyped for 585,576 SNPs as part of a GWAS of kidney cancer,21 including the ALAD SNPs rs1805313 and rs2761016. We imputed the ALAD SNPs rs1800435 and rs8177796 (imputation information scores 0.98 and 0.84 respectively) using IMPUTE2 version 2.2.2,22 with 1000 Genomes Project data (phase 1 release 3) used as a reference set. We included 652 cases and 559 controls with both lead exposure and genetic data in our analyses investigating gene*environment interactions. Statistical analysis We developed sample weights to reduce the potential for bias from differential sampling rates for cases and controls. Sample weights for controls also included a poststratification adjustment so that the weighted distribution of controls across the matching variables matched exactly the weighted distribution of cases. Poststratification adjustment reduces the variability of the weights and is consistent with the objective of frequency matching.23 Case-control differences across metrics of exposure were assessed using odds ratios (ORs) and 95% confidence intervals (CIs) estimated from unconditional logistic regression modeling with adjustment for the poststratification weights using the jackknife replicate weight method to estimate standard errors.24 Models were adjusted for age at reference date (20–44, 45–54, 55–64, 65–74, ≥75 years), race, study center, sex, education (<12 years, high school graduate, some college, 4+ years of college), smoking status two years before the reference date [never, occasional (smoked more than 100 cigarettes but never smoked one cigarette daily for 6 months or could not provide a definitive answer), former regular, or current regular], body mass index (BMI) based on height at interview and weight 5 years prior to interview, <25, 25–<30, 30–<35, 35kg/m2, unknown), and self-reported history of hypertension as of two years prior to the reference date. Analyses of cumulative exposure, years of exposure, average yearly exposure, and maximum yearly exposure were based on jobs with a ≥80% probability of lead exposure; these metrics were categorized using quartiles of the exposed controls as cut-points. Trends were tested for statistical significance by modelling the intra-category medians as a continuous variable, with values for unexposed participants set to zero. We also conducted sensitivity analyses including cumulative analyses of specific sources of lead exposure, stratifying by study region (Detroit vs. Chicago), stratifying by age (<65 vs. >65 years), using tertiles or quintiles as cut-points, including jobs with a medium (50–89%) probability of exposure, incorporating ten- or twenty-year lags, excluding exposures that occurred for less than 0.5 hours per week, excluding exposures from leaded gas, and excluding hypertension as a covariate. We further evaluated the joint effects of lead and ALAD SNP genotype (rs1805313, rs1800435, rs8177796, and rs2761016), smoking, sex, race, study center, history of hypertension, body mass index, age group, or histologic subtype (clear cell vs. other). Potential interaction on the multiplicative scale was assessed by the significance of interaction terms using the Wald χ2 test. All analyses were conducted with SAS 9.3 using procedures appropriate for sample-weighted data. Statistical tests were determined to be significant at a two-sided p-value <0.05 that was not corrected for multiple comparisons. Results Characteristics of cases and controls are presented in Table 1. Cases generally had a higher BMI and less education than controls and were more likely to have a history of hypertension and smoking. Forty-five percent of USKC participants had a greater than 80% likelihood of occupational exposure to lead (Table 2). Of jobs with ≥80% probability of exposure the most common industries were manufacturing (28%), public administration (17%), services (13%), retail (12%), and transportation (13%). Of the public administration jobs with a >80% probability of lead exposure, 83% were national security. The industries with the highest median estimated cumulative blood lead from occupational exposures were mining (6.4 μg/dL-FTE), construction (1.9 μg/dL-FTE), and manufacturing (1.6 μg/dL-FTE). As presented in Table 2, there was no difference in lead exposure probability between cases and controls (≥80% probability of exposure vs. non-exposed: OR 1.0, 95% CI 0.8, 1.2). Analyses of cumulative, average, maximum, and duration of lead exposure were also null (Table 2). Source-specific analyses of cumulative exposure were similarly null (Supplemental Table 1). Only one subject had greater than 50% exposure to pigments, thus we did not assess pigments separately. We observed similar results in the following sensitivity analyses: stratifying by study region (Detroit vs. Chicago), stratifying by age (<65 vs. ≥65 years), including jobs with greater than 50% likelihood of exposure in analyses of duration and cumulative lead exposure, excluding exposures that occurred for less than 0.5 hours per week, excluding exposures from leaded gas (the major source of organic lead in this population), incorporating ten- or twenty-year lags, not adjusting for hypertension, and parameterizing continuous exposure metrics as tertiles or quintiles (results not shown). None of the four ALAD SNPs were significantly associated with kidney cancer, although we did observe a suggestive association with rs8177796 (OR per rare-allele 1.3, 95%CI 1.0,1.8; p=0.08, Supplemental Table 2). We did not observe evidence of effect modification from these SNPs for cumulative lead exposure (Table 3) or other metrics (Supplemental Table 3). In joint analyses, we observed evidence of effect modification in the association between lead exposure and kidney cancer by smoking status; among never or occasional smokers, we observed associations with increased risk for the highest categories of duration, maximum, and cumulative exposure, while among current smokers, these measures were not associated with risk (Table 4). We did not observe evidence of effect modification by sex, race, study center, history of hypertension, body mass index, age group, or histologic subtype (clear cell vs. other) (results not shown). Discussion The results of this large population-based case-control study with detailed exposure assessment do not provide clear support for the hypothesis that occupational lead exposure is associated with kidney cancer. We did not observe evidence that the association between lead and kidney cancer was modified by variation in four ALAD SNPs, although we did observe evidence of effect modification by smoking status where there was an association between duration of, cumulative, and maximum lead exposure and increased risk of kidney cancer among never smokers only. Findings from the six prior epidemiologic studies evaluating lead exposure and kidney cancer are inconsistent. Two case-control studies that conducted expert-based assessments of lead exposure had conflicting results; occupational lead exposure was associated with kidney cancer in a large multi-center study conducted in central and Eastern Europe5, but not in a smaller study of male participants in Canada.7 Southard et al. conducted a nested case-control study of Finnish male smokers and reported that higher blood lead levels, measured from samples drawn between five years and two months prior to kidney cancer diagnosis, were associated with increased risk of kidney cancer.6 Since smoking is a predictor of blood lead,25 the potential for residual confounding by other kidney carcinogens in cigarette smoke cannot be ruled out as an explanation for these findings. Exposure to lead, as estimated by a job-exposure matrix, was associated with a non-significant increase in kidney cancer risk among participants in the Shanghai Men’s and Women’s Health Study Cohorts.4 Kidney cancer has been assessed in two occupational cohort studies of lead-exposed workers that did not adjust for smoking. A retrospective cohort study of male lead-exposed workers in Australia considered both kidney cancer incidence and mortality, neither of which were elevated when compared to the general population of Australia.8 Kidney cancer mortality was not elevated in an occupational cohort of workers in Great Britain and was not associated with measured blood lead.9 Our findings provide additional evidence not supportive of an association between lead exposure and kidney cancer. In analyses restricted to white subjects, we observed no evidence that the four ALAD SNPs we assessed modify the observed null association between lead exposure and kidney cancer. In a previous investigation of lead exposure and ALAD variants within the kidney cancer case-control study conducted in Central and Eastern Europe, van Bemmel et al. reported that the minor allele (T) of rs8177796CT/TT was associated with kidney cancer in their study population and that variation in rs2761016 appeared to modify the lead- kidney cancer association, with a stronger association observed among participants with the GG or GA genotype than the AA genotype.10 We observed a non-significant association for rs8177796 in the same direction, but did not otherwise replicate their findings. Variation in rs1800435, which has been reported to alter susceptibility to lead poisoning,11 did not modify the association between lead exposure and kidney cancer in our study, the study by van Bemmel et al.10, nor in the nested case-control study of Finnish smokers by Southard et al.6. Our analysis of rs1805313, which was recently associated with blood lead in a GWAS,12 was similarly null. Although we cannot rule out the existence of subtle patterns of interaction given study power limitations, our findings do not support the existence of lead-kidney cancer effect modification by ALAD variants. In the most notable finding from our joint analyses, we observed associations with increased kidney cancer risk for high lead exposure among non-smokers only. Interestingly, other studies have reported the associations between blood lead and periodontitis,26 cardiovascular disease27 and chronic kidney disease28 to similarly be present only among non-smokers. It is unclear what biologic mechanism would underlie this finding. Metals can interact in a variety of ways. For instance, experimental studies have indicated that of cadmium, another metal found in cigarette smoke may diminish the biologic effects of lead.29 30 However, as we conducted secondary analyses exploring possible effect modification across several factors, we cannot rule out the possibility that these findings arose due to chance.31 It is important for this result to be confirmed in an independent study before meaningful causal inferences can be drawn. Our study has several important strengths. Our detailed retrospective assessment of occupational lead exposure, involving in-depth subject-specific information on workplace exposure determinants assessed by a team of industrial hygienists, likely minimized bias from exposure misclassification as compared to cruder exposure assessment approaches employed in some past studies. Study participants provided detailed information on potential confounders, such as smoking, BMI, and hypertension, which were not addressed in previous industry-based studies.8 9 The opportunity to incorporate genetic information on ALAD variants to assess gene*environment interaction is another strength. Our study also had limitations. We did not directly measure levels of lead in kidney tissue or blood. In addition to the occupational exposures we considered, several other sources of lead exposure such as diet, genetic variability other than ALAD, and environmental exposure to leaded gas were not considered in our analyses. Furthermore, Black Americans have consistently higher levels of blood lead in NHANES than white Americans32, which suggests a potential for residual confounding by race; however we did not observe any significant results in race-stratified analyses. Although we used a detailed exposure assessment approach, given the inherent reliance on participants’ recall of past occupational events, we cannot eliminate the possibility that exposure misclassification could have biased results towards the null. We did not validate our exposure assessment approach. To reduce the impact of potential exposure misclassification, we restricted analyses to participants with ≥80% probability of exposure and conducted several sensitivity analyses. Despite using this stringent exposure definition, we did not observe even a suggestion of an association with kidney cancer for the highest levels of the lead exposure metrics. Our exposure assessment method did not distinguish between organic and inorganic lead; we note that analyses that excluding exposure to leaded gas, the main source of organic lead in our population, were similar to our main analyses. Estimates of lead exposure intensity were based on published data from US work sites, which despite the exclusion of worst-case events, may not be representative of typical workplace exposures.18 Another limitation was the low response rate among controls, which is typical of recent population-based case-control studies. The use of sample weights can help to reduce the potential for bias arising from nonresponse, as the weights account for differential nonresponse across subgroups defined by factors such as age, sex, and county of residence, for which data were available for both respondents and nonrespondents. However, we cannot entirely rule out the possibility that nonresponse bias or residual confounding influenced our results. Although our overall study size was relatively large, we had limited statistical power to detect effect modification, particularly in the analyses with ALAD variants. In conclusion, our large population-based study that used detailed exposure assessment methods does not offer clear support for an association between occupational lead exposure and kidney cancer, nor an interaction with ALAD variants. Our finding suggesting effect modification by smoking status warrants further investigation. Supplementary Material Supplementary Material Table 1. Selected characteristics of kidney cancer cases and controls in the US Kidney Cancer Study Cases Controls Age (years) N (%)a N (%)a   20–44 147 (12) 179 (12)   45–54 287 (24) 270 (22)   55–64 372 (31) 350 (28)   65–74 303 (25) 329 (27)   ≥75 108 (9) 107 (9) Sex   Male 720 (59) 689 (56)   Female 497 (41) 546 (44) Race   White 856 (70) 712 (58)   Black 361 (30) 523 (42) Study center   Chicago 199 (16) 197 (16)   Detroit 1018 (84) 1038 (84) Education   ≤11 years 200 (16) 165 (13)   12 years 419 (34) 390 (32)   1–3 years’ college 328 (27) 356 (29)   4+ years’ college 270 (22) 324 (26) Smoking status   Never 432 (36) 471 (38)   Occasional 55 (5) 55 (5)   Regular, former 410 (34) 445 (36)   Regular, current 320 (26) 264 (21) Body mass index (kg/m2)   <25 240 (20) 366 (30)   25–30 436 (36) 493 (40)   30–35 298 (25) 221 (18)   ≥35 230 (19) 147 (12)   Unknown 13 (1) 8 (1) History of hypertension   No 500 (41) 718 (58)   Yes 701 (58) 508 (41)   Unknown 16 (1) 9 (1) Age at first job with mean (SD) mean (SD) ≥50% probability of lead exposure 22.5 (7.7) 22.6 (8.0) ≥80% probability of lead exposure 22.2 (7.3) 22.4 (7.9) a percentages adjusted for post-stratified sample weights. Abbreviations: SD, standard deviation. Table 2. Estimated occupational lead exposure and kidney cancer risk, US Kidney Cancer Study Cases Controls ORb (95% CI) Exposure probability n (%)a n (%)a   Unexposed 432 (34) 490 (38) 1.0   <50% 193 (16) 184 (14) 1.2 (0.9, 1.6)   50–80% 56 (5) 44 (4) 1.2 (0.8, 1.9)   ≥80% 536 (45) 517 (45) 1.0 (0.8, 1.2) Average yearly exposure (μg/dL) c   Unexposed 432 (43) 490 (46) 1.0   ≤1.2 111 (11) 128 (14) 0.8 (0.6, 1.2)   1.3–2.9 146 (16) 130 (14) 1.1 (0.8, 1.5)   3.0–6.6 167 (18) 129 (13) 1.3 (0.9, 1.7)   ≥6.7 112 (12) 130 (13) 0.8 (0.6, 1.1)   p for trend 0.32 Duration of lead exposure (years)d   Unexposed 432 (43) 490 (46) 1.0   ≤4 153 (16) 156 (16) 1.0 (0.8, 1.4)   4.5–8.5 92 (10) 103 (11) 0.9 (0.6, 1.3)   9–20.5 151 (16) 130 (13) 1.0 (0.7, 1.4)   >20.5 140 (15) 128 (14) 1.0 (0.7, 1.4)   p for trend 0.98 Maximum exposure (μg/dL)e   Unexposed 432 (42) 490 (45) 1.0   <1.6 113 (11) 134 (14) 0.8 (0.6, 1.1)   1.6–3.8 131 (14) 134 (14) 1.1 (0.8, 1.5)   3.9–9.3 147 (16) 123 (13) 1.2 (0.9, 1.6)   ≥9.4 162 (17) 146 (15) 1.0 (0.7, 1.3)   p for trend 0.76 Cumulative lead exposure (μg/dL-FTE )f   Unexposed 432 (42) 490 (46) 1.0   ≤6.7 123 (13) 129 (13) 1.0 (0.7, 1.4)   6.8–21.2 115 (12) 129 (14) 0.9 (0.6, 1.3)   21.3–78.4 163 (18) 129 (14) 1.2 (0.9, 1.6)   ≥78.5 135 (15) 130 (13) 0.9 (0.7, 1.3)   p for trend 0.66 a percentages are adjusted for poststratified sample weights. b Adjusted for age, race, study center, sex, education, smoking status, body mass index, and history of hypertension; weighted with poststratified sample weights. c Estimated average annual blood lead (μg/dL) from all high probability sources of exposure, cut-points based on quartiles of exposed controls. d Years of high probability of exposure to lead, cut-points based on quartiles of exposed controls. e Maximum estimated blood lead, cut-points based on quartiles of exposed controls. f Estimated cumulative blood lead (μg/dL) from all high probability sources of exposure, cut-points based on quartiles of exposed controls. Table 3. Joint analysis of cumulative occupational lead exposure (μg/dL), ALAD genotype, and kidney cancer risk, white US Kidney Cancer Study subjects only. SNP Nca Nco ORa (95%CI) Nca Nco ORa (95% CI) Nca Nco ORa (95% CI) Pinteraction rs1805313 AA GA GG  Unexposed 118 81 1.0 referent 104 101 0.8 (0.6, 1.2) 30 30 0.9 (0.5, 1.5) 0.21  ≤6.7 31 35 0.8 (0.4, 1.5) 31 27 0.9 (0.5, 1.6) 6 7 0.7 (0.3, 1.7)  6.8–21.2 22 33 0.5 (0.3, 1.0) 30 23 1.2 (0.6, 2.3) 7 8 0.6 (0.2, 2.2)  21.3–78.4 38 23 1.2 (0.7, 2.2) 40 25 1.2 (0.6, 2.5) 11 9 0.7 (0.3, 1.7)  ≥78.5 29 33 0.7 (0.4, 1.2) 28 29 0.5 (0.2, 1.1) 9 1 9.2 (2.0, 43.3)  Ptrendb 0.48 0.14 0.04 rs1800435 CC CG GG  Unexposed 208 185 1.0 referent 42 25 1.3 (0.8, 1.9) 2 3 0.9 (0.2, 4.0) 0.14  ≤6.7 56 57 0.9 (0.6, 1.4) 11 12 0.9 (0.4, 2.2) 1 0  6.8–21.2 49 54 0.9 (0.5, 1.4) 10 8 1.1 (0.5, 2.4) 0 2  21.3–78.4 82 45 1.4 (0.9, 2.4) 6 10 0.4 (0.2, 1.2) 1 2  ≥78.5 55 52 0.9 (0.5, 1.4) 11 11 0.5 (0.2, 1.3) 0 0  Ptrendb 0.52 0.02 rs8177796 GG GA AA  Unexposed 215 186 1.0 referent 35 26 1.4 (0.9, 2.2) 2 1 0.53  ≤6.7 56 57 1.0 (0.7, 1.5) 12 12 0.7 (0.3, 1.8) 0 0  6.8–21.2 49 53 0.8 (0.5, 1.3) 10 10 1.3 (0.5, 3.4) 0 1  21.3–78.4 76 48 1.2 (0.7, 2.0) 13 9 1.8 (0.7, 4.7) 0 0  ≥78.5 55 56 0.8 (0.5, 1.2) 11 7 1.3 (0.5, 3.4) 0 0  Ptrendb 0.15 0.71 rs2761016 CC CA AA  Unexposed 69 59 1.0 referent 139 112 1.0 (0.6, 1.5) 44 42 0.8 (0.5, 1.3) 0.31  ≤6.7 16 19 0.7 (0.3, 1.6) 36 35 0.9 (0.5, 1.6) 16 15 1.0 (0.5, 2.3)  6.8–21.2 19 20 0.9 (0.4, 2.1) 30 32 0.8 (0.4, 1.5) 10 12 0.7 (0.3, 1.5)  21.3–78.4 26 20 0.8 (0.4, 1.5) 39 28 1.2 (0.7, 2.2) 24 9 2.0 (0.8, 5.1)  ≥78.5 21 20 0.7 (0.4, 1.4) 34 32 0.7 (0.4, 1.3) 11 11 1.0 (0.4, 2.5)  Ptrendb 0.31 0.52 0.66 a Adjusted for age, sex, education, body mass index, study center, and history of hypertension; weighted with poststratified sample weights, cut-points based on quartiles of exposed controls. b Tests of trend based on modeling the inta-category medians in genotype-stratified analyses. Table 4. Joint analysis of estimated occupational lead exposure, smoking status, and kidney cancer risk, US Kidney Cancer Study Nca Nco ORa (95%CI) Nca Nco ORa (95%CI) Nca Nco ORa (95%CI) Pinteraction Lead exposure Never/Occasional Former Current Probability Unexposed 205 265 1.0 referent 133 149 1.2 (0.8, 1.6) 94 76 1.8 (1.1, 2.7) 0.41 <50% 87 83 1.4 (1.0, 2.1) 56 61 1.3 (0.8, 2.1) 50 40 1.8 (1.1, 2.9) 50–80% 20 22 1.0 (0.5, 2.0) 17 16 1.2 (0.5, 2.9) 19 6 4.5 (1.6, 12.5) ≥80% 175 156 1.2 (0.9, 1.7) 204 219 0.9 (0.7, 1.3) 157 142 1.3 (0.9, 2.0) Duration (years)b Unexposed 205 265 1.0 referent 133 149 1.1 (0.8, 1.4) 94 76 1.7 (1.2, 2.5) 0.08 ≤4 53 59 1.1 (0.7, 1.7) 59 61 1.1 (0.7, 1.8) 41 36 1.4 (0.8, 2.5) 4.5–8.5 22 20 1.1 (0.6, 2.1) 39 48 0.8 (0.5, 1.4) 31 35 1.0 (0.5, 1.8) 9–20.5 44 40 1.2 (0.7, 1.9) 54 46 1.2 (0.8, 1.8) 53 44 1.5 (0.8, 2.6) >20.5 56 37 1.6 (0.9, 2.8) 52 64 0.9 (0.5, 1.4) 32 27 1.6 (0.9, 2.8) Ptrende 0.09 0.35 0.18 Maximum (μg/dL)c Unexposed 205 265 1.0 referent 133 149 1.1 (0.8, 1.4) 94 76 1.7 (1.2, 2.5) 0.02 <1.6 43 58 0.9 (0.5, 1.5) 32 48 0.6 (0.4, 1.1) 38 28 1.6 (0.9, 2.9) 1.6–3.8 45 46 1.3 (0.8, 2.1) 49 58 1.0 (0.6, 1.5) 37 30 1.6 (0.9, 2.9) 3.9–9.3 47 34 1.4 (0.8, 2.4) 60 48 1.4 (0.9, 2.2) 40 41 1.3 (0.7, 2.2) ≥9.4 47 31 1.6 (0.9, 2.9) 67 69 0.9 (0.6, 1.4) 48 46 1.1 (0.7, 1.8) Ptrende 0.04 0.88 0.02 Cumulative (μg/dL-FTE)d Unexposed 205 265 1.0 referent 133 149 1.1 (0.8, 1.4) 94 76 1.7 (1.2, 2.5) 0.03 ≤6.7 45 50 1.0 (0.6, 1.6) 42 53 0.8 (0.5, 1.3) 36 26 1.9 (1.0, 3.5) 6.8–21.2 37 44 1.1 (0.7, 1.9_ 44 49 0.9 (0.5, 1.5) 34 36 1.3 (0.7, 2.5) 21.3–78.4 48 33 1.6 (1.0, 2.6) 60 54 1.3 (0.8, 2.0) 55 42 1.3 (0.8, 2.2) ≥78.5 45 29 1.4 (0.8, 2.7) 58 36 0.9 (0.6, 1.4) 32 38 1.1 (0.6, 1.9) Ptrende 0.18 0.77 0.03 Abbreviations: Nca, number of cases; Nco, number of controls; OR, odds ratio; CI, confidence interval; FTE, full time equivalent a Adjusted for age, sex, race, education, body mass index, study center, and history of hypertension; weighted with poststratified sample weights. b Years of high probability of exposure to lead, cut-points based on quartiles of exposed controls. c maximum estimated high probability of exposure to blood lead, cut-points based on quartiles of exposed controls. c cumulative estimated high probability of exposure to blood lead, cut-points based on quartiles of exposed controls. e Tests of trend based on modeling the inta-category medians in smoking-stratified analyses. KEY FINDINGS What is already known about this subject? Lead has been inconsistently associated with kidney cancer in previous studies with limited exposure assessment. We employed detailed exposure assessment within a case-control study to investigate whether occupational lead exposure is associated with kidney cancer, and whether the association is modified by ALAD genetic polymorphisms affecting lead toxicokinetics. What are the new findings? In this study, exposure to lead was unrelated to kidney cancer, both overall and across ALAD variants, although an association with increased risk was observed among never smokers. How might this impact on policy or clinical practice in the foreseeable future? Inorganic lead was classified as a probable human carcinogen by the International Agency for Research on Cancer 2006, based on sufficient animal evidence and limited human evidence. Our findings will be informative for any future evaluation of carcinogenicity for this metal. References 1. Inorganic and organic lead compounds. IARC monographs on the evaluation of carcinogenic risks to humans / World Health Organization, International Agency for Research on Cancer 2006;87 :1–471. [published Online First: 2006/12/29] 17191367 2. Kennedy C , Yard E , Dignam T , Blood Lead Levels Among Children Aged <6 Years - Flint, Michigan, 2013–2016. MMWR Morbidity and mortality weekly report 2016;65 (25 ):650–4. doi: 10.15585/mmwr.mm6525e1 [published Online First: 2016/07/01] 27359350 3. ATSDR Toxicological Profile for Lead. 2007 4. Liao LM , Friesen MC , Xiang YB , Occupational Lead Exposure and Associations with Selected Cancers: The Shanghai Men’s and Women’s Health Study Cohorts. Environmental health perspectives 2016;124 (1 ):97–103. doi: 10.1289/ehp.1408171 [published Online First: 2015/06/20] 26091556 5. Boffetta P , Fontana L , Stewart P , Occupational exposure to arsenic, cadmium, chromium, lead and nickel, and renal cell carcinoma: a case-control study from Central and Eastern Europe. Occupational and environmental medicine 2011;68 (10 ):723–8. doi: 10.1136/oem.2010.056341 [published Online First: 2011/01/11] 21217163 6. Southard EB , Roff A , Fortugno T , Lead, calcium uptake, and related genetic variants in association with renal cell carcinoma risk in a cohort of male Finnish smokers. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology 2012;21 (1 ):191–201. doi: 10.1158/1055-9965.epi-11-0670 [published Online First: 2011/11/17] 7. Rousseau MC , Parent ME , Nadon L , Occupational exposure to lead compounds and risk of cancer among men: a population-based case-control study. Am J Epidemiol 2007;166 (9 ):1005–14. doi: 10.1093/aje/kwm183 [published Online First: 2007/08/11] 17690218 8. Gwini S , MacFarlane E , Del Monaco A , Cancer incidence, mortality, and blood lead levels among workers exposed to inorganic lead. Annals of epidemiology 2012;22 (4 ):270–6. doi: 10.1016/j.annepidem.2012.01.003 [published Online First: 2012/01/31] 22285868 9. McElvenny DM , Miller BG , MacCalman LA , Mortality of a cohort of workers in Great Britain with blood lead measurements. Occupational and environmental medicine 2015;72 (9 ):625–32. doi: 10.1136/oemed-2014-102637 [published Online First: 2015/04/16] 25872777 10. van Bemmel DM , Boffetta P , Liao LM , Comprehensive Analysis of 5-Aminolevulinic Acid Dehydrogenase (ALAD) Variants and Renal Cell Carcinoma Risk among Individuals Exposed to Lead. PLoS ONE 2011;6 (7 ):e20432. doi: 10.1371/journal.pone.0020432 21799727 11. Onalaja AO , Claudio L . Genetic susceptibility to lead poisoning. Environmental health perspectives 2000;108 (Suppl 1 ):23–28.10698721 12. Warrington NM , Zhu G , Dy V , Genome-wide association study of blood lead shows multiple associations near ALAD. Human molecular genetics 2015;24 (13 ):3871–9. doi: 10.1093/hmg/ddv112 [published Online First: 2015/03/31] 25820613 13. Colt JS , Schwartz K , Graubard BI , Hypertension and risk of renal cell carcinoma among white and black Americans. Epidemiology (Cambridge, Mass) 2011;22 (6 ):797–804. doi: 10.1097/EDE.0b013e3182300720 [published Online First: 2011/09/02] 21881515 14. US Department of Commerce. Standard Occupational Classification Manual. In: US Department of Commerce, ed. Washington, DC, 1980. 15. US Department of Commerce. Standard Industrial Classification Manual. In: Office USGP, ed. Washington, DC, 1987. 16. Callahan CL , Locke SJ , Dopart PJ , Decision rule approach applied to estimate occupational lead exposure in a case-control study of kidney cancer. American Journal of Industrial Medicine 2018;In press 17. Friesen MC , Locke SJ , Tornow C , Systematically extracting metal- and solvent-related occupational information from free-text responses to lifetime occupational history questionnaires. The Annals of occupational hygiene 2014;58 (5 ):612–24. doi: 10.1093/annhyg/meu012 [published Online First: 2014/03/05] 24590110 18. Koh DH , Locke SJ , Chen YC , Lead exposure in US worksites: A literature review and development of an occupational lead exposure database from the published literature. American journal of industrial medicine 2015;58 (6 ):605–16. doi: 10.1002/ajim.22448 [published Online First: 2015/05/15] 25968240 19. Koh DH , Nam JM , Graubard BI , Evaluating temporal trends from occupational lead exposure data reported in the published literature using meta-regression. The Annals of occupational hygiene 2014;58 (9 ):1111–25. doi: 10.1093/annhyg/meu061 [published Online First: 2014/09/07] 25193938 20. Locke SJ , Deziel NC , Koh DH , Evaluating predictors of lead exposure for activities disturbing materials painted with or containing lead using historic published data from US workplaces. American journal of industrial medicine 2017;60 (2 ):189–97.28079279 21. Purdue MP , Johansson M , Zelenika D , Genome-wide association study of renal cell carcinoma identifies two susceptibility loci on 2p21 and 11q13.3. Nature genetics 2011;43 (1 ):60–65. doi: http://www.nature.com/ng/journal/v43/n1/abs/ng.723.html#supplementary-information 21131975 22. Howie BN , Donnelly P , Marchini J . A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS Genet 2009;5 (6 ):e1000529. doi: 10.1371/journal.pgen.1000529 [published Online First: 2009/06/23] 19543373 23. Li Y , Graubard BI , DiGaetano R . Weighting methods for population-based case–control studies with complex sampling. Journal of the Royal Statistical Society: Series C (Applied Statistics) 2011;60 (2 ):165–85. 24. Rust KF , Rao JN . Variance estimation for complex surveys using replication techniques. Statistical methods in medical research 1996;5 (3 ):283–310. [published Online First: 1996/09/01] 8931197 25. Jain RB . Trends and variability in blood lead concentrations among US adults aged 20–64 years and senior citizens aged ≥65 years. Environmental Science and Pollution Research 2016;23 (14 ):14056–67. doi: 10.1007/s11356-016-6583-7 27044289 26. Young-Soon W , Ji-Hyun K , Young-Soo K , Association of internal exposure of cadmium and lead with periodontal disease: a study of the Fourth Korean National Health and Nutrition Examination Survey. Journal of Clinical Periodontology 2013;40 (2 ):118–24. doi: doi:10.1111/jcpe.12033 23167252 27. Lanphear BP , Rauch S , Auinger P , Low-level lead exposure and mortality in US adults: a population-based cohort study. The Lancet Public health 2018 doi: 10.1016/s2468-2667(18)30025-2 [published Online First: 2018/03/17] 28. Muntner P , He J , Vupputuri S , Blood lead and chronic kidney disease in the general United States population: results from NHANES III. Kidney international 2003;63 (3 ):1044–50. doi: 10.1046/j.1523-1755.2003.00812.x [published Online First: 2003/03/13] 12631086 29. Goering PL , Fowler BA . Metal constitution of metallothionein influences inhibition of delta-aminolaevulinic acid dehydratase (porphobilinogen synthase) by lead. The Biochemical journal 1987;245 (2 ):339–45. [published Online First: 1987/07/15] 3663161 30. Wang G , Fowler BA . Roles of biomarkers in evaluating interactions among mixtures of lead, cadmium and arsenic. Toxicology and applied pharmacology 2008;233 (1 ):92–9. doi: 10.1016/j.taap.2008.01.017 [published Online First: 2008/03/08] 18325558 31. Weiss NS . Subgroup-Specific Associations in the Face of Overall Null Results: Should We Rush In or Fear to Tread? Cancer Epidemiology Biomarkers & Prevention 2008;17 (6 ):1297–99. doi: 10.1158/1055-9965.epi-08-0144 32. White BM , Bonilha HS , Ellis C Jr. Racial/Ethnic Differences in Childhood Blood Lead Levels Among Children <72 Months of Age in the United States: a Systematic Review of the Literature. Journal of racial and ethnic health disparities 2016;3 (1 ):145–53. doi: 10.1007/s40615-015-0124-9 [published Online First: 2016/02/21] 26896114

PMC010xxxxxx/PMC10364142.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9422759 2553 Occup Environ Med Occup Environ Med Occupational and environmental medicine 1351-0711 1470-7926 29588333 10364142 10.1136/oemed-2017-104890 NIHMS1914834 Article A case-control investigation of occupational exposure to chlorinated solvents and non-Hodgkin lymphoma Callahan Catherine L. 1 Stewart Patricia 2 Friesen Melissa C. 1 Locke Sarah 1 De Roos Anneclaire J. 3 Cerhan James R. 4 Severson Richard K. 5 Rothman Nathaniel 1 Purdue Mark P. 1 1 Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland, USA 2 Stewart Exposure Assessments, LLC, Arlington, Virginia, USA 3 Department of Environmental and Occupational Health, Drexel University School of Public Health, Philadelphia, Pennsylvania, USA 4 Department of Health Sciences Research, Mayo Clinic, Rochester, Minnesota, USA 5 Department of Family Medicine and Public Health Sciences, Wayne State University, Detroit, Michigan, USA CONTRIBUTORSHIP PS, AJD, JRC, RKS, and NR contributed to the design, enrollment and/or data collection of the case-control study. PS, MPP, MCF, SJL contributed to the exposure assessment. MPP and CLC conducted the statistical analysis for this project. MPP and CLC drafted the manuscript. All authors contributed to the writing of the manuscript. Address correspondence to C. Callahan, Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 9609 Medical Center Dr., Room 6E644, Rockville, MD 20850 USA. Telephone: (240) 276-5040. catherine.callahan@nih.gov 19 7 2023 6 2018 27 3 2018 24 7 2023 75 6 415420 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Objectives: Although many studies have investigated the association between trichloroethylene (TCE) exposure and non-Hodgkin lymphoma (NHL), less is known about other chlorinated solvents. We extended our previous analysis of occupational TCE exposure in a multi-center population-based case-control study of NHL to investigate associations with five additional chlorinated solvents: 1,1,1,-trichloroethane, carbon tetrachloride, chloroform, methylene chloride, and perchloroethylene. Methods: Cases (n= 1189) and controls (n = 982) provided detailed information on their occupational histories and workplace exposure to chlorinated solvents for selected occupations using job-specific interview modules. An industrial hygienist used this information and a review of the literature to assess occupational exposure to chlorinated solvents. We computed odds ratios (ORs) and 95% confidence intervals (CIs) for different exposure metrics, with the unexposed group as the referent. We also computed ORs by NHL subtype. Results: High cumulative hours exposed to carbon tetrachloride was associated with NHL (>520 hours: OR 1.9; 95%CI 1.0–3.6; Ptrend = 0.04). This association remained after restricting to jobs with high-intensity exposure (2.0; 1.1–3.8; 0.03) and ≥90% exposure probability (2.1;1.0–4.3; 0.03), adjusting for TCE (2.1; 1.0–4.1; 0.04) and incorporating a 15-year lag (1.9; 1.0–3.6; 0.06). The other evaluated chlorinated solvents were not associated with NHL. Conclusions: This is the first study using high-quality quantitative exposure assessment methods to identify a statistically significant elevated association between occupational exposure to carbon tetrachloride and NHL. Our findings, though limited by a small number of exposed cases, offer evidence that carbon tetrachloride may be a lymphomagen. pmcINTRODUCTION Chlorinated solvents are a family of chlorine-containing synthetic compounds that have been used in a wide variety of industrial applications, including vapor degreasing, dry cleaning, paint removers, glues, and other operations.1 2 The production and use of these solvents declined over the course of the 20th century due to concerns over their toxicity, including their carcinogenic potential.1 2 Trichlorethylene (TCE) is classified as a human carcinogen (Group 1) by the International Agency for Research on Cancer (IARC) based on sufficient epidemiologic evidence of an association with kidney cancer and suggestive evidence for non-Hodgkin lymphoma (NHL) and liver cancer,3 while other chlorinated solvents have been classified by IARC as probably carcinogenic to humans (Group 2A; perchloroethylene), possibly carcinogenic (Group 2B; carbon tetrachloride, chloroform, methylene chloride) or unclassifiable with regards to carcinogenicity (Group 3; 1,1,1-trichloroethane).3–5 Evidence that several chlorinated solvents can induce immunotoxic effects has motivated epidemiologic research to investigate associations with NHL, a malignancy linked to immune dysregulation.6 While the relationship between TCE exposure and NHL has frequently been the focus of investigation (reviewed by:7), fewer studies have evaluated NHL associations for other chlorinated solvents.8–15 Moreover, findings from several studies have been limited by use of relatively non-specific exposure assessment methods, such as job-exposure matrices or self-reported use of chlorinated solvents. In a previous analysis of occupational TCE exposure within a population-based case-control study of NHL using job-by-job expert review of occupational exposure, we found high TCE exposure to be associated with increased risk of NHL.16 To clarify the association between occupational exposure to other chlorinated solvents and risk of NHL, we have expanded our investigation within this study to assess exposure to 1,1,1,-trichloroethane, carbon tetrachloride, chloroform, methylene chloride, and perchloroethylene. METHODS Details of the study have been described.17 18 Briefly, the National Cancer Institute – Surveillance, Epidemiology, and End Results (NCI-SEER) Study is a large population-based multi-center case-control study of NHL designed to obtain detailed information regarding workplace exposure to solvents. Participants were enrolled from four US SEER registry areas: the State of Iowa; Los Angeles County, California; and the metropolitan areas of Seattle, Washington; and Detroit, Michigan. Eligible cases were individuals between ages 20–74, diagnosed with incident, histologically confirmed NHL (ICD-O-3 codes 967–972) between July 1998 and June 2000, without known HIV infection. Cases of multiple myeloma were not enrolled in this study, which predates the 2001 World Health Organization (WHO) classification that expanded the definition of NHL to include this malignancy.19 Eligible controls were identified from the general population in the four registry areas via random digit dialing (RDD; < 65 years of age) or from Medicare files (65–74 years of age), with stratification on age (5-year intervals), sex, race, and SEER area to match the distribution of the cases. The study was approved by the institutional review boards at the NCI and the participating institutions, and study participants provided informed consent. We identified 2,248 potentially eligible cases. Of these 320 (14%) died before they could be interviewed, 127 (6%) could not be found, 16 (1%) had moved away, and 57 (3%) had physician refusals. Of the remaining 1,728 cases; 1,321 agreed to participate (participation rate of 76% and an overall response rate of 59%). Sixty-one percent of the cases were interviewed within 6 months after the diagnosis date, and 84% within 12 months after diagnosis. Of the 2,409 potentially eligible controls identified from RDD and Medicare files, 28 (1%) died before they could be interviewed, 311 (13%) could not be located, and 24 (1%) had moved away. Of the remaining 2,046 potential controls, 1,057 participated, which yielded a participation rate of 52% and an overall response rate of 44%. We excluded 132 cases and 75 controls who were never employed or had unknown occupations, thus 1,189 cases and 982 controls were included in our analyses. Exposure assessment Participants were mailed an occupational history calendar prior to the home visit. During the home visit, a trained interviewer administered a computer-assisted personal interview that gathered information on a variety of topics including occupational history. The occupational history queried the name of the employer, dates of employment, job title, number of hours worked (full or part time), type of business or service, tasks performed, chemicals or materials handled, and tools and equipment used for each job a participant held for at least 12 months since the age of 16. All jobs were coded using the US Standard Industrial Classification (SIC) and US Standard Occupational Classification (SOC) systems.20 21 Additionally, for selected occupations, one of 32 job- or industry-specific interview modules was administered. Twenty-eight of the modules included questions related to exposure to solvents, collecting information on the solvent(s) used, average frequency of and length of time spent on a variety of solvent-related tasks, work practices, potential for dermal exposure, and use of personal protective equipment.22–24 To reduce participant burden, a maximum of five modules was administered in an interview; six cases and four controls who reached the five-module limit had at least one additional job that would have prompted a module. Exposure metrics for the five newly assessed chlorinated solvents (chloroform, methylene chloride, perchloroethylene, carbon tetrachloride, and 1,1,1-trichloroethane) were developed using previously described methods.25 An expert industrial hygienist used information from systematic reviews of the industrial hygiene literature on the uses of perchloroethylene2 and similar unpublished reviews for the other substances to develop 38 job and task exposure matrices for each of the five newly assessed solvents. These matrices provided an initial estimate of probability and frequency of exposure to each of the five solvents for different combinations of occupation, industry, and decade of employment. The industrial hygienist used the information gathered from the occupational histories, and the job modules, the literature review, and the exposure matrices, to assess the probability, frequency, and determinants of exposure intensity for each chlorinated solvent for each job. Probability, defined as the theoretical probability of exposure to the solvent, was assigned to one of five categories: 0%, <10%, 10–49%, 50–89%, or ≥90%. A probability of ≥90% was assigned when the participant specifically reported using a given solvent. Otherwise, the industrial hygienist assigned a probability that the specific solvent was used when the job was held within the context of the other available information. Jobs with a probability ≥50% were assigned an exposure frequency and intensity. Exposure frequency was assigned to one of four categories: <2, 2–9, 10–19, or ≥20 hours per week. The assigned frequency was either the reported frequency of performing a solvent-related task or, if missing, the average frequency of all reports for that task. If frequency was not asked about in the occupational questionnaire, it was assigned based on the industrial hygienist’s knowledge of the job and workplace. Determinants of chlorinated solvent exposure intensity, defined as the solvent concentration in a participant’s breathing zone while exposed, were identified from previously developed predictive intensity models for three chlorinated solvents26 and estimated from the literature review and the industrial hygienist’s knowledge. These determinants of exposure intensity included job location (indoor, outdoors, both), local exhaust ventilation (effective, ineffective, absent), mechanism of solvent release (evaporation, aerosolized, other active), proximity (near, far, both), and process temperature (room temperature, elevated, both). An algorithm using these parameters was developed to assign a qualitative job exposure intensity of “high” or “low” (Supplemental Table 1). Levels of confidence ranging from 1 (lowest) to 4 were assigned to the values for probability, frequency, and intensity parameters. The exposure assessment was performed without knowledge of case/control status. Job-specific estimates of probability, frequency, and intensity were combined to develop participant level metrics of exposure. The highest exposure probability across all jobs was assigned as a participant’s exposure probability. Participants with an exposure probability ≥50% for a given solvent were assigned the following exposure metrics: duration of exposure (years), the sum of the number of years worked in jobs with an exposure probability ≥50%; cumulative hours exposed (hours), the sum of the product of the job-specific frequency midpoint (1, 6, 15, or 30 hours/week) and the job duration in weeks across all jobs with an exposure probability ≥50%; and average weekly exposure (hours per week), the cumulative hours exposed divided by the duration of exposure in weeks. These exposure metrics were defined as zero for participants with an exposure probability of 0% (unexposed). Participants with an exposure probability between >0% and <50% for a given solvent were excluded from analyses of that solvent. We also calculated these metrics restricted to jobs with ≥90% exposure probability and with a high exposure intensity and incorporating a five or 15-year exposure lag. Statistical analysis Differences between cases and controls across exposure metrics were assessed using odds ratios (OR) and 95% confidence intervals (CI) computed from unconditional logistic regression models. These models were adjusted for age (<35, 35–44, 45–54, 55–64, ≥65), sex, study center, race (black, white, or other), and education (<12 years, 12–15 years, >16 years, or missing). Estimates of exposure duration for each solvent were dichotomized at the median value of exposed controls with unexposed participants as the referent category. Tests for trend were performed by modelling the intra-category median among controls as a continuous variable, with values for unexposed participants set to zero. We note that our results for tests of trend should not be interpreted as independent of the results of exposure categories. We stratified analyses by sex and tested for interaction by including multiplicative terms for the sex and the solvent variables. We conducted analyses of specific histologic NHL subtypes, as defined by the WHO27 using polytomous logistic regression to compute ORs for the four most common NHL subtypes in our study [diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL), chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL), mantle cell lymphoma (MCL)]. Tests for heterogeneity across these subtypes were conducted using the Wald test. To assess the robustness of our findings we conducted certain analyses stratified by exposure probability, restricted to jobs with high exposure intensity, ≥90% exposure probability, restricting to exposed participants for whom the industrial hygienist assessed exposure with high confidence (≥3), incorporating a five- and 15-year lag, restricting to participants less than 65 years of age, and adjusting for cumulative hours exposed to TCE. RESULTS Characteristics of cases and controls in the NCI-SEER study are described in Table 1. Compared to controls, cases tended to be slightly younger and were less likely to be African American but otherwise were comparable with regards to sex, SEER site, and education level. In analyses among controls, after excluding participants unexposed to any chlorinated solvent, solvent exposure probabilities were moderately correlated with one another (Supplemental Table 2), with Spearman coefficients ranging from 0.05 (TCE and 1,1,1-trichloroethane) to 0.67 (1,1,1-trichloroethane and methylene chloride). Table 2 summarizes the association between exposure to the five newly assessed chlorinated solvents and risk of NHL. The prevalence of exposure to these solvents was low in this study population, with the proportion of controls with an exposure probability ≥50% ranging from 1.2% (1,1,1-trichloroethane) to 9.1% (methylene chloride). Degreasing was the most common exposure-related task among participants with exposure probability ≥50% to methylene chloride (45%), carbon tetrachloride (65%), and 1,1,1-trichloroethane (74%). The most common exposure-related tasks assessed for the remaining solvents were: for chloroform, surgery (28%), degreasing (15%), and dry cleaning (10%); for perchloroethylene, dry cleaning (42%) and degreasing (29%). While we did not observe evidence of an association with NHL for most of the solvents, high cumulative hours exposed to carbon tetrachloride was associated with increased risk (Table 2; >520 exposure hours vs. unexposed: OR 1.9, 95% CI 1.0–3.6; Ptrend = 0.04). In analyses of other carbon tetrachloride exposure metrics, weak, non-significant associations between duration (>9 years vs. unexposed: OR 1.3, 95% CI 0.7–2.4; Ptrend= 0.36) and average weekly exposure (>1hour/week vs. unexposed: OR 1.4, 95% CI 0.7–3.0; Ptrend=0.28) and increased risk were observed. When we further stratified by exposure probability (Table 3), the association with >520 cumulative hours exposed to carbon tetrachloride was confined to subjects with ≥90% exposure probability (OR 2.1, 95% CI 1.0–4.3). Additional analyses of carbon tetrachloride by NHL subtype are presented in Table 4. The association with high cumulative hours exposed was observed for FL (>520 exposure hours vs. unexposed: OR 2.7, 95% CI 1.2–6.3; Ptrend = 0.02) and MCL (3.5, 1.0–11.5; Ptrend = 0.04), but not DLBCL (1.1, 0.4–2.9; Ptrend = 0.87) or CLL/SLL (0.9, 0.2–3.9; Ptrend = 0.91). However, tests for heterogeneity were not statistically significant. Subtype specific analyses for exposure to other solvents were null (results not shown). We conducted several sensitivity analyses to assess the robustness of the association between >520 cumulative hours exposed to carbon tetrachloride and NHL. The association persisted in analyses restricted to high intensity exposure (OR 2.0, 95% CI 1.1–3.8; Ptrend = 0.03), restricted to participants the hygienist assessed with high confidence (2.1, 1.1–4.3; Ptrend = 0.04), incorporating a five- and fifteen-year exposure lags (2.0, 1.1–3.8; Ptrend = 0.03 and 1.9, 1.0–3.6; Ptrend = 0.06, respectively), restricting to participants less than 65 years of age (3.9, 1.1–13.9; Ptrend = 0.03), and adjusting for cumulative hours exposed to TCE (2.1, 1.0–4.1; Ptrend = 0.04). DISCUSSION In this expanded investigation of chlorinated solvents within the NCI-SEER case-control study of NHL, cumulative hours exposed to carbon tetrachloride was associated with NHL. We observed null findings for chloroform, methylene chloride, perchloroethylene, and 1,1,1-trichloroethane and other metrics of carbon tetrachloride exposure. Carbon tetrachloride was widely used between the 1930’s and 1980’s in the US as a degreasing and cleaning agent, in the production of chlorinated refrigerants and propellants, as a fire extinguisher, and as a grain fumigant.28 Since it is an ozone-depleting chemical, dispersive use of carbon tetrachloride was eliminated in the US in 1996 and non-dispersive uses have been strictly regulated.29 The findings from previous epidemiologic studies of carbon tetrachloride (one cohort, three case-control) support an association with NHL. NHL mortality was non-significantly elevated in a study of aircraft maintenance workers exposed to carbon tetrachloride.30 Self-reported use of carbon tetrachloride as a fumigant was associated with NHL in a study of 517 male NHL cases and 1,506 controls (OR 2.4, 95% CI 1.2–5.1).11 Occupational exposure to carbon tetrachloride, estimated from a job-exposure matrix, was associated with NHL (OR 2.3, 95% CI 1.3– 4.0) in a study of only women.9 Another study that used an expert assessment of work histories reported a non-significant association with this chemical (OR 2.0, 95% CI 0.6–6.9).8 Although our study and the two previous studies that considered subtype of NHL were underpowered to detect associations with individual subtypes, our observed carbon tetrachloride association with FL and MCL, and absence of an association with DLBCL and CLL/SLL, is not consistent with the findings from the previous studies.8 31 However, TCE exposure has been more strongly associated with FL in our study16 and in a pooled analysis of 3788 NHL cases and 4279 controls that included participants from our study.32 The biologic basis for an association between carbon tetrachloride and NHL is unclear, as there have been no human studies investigating immunologic and other biologic effects among exposed workers. However, rodent studies have indicated that hepatotoxic doses of carbon tetrachloride (≥500 mg/kg) suppress T-cell activity, which could be due in part to the transforming growth factor beta (TGF-β) secreted by the liver during repair.33 34 To our knowledge, no prior studies of carbon tetrachloride and NHL attempted to control for possible confounding by exposure to TCE or other solvents. Use of specific chlorinated solvents for a given task has overlapped in various industries over time, and has complicated the interpretation of studies regarding the carcinogenicity of specific solvents.32 Our observed association between carbon tetrachloride and NHL remained upon adjustment for exposure to TCE, thus arguing against confounding from this solvent as an explanation for our findings. We did not observe evidence of NHL associations for the other evaluated chlorinated solvents. We acknowledge, however, that we had limited power to detect modest associations with these chemicals, particularly for 1,1,1-trichloroethane, perchloroethylene, and chloroform, for which less than three percent of cases or controls were exposed. We thus cannot rule out the existence of subtle effects on NHL overall or on a certain subtype. Inferences from our study should be made in the context of several limitations. Despite the relatively large sample size of this case-control study, we identified only small numbers of subjects with high exposure probabilities, which is not unexpected given the rarity of these occupational exposures in the general population. Sample size limitations were particularly severe in our analyses of NHL subtypes. We cannot rule out selection bias as an explanation for our finding, because the participation rate among controls was comparatively low, although we previously estimated demographic and socioeconomic differences between control participants and nonparticipants to be generally minor.35 As with all retrospective case-control studies, it is possible that cases may have ruminated more deeply on past exposures leading to recall bias. However, we did not observe an association between NHL and four out of the six solvents we assessed, which would be expected if cases were systematically overreporting occupational exposures. We made many comparisons in our analyses, thus the association between carbon tetrachloride could be due to chance, however prior epidemiologic studies have also observed an association with carbon tetrachloride and NHL risk, which provides some reassurance that chance alone is not an alternate explanation. Lastly, our occupational history only collected information on jobs held by participants for six months or longer, thus exposures from shorter-term jobs may not have been captured. Our study has several important strengths. Importantly, we had detailed information on workplace tasks gathered from a general work history and job- and industry-specific modules designed to elicit information on solvent exposure. An extensive literature review informed the development of task-, job-, industry-, and decade-specific exposure matrices and assessment rules a priori. This rich dataset was used by an industrial hygienist to assign several parameters of exposure. These expected improvements in exposure assessment facilitated several sensitivity analyses, including the calculation of exposure metrics among participants who were rated as having a high probability of exposure to carbon tetrachloride. In addition to increasing the sensitivity of exposure estimates, our method likely increased the specificity of exposure assessment, which is essential to reduce the potential for bias from exposure misclassification in studies, such as ours, with a low exposure prevalence.36 Furthermore, we were able to adjust for a broad range of chlorinated solvents. In conclusion, we observed a significantly elevated association between cumulative hours exposed to carbon tetrachloride exposure and NHL, which was not attenuated after adjustment for TCE. These findings offer further support that carbon tetrachloride may be a lymphomagen. While additional investigations of this relationship may not be feasible given the reduction of use of carbon tetrachloride, a meta-analysis or other quantitative summary of the epidemiologic literature may be informative. Supplementary Material Supplementary Material FUNDING This research was supported by the Intramural Research Program of the NIH, National Cancer Institute, with Public Health Service contracts N01-PC-65064, N01-PC-67008, N01-PC-67009, N01-PC-67010, and N02-PC-71105. Table 1. Selected characteristics of participants in the NCI-SEER study, 1998–2001. Characteristic Cases (n = 1189) Controls (n = 982) n (%) Pa Age (years)  <35 68 (5.7) 53 (5.4)  35–44 153 (12.9) 98 (10.8)  45–54 261 (22.0) 185 (18.8)  55–64 316 (26.6) 230 (23.4)  ≥65 391 (32.9) 416 (42.4) 0.0002 Sex  Female 523 (44.0) 458 (46.6)  Male 666 (56.0) 524 (53.4) 0.22 Study center  Detroit 209 (17.6) 144 (14.7)  Iowa 352 (29.6) 273 (27.8)  Los Angeles 310 (26.1) 273 (27.8)  Seattle 318 (26.8) 292 (29.7) 0.13 Race  White 1014 (85.3) 787 (80.1)  African American 91 (7.7) 132 (13.4)  Other 84 (7.1) 63 (6.4) <0.0001 Years of education  <12 118 (9.9) 97 (9.9)  12–15 734 (61.7) 584 (59.5)  ≥16 336 (28.3) 301 (30.7) Missing 1 0.47 Category percentages may not sum to 100% because of rounding. a P-value from chi-square test of independence between cases and controls. Table 2. Associations between measures of exposure to individual chlorinated solvents (exposure probability, cumulative hours exposed) and non-Hodgkin’s lymphoma risk, the NCI-SEER study, 1998–2001. Solvent Exposure probability Cumulative hours Cases Controls ORa (95% CI) N (%) N (%) Chloroform Unexposed 836 (70.6) 711 (72.5) 1.0 <50% 329 (27.8) 255 (26.0) 1.0 (0.8, 1.3) ≥50% 19 (1.6) 15 (1.5) 1.2 (0.6, 2.3) ≤1560 12 (1.4) 8 (1.1) 1.3 (0.5, 3.2) >1560 7 (0.8) 7 (1.0) 1.0 (0.3, 2.8) Ptrend 0.98 Methylene chloride Unexposed 602 (50.8) 536 (54.6) 1.0 <50% 468 (39.5) 356 (36.3) 1.1 (0.9, 1.3) ≥50% 115 (9.7) 89 (9.1) 1.0 (0.7, 1.3) ≤1560 68 (9.5) 44 (7.1) 1.2 (0.8, 1.9) >1560 44 (6.2) 42 (6.8) 0.8 (0.5, 1.3) Ptrend 0.47 Perchloroethylene Unexposed 693 (58.5) 608 (62.0) 1.0 <50% 468 (39.5) 353 (36.0) 1.1 (0.9, 1.3) ≥50% 24 (2.0) 20 (2.0) 1.0 (0.6, 1.9) ≤2652 11 (1.5) 10 (1.6) 1.0 (0.4, 2.3) >2652 11 (1.5) 9 (1.4) 1.1 (0.4, 2.7) Ptrend 0.87 Carbon tetrachloride Unexposed 829 (70.0) 712 (72.6) 1.0 <50% 301 (25.4) 234 (23.9) 1.0 (0.8, 1.3) ≥50% 54 (4.6) 35 (3.6) 1.3 (0.8, 2.0) ≤520 17 (1.9) 16 (2.2) 0.9 (0.4, 1.8) >520 34 (3.9) 16 (2.2) 1.9 (1.0, 3.6) Ptrend 0.04 1,1,1-trichloroethane Unexposed 619 (52.3) 555 (56.6) 1.0 <50% 551 (46.5) 414 (42.2) 1.1 (0.9, 1.3) ≥50% 14 (1.2) 12 (1.2) 1.0 (0.4, 2.1) ≤312 2 (0.3) 6 (1.1) 0.3 (0.1, 1.6) >312 11 (1.7) 6 (1.1) 1.5 (0.6, 4.3) Ptrend 0.47 a adjusted for age, study center, sex, race, and education. “Exposure probability” percentages may not sum to 100% due to rounding error. “Cumulative hours exposed” percentages may not sum to “≥50% probability” percentage due to rounding and subjects with missing data for this metric. Tests for trend were performed by modelling the intra-category median among controls as a continuous variable, with values for unexposed participants set to zero. Table 3. Associations between exposure probability and cumulative hours exposed to carbon tetrachloride and non-Hodgkin’s lymphoma risk, the NCI-SEER study, 1998–2001. Exposure probability Cumulative hours Cases N Controls N ORa (95% CI) Unexposed 829 712 1.0 <50% 300 234 1.0 (0.8, 1.3) 50–89% ≤520 6 4 1.2 (0.3, 4.4) >520 7 5 1.2 (0.4, 3.9) ≥90% ≤520 11 12 0.7 (0.3, 1.7) >520 27 11 2.1 (1.0, 4.3) a adjusted for age, study center, sex, race, and education. Table 4. Analysis of estimated occupational exposure to carbon tetrachloride and selected NHL histologic subtypes within the NCI-SEER study, 1998–2001. Exposure probability Cumulative hours Controls DLBCL FL CLL/SLL MCL P heterogeneity n n ORa (95% CI) n ORa (95% CI) n ORa (95% CI) n ORa (95% CI) Unexposed 712 260 1.0 210 1.0 91 1.0 33 1.0 <50% 234 95 1.0 (0.7, 1.3) 70 0.9 (0.7, 1.3) 42 1.2 (0.8, 1.8) 12 0.8 (0.4, 1.6) 0.71 ≥50% 35 10 0.7 (0.3, 1.5) 12 1.2 (0.6, 2.4) 8 1.4 (0.6, 3.3) 4 1.9 (0.7, 5.4) 0.31 ≤520 16 3 0.5 (0.1, 1.6) 2 0.4 (0.1, 1.8) 6 2.5 (0.9, 6.7) 1 0.8 (0.1, 6.5) 0.06 >520 16 6 1.1 (0.4, 2.9) 10 2.7 (1.1, 6.0) 2 0.9 (0.2, 3.9) 4 3.5 (1.0, 11.5) 0.17 Ptrend 0.87 Ptrend 0.02 Ptrend 0.91 Ptrend 0.04 0.17 Abbreviations: NHL, non-Hodgkin’s lymphoma; DLBCL, diffuse large B-cell lymphoma; CLL/SLL, chronic lymphocytic leukemia/small lymphocytic lymphoma; MCL, mantle cell lymphoma; OR, odds ratio; CI, confidence interval. a adjusted for age, study center, sex, race, and education. Tests for trend were conducted by modelling the intra-category median among controls as a continuous variable, with values for unexposed participants set to zero. Tests for heterogeneity were based on Wald test of beta coefficients across subtypes from polytomous logistic regression models. KEY FINDINGS What is already known about this subject? Trichloroethylene, a chlorinated solvent used to degrease metal parts, has been associated with non-Hodgkin lymphoma (NHL) in several studies. We extended our previous analyses of trichloroethylene within a case-control study to include expert assessment of occupational exposure to five additional solvents. What are the new findings? NHL cases were more likely than controls to have high cumulative hours exposed to carbon tetrachloride, a solvent previously used in several industrial applications. The association between carbon tetrachloride and NHL persisted after adjustment for co-exposure to trichloroethylene. How might this impact on policy or clinical practice in the foreseeable future? Carbon tetrachloride was classified in 1999 as a possible human carcinogen by the International Agency for Research on Cancer (IARC), our findings will inform any future evaluation of carcinogenicity for this chemical. COMPETING INTERESTS P.S. is employed by Stewart Exposure Assessments, LLC (Arlington, VA, USA). The remaining authors declare they have no actual or potential competing financial interests. REFERENCES 1. Bakke B , Stewart PA , Waters MA . Uses of and exposure to trichloroethylene in U.S. industry: a systematic literature review. Journal of occupational and environmental hygiene 2007;4 (5 ):375–90.17454505 2. Gold LS , De Roos AJ , Waters M , Systematic literature review of uses and levels of occupational exposure to tetrachloroethylene. Journal of occupational and environmental hygiene 2008;5 (12 ):807–39.18949603 3. IARC. Trichloroethylene, Tetrachloroethylene, and Some Other Chlorinated Agents. IARC monographs on the evaluation of carcinogenic risks to humans / World Health Organization, International Agency for Research on Cancer 2014;106 . 4. IARC. Some Chemicals Used as Solvents and in Polymer Manufacture. IARC monographs on the evaluation of carcinogenic risks to humans / World Health Organization, International Agency for Research on Cancer 2016;110 :189–243. 5. IARC. Re-evaluation of some organic chemicals, hydrazine and hydrogen peroxide. IARC monographs on the evaluation of carcinogenic risks to humans / World Health Organization, International Agency for Research on Cancer 1999;71 . 6. Hartge P , Wang S , Bracci P , In: Schottenfeld DFJJ , ed. Cancer Epidemiology and Prevention. Third ed. New York: Oxford University Press, 2006:898–918. 7. Karami S , Bassig B , Stewart PA , Occupational trichloroethylene exposure and risk of lymphatic and haematopoietic cancers: a meta-analysis. Occupational and environmental medicine 2013;70 (8 ):591–9.23723297 8. Seidler A , Möhner M , Berger J , Solvent exposure and malignant lymphoma: a population-based case-control study in Germany. Journal of Occupational Medicine and Toxicology (London, England) 2007;2 :2–2.17407545 9. Wang R , Zhang Y , Lan Q , Occupational exposure to solvents and risk of non-Hodgkin lymphoma in Connecticut women. Am J Epidemiol 2009;169 (2 ):176–85.19056833 10. Fritschi L , Benke G , Hughes AM , Risk of non-Hodgkin lymphoma associated with occupational exposure to solvents, metals, organic dusts and PCBs (Australia). Cancer causes & control : CCC 2005;16 (5 ):599–607.15986116 11. McDuffie HH , Pahwa P , McLaughlin JR , Non-Hodgkin’s lymphoma and specific pesticide exposures in men: cross-Canada study of pesticides and health. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology 2001;10 (11 ):1155–63. 12. Miligi L , Costantini AS , Benvenuti A , Occupational exposure to solvents and the risk of lymphomas. Epidemiology (Cambridge, Mass) 2006;17 (5 ):552–61.16878041 13. Tranah GJ , Holly EA , Bracci PM . Solvent exposure and non-Hodgkin lymphoma: no risk in a population-based study in the San Francisco Bay Area. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology 2009;18 (11 ):3130–2. 14. Kato I , Koenig KL , Watanabe-Meserve H , Personal and occupational exposure to organic solvents and risk of non-Hodgkin’s lymphoma (NHL) in women (United States). Cancer causes & control : CCC 2005;16 (10 ):1215–24.16215872 15. Vlaanderen J , Straif K , Pukkala E , Occupational exposure to trichloroethylene and perchloroethylene and the risk of lymphoma, liver, and kidney cancer in four Nordic countries. Occupational and environmental medicine 2013;70 (6 ):393–401.23447073 16. Purdue MP , Bakke B , Stewart P , A case-control study of occupational exposure to trichloroethylene and non-Hodgkin lymphoma. Environmental health perspectives 2011;119 (2 ):232–8.21370516 17. Chatterjee N , Hartge P , Cerhan JR , Risk of non-Hodgkin’s lymphoma and family history of lymphatic, hematologic, and other cancers. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology 2004;13 (9 ):1415–21. 18. Schenk M , Purdue MP , Colt JS , Occupation/industry and risk of non-Hodgkin’s lymphoma in the United States. Occupational and environmental medicine 2009;66 (1 ):23–31.18805886 19. Jaffe ES . Pathology and genetics of tumours of haematopoietic and lymphoid tissues: Iarc, 2001. 20. US Department of Commerce. Standard Occupational Classification Manual. In: US Department of Commerce, ed. Washington, DC, 1980. 21. US Department of Commerce. Standard Industrial Classification Manual. In: Office USGP, ed. Washington, DC, 1987. 22. Gerin M , Siemiatycki J , Kemper H , Obtaining occupational exposure histories in epidemiologic case-control studies. Journal of occupational medicine : official publication of the Industrial Medical Association 1985;27 (6 ):420–6.4020500 23. Siemiatycki J , Day NE , Fabry J , Discovering carcinogens in the occupational environment: a novel epidemiologic approach. 1981. 24. Stewart PA , Stewart WF , Heineman EF , A novel approach to data collection in a case-control study of cancer and occupational exposures. Int J Epidemiol 1996;25 (4 ):744–52.8921451 25. Purdue MP , Stewart PA , Friesen MC , Occupational exposure to chlorinated solvents and kidney cancer: a case-control study. Occupational and environmental medicine 2017;74 (4 ):268–74.27803178 26. Hein MJ , Waters MA , Ruder AM , Statistical modeling of occupational chlorinated solvent exposures for case-control studies using a literature-based database. The Annals of occupational hygiene 2010;54 (4 ):459–72.20418277 27. Morton LM , Turner JJ , Cerhan JR , Proposed classification of lymphoid neoplasms for epidemiologic research from the Pathology Working Group of the International Lymphoma Epidemiology Consortium (InterLymph). Blood 2007;110 (2 ):695–708.17389762 28. Doherty RE . A History of the Production and Use of Carbon Tetrachloride, Tetrachloroethylene, Trichloroethylene and 1,1,1-Trichloroethane in the United States: Part 1--Historical Background; Carbon Tetrachloride and Tetrachloroethylene. Environmental Forensics 2000;1 (2 ):69–81. 29. Hu L , Montzka SA , Miller BR , Continued emissions of carbon tetrachloride from the United States nearly two decades after its phaseout for dispersive uses. Proceedings of the National Academy of Sciences of the United States of America 2016;113 (11 ):2880–5.26929368 30. Blair A , Hartge P , Stewart PA , Mortality and cancer incidence of aircraft maintenance workers exposed to trichloroethylene and other organic solvents and chemicals: extended follow up. Occupational and environmental medicine 1998;55 (3 ):161–71.9624267 31. Barry KH , Zhang Y , Lan Q , Genetic variation in metabolic genes, occupational solvent exposure, and risk of non-hodgkin lymphoma. Am J Epidemiol 2011;173 (4 ):404–13.21228414 32. Cocco P , Vermeulen R , Flore V , Occupational exposure to trichloroethylene and risk of non-Hodgkin lymphoma and its major subtypes: a pooled InterLymph analysis. Occupational and environmental medicine 2013;70 (11 ):795–802.23881218 33. Delaney B , Strom SC , Collins S , Carbon Tetrachloride Suppresses T-Cell-Dependent Immune Responses by Induction of Transforming Growth Factor-1. Toxicology and Applied Pharmacology 1994;126 (1 ):98–107.8184439 34. Guo TL , McCay JA , Brown RD , Carbon tetrachloride is immunosuppressive and decreases host resistance to Listeria monocytogenes and Streptococcus pneumoniae in female B6C3F1 mice. Toxicology 2000;154 (1–3 ):85–101.11118673 35. Shen M , Cozen W , Huang L , Census and geographic differences between respondents and nonrespondents in a case-control study of non-Hodgkin lymphoma. Am J Epidemiol 2008;167 (3 ):350–61.17989060 36. Teschke K , Olshan AF , Daniels JL , Occupational exposure assessment in case-control studies: opportunities for improvement. Occupational and environmental medicine 2002;59 (9 ):575–93; discussion 94. 12205230

PMC010xxxxxx/PMC10364143.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0402313 3651 Exp Hematol Exp Hematol Experimental hematology 0301-472X 1873-2399 36183966 10364143 10.1016/j.exphem.2022.09.003 NIHMS1917473 Article Role of ASXL1 in Hematopoiesis and Myeloid Diseases Gao Xin 1 You Xiaona 2 Droin Nathalie 3 Banaszak Lauren G 4 Churpek Jane 4 Padron Eric 5 Geissler Klaus 6 Solary Eric 378 Patnaik Mrinal M. 9 Zhang Jing 1* 1 McArdle Laboratory for Cancer Research, University of Wisconsin-Madison, Madison, WI 2 Institute of Immunopharmaceutical Sciences, School of Pharmaceutical Sciences, Shandong University, Jinan, China 3 INSERM U1287, Gustave Roussy Cancer Center, Villejuif, France 4 Department of Medicine, University of Wisconsin School of Medicine and Public Health, University of Wisconsin Carbone Cancer Center, Madison, WI, USA 5 Chemical Biology and Molecular Medicine Program, Moffitt Cancer Center, FL, USA 6 Medical School, Sigmund Freud University; Vienna; Austria 7 Department of Hematology, Gustave Roussy Cancer Center, Villejuif, France 8 Université Paris-Saclay, Faculté de Médecine, Le Kremlin-Bicêtre, France 9 Division of Hematology, Department of Internal Medicine, Mayo Clinic, Rochester, MN * Correspondence: zhang@oncology.wisc.edu 17 7 2023 11 2022 30 9 2022 24 7 2023 115 1419 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Next-generation sequencing technology (NGS), including whole-exome or whole-genome sequencing and target gene sequencing, has allowed the molecular characterization of somatic mutation spectrums in hematologic diseases. Mutations in Additional sex combs-like 1 (ASXL1), a chromatin regulator, are identified in clonal hematopoiesis of indeterminate potential (CHIP), indicating ASXL1 mutations as early events in leukemogenesis. Not surprisingly, they occur at high frequency in myeloid malignancies and associated with poor prognosis. Therefore, understanding how mutant ASXL1 drives clonal expansion and leukemogenesis will serve as the basis for future development of preventative and/or therapeutic strategies for myeloid diseases with ASXL1 mutations. Here, we discuss the biology of ASXL1 and its role in controlling normal and malignant hematopoiesis. In addition, we review the clinical relevance of ASXL1 mutations in CHIP and myeloid diseases. pmcIntroduction of ASXL1 gene and protein Mammalian ASXL family genes (ASXL1, ASXL2 and ASXL3) are the mammalian homologs of Drosophila Additional sex combs Asx (1). Asx deletion leads to a homeotic phenotype characteristic of both Polycomb group (PcG, repressive complex associated with H3K27me3) and Trithorax group (TrxG, activating function associated with H3K4me3) gene deletions (1–3). Both Asxl1 and Asxl2 expression is virtually ubiquitous throughout embryogenesis and in adult tissues, whereas Asxl3 expression is more restricted and only detectable in lymph node, eye, lung, skin, brain, and pituitary gland (4). The human ASXL1 gene is located on the chromosome 20q11.21 and encodes a 1541 amino acid protein (Figure 1) (5). ASXL1 contains an ASXN domain in the N-terminal region, an ASX homology (ASXH) domain in the N-terminal adjoining region, and a plant homeodomain (PHD) in the C-terminal region. ASXL family proteins share highly conserved ASXN, ASXH and PHD domains. The ASXN and PHD domains are putative DNA and histone binding domains, respectively. The ASXH domain (also referred to as DEUBAD, deubiquitinase adaptor) interacts with a partner protein BAP1 to confer deubiquitinase activity, leading to gene repression (6). At the endogenous level, truncated ASXL1 proteins resulting from ASXL1 mutations are rapidly degraded, and the ASXL1-BAP1 complex is undetectable (7). By contrast, overexpression of truncated ASXL1 increases the stability of BAP1 and enhances the deubiquitination activity of ASXL1-BAP1 complex. It is unclear whether overexpression of mutant ASXL1 recapitulates its function at the physiological level (8, 9). In addition to BAP1, ASXL1 interacts with core polycomb repressive complex 2 (PRC2) components EZH2 and SUZ12, which are involved in the deposition of H3K27me3 histone repressive marks (7). The functions of the long stretch of amino acids between the ASXH and PHD domains have been poorly understood. A recent study revealed that the C-terminal intrinsic disordered region is important for the formation of nuclear paraspeckles. Deletion of this region disrupts these paraspeckles, leading to the attenuated repopulation capability of hematopoietic stem cells (HSCs) (10). Interestingly, ASXL1 mutations identified in myeloid diseases are predominantly located within this region, generating C-terminal truncated proteins (see below). The functions of the other ASXL proteins are less known. ASXL2 has been shown to be essential for cardiac function and bone development (11–13). Recent studies revealed a high-frequency of ASXL2 mutations in acute myeloid leukemia (AML) patients bearing the RUNX1::RUNX1T1 (AML/ETO) fusion. Loss of Asxl2 in mouse leads to development of myelodysplastic syndrome (MDS)-like disease and promotes leukemogenesis driven by RUNX1::RUNX1T1 (14, 15). Unlike ASXL1 and ASXL2 mutations, ASXL3 mutations have not been detected in AML patients (16). ASXL1 mutations in CHIP and myeloid diseases (clinal relevance) Mutations in ASXL1 are identified in clonal hematopoiesis of indeterminate potential (CHIP) and significantly associated with smoking (17, 18). CHIP initially referred to the expansion of peripheral blood cells derived from hematopoietic stem cells (HSCs) with at least one somatic driver mutation in healthy elderly individuals (19–21). CHIP is strongly linked to aging and confers to an increased risk for blood cancers, non-hematological diseases (e.g. cardiovascular disease), and all-cause mortality (19–23). Although CHIP confers an approximately 10-fold increased risk to develop hematologic malignancies, such risk remains low (0.5-1% per year) (19). Therefore, it cannot explain the increased overall mortality associated with CHIP. A cause-specific mortality analysis revealed that non-leukemic mortality (e.g. cardiovascular diseases) in CHIP patients is higher than that due to blood cancers (21). DNMT3A, TET2, and ASXL1 are among the most frequently mutated genes in CHIP. They are associated with initiation of acute myeloid leukemia (AML) and other myeloid diseases. Corroborating the human data, HSCs with Tet2 or Dnmt3a mutations robustly undergo expansion in transplant recipient mice (reviewed in (24)). Subsequently, CHIP was identified in patients previously treated for solid tumors and myeloid malignancy-associated CHIP mutations were also present in patients with lymphoid malignancies (24). The CHIP mutation spectrum in these patients is distinct from that in healthy individuals. The selection and expansion of preleukemic-HSC clones precede the development of myeloid leukemia. Not surprisingly, ASXL1 mutations (and 20q deletion) are frequently identified in myeloid malignancies, in particular ~20% in MDS, ~45% in chronic myelomonocytic leukemia (CMML), ~10% in myeloproliferative neoplasms (MPNs), and ~20% in AML (25–28). Interestingly, ASXL1 mutations identified in CHIP are enriched around codons R404 (nonsense), Y591 (nonsense/frameshift), H630 (frameshift), and R693 (nonsense/frameshift). By contrast, ASXL1 mutations identified in myeloid diseases (including MDS and CMML) are predominantly frameshift mutations around codon G646 (G646: 18%; codon 630-660: 42%) (Figure. 2). Controversy has surrounded molecular testing of c.1934dupG p.Gly646fs ASXL1 variant. Its location within an 8 base-pair guanine mononucleotide repeat sequence made it suspicious for an artifact of PCR and/or sequencing rather than a true somatic mutation (29). However, the variant allele frequency of this mutation is >5% in many cases, arguing against PCR artefacts. Moreover, subsequent reports using NGS sequencing confirmed that the ASXL1 c.1934dupG is only detected in leukemia cells, but not in matched germline samples or healthy controls (30–33). Clearly, the ASXL1 mutations around codon G646 are prevalent in myeloid diseases but much less common in CHIP. Similarly, the hotspot DNMT3A R882H mutation in AML is rarely seen in CHIP (34). We and others hypothesize that unlike majority of CHIP mutations that are fairly stable and less pathogenic in elderly patients, the hotspot ASXL1 and DNMT3A mutations represent pathological CHIP mutations with high risk for accumulating additional driver mutations and developing myeloid diseases. In support of this idea, CMML cells with hotspot ASXL1 mutations around G646 display distinct transcriptomic changes from normal BM cells and these changes are absent in CMML cells with non-hotspot ASXL1 mutations (35). Asxl1−/− and Asxl1 G643Wfs (corresponding to human G646Wfs) knockin mice develop MDS and a fraction of them transform to myeloid leukemia (36, 37) (see below). ASXL1 germline mutations and Bohring-Opitz syndrome Bohring-Opitz syndrome (BOS) is a rare genetic disorder first reported by Bohring et al. in 1999, to describe four individuals with Opitz trigonocephaly (C)-like syndrome (38). BOS is a clinically recognizable syndrome characterized by facial dysmorphism, microcephaly, limb anomalies, postnatal failure to thrive, severe developmental delays and intellectual disability. To date ~100 cases have been described, almost half of which were molecularly confirmed to carry a heterozygous constitutive ASXL1 mutation, suggesting that constitutive mutations in ASXL1 are a major cause of BOS. Similar as CHIP and myeloid diseases-associated ASXL1 mutations, most of BOS-related ASXL1 mutations are de novo nonsense or frameshift mutation. Emma Bedoukiann and colleagues presented the first report of BOS caused by a pathogenic ASXL1 mutation inherited from a germline mosaic mother (39). Later, Karen Seiter and colleagues reported that a father and son were found to have the identical ASXL1 mutation (40), supporting the diagnosis of a germline mutation of ASXL1. Both of them developed AML without BOS symptoms. Therefore, how the same germline ASXL1 mutations cause different diseases remains unknown. Biological function of Asxl1 (mouse work) To evaluate the functions of Asxl1 in hematopoiesis and leukemogenesis, five different mouse models have been generated using different approaches (36, 37, 41–43). Conditional Asxl1 knockout mice were created to study loss of Asxl1 function in adult hematopoietic system (36). Asxl1−/− bone marrow (BM) cells display increased number of HSCs and decreased re-plating capability as compared to wildtype (WT) cells. Upon transplantation, Asxl1−/− BM cells show reduced reconstitution in young recipients. Deletion of Asxl1 leads to significant down-regulation of H3K27me3 due to loss of ASXL1-mediated recruitment of PRC2 key components, such as EZH2, to the chromatin (7, 36). In addition, a novel ASXL1-OGT(O-GlcNAc transferase) axis was identified to regulate H3K4 methylation in myeloid malignancies (44). ASXL1 interacts with HCFC1 and OGT and is stabilized via OGT-mediated O-GlcNAcylation. Disruption of this novel axis inhibits myeloid differentiation and H3K4 methylation(44). Consistent with the previous results, we reported that global H3K4me1, H3K4me3, and H3K27me3 levels were significantly decreased in Asxl1−/− BM cells (45). Although global H3K27Ac level in Asxl1−/− BM cells was comparable to that in control cells, H3K27Ac level was increased at specific gene loci. Two transgenic overexpression models use different exogenous promoters (Rosa26 vs Vav1) to drive the transcription of different Asxl1 mutants (E635RfsX15 vs Y558X) (42, 43). Therefore, it is difficult to compare and interpret their results. Nonetheless, these transgenic overexpression models and in vitro overexpression studies (46) suggest that ASXL1 mutations may be dominant negative or gain-of-function. However, it is questionable whether these overexpression studies truefully reflects the physiology function of truncated ASXL1 proteins. To overcome this problem, two groups independently generated Asxl1tm knock-in mouse models (37, 41). In both models, the same Asxl1 guanine duplication was introduced into the endogenous Asxl1 locus, closely resembling patient-derived ASXL1 G646WfsX12 mutation. This hotspot frameshift mutation creates a truncated protein of 655aa (658aa in human) in contrast to the full length ASXL1 protein of 1514aa (1541aa in human). Studies with these two knock-in mouse models yielded highly consistent results, some of which are distinct from Asxl1−/− data. For example, in comparison to WT cells, Asxl1tm/+ BM cells exhibit reduced number of HSCs, increased re-plating capability, and largely comparable reconstitution in young recipients, suggesting that in addition to losing part of WT ASXL1 functions, Asxl1tm instills some new functions. However, it remains unclear what epigenetic alterations this mutation causes and how this mutation could drive CH in humans. Genetic interaction of ASXL1 with NRAS ASXL1 mutations frequently coexist with other mutations, such as TET2 (47), RUNX1 (48), SETBP1 (49–51) and NRAS (25–28). Asxl1 loss in mice results in MDS that could transform to myeloid leukemia with age, suggesting that Asxl1 deficiency cooperates with additional mutations to induce myeloid leukemias. ASXL1 mutations predict inferior outcomes in all myeloid diseases (26, 52, 53). They significantly co-occur with NRAS mutations in CMML (25–28). We showed that concurrent ASXL1 and NRAS mutations define a population of CMML patients with shorter leukemia-free survival compared to patients with ASXL1 mutation only (45). Corroborating these human data, we discovered that Asxl1−/− accelerates CMML progression and promotes CMML transformation to AML (secondary AML, sAML) in NrasG12D/+ mice. Although NrasG12D/+; Asxl1−/− (NA) model shares common genetic mutations with the published NrasG12D/+; Ezh2−/− (54) and Nf1+/−; Asxl1+/− (55) models, it displays distinct phenotypes and molecular mechanisms from the other two. NrasG12D/+; Asxl1−/− (NA) leukemia cells exhibited hyperactivation of MEK/ERK signaling and increased global level of H3K27Ac, a histone mark bound by bromodomain and extra-terminal domain (BET) proteins for gene transcriptional activation (45). NA-sAML cells were more immunosuppressive than NA-CMML cells and overexpressed all the major inhibitory immune checkpoint ligands, PD-L1/L2, CD155, and CD80/86 (45). Among them, overexpression of PD-L1 and CD86 correlated with upregulation of AP-1 transcription factors (TFs) in NA-sAML cells (45). An AP-1 inhibitor and shRNAs against AP-1 TF Jun decreased PD-L1 and CD86 expression in NA-AML cells. Once NA-sAML cells were transplanted into syngeneic recipients, NA-derived T cells were not detectable (45). Host-derived wildtype T cells overexpressed inhibitory immune checkpoint receptors, PD-1 and TIGIT, and displayed an exhausted T cell phenotype (45). Combined inhibition of MEK and pan-BET proteins led to downregulation of AP-1 TF expression, mitigation of the suppressive immune microenvironment, enhancement of CD8 T cell cytotoxicity, and prolonged survival in NA-sAML mice. Given the distinct phenotypes observed in Asxl1−/− and Asxl1tm/+ mice, it would be interesting to determine if NrasG12D/+; Asxl1tm/+ mice display similar phenotypes as NA mice and if the underlying mechanisms are distinct. Treatment response of ASXL1 mutant leukemic cells Recent studies revealed that patients with ASXL1 mutations are associated with distinct sensitivity to drug treatment. Hypomethylating agents (HMA) have been a standard treatment for CMML. A retrospective study of 177 CMML patients revealed that ASXL1 mutations predict a lower overall response rate to HMAs (azacitidine or decitabine) (ASXL1mt 42% versus ASXL1wt 60%, p = 0.02) (58). This clinical observation may be explained, at least partially, by the increased expression of anti-apoptotic gene BCL2 and elevated global cytosine methylation in ASXL1mt leukemia cells (59). Not surprisingly, combined veneteoclax, a selective BCL2 inhibitor, and azacitidine effectively inhibit ASXL1mt leukemia cell growth in vitro (59). This combination was approved by FDA to treat AML in 2018. It would be interesting to see if it treats ASXL1mt CMML patients better than HMA alone. In summary, ASXL1 hotspot mutations around codon G646 are prevalent in myeloid diseases but rarely identified in CHIP, suggesting that they are highly pathogenic and confers higher risk to develop myeloid diseases. The nature of these mutations remains elusive. While mouse genetic studies suggest that they are loss-of-function and gain-of-new function at the physiological level, overexpression studies in transgenic mice and cell lines indicate that they are dominant negative and gain-of-function. Additional rigorous investigations are needed to provide a definitive answer to this question. ASXL1 mutations are associated with poor prognosis in all myeloid diseases, perhaps due to the reduced response to current treatment options (e.g. HMA in CMML). We recently discover that concurrent ASXL1 and NRAS mutations define dismal outcomes in CMML patients. Correspondingly, Asxl1 loss cooperates with oncogenic Nras in mice to reprogram immune microenvironment and drive leukemic transformation. Our study provides a strong rational to develop combined targeted therapy and immunotherapy for treating leukemia patients with concurrent ASXL1 and NRAS mutations. Figure 1. Schematic representation of structure of wildtype ASXL1 and mutant ASXL1. Figure 2. Distinct ASXL1 mutation spectrums in CHIP vs myeloid diseases. This figure summarizes published datasets of CHIP in healthy elderlies and cancer patients and of patients with myeloid diseases (including MPN, MDS, CMML, and AML). References 1. Sinclair DA , The Additional sex combs gene of Drosophila encodes a chromatin protein that binds to shared and unique Polycomb group sites on polytene chromosomes. Development 125 , 1207–1216 (1998).9477319 2. Schuettengruber B , Bourbon HM , Di Croce L , Cavalli G , Genome Regulation by Polycomb and Trithorax: 70 Years and Counting. Cell 171 , 34–57 (2017).28938122 3. Fisher CL , Additional sex combs-like 1 belongs to the enhancer of trithorax and polycomb group and genetically interacts with Cbx2 in mice. Dev Biol 337 , 9–15 (2010).19833123 4. Fisher CL , Randazzo F , Humphries RK , Brock HW , Characterization of Asxl1, a murine homolog of Additional sex combs, and analysis of the Asx-like gene family. Gene 369 , 109–118 (2006).16412590 5. Fisher CL , Berger J , Randazzo F , Brock HW , A human homolog of Additional sex combs, ADDITIONAL SEX COMBS-LIKE 1, maps to chromosome 20q11. Gene 306 , 115–126 (2003).12657473 6. Scheuermann JC , Histone H2A deubiquitinase activity of the Polycomb repressive complex PR-DUB. Nature 465 , 243–247 (2010).20436459 7. Abdel-Wahab O , ASXL1 mutations promote myeloid transformation through loss of PRC2-mediated gene repression. Cancer Cell 22 , 180–193 (2012).22897849 8. Asada S , Mutant ASXL1 cooperates with BAP1 to promote myeloid leukaemogenesis. Nat Commun 9 , 2733 (2018).30013160 9. Balasubramani A , Cancer-associated ASXL1 mutations may act as gain-of-function mutations of the ASXL1-BAP1 complex. Nat Commun 6 , 7307 (2015).26095772 10. Yamamoto K , A histone modifier, ASXL1, interacts with NONO and is involved in paraspeckle formation in hematopoietic cells. Cell Rep 36 , 109576 (2021).34433054 11. Baskind HA , Functional conservation of Asxl2, a murine homolog for the Drosophila enhancer of trithorax and polycomb group gene Asx. PLoS One 4 , e4750 (2009).19270745 12. Farber CR , Mouse genome-wide association and systems genetics identify Asxl2 as a regulator of bone mineral density and osteoclastogenesis. PLoS Genet 7 , e1002038 (2011).21490954 13. Izawa T , ASXL2 Regulates Glucose, Lipid, and Skeletal Homeostasis. Cell Rep 11 , 1625–1637 (2015).26051940 14. Li J , Loss of Asxl2 leads to myeloid malignancies in mice. Nat Commun 8 , 15456 (2017).28593990 15. Micol JB , ASXL2 is essential for haematopoiesis and acts as a haploinsufficient tumour suppressor in leukemia. Nat Commun 8 , 15429 (2017).28516957 16. Duployez N , Unlike ASXL1 and ASXL2 mutations, ASXL3 mutations are rare events in acute myeloid leukemia with t(8;21). Leuk Lymphoma 57 , 199–200 (2016).25856206 17. Dawoud AAZ , Tapper WJ , Cross NCP , Clonal myelopoiesis in the UK Biobank cohort: ASXL1 mutations are strongly associated with smoking. Leukemia 34 , 2660–2672 (2020).32518416 18. van Zeventer IA , Prevalence, predictors, and outcomes of clonal hematopoiesis in individuals aged >/=80 years. Blood Adv 5 , 2115–2122 (2021).33877299 19. Genovese G , Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med 371 , 2477–2487 (2014).25426838 20. Xie M , Age-related mutations associated with clonal hematopoietic expansion and malignancies. Nat Med 20 , 1472–1478 (2014).25326804 21. Jaiswal S , Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med 371 , 2488–2498 (2014).25426837 22. Jaiswal S , Clonal hematopoiesis and nonhematologic disorders. Blood 136 , 1606–1614 (2020).32736379 23. Jaiswal S , Clonal Hematopoiesis and Risk of Atherosclerotic Cardiovascular Disease. N Engl J Med 377 , 111–121 (2017).28636844 24. Asada S , Kitamura T , Clonal hematopoiesis and associated diseases: A review of recent findings. Cancer Sci 112 , 3962–3971 (2021).34328684 25. Chen TC , Dynamics of ASXL1 mutation and other associated genetic alterations during disease progression in patients with primary myelodysplastic syndrome. Blood cancer journal 4 , e177 (2014).24442206 26. Elena C , Integrating clinical features and genetic lesions in the risk assessment of patients with chronic myelomonocytic leukemia. Blood 128 , 1408–1417 (2016).27385790 27. Itzykson R , Prognostic score including gene mutations in chronic myelomonocytic leukemia. J Clin Oncol 31 , 2428–2436 (2013).23690417 28. Patnaik MM , ASXL1 and SETBP1 mutations and their prognostic contribution in chronic myelomonocytic leukemia: a two-center study of 466 patients. Leukemia 28 , 2206–2212 (2014).24695057 29. Abdel-Wahab O , Kilpivaara O , Patel J , Busque L , Levine RL , The most commonly reported variant in ASXL1 (c.1934dupG;p.Gly646TrpfsX12) is not a somatic alteration. Leukemia 24 , 1656–1657 (2010).20596031 30. Montes-Moreno S , Clinical molecular testing for ASXL1 c.1934dupG p.Gly646fs mutation in hematologic neoplasms in the NGS era. PLoS One 13 , e0204218 (2018).30222780 31. Alberti MO , Discriminating a common somatic ASXL1 mutation (c.1934dup; p.G646Wfs*12) from artifact in myeloid malignancies using NGS. Leukemia 32 , 1874–1878 (2018).29959414 32. Metzeler KH , ASXL1 mutations identify a high-risk subgroup of older patients with primary cytogenetically normal AML within the ELN Favorable genetic category. Blood 118 , 6920–6929 (2011).22031865 33. Thol F , Prognostic significance of ASXL1 mutations in patients with myelodysplastic syndromes. J Clin Oncol 29 , 2499–2506 (2011).21576631 34. Patnaik MM , DNMT3A mutations are associated with inferior overall and leukemia-free survival in chronic myelomonocytic leukemia. Am J Hematol 92 , 56–61 (2017).27733013 35. Binder M , Oncogenic gene expression and epigenetic remodeling of cis-regulatory elements in ASXL1-mutant chronic myelomonocytic leukemia. Nat Commun 13 , 1434 (2022).35301312 36. Abdel-Wahab O , Deletion of Asxl1 results in myelodysplasia and severe developmental defects in vivo. J Exp Med 210 , 2641–2659 (2013).24218140 37. Uni M , Modeling ASXL1 mutation revealed impaired hematopoiesis caused by derepression of p16Ink4a through aberrant PRC1-mediated histone modification. Leukemia 33 , 191–204 (2019).29967380 38. Bohring A , Severe end of Opitz trigonocephaly (C) syndrome or new syndrome? Am J Med Genet 85 , 438–446 (1999).10405439 39. Bedoukian E , Copenheaver D , Bale S , Deardorff M , Bohring-Opitz syndrome caused by an ASXL1 mutation inherited from a germline mosaic mother. Am J Med Genet A 176 , 1249–1252 (2018).29681100 40. Seiter K , Htun K , Baskind P , Liu Z , Acute myeloid leukemia in a father and son with a germline mutation of ASXL1. Biomark Res 6 , 7 (2018).29456859 41. Hsu YC , The distinct biological implications of Asxl1 mutation and its roles in leukemogenesis revealed by a knock-in mouse model. J Hematol Oncol 10 , 139 (2017).28697759 42. Nagase R , Expression of mutant Asxl1 perturbs hematopoiesis and promotes susceptibility to leukemic transformation. J Exp Med 215 , 1729–1747 (2018).29643185 43. Yang H , Gain of function of ASXL1 truncating protein in the pathogenesis of myeloid malignancies. Blood 131 , 328–341 (2018).29113963 44. Inoue D , A novel ASXL1-OGT axis plays roles in H3K4 methylation and tumor suppression in myeloid malignancies. Leukemia 32 , 1327–1337 (2018).29556021 45. You X , Asxl1 loss cooperates with oncogenic Nras in mice to reprogram the immune microenvironment and drive leukemic transformation. Blood 139 , 1066–1079 (2022).34699595 46. Inoue D , Truncation mutants of ASXL1 observed in myeloid malignancies are expressed at detectable protein levels. Exp Hematol 44 , 172–176 e171 (2016).26700326 47. Bejar R , Clinical effect of point mutations in myelodysplastic syndromes. N Engl J Med 364 , 2496–2506 (2011).21714648 48. Schnittger S , ASXL1 exon 12 mutations are frequent in AML with intermediate risk karyotype and are independently associated with an adverse outcome. Leukemia 27 , 82–91 (2013).23018865 49. Inoue D , SETBP1 mutations drive leukemic transformation in ASXL1-mutated MDS. Leukemia 29 , 847–857 (2015).25306901 50. Makishima H , Somatic SETBP1 mutations in myeloid malignancies. Nat Genet 45 , 942–946 (2013).23832012 51. Meggendorfer M , SETBP1 mutations occur in 9% of MDS/MPN and in 4% of MPN cases and are strongly associated with atypical CML, monosomy 7, isochromosome i(17)(q10), ASXL1 and CBL mutations. Leukemia 27 , 1852–1860 (2013).23628959 52. Gelsi-Boyer V , Mutations in ASXL1 are associated with poor prognosis across the spectrum of malignant myeloid diseases. J Hematol Oncol 5 , 12 (2012).22436456 53. Pratcorona M , Acquired mutations in ASXL1 in acute myeloid leukemia: prevalence and prognostic value. Haematologica 97 , 388–392 (2012).22058207 54. Gu Z , Loss of EZH2 Reprograms BCAA Metabolism to Drive Leukemic Transformation. Cancer Discov 9 , 1228–1247 (2019).31189531 55. Zhang P , Chromatin regulator Asxl1 loss and Nf1 haploinsufficiency cooperate to accelerate myeloid malignancy. J Clin Invest 128 , 5383–5398 (2018).30226831 56. Guo Y , Reduced BAP1 activity prevents ASXL1 truncation-driven myeloid malignancy in vivo. Leukemia 32 , 1834–1837 (2018).29743720 57. Saika M , ASXL1 and SETBP1 mutations promote leukaemogenesis by repressing TGFbeta pathway genes through histone deacetylation. Sci Rep 8 , 15873 (2018).30367089 58. Duchmann M , Prognostic Role of Gene Mutations in Chronic Myelomonocytic Leukemia Patients Treated With Hypomethylating Agents. EBioMedicine 31 , 174–181 (2018).29728305 59. Rahmani NE , ASXL1 mutations are associated with distinct epigenomic alterations that lead to sensitivity to venetoclax and azacytidine. Blood Cancer J 11 , 157 (2021).34548471

PMC010xxxxxx/PMC10364185.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 0376340 5287 J Surg Res J Surg Res The Journal of surgical research 0022-4804 1095-8673 35033819 10364185 10.1016/j.jss.2021.11.015 NIHMS1914303 Article Distal Small Bowel Resection Yields Enhanced Intestinal and Colonic Adaptation Tecos Maria E. MD ab Steinberger Allie E. MD ac Guo Jun PhD a Warner Brad W. MD a* a Division of Pediatric Surgery, Department of Surgery, St. Louis Children’s Hospital, Washington, University in St. Louis School of Medicine, St. Louis, Missouri b Division of General Surgery, Department of Surgery, University of Nebraska Medical Center, Omaha, Nebraska c Division of General Surgery, Department of Surgery, Barnes Jewish Hospital, Washington, University in St. Louis School of Medicine, St. Louis, Missouri * Corresponding author. Division of Pediatric Surgery, Department of Surgery, Washington University School of Medicine, St. Louis Children’s Hospital, One Children’s Place, Suit 6110, St. Louis, MO 63110. Tel.: (314) 454 0622; fax: (314) 454 2442. brad.warner@wustl.edu (B.W. Warner). 20 7 2023 5 2022 13 1 2022 24 7 2023 273 100109 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Background: Murine ileocecal resection (ICR) has been employed to investigate intestinal adaptation. The established model often includes the sacrifice of significant length of proximal colon. Here, we optimized a highly selective vascular approach to the ICR, with primary jejunal-colic anastomosis yielding maximal colonic preservation. Materials and Methods: 40 C57BL/6 mice underwent a highly vascularly selective ICR. The terminal branches of the ileocecal artery are isolated apart from the mesenteric branches supplying the small bowel to be resected. The distal 50% of small bowel and cecum are resected; a primary jejuno-colonic anastomosis is performed. Animals were sacrificed at postoperative weeks 2 (n=10) and 10 (n=29). Proximal 50% small bowel resection (SBR) with jejuno-ileal anastomosis was also performed for comparison. Results: The entire colon (with exception of the cecum) was preserved in 100% of animals. 97% of animals survived to postoperative week 10, and all exhibited structural adaptation in the remnant small intestine epithelium. Crypts deepened by 175% and villi lengthened by 106%, versus 39% and 29% in the proximal SBR cohort, respectively. Colonic proliferation, structural adaptation, and functional adaptation (measured by p-histone 3, luminal-facing apical crypt border size, and sucrase isomaltase, respectively) were increased in ICR compared to proximal SBR. Conclusions: Highly selective isolation of the cecal vasculature allows for greater colon preservation and yields enhanced remnant intestine epithelial adaptation. ICR is also associated with greater colonic adaptation and unique plasticity toward an intestinal phenotype. These findings underscore major differences between resection sites and offer insights into the critical adaptive mechanisms in response to massive intestinal loss. Short bowel syndrome intestinal adaptation colonic adaptation small bowel resection ileocecal resection intestinal failure pmcIntroduction Murine SBR has long been utilized to study a range of intestinal failure associated disease processes. By experimentally manipulating elements such as diet, remnant intestinal length, and residual intestine location, it is possible to create scenarios that more closely mirror the clinical variables that are present in patients with short bowel syndrome. To date, there has been no direct comparison of the impact of remnant intestinal location in these models, nor has the potential role of the colon in intestinal adaptation been fully elucidated. Several iterations of murine SBR techniques are employed to isolate specific portions of the intestine. In the standard proximal resection model, 50–75% of the small intestine is resected, leaving behind the ileum.1 One advantage of this model is that the remnant ileum has long been considered to demonstrate greater degrees of structural adaptation than the remnant jejunum. As such, the magnitude of alterations in gene expression that contribute to adaptation are likely to be greater.2–6 In addition, studies of adapting remnant bowel distal to an anastomosis eliminate the potential confounder of partial anastomotic obstruction. Partial obstruction alone is known to augment several features of structural adaptation such as villus and crypt growth.7 A 50% distal resection has been previously described that leaves behind mostly jejunal tissue. Historically, this 50% distal resection removes the cecum and at least 1–2 cm of proximal colon beyond the cecum.8 The sacrificed ileocecal valve (ICV) and length of proximal colon significantly contribute to overall colon function when intact. A highly selective vascular approach to the ICR has been previously described that allows for the cecum to be removed while retaining the entire proximal colon.9 However, this isolated approach was limited to the immediate peri-cecal tissue and was not combined with an SBR. The retention of the entirety of the colon may be important for multiple reasons. Firstly, the ability to hone the resected segments of tissue intraoperatively allows for more highly specific surgical models of disease. Secondly, an aboral gradient of gene expression is known to occur along the gastrointestinal tract for many genes. For example, toll-like receptor (TLR) 4 has a strong expression throughout the colon, while TLRs 2 and 5 have more dense expression in the small intestine and proximal colon.10–11 Our previous work has identified abrogation of hepatic injury12 and paracellular intestinal permeability13 after knockout of TLR 4. The impact of resection locus with inherent variations in genetic expression therefore requires further investigation. The ability of the remnant small intestine to adapt after massive SBR is characterized by both morphological increases in the size of its functional units, the crypts and villi,14 as well as alterations in cell signaling.15 Enhanced physical adaptation has been observed in the presence of dietary fat,16 and interleukin 22 (IL22; mediated by lipocalin 2).17 The ability to harness and modify these structural and functional changes with measurable impact speaks to the translatability of the SBR model, and offers insight on future subjects of manipulation. Here, we detail a modification on the previously established ICR techniques that combines aspects of the aforementioned distal resection approaches to yield a novel operative model. Our highly selective vascular approach allows for distal massive SBR including the cecum, while maintaining the proximal colon in its entirety, with consistently enhanced adaptation and resultant survival that approached 100%. Methods Mice C57BL/6 mice were purchased from the Jackson Laboratory (Bar Habor, ME) and housed at Washington University in St. Louis animal holding facilities with a 12–12 hour light-dark cycle (St. Louis, MO). They were provided ample food and water, available 24 hours per day. All 30 mice in the 10-week cohort were included in the survival analysis. However, only data from male C57BL/6 mice that underwent surgery at the eighth week of life were included in experimental quantification assessments, as an age and gender matched cohort to the proximal SBR group. All experiments were performed within the parameters of protocol 20–0197 as approved by the Washington University Animal Studies Committee and are in compliance with National Institute of Health animal care guidelines. The inclusion of only male mice in quantification assessments may be viewed as a limitation of this study. Analyses were limited to male mice to provide the most accurate data comparisons across current and historical experimental cohorts without the introduction of potential confounding variables from hypothetical physiological and hormonal variations in a mixed gender sampling. Gender differences in murine short gut models is an area of active research. Operations A highly selective vascular approach to the distal massive SBR was constructed in an effort to enhance resection locus specificity and optimize survival. (Table 1, Figures 1–2) At 24 hours prior to operation, mice were switched from solid pellet chow to a liquid diet (LD; PMI Micro-Stabilized Rodent Liquid Diet LD 101; TestDiet, St. Louis, MO). Under isoflurane anesthesia, all mice underwent 50% highly selective distal SBR to include the cecum while sparing the remaining proximal colon. Preoperatively, mice were weighed, given a 1 cc bolus of normal saline, and placed on a heating pad. A midline laparotomy is made, and the bowel externalized. The bowel is splayed and oriented in a proximal to distal fashion, with the cecum and proximal colon as the distal points. At the distal point of the dissection, the terminal branches of the ileocecal artery are identified, isolated, and encircled with a 3–0 silk suture. Attention is then turned to the proximal point of dissection. Twelve centimeters are measured proximally from the ileocecal junction, a window made in the mesentery, and the mesenteric vessels supplying the segment of small bowel to be removed are similarly ligated. The bowel is then transected proximally at the junction of ischemic and non-ischemic bowel. The distal transection is made at the junction of the cecum and proximal colon. A hand-sewn end-to-end anastomosis is then performed with 9–0 nylon in an interrupted fashion. Postoperatively, animals are given a 1 cc normal saline bolus at 24 hours after surgery and housed for 1 week in an incubator. For the first 24 hours after operation, they are provided water ad libitum. After this period, they are again given liquid diet. All animals are weighed weekly and harvested at postoperative week 10. The proximal resection is performed in the same manner, with the resection and anastomosis being performed on the proximal 50% of small bowel rather than the distal 50% of small bowel, as we have previously described.1 Histology Intraoperative (IO) samples of small bowel were taken during distal small bowel resection. At the time of harvest, postoperative (PO) samples of remnant small bowel, liver, and colon were taken. The PO small bowel samples are obtained from the same 4 cm mid-small intestine segment of transition between jejunum and ileum. (Figure 1b and d) The rationale for these sampling locations is to show the areas of most respective change for each type of surgery immediately about the same intestinal region. As such, the PO sections from the proximal SBR are referred to as remnant ileum, while the PO sections from the ICR are referred to as remnant jejunum. All samples were formalin-fixed and paraffin embedded. Slides were cut and sections were stained with hematoxylin and eosin or left unstained for further specific staining. Adaptation Intraoperative small bowel and remnant postoperative small bowel hematoxylin and eosin-stained slides were examined at 20x magnification. Crypt depth and villus length were measured in at least 30 well-oriented crypts and villi in each sample. All slides were analyzed by a single blinded investigator to maintain consistency in length counting method while eliminating selection bias. Crypt depth and luminal-facing apical crypt border were similarly measured in the colon. Quantification of data was performed in GraphPad Prism. 2 week and 10-week postoperative C57BL/6 mice were used in the ICR cohort for adaptation analyses, as structural intestinal adaptation plateaus by postoperative week 1.1 Expression of sucrase isolmaltase (Life Technologies; Carlsbad, CA) was measured in the colon to assess functional adaptation. Proliferation IO and PO small bowel slides were stained for p-histone 3 (Cell Signaling Technologies; Danvers, MA). The average number of positively stained cells per section was measured in each sample. All slides were analyzed by a single investigator. Quantification of data was performed in GraphPad Prism. Samples from 2-week postoperative C57BL/6 mice were used in both cohorts for proliferation analyses. Western Blot was carried out for p-histone 3 (Cell Signaling Technologies; Danvers, MA). Quantification for Western Blot analysis was performed via Image Lab software (Bio-Rad; Hercules, CA). Real Time PCR Crypts and villi were isolated after harvest and homogenized in lysis buffer as previously described.13 The provided manufacturer’s protocol was followed for RNA extraction (RNAqueous Kit, Ambion; Austin, TX). A NanoDrop Spectophotomer was used to measure sample RNA concentration (ND-1000, NanoDrop Technologies; Wilmington, DE). Lipocalin 2, IL22, and sucrase isomaltase primers and endogenous control primer GAPDH were purchased from Applied Biosystems (Life Technologies; Carlsbad, CA). RT PCR was carried out via the Applied Biosystems 7500 Fast Real Time PCR system. All results were normalized to the GAPDH endogenous control. Western Blot Tissue harvested for protein analysis was extracted via bead homogenization at 60 hz for 60 seconds, and heated at 100°C for 5 minutes, as previously described.13 Protein concentration was measured per the manufacturer’s protocol using protein assay RC DC kit (Bio-Rad; Hercules, CA). 80 ug per sample were loaded into a 12% polyacrylamide gel, with Novex Sharp Pre-Stained Protein used as standard (Invitrogen; Carlsbad, CA). Dry transfer onto a nitrocellulose membrane (Invitrogen; Carlsbad, CA) was carried out via the iBLOT Gel Transfer Device (Invitrogen; Carlsbad, CA). Primary antibodies for p-histone 3 and GAPDH, and rabbit secondary antibody were purchased from Cell Signaling Technologies (Danvers, MA). The membrane was developed with Amersham ECL Western Blotting Detection Reagents (GE Healthcare; Chicago, IL). Protein detection and quantification was performed with Image Lab Software (Bio-Rad; Hercules, CA). Statistical Analysis Individual values for all data points were imported into GraphPad Prism and distribution was assessed via a panel including Anderson-Darling, D’Agostino and Pearson, Shapiro-Wilk, and Kolmogorov-Smirnov testing. T-tests were used to compare datasets found to have normal distribution, and Mann-Whitney tests were employed for datasets with non-gaussian distribution. ANOVA was utilized for multiple comparisons. A p-value of 0.05 was used as the threshold for statistical significance. Normally distributed data parameters are reported as mean and standard deviation (SD), while non-normally distributed data parameters are reported as median and interquartile range (IQR). Results Survival A diverse cohort of mice were used to establish general durable survival of the operative technique. Ten male 8-week-old C57BL/6 mice and 20 male and female C57BL/6 mice of various ages underwent the novel ICR procedure. Twenty-nine of 30 distal SBR mice survived to postoperative week 10, at which point they were harvested. One mouse died at postoperative week 6 as a result of cage fighting. There was no difference in survival when data was stratified for gender, or age (data not shown). Ten of 10 proximal SBR mice survived to postoperative week 10. Each distal SBR subject successfully regained and stabilized weight without extensive postoperative fluid resuscitation (Figure 3). Further analyses regarding differences between proximal and distal resection were carried out between age and gender matched cohorts of male C57BL/6 mice. There were no differences in food intake or weight between the two groups (data not shown). Structural and Functional Adaptation Every animal at postoperative weeks 2 and 10 demonstrated statistically significant adaptation. Histological evidence of adaptation was assessed microscopically at 20x magnification, by measuring change in crypt depth and villus height. (Figure 4 e–h) Crypt depth and villus length were counted by a blinded single investigator. When IO and PO sizes were compared, crypts deepened on average 175% (p < 0.0001, Figure 4a), and villi lengthened an average of 106% (p < 0.0001, Figure 4b). In comparison, proximal massive SBR produced a 39% and 29% change in crypt and villus sizes, respectively. The non-significant difference between IO crypt and villus sizes when proximal and distal resections were compared ensures that each set of samples was obtained from the same approximate location along the mid-small intestine jejunal-ileal transition zone, and that a 50% resection was indeed performed during each operation. The colon was similarly assessed for structural adaptation. Crypt depth was not significantly different between proximal and distal SBR (p = 0.8, data not shown). However, obvious morphological differences were evident at the luminal-facing apical border of the distal SBR colon crypts (comprised of lamina propria and columnar epithelium), which appeared thicker and were organized into villus-like projections (Figure 5a and b). When these differences were compared, the luminal-facing apical border of the distal SBR colon crypts was significantly increased compared to their proximal SBR counterparts (p < 0.0001, Figure 5c). Further, these villus-like projections found in the distal SBR colon did indeed express a greater amount of sucrase isomaltase, as measured by PCR (p = 0.0312, Figure 6d). Proliferation Both remnant small intestine and colon proliferation after proximal and distal SBR were assessed by measuring mitosis marker p-histone 3 via Western Blot (Figure 6a and b). Western Blot quantification revealed an increase in small intestine proliferation after proximal resection compared to distal resection (p = 0.0262). Taken together with the increased histological adaptation seen after distal resection compared to the proximal resection, it follows that the postoperative adaptation process may be completed more efficiently in the distal resection model. A significant increase in colonic proliferation was observed after distal resection compared to proximal resection (p = 0.0111; Figure 6c and d). Immunohistochemical staining for p-histone 3 was also performed in post-operative remnant small intestine and colon for each surgical model. Colonic p-histone 3 staining was markedly pronounced in the colon after distal SBR compared to proximal SBR, as viewed at 40x magnification (Figure 6e and f). Small bowel immunohistochemistry staining for p-histone 3 appeared grossly equivocal prior to quantification (data not shown). Discussion In the present study, we have described a unique distal ICR model which preserves the post-cecal proximal colon. We show that this model is well tolerated and associated with excellent survival. We also revealed greater structural features of remnant small bowel and colon adaptation when compared with a more proximal SBR. Colon preservation and greater adaptation of the small bowel and colon would therefore be clinically beneficial for patients with short gut syndrome. Studies that can discriminate the contributions of small bowel and colon are therefore likely to provide a more precise understanding of physiologic responses to massive intestinal loss. It has previously been shown that intestinal lipocalin 2 is increased after proximal SBR compared to sham operation, and that its presence may act to slow postoperative adaptation via a mechanism that includes downregulation of IL22. Knockout of lipocalin 2 was also found to promote adaptation.17 Here, we observed no differences in remnant small bowel lipocalin 2 or IL22 when proximal and distal SBR were compared (p = 0.4176 and p = 0.6068, data not shown). These disparate findings may be due to several factors including different lengths of resection (75% versus 50%), varied postoperative times for study (1 and 3 week versus 2 and 10 week), as well as site of resection. Further, a major factor to consider and a current focus of study in our laboratory is to elucidate changes in gut microbiota between proximal SBR and distal ICR models. Since the ileum is the major site of bile acid transport and there is significant interplay between bile acids and gut microbiota, it is logical to characterize these differences. While the differences in small bowel lipocalin 2 and IL22 have been shown between sham and proximal SBR, the lack of difference between proximal and distal SBR is likely multifactorial in nature, with potential influences by resection locus, resection length, postoperative timepoints, microbiome changes, and bile acid recycling. The cecum is removed along with the distal ileum in our model. We therefore are not able to test the significance of the ICV in resection associated adaptation responses. Clinically, the ICV may be beneficial as its presence is associated with enhanced ability to wean from parenteral nutrition.18–19 The mechanism for the ICV is likely due to slowing of intestinal transit, thereby allowing for longer contact of nutrient with the absorptive intestinal surface. In addition, the large murine cecum is a site of intestinal stasis and therefore has a significant impact on the resident microbial community. It must be considered that the lack of an ICV is typically associated with varied degrees of colon resection. Greater magnitudes of colon resection results in diminished capacity for fluid and electrolyte absorption. Since our model maximally preserves the colon, we are able to study the maximal contribution of the colon to many adaptive responses, independent of the confounder of the ICV. Our highly selective vascular approach to the ICR combines two established murine surgical models for an optimal result. In doing so, it allows for 50% massive distal SBR, including the cecum, with retention of the entirety of the proximal colon. The significance of the colon and its possible impact on intestinal adaptation after resection have yet to be fully investigated. By preserving the colon in its entirety distal to the cecum, any potential colonic-driven mechanisms that could impact adaptation remain unadulterated. With typical total colon length of 6–8 cm as observed at harvest, the sacrifice or preservation of even 1–2 cm (12.5–33.3% of the functional colon) during distal SBR may be significant. This may account for the greater structural adaptation of the remnant bowel in our experiments, compared with the disparate degrees of adaptation in the historic ICR model. Past distal SBR adaptation analysis has shown peak villus heights near 500 μm by postoperative week 6, and crypt depths approaching 200 μm at postoperative days 4–5.8 Staining for actively dividing cells after resection revealed increased proliferation within the colon of distal compared to proximal SBR subjects. Further, after ICR we showed structural adaptation not only marked by increase in the luminal-facing apical border of colon, but also organization of the lamina propria and columnar epithelium into villus-like structures. This apparent morphological intestinalization was correlated with functional adaptation by expressing increased levels of sucrase isomaltase, an enzyme the activity of which is largely specific to small intestinal villi.20 Prior comparisons of proximal and distal SBR have failed to demonstrate a morphological difference in their impact on the colon and were also limited in their functional assessment.2 A possible explanation for the profound adaptive changes observed in our model compared to others may be found in the structural alteration inherent in the ICR. The free passage of enteric contents without the hindrance of the ICV functioning as a physiologic checkpoint provides increased stimulation to the downstream colonic tissue, which appears to translate to enhanced proliferative activation. Preservation of the cecum and ICV in a massive distal SBR would present technical challenges due to the caliber discrepancies between the distal ileum and the more proximal jejunum. Recent surgical models have targeted manipulations that can serve to harness the colon to become more functionally similar to the small intestine.21 Indeed, this is the rationale to generate a functional small intestinalized colon by replacing the native colonic epithelium with ileum-derived organoids.22 Enhancing colonic adaptation is therefore an attractive area of investigation. Our findings of enhanced structural and functional colon adaptation after ICR will pave the way toward understanding the pathogenesis of this important response. This concept is supported by the documented proximalization of remnant distal small intestine after proximal SBR.23 The standard SBR models have been extensively characterized in terms of metabolic derangements, intestinal permeability, proliferation, and adaptation.24–31 However, a direct comparison of proximal versus distal resection, and their respective resultant sequelae has yet to be studied. The findings we have demonstrated challenge the previously accepted dogma of remnant ileum having more profound post-resection adaptation potential compared to remnant jejunum.2–6 It is possible that our unique outcome of enhanced remnant small bowel adaptation after ICR compared to proximal SBR is influenced by both environment and physiological factors inherent to our model. The combination of remnant small bowel long-term exposure to a high fat diet,16 potential intestinal microbiome differences after proximal versus distal SBR,32 increased inflammation secondary to augmented SBR-associated bacterial translocation,13,33 and resection-induced disruption of site-specific physiologic functioning6 may well contribute to a post-SBR milieu that favors enhanced adaptation of the remnant jejunum in our model. Continued optimization of locus-specific massive SBR surgical models will provide for the most stable foundation upon which to measure these techniques against each other in terms of cellular signaling, adaptation, metabolic effects, and systemic impact. As demonstrated, the durability of this model is enhanced. This precision of targeted locus-specific resection directly translates to improved postoperative outcomes and the ability to create unique surgical models. Acknowledgements This project was generously funded by NIH RO1 DK104698, NIH T32 DK007120, and NIH T32 DK077653. Figure 1. Proximal SBR and Distal SBR Surgery Schematics A. Schematic highlighting traditional 50% proximal massive SBR in blue, and location of post-resection anastomosis. B. Schematic depiction of resected tissue, as well as intra- and postoperative sample segments taken for histological analysis at the time of resection and harvest, respectively. The pink bar represents the entire length of small bowel. The blue bar correlates to the 50% proximal segment of small bowel that is resected. The red bar denotes the intra-operative segment taken during resection. The yellow bar portrays the post-operative segment taken at harvest. C. Intestinal tract schematic highlighting novel colon-sparing 50% massive ICR in green, and location of post-resection anastomosis. D. Schematic depiction of resected tissue, as well as intra- and post-operative sample segments taken for histological analysis at the time of resection and harvest, respectively. The green bar correlates to the 50% distal segment of small bowel that is resected. The red bar denotes the intra-operative segment taken during resection. The yellow bar portrays the post-operative segment taken at harvest. Figure 2. Operative Photos A. Initial orientation of bowel after extravasation from abdomen (Table 1, Step 1). B. Complete resected length of cecum and small intestine, measuring 12 cm (50% total small bowel length). Most proximal 1–2 cm taken for intraoperative histology sample (Figure 1d). C. Completed 50% massive ICR, with anastomosis circled in white. D. Identification and encircling of terminal branch of ileocecal artery (Table 1, Step 2). E. Orientation of both vascular ties, showing the sparing of the proximal portion of the ileocecal artery (Table 1, Step 3). F. Transection of the mesentery supplying the small bowel to be resected (Table 1, Step 5), the terminal branch of the ileocecal artery (Table 1, Step 6), and the cecal-proximal colon junction (Table 1, Step 7). G. Completion of single layer handsewn end-to-end jejunal-colonic anastomosis, with anastomosis circled in white (Table 1, Step 8). Figure 3. Postoperative Weight Trajectory No significant difference was found when the postoperative weight trajectory of massive proximal and distal SBRs were compared (p = 0.1061). Proximal resection n = 10. Distal resection n = 9. Error bars represent standard deviations. Figure 4. Intraoperative and Postoperative Histology A. Average percent change in crypt depth from intraoperative (IO) to postoperative (PO) tissue sample was significantly increased in distal SBR compared to proximal SBR (mean 175% ± SD 81% versus mean 39% ± SD 16%; ****, p ≤ 0.0001). B. PO crypt depth was significantly increased in both resection types when compared to IO crypt depth (proximal SBR; ***, p ≤ 0.001, distal SBR ****, p ≤ 0.0001). Crypts deepened more after distal SBR than after proximal SBR (****, p ≤ 0.0001). There was no difference in IO crypt depth between resection types. Mean proximal IO crypt size 69 ± SD 3 μM, mean proximal PO crypt size 104 ± SD 9 μM, mean distal IO crypt size 52 ± SD 8 μM, mean distal PO crypt size 138 ± SD 28 μM. C. Median percent change in villus height from IO to PO tissue sample was significantly increased in distal Revised 9/27/21 SBR compared to proximal SBR (median 106% with IQR 137% versus median 29% with IQR 36%; ****, p ≤ 0.0001). D. PO villus height was significantly increased in both resection types when compared to IO villus height (proximal SBR, ****, p ≤ 0.0001; distal SBR, ****, p ≤ 0.0001). Villi lengthened more after distal SBR than after proximal SBR (****, p ≤ 0.0001). There was no difference in IO villus between resection types. Proximal SBR resection n = 10, distal resection n = 19 for all comparisons. Mean proximal IO villus size 219 ± SD 24 μM, mean proximal PO villus size 306 ± SD 29 μM, mean distal IO villus size 190 ± SD 25 μM, mean distal PO villus size 384 ± 50 μM. E. Hematoxylin and eosin-stained intraoperative sample from 50% proximal SBR (10x magnification). F. Hematoxylin and eosin-stained postoperative sample from 50% proximal SBR (10x magnification). G. Hematoxylin and eosin-stained intraoperative sample from 50% distal SBR (10x magnification). H. Hematoxylin and eosin-stained postoperative sample from 50% distal SBR (10x magnification). Figure 5. Functional and Structural Distal Small Bowel Resection-Induced Adaptation A. Hematoxylin and eosin-stained colon tissue after 50% proximal SBR (40x magnification). B. Hematoxylin and eosin-stained colon sample after 50% distal SBR (40x magnification). C. Luminal-facing apical colon crypt border (identified as projection of lamina propria and columnar epithelium at apex of crypt) was significantly larger after distal SBR compared to proximal SBR (****, p ≤ 0.0001). Mean proximal SBR PO luminal colon crypt border size 37 ± SD 6 μM, mean distal SBR PO luminal colon crypt border size 80 ± SD 19 μM. D. Colonic sucrase isomaltase expression was significantly increased after distal SBR compared to proximal SBR (*, p ≤ 0.05). Median proximal SBR PO colon sucrase isomaltase 5 with IQR 31. Median distal SBR PO colon sucrase isomaltase 36 with IQR 111. Figure 6. Quantification of Proliferation A. Western Blot comparing proliferation in remnant small intestine after proximal and distal SBR as characterized by p-histone expression, with GAPDH as an internal control. B. Quantification of remnant small bowel proliferation as measured by p-histone 3 protein expression after proximal and distal SBR, showing increased proliferation after proximal SBR (*, p ≤ 0.05). Median proximal SBR small bowel p-H3 7 with IQR 10, median distal SBR small bowel p-H3 5 with IQR 6. C. Western Blot comparing proliferation in colon after proximal and distal SBR as characterized by p-histone 3 expression, with GAPDH as an internal control. D. Quantification of colon proliferation as measured by p-histone 3 protein expression after proximal and distal SBR, showing greater proliferation after distal SBR (*, p ≤ 0.05). Median proximal SBR colon p-H3 2 with IQR 10, median distal SBR colon p-H3 9 with IQR 3. E. Immunohistochemistry staining for p-histone 3 in colon tissue after proximal small bowel resection (40x magnification). F. Immunohistochemistry staining for p-histone 3 in colon tissue after distal small bowel resection (40x magnification). Table 1 HIGHLY VASCULAR SELECTIVE MURINE MASSIVE ILEOCECAL RESECTION OPERATIVE TECHNIQUE STEP 1 Extravasate small bowel and cecum, splaying the intestine such that it progresses from proximal to distal in a counter-clockwise orientation. (Figure 2a) STEP 2 Identify the most terminal branch of the ileocecal artery, create a window in the mesentery, and encircle the vessel with a 3–0 silk tie. STEP 3 Measure 12 cm (approximately 50%) of small bowel beginning from the ileocecal junction. Create a window in the mesentery, and encircle the mesenteric vessels supplying this section of small bowel with a second 3–0 silk tie. Take care to spare the proximal aspect of the ileocecal artery; only the terminal end of the ileocecal artery should be ligated, as was performed in Step 3. STEP 4 Transect the bowel proximally at the junction of ischemic and viable tissue. STEP 5 Transect the mesenteric vessels supplying the 50% of small bowel to be resected. STEP 6 Transect the isolated terminal branch of the ileocecal artery. STEP 7 Transect the cecal-proximal colon junction. STEP 8 Arrange the proximal and distal cut ends such that their respective mesenteries are aligned. Create a single layered end-to-end handsewn anastomosis with 9–0 nylon suture. STEP 9 Test integrity of anastomosis by gently compressing the peri-anastomotic tissue, noting any leakage of enteric contents. Place additional sutures where necessary. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. References 1. Helmrath MA , VanderKolk WE , Can G , Erwin CR , Warner BW . Intestinal adaptation following massive small bowel resection in the mouse. J Am Coll Surg 183 : 441–449, 1996.8912612 2. Buccigrossi V , Armellino C , Tozzi A , Nicastro E , Esposito C , Alicchio F , Cozzolino S , Guarino A . Time-and segment-related changes of postresected intestine: a 4-dimensional model of intestinal adaptation. Journal of pediatric gastroenterology and nutrition. 2013 Jan 1;56 (1 ):40–5.22820122 3. Haxhija EQ , Yang H , Spencer AU , Sun X , Teitelbaum DH . Intestinal epithelial cell proliferation is dependent on the site of massive small bowel resection. Pediatric surgery international. 2007 May 1;23 (5 ):379–90.17205293 4. Křížová J , Trachta P , Kotrlíková E . Intestinal adaptation following gut resection. Vnitrni lekarstvi. 2017 Jan 1;63 (10 ):703–6.29127753 5. Gutierrez IM , Fisher JG , Ben-Ishay O , Jones BA , Kang KH , Hull MA , Shillingford N , Zurakowski D , Modi BP , Jaksic T . Citrulline levels following proximal versus distal small bowel resection. Journal of Pediatric surgery. 2014 May 1;49 (5 ):741–4.24851760 6. Tappenden KA . Pathophysiology of short bowel syndrome: considerations of resected and residual anatomy. Journal of Parenteral and Enteral Nutrition. 2014 May;38 :14S–22S.24500909 7. Collins III J , Vicente Y , Georgeson K , Kelly D . Partial intestinal obstruction induces substantial mucosal proliferation in the pig. Journal of pediatric surgery. 1996 Mar 1;31 (3 ):415–9.8708915 8. Dekaney CM , Fong JJ , Rigby RJ , Lund PK , Henning SJ , Helmrath MA . Expansion of intestinal stem cells associated with long-term adaptation following ileocecal resection in mice. American Journal of Physiology-Gastrointestinal and Liver Physiology. 2007 Nov;293 (5 ):G1013–22.17855764 9. Perry T , Borowiec A , Dicken B , Fedorak R , Madsen K . Murine ileocolic bowel resection with primary anastomosis. JoVE (Journal of Visualized Experiments). 2014 Oct 29 (92 ):e52106. 10. Price AE , Shamardani K , Lugo KA , Deguine J , Roberts AW , Lee BL , Barton GM . A map of Toll-like receptor expression in the intestinal epithelium reveals distinct spatial, cell type-specific, and temporal patterns. Immunity. 2018 Sep 18;49 (3 ):560–75.30170812 11. Gourbeyre P , Berri M , Lippi Y , Meurens F , Vincent-Naulleau S , Laffitte J , Rogel-Gaillard C , Pinton P , Oswald IP . Pattern recognition receptors in the gut: analysis of their expression along the intestinal tract and the crypt/villus axis. Physiological Reports. 2015 Feb;3 (2 ):e12225.25677543 12. Barron LK , Bao JW , Aladegbami BG , Colasanti JJ , Guo J , Erwin CR , Warner BW . Toll-like receptor 4 is critical for the development of resection-associated hepatic steatosis. Journal of pediatric surgery. 2017 Jun 1;52 (6 ):1014–9.28351520 13. Courtney CM , Onufer EJ , McDonald KG , Steinberger AE , Sescleifer AM , Seiler KM , Tecos ME , Newberry RD , Warner BW . Small Bowel Resection Increases Paracellular Gut Barrier Permeability via Alterations of Tight Junction Complexes Mediated by Intestinal TLR4. Journal of Surgical Research. 2021 Feb 1;258 :73–81.33002664 14. Helmrath MA , VanderKolk WE , Can G , Erwin CR , Warner BW . Intestinal adaptation following massive small bowel resection in the mouse. Journal of the American College of Surgeons. 1996 Nov 1;183 (5 ):441–9.8912612 15. Seiler KM , Waye SE , Kong W , Kamimoto K , Bajinting A , Goo WH , Onufer EJ , Courtney C , Guo J , Warner BW , Morris SA . Single-cell analysis reveals regional reprogramming during adaptation to massive small bowel resection in mice. Cellular and molecular gastroenterology and hepatology. 2019 Jan 1;8 (3 ):407–26.31195149 16. Choi PM , Sun RC , Guo J , Erwin CR , Warner BW . High-fat diet enhances villus growth during the adaptation response to massive proximal small bowel resection. Journal of Gastrointestinal Surgery. 2014 Feb 1;18 (2 ):286–94.24002772 17. Zhang A , Sodhi CP , Wang M , Shores DR , Fulton W , Prindle T , Brosten S , O’Hare E , Lau A , Ding H , Jia H , Lu P , White JR , Hui J , Sears CL , Hackam DJ , Alaish SM . A Central Role for Lipocalin-2 in the Adaptation to Short-Bowel Syndrome Through Down-Regulation of IL22 in Mice. Cellular and molecular gastroenterology and hepatology. 2020 Jan 1;10 (2 ):309–26.32330729 18. Enman MA , Wilkinson LT , Meloni KB , Shroyer MC , Jackson TF , Aban I , Dimmitt RA , Martin CA , Galloway DP . Key determinants for achieving enteral autonomy and reduced parenteral nutrition exposure in pediatric intestinal failure. Journal of Parenteral and Enteral Nutrition. 2020 Sep;44 (7 ):1263–70.31840829 19. Jo SY , McCallum Z , Shalley H , Peterkin M , Rogers E , Paxton G , Wong T , Bines JE . Outcomes of children with chronic intestinal failure: Experience over 2 decades at a tertiary paediatric hospital. Journal of pediatric gastroenterology and nutrition. 2019 Sep 1;69 (3 ):e79–87.31169663 20. Traber PG , Yu L , Wu GD , Judge TA . Sucrase-isomaltase gene expression along crypt-villus axis of human small intestine is regulated at level of mRNA abundance. American Journal of Physiology-Gastrointestinal and Liver Physiology. 1992 Jan 1;262 (1 ):G123–30. 21. Deng G , Deng Z . enhancement of colonic Absorptive function after the Massive Resection of the Small intestine Based on the creation of an Artificial Colonic Valve. Scientific reports. 2020 Jan 21;10 (1 ):1–9.31913322 22. Sugimoto S , Kobayashi E , Fujii M , Ohta Y , Arai K , Matano M , Ishikawa K , Miyamoto K , Toshimitsu K , Takahashi S , Nanki K . An organoid-based organ-repurposing approach to treat short bowel syndrome. Nature. 2021 Apr;592 (7852 ):99–104.33627870 23. Seiler KM , Waye SE , Kong W , Kamimoto K , Bajinting A , Goo WH , Onufer EJ , Courtney C , Guo J , Warner BW , Morris SA . Single-cell analysis reveals regional reprogramming during adaptation to massive small bowel resection in mice. Cellular and molecular gastroenterology and hepatology. 2019 Jan 1;8 (3 ):407–26.31195149 24. Seiler KM , Goo WH , Zhang Q , Courtney C , Bajinting A , Guo J , Warner BW . Adaptation of extracellular matrix to massive small bowel resection in mice. Journal of pediatric surgery. 2020 Jun 1;55 (6 ):1107–12.32164986 25. Seiler KM , Schenhals EL , von Furstenberg RJ , Allena BK , Smith BJ , Scaria D , Bresler MN , Dekaney CM , Henning SJ . Tissue underlying the intestinal epithelium elicits proliferation of intestinal stem cells following cytotoxic damage. Cell and tissue research. 2015 Aug;361 (2 ):427–38.25693894 26. Berlin P , Reiner J , Wobar J , Bannert K , Glass Ä , Walter M , Bastian M , Willenberg HS , Vollmar B , Klar E , Seidler U . Villus growth, increased intestinal epithelial sodium selectivity, and hyperaldosteronism are mechanisms of adaptation in a murine model of short bowel syndrome. Digestive diseases and sciences. 2019 May;64 (5 ):1158–70.30569336 27. Courtney CM , Onufer EJ , Seiler KM , Warner BW . An anatomic approach to understanding mechanisms of intestinal adaptation. InSeminars in pediatric surgery 2018 Aug (Vol. 27 , No. 4 , pp. 229–236). 28. Onufer EJ , Han YH , Courtney CM , Steinberger AE , Tecos ME , Sutton S , Sescleifer A , Ou J , Sanguinetti Czepielewski R , Randolph GJ , Warner BW . Liver injury after small bowel resection is prevented in obesity-resistant 129S1/SvImJ mice. American Journal of Physiology-Gastrointestinal and Liver Physiology. 2021 Mar 17. 29. Onufer EJ , Han YH , Czepielewski RS , Courtney CM , Sutton S , Randolph GJ , Warner BW . Effects of high-fat diet on liver injury after small bowel resection. Journal of pediatric surgery. 2020 Jun 1;55 (6 ):1099–106.32164985 30. Berlin P , Reiner J , Witte M , Wobar J , Lindemann S , Barrantes I , Kreikemeyer B , Bastian M , Schäffler H , Bannert K , Jaster R . Nod2 deficiency functionally impairs adaptation to short bowel syndrome via alterations of the epithelial barrier function. American Journal of Physiology-Gastrointestinal and Liver Physiology. 2019 Nov 1;317 (5 ):G727–38.31509436 31. Reiner J , Berlin P , Wobar J , Schäffler H , Bannert K , Bastian M , Vollmar B , Jaster R , Lamprecht G , Witte M . Teduglutide promotes epithelial tight junction pore function in murine short bowel syndrome to alleviate intestinal insufficiency. Digestive diseases and sciences. 2020 Dec;65 (12 ):3521–37.32072437 32. Barron LK , Gayer CP , Roberts A , Golden JM , Aladegbami BG , Guo J , Erwin CR , Warner BW . Liver steatosis induced by small bowel resection is prevented by oral vancomycin. Surgery. 2016 Dec 1;160 (6 ):1485–95.27592213 33. Sukhotnik I , Haj B , Pollak Y , Dorfman T , Bejar J , Matter I . Effect of bowel resection on TLR signaling during intestinal adaptation in a rat model. Surgical endoscopy. 2016 Oct;30 (10 ):4416–24.26895894

PMC010xxxxxx/PMC10364190.txt

LICENSE: This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. 9805456 20853 Mol Genet Metab Mol Genet Metab Molecular genetics and metabolism 1096-7192 1096-7206 36434903 10364190 10.1016/j.ymgme.2022.11.004 NIHMS1914394 Article Chloroquine corrects enlarged lysosomes in FIG4 null cells and reduces neurodegeneration in Fig4 null mice Lenk Guy M. 1 Meisler Miriam H. 12 1 Department of Human Genetics, University of Michigan, Ann Arbor MI 48109-5618 2 Department of Neurology, University of Michigan, Ann Arbor MI 48109 Correspondence to: Guy M. Lenk, Ph. D., Department of Human Genetics, 4909 Buhl Box 5618, University of Michigan, Ann Arbor, MI 48109-5618, glenk@umich.edu 16 7 2023 12 2022 12 11 2022 24 7 2023 137 4 382387 This file is available for text mining. It may also be used consistent with the principles of fair use under the copyright law. Loss-of-function mutations of FIG4 impair the biosynthesis of PI(3,5)P2 and are responsible for rare genetic disorders including Yunis-Varón Syndrome and Charcot-Marie-Tooth Disease Type 4J. Cultured cells deficient in FIG4 accumulate enlarged lysosomes with hyperacidic pH, due in part to impaired regulation of lysosomal ion channels and elevated intra-lysosomal osmotic pressure. We evaluated the effects of the FDA approved drug chloroquine, which is known to reduce lysosome acidity, on FIG4 deficient cell culture and on a mouse model. Chloroquine corrected the enlarged lysosomes in FIG4 null cells. In null mice, addition of chloroquine to the drinking water slowed progression of the disorder. Growth and mobility were dramatically improved during the first month of life, and spongiform degeneration of the nervous system was reduced. The median survival of Fig4 null mice was increased from 4 weeks for untreated mutants to 8 weeks with chloroquine treatment (p<0.009). Chloroquine thus corrects the lysosomal swelling in cultured cells and ameliorates Fig4 deficiency in vivo. The improved phenotype of mice with complete loss of Fig4 suggests that chloroquine might be beneficial in partial loss-of-function disorders such as Charcot-Marie-Tooth Type 4J. PI(3,5)P2 FIG4 chloroquine lysosomal disorder neurodegeneration CMT4J pmcIntroduction FIG4 is a subunit of the protein complex responsible for biosynthesis and turnover of the signaling lipid PI(3,5)P2 at the lysosome membrane (1). PI(3,5)P2 regulates the activity of several lysosomal ion channels and transporters, including TRPML1, TPC1, TPC2 and CLCN7 (2–5). In cultured cells and in mice, mutations of FIG4 result in deficiency of PI(3,5)P2 and accumulation of greatly enlarged lysosomes (1, 6). In addition to the enlarged size and lack of stored material, lysosomal pH in FIG4 null cells is reduced from wildtype lysosomes by 0.9 pH unit (7). Pathogenic variants of FIG4 have been identified in patients with recessive disorders. Complete loss of human FIG4 results in Yunis-Varón Syndrome, a lethal multisystem disorder with major effects in CNS and bone (8). Compound heterozygosity for one null allele and one partially functional allele results in a variable level of neurological dysfunction (9). Individuals with Charcot-Marie-Tooth disorder type 4J carry a null allele of FIG4 and the missense variant p.Ile41Thr that destabilizes interaction with VAC14. The p.Ile41Thr variant is present at an allele frequency of 0.001 in European populations, and homozygous individuals have been identified (10). Other partial loss-of-function variants result in polymicrogyria with epilepsy (OMIM 612619) and pediatric neurodegeneration with hypomyelination (11). Cultured fibroblasts from patients with these disorders contain characteristic enlarged lysosomes (12, 13). There are currently no treatments available for FIG4 deficiency disorders. The spontaneous mouse mutant pale tremor (plt) is a model of Yunis-Varón Syndrome. The null mutation was caused by retrotransposon insertion into intron 18 of Fig4, resulting in complete loss of transcript and protein (6). The mice exhibit spongiform neurodegeneration of the CNS and PNS and juvenile lethality. CNS myelination is defective due to failed maturation of oligodendrocyte precursor cells (14, 15). The Fig4 null phenotype can be rescued by neuron-specific expression of a Fig4 cDNA transgene (16). The lysosomotrophic agent chloroquine is an anti-malarial drug that has been used in cell culture to alkalinize the lysosome and inhibit autophagy (5, 17, 18). Rescue of enlarged, hyperacidified lysosomes by chloroquine was recently demonstrated in cells with mutation of the lysosomal choride transporter CLCN7 (5, 19). Chloroquine crosses the blood brain barrier and the placenta, and is secreted by nursing mice into the milk, without ill effects on developing embryos or pups (20, 21). Here we describe the positive effects of chloroquine on FIG4 null cells and mice. In FIG4 null HAP1 cells, chloroquine rescued the enlarge lysosomes associated with reduced PI(3,5)P2. In Fig4 null mice, chloroquine improved growth and mobility, reduced neurodegeneration and increased lifespan. Chloroquine may thus be useful for treatment of individuals with genetic disorders resulting from deficiency of FIG4. Results Treatment of FIG4 null cells with chloroquine CRISPR/Cas9 targeting of human HAP1 cells generated the FIG4−/− clonal cell line 3D4 (7). 3D4 cells were plated at equal cell density in medium containing 10 uM chloroquine or in control medium. After 18 hours in the presence of chloroquine, the prominent enlarged lysosomes characteristic of Fig4 null cells were no longer visible (Figure 1A). To quantitate the effect, cells were stained with LysoSensor DND160, trypsinized, and analyzed by FACS sorting (7). In the absence of chloroquine, 82% of 3D4 cells exhibit elevated fluorescence due to accumulation of LysoSensor in the enlarged lysosomes (Figure 1B). In the presence of chloroquine, the proportion of cells with elevated fluorescence was reduced to 20% (Figure 1B). Rescue of vacuolization persisted during 7 days of culture, through several cell passages (Figure 1C). Data from replica experiments are summarized in Figure 1D. The extent of rescue was dependent on the concentration of chloroquine in the cell media (Figure 1E). The striking rescue of the cellular defect in Fig4 null cells suggested that chloroquine might improve the in vivo phenotypes of null mice. In vivo effects of chloroquine on Fig4 null mice Congenic male C57BL/6J.Fig4+/− mice were crossed with congenic female C3HeB/FeJ.Fig4+/− mice to generate Fig4−/− null mice on the (BL/6XC3H)F1 strain background (22). Drinking water supplemented with chloroquine (0.11 mg/ml) was provided prior to conception and continued throughout the lifetime of the mice. Cages were monitored twice daily to detect all births, including perinatal deaths. Homozygous Fig4−/− null pups were recovered at the predicted Mendelian frequency of 25% in chloroquine treated cages (11/42) and untreated cages (18/73). Homozygous Fig4−/− offspring treated with chloroquine were larger and more robust than untreated homozygotes (Figure 2A). The average weight of 3 week old homozygotes receiving chloroquine was 8.9 ± 1.2 g (mean ± SD, n=9), compared with 5.3 ± 0.1 g (n=17) for untreated homozygotes (Figure 2B) (p<0.0001). The body weight of treated Fig4−/− mice remained lower than wildtype littermates (14.9 ± 1.9 g; n=19) (Figure 2B). Untreated Fig4−/− homozygotes exhibit hunched posture, tremor, spasticity, swimming gait, periods of immobility, and hydrocephalus (see Supplemental Video) (6). At 30 days of age, chloroquine-treated mice were dramatically improved in mobility, and demonstrated spontaneous running, rearing and grooming behaviors (see Supplemental Video). The substantial effects of chloroquine on size and activity at 30 days of age were striking. Reduced neurodegeneration after treatment with chloroquine Tissues of Fig4 null mice accumulate enlarged lysosomes that can be immunostained for the lysosomal membrane markers LAMP1 and LAMP2 (6, 22). Accumulation of these vacuoles in the central and peripheral nervous system is prevented by neuronal expression of a wildtype Fig4 transgene (16). Consistent with the improved movement, the accumulation of vacuoles in brain and spinal cord was improved by the chloroquine treatment. The extent of vacuolization was quantitated in brain, spinal cord and dorsal root ganglion. The fraction of dorsal root ganglion neurons containing vacuoles was significantly reduced in the treated mice at P21 (Figure 3). The area occupied by vacuoles was significantly reduced in spinal cord and approached significant reduction in cerebral cortex (Figure 3). In Fig4 null mice, nerve conduction velocity in sural and sciatic nerves is reduced by 50% compared with wildtype mice (6, 15, 23, 24). However, nerve conduction velocity was not improved in the treated mice (Table 1). Effect of chloroquine on survival of Fig4 null mice Fig4−/− mice with complete loss of Fig4 exhibit median survival of 4 weeks on the (B6/J XC3H)F1 strain background (Figure 4). Treatment with chloroquine increased the length of survival of Fig4−/− mice to 8 weeks (Figure 4) (p=0.009, Log-Rank test). Discussion We observed substantial positive effects of chloroquine on the abnormalities seen in Fig4 null cells in culture and Fig4 null mice in vivo. In FIG4 null cells, culture in the presence of 10 uM chloroquine resulted in more than 50% reduction in the content of enlarged lysosomes. This effect persisted through one week of culture with many cell divisions and several rounds of replating. In vivo, chloroquine treatment doubled the rate of growth of Fig4 null mice during the first month after birth. The gait and mobility of null mice was largely rescued at 4 weeks of age, an age when untreated homozygotes are significantly impaired. Consistent with improved motor function, spongiform degeneration in spinal cord and dorsal root ganglia was reduced by approximately 50%. Overall, chloroquine delayed disease progression and lengthened survival. However, the positive effects of treatment did not persist, and the null mice did not survive beyond 2 months. It is possible that a higher dose of chloroquine would be more effective. The Fig4 null mouse studied here provides a stringent test of the effectiveness of chloroquine, since the mouse completely lacks Fig4 function. Most patients with FIG4 disorders retain partial gene function, with the exception of the early lethal Yunis-Varón Syndrome. Fig4 expression is required throughout life, and inactivation of a floxed allele in adult mice was lethal within two months (24). The chloroquine treatment itself was well tolerated in wildtype mice. Chloroquine has a good safety profile in clinical use as an anti-malarial drug and also in long term application for systemic lupus erythematosus (25) and as an anti-inflammatory agent that inhibits cytokine production (26). Chloroquine is a small lipophilic molecule that permeates cells and organelles and accumulates in protonated form in lysosomes (27). It is a weak base that elevates lysosomal pH, possibly by acting as a proton sink (17). The effect of chloroquine in reducing lysosome swelling suggests that the elevated proton concentration in Fig4 null lysosomes is partially responsible for swelling (7). Alternatively, the elevated proton concentration in Fig4 lysosomes, or the reduction in PI(3,5)P2 concentration, may influence the flux of other ions. Fig4 null mice were rescued by gene replacement using a Fig4 cDNA transgene (16). Expression of the Fig4 cDNA from the neuron-specific Nse1 promoter rescued the null phenotype, indicating that lethality is a consequence of the neuronal defect (16). A cDNA with a mutation of the Fig4 phosphatase active site rescued the enlarged lysosomes in cultured cells. The mutant cDNA also rescued the survival of null mice for 5 months, indicating that the FIG4 protein has additional function beyond its phosphatase activity (28). The enzymatically inactive protein may act to stabilize the PI(3,5)P2 biosynthesis complex, as it does in yeast (23, 29, 30). Gene replacement by viral delivery of a Fig4 cDNA was recently shown to produce longterm survival of Fig4 null mice (31), and efforts to translate this result to the clinic are in progress. At present, however, there is no treatment for patients with FIG4 deficiency. The improved phenotypes reported here for the null mouse suggest that chloroquine could be useful in less severe disorders such as Charcot-Marie-Tooth Type 4J (32) and cerebral hypomyelination (11). Based on the positive observations in the preclinical model, the safety profile in the clinic, and the lack of alternative treatments, we suggest that chloroquine be evaluated for beneficial effects in individuals with FIG4 deficiency disorders. Methods Reagents. Chloroquine phosphate was purchased from Sigma Pharmaceuticals (# C6628) and stored at −20°C. Concentrated aqueous stock solutions containing 22 mg/ml chloroquine were prepared weekly, filter sterilized, and stored at 4°C. Diluted working solutions were prepared daily in cell culture medium or drinking water. Drinking water contained 0.11 mg/ml chloroquine and 15 g/l glucose, which was estimated to provide an in vivo chloroquine dose of 15 mg/kg and was compatible with live birth after prenatal treatment (21). Cell Assays. The FIG4 null clonal cell clone 3D4 is homozygous for a 13 bp deletion in exon 6 that was generated by CRISPR/Cas9 targeting in human HAP1 cells (7). Cells were cultured in a humidified incubator at 37°C with 5% CO2 in IMDM culture medium (Gibco 12440053) supplemented with 1X Antibiotic-Antimycotic (Gibco 15240062) and 10% fetal bovine serum (Corning 35-010-CV). Cells with enlarged lysosomes were quantitated by FACS as previously described (7). Cells were incubated for 1 hour with LysoSensor DND160 (Invitrogen L7545) to stain lysosomes followed by cell sorting in an MoFlo Astrios sorter. Fluorescence was measured with excitation at 355 nm and emission at 546 nm. Animals. The spontaneous Fig4 null mutation arose on a mixed genetic background and was named pale tremor (plt) based on its diluted pigmentation and intention tremor (6). The mutant allele has been backcrossed separately to strains C57BL/6J and C3HeB/FeJ for more than 30 generations each, resulting in two congenic lines with different lifespan. Fig4−/− homozygotes survive to embryonic day E18 on strain C57BL/6J, to postnatal day P10 on strain C3HeB/FeJ, and to 4 weeks of age on the hybrid (C57BL/6J X C3HeB/FeJ)F1 background (22, 31). The experiments described here were carried out on homozygous null mice with the F1 background obtained from crosses between the two congenic lines. Mice were maintained in static housing. To standardize maternal care, litters were reduced to 4 to 6 pups at postnatal day 4. Fig4 pups were identified by their diluted pigmentation at P3 (6). All offspring were subsequently genotyped. Mice were housed and cared for in accordance with NIH guidelines in a 12/12h light/dark cycle with standard mouse chow and water available ad libitum. All experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Michigan. Histology. Brain, spinal cord and dorsal root ganglia from 21 day old mice were fixed at 4°C for 24 hours in phosphate buffered 10% formalin and for an additional 24 hours in 70% ethanol. Paraffin embedding, decalcification of spine, H&E staining and imaging were carried out at Histoserv Inc. (Germantown, Maryland). The area occupied by vacuoles was quantitated with ImageJ software. Nerve conduction velocity. Nerve conduction velocity in sural and sciatic nerve was measured in 30 day old mice as previously described (6, 23, 33). Supplementary Material Supplemental Video 1 Supplemental video. Chloroquine improves motor coordination of Fig4 null mice. Video recording of an untreated Fig4 null mouse (red mark) and a chloroquine-treated null mouse at 30 days of age. Time, 1 min 10 sec. Acknowledgements. We thank Drs. James Shayman, Andrew Lieberman and Jess Thoene for helpful discussions of the manuscript, John Hayes for measurement of nerve conduction velocity and Ann Marie DesLauriers-Cox for assistance with FACs sorting. We are grateful to Jocelyn Duff for advice and support. Funding. This work was supported by NIH R01 GM024872 and the Cure CMT4J Foundation. Figure 1. Chloroquine corrects enlarged lysosomes in FIG4−/− null cells. (A) Enlarged lysosomes in FIG4 null HAP1 cell line 3D4 are rescued by culture for 18 hours in the presence of 10 uM chloroquine. (B, C) Quantitation of enlarged lysosomes by FACS sorting after culture with chloroquine for 18 hours (B) or 7 days (C). Wildtype HAP1 cells with normal lysosome content were clustered to the left of the dotted line. (D) Data from replicate FACS sorting. (E) The extent of correction of enlarged vacuoles is directly proportional to chloroquine concentration in the cell medium. Figure 2. In vivo effects of chloroquine in Fig4 null mice. Fig4 null mice were given drinking water containing 0.11 mg/ml chloroquine throughout prenatal and postnatal life. The supplemented water was provided to pregnant females during gestation. (A) Increase in body size and posture of treated mice at 4 weeks of age. The improved mobility of the treated mice is shown in the Supplemental Video. (B) Partial rescue of body weight in chloroquine-treated mice at postnatal day 21. Figure 3. Chloroquine treatment reduced spongiform degeneration in brain, spinal cord and dorsal root ganglia at three weeks of age. Individual neurons were clearly visible in dorsal root ganglia and the neurons containing vacuoles were counted. In the sections of cerebral cortex and spinal cord, the area occupied by vacuoles was quantitated using ImageJ software. In cerebral cortex, the difference between untreated and chloroquine-treated mice approached significance. Figure 4. Increased postnatal survival of Fig4 null mice after chloroquine treatment. Fig4 null mice treated with chloroquine from conception survived longer than untreated Fig4 null controls (p=0.009; Log Rank test). Table 1: Chloroquine does not rescue nerve conduction velocity in Fig4 null mice. Measurements were carried out at 21 days of age. Values are m/sec ± standard deviation (n). Genotype Sural nerve Sciatic nerve +/+ 18.8 ± 0.6 (2) 28.3 ± 2.1 (2) Fig4−/− 12.1 ± 1.2 (2) 17.2 ± 1.1 (2) Fig4−/− + CQ 12.5 ± 1.1 (4) 18.4 ± 4.6 (4) Competing interests. The authors report no competing interests. Supplementary material. The supplementary video is available online. References 1. Hasegawa J , Strunk BS , Weisman LS . (2017). PI5P and PI(3,5)P2: Minor, but Essential Phosphoinositides. Cell structure and function, 42 , 49–60.28302928 2. Dong XP , (2010). PI(3,5)P(2) controls membrane trafficking by direct activation of mucolipin Ca(2+) release channels in the endolysosome. Nature communications, Jul 13;1 (4 ):38. 3. She J (2018). Structural insights into the voltage and phospholipid activation of the mammalian TPC1 channel. Nature, 556 , 130–134.29562233 4. Patel S , (2022). Activation of endo-lysosomal two-pore channels by NAADP and PI(3,5)P2. Five things to know. Cell calcium, 103 , 102543.35123238 5. Leray X , (2022). Tonic inhibition of the chloride/proton antiporter ClC-7 by PI(3,5)P2 is crucial for lysosomal pH maintenance. eLife, 11 , e74136.35670560 6. Chow CY , (2007) Mutation of FIG4 causes neurodegeneration in the pale tremor mouse and patients with CMT type 4G. Nature 448 :68–72.17572665 7. Lenk GM , (2019) CRISPR knockout screen implicates three genes in lysosome function. Scientific Reports 9 :9609.31270356 8. Campeau PM , (2013). Yunis-Varón syndrome is caused by mutations in FIG4, encoding a phosphoinositide phosphatase. American journal of human genetics, 92 , 781–791.23623387 9. Zimmermann M , (2020). FIG4 mutations leading to parkinsonism and a phenotypical continuum between CMT4J and Yunis Varón syndrome. Parkinsonism & related disorders 74 , 6–11.32268254 10. Lafontaine M , (2021). Clinical features of homozygous FIG4-p.Ile41Thr Charcot-Marie-Tooth 4J patients. Ann Clin Transl Neurol 8 , 471–476.33405357 11. Lenk GM , (2019). Cerebral hypomyelination associated with biallelic variants of FIG4. Human Mutation, 40 , 619–630.30740813 12. Zhang X , (2008). Mutation of FIG4 causes a rapidly progressive, asymmetric neuronal degeneration. Brain, 131 , 1990–2001.18556664 13. Baulac S , (2014). Role of the phosphoinositide phosphatase FIG4 gene in familial epilepsy with polymicrogyria. Neurology, 82 , 1068–1075.24598713 14. Winters JJ , (2011). Congenital CNS hypomyelination in the Fig4 null mouse is rescued by neuronal expression of the PI(3,5)P(2) phosphatase Fig4. J Neurosci. 31 (48 ), 17736–17751.22131434 15. Mironova YA , (2016). PI(3,5)P2 biosynthesis regulates oligodendrocyte differentiation by intrinsic and extrinsic mechanisms. eLife, 5 , e13023.27008179 16. Ferguson CJ , (2012). Neuronal expression of Fig4 is both necessary and sufficient to prevent spongiform neurodegeneration. Human Molecular Genetics, 21 , 3525–3534.22581779 17. Homewood CA , (1972). Lysosomes, pH and the anti-malarial action of chloroquine. Nature, 235 (5332 ), 50–52.4550396 18. Yayon A , Cabantchik ZI , Ginsburg H . (1985). Susceptibility of human malaria parasites to chloroquine is pH dependent. PNAS, 82 , 2784–2788.3887411 19. Nicoli ER , (2019). Lysosomal Storage and Albinism Due to Effects of a De Novo CLCN7 Variant on Lysosomal Acidification. American journal of human genetics, 104 (6 ), 1127–1138.31155284 20. Compter I (2021). Chloroquine combined with concurrent radiotherapy and temozolomide for newly diagnosed glioblastoma: a phase IB trial. Autophagy, 17 , 2604–2612.32866424 21. Keyaerts E , (2009). Antiviral activity of chloroquine against human coronavirus OC43 infection in newborn mice. Antimicrob Agents Chemother, 53 , 3416–3421.19506054 22. Lenk GM , Meisler MH . (2014). Mouse models of PI(3,5)P2 deficiency with impaired lysosome function. Methods in enzymology, 534 , 245–260.24359958 23. Lenk GM , (2011). Pathogenic mechanism of the FIG4 mutation responsible for Charcot-Marie-Tooth disease CMT4J. PLoS genetics June 7 (6 ), e1002104. 24. Mironova YA , (2018). Protective role of the lipid phosphatase Fig4 in the adult nervous system. Human molecular genetics, 27 , 2443–2453.29688489 25. Lee SJ , Silverman E , Bargman JM . The role of antimalarial agents in the treatment of SLE and lupus nephritis. Nat Rev Nephrol. 2011 Oct 18;7 (12 ):718–29.22009248 26. Schrezenmeier E , Dörner T . (2020). Mechanisms of action of hydroxychloroquine and chloroquine: implications for rheumatology. Nature Reviews Rheumatology, 16 , 155–166.32034323 27. Solomon VR , Lee H . (2009). Chloroquine and its analogs: a new promise of an old drug for effective and safe cancer therapies. European journal of pharmacology, 625 (1-3 ), 220–233.19836374 28. Lenk GM , (2016). Rescue of neurodegeneration in the Fig4 null mouse by a catalytically inactive FIG4 transgene. Human Molecular Genetics, 25 , 340–347.26604144 29. Jin N , (2008). VAC14 nucleates a protein complex essential for the acute interconversion of PI3P and PI(3,5)P(2) in yeast and mouse. The EMBO journal, 27 , 3221–3234.19037259 30. Botelho RJ , (2008). Assembly of a Fab1 phosphoinositide kinase signaling complex requires the Fig4 phosphoinositide phosphatase. Molecular biology of the cell 19 , 4273–4286.18653468 31. Presa M , (2021). AAV9-mediated FIG4 delivery prolongs life span in Charcot-Marie-Tooth disease type 4J mouse model. The Journal of Clinical Investigation, 131 , e137159.33878035 32. Nicholson G , (2011). Distinctive genetic and clinical features of CMT4J: a severe neuropathy caused by mutations in the PI(3,5)P2 phosphatase FIG4. Brain 134 , 1959–1971.21705420 33. Sullivan KA , Mouse models of diabetic neuropathy. Neurobiol Dis. 2007 Dec;28 (3 ):276–85.17804249